【关键词】 Distinct
Selective modulation of 7 nicotinic acetylcholine receptors (nAChRs) is thought to regulate processes impaired in schizophrenia, Alzheimer's disease, and other dementias. One approach to target 7 nAChRs is by positive allosteric modulation. Structurally diverse compounds, including PNU-120596, 4-naphthalene-1-yl-3a,4,5,9b-tetrahydro-3-H-cyclopenta[c]quinoline-8-sulfonic acid amide (TQS), and 5-hydroxyindole (5-HI) have been identified as positive allosteric modulators (PAMs), but their receptor interactions and pharmacological profiles remain to be fully elucidated. In this study, we investigated interactions of these compounds at human 7 nAChRs, expressed in Xenopus laevis oocytes, along with genistein, a tyrosine kinase inhibitor. Genistein was found to function as a PAM. Two types of PAM profiles were observed. 5-HI and genistein predominantly affected the apparent peak current (type I) whereas PNU-120596 and TQS increased the apparent peak current and evoked a distinct weakly decaying current (type II). Concentration-responses to agonists [ACh, 3-[(3E)-3-[(2,4-dimethoxyphenyl)methylidene]-5,6-dihydro-4H-pyridin-2-yl]pyridine dihydrochloride (GTS-21), and N-[(3R)-1-azabicyclo[2.2.2]oct-3-yl]-4-chlorobenzamide hydrochloride (PNU-282987)] were potentiated by both types, although type II PAMs had greater effects. When applied after 7 nAChRs were desensitized, type II, but not type I, PAMs could reactivate 7 currents. Both types of PAMs also increased the ACh-evoked 7 window currents, with type II PAMs generally showing larger potentiation. None of the PAMs tested increased nicotine-evoked Ca2+ transients in human embryonic kidney 293 cells expressing human 42 or 34 nAChRs, although some inhibition was noted for 5-HI, genistein, and TQS. In summary, our studies reveal two distinct 7 PAM profiles, which could offer unique opportunities for modulating 7 nAChRs in vivo and in the development of novel therapeutics for central nervous system indications.
Nicotinic acetylcholine receptors (nAChRs) belong to the pentameric superfamily of ligand-gated ion channels that includes 5HT3, GABAA, and glycine receptors. Twelve neuronal nicotinic subunits have been identified thus far (2–10; 2–4) of which nine subunits, 2–7 and 2–4, predominate in the mammalian brain (Paterson and Nordberg, 2000). Multiple functionally distinct nAChR complexes can be assembled either as homomeric functional pentamers, such as 7 nAChRs (Couturier et al., 1990), or as heteropentamers with at least two different subunits, such as 42 nAChRs (Gotti et al., 2006).
The role of 7 nAChRs in the CNS has received much attention since their discovery (Couturier et al., 1990). These subunits, when expressed in heterologous expression systems, activate and desensitize rapidly and, furthermore, exhibit relatively higher calcium permeability compared with other nAChR combinations (Dajas-Bailador and Wonnacott, 2004). In the brain, the 7 subunit is distributed at high levels, including in regions involved in learning and memory, hippocampus and cerebral cortex (Rubboli et al., 1994; Wevers et al., 1994; Breese et al., 1997). At the cellular level, activation of 7 nAChRs is thought to regulate interneuron excitability (Frazier et al., 1998), modulate the release of excitatory and inhibitory neurotransmitters (Alkondon et al., 2000), and contribute to neuroprotective effects in experimental in vitro models of cellular damage (Levin and Rezvani, 2002). Antisense (Curzon et al., 2006) and more recent gene knock-out studies (Wehner et al., 2004; Keller et al., 2005) have demonstrated that 7 nAChRs could play important roles in certain cognitive and attentive tasks. For example, 7 nAChR genetic knockout mice have shown impaired performance in ethanol-induced contextual fear conditioning (Wehner et al., 2004) and showed further deterioration in hippocampus-selective associative learning and memory when crossed with Tg2576 animals (Dineley et al., 2005). Selective 7 nAChR agonists such as PNU-282987 (Hajós et al., 2005), PHA-543613 (Wishka et al., 2006), and AR-R17779 (Felix and Levin, 1997; Van Kampen et al., 2004) improve performance in sensory gating, novel object recognition, social recognition, water maze performance, or inhibitory avoidance models of cognitive function. Given these roles, targeting 7 nAChRs has been considered as a viable strategy for a variety of diseases involving cognitive deficits and neurodegeneration (for review, see Levin and Rezvani, 2002; Gotti et al., 2006).
An alternate approach to enhance 7 nAChR function is by augmenting effects of the neurotransmitter ACh via positive allosteric modulation that could reinforce the endogenous cholinergic neurotransmission without directly activating 7 nAChRs. Indeed, such positive allosteric modulator (PAM) approach to enhance channel activity has been proven clinically successful for GABAA receptors (Hevers and Luddens, 1998). The preclinical validation of 7 nAChR PAMs will require selective compounds yet to be identified because many of the compounds identified so far are weak, nonselective or incompletely characterized pharmacologically. Various molecules have been reported to positively modulate 7 nAChR, including PNU-120596 (Hurst et al., 2005), 5-hydroxyindole (5-HI) (Zwart et al., 2002), ivermectin (Krause et al., 1998), galantamine (Samochocki et al., 2003), bovine serum albumin (Conroy et al., 2003), SLURP-1 (Chimienti et al., 2003), an acetylcholinesterase derived peptide (Zbarsky et al., 2004), (2-amino-5-keto)thiazole compounds (Broad et al., 2006), and compound 6 (Ng et al., 2007). Among these compounds, PNU-120596 and compound 6 improved auditory gating and other cognitive deficits (Hurst et al., 2005; Ng et al., 2007), supporting the concept that 7 nAChR PAMs may be effective in vivo. Genistein, a nonselective kinase inhibitor (Akiyama et al., 1987) has been shown to increase 7 responses (Charpantier et al., 2005; Cho et al., 2005). Although there was evidence that effects of genistein could be mediated through kinase inhibition, direct allosteric modulatory effects on 7 nAChR may be involved, and detailed studies aimed at identifying direct effects of genistein on 7 nAChR have yet to be carried out.
This study describes the pharmacological profiles of structurally diverse PAMs: 5-HI, PNU-120596, and TQS. In addition, evidence is presented that genistein also functions as an 7 nAChR PAM. The effects of these compounds were determined on recombinant 7 current evoked by diverse 7 agonists [ACh, GTS-21 (de Fiebre et al., 1995), and PNU-282987 (Bodnar et al., 2005)] as well as on 42 and 34 nAChRs to investigate their selectivity. Our results demonstrate the existence of at least two types of 7 nAChR PAMs based upon differential effects on current responses, reactivation of desensitized 7 nAChRs, augmentation of ACh window current, and agonist concentration-response characteristics. This study provides an insight into the understanding of PAM actions relevant to the design of novel compounds with potential therapeutic utility in diseases such as: Alzheimer's disease, schizophrenia, and attention deficit hyperactivity disorder, where 7 nAChRs are thought to play important roles.
Materials. Oocytes were obtained from adult female Xenopus laevis frogs (Blades Biological Ltd., Cowden, Edenbridge, Kent, UK) and cared for in accordance with the Institutional Animal Care and Use Committee guidelines. Genistein, 5-hydroxindole, herbimycin A, ACh, nicotine, choline, MLA, and BAPTA-AM were obtained from Sigma (St. Louis, MO). GTS-21 and staurosporine were purchased from Tocris (London, UK). PP2 and SU6656 were obtained from Biaffin GmbH and Co KG (Kassel, Germany). PNU-120596, TQS, and PNU-282987 were synthesized in-house. All other chemicals and reagents were obtained from Sigma or Fisher Scientific (Essex, UK).
Two-Electrode Voltage-Clamp on X. laevis Oocytes. X. laevis oocytes were prepared for electrophysiological experiments as described previously (Briggs et al., 1995; Briggs and McKenna, 1998). In brief, three to four lobes from ovaries of female adult X. laevis frogs were removed and defolliculated after treatment with collagenase type 1A (2 mg/ml; Sigma) prepared in low-Ca2+ Barth's solution [90 mM NaCl, 1.0 mM KCl, 0.66 mM NaNO3, 2.4 mM NaHCO3,10 mM HEPES, 2.5 mM sodium pyruvate, 0.82 mM MgCl2, and 0.5% (v/v) penicillin-streptomycin solution, pH = 7.55 (Sigma)] for 1.5 to 2 h at 18°C under constant agitation to obtain isolated oocytes. The oocytes were injected with 4 to 6 ng of human 7 nAChR cRNA, kept at 18°C in a humidified incubator in modified Barth's solution [90 mM NaCl, 1.0 mM KCl, 0.66 mM NaNO3, 2.4 mM NaHCO3,10 mM HEPES, 2.5 mM sodium pyruvate, 0.74 mM CaCl2, 0.82 mM MgCl2, 0.5% (v/v) penicillin-streptomycin solution, pH 7.55] and used 2 to 7 days after injection. Responses were measured by two-electrode voltage clamp using parallel oocyte electrophysiology test station (Abbott, Abbott Park, IL) (Trumbull et al., 2003). During recordings, the oocytes were bathed in Ba2+-OR2 solution (90 mM NaCl, 2.5 mM KCl, 2.5 mM BaCl2, 1.0 mM MgCl2, 5.0 mM HEPES, and 0.0005 mM atropine, pH 7.4) to prevent activation of Ca2+-dependent currents and held at –60 mV at room temperature (20°C). Modulators were given for 60 s before agonist application. Agonists were applied for 1 s at 6 ml/s with or without modulators to the recording chambers. The buffer flow to the chamber, however, did not resume until at least 3 s had passed. The parallel oocyte electrophysiology test station system, similar to any other electrophysiological setup using X. laevis oocytes, cannot apply 7 agonists fast enough to cause rapid and complete activation of 7 channels without desensitization; hence, the measured maximum peak current responses underestimate the maximum achievable current mediated by 7 nAChRs. For this reason, we use the term apparent peak current to describe the maximum observed peak current amplitude response. In inhibition experiments, which were carried out as part of the window current analysis, a three-step protocol was used. In the first addition, 1 mM ACh without PAM was applied to obtain a control response. In the second addition, different concentrations of ACh were given in the presence of 3 µM PNU-120596, 5 µM TQS, 50 µM genistein, 1 mM 5-HI, or no PAM (buffer control) for 10 min. After this preincubation, 1 mM ACh in the continual presence or absence of PAM was applied for at least 3 s. This protocol allowed for normalization of the concentration-inhibition curves to 1 mM ACh without PAM. The agonist responses obtained in the presence or absence of PAM were also normalized to 1 mM ACh without PAM ensuring that the same control condition was used in comparing the window current effects. In current-voltage experiments aimed at identifying reversal potentials for initial and secondary components, 7 currents were evoked by 100 µM ACh in the presence of 1 µM TQS while changing the holding potential from –140 to +80 mV in steps of 20 mV and normalized to the respective initial and secondary current responses measured at –100 mV taken as –1.0 for each cell.
Calcium Imaging. Functional activities were assessed in human embryonic kidney 293 cell lines expressing human 42 or 34 subunits by measuring intracellular calcium changes using a fluorometric imaging plate reader (FLIPR; Molecular Devices, Sunnyvale, CA). Cells were plated at densities of 25 to 60 x 103 cells/well in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum in 96-well clear-bottomed, black-walled plates precoated with poly(D-lysine) (75 µl/well of 0.01 g/l solution 30 min) and allowed to incubate for 24 to 48 h at 37°C in 5% CO2 in a humidified environment. After aspirating the media, cells were incubated for 45 to 60 min with Fluo-4 AM calcium indicator dye in the dark at room temperature (Invitrogen, Carlsbad, CA) dissolved in N-methyl-D-glucamine (NMDG)/Ringer buffer (140 mM NMDG, 5 mM KCl, 1 mM MgCl2, 10 mM HEPES, and 10 mM CaCl2, pH 7.4). After dye loading, cells were gently washed with the same buffer removing extracellular dye, leaving 100 µl/well after the final wash. Cells were placed in the FLIPR chamber, where 50 µl of 3x stock concentration of test modulators or buffer prepared in the same NMDG/Ringer buffer were added to the wells in the first addition for 5 min. In the second addition, also for 5 min, 50 µl of 4x stock concentrations of nicotine (3–10 µM) or buffer were added.
Data Analysis. In two-electrode voltage-clamp studies, responses were quantified by measuring apparent peak current amplitude. Apparent peak current responses were expressed as percentage response to 100 µM ACh when assessing PAM responses or to 1 mM ACh when determining agonist concentration responses. In inhibition experiments, the concentration-responses to preapplied ACh concentrations in the presence or absence of PAM were plotted against the observed effects on 1 mM ACh without PAM as explained above. In calcium imaging studies, raw fluorescence data were corrected by subtracting fluorescence values from wells with buffer only added. Peak fluorescent responses were determined using FLIPR software and expressed as -fold increases over the submaximum nicotine response (3–10 µM, corresponding to EC30 to EC50); 1-fold indicates no change in the response. Data were analyzed and fitted using Prism (GraphPad Software, San Diego, CA). Sigmoidal doseresponse (variable slope) function was used to fit the replicates. The pEC50 (–log EC50) or pIC50 (–log IC50) values and associated S.E.M. values were obtained from fitted results. The maximum mean ± S.E.M. values were calculated from individual experiments. p <0.05 was considered statistically significant. Student's t test (Microsoft Excel; Microsoft Corp., Redmond, WA) was used to compare data sets.
Modulation of 7 nAChRs by PNU-120596, TQS, and 5-HI. First, the effects of representative compounds from structurally diverse chemotypes, including 5-HI (Zwart et al., 2002), PNU-120596 (Hurst et al., 2005), and TQS (Becker et al., 2006) were assessed on 7 function as agonists and then as PAMs (see Fig. 1 for structures). None of these compounds alone induced activation of 7 currents up to the maximum concentrations tested (30 µM except for 5-HI, which was tested up to 10 mM) indicating that they are not 7 agonists. Under similar conditions, ACh and 7 selective agonists such as PNU-282987 were effective in evoking currents (see Modulation of Agonist Concentration-Responses by PAMs). When 5-HI, PNU-120596, and TQS were added to the cells during preincubation and then 7 currents were obtained by submaximal concentration of ACh (100 µM), concentration-dependent potentiation of current responses was obtained. As shown in Figs. 2 and 3, the rank order of potency, based on apparent peak current amplitude analysis, was PNU-120596 (pEC50 = 5.8 ± 0.09) > TQS (pEC50 = 5.5 ± 0.07) > 5-HI (pEC50 = 3.2 ± 0.06). A similar rank order of potency was obtained when total current charge (integral or area under the current response) was analyzed.
Fig. 1. Diversity of 7 nAChR PAMs. Depicted are structures of PNU-120596, TQS, genistein, and 5-HI.
Fig. 2. Enhancement of ACh-evoked 7 responses by PNU-120596, TQS, 5-HI, and genistein. Representative traces showing the effects of PNU-120596 (a), TQS (b), 5-HI (c), and genistein (d). Concentrations of the modulators are indicated on the right within each panel. The horizontal bars indicate when 100 µM ACh was added in the presence or absence of the specified concentration of PAM. The holding potential was –60 mV.
Fig. 3. Summary of PAM concentration-responses potentiating submaximum ACh evoked 7 currents by PNU-120596, TQS, 5-HI, and genistein. The respective mean pEC50 and maximum efficacy values are 3.2 ± 0.1 and 541 ± 26% for 5-HI, 4.7 ± 0.11 and 267 ± 16% for genistein, 5.5 ± 0.2 and 418 ± 25% for TQS, and 5.8 ± 0.1 and 455 ± 20% for PNU-120596. The n value for each data point is n = 5 to 12.
PAMs had qualitatively different effects on ACh responses, as exemplified by the traces depicted in Fig. 2, and could be classified into two types. PNU-120596 and TQS dramatically increased the apparent peak current response and seemed to reduce the current decay rate (designated as type II). At highest concentrations tested, these compounds in the presence of ACh evoked a nondecaying or weakly decaying current during the recording interval (usually 3 s). Typically, at lower concentrations of PNU-120596 (e.g., 1 µM; Fig. 2a, trace B) and TQS (e.g., 1 µM; Fig. 2b, trace B), the effects on the amplitude were minimal and an apparent secondary component with amplitude similar to that of the initial apparent peak was identifiable. The onset of this secondary component was clearly distinct from that of the initial component. With increasing concentrations of these PAMs, the apparent peak and secondary components overlapped, producing an apparent single current profile with relatively rapid onset and very weak current decay. During washout, when both agonist and PAM were removed, it typically took 50 to 100 s for TQS (10 µM) and longer than 200 to 250 sec for PNU-120596 (10 µM) for the holding current to return to pretreatment levels, suggesting relatively prolonged effects.
In contrast, 5-HI (and genistein see below) predominantly increased 7 nAChR apparent peak amplitude response without robustly affecting current decay rate (designated as type I). Although the decay rate could have been slightly altered, especially at the highest concentrations tested (Fig. 2c, traces C and D); the effects, however, were modest. Furthermore, unlike PNU-120595 and TQS, there was no secondary component identifiable with onset separate from that of the initial apparent peak component. This suggests that the mechanism by which 5-HI allosterically potentiates 7 nAChR response is distinct from that mediated by PNU-120596 and TQS.
To further characterize the nature of the ACh-evoked secondary component, current-voltage experiments were carried out in which the holding potential was varied from –140 up to +80 mV, and ACh evoked 7 currents measured in the presence of 1 µM TQS (see Fig. 4). At this concentration of TQS, both initial and secondary components are easily separable. As shown, both initial and secondary current components reversed at 0 mV consistent with their being mediated directly by 7 nAChR.
Fig. 4. Current-voltage relationship for ACh evoked initial and secondary 7 component responses in the presence of TQS. a, representative current traces obtained by varying the holding potentials from –140 to +60 mV in steps of 20 mV. For each voltage, ACh (100 µM) was applied in the presence of TQS (1 µM). The interval between the traces was at least 3 min. b, the mean current-voltage relationship for the ACh evoked initial and secondary components (n = 2). The responses were normalized to –100 mV for each cell (taken as the normalized current of –1.0 at this voltage) and illustrate that both initial and secondary components reverse at 0 mV consistent with both being mediated directly by 7 nAChR.
Mechanism of 7 nAChR Modulation by Genistein. Genistein is a nonspecific kinase inhibitor (Akiyama et al., 1987) that also increases 7 nAChR current response. This effect has been attributed to inhibition of Src kinase, although a direct mechanism involving positive allosteric modulation may be involved (Charpantier et al., 2005; Cho et al., 2005). To examine whether effects of genistein are due to positive allosteric modulation, three types of experiments were conducted: 1) pre- and coapplication of genistein and ACh to test onset of effects; 2) comparison with other kinase inhibitors; and 3) interaction with 5-HI, another type I PAM.
When added directly, genistein did not activate 7 nAChR current up to the maximum tested concentration of 300 µM (n = 10). In the continued preincubation with genistein, the apparent peak current 7 response was potentiated in a concentration-dependent manner with a pEC50 value of 4.6 ± 0.1 and maximum potentiation of 2.6-fold (Figs. 2d and 3). When preincubation was eliminated and genistein was simply coapplied with ACh, to limit time for potential kinasemediated effects to develop, genistein was still effective in potentiating the 7 nAChR response. The degree of potentiation was 2.2-fold without preincubation (Fig. 5a) compared with 2.6-fold with preincubation (Fig. 3), hence 15% less. We also examined the effect of 5-HI under preapplication and coapplication conditions with ACh and determined that this compound exhibited 35% lesser potentiation when coapplied. The -fold increases were 5.4 (Fig. 3) and 3.5 (Fig. 5a) for pre- and coapplication conditions, respectively.
Fig. 5. Potentiation of 7 nAChR currents by genistein involves direct effects. a, the concentration-responses to 5-HI and genistein added as coapplication without any preincubation. The respective mean pEC50 and maximum efficacy values are 2.8 ± 0.1 and 350 ± 7% for 5-HI and 4.4 ± 0.1 and 227 ± 10% for genistein; each data point is n = 4 to 6. b, representative 7 current traces obtained before and after prolonged treatment with staurosporine for either ACh alone or ACh with genistein treatment. Currents in response to 100 µM ACh in traces i and ii are before and after, respectively, 60-min treatment with 30 nM staurosporine. Currents in response to 100 µM ACh and 100 µM genistein in traces iii and iv are before and after, respectively, 60-min treatment with 30 nM staurosporine. c, the effects of different tyrosine kinase inhibitors on 7 nAChR currents after at least 5-min preincubation. Among these inhibitors, only genistein potentiated the current evoked by 100 µM ACh. Each data point is n 4, * indicates p < 0.05. d, the concentration responses to potentiate 7 currents by 5-HI in the presence of nearly maximal concentration of genistein and by genistein in the presence of nearly maximal concentration of 5-HI. Each data point is n = 3.
Staurosporine and herbimycin A, two nonspecific kinase inhibitors (Yanagihara et al., 1991; Zakar et al., 1999), were also tested to determine their effects on 7 currents. Oocytes were exposed to these two inhibitors for 5- to 10 or 60-min preincubation followed by ACh application. At both time points, staurosporine (up to 30 nM) and herbimycin A (up to 10 µM) failed to increase or inhibit the 7 currents evoked by ACh (n 2) (see Fig. 5c). When genistein was coapplied together with ACh after short- or long-term exposure to either staurosporine (30 nM) or herbimycin A (10 µM), the maximum potentiation of the current was similar to that of genistein alone (see Fig. 5b for example). This observation supports a direct allosteric effect of genistein because this compound was still able to increase 7 currents, even though intracellular kinases were inhibited by the treatment with staurosporine or herbimycin A. In addition, we studied the effects of PP2 and SU6656, two Src tyrosine kinase inhibitors that increased 7 currents in one study (Charpantier et al., 2005) but not in another (Cho et al., 2005). In this study, PP2 and SU6656, similarly to herbimycin A and staurosporine, had no effect on 7 nAChRs (see Fig. 5c). These experiments also support a direct allosteric effect of genistein. We rationalized that if genistein effects on 7 currents were due primarily to inhibition of protein kinases, then staurosporine, herbimycin A, PP2, or SU6656 should mimic the effects of genistein, and they should abolish or attenuate the increased current responses to genistein. As shown, the data support the contrary hypothesis that genistein effects are primarily mediated by a direct allosteric effect on the 7 nAChR.
Finally, the interaction of genistein and 5-HI, both type I PAMs, was evaluated by exposing oocytes to a nearly fully efficacious concentration of either 5-HI or genistein followed by determination of the concentration-dependent effects of the other modulator. As shown in Fig. 5d, the net modulatory effect of 5-HI was attenuated by pretreatment with 50 µM genistein. Likewise, the effect of genistein was occluded by pre-exposure with 3 mM 5-HI. This lack of additivity is consistent with the hypothesis that the 5-HI and genistein act through a similar mechanism, but the results do not exclude the possibility that the nonadditivity was due to a ceiling effect such that either compound alone could exert the maximum possible effect.
Differential Reactivation of Desensitized 7 nAChR by PAMs. To investigate the effects of modulators on desensitized 7 nAChRs, oocytes were first exposed to 100 µM ACh for at least 60 s (and up to 5 min) to desensitize the channels. Subsequently, in the continued presence of ACh, modulators were applied (for a 4-min interval) followed by washout of the modulator and ACh. As exemplified by Fig. 6, the addition of either 100 µM genistein (n = 4) or 3 mM 5-HI (n = 4) caused no change in current responses. However, when 3 µM PNU-120596 (n = 4) or 5 µM TQS (n = 4) was added, there was an increase in the 7 current. This indicates that modulators that affect both apparent peak current response and evoke the secondary component are able to re-activate currents when 7 channels are desensitized. On the other hand, 5-HI and genistein, neither of which evokes the secondary component, do not exhibit this property.
Fig. 6. Preferential activation of desensitized 7 nAChRs by type II PAMs. a and b, the effects of 5-HI and genistein added after ACh treatment indicating their inability to affect desensitized channels. c and d, the effects of TQS and PNU-120596, respectively. Compounds were added during the intervals indicated by the horizontal bars.
Experiments were also done in presence of MLA, an 7 antagonist, or BAPTA-AM, a membrane permeable intracellular Ca2+ chelator. MLA at 100 nM completely abolished the ability of 3 µM PNU-120596 or 5 µM TQS (n 2) to reactivate desensitized 7 nAChR in the presence of ACh (data not shown). In BAPTA-AM experiments, oocytes were incubated with 100 µM BAPTA-AM for at least 3 h, allowing for sufficient chelation of cytosolic Ca2+. Both 3 µM PNU-120596 and 5 µM TQS (n 2) were able to reactivate 7 channels without obvious differences in responses, whether treated with BAPTA-AM or not (data not shown). These experiments indicate that the reactivated current by type II PAMs in oocytes is indeed mediated by 7 nAChRs and unaffected by chelation of intracellular Ca2+ and related Ca2+ dependent currents such as those mediated Ca2+ dependent Cl– channels.
Modulation of Agonist Concentration-Responses by PAMs. Next, we evaluated the effects of 7 modulators on current responses to different 7 agonists. Modulators were preincubated at a fixed concentration (corresponding to EC70–80 in modulator concentration-response experiments: 3 µM PNU-120596, 5 µM TQS, 50 µM genistein, and 1 mM 5-HI) followed by determination of agonist concentration-responses in the continued presence of the modulator. In particular, we aimed to compare the effects of the different modulators on concentration responses to full agonists, ACh and PNU-282987 (Bodnar et al., 2005), and a partial 7 agonist, GTS-21 (de Fiebre et al., 1995). As summarized in Table 1 and Figs. 7 and 8, the modulators affected the concentration-responses to all three agonists by shifting the potencies and increasing the maximum responses. The highest enhancement in efficacy was observed with GTS-21. In the absence of any PAM, GTS-21 behaved as a weak partial agonist (pEC50 < 3, max = 26.5% at 300 µM). However, in the presence of any of the four modulators, GTS-21 became more efficacious, with maximum responses in the range of 70 to 170%. Table 1 also indicates that 5-HI and genistein affected the agonist potencies to a lesser extent than did PNU-120596 and TQS, differences ranging from 0.2 to 0.6-log units. For example, for ACh (pEC50 of 3.9 without any modulator), 5-HI and genistein shifted the potency by 0.3 to 0.4 log units, whereas PNU-120596 and TQS shifted by 0.8 to 0.9 log units. This suggests that modulators affecting both apparent peak current and secondary component generation are more likely to shift the concentration-response profile to 7 agonists to a greater extent than would modulators altering only the apparent peak current response.
TABLE 1 Potencies and efficacies of ACh, GTS-21, and PNU-282987 on 7 currents in the absence or presence of test concentrations of PAMs
Fig. 7. Enhancement of 7 agonist responses by PNU-120596. a, representative traces in X. laevis oocytes expressing 7 evoked by ACh (1 mM, normalizing control) or in the presence of PNU-120596 (3 µM) for PNU-282987 (0.1, 1, and 10 µM) added as indicated by the horizontal bars. b, c, and d, the concentration responses to ACh, GTS-21, and PNU-282987, respectively, in the presence or absence of PNU-120596 (3 µM). Each data point is n = 3 to 6. Summary of potency and maximum efficacy is given in Table 1.
Fig. 8. Enhancement of 7 responses by genistein. (a) shows representative traces in X. laevis oocytes expressing 7 evoked by ACh (1 mM, normalizing control) or in the presence of genistein (50 µM) for PNU-282987 (0.1, 1, and 10 µM) added as indicated by the horizontal bars. b, c, and d, the concentration responses to ACh, GTS-21, and PNU-282987, respectively, in the presence or absence of genistein (50 µM). Each data point is n = 3 to 6. Summary of potency and maximum efficacy is summarized in Table 1.
Effect of PAMs on ACh Window Current. It is well established that 7 nAChRs are activated and desensitized by agonists. In fact, the constants for half-maximum inhibition or desensitization (pIC50) are 1 or 2 orders of magnitude higher that those for activation (pEC50), resulting in a very minimal window current (i.e., the overlap between the activation and inactivation) (Briggs and McKenna, 1998). The effects of modulators on the ACh window current were therefore examined. The activation curves to ACh, discussed above and summarized in Fig. 9 and Table 1, showed differential abilities of PAMs to shift the potencies and efficacies. In contrast, the PAMs did not seem to have any significant effect on the ACh concentration-inhibition curves (see Fig. 9). As shown, the pIC50 for the inactivation curve of ACh was 5.1 ± 0.05 (n = 4). In the presence of TQS, PNU-120596, genistein, and 5-HI, the pIC50 values were 4.9 ± 0.03 (n = 3), 4.8 ± 0.04 (n = 4), 4.7 ± 0.1 (n = 3), and 4.8 ± 0.04 (n = 3), respectively. As a measure of the window currents, we have calculated the integral of the overlapping area of the inhibition and activation curves for the different modulators tested. For ACh, the normalized ratios in the absence of any modulator and in the presence of genistein, 5-HI, TQS, and PNU-120596 were 1 (control, ACh alone), 6, 7, 12, and 17, respectively. This analysis indicates that PNU-120596 and TQS, both type II PAMs, produced more robust effects on ACh window currents than 5-HI and genistein, both type I PAMs.
Fig. 9. Effects of 7 PAMs on window currents evoked by ACh. a, ACh activation and inhibition concentration response graphs without PAM. b and c, ACh activation and inhibition concentration graphs, respectively, in the presence of genistein (50 µM) or PNU-120596 (3 µM). Each data point is n = 2 to 6.
Effects of PAMs on 42 and 34 nAChRs. To investigate the selectivity of modulation, the effects of 5-HI, genistein, TQS, and PNU-120596 were measured in recombinant human embryonic kidney 293 nAChR cell lines expressing either human 42 or human 34 using submaximum concentrations of nicotine (3–10 µM) to evoke Ca2+ transients. Genistein, TQS, or PNU-120596 alone did not affect basal Ca2+ in either of the two cell lines. 5-HI up to 1 mM had no effect. However, at 3 and 10 mM, it alone transiently decreased fluorescence in both cell lines followed by a slow recovery in the signal. Overall, submaximum nicotine evoked Ca2+ signals were not increased by the tested PAMs. PNU-120596 produced only small 0.1- to 0.2-fold decreases in the nicotine-evoked Ca2+ signals in the two cell lines tested (Table 2). TQS also produced a small maximum reduction of 0.1-fold on 34 responses and a decrease of 0.7- fold in 42 responses. Likewise, genistein and 5-HI decreased nicotine-evoked signals mediated by the two subunit combinations by 0.4- to 0.8-fold. Hence, modulators at the concentrations showing positive effects on 7 function did not potentiate or increase Ca2+ signals mediated by 34 and 42 nAChRs. However, at comparable or higher concentrations inhibition of nicotine evoked Ca2+ signals at both 42 or 34 subunits were observed except for PNU-120596.
TABLE 2 Selectivity of PAMs at other human nAChR subtypes studied by Ca2+ imaging
This study compares and contrasts the properties of four structurally distinct 7 nAChR modulators—PNU-120596, TQS, 5-HI, and genistein—and demonstrates important distinctions in their pharmacological profiles. We extend earlier observations made with PNU-120596 and 5-HI (Zwart et al., 2002; Hurst et al., 2005) and provide evidence that genistein effects on 7 nAChR are due primarily to a positive allosteric mechanism rather than via inhibition of protein kinases. We also characterize for the first time the properties of TQS as an 7 PAM. All four compounds increased currents evoked by 7 nAChR agonists. Based on their profiles, two types were recognized. Type I PAMs, exemplified by genistein and 5-HI, predominantly affected the apparent peak current response. Type II PAMs, illustrated by PNU-120596 and TQS, increased the apparent peak current amplitude and strongly evoked the secondary component with onset distinguishable from the initial apparent peak component especially at lower concentrations. Both types exhibit differential properties. Type II modulators were able to reactivate desensitized 7 nAChRs, whereas type I did not. The former had also greater effects on the 7 activation concentration-response curves. The ACh inhibition curves for 7 currents were affected similarly by type I and II compounds resulting in greater ACh window current effects by PNU-120596 and TQS rather than genistein and 5-HI. None of the four compounds, at concentrations active on 7, potentiated 42 and 34 nAChRs indicating that positive allosteric effects are selective for the 7 subtype.
Genistein Is a PAM of 7 nAChR. Genistein, a tyrosine kinase inhibitor, has been shown to increase 7 currents expressed in X. laevis oocytes, in rat hippocampus brain slice interneurons, and stably expressed in SH-SY5Y neuroblastoma cells (Charpantier et al., 2005; Cho et al., 2005). We have confirmed this increase of 7 currents. However, the mechanism underlying this effect remains controversial. In one study, evidence was provided for genistein causing rapid up-regulation of 7 receptors at the cell surface membrane (Cho et al., 2005). In contrast, another report identified no changes in cell surface labeling on neurons with 125I--bungarotoxin (Charpantier et al., 2005). Genistein coapplication with agonist was found to be either effective (Charpantier et al., 2005) or ineffective (Cho et al., 2005) in potentiating 7 currents. Effects of genistein have been interpreted to occur via tyrosine dephosphorylation of non-7 nAChR protein(s) rather than direct allosteric effect on 7 nAChR. Effects of PP2, another tyrosine kinase inhibitor, on 7 currents are also inconclusive, with no (Cho et al., 2005) or potentiating (Charpantier et al., 2005) effects reported. In our study, genistein produced effects similar to those of 5-HI under similar testing conditions consistent with a direct allosteric modulation of 7 nAChR on the basis of several lines of evidence. First, genistein was effective when coapplied with ACh, indicating that pre-exposure to genistein was not required. Second, only genistein among the kinase inhibitors studied (staurosporine, herbimycin A, PP2, or SU6656), examined at concentrations showing effective kinase inhibition (Yanagihara et al., 1991; Hanke et al., 1996; Zakar et al., 1999; Blake et al., 2000), increased 7 currents. Third, the pretreatment with other tyrosine kinases inhibitors did not abolish or attenuate the modulatory effect of genistein. Fourth, the modulatory effect of genistein on 7 nAChR was occluded by effective concentrations of 5-HI and vice versa. This indicates that either both compounds bind to the same modulatory binding site or that there are two separate modulatory sites and activation of either is sufficient to allosterically potentiate 7 nAChR to a certain level that cannot be surpassed by activation of the other modulatory site.
Distinct Profiles of nAChR PAMs: Type I and II. The electrophysiological analysis of the effects of PAMs indicates that there are at least two distinct modulator profiles: type I, exemplified by 5-HI and genistein, and type II, exemplified by PNU-120506 and TQS. The primary difference between these two types is in their ability to evoke the secondary component. At high concentrations of type II PAMs, the initial and secondary components overlap, producing an apparent single complex. In the study by Hurst et al. (2005), the effects of PNU-120596 were judged to occur by slowing down the current decay rate. In this study, the concentration-responses to PNU-120596 and TQS show that at lower concentrations, there are two separate identifiable components. An initial component similar in time course to that of 7 agonists evoked in the absence of any PAM, and a secondary nondecaying or weakly decaying current component, which activates with a slower onset. With increasing concentrations of PAMs, the time courses overlap.
The concept of distinct PAM profiles has been postulated earlier although no previous study has compared and demonstrated such differences. For example, 5-HI (Zwart et al., 2002), ivermectin (Krause et al., 1998), galantamine (Samochocki et al., 2003), bovine serum albumin (Conroy et al., 2003), SLURP-1 (Chimienti et al., 2003), (2-amino-5-keto)thiazole compounds (LY2087101, LY1078733, and LY2087133) (Broad et al., 2006), and compound 6 (Ng et al., 2007) have been reported as 7 PAMs exhibiting profile characteristic of type I PAMs. PNU-120596 has been shown to exhibit a different profile (Hurst et al., 2005). This compound increased the apparent peak 7 current response and robustly affected the time course of current response. At the single-channel level, PNU-120596 increased mean open time, had no effect on ion selectivity, and had relatively little effect on unitary conductance (Hurst et al., 2005). PNU-120596 also increased ACh-evoked GABAergic synaptic activity recorded in pyramidal cells (Hurst et al., 2005) similar to effects of 5-HI in interneurons in hippocampus slices (Mok and Kew, 2006). 5-HI also enhanced ACh-stimulated glutamate evoked postsynaptic currents in cerebellar slices (Zwart et al., 2002) illustrating 7 PAM effects on synaptic activity. In this study, we showed that PNU-120596 potentiated the 7 currents expressed in X. laevis oocytes with an EC50 value of 1.6 µM (or pEC50 of 5.8) and maximum poteniation of 4.5- fold, and we demonstrate that TQS is also a type II PAM exhibiting potency and efficacy similar to PNU-120596.
Comparison of Pharmacological Properties of Type I and II PAMs. Our study provides further insight into the pharmacological properties of type I and II PAMs. Compounds belonging to both types were effective in shifting the potencies of agonists to the left and in increasing their maximum efficacies, although type II PAMs were generally more effective (see Table 1). These changes are comparable with those reported by others for 5-HI (Zwart et al., 2002) and PNU-120596 (Hurst et al., 2005). Among the agonists tested, the greatest effect was observed for GTS-21. In the absence of any modulator, this compound was a partial agonist, and in the presence of any one of the four modulators, GTS-21 turned out to be a very efficacious agonist. Analysis of the inhibition concentration-response curves to ACh revealed that both types of modulators affected the inhibition similarly. In this study, very little window current (overlap between inhibition and activation curves) to ACh alone (Fig. 9) was observed, consistent with previous observations (Briggs and McKenna, 1998). To our knowledge, effects of modulators on window currents have not been evaluated for any 7 PAM. Our studies demonstrate that type II PAMs, PNU-120596, and TQS, had in general larger effects on window currents than type I PAMs, genistein, and 5-HI.
In cultured rat hippocampal neurons, PNU-120956 activated desensitized rat 7 currents when studied electrophysiologically using whole-cell, patch-clamp recordings (Hurst et al., 2005). In addition to confirming this observation at human 7 currents, we found that PNU-120596 and TQS produced similar effects evoking current reactivation from desensitized 7 nAChRs in contrast to genistein and 5-HI that did not. The characteristics of this reactivated current (specifically its onset and weakly decaying nature) are similar to those of the secondary component described above (see Fig. 3 and Distinct Profiles of nAChR PAMs: Types I and II) suggesting that the same "activated" channel state is responsible for both. The mechanisms responsible for the induction of this activated state remain to be identified and require further investigation. Potential explanations could be that 7 nAChR modulators stabilize a new "desensitized-open" state, as in the case of the 7V274T mutant (Galzi et al., 1992; Briggs et al., 1999), or promote a shift in the equilibrium from a desensitized state to the "active" open state and stabilizing the receptor in the latter state as suggested for ivermectin (Krause et al., 1998).
Selectivity of 7 PAMs. Targeting PAMs rather than direct agonist could offer a potential advantage in terms of selectivity because PAM binding sites are likely distinct from agonist/competitive antagonist binding sites that show considerable homology among various nAChR and related ligand-gated ion channels of the cys-loop family. Determination of selectivity of PAMs will be important to avoid potential non-7 nAChR interactions. For example, 34* receptors are thought to be involved in the control of bladder and cardiac function and 42* subunits in reinforcing effects of nicotine related to addiction (see review by Dani and Bertrand, 2007). In this study, we tested the effects of PNU-120596, TQS, genistein, and 5-HI on 42 and 34 subunits using Ca2+ flux measurements. None of these compounds evoked increases in the signals mediated by 42 and 34 nAChRs, indicating that they are selective PAMs for 7. Our observations are similar to those reported for PNU-120596 (tested only at 1 µM) in recordings from X. laevis oocytes expressing h42, h34, or h910 nAChR (Hurst et al., 2005) and for genistein (at 10 µM) on ACh evoked currents in X. laevis oocytes expressing 42 and 34 subunits (Cho et al., 2005). In this study, significant inhibition of Ca2+ responses mediated by h42 or h34 subunits were noted for all compounds except PNU-120596 (see Table 2) at concentrations similar or slightly higher than those required for modulation of 7; the significance of which remains to be clarified. In addition to effects of 5-HI on nAChRs, this compound also positively modulates 5-HT3 currents endogenously expressed in NCB-20 cells and N1E-115 neuroblastoma cells (van Hooft et al., 1997), limiting its usefulness as a tool compound.
In summary, this study shows that structurally distinct 7 PAMs can be divided into two types based on their effects on 7 currents. Type I PAMs—5-HI and genistein—predominantly affected the apparent peak current response, whereas type II PAMs—PNU-120596 and TQS—increased apparent peak current response and strongly evoked a secondary weakly decaying current. In general, type II but not type I PAMs could reactivate desensitized 7 currents and had greater effects shifting 7 agonist concentration-responses and on ACh window currents. The identification of distinct functional profiles of 7 PAMs and the reported demonstration of PAM efficacy in preclinical in vivo models of cognition provide basis for the development of novel therapeutics for CNS indications for which the 7 nAChR is considered a viable target.
ABBREVIATIONS: nAChR, nicotinic acetylcholine receptor; CNS, central nervous system; PNU-282987, N-[(3R)-1-azabicyclo[2.2.2]oct-3-yl]-4-chlorobenzamide hydrochloride; PAM, positive allosteric modulator; PNU-120596, 1-(5-chloro-2,4-dimethoxy-phenyl)-3-(5-methyl-isoxazol-3-yl)-urea; 5-HI, 5-hydroxyindole; TQS, 4-naphthalen-1-yl-3a,4,5,9b-tetrahydro-3H-cyclopenta[c]quinoline-8-sulfonic acid amide; compound 6, N-(4-chlorophenyl)--[[(4-chloro-phenyl)amino]methylene]-3-methyl-5-isoxazoleacet-amide; MLA, methyllycaconitine; PP2, protein phosphatase 2; SU6656, 2-oxo-3-(4,5,6,7-tetrahydro-1H-indol-2-ylmethylene)-2,3-dihydro-1H-indole-5-sulfonic acid dimethylamide; FLIPR, fluorometric imaging plate reader; NMDG, N-methyl-D-glucamine; BAPTA-AM, 1,2-bis(2-aminophenoxy)ethane-N,N,N',N'-tetraacetic acid, acetoxymethyl ester; ARR17779, (5S)-spiro[1,3-oxazolidine-5,8'-1-azabicyclo[2.2.2]octane]-2-one.
【参考文献】
Akiyama T, Ishida J, Nakagawa S, Ogawara H, Watanabe S, Itoh N, Shibuya M, and Fukami Y (1987) Genistein, a specific inhibitor of tyrosine-specific protein kinases. J Biol Chem 262: 5592–5595.[Abstract/Free Full Text]
Alkondon M, Braga MF, Pereira EF, Maelicke A, and Albuquerque EX (2000) Alpha7 nicotinic acetylcholine receptors and modulation of gabaergic synaptic transmission in the hippocampus. Eur J Pharmacol 393: 59–67.[CrossRef][Medline]
Becker C, Comstock J, Michne WF, Murphy M, Philips E, Rosamond JD, and Simpson TR (2006), inventors; AstraZeneca, assignee. Positive modulators of nicotinic acetylcholine receptors. World Patent no. W02004098600(A1). 2006 Feb 1.
Blake RA, Broome MA, Liu X, Wu J, Gishizky M, Sun L, and Courtneidge SA (2000) SU6656, a selective src family kinase inhibitor, used to probe growth factor signaling. Mol Cell Biol 20: 9018–9027.[Abstract/Free Full Text]
Bodnar AL, Cortes-Burgos LA, Cook KK, Dinh DM, Groppi VE, Hajos M, Higdon NR, Hoffmann WE, Hurst RS, Myers JK, et al. (2005) Discovery and structure-activity relationship of quinuclidine benzamides as agonists of alpha7 nicotinic acetylcholine receptors. J Med Chem 48: 905–908.[CrossRef][Medline]
Breese CR, Adams C, Logel J, Drebing C, Rollins Y, Barnhart M, Sullivan B, Demasters BK, Freedman R, and Leonard S (1997) Comparison of the regional expression of nicotinic acetylcholine receptor alpha7 mRNA and [125I]-alpha-bungarotoxin binding in human postmortem brain. J Comp Neurol 387: 385–398.[CrossRef][Medline]
Briggs CA and McKenna DG (1998) Activation and inhibition of the human alpha7 nicotinic acetylcholine receptor by agonists. Neuropharmacology 37: 1095–1102.[CrossRef][Medline]
Briggs CA, McKenna DG, Monteggia LM, Touma E, Roch JM, Arneric SP, Gopalakrishnan M, and Sullivan JP (1999) Gain of function mutation of the alpha7 nicotinic receptor: distinct pharmacology of the human alpha7V274T variant. Eur J Pharmacol 366: 301–308.[CrossRef][Medline]
Briggs CA, McKenna DG, and Piattoni-Kaplan M (1995) Human alpha7 nicotinic acetylcholine receptor responses to novel ligands. Neuropharmacology 34: 583–590.[CrossRef][Medline]
Broad LM, Zwart R, Pearson K, Lee M, Wallace L, Mc PG, Emkey R, Hollinshead S, Dell C, Baker R, et al. (2006) Identification and pharmacological profile of a new class of selective nicotinic acetylcholine receptor potentiators. J Pharmacol Exp Ther 318: 1108–1117.[Abstract/Free Full Text]
Charpantier E, Wiesner A, Huh KH, Ogier R, Hoda JC, Allaman G, Raggenbass M, Feuerbach D, Bertrand D, and Fuhrer C (2005) Alpha7 neuronal nicotinic acetylcholine receptors are negatively regulated by tyrosine phosphorylation and Src family kinases. J Neurosci 25: 9836–9849.[Abstract/Free Full Text]
Chimienti F, Hogg RC, Plantard L, Lehmann C, Brakch N, Fischer J, Huber M, Bertrand D, and Hohl D (2003) Identification of SLURP-1 as an epidermal neuromodulator explains the clinical phenotype of Mal de Meleda. Hum Mol Genet 12: 3017–3024.[Abstract/Free Full Text]
Cho CH, Song W, Leitzell K, Teo E, Meleth AD, Quick MW, and Lester RA (2005) Rapid upregulation of alpha7 nicotinic acetylcholine receptors by tyrosine dephosphorylation. J Neurosci 25: 3712–3723.[Abstract/Free Full Text]
Conroy WG, Liu QS, Nai Q, Margiotta JF, and Berg DK (2003) Potentiation of 7-containing nicotinic acetylcholine receptors by select albumins. Mol Pharmacol 63: 419–428.[Abstract/Free Full Text]
Couturier S, Bertrand D, Matter JM, Hernandez MC, Bertrand S, Millar N, Valera S, Barkas T, and Ballivet M (1990) A neuronal nicotinic acetylcholine receptor subunit (alpha 7) is developmentally regulated and forms a homo-oligomeric channel blocked by alpha-BTX. Neuron 5: 847–856.[CrossRef][Medline]
Curzon P, Anderson DJ, Nikkel AL, Fox GB, Gopalakrishnan M, Decker MW, and Bitner RS (2006) Antisense knockdown of the rat alpha7 nicotinic acetylcholine receptor produces spatial memory impairment. Neurosci Lett 410: 15–19.[CrossRef][Medline]
Dajas-Bailador F, and Wonnacott S (2004) Nicotinic acetylcholine receptors and the regulation of neuronal signalling. Trends Pharmacol Sci 25: 317–324.[CrossRef][Medline]
Dani JA and Bertrand D (2007) Nicotinic acetylcholine receptors and nicotinic cholinergic mechanisms of the central nervous system. Annu Rev Pharmacol Toxicol 47: 699–729.[CrossRef][Medline]
de Fiebre CM, Meyer EM, Henry JC, Muraskin SI, Kem WR, and Papke RL (1995) Characterization of a series of anabaseine-derived compounds reveals that the 3-(4)-dimethylaminocinnamylidine derivative is a selective agonist at neuronal nicotinic 7/[125I]--bungarotoxin receptor subtypes. Mol Pharmacol 47: 164–171.[Abstract]
Dineley KT, Hogan D, Hernandez CM, Song W, Zheng H, and Sweatt JD (2005) Genetic deletion of the 7 nicotinic acetylcholine receptor decreases hippocampal A accumulation but exacerbates hippocampal learning and memory deficits in APPSWE mice. Soc Neurosci Abstr 33: 586.7
Felix R and Levin ED (1997) Nicotinic antagonist administration into the ventral hippocampus and spatial working memory in rats. Neuroscience 81: 1009–1017.[CrossRef][Medline]
Frazier CJ, Rollins YD, Breese CR, Leonard S, Freedman R, and Dunwiddie TV (1998) Acetylcholine activates an alpha-bungarotoxin-sensitive nicotinic current in rat hippocampal interneurons, but not pyramidal cells. J Neurosci 18: 1187–1195.[Abstract/Free Full Text]
Galzi JL, Devillers-Thiery A, Hussy N, Bertrand S, Changeux JP, and Bertrand D (1992) Mutations in the channel domain of a neuronal nicotinic receptor convert ion selectivity from cationic to anionic. Nature 359: 500–505.[CrossRef][Medline]
Gotti C, Riganti L, Vailati S, and Clementi F (2006) Brain neuronal nicotinic receptors as new targets for drug discovery. Curr Pharm Des 12: 407–428.[CrossRef][Medline]
Hajós M, Hurst RS, Hoffmann WE, Krause M, Wall TM, Higdon NR, and Groppi VE (2005) The selective alpha7 nicotinic acetylcholine receptor agonist PNU-282987 [N-[(3R)-1-azabicyclo[2.2.2]oct-3-yl]-4-chlorobenzamide hydrochloride] enhances GABAergic synaptic activity in brain slices and restores auditory gating deficits in anesthetized rats. J Pharmacol Exp Ther 312: 1213–1222.[Abstract/Free Full Text]
Hanke JH, Gardner JP, Dow RL, Changelian PS, Brissette WH, Weringer EJ, Pollok BA, and Connelly PA (1996) Discovery of a novel, potent, and Src family-selective tyrosine kinase inhibitor. Study of Lck- and FynT-dependent T cell activation. J Biol Chem 271: 695–701.[Abstract/Free Full Text]
Hevers W and Luddens H (1998) The diversity of GABAA receptors. Pharmacological and electrophysiological properties of GABAA channel subtypes. Mol Neurobiol 18: 35–86.[Medline]
Hurst RS, Hajos M, Raggenbass M, Wall TM, Higdon NR, Lawson JA, Rutherford-Root KL, Berkenpas MB, Hoffmann WE, Piotrowski DW, et al. (2005) A novel positive allosteric modulator of the alpha7 neuronal nicotinic acetylcholine receptor: in vitro and in vivo characterization. J Neurosci 25: 4396–4405.[Abstract/Free Full Text]
Keller JJ, Keller AB, Bowers BJ, and Wehner JM (2005) Performance of alpha7 nicotinic receptor null mutants is impaired in appetitive learning measured in a signaled nose poke task. Behav Brain Res 162: 143–152.[CrossRef][Medline]
Krause RM, Buisson B, Bertrand S, Corringer PJ, Galzi JL, Changeux JP, and Bertrand D (1998) Ivermectin: a positive allosteric effector of the 7 neuronal nicotinic acetylcholine receptor. Mol Pharmacol 53: 283–294.[Abstract/Free Full Text]
Levin ED and Rezvani AH (2002) Nicotinic treatment for cognitive dysfunction. Curr Drug Targets CNS Neurol Disord 1: 423–431.[CrossRef][Medline]
Mok MHS and Kew JN (2006) Excitation of rat hippocampal interneurons via modulation of endogenous agonist activity at the alpha7 nicotinic ACh receptor. J Physiol 574: 699–710.[Abstract/Free Full Text]
Ng HJ, Whittemore ER, Tran MB, Hogenkamp DJ, Broide RS, Johnstone TB, Zheng L, Stevens KE, and Gee KW (2007) Nootropic alpha7 nicotinic receptor allosteric modulator derived from GABAA receptor modulators. Proc Natl Acad Sci U S A 104: 8059–8064.[Abstract/Free Full Text]
Paterson D and Nordberg A (2000) Neuronal nicotinic receptors in the human brain. Prog Neurobiol 61: 75–111.[CrossRef][Medline]
Rubboli F, Court JA, Sala C, Morris C, Chini B, Perry E, and Clementi F (1994) Distribution of nicotinic receptors in the human hippocampus and thalamus. Eur J Neurosci 6: 1596–1604.[CrossRef][Medline]
Samochocki M, Hoffle A, Fehrenbacher A, Jostock R, Ludwig J, Christner C, Radina M, Zerlin M, Ullmer C, Pereira EF, et al. (2003) Galantamine is an allosterically potentiating ligand of neuronal nicotinic but not of muscarinic acetylcholine receptors. J Pharmacol Exp Ther 305: 1024–1036.[Abstract/Free Full Text]
Trumbull JD, Maslana ES, McKenna DG, Nemcek TA, Niforatos W, Pan JY, Parihar AS, Shieh CC, Wilkins JA, Briggs CA, et al. (2003) High throughput electrophysiology using a fully automated, multiplexed recording system. Receptors Channels 9: 19–28.[CrossRef][Medline]
van Hooft JA, van der HE, and Vijverberg HP (1997) Allosteric potentiation of the 5-HT3 receptor-mediated ion current in N1E-115 neuroblastoma cells by 5-hydroxyindole and analogues. Neuropharmacology 36: 649–653.[CrossRef][Medline]
Van Kampen M, Selbach K, Schneider R, Schiegel E, Boess F, and Schreiber R (2004) AR-R 17779 improves social recognition in rats by activation of nicotinic alpha7 receptors. Psychopharmacology (Berl) 172: 375–383.[CrossRef][Medline]
Wehner JM, Keller JJ, Keller AB, Picciotto MR, Paylor R, Booker TK, Beaudet A, Heinemann SF, and Balogh SA (2004) Role of neuronal nicotinic receptors in the effects of nicotine and ethanol on contextual fear conditioning. Neuroscience 129: 11–24.[CrossRef][Medline]
Wevers A, Jeske A, Lobron C, Birtsch C, Heinemann S, Maelicke A, Schroder R, and Schroder H (1994) Cellular distribution of nicotinic acetylcholine receptor subunit mRNAs in the human cerebral cortex as revealed by non-isotopic in situ hybridization. Brain Res Mol Brain Res 25: 122–128.[Medline]
Wishka DG, Walker DP, Yates KM, Reitz SC, Jia S, Myers JK, Olson KL, Jacobsen EJ, Wolfe ML, Groppi VE, et al. (2006) Discovery of N-[(3R)-1-Azabicyclo[2.2.2]oct-3-yl]furo[2,3-c]pyridine-5-carboxamide, an agonist of the alpha7 nicotinic acetylcholine receptor, for the potential treatment of cognitive deficits in schizophrenia: synthesis and structure-activity relationship. J Med Chem 49: 4425–4436.[CrossRef][Medline]
Yanagihara N, Tachikawa E, Izumi F, Yasugawa S, Yamamoto H, and Miyamoto E (1991) Staurosporine: an effective inhibitor for Ca2+/calmodulin-dependent protein kinase II. J Neurochem 56: 294–298.[Medline]
Zakar T, Mijovic JE, Bhardwaj D, and Olson DM (1999) Tyrosine kinase inhibitors block the glucocorticoid stimulation of prostaglandin endoperoxide H synthase expression in amnion cells. Can J Physiol Pharmacol 77: 138–142.[CrossRef][Medline]
Zbarsky V, Thomas J, and Greenfield S (2004) Bioactivity of a peptide derived from acetylcholinesterase: involvement of an ivermectin-sensitive site on the alpha7 nicotinic receptor. Neurobiol Dis 16: 283–289.[CrossRef][Medline]
Zwart R, De Filippi G, Broad LM, McPhie GI, Pearson KH, Baldwinson T, and Sher E (2002) 5-Hydroxyindole potentiates human alpha 7 nicotinic receptor-mediated responses and enhances acetylcholine-induced glutamate release in cerebellar slices. Neuropharmacology 43: 374–384.[CrossRef][Medline]
作者单位：Neuroscience Research, Global Pharmaceutical Research and Development, Abbott, Abbott Park, Illinois (C.A.B., M.G., J.M.) and Oslo Research Park, Oslo, Norway (J.H.G., M.H., H.W., K.T.H.)
【关键词】 Efficacy
Drug efficacy is typically considered an intrinsic property of a ligand/receptor couple. However, recent observations suggest that efficacy may also be influenced by the signaling effectors engaged by a unique receptor. To directly and systematically test this possibility, we assessed the ability of a panel of -adrenergic ligands to modulate the activity of two effector systems, the adenylyl cyclase (AC) and the mitogen-activated protein kinase (MAPK), via 1 and 2 adrenergic receptors. Although some compounds displayed similar efficacies toward the two pathways, others showed complex efficacy profiles. For example, compounds that are inverse agonists for the AC activity were found to be either agonists, neutral antagonists, or inverse agonists for the MAPK pathway. Likewise, agonists for the AC were either agonists or neutral antagonists for MAPK. Given this complexity, we propose a Cartesian representation of the efficacies that takes into account the activities of the different effectors that can be engaged by a given receptor. In addition, compounds considered as nonselective for 1 and 2 adrenergic receptors, based on their binding affinities, showed distinct relative efficacy profiles toward AC and MAPK, adding a new dimension to the concept of ligand selectivity. Taken together, the results suggest that binding of different ligands promote distinct conformational changes leading to specific signaling outcomes. Our data therefore clearly illustrate that efficacy is a pluridimensional parameter that is not an intrinsic characteristic of a ligand/receptor couple. This should have important implications for the future design of screening assays used in drug discovery campaigns.G protein-coupled receptors (GPCRs) represent the largest class of proteins involved in signal transduction across biological membranes. As such, they are among the main molecular targets considered for the development of therapeutic agents. Drugs acting on GPCR have traditionally been classified into two main categories: agonists and antagonists, which promote or block receptor activation, respectively. In the last 10 years, however, the recognition that GPCR can exhibit constitutive activity led to the discovery of a third class of compounds that can decrease such spontaneous activity and are known as inverse agonists. In the framework of an allosteric model whereby receptors are in equilibrium between inactive (R) and active (R*) conformations, agonists and inverse agonists are believed to stabilize R* and R, respectively. Neutral competitive antagonists, for their part, presumably compete for the binding of agonists or inverse agonists but do not affect the equilibrium and thus have no intrinsic activity (for review, see Bond, 1997; Strange, 2002; Milligan, 2003).
It is noteworthy that the extended and cubic ternary complex models, which were developed to formalize ligand behaviors, included terms that qualified the affinity of the activated receptor for the G protein, opening the possibility that various ligands may stabilize different active conformations resulting in distinct signaling properties (for review, see Kenakin 2004). Consistent with this theoretical possibility, some studies reported that the order of potency/efficacy of compounds acting through a unique receptor can be different depending of the effector system considered (Spengler et al., 1993; Kenakin, 1995; Berg et al., 1998; Hall et al., 1999, Kurrasch-Orbaugh et al., 2003; Gay et al., 2004; Moniri et al., 2004; Krueger et al., 2005; McLaughlin et al., 2005). This phenomenon, often referred to as "ligand-directed stimulus trafficking", "functional selectivity", or "biased agonism", has been taken as evidence that more than one active receptor conformation exists (Kenakin, 2002; Urban et al., 2006). Particularly striking in this respect are recent studies reporting that ligands can have opposite efficacies toward two different signaling pathways. For example, ICI118,551 and propranolol, which act as inverse agonists on the 2-adrenergic receptor (2AR) toward the adenylyl cyclase (AC) signaling pathway, were shown to be partial agonists when tested on the extracellular signal-regulated kinase (ERK) activity (Azzi et al., 2003; Baker et al., 2003). Similar dual efficacies for distinct signaling pathways were also reported for ligands acting on the H3-histamine receptor (Gbahou et al., 2003), the -opioid receptor (Audet et al., 2005), and the serotonin 5-HT2C receptor (Werry et al., 2005).
Taken together, these observations suggest that efficacy is a more complex parameter than was originally anticipated and that the effector systems may need to be included in its description. When considering two distinct signaling pathways modulated by a single receptor, multiple efficacy combinations are theoretically possible. Compounds could be agonist for the two pathways, inverse agonist for the two pathways, or have opposite efficacies on each of the pathways. The present study, therefore, was initiated to test whether these different theoretical efficacy profiles can exist. For this purpose, the ability of various -adrenergic ligands to modulate the activity of the adenylyl cyclase and ERK1/2 was assessed in cells expressing either the human 1 adrenergic receptor (1AR) or 2AR. In the case of each receptor subtype, ligands that activate both, inhibit both, or have opposite effects on each of the two pathways were indeed identified. The wide diversity of efficacy and potency profiles revealed by our study clearly illustrates the notion of signaling pluridimensionality that complicates the classification of ligands according to unique efficacy terms. The systematic comparison of the efficacy profiles for the two closely related receptor subtypes also demonstrated that drug efficacy can only be considered in the context of the diverse signaling pathways that can be engaged by a specific receptor subtype.
In addition to shedding light on the concept of drug efficacy, these results could have important clinical implications because the distinct efficacy profiles of -blockers, which are widely used in the treatment of hypertension and heart failure, could underlie differences in therapeutic and side effect patterns. Additional studies linking defined physiological actions of various compounds to their efficacies toward specific effector systems should allow to explore the potential therapeutic impact of pluridimensional signaling efficacy.
Reagents. (-)-Isoproterenol, labetalol, DL-propranolol, S(-)-atenolol, and (±)-metoprolol were purchased from Sigma-Aldrich (St Louis, MO), and (±)-bisoprolol was from Tocris Cookson Inc. (Ellisville, MO). Carvedilol and bucindolol were a generous gift from GlaxoSmithKline (Research Triangle Park, NC) and Dr. Michael Bristow (University of Colorado Health Sciences Center, Denver, CO), respectively. Mouse anti-phosphorylated ERK1/2 and rabbit anti-ERK1/2 antibodies were from Santa Cruz Biotechnology (Santa Cruz, CA). Horseradish peroxidase (HRP)-anti-mouse and HRP-anti-rabbit polyclonal antibodies were from GE Healthcare (Baie d'Urfé, QC, Canada). All other reagents were of analytical grade and obtained from various suppliers.
Stable Cell Lines and Cell Culture. Stable cell lines expressing human-1AR and human-2AR were generated by transfection of pcDNA3.1-HA-1AR or pcDNA3.1-HA-2AR plasmids (Lavoie et al., 2002) into human embryonic kidney (HEK) 293S cells (Reeves et al., 1996) using the calcium-phosphate precipitation method. Isolates that stably incorporated the plasmids were selected on the basis of their resistance to G418 by treating cells with G418 (400 µg/ml). Receptor level expression was deduced from binding experiments carried out on whole cells using 125I-cyanopindolol as radioligand. The two cell lines used expressed 7 to 8 pmol and 4 to 5 pmol receptor/mg of protein for the 1AR and 2AR, respectively.
Cells were routinely grown in Dulbecco's modified Eagle's medium supplemented with 5% fetal bovine serum, 100 U/ml penicillin and streptomycin, 2 mM L-glutamine, and 400 µg/ml G418 in a 37°C humidified 5% CO2 atmosphere.
Quantification of cAMP Accumulation. Cells at 80% confluence were serum-starved for 16 h. The day of the experiment, cells were resuspended in phosphate-buffered saline (PBS)/0.1% glucose/1 mM 3-isobutyl-1-methylxanthine and then treated for 30 min at 37°C with the indicated drugs. Compounds behaving as inverse agonists were tested in the presence of 0.3 µM forskolin to increase the window of inhibition. After drug treatment, cells were immediately lysed, and cAMP levels were measured using the CatchPoint cAMP Kit (Molecular Devices, Sunnyvale, CA) according to manufacturer's recommendations. In brief, cell lysates were incubated in 384-well plates coated with anti-cAMP antibodies in the presence of known amounts of HRP-cAMP. cAMP from cell lysates was allowed to compete with the HRP-cAMP for 2 h, and the remaining peroxidase activity was measured after 3 washes. The cAMP generated under the different conditions was interpolated from a cAMP standard curve generated in parallel for each experiment. Triplicates were used for each condition, and all experiments were repeated at least three times. For the determination of the EC50, data are expressed as a percentage of the maximal response reached for each compound. For the determination of the relative activities, maximal agonist responses (Emax) are expressed as the percentage of maximal (10 µM) isoproterenol-promoted stimulation. For the compounds behaving as inverse agonists, data were expressed in percentage of forskolin (0.3 µM) inhibition.
Detection of Phosphorylated ERK1/2. Cells expressing 1AR or 2AR were seeded in poly-D-lysine-coated six-well plates. The day after, cells were washed once with PBS and incubated with serumfree media for 16 hours. Cells at 80% confluence were then stimulated for the indicated times at 37°C. The media were then rapidly removed and cells were washed with ice-cold PBS before being lysed using 100 µl per well of Laemmli sample buffer (62.5 mM Tris-HCl, 2% SDS, 10% glycerol, 50 mM dithiothreitol, and 0.1% bromphenol blue, pH 6.8). Whole-cell lysates were sonicated, resolved by SDS-PAGE, and transferred onto nitrocellulose. The blots were blocked at room temperature for 1 h with TBS-T buffer [50 mM Tris, pH 7.4, 150 mM NaCl, and 0.1% (v/v) Tween 20], 5% fat-free milk. Phospho-ERK1/2 was detected using mouse polyclonal anti-phospho p42/p44 ERK-specific antibody (1:3000, overnight at 4°C in TBS-T, 5% fatfree milk). The immunoreactivity was revealed using a secondary HRP-conjugated anti-mouse antibody (1:10,000, 1 h at room temperature in TBS-T, 5% fat-free milk) and the peroxidase activity detected by chemiluminescence (PerkinElmer Life and Analytical Sciences, Boston, MA). Blots were stripped and re-probed for total ERK1/2 with rabbit polyclonal anti-ERK1/2 antibody (1:25,000, 1 h at room temperature in TBS-T, 5% fat-free milk) followed by HRP-anti-rabbit antibody (1:20,000, 1 h at room temperature in TBS-T, 5% fat-free milk). Films were scanned, and band intensities were quantified using Quantity One (Bio-Rad Laboratories, Hercules, CA) software. ERK1/2 phosphorylation was normalized according to the loading of proteins by expressing the data as a ratio of P-ERK1/2 over total ERK1/2.
Statistical Analysis. Statistical analysis and curve fitting were done using Prism 2.01 (GraphPad Software, San Diego, CA). Statistical significance of the differences was assessed using one-way analysis of variance (ANOVA) and post hoc Bonferonni or Dunnett's test.
The efficacy and potency profiles of eight -adrenergic ligands were tested on both AC and ERK signaling pathways in cells expressing either the human 1- or 2-adrenergic receptors (ARs). The compounds were selected based on their prevalent use in pharmacological studies and clinical settings. Their reported affinities for both 1 and 2AR are listed in Table 1. Five of the compounds bind with similar affinities to both receptors, whereas three display a strong selectivity toward 1AR.
TABLE 1 Compounds tested and their relative binding affinities for the two -adrenergic subtypes
The data from Hoffmann et al. (2004) and Ponicke et al. (2002) are from 125I-cyanopindolol binding, and those from Pauwels et al. (1988) are from [3H]CGP-12177 binding.
Ligand Profiles on the 1 Adrenergic Receptor. The eight selected compounds were first tested for their ability to modulate cAMP production in HEK293S cells stably expressing the human-1AR. In a first round, their activity was assayed to establish whether they behave as agonists, neutral antagonists, or inverse agonists (data not shown). In the case of antagonists and inverse agonists, their potency and efficacy were tested in the presence of 0.3 µM forskolin to enhance the response window. As shown in Fig. 1, isoproterenol, labetalol, bucindolol, and carvedilol behaved as agonists, whereas propranolol, metoprolol, bisoprolol, and atenolol were inverse agonists for the AC pathway. Using maximal concentration (at least 10x the EC50 for each compound determined from dose response experiments; Fig. 1A and Table 2, left column), the efficacy of each compound was determined. When considering the agonists, isoproterenol was the most efficacious compound, whereas labetalol, bucindolol, and carvedilol had equivalent partial efficacy, corresponding to 33 to 44% of the maximal isoproterenol-stimulated response (Fig. 1B, top, and Table 3, left column). For inverse agonists, metoprolol, bisoprolol, and atenolol had similar high inverse efficacies. Propranolol behaved as a partial inverse agonist, leading to an inhibition of only 32% of the forskolin-stimulated AC compared with the 63 to 72% inhibition observed for the other three compounds (Fig. 1B, bottom).
Fig. 1. Profile of the different ligands on the 1AR adenylate cyclase pathway. cAMP accumulation experiments were performed in HEK293S cells stably expressing 1AR. A, EC50 of compounds were obtained using increasing concentrations of ligands in the presence (bottom) or absence (top) of 0.3 µM forskolin. B, Emax of compounds were tested using maximal concentrations of ligands, determined from EC50 curves, with (bottom) or without (top) 0.3 µM forskolin. Data represent the mean ± S.E.M. of at least three experiments performed in triplicates. Iso, isoproterenol; Lab, labetalol; Buc, bucindolol; Carv, carvedilol; Prop, propranolol; Met, metoprolol; Bis, bisoprolol; At, atenolol.
TABLE 2 EC50 on cAMP and on MAPK of compounds tested on 1AR
Experimental data from Figs. 1A and 2B presented as mean ± S.E.M.
TABLE 3 Emax on cAMP and on MAPK of compounds tested on 1AR.
Experimental data from Figs. 1B and 2A, presented as maximal response ± S.E.M. Data are expressed as a percentage of the isoproterenol-stimulated response unless otherwise specified.
Fig. 2. Profile of the different ligands on the 1AR MAPK ERK1/2 pathway. ERK1/2 phosphorylation was assessed in HEK293S cells stably expressing 1AR. Compounds were tested at their appropriate maximal stimulation time (4 min for isoproterenol and 2 min for other compounds). A, Emax of compounds were obtained using maximal ligand concentration (10-4 M for isoproterenol and 10-5 M for other ligands). B, EC50 of compounds were obtained using increasing concentrations of ligands. Data represent the mean ± S.E.M. of at least four experiments. Statistical significance was determined by two-way ANOVA, followed by a Dunnett test with basal as control column. ***, p < 0.001.
The same compounds were then assessed for their ability to modulate the ERK1/2 activity in the 1AR-expressing cells. Because ERK1/2 activation is known to be fast and transient, we first performed time course experiments to determine the time of maximal activation for each compound. These times of activation (between 2 and 4 min depending on the compound) were then selected to assess relative efficacies. As shown in Fig. 2, the efficacy pattern found in the ERK1/2 assay was significantly different from that observed for AC activation. When considering the compounds that were agonists for the AC pathway, isoproterenol, bucindolol, and carvedilol also acted as agonists and partial agonists for ERK1/2; bucindolol and carvedilol showed 25 to 28% of isoproterenol's efficacy (Fig. 2A and Table 3, right column). Labetalol, however, which was as efficacious as bucindolol and carvedilol for the AC, did not significantly stimulate ERK1/2 phosphorylation (the small increase seen in Fig. 2A did not reach statistical significance). For the inverse agonists in the AC assay, three compounds (metoprolol, bisoprolol, and atenolol) acted as neutral antagonists (the small increases did not reach statistical significance) on the ERK1/2 pathway. Propranolol showed an opposite efficacy, stimulating ERK1/2 phosphorylation as efficiently as bucindolol and carvedilol and reaching 22% of the isoproterenol efficacy (Fig. 2A and Table 3, right column).
When considering the potency of the compounds in the two signaling pathways (AC and ERK1/2), similar EC50 values were found for all compounds except isoproterenol (Table 2). Even propranolol, which acted as an inverse agonist on AC but as antagonist on ERK1/2, did so with similar potency. In contrast, isoproterenol activated ERK1/2 with a potency markedly lower (450x) than that observed for its stimulation of AC (Fig. 2B, Table 2). The possible meanings of this difference are considered under Discussion.
Ligand Profiles on the 2 Adrenergic Receptor. The -adrenergic ligands were then assessed for their ability to modulate AC and ERK1/2 activity in HEK293S cells expressing the human 2AR. As shown in Fig. 3, only isoproterenol and labetalol were able to increase cAMP levels. Isoproterenol behaved as an agonist, whereas labetalol acted as partial agonist for this pathway, possessing an efficacy of 25% of isoproterenol. Propranolol, metoprolol, bisoprolol, and atenolol all acted as inverse agonists on the cyclase response (Fig. 3 A and B, bottom). Their efficacy to inhibit forskolin-stimulated AC was similar, ranging from 44 to 59% (Table 5, left column). Bucindolol and carvedilol showed no significant efficacy toward the AC pathway up to a concentration of 10-5 M.
Fig. 3. Profile of the different ligands on the 2AR adenylate cyclase pathway. cAMP accumulation experiments were performed in HEK293S cells stably expressing 2AR. A, EC50 of compounds were obtained using increasing concentrations of ligands in the presence (bottom) or absence (top) of 0.3 µM forskolin. B, Emax of compounds were tested using maximal concentrations of ligands, determined from EC50 curves, with (bottom) or without (top) 0.3 µM forskolin. Data represent the mean ± S.E.M of at least three experiments performed in triplicate.
TABLE 5 Emax on cAMP and on MAPK of compounds tested on 2AR
Experimental data from Fig. 3B and 4A, presented as maximal response ± S.E.M. Data are expressed as percentage of the isoproterenol-stimulated response unless otherwise specified.
Fig. 4. Profile of the different ligands on the 2AR MAPK ERK1/2 pathway. ERK1/2 phosphorylation was assessed in HEK293S cells stably expressing 2AR. Compounds were tested at their appropriate maximal stimulation time (1 min for isoproterenol and 2 min for other compounds). A, Emax of compounds were obtained by the use of a maximal ligand concentration (10-5 M for all compounds). Inset, inverse efficacy of compounds was tested in the presence of 100 ng/ml PMA, 90 s. B, EC50 of compounds were determined using increasing concentrations of ligands. Data represent the mean ± S.E.M. of at least four experiments. Statistical significance was determined by two-way ANOVA, followed by a Dunnett test with basal (A) or with PMA (A, inset) as control column. ***, p < 0.001.
Significant differences in the potency of the compounds to modulate cAMP production were observed in cells expressing the 2AR (Table 4, left column). Indeed, whereas the nonselective -adrenergic ligands showed similar high potencies to either stimulate (isoproterenol and labetalol) or inhibit (propranolol) AC activity, the 1AR selective ligands (metoprolol, bisoprolol, and atenolol) showed poor potency to inhibit AC (Fig. 3A and Table 4, left column).
TABLE 4 EC50 on cAMP and on MAPK of compounds tested on 2AR Experimental data from Figs. 3A and 4B presented as mean ± S.E.M.
After determination of the time of maximal activation for each compound (1 or 2 min, depending of the compounds), the relative efficacies and potencies of the compounds to modulate ERK1/2 activity were assessed in the 2AR-expressing cells. As shown in Fig. 4, and similar to what was observed in 1AR-expressing cells, the efficacy pattern found in the ERK1/2 assay was significantly different from that observed for AC activation. In addition to isoproterenol and labetalol (which acted as agonists for the AC pathway), bucindolol and carvedilol (which did not have efficacy for the AC pathway) were also able to significantly increase the ERK1/2 phosphorylation (Fig. 4A). Moreover, propranolol, which was an inverse agonist for the AC, was able to stimulate the ERK1/2 signaling pathway (Fig. 4A). Labetalol, bucindolol, carvedilol, and propranolol were thus classified as partial agonists on the ERK1/2 pathway, their maximal stimulatory activity reached between 38% and 64% of isoproterenol efficacy (Fig. 4A; Table 5, MAPK). Metoprolol, bisoprolol, and atenolol, which were efficacious inverse agonists on the AC, did not seem to modulate ERK1/2 activity. However, because the basal level of ERK1/2 phosphorylation was close to the detection limit, inverse efficacy could have gone unnoticed. Hence, the efficacies of these compounds were reassessed in the presence of phorbol 12-myristate 13-acetate (PMA), a compound known to elevate ERK1/2 activity as a result of protein kinase C stimulation. Under these conditions, metoprolol, bisoprolol, and atenolol significantly decreased the PMA-stimulated ERK1/2 phosphorylation, thus acting as inverse agonists for this pathway (Fig. 4A, inset). This inverse agonist effect was also observed when these compounds were tested after elevation of the ERK1/2 activity by a different activator, the epidermal growth factor (data not shown). The revealed inverse efficacy of metoprolol, bisoprolol, and atenolol did not result from the artificially elevated ERK1/2 activity, because propranolol still acted as an agonist on the PMA-stimulated ERK1/2 activity (data not shown). Moreover, PMA treatment did not reveal inverse efficacy toward the 1AR-regulated ERK1/2 activity for any of the compounds (data not shown), clearly indicating the subtype selectivity of the effect observed.
As was the case for the 1AR-expressing cells, isoproterenol had a much lower (43x) potency to activate ERK1/2 than AC in cells expressing 2AR (Table 4). In contrast, labetalol and propranolol had similar potencies for the two signaling pathways. It is noteworthy that bucindolol and carvedilol, which had no detectable efficacy toward the AC pathway, activated the ERK1/2 with high potency.
Combining the use of multiple ligands, two closely related receptor subtypes, and two effector systems, the present study, in line with other recent observations, illustrates a novel level of complexity in the definition of signaling efficacy and selectivity. When considering a single receptor, several ligands showed a complex efficacy profile, in some cases resulting in the opposite regulation of the two pathways considered by the same ligand. Our results also clearly indicate the existence of subtype-dependent efficacy profiles toward distinct signaling pathways that are not a simple reflection of the binding affinities of the ligands for each of the receptor subtype. Taken together, our data forcefully support the emerging notion that signaling efficacy can no longer be defined as a function of a ligand/receptor couple but needs to include the specific effector(s) considered, thus revealing the pluridimensionality of efficacy.
One of the most striking observations of this study is that, although some ligands have similar efficacies toward ERK and AC, others displayed effectors-specific efficacies. In some cases, effector-specific efficacy profiles were revealed by different rank orders of efficacy for the two pathways. For example, when considering the 1AR, labetalol is an agonist as efficacious as bucindolol and carvedilol for AC, whereas it is much less efficacious than these two compounds toward the ERK1/2 stimulation. In other cases, the efficacy of a given compound is completely different, depending on the pathway considered. This is particularly striking for propranolol, which is an inverse agonist for AC but an agonist of the ERK1/2 pathway for both 1AR and 2AR. The effector-specific efficacy profiles for the two receptors can easily be appreciated by a Cartesian representation of the data that displays the relative efficacy of each compound toward the two signaling pathways (Fig. 5). As can be seen, three of the four possible scenarios have been observed: ACago-ERKago, ACinv-ERKago, and ACinv-ERKinv. Although the case scenario ACago-ERKinv was not observed in the present study, such possibility cannot be excluded on theoretical grounds and is likely to be observed for other ligands and/or other receptors. We and others have previously shown that 2-adrenergic, V2-vasopressin, serotonin 5-HT2C, and -opioid receptor ligands can act as inverse agonists on the adenylyl cyclase pathway but as agonists for the mitogen-activated protein kinase (MAPK) (Azzi et al., 2003; Baker et al., 2003; Audet et al., 2005; Werry et al., 2005). Likewise, Gbahou et al. (2003) reported that proxyfan, a high-affinity histamine H3 receptor ligand, is a partial agonist on AC and ERK1/2 but acts as an inverse agonist for the arachidonic acid release pathway. Although these studies introduced the concept of dual-efficacy ligands, the study of several ligands presented here clearly illustrates the pluridimensionality of ligand efficacy. In the present study, the efficacy description relied on two signaling pathways leading to a bidimensional representation, resulting in four possible efficacy quadrants (Fig. 5). However, more complete descriptions could take into account multiple possible signaling pathways and could thus offer a theoretical representation of efficacy in "n" dimension, resulting in "2"n efficacy quadrants. In addition to defining the qualitative nature of the efficacy toward various signaling pathways, such graphical representation allows us to attribute quantitative terms to these efficacies by using the spatial coordinates of the point corresponding to a given ligand, thus conferring a new meaning to the term efficacy. An example of how efficacies could be reported using such a coordinate system is presented at the bottom of Fig. 5.
Fig. 5. Cartesian representation (AC pathway in abscise and MAPK ERK1/2 pathway in ordinate) of compounds efficacy profiles and their appropriate efficacy coordinates. A, efficacy profile and coordinates of the compounds tested on the 1AR. B, efficacy profile and coordinates of the compounds tested on the 2AR. ACago-ERKago, compounds that are agonists on the AC and on the ERK1/2; ACinv-ERKago, compounds that are inverse agonists on the AC and agonists on the ERK1/2; ACinv-ERKinv, compounds that are inverse agonists on the AC and on the ERK1/2; ACago-ERKinv, compounds that are agonist on the AC and inverse agonists on the ERK1/2. Coordinates are obtained from Tables 3 and 5.
The distribution of the efficacy points presented in Fig. 5 do not follow the theoretical diagonal line predicted by a classic two-state model in which the receptor is either active or inactive for all the effectors considered. The dispersed distribution rather supports the emerging concept that multiple discrete conformations can be active for one pathway but inactive for another and that these conformations can be differentially stabilized/induced by distinct ligands (for review, see Kenakin, 1995, 2001, 2003). This is in line with the concept of ligand-directed stimulus trafficking that was first introduced based on studies reporting that a given agonist can activate two different signaling pathways with distinct orders of potency while acting on a single receptor. Similar differences in the orders of potency were also observed in the present study. For example, the rank order of potency for the 2AR-modulated AC (established independently of their efficacy; ie, combining agonists and inverse agonists) was propranolol isoproterenol = labetalol carvedilol = bucindolol, whereas it was bucindolol carvedilol = propranolol = labetalol > isoproterenol for ERK1/2. The case of isoproterenol is particularly interesting when considering that its potencies to stimulate the two pathways are greatly different. Whereas its potency toward ERK1/2 (EC50 = 214 nM) is in line with its affinity for the receptor (Ki 450 nM; see Table 1), its potency to promote AC activation is much higher (EC50 = 5.0 nM, see Table 4). This suggests that the conformation stabilized/induced by isoproterenol favors the coupling to the AC signaling pathway so that a smaller fraction of the receptor population needs to be in this active conformation to fully activate the AC than to activate the ERK pathway. When considered in the context of the "spare receptors" concept, the isoproterenol-promoted conformation leads to a greater number of spare receptors for the AC than the ERK pathway.
Overall, our data are consistent with the emerging concept that different ligands can stabilize distinct receptor conformations (Vauquelin and Van, 2005) that may differ in their signaling partner preference (Kenakin, 2002). Such behavior can easily be formalized using either the extended or the cubic ternary complex models of G protein activation. Indeed, by including factors controlling the affinity of the receptor for its cognate G protein, these models allow different ligands to induce/stabilize distinct receptor/G protein affinity states (Kenakin, 2004). Given that many receptors can couple to more than one G protein, it can easily be foreseen that the different ligand-promoted receptor conformations could yield differential signaling efficacies through distinct effector systems. Such behavior has indeed been observed in native tissues supporting its potential physiological relevance. For instance, whereas several -adrenergic agonists (e.g., salbutamol, terbutaline, procaterol, zinterol) have been found to promote both Gi and Gs activation, fenoterol, which is an efficacious agonist for the Gs pathway, was found to be inactive toward Gi in rat cardiomyocytes (Xiao et al., 2003; Ponicke et al., 2006). There is no theoretical reason to limit this conformationally based selection to G proteins. It could therefore be proposed that the individual conformations stabilized by each ligand could display a specific set of affinities for the various partners involved in the different behaviors of the receptor, including signaling via multiple effectors, receptor phosphorylation, endocytosis, and desensitization. In this context, it is not surprising that signaling efficacies are often found to be context-dependent. Indeed, the different levels of expression or subcellular distribution of specific partners among cells or tissues would be predicted to affect the signaling profiles observed in these individual systems (Watson et al., 2000).
The capacity of different ligands to promote distinct conformational rearrangements has been clearly confirmed by biophysical studies monitoring the fluorescent properties of intramolecular probes within purified 2AR (Ghanouni et al., 2001; Swaminath et al., 2005). However, the link between the conformations stabilized by specific ligands and their efficacy pattern toward different signaling pathways remains to be established. Speaking intuitively, compounds with the greatest chemical similarity would be expected to stabilize similar conformation, thus resulting in comparable efficacy patterns. As can be seen in Table 6, this prediction seems to be borne out, because compounds with identical efficacy patterns present similar chemical structures.
TABLE 6 Structures of the ligands tested and their efficacy pattern towards AC and ERK1/2 pathways on the two -adrenergic subtypes.
The comparison between 1AR and 2AR led to the observation that in addition to their traditionally defined subtypespecific affinities, -adrenergic ligands also display subtypespecific efficacy patterns (Fig. 5). This is particularly striking when considering labetalol. This compound binds 1AR and 2AR with similar affinities (Table 1) and has similar potencies and efficacies toward the AC pathway for the two receptor subtypes (Figs. 1 and 3, and Tables 3 and 5). However, labetalol is 2AR-selective ERK1/2 agonist is unable to activate this pathway in 1AR-expressing cells. This clearly indicates that subtype specificity cannot be established only on the basis of ligand binding affinities or the relative potency determined for a single signaling pathway. To be pharmacologically exhaustive, the receptor subtype selectivity would therefore need to be determined for each effector system that can be engaged by the receptors considered.
Although the complexity of signaling efficacy and selectivity can be more easily dissected in heterologous expression systems in which specific receptor expression can be controlled, efficacy profiles that differ depending of the effector systems considered have also been observed in native environments. For instance, propranolol was found to act as an inverse agonist toward the AC pathway while being an agonist on the ERK1/2 activity in canine cardiomyocytes, S49 lymphoma cells, and Cos cells endogenously expressing the AR (Azzi et al., 2003). Likewise, proxyfan was found to be a partial agonist when assessed in the context of H3 histaminereceptor-promoted histamine release but an inverse agonist in a GTPS binding assay (Gbahou et al., 2003). These observations suggest that the pluridimensionality of signaling efficacies observed herein is not simply an artifact of overexpression in artificial systems but could be physiologically relevant. Additional studies, however, are needed to assess the extent to which the effector-dependent signaling efficacies will be detectable in normal and pathophysiological conditions.
It should also be noted that primary drug screenings are often carried out using unique cell-based assays, taking advantage of robust signals detected in heterologous expression systems. The observation that different efficacies can be observed depending on the signaling outcome considered has obvious implication on the conclusions that can be drawn from such screening campaigns. On the one hand, it stresses the importance of selecting an effector pathway that is pertinent for the pathological condition considered. On the other hand, taking the pluridimensionality of efficacy into account, testing the signaling efficacies of compounds into multiple assay systems could allow us to link specific signaling properties to given therapeutic activities or side effects, thus increasing the chances of identifying compounds with efficacy profiles that may have better therapeutic outcomes. A better description of the efficacy profiles for each member of a drug family could also help us understand why individual members of a drug class sometimes have different therapeutic indications. For example, clinical trials assessing the efficacy of -blockers in the treatment of congestive heart failure revealed that carvedilol, metoprolol, and bisoprolol, but not bucindolol, decreased mortality, and the reason for the lack of beneficial action of bucindolol remains elusive. In fact, although carvedilol and metoprolol are approved in many countries for the treatment of heart failure, the molecular basis of their beneficial effects is still poorly understood. Based on the efficacy profiles determined for two receptor subtypes and two signaling pathways, the present study did not provide a signature that could predict efficacy in heart failure. Whether increasing the number of signaling modalities measured for each compounds will allow such predictive signatures to emerge remains to be investigated.
One of the major conclusions of the present study is that efficacy and selectivity toward GPCRs cannot be defined based simply on a ligand/receptor couple but should also include the effector system in its operational definition. Given the increasing number of signaling possibilities that have recently surfaced for GPCRs, the pluridimensionality of efficacy will certainly become more and more evident and will be integrated in our way of thinking about drug action and classification.
Acknowledgements
We thank GlaxoSmithKline (Research Triangle Park, NC) and Dr. Michael Bristow (University of Colorado Health Sciences Center, CO) for kindly providing carvedilol and bucindolol, respectively. We are grateful to Dr. Monique Lagacé for her critical reading of the manuscript.
ABBREVIATIONS: GPCR, G protein-coupled receptor; ICI118,551, (±)-1-[2,3-(dihydro-7-methyl-1H-inden-4-yl)oxy]-3-[(1-methylethyl)amino]-2-butanol; AR, adrenergic receptor; AC, adenylyl cyclase; ERK, extracellular signal-regulated kinase; HRP, horseradish peroxidase; HEK, human embryonic kidney; PBS, phosphate-buffered saline; TBS-T, Tris-buffered saline-Tween 20; PMA, phorbol 12-myristate 13-acetate; MAPK, mitogen-activated protein kinase; ANOVA, analysis of variance; CGP-12177, 4-[3-[(1,1-dimethylethyl)amino]-2-hydroxypropoxy]-1,3-dihydro-2H-benzimidazol-2-one; Iso, isoproterenol; Lab, labetalol; Buc, bucindolol; Carv, carvedilol; Prop, DL-propanolol; Bis, bisoprolol; Met, (±)-metoprolol; At, S(-)-atenolol.
【参考文献】
Audet N, Paquin-Gobeil M, Landry-Paquet O, Schiller PW, and Pineyro G (2005) Internalization and Src activity regulate the time course of ERK activation by delta opioid receptor ligands. J Biol Chem 280: 7808-7816.[Abstract/Free Full Text]
Azzi M, Charest PG, Angers S, Rousseau G, Kohout T, Bouvier M, and Pineyro G (2003) Beta-arrestin-mediated activation of MAPK by inverse agonists reveals distinct active conformations for G protein-coupled receptors. Proc Natl Acad Sci USA 100: 11406-11411.[Abstract/Free Full Text]
Baker JG, Hall IP, and Hill SJ (2003) Agonist and inverse agonist actions of -blockers at the human 2-adrenoceptor provide evidence for agonist-directed signaling. Mol Pharmacol 64: 1357-1369.[Abstract/Free Full Text]
Berg KA, Maayani S, Goldfarb J, Scaramellini C, Leff P, and Clarke WP (1998) Effector pathway-dependent relative efficacy at serotonin type 2A and 2C receptors: evidence for agonist-directed trafficking of receptor stimulus. Mol Pharmacol 54: 94-104.[Abstract/Free Full Text]
Bond RA (1997) Do recent operational studies indicate that a single state model is no longer applicable to G protein-coupled receptors? Ann N Y Acad Sci 812: 92-97.[Medline]
Gay EA, Urban JD, Nichols DE, Oxford GS, and Mailman RB (2004) Functional selectivity of D2 receptor ligands in a Chinese hamster ovary HD2L cell line: evidence for induction of ligand-specific receptor states. Mol Pharmacol 66: 97-105.[Abstract/Free Full Text]
Gbahou F, Rouleau A, Morisset S, Parmentier R, Crochet S, Lin JS, Ligneau X, Tardivel-Lacombe J, Stark H, Schunack W, Ganellin CR, Schwartz JC, and Arrang JM (2003) Protean agonism at histamine H3 receptors in vitro and in vivo. Proc Natl Acad Sci USA 100: 11086-11091.[Abstract/Free Full Text]
Ghanouni P, Gryczynski Z, Steenhuis JJ, Lee TW, Farrens DL, Lakowicz JR, and Kobilka BK (2001) Functionally different agonists induce distinct conformations in the G protein coupling domain of the beta 2 adrenergic receptor. J Biol Chem 276: 24433-24436.[Abstract/Free Full Text]
Hall DA, Beresford IJ, Browning C, and Giles H (1999) Signalling by CXC-chemokine receptors 1 and 2 expressed in CHO cells: a comparison of calcium mobilization, inhibition of adenylyl cyclase and stimulation of GTPgammaS binding induced by IL-8 and GROalpha. Br J Pharmacol 126: 810-818.[CrossRef][Medline]
Hoffmann C, Leitz MR, Oberdorf-Maass S, Lohse MJ, and Klotz KN (2004) Comparative pharmacology of human beta-adrenergic receptor subtypes-characterization of stably transfected receptors in CHO cells. Naunyn-Schmiedeberg's Arch Pharmacol 369: 151-159.[Medline]
Kenakin T (1995) Agonist-receptor efficacy. II. Agonist trafficking of receptor signals. Trends Pharmacol Sci 16: 232-238.[CrossRef][Medline]
Kenakin T (2001) Inverse, protean, and ligand-selective agonism: matters of receptor conformation. FASEB J 15: 598-611.[Abstract/Free Full Text]
Kenakin T (2002) Drug efficacy at G protein-coupled receptors. Annu Rev Pharmacol Toxicol 42: 349-379.[CrossRef][Medline]
Kenakin T (2003) Ligand-selective receptor conformations revisited: the promise and the problem. Trends Pharmacol Sci 24: 346-354.[CrossRef][Medline]
Kenakin T (2004) Principles: receptor theory in pharmacology. Trends Pharmacol Sci 25: 186-192.[CrossRef][Medline]
Krueger KM, Witte DG, Ireland-Denny L, Miller TR, Baranowski JL, Buckner S, Milicic I, Esbenshade TA, and Hancock AA (2005) G protein-dependent pharmacology of histamine H3 receptor ligands: evidence for heterogeneous active state receptor conformations. J Pharmacol Exp Ther 314: 271-281.[Abstract/Free Full Text]
Kurrasch-Orbaugh DM, Watts VJ, Barker EL, and Nichols DE (2003) Serotonin 5-hydroxytryptamine 2A receptor-coupled phospholipase C and phospholipase A2 signaling pathways have different receptor reserves. J Pharmacol Exp Ther 304: 229-237.[Abstract/Free Full Text]
Lavoie C, Mercier JF, Salahpour A, Umapathy D, Breit A, Villeneuve LR, Zhu WZ, Xiao RP, Lakatta EG, Bouvier M, and Hebert TE (2002) 1/2-Adrenergic receptor heterodimerization regulates 2-adrenergic receptor internalization and ERK signaling efficacy. J Biol Chem 277: 35402-35410.[Abstract/Free Full Text]
McLaughlin JN, Shen L, Holinstat M, Brooks JD, Dibenedetto E, and Hamm HE (2005) Functional selectivity of G protein signaling by agonist peptides and thrombin for the protease-activated receptor-1. J Biol Chem 280: 25048-25059.[Abstract/Free Full Text]
Milligan G (2003) Constitutive activity and inverse agonists of G protein-coupled receptors: a current perspective. Mol Pharmacol 64: 1271-1276.[Abstract/Free Full Text]
Moniri NH, Covington-Strachan D, and Booth RG (2004) Ligand-directed functional heterogeneity of histamine H1 receptors: novel dual-function ligands selectively activate and block H1-mediated phospholipase C and adenylyl cyclase signaling. J Pharmacol Exp Ther 311: 274-281.[Abstract/Free Full Text]
Pauwels PJ, Gommeren W, Van Lommen G, Janssen PA, and Leysen JE (1988) The receptor binding profile of the new antihypertensive agent nebivolol and its stereoisomers compared with various -adrenergic blockers. Mol Pharmacol 34: 843-851.[Abstract]
Ponicke K, Groner F, Heinroth-Hoffmann I, and Brodde OE (2006) Agonist-specific activation of the beta2-adrenoceptor/Gs-protein and beta2-adrenoceptor/Giprotein pathway in adult rat ventricular cardiomyocytes. Br J Pharmacol 147: 714-719.[CrossRef][Medline]
Ponicke K, Heinroth-Hoffmann I, and Brodde OE (2002) Differential effects of bucindolol and carvedilol on noradenaline-induced hypertrophic response in ventricular cardiomyocytes of adult rats. J Pharmacol Exp Ther 301: 71-76.[Abstract/Free Full Text]
Reeves PJ, Thurmond RL, and Khorana HG (1996) Structure and function in rhodopsin: high level expression of a synthetic bovine opsin gene and its mutants in stable mammalian cell lines. Proc Natl Acad Sci USA 93: 11487-11492.[Abstract/Free Full Text]
Spengler D, Waeber C, Pantaloni C, Holsboer F, Bockaert J, Seeburg PH, and Journot L (1993) Differential signal transduction by five splice variants of the PACAP receptor. Nature (Lond) 365: 170-175.[CrossRef][Medline]
Strange PG (2002) Mechanisms of inverse agonism at G-protein-coupled receptors. Trends Pharmacol Sci 23: 89-95.[CrossRef][Medline]
Swaminath G, Deupi X, Lee TW, Zhu W, Thian FS, Kobilka TS, and Kobilka B (2005) Probing the 2 adrenoceptor binding site with catechol reveals differences in binding and activation by agonists and partial agonists. J Biol Chem 280: 22165-22171.[Abstract/Free Full Text]
Urban JD, Clarke WP, von Zastrow M, Nichols DE, Kobilka BK, Weinstein H, Javitch JA, Roth BL, Christopoulos A, Sexton P, Miller K, Spedding M, and Mailman RB (2006) Functional selectivity and classical concepts of quantitative pharmacology. J Pharmacol Exp Ther, in press.
Vauquelin G and Van LI (2005) G protein-coupled receptors: a count of 1001 conformations. Fundam Clin Pharmacol 19: 45-56.[CrossRef][Medline]
Watson C, Chen G, Irving P, Way J, Chen WJ, and Kenakin T (2000) The use of stimulus-biased assay systems to detect agonist-specific receptor active states: implications for the trafficking of receptor stimulus by agonists. Mol Pharmacol 58: 1230-1238.[Medline]
Werry TD, Gregory KJ, Sexton PM, and Christopoulos A (2005) Characterization of serotonin 5-HT2C receptor signaling to extracellular signal-regulated kinases 1 and 2. J Neurochem 93: 1603-1615.[CrossRef][Medline]
Xiao RP, Zhang SJ, Chakir K, Avdonin P, Zhu W, Bond RA, Balke CW, Lakatta EG, and Cheng H (2003) Enhanced G(i) signaling selectively negates beta2-adrenergic receptor (AR)-but not beta1-AR-mediated positive inotropic effect in myocytes from failing rat hearts. Circulation 108: 1633-1639.[Abstract/Free Full Text]
作者单位：Department of Biochemistry and Groupe de Recherche Universitaire sur le Médicament, Institute for Research in Immunology and Cancer, Université de Montréal, Montréal, Québec, Canada
1 From the Division of Epidemiology, University of Minnesota, Minneapolis (CWK), and the Carolina Population Center, University of North Carolina, Chapel Hill (LSA).
2 Supported by the National Science Foundation, The Nestle Foundation, and The Emory Internationalization Program. 3 Address reprint requests to CW Kuzawa, Department of Anthropology, Northwestern University, 1810 Hinman Avenue, Evanston, IL 60208. E-mail: kuzawa{at}northwestern.edu.
ABSTRACT
Background: The finding that persons with low birth weight have a higher cardiovascular disease (CVD) risk than do persons with higher birth weight remains poorly understood.
Objective: We tested the hypothesis that maternal arm fat area (MAFA) in the third trimester of pregnancy and birth weight of offspring are inversely related to the offspring’s risk of CVD.
Design: In a 1-y birth cohort study (1983–1984), 296 male and 307 female offspring were followed up (1998–1999) to measure their lipid profiles. Participants came from randomly selected communities of Cebu, Philippines.
Results: MAFA (log cm2) was positively associated (ß) with HDL cholesterol (0.12 log mg/dL; P < 0.01) and inversely associated with total cholesterol (-10.0 mg/dL; P < 0.10), LDL cholesterol (-13.1 mg/dL; P < 0.01), and the ratios of total to HDL cholesterol and LDL to HDL cholesterol (both P < 0.001) in males. These relations were independent of birth weight, present adiposity, energy and fat intakes, maturity, and income. Birth weight 2.6 kg was associated with elevated LDL cholesterol (9.9 mg/dL; P < 0.01) and an elevated ratio of LDL to HDL cholesterol (0.22; P < 0.10) only in males. In females, MAFA related positively to total (15.5 mg/dL; P < 0.05) and LDL (11.9 mg/dL; P < 0.05) cholesterol.
Conclusions: In this Filipino population, mothers with low energy status during pregnancy gave birth to male offspring who had a high CVD risk in adolescence, as indicated by lipid profiles. The findings in females are less consistent with the fetal origins hypothesis and suggest sex differences in the relation between fetal nutrition and postnatal lipid metabolism.
Key Words: Cardiovascular diseases • birth weight • fetal nutrition • pregnancy • cholesterol • lipoproteins • sex differences • adolescence • Philippines • risk factors • dietary fats • programming • fetal origins hypothesis • nutrition transition • maternal energy status • Cebu Longitudinal Health and Nutrition Survey
INTRODUCTION
Cholesterol profiles are among the suite of cardiovascular disease (CVD) risk factors believed to contribute to the inverse relation between birth weight and CVD mortality in human populations (1). The most common finding among studies to date is an elevation of total cholesterol in relation to markers of poor birth outcome, typically indexed by weight at birth (2–5). It is widely assumed that low birth weight (LBW) indicates fetal growth restriction and underlying nutritional insufficiency, which is viewed as the stimulus that triggers fetal adaptations that have long-term physiologic effects on cholesterol metabolism (6).
Because fetal nutritional sufficiency is in part a reflection of the maternal capacity to supply the energy and nutrients necessary for growth, maternal nutritional status during pregnancy may be an important factor influencing patterns of CVD risk in offspring (6). Although maternal dietary manipulation during pregnancy has long-term effects on lipid metabolism in offspring in animal models (7, 8), few studies in humans have investigated the possible role of maternal nutrition during pregnancy in these relations (9). Of the 3 studies of cholesterol that included markers of maternal nutritional status during pregnancy, poor maternal nutritional status was alternately found to increase (10), to decrease (3), or to have no association with CVD risk in offspring (11), as indicated by lipid profiles. Of the 2 studies that investigated famine-exposed populations, 1 showed evidence of an effect of famine exposure during the first trimester on the ratio of LDL to HDL cholesterol in offspring, whereas the second found no significant famine-related differences in lipid profiles in offspring (12, 13).
Despite evidence that LBW is associated with an atherogenic lipid profile and an elevated risk of CVD mortality, the contribution of maternal nutritional status during pregnancy to cholesterol metabolism in offspring remains a matter of speculation. In the present study, we used data collected prospectively over a 17-y period in the Philippines to test the hypothesis that poor maternal energy status during pregnancy and LBW predict elevated CVD risk in offspring during adolescence, as indexed by cholesterol lipid profiles.
SUBJECTS AND METHODS
Sample characteristics
Participants were enrolled in the Cebu Longitudinal Health and Nutrition Survey, a community-based birth cohort study of infants born in 1983–1984. The study area was metropolitan Cebu, the second largest metropolitan area in the Philippines. The 33 metropolitan Cebu communities randomly selected for the survey included densely populated urban neighborhoods, periurban neighborhoods, and rural villages in the surrounding mountains and islands. All pregnant women in the selected communities were initially invited to participate and were included in the longitudinal study if they gave birth between 1 May 1983 and 30 April 1984 (n = 3327). The child sample (3080 single live births) was thus representative of singletons born during that 1-y interval. Initial refusal rates were low (3%), but we do not have information on those who refused to participate. The present analysis used data collected from mothers during the third trimester of pregnancy (1983–1984) and from their offspring at birth (1983–1984) and at 14–16 y of age (1998–1999).
Maternal arm fat area (MAFA) was calculated from triceps skinfold thickness and midupper arm circumference measured during the baseline survey (1983–1984) at a mean (± SD) of 30 ± 5 wk of gestation. Maternal height was measured with a folding stadiometer. Infant length was measured within 6 d of birth by trained project staff using custom-designed length boards. For infants born in hospitals, birth weight was measured by birth attendants using hospital scales. For infants born at home, birth weight was measured by birth attendants who had been provided with Salter (London) hanging-type scales and trained in their use. Gestational age was estimated from the mother’s report of the date of her last menstrual period. In cases where this date was unknown, when pregnancy complications occurred, or if the infant weighed < 2.5 kg at birth, gestational age was determined by using the Ballard method (14).
At the most recent follow-up survey in 1998, 2089 adolescents aged 14–16 y were located and interviewed. Of these, 2056 had birth weight, gestational age, and current measurements. From those subjects, a subsample of 307 females and 296 males were selected at random within 2 birth weight strata for blood sample collection. To avoid including subjects who were small at birth because of prematurity, we limited the subsample to subjects who were carried to term, which was defined as a gestational age at birth 
We compared the baseline characteristics of the adolescents who were included in the 1998 follow-up with those of the subjects who were in the sample at baseline (single live births). The mean birth weight of the subjects lost to follow-up was 50 g lower than that of those retained in the sample. This is most likely attributable to the higher mortality rates among the low-birth-weight infants. Birth length did not differ significantly between the 2 groups. The subjects lost to follow-up were more likely than those retained in the sample to have been urban residents (82.5% compared with 73.5%), but there were no significant differences in household assets or in maternal education, height, age, or parity. We also assessed potential biases in the subsample selected for lipid analysis. The females who were included in the CVD study had significantly lower birth weight, current height, and current weight (all P < 0.05) than did the females who were excluded from the study, and this result was consistent with our sampling design; however, among the males, the 2 groups did not differ significantly in these variables. When the subjects who were included in the CVD substudy were compared with those who were not included, there were no significant differences among either sex in body mass index (BMI), skinfold thickness, income, or dietary fat intake.
Data measurement protocol
For measurement of lipid profiles, participants were asked to fast overnight for 12 h, and blood samples were collected in clinics the following morning with the use of EDTA-coated tubes. After separation, samples were frozen and shipped on dry ice to the Emory Lipid Research Laboratory (Atlanta) for analysis of lipid profiles. All samples remained frozen at -80 °C until ready for analysis. Total lipid concentrations were measured by using enzymatic methods with reagents from Beckman Diagnostics on the Beckman Diagnostics CX5 chemistry analyzer (Fullerton, CA). HDL- and LDL-cholesterol concentrations were measured by using the homogenous assays Direct HDL Cholesterol and Direct LDL Cholesterol (Equal Diagnostics, Exton, PA). Total cholesterol concentrations were measured with an enzymatic kit, and triacylglycerol concentrations were measured with a glycerol blank as a 2-step reaction (Beckman Coulter Diagnostics, Fullerton, CA). The atherogenic ratios of total to HDL cholesterol and LDL to HDL cholesterol were also calculated (15). The Emory Lipid Research Laboratory participates in the Lipid Standardization Program of the Centers for Disease Control and Prevention and the National Heart, Lung, and Blood Institute to ensure accuracy and precision of measurements.
Body weight, height, waist circumference, midupper arm circumference, and triceps skinfold thickness were measured by using standard anthropometric techniques (16). BMI was calculated as weight (kg)/height2 (m). In this sample, the child’s arm fat area (CAFA), which was calculated from triceps skinfold thickness and midupper arm circumference (16), was more strongly correlated with current lipids than was current triceps skinfold thickness. Thus, arm fat area was used to control the effects of current adiposity on lipids. Prior work showed that third trimester MAFA is a positive predictor of birth weight in this population, and MAFA was therefore used as a marker of maternal energy status (17). During the 1998–1999 follow-up, the adolescents’ dietary intake was measured by using two 24-h recalls on consecutive days, and the mean was used in analyses. Energy and fat intakes were calculated by using the Food Composition Tables Recommended for Use in the Philippines, which is published by the Food and Nutrition Research Institute of the Philippines (18). Among the males, maturational status was assessed by self-rated, 5-level pubic-hair staging, which was validated against physician assessments (CW Kuzawa and LS Adair, unpublished observations, 2002). For the females, longitudinally assessed menarcheal age estimates were used to construct a 5-level maturational status variable. When data were collected for the 1998–2000 survey, the females were completely surveyed before the males. Consequently, the males were 1 y older than the females. Informed consent was obtained from all participants, and human subjects clearance was obtained from the Institutional Review Boards of the Emory University Medical School and the University of North Carolina, Chapel Hill.
Statistical analyses
All analyses were performed with version 7 of the STATA Statistical Package (19). The population was sampled at random within each of 2 strata. In the 1998 survey, we randomly selected 50% (154 of 327) of the subjects with LBW (birth weight 2.6 kg) and 25% (449 of 1729) of the subjects with high birth weight for lipid measurement. We used 2.6 kg, rather than the more conventional LBW cutoff of 2.5 kg, to ensure adequate numbers of LBW subjects in our sample. To get an unbiased estimate of distributional characteristics (eg, mean LDL cholesterol or dietary energy intake of the adolescents), we down-weighted the observations in the LBW subjects, so that both birth weight strata would be represented in the estimate in proportion to their occurrence in the population. Probability sampling weights for each stratum were calculated as the inverse of the within-stratum sampling fraction, and the survey mean procedure (svymean) of STATA was used to estimate distributional characteristics.
We used regression analysis to model the relations between birth weight and MAFA as predictor variables with offspring lipid profiles as outcomes. The expansion of the proportion of subjects with LBW in our design adds information, and therefore precision, to an estimate of the relation of blood lipid concentrations to other factors, such as MAFA. However, the oversampling of LBW subjects may create confounding if the LBW variable is related to both the dependent and the independent variables. Therefore, to estimate relations of adolescent blood lipid concentrations to MAFA, we assessed the extent of bias caused by the design by using regression models with and without the variable defining the LBW sampling stratum. An additional difficulty arises because the design variable, LBW, may be in the causal pathway between blood lipid concentrations and several of the predictor variables. We carefully considered the extent to which any effect of including the LBW variable on the regression coefficient of MAFA reflected design-induced confounding, naturally occurring confounding, or overadjustment for a factor in the causal pathway. However, sensitivity analyses for all models relating MAFA to offspring blood lipids in this study showed that the design variable (LBW) had a negligible effect on the estimates of regression coefficients, suggesting that overrepresentation of LBW subjects in our sample was not a significant confounder of MAFA-lipid relations. These findings are reported in Results.
Previous studies report sex differences in the association between prenatal factors and later lipid profiles (eg, reference 2). Because initial pooled models (males and females) indicated significant sex x MAFA interactions for 4 of the 6 lipid outcomes analyzed, all subsequent models were stratified by sex. We present a series of regression models designed to assess the relative and independent effects of LBW and MAFA on offspring lipids while investigating any biases in our estimates due to sampling design. Thus, we report results of a base model including only control variables (model 1), to which was added LBW (model 2) or MAFA (model 3), and finally, both LBW and MAFA simultaneously (model 4).
RESULTS
Descriptive statistics for the study participants are shown in Table 1. According to growth charts for US children and adolescents from the Centers for Disease Control and Prevention (20), 14 of the 603 adolescents in our sample (2%) were classified as overweight (7 males and 7 females), but only 2 of these adolescents (both male) were classified as obese. American adolescents of comparable age obtained 50% more of their energy from fat than did the participants in the Cebu study (33% compared with 22%) (21). Despite the favorable levels of obesity and the favorable fat intake in the Cebu sample, lipid profiles suggested a relatively high risk of CVD. The mean lipid concentrations in the Cebu sample were roughly comparable to those of US adolescents of the same age (22). For both sexes, mean total cholesterol and LDL-cholesterol concentrations were higher and mean HDL-cholesterol concentrations were lower than the values reported for adolescents in Taiwan (23), Japan (24), and Singapore (25). As expected, most lipid measures were higher in the females than in the males. The characteristics of the mothers during pregnancy are shown in Table 2.
View this table:
TABLE 1 . Anthropometric measures, dietary intakes of energy and fat, and lipid concentrations in adolescents who participated in the Cebu Longitudinal Health and Nutrition Survey1
View this table:
TABLE 2 . Mothers’ characteristics during pregnancy, 1983–19841
Because prior studies reported significant sex differences in the association between prenatal nutrition variables and later lipid concentrations (2), we first tested for sex x MAFA interactions in pooled models (n = 603) controlling for male sex (male = 1), CAFA, energy intake, percentage of energy from fat, maturity, and household income (at birth and in 1998). There were significant male x MAFA interactions for total cholesterol (P < 0.05), LDL cholesterol (P < 0.01), and the ratio of total to HDL cholesterol (P < 0.01) and LDL to HDL cholesterol (P < 0.01). Therefore, all models reported are stratified by sex.
In Tables 3 and 4, results are shown for multivariate regression models in the males and females, respectively, that relate each lipid outcome to a set of control variables (model 1), to which was added LBW (model 2) or MAFA (model 3) or both LBW and MAFA simultaneously (model 4). In multivariate models, LBW was associated with a significant elevation in LDL cholesterol in the males, yielding a borderline significant elevation in the ratio of LDL to HDL cholesterol (model 2). Results were similar when birth weight was modeled as a continuous variable, with birth weight (in kg) modestly related only to LDL cholesterol in the males (ß ± SE = 7.2 ± 3.7 mg/dL; P = 0.051). In the females, neither LBW nor birth weight modeled as a continuous variable were significantly associated with any lipid outcome. Although all participants were from term births, controlling for gestational age at birth had no measurable effect on the ß coefficients for either LBW or birth weight as a continuous variable in any lipid model.
View this table:
TABLE 3 . Regression models relating low birth weight (LBW) and maternal arm fat area (MAFA) in the third trimester of pregnancy to lipid concentrations in male adolescents from Cebu, Philippines1
View this table:
TABLE 4 . Regression models relating low birth weight (LBW) and maternal arm fat area (MAFA) in the third trimester of pregnancy to lipid concentrations in female adolescents from Cebu, Philippines1
There were several significant relations between MAFA and offspring lipids in both sexes (model 3). In the males, MAFA was significantly inversely related to LDL and total cholesterol and positively related to HDL cholesterol. As a result of these relations, MAFA was strongly inversely related to both atherogenic ratios (total cholesterol:HDL cholesterol and LDL cholesterol:HDL cholesterol) in the males and roughly doubled the explained variance in both ratios. In the females, there were significant positive relations between MAFA and both LDL and total cholesterol (model 3). Including a variable defining the gestational age at which the MAFA measurement was made had no effect on any of the regression coefficients relating MAFA to offspring lipids in either sex.
For all the lipid outcomes investigated, inclusion of the LBW variable had a negligible effect on the ß coefficient or significance level for the relation with MAFA (model 4). Because the MAFA-lipid relations in this sample were independent of LBW status, overrepresentation of LBW subjects in our sample was not a confounder of the MAFA-lipid relations documented here. There were no significant MAFA x LBW interactions in any of the models (data not shown).
In this sample, offspring adiposity correlated with lipids and both MAFA (r = 0.26 in both sexes) and birth weight (r = 0.09 and 0.15 in the males and the females, respectively), which could confound associations between MAFA or LBW and offspring lipids. We investigated this potential source of confounding by considering the change in the ß coefficients for MAFA and LBW after removal of CAFA from multivariate models (models 2 and 3). In the males, the coefficients relating log MAFA to LDL (-7.9 mg/dL; 95% CI: -17.2, 1.5; P < 0.09) and total (-4.5 mg/dL; 95% CI: -15.5, 6.4; P > 0.4) cholesterol were reduced to nonsignificance after removal of CAFA. Coefficients relating MAFA to HDL cholesterol (0.11 log mg/dL; 95% CI: 0.02, 0.17) and to the ratios of total to HDL cholesterol (-0.49; 95% CI: -0.84, 0.14) and LDL to HDL cholesterol (-0.47; 95% CI: -0.79, -0.15) were attenuated but remained significant (all P < 0.02). There were similar effects of removing CAFA on the coefficients relating LBW to LDL cholesterol (9.5 mg/dL; 95% CI: 1.8, 17.1; P < 0.02) and the ratio of LDL to HDL cholesterol (0.20; 95% CI: -0.06, 0.47; P < 0.121). In the females, the slope and precision of ß coefficients relating MAFA to LDL (13.7 mg/dL; 95% CI: 3.0, 24.3; P < 0.01) and total (16.9 mg/dL; 95% CI: 2.7, 31.1; P < 0.02) cholesterol were strengthened after removal of CAFA. Thus, the inverse relations between MAFA and CVD risk in the males were strengthened after the adolescents’ own adiposity was controlled for, and some portion of the positive relations between MAFA and CVD risk in the females was probably confounded by positive associations between MAFA and the adolescents’ own adiposity.
DISCUSSION
In this sample of adolescent Filipinos, maternal energy status during pregnancy had opposite relations with CVD risk in the male and female offspring. The relations in the males were consistent with the expectations of the fetal origins hypothesis: poor maternal energy status, as reflected in third trimester MAFA, predicts elevated CVD risk in offspring, as indicated by LDL-, HDL-, and total cholesterol concentrations and the ratios of total to HDL cholesterol and LDL to HDL cholesterol. The relations with MAFA in the males were strong and independent of the adolescents’ own energy and fat intakes, adiposity, maturity, and household income (both at the age of the cholesterol measurement and at birth). Our results in the males provide some of the strongest evidence yet published that poor maternal energy status during pregnancy has persistent effects on offspring lipids, thus elevating CVD risk.
In contrast with the males, MAFA was positively related to LDL cholesterol and total cholesterol in the female offspring. The finding of a positive relation between maternal nutritional status during pregnancy and offspring CVD risk is infrequently reported (10). It is unclear whether the relations in the females in our sample reflect a positive effect of MAFA on CVD risk in offspring (through "programming") or merely the effect on blood cholesterol of the adolescents’ own adiposity, which is positively correlated with MAFA during pregnancy. The latter interpretation is supported by the finding that the slope and significance of the relations between MAFA and both total and LDL cholesterol were strengthened after removal of CAFA from the models in females. Thus, we are unable to rule out confounding of the positive associations of MAFA with LDL and total cholesterol in the females by residual variance in the adolescents’ adiposity that was not indexed by their arm fat area. In the males, removal of CAFA from the models had the opposite effect on the regression coefficients with MAFA, reducing both their slope and precision. A tentative interpretation of these findings is that there is a pathway relating low MAFA during pregnancy to elevated CVD risk in male but not female offspring and that this pathway is strongest once positive correlations between maternal and child adiposity are held constant.
Although the slope and significance of the ß coefficients were attenuated in the models that did not include CAFA, most MAFA-lipid relations documented in the males remained significant after removal of this measure of postnatal body size. This may indicate that lipid programming is less dependent on postnatal growth than has been suggested for other CVD risk factors (26). For instance, the finding that inverse relations between birth weight and blood pressure are often significant only when postnatal body size is controlled for has been interpreted as evidence for a role of postnatal growth in these relations (26, 27).
Few prior studies have explored the association between maternal pregnancy nutrition and offspring lipids, and contradictory findings limit our ability to generalize. Our results in the males are most similar to findings among Chinese adults studied by Mi et al (3), who found that maternal BMI measured at 15 wk of gestation was inversely associated with total and LDL cholesterol and triacylglycerol in offspring. In this sample, BMI measured close to term (38 wk of gestation) was not significantly related to any lipid in offspring. The positive association between MAFA and total cholesterol in the females in our sample is similar to findings from Jamaica in which first-trimester maternal BMI was positively correlated with total cholesterol in offspring during childhood (10). Cowin and Emmett’s (11) finding of no association between maternal BMI before pregnancy and offspring lipids may be less comparable because of the young age of the sample (31–43 mo).
Differences between studies in the age at which maternal nutritional status measurements were made may have contributed to some of these inconsistencies. Maternal nutritional status may have different implications for fetal growth or nutritional sufficiency depending on when it is measured because both preconception nutritional status and nutrition during pregnancy may have independent as well as interactive effects (28). There is presently no consensus on the timing of critical periods in lipid metabolism programming. The studies of Mi et al (3) and the Dutch famine study (13) suggest a role for early pregnancy nutritional status or diet in offspring lipid programming, whereas animal model research, though sparse, suggests an effect of nutrition during late gestation and the early postnatal period (7). In the present analysis, most of the women were measured between 27 and 34 wk of gestation, thus representing early third-trimester nutritional status. However, accounting for gestational age at the time of measurement did not significantly alter the relations documented here. Our third-trimester measure of MAFA is probably reflective of cumulative energy sufficiency throughout pregnancy and is therefore incapable of clarifying the timing of critical periods in lipid metabolism programming. Prospective studies with detailed repeat measures of maternal nutritional status and diet during pregnancy will be necessary to establish more definitively whether human lipid metabolism is sensitive to maternal nutrition, and if so, during which stage or stages of gestation.
Among the males, adding MAFA to the base model improved the model R2 25% for LDL cholesterol (from 8.6% to 10.8% of variance explained) and 82% for the ratio of total to HDL cholesterol (from 5.7% to 10.4%). Thus, although much of the variance in lipid concentrations remained unexplained by our models, MAFA contributed substantially to the variance that was explained. The magnitude of the relations in the males is in line with that of prior research. In the sample studied by Mi et al (3), the offspring of mothers in the lowest quartile of the 15-wk BMI distribution had mean LDL- and total cholesterol concentrations that were 15 and 17 mg/dL higher, respectively, than those of the offspring of mothers in the highest quartile. Although we used a different marker of maternal nutritional status, there were similar lipid differences across the MAFA distribution in the males. In our sample, a 1-SD change in MAFA was associated with a 4.6-mg/dL change in LDL-cholesterol concentration and a 0.25-unit change in the ratio of total to HDL cholesterol.
Our finding of significant sex x MAFA interactions for most lipid measures is also in agreement with past research documenting sex differences in the relations between prenatal factors and postnatal lipids. Studies of fetal influences document sex differences more often for lipids than for other CVD risk factors, such as blood pressure, and associations between lipids and birth weight are more common among males than among females (2, 29–33). The few animal model studies that investigated lipid metabolism programming by maternal pregnancy nutrition showed similar evidence for sex differences, with effects most (7), or only (8), evident in males. In the sample in the present study, we also found that relations between maternal adiposity during pregnancy and blood pressure in offspring during adolescence were more consistent among males (27). It will be important to follow up our population as they enter adulthood to establish whether these sex differences are persistent or are transitory effects of the hormonal changes of puberty (34). However, the consistency of sex differences in the relation between prenatal factors and lipid profiles in humans and animal models provides a strong incentive to investigate mechanisms of sex differences in lipid programming.
Relations between MAFA and offspring lipids were independent of birth weight, suggesting that maternal energy status during pregnancy may influence lipid profiles in offspring in the absence of measurable changes in birth outcomes. This finding is in agreement with the findings of past research on lipid metabolism in other populations (3, 13). Among the survivors of the Dutch famine (13), the ratio of LDL to HDL cholesterol was elevated in offspring of mothers who experienced the famine during the first trimester of pregnancy, despite no significant decline in birth weight. In the Cebu sample, we previously documented inverse relations between maternal triceps skinfold thickness during pregnancy and blood pressure in offspring that were independent of birth weight (27). The present findings underscore the limitations of birth weight as a marker of the nutritional or related prenatal exposures that influence long-term CVD risk.
CVD is rapidly becoming a key public health challenge in many developing nations, particularly in the Asia-Pacific region (35). Our findings illustrate significant relations between maternal energy status and CVD risk in offspring in an adolescent population with low obesity levels and a low intake of dietary fat. The strongest postnatal predictors of individual lipid outcomes in our sample were current body fat measures, which in turn are responsive to diet and activity patterns. On the basis of trends observed in the full sample of mothers in the Cebu Longitudinal Health and Nutrition Survey, the percentage of energy consumed from fat has increased from 17% to 22% in the past 15 y, and rates of overweight are also increasing. Thus, we expect a trend toward more atherogenic risk profiles in the population. Our findings suggest that males in particular will experience this transition to a more atherogenic lifestyle differentially, depending on the energy status of their mothers during pregnancy. Optimizing the cardiovascular health of future generations of Filipinos may require efforts to attenuate the emergence of overweight among youth and adults while ensuring that mothers are well nourished during pregnancy.
ACKNOWLEDGMENTS
We thank the health departments of the cities of Cebu, Mandaue, and Lapu Lapu and Cebu Provincial Health for Region 7 for their assistance and for the participation of their clinics and staff in this research. The employees of the Office of Population Studies, Cebu, are warmly thanked for their support and significant contribution to data collection. Josephine Avila and Joseph Harvey Canete Cadungog assisted in blood sample collection. Lipid values were measured by Ngoc-Anh Le of the Emory University Medical School.
CWK and LSA were responsible for planning, funding, and implementing data collection. CWK was primarily responsible for data analysis and the writing of the manuscript. The authors had no conflicts of interest.
REFERENCES
1 From the Departments of Clinical Epidemiology and Biostatistics and Obstetrics and Gynecology, the Academic Medical Center, the University of Amsterdam, and the MRC Environmental Epidemiology Unit, the University of Southampton, United Kingdom.
2 Supported by the Medical Research Council, United Kingdom; the Diabetes Fonds Nederland; Wellbeing, United Kingdom; and the Academic Medical Center, Amsterdam. 3 Address reprint requests to TJ Roseboom, Department of Clinical Epidemiology and Biostatistics, Academic Medical Center, University of Amsterdam, PO Box 22700, 1100 DE Amsterdam, Netherlands. E-mail: t.j.roseboom{at}amc.uva.nl.
ABSTRACT
Background: Small body size at birth has been reported to be associated with an atherogenic lipid profile in humans, and animal experiments have shown that undernutrition during pregnancy permanently alters cholesterol metabolism in the offspring. There is no direct evidence in humans that maternal malnutrition during pregnancy affects the lipid profiles of the offspring.
Objectives: We assessed the effects of maternal malnutrition during specific periods of gestation on plasma lipid profiles in persons aged 50 y.
Design: This was a follow-up study of men and women born at term as singletons in a university hospital in Amsterdam between 1 November 1943 and 28 February 1947 around the time of a severe famine.
Results: Persons exposed to famine in early gestation had a more atherogenic lipid profile than did those who were not exposed to famine in utero. Their LDL-HDL cholesterol ratios were significantly higher (by 13.9%; 95% CI: 2.6–26.4%). Additionally, their plasma HDL-cholesterol and apolipoprotein A concentrations tended to be lower, and their plasma total cholesterol, LDL-cholesterol, and apolipoprotein B concentrations tended to be higher, although these differences were not statistically significant. The effect of famine was independent of size at birth and adult obesity.
Conclusions: An atherogenic lipid profile might be linked to a transition from poor maternal nutrition in early gestation to adequate nutrition later on. This suggests that maternal malnutrition during early gestation may program lipid metabolism without affecting size at birth.
Key Words: Cholesterol • lipid profile • famine • undernutrition • fetal growth • fetal origins • the Netherlands
INTRODUCTION
Small body size at birth has been reported to be associated with an atherogenic lipid profile (high plasma LDL-cholesterol and low plasma HDL-cholesterol concentrations). Some investigators found associations between low birth weight and low HDL-cholesterol or high plasma triacylglycerol concentrations (1–3); others found an association between short body length at birth or reduced abdominal circumference and elevated total cholesterol, LDL-cholesterol, and apolipoprotein B concentrations (4, 5).
Observations in guinea pigs and rats suggest that manipulations of maternal dietary intake during gestation permanently alter cholesterol synthesis and plasma cholesterol concentrations (6–8; JA Owens, A Sohlstrom, A Katsman, et al, unpublished observations, 1991). So far, the only study in humans on the effect of maternal nutrition during gestation on later cholesterol concentrations was performed in persons prenatally exposed to famine at the time of the 900-d Leningrad siege (1941–1944), and this study showed no significant effects (9).
We present the effects of prenatal undernutrition during specific periods in pregnancy on lipid profiles in adults born around the time of famine in the Netherlands (1944–1945). The Dutch famine was a 5-mo period of extreme malnutrition in the western part of the Netherlands that was clearly delineated in time. We showed previously that glucose tolerance in this group decreased after prenatal exposure to famine, especially in late or mid gestation (10), and that women exposed to famine in early gestation had a higher body mass index (BMI; in kg/m2) than did those not exposed to famine (11). We assessed the lipid profiles of adults exposed to the famine in utero during late, mid, or early gestation (exposed subjects); of those born in the year before the famine began; and of those conceived in the year after the famine (nonexposed subjects).
SUBJECTS AND METHODS
Selection procedures
All 5425 babies born in the Wilhelmina Gasthuis in Amsterdam between 1 November 1 1943 and 28 February 1947 were possible candidates for study. Most patients in this hospital were of lower-to-middle social classes, but little is known about the actual referral pattern during the period of our study. First, we excluded 349 babies who were stillborn or part of a multiple pregnancy. Second, we retrieved the medical records of all 1380 babies born between 1 November 1944 and 28 February 1946 who were potentially exposed to famine during gestation. Third, we retrieved the records of a random sample of 650 of the 1305 babies born in the year before that period (born between 1 November 1943 and 31 October 1944) and a random sample of 650 of the 2391 babies conceived in the year after that period (born between 1 March 1946 and 28 February 1947). Of these 2680 babies, 27 (1.0%) were excluded because their birth record was missing and 239 (8.9%) were excluded because they were born prematurely (gestational age at birth <259 d, either as computed from the date of the last menstrual period or as estimated by the obstetrician at the first prenatal visit and at the physical examination of the baby just after birth). In all, 2414 live-born singletons were included in the study.
The Bevolkingsregister of Amsterdam (population registry) traced 2155 (89%) of the 2414 infants included. Of these, 265 had died, 199 had emigrated from the Netherlands, and 164 did not allow the population registry to give us their address. Of the remaining 1527 infants, we asked 912 persons who lived in or close to Amsterdam to participate: 741 attended the clinic, and plasma lipid and lipoprotein concentrations were measured in fasting blood samples from 704 of them. Birth weights in this group of 704 subjects (mean birth weight: 3348 g) were not significantly different from those of the 1710 infants who were not included (mean birth weight: 3332 g; P adjusted for exposure = 0.3).
Exposure to famine
We defined the famine period according to the daily official food rations for the general population aged >21 y. The amount of protein, carbohydrate, and fat decreased more or less proportionately. The official rations reflected rather accurately the variation over time in the total amount of food available in the west of the Netherlands (12). In addition to the official rations, food came from other sources (eg, church organizations, central kitchens, and the black market), and the amount of food actually available to individuals was roughly twice as much as the official rations. Pregnant and lactating women were entitled to an extra 600 kcal (2520 kJ)/d, but at the peak of the famine, this additional energy could not always be provided. It is also likely that most women shared these extra supplies with their families. Therefore, the rations should be considered only a relative measure of nutritional intake for the population as a whole.
The official rations provided 1800 kcal (7560 kJ)/d in December 1943. This figure gradually decreased to 1400 kcal (5880 kJ)/d by October 1944, and to <1000 kcal (4200 kJ) by 26 November 1944. The energy content of the rations varied between 400 kcal (1680 kJ)/d and 800 kcal (3360 kJ)/d from December 1944 to April 1945 and rose to >1000 kcal (4200 kJ)/d by 12 May 1945, 1 wk after liberation by the Allied forces. In June 1945, rations provided >2000 kcal (8400 kJ)/d. Children younger than 1 y were relatively protected during the famine because their official daily rations always provided >1000 kcal (4200 kJ)/d (13).
We considered fetuses to have been exposed to famine if the average energy content of the daily rations for persons older than 21 y during any 13-wk period of gestation was <1000 kcal (4200 kJ)/d. Therefore, babies born between 7 January 1945 and 8 December 1945 were exposed in utero. We used 3 periods of 16 wk to differentiate between persons who were exposed in late gestation (born between 7 January 1945 and 28 April 1945), in mid gestation (born between 29 April 1945 and 18 August 1945), and in early gestation (born between 19 August 1945 and 8 December 1945).
Procedures
The medical birth records provided information about the mother, the course of the pregnancy, and the size of the baby at birth (for detailed information see reference 10). We also recorded the method of infant feeding at discharge, which took place 10 d after delivery, and classified it as exclusive breast-feeding, partial bottle-feeding, or exclusive bottle-feeding (14). Maternal weight gain in the third trimester was calculated as the difference in weight at the beginning and end of the third trimester divided by the duration of the time interval between the 2 measurements, multiplied by the duration of the trimester (13 wk). The socioeconomic status at birth was dichotomized into manual and nonmanual labor according to the occupation of the head of the family (15).
Total plasma cholesterol, HDL-cholesterol, LDL-cholesterol, triacylglycerol, apolipoprotein A, and apolipoprotein B concentrations were measured by standard enzymatic methods (16, 17). We measured height with a fixed stadiometer and weight with a Seca scale (Hamburg, Germany). All subjects were interviewed about their medical history, lifestyle, and use of medication. Current socioeconomic status was coded by using the International Socio-economic Index of occupational status according to the occupation of the participants or their partners, whichever was highest (18). Values ranged from 16 (low status) to 87.
Statistical methods
We calculated the differences between the lipid profiles of unexposed subjects and those exposed in late, mid, or early gestation. The variables HDL cholesterol, LDL-HDL ratio, serum triacylglycerol, and BMI had a skewed distribution and were log transformed before analysis. The results for these variables are given as geometric means ± SDs and the differences are given as relative differences expressed as percentages of the means of nonexposed participants. First, we used multiple linear regression analysis to adjust for sex. Second, we also adjusted for adult BMI, then for adult (current socioeconomic status, smoking status, and use of lipid-lowering medication) and maternal (age, parity, weight at last prenatal visit, and socioeconomic status at birth) characteristics. We computed 95% CIs. When we compared separately the 3 prenatally exposed groups with the nonexposed group, the P values were Bonferroni adjusted for multiple comparisons.
For a relatively large number of participants, information on maternal weight at the end of pregnancy, weight gain, or socioeconomic status at birth was missing. Therefore, when adjusting for maternal weight or weight gain, we set the value for that variable with missing values to the mean of the nonmissing values and entered an extra variable into the regression model with a value of 1 for those with missing values for that variable and a value of 0 for the rest. When adjusting for categorical variables (parity, socioeconomic status at birth, smoking status, and use of lipid-lowering medication), we added an extra category for those participants with missing values. SPSS (version 9.0.0; SPSS Inc, Chicago) was used for the analyses.
RESULTS
Of the 704 participants included in the study, 283 (40.2%) had been exposed to famine in utero (Table 1). Because it was more difficult to contact men, they were underrepresented in the groups that were exposed to famine in utero. Weight at the last prenatal visit was lower in mothers exposed to famine during late and mid pregnancy than in nonexposed mothers. Weight gain during the last trimester of pregnancy was lower in mothers exposed to famine during late pregnancy (those who gave birth during the famine) and higher in those exposed in mid pregnancy (those who conceived before the famine and gave birth after the famine) and early pregnancy (those who conceived during the famine). Babies exposed to famine during late or mid gestation were lighter and shorter and had smaller heads than did babies who were not exposed. The percentage of babies who were exclusively breast-fed in the first weeks after birth tended to be higher for those babies exposed during mid or early gestation. Adult BMI tended to be higher in those exposed to famine in early gestation, especially in women.
View this table:
TABLE 1. Maternal characteristics, birth outcomes, and adult characteristics according to time of prenatal exposure to famine
Participants exposed to famine in late or mid gestation tended to have lower total cholesterol concentrations but none of the lipid or lipoprotein concentrations were significantly different from those of the nonexposed participants (born before or conceived after the famine) (Table 2). Participants exposed to famine in early gestation, however, had a more atherogenic lipid profile than did those who were not exposed. After adjustment for sex, the subjects' LDL-HDL cholesterol ratios were significantly higher than those of nonexposed participants. Plasma HDL-cholesterol and apolipoprotein A (the structural apolipoprotein linked to HDL cholesterol) concentrations tended to be lower and total cholesterol, LDL-cholesterol, and apolipoprotein B (the structural apolipoprotein linked to LDL cholesterol) concentrations tended to be higher than in nonexposed participants. Triacylglycerol concentrations were not affected significantly.
View this table:
TABLE 2. Differences (and 95% CIs), adjusted for sex, between participants prenatally exposed to famine (in late, mid, or early gestation) and nonexposed participants (those born before or conceived after the famine)
The slightly higher percentage of exclusive breast-feeding in persons exposed to famine in mid and early gestation did not explain the observed effects of prenatal exposure to famine. We found, for example, after adjustment for the method of infant feeding that the LDL-HDL cholesterol ratio was 6.4% (15.6–2.8%) lower in those exposed to famine in mid gestation and 13.1% (2.4–23.8%; P = 0.017) lower in those exposed to famine in early gestation than in those not exposed. Because women exposed to famine in early gestation tended to have a higher BMI than did those exposed to famine in mid or late gestation, their more atherogenic lipid profile might also be explained by their higher incidence of obesity. However, adjustment for BMI reduced the magnitude of the effect only minimally. After adjusting for BMI, we found that the LDL-HDL cholesterol ratio differed, although not significantly, by 7.6% (7.0–24.5%) in men and by 12.4% (2.2–29.3%) in women exposed to famine in early gestation from that in nonexposed men or women, respectively. Further adjustment for adult characteristics (socioeconomic status, smoking status, and use of lipid-lowering medication) did not alter the results. The effects of prenatal exposure to famine on plasma total, LDL-, and HDL-cholesterol concentrations; the LDL-HDL cholesterol ratio; and apolipoprotein A and B concentrations were not significantly different for men and women.
Maternal weight at the last prenatal visit and maternal weight gain were not associated with any of the plasma lipid or lipoprotein concentrations (P for trend adjusted for sex >0.5), and adjustment for these maternal characteristics, therefore, did not alter the results appreciably. We also found that adjustment for other maternal characteristics (maternal age, parity, and socioeconomic status) as well as gestational age at birth were not associated with any of the plasma lipid or lipoprotein concentrations and hardly affected our results.
Birth weight was positively associated with apolipoprotein A concentration (Table 3). The ponderal index (in kg/m3) was positively associated with HDL cholesterol, apolipoprotein A, and total cholesterol. Additional adjustment for adult BMI did not alter these associations. Other measures at birth were not significantly associated with plasma lipid or lipoprotein concentrations. The effects of exposure to famine in utero on the plasma lipid profile were hardly affected, however, after adjustment for any body measure at birth.
View this table:
TABLE 3. Means of plasma lipid and lipoprotein concentrations by size at birth1
DISCUSSION
In this study we assessed the effect of maternal malnutrition during specific periods in gestation on the lipid profiles of 50-y-old persons. We found that men and women exposed to famine in early gestation had a more atherogenic plasma lipid profile than did those who were not exposed to famine in utero. Women in this group also tended to have the highest BMI, but adjustment for BMI altered the size of this effect only slightly. Persons exposed to famine in late or mid gestation tended to have lower total cholesterol concentrations, but this difference was not paralleled by differences in other lipid or lipoprotein concentrations. The effect of exposure to famine in early gestation on adult lipid profiles could not be explained by differences in maternal weight or weight gain, body size at birth, gestational age at birth, or method of infant feeding among the exposure groups.
The Dutch famine can be considered a unique "experiment of history" to study the effects of maternal malnutrition during different stages of gestation in humans. The famine, however, had a profound effect on the birth rate and early mortality. The number of births corresponding to conceptions at the peak of the famine—and consequently also to exposure during early gestation—was 50% lower than the number prefamine (15). Perinatal mortality and mortality in the first year after birth were highest in those who were born during the famine period (15). We cannot exclude potential selection effects of increased abortion rates in babies who were conceived during the famine, but we consider it unlikely that the differences in birth rate or early mortality fully explained our results. First, maternal characteristics that might relate to the biological or behavioral determinants of fertility (maternal age, parity, maternal weight, and socioeconomic status) were not associated with plasma lipid concentrations in the adult offspring. Second, early mortality rates were highest in those born during the famine (15), whereas we found the greatest effects on plasma lipid concentrations among those who were conceived during the famine and born after it (those exposed in early gestation).
A study in persons who were born in or around Leningrad at the time of the siege (1941–1944) showed that lipid and lipoprotein concentrations were not affected by prenatal undernutrition (9). The essentially different circumstances during the famines, however, did not allow a direct comparison between our findings and those of the Leningrad study. First, the Dutch famine was not only shorter but it was also preceded and followed by adequate nutrition; persons in Leningrad were also undernourished before the siege. Second, the rations for infants aged <1 y were found to be adequate throughout the famine (13), which indicates that babies born before or during the famine were not exposed in their first year of life. Finally, the Dutch people grew up in a period of increasing affluence, whereas the Russian standard of living remained relatively poor (19).
Our finding that persons exposed to famine in early gestation had a more atherogenic lipid profile seems to agree with the results from animal experiments. Observations in animals show that maternal undernutrition just before and throughout pregnancy permanently alters cholesterol metabolism, although plasma total cholesterol concentrations increased in guinea pigs (JA Owens, A Sohlstrom, A Katsman, et al, unpublished observations, 1991) and decreased in rats (8). This suggests that the effects of maternal diet during gestation are complex and may be different between species (20). It was also shown in rats that the composition of the maternal diet during pregnancy influences the activity of hepatic enzymes crucially involved in cholesterol metabolism in the offspring (6, 21). These results in animals suggest that the transition from nutritional deprivation in early gestation to nutritional adequacy later on has led to metabolic conflicts resulting in an altered cholesterol metabolism in persons conceived during the Dutch famine.
Our study showed for the first time in humans that maternal nutrition during early gestation can permanently influence the lipid profile in later life. Exposure to famine in early gestation did not affect body size at birth but led to a higher LDL-HDL cholesterol ratio in adult life. It confirmed findings from other studies in humans that maternal nutritional intake during pregnancy can have permanent effects on health in later life without affecting size at birth (10, 22). It is therefore difficult to predict the long-term effects of maternal starvation during gestation on the basis of its effects on size at birth. Furthermore, experiments in sheep have shown that different patterns of fetal growth can result in the same size at birth (23). This might explain how several studies in humans have reported that small size at birth is linked with a more atherogenic lipid profile in adult life (1–4), whereas we found that a high ponderal index at birth was associated with increased plasma total cholesterol concentrations in adult life.
Our findings may have important implications for public health. The nutritional experience of babies who were exposed to famine in early gestation may resemble that of babies in developing countries whose mothers are undernourished in early pregnancy and receive supplementation in the second half of pregnancy, but also of babies in developed countries whose mothers suffered from hyperemesis gravidarum or followed a strict diet just before conception or early in pregnancy. Furthermore, our findings suggest that the long-term effect that these imbalances in women's nutritional intakes during pregnancy have on the health of their children may be underestimated by the known associations between small size at birth and adult disease.
We showed previously in the same group of persons that those exposed to famine in late or mid gestation have a lower glucose tolerance than do those not exposed to famine (10), and that women exposed to famine in early gestation are more obese (11). We found that cholesterol metabolism was most affected in those exposed to famine in early gestation, and was largely independent of the effect of famine on obesity. This finding suggests that there are distinct sensitive periods during gestation for the programming of glucose and cholesterol metabolism. Animal experiments and prospective studies of mothers and their offspring are needed to unravel the mechanisms involved in nutritional programming.
ACKNOWLEDGMENTS
We are grateful for the willing cooperation of all participants. We especially thank Marjan Loep, Mieneke Vaas, Lydia Stolwijk, Yvonne Graafsma, Jokelies Knopper, Maartje de Ley, and the nurses at the Special Research Unit for collecting the data and at the Research Laboratory of Internal Medicine for analyzing the blood samples. We thank the Gemeentearchief and the Bevolkingsregister, Amsterdam, for their help in tracking the persons in our cohort.
REFERENCES
【摘要】
Objective- The similarities between valvular and vascular lesions suggest pathological initiation mediated through endothelium, but the role of hemodynamics in valvular endothelial biology is poorly understood.
Methods and Results- Monolayers of porcine aortic endothelial cells (PAECs) or porcine aortic valve endothelial cells (PAVECs) were exposed to 20 dyne/cm 2 steady laminar shear stress for 48 hours, with static cultures serving as controls. Multiple microarray comparisons were made using RNA from sheared and control batches of both cell types. More than 400 genes were significantly differentially expressed in each comparison group. The resulting profiles were validated at the transcription and protein level and expression patterns confirmed in vivo by immunohistochemistry. PAVECs were found to be less intrinsically inflammatory than PAECs, but both cell types expressed similar antioxidant and antiinflammatory genes in response to shear stress. PAVECs expressed more genes associated with chondrogenesis, whereas PAECs expressed osteogenic genes, and shear stress had a protective effect against calcification.
Conclusions- Transcriptional differences between PAVECs and PAECs highlight the valvular endothelial cell as a distinct organ system and suggest more attention needs to be given to valvular cells to further our understanding of similarities and differences between valvular and vascular pathology.
Aortic and aortic valve endothelial cell gene expression was compared in static and steady shear environments. Transcriptional profiles suggested that valvular endothelial cells are similar in some respects but distinct in other ways that may have important implications for the understanding of valvular pathology and therapeutic strategies.
【关键词】 aortic valve shear stress inflammation calcification endothelial cell
Introduction
Aortic valve disease is associated with significant mortality and morbidity and is a strong risk factor for additional cardiovascular events. 1,2 Valvular degeneration is characterized by the development of stenosis or insufficiency, and by the time it is clinically manifested, it is usually only treatable by prosthetic valve replacement. 3 Explants of diseased valves reveal a wide spectrum of pathology, including sclerotic and calcific lesions, thrombus formations, bacterial vegetations, and fractured matrix fibers. 4,5 Aortic valve disease was originally thought to be the result of the continuous barrage of hemodynamic and mechanical forces over time, but recent evidence suggests a much more active biological progression involving inflammation, oxidation, angiogenesis, calcification, and osteogenesis. 6-8
The vascular endothelium is a critical mediator of hemodynamic and humoral stimuli, and that endothelial inflammation and atherosclerosis occur preferentially at sites of disturbed or oscillatory flow. 9 Valvular endothelial dysfunction is also a hallmark of leaflet degeneration, and similarly characterized by the expression of proinflammatory adhesion receptors. 10,11 Interestingly, much of the aforementioned valvular pathology seems to occur preferentially on the aortic surface of the leaflet, which experiences a complex circulating flow that is different from the unidirectional flow on the ventricular side of the leaflet. This suggests that disturbed flow may play a causal role in the initiation of valvular pathology through activation of valvular endothelium, and likewise that valvular endothelium may be protected from dysfunctional activation by unidirectional flow, but as yet, no studies have been done to investigate this.
We have previously shown that valvular endothelial cells respond to shear stress by aligning perpendicular to the direction of flow, this in contrast to vascular endothelial cells, which align parallel to flow. 12 This alignment was associated with differences in focal adhesion arrangement and differential involvement of signal kinases, suggesting that these different endothelial cell types may interpret mechanical signals heterogeneously.
The objective of this article, therefore, is to characterize the similarities and differences between these endothelial populations through transcriptional profiles in static and shear conditions and identify putative mechanosensitive proteins that may be involved in the regulation of these differences. The resulting expression profiles suggest that valvular endothelium are similarly protected from oxidative stress, inflammatory stress, and calcification by shear stress, but intrinsic differences in the susceptibilities of these cells to the aforementioned pathologies may exist.
Materials and Methods
Cell Culture and Shear Exposure
Porcine aortic valve endothelial cells (PAVECs) and porcine aortic endothelial cells (PAECs) were isolated from tissue obtained at a local slaughterhouse and cultured as described previously. 12 Passage 5 monolayers of either PAVECs or PAECs were grown to confluence (48 hours) on glass slides coated with Collagen I (Becton-Dickenson; rat tail; 50 µg/mL). Slides were then placed in a parallel plate flow system and subjected to 20 dyne/cm 2 steady shear stress for 48 hours, whereas statically cultured slides served as controls (justification in online supplement, available at http://atvb.ahajournals.org).
Transcript Profiling Studies and Data Analyses
Four experimental groups were created for gene expression comparison: group 1 included PAVEC shear versus PAVEC static; group 2 included PAEC shear versus static; group 3 included PAVEC static versus PAEC static; and group 4, PAVEC shear versus PAEC shear (Figure I, available online at http://atvb.ahajournals.org). After shear or static conditions, total RNA was extracted from cell pellets pooled from 2 identical and independent experiments using the RNeasy mini column (Qiagen). Ten micrograms total RNA pools from 2 different experimental conditions were independently and randomly labeled with either Cy3 or Cy5 fluorophores using the Agilent Fluorescent Direct Label kit (Agilent Technologies). Labeled RNA pools were then competitively hybridized to Agilent Human 1 cDNA microarray slides (Agilent #G4100A), which contain 2 identical array areas. The above procedures were repeated 3 times using a different cell isolation batch each time, giving 6 arrays in an n=3 (biological replicates), n=2 (technical replicates) arrangement for each comparison group, for a total of 24 arrays. Fluorescence intensities of each hybridized spot were determined by using the Agilent Array Scanner and the Agilent Feature Extraction Software. 13 Two different statistical methods were used to determine differentially expressed genes: mixed ANOVA (MxANOVA) and significance analysis of microarrays (SAM). 14,15 The list of differentially expressed genes in each comparison group was analyzed for statistically enriched or depleted biological classifications using the GoMiner database engine. 16 Additional details are provided in the methods supplement.
Quantitative Real-Time RT-PCR, Western Blots, and Immunostaining
The expression trends of a panel of genes were confirmed at the transcript (quantitative real-time RT-PCR [QRT-PCR]) and protein levels (Western blot) using additional samples not used in the microarrays. Additional normal adult porcine valve leaflets and aortic wall tissue were excised, fixed in 10% neutral buffered formalin, paraffin embedded, and sectioned at 5 µm. Immunohistochemistry was performed for selected genes as indicated in the results using fluorescence-based detection methods (see online supplement).
Results
Transcriptional Response Differences Between Endothelial Cells
Microarray hybridizations were performed using sheared and static control samples of PAVECs and PAECs and analyzed statistically using MxANOVA. Shear exposure upregulated 545 and 324 genes in PAVECs and PAECs, respectively, and of those, 260 and 62 genes were unique to the valvular and vascular cell types. On the other hand, shear exposure downregulated expression of 311 and 157 genes in PAVECs and PAECs, respectively, with 184 and 37 genes unique to each cell type. A total of 236 and 615 genes were expressed more in PAVECs compared with PAECs in static culture and in response to shear stress, respectively, with 180 and 537 genes expressed more in PAECs under those same conditions. A comparison of the MxANOVA method with the permutation based system (SAM) and single ANOVA shows that the MxANOVA calls genes at a dramatically higher level of sensitivity compared with single ANOVA and maintained &75% fidelity with SAM (Table III, available online at http://atvb.ahajournals.org). Because SAM returned 3 x more genes than the MxANOVA that were not present in both, we decided that the MxANOVA results were a more conservative estimate of the true called genes. Complete lists of significantly changed genes from each comparison group as determined by MxANOVA are available in the GEO database.
We next validated some of the microarray data at the mRNA and protein levels by QRT-PCR and Western blots, respectively ( Figure 1 ). Based on the microarray data, we initially chose 3 known shear-regulated genes. Bone morphogenic protein 4 (BMP-4) downregulation in response to shear as shown by the microarray analysis was confirmed in both cell types and in each method ( P <0.05; Figure 1 A). Cytochrome P450 peptide 1A1 (CYP1A1) upregulation found in the microarray analysis (&10-fold) by shear was also confirmed by QRT-PCR in both cell types ( P <0.05), although the protein level was upregulated to a much smaller degree (1.6-fold upregulation; Figure 1 B). As shown in Figure 1 C, caveolin-1 (CAV-1) gene transcript level determined by the microarray was not affected by shear in PAECs but was downregulated by shear in PAVECs. However, QRT-PCR and Western blot studies showed that shear exposure decreased the mRNA and protein expression levels by 2.6 and 1.7-fold, respectively ( P <0.05; Figure 1 C).
Figure 1. Confirmation of microarray data by real-time PCR and Western blot. NS denotes not significantly changed, whereas all other fold expressions are significantly different from matched controls ( P <0.05). Boxes display Western blots. A indicates PAECs; V, PAVECs; +, shear; -, static.
We further confirmed the array data by Western blots for 2 additional genes: periostin (POSTN) and cadherin 11 (CAD11), in both cell types exposed to static and shear conditions ( Figure 2 ). In total, the microarray trends were confirmed in 9 of 10 cases (BMP-4, CYP1A1, POSTN, and CAD11 in both cell types and CAV1 in PAVECs) but not in 1 case (CAV1 in PAECs) by QRT-PCR and Western blotting ( Figures 1 and 2 ).
Figure 2. Divergent expression of mechanosensitive proteins involved in differentiation by aortic and aortic valve endothelial cells. A, Two POSTN isoforms (75 and 77kDa) are expressed in PAECs, whereas PAVECs express the 75-kDa form only. B, CAD11 expression is significantly higher in PAVECs than PAECs. C, POSTN is downregulated by shear in PAECs. D, CAD11 is downregulated by shear in PAVECs. Actin blots are used as loading controls. A indicates aortic endothelial cells; V, aortic valve endothelial cells; +, shear; -, static. *Denotes significance P <0.05.
In addition, we validated the in vitro results in intact aortic valve (AV) tissues of normal pigs by immunohistochemical staining. We first used endothelial NO synthase (eNOS) as a positive control for endothelial cells and shear responses, although its transcript levels in our array studies could not be analyzed because of poor quality of some of the spots. The eNOS is expressed in the endothelium of the aorta and both sides of the aortic valve with higher expression at the ventricularis surface, where shear stress is believed to be more stable and unidirectional than that of fibrosa. BMP-4 was detected on the endothelium of normal aortas and valves ( Figure 3 ). Expression was greater on the fibrosa side of the leaflet. Vascular cell adhesion molecule 1 (VCAM1) was not detected on normal aorta or aortic valves (data not shown). These results are similar to our previous report with aortic endothelial cells in response to laminar or oscillatory shear stresses. 17
Figure 3. Confirmation of protein expression in vivo through immunohistochemistry. Panels A and B, F and G, K and L, and P and Q are from normal porcine ascending aortas, and the rest are of normal porcine aortic valve leaflets. Panels C, H, M, and R are the whole leaflet, whereas D, I, N, and S magnify the fibrosa side (f), and E, J, O, and T magnify the ventricularis (v) endothelium. Endothelial cells are marked with arrows. Panels A through E demonstrate eNOS expression; F through J, BMP-4 expression; K through O, Cad11 expression; and P through T, POSTN expression (green color). DAPI staining (blue) denotes cell nuclei. Bars=200 µm (B, C, G, and H) or 40 µm.
Analysis of Biological Classifications
The coordinated overexpression or underexpression of a group of genes in an ontological category may indicate certain functional responses to shear stress or differences between cell types. Biological classifications related to several important known endothelial functions using the GoMiner program were significantly changed ( Table 1 ). For example, groups of genes related to oxioreductase activity, cell proliferation, apoptosis, cell migration, and cell signaling were significantly changed in PAVECs and PAECs in response to shear stress (groups 1 and 2) and differentially expressed between PAVECs and PAECs in static and shear flow conditions (groups 3 and 4). Surprisingly, there were also many biological classifications that were significantly changed in these groups related to developmental and differentiation events, including morphogenesis, angiogenesis, skeletal and muscle development. A complete list of the changed gene categories for each comparison group is available in the GEO database. Several additional classes of functional groups related to endothelial physiology and pathology were identified, including the genes involved in skeletal and mesenchymal development.
TABLE 1. Prominent Endothelial Gene Classifications Identified by GOMiner *
TABLE 1. Continued
Expression of Antioxidant and Antiinflammatory Gene Transcripts in Aortic Valve Endothelium
As shown in Table 2, PAVECs expressed a similar number of antioxidant genes but less inflammatory ones compared with PAECs. PAVECs expressed 5 of 6 antiinflammatory genes to a greater degree than PAECs and 10 of 13 proinflammatory genes to a lesser degree than PAECs. PAECs expressed interleukins (IL-1A, IL-8), connexin 43 (GJA1), activated leukocyte adhesion molecule, BMP-4, and type III collagen (COL3A1) to a greater degree than PAVECs, all of which have been shown to contribute to vascular endothelial cell inflammatory atherosclerosis (see references in Table IV, available online at http://atvb.ahajournals.org). Expression of prooxidant and antioxidant genes was more evenly distributed between the 2 cell types, with 9 antioxidant genes expressed to a greater degree in PAECs and 8 in PAVECs. Antioxidant and antiinflammatory genes were uniformly upregulated in response to shear stress in both cell types, including peroxiredoxins (PRDX1 and PRDX2) superoxide dismutase (SOD2), and cytochromes (CYP1A1, CYP1B1). Shear stress regulated proinflammatory and prooxidant genes in a more complex and heterogeneous manner.
TABLE 2. Differential Genes Related to Oxidation and Inflammation *
Aortic and Aortic Valve Endothelial Cells Differentially Regulate Chondro/Osteogenic Genes
The analysis of significantly differentially expressed genes revealed an unexpected number of genes associated with chondrogenesis and osteogenesis ( Table 2 ). Surprisingly, PAECs expressed 5 of 6 proosteogenic genes to a greater degree than PAVECs, whereas PAVECs expressed 6 of 8 chondrogenic genes to a greater degree 100-fold greater 400-fold greater than PAVECs ( Table 3 ). These POSTN and CAD11 array results were confirmed by Western blots using cell lysates. As shown in Figure 2, the 2 osteogenic proteins were expressed almost exclusively in 1 endothelial cell type. PAECs express 2 POSTN isoforms (75- and 77-kDa bands), whereas only the 75-kDa form was expressed in PAVECs ( Figure 2 A). The specificity of both bands has been confirmed by antigen competition blotting (Figure II, available online at http://atvb.ahajournals.org). Shear exposure downregulated the 77-kDa form but not the 75-kDa form, suggesting that the 77-kDa form corresponds to the shear-sensitive transcripts ( Figure 2 C). In contrast, CAD11 protein was highly expressed in PAVECs but not in PAECs ( Figure 2 B). The downregulation of CAD11 protein by shear in PAVEC was also confirmed ( Figure 2 D). These results are consistent with the notion that shear protects both endothelial cell types from chondro/osteogenic differentiation ( Figure 2 ). The microarray and Western blot results of PAECs and PAVECs were further validated in vivo by immunostaining of normal porcine aorta and aortic valve ( Figure 3 ). POSTN protein was easily detected in aortic endothelium ( Figure 3P and 3 Q) but not in valvular endothelium ( Figure 3R through 3 T). In contrast, CAD11 protein expression in aortic endothelium was not detectable ( Figure 3K and 3 L), whereas it was easily detected in valvular endothelium of fibrosa and ventricularis surfaces ( Figure 3M through 3 O).
TABLE 3. Differentially Expressed Genes Related to Calcification *
Discussion
Although it is well recognized that vascular endothelial cells are critical mediators of vascular function and dysfunction through response to changes in hemodynamics, a similar role for valvular endothelial cells has yet to be determined. The differences between embryonic valvular and vascular endothelial cells suggest that these cell types may represent unique phenotypes in adult tissues. Using unamplified RNA samples pooled from multiple experiments and 4 separate microarray comparisons, we generated a comprehensive picture of valvular and vascular endothelial transcriptional profiles in static and steady shear stress conditions. Many studies have used microarray analysis to investigate the response of vascular endothelial cells to steady shear stress. 18,19 A study by Farivar et al compared the transcriptional profile of valvular and vascular endothelial cells in static conditions and identified several genes with putative expression in 1 cell type or the other. 20 However, their study did not permit a statistical analysis of the generated expression profiles. A recent study by Simmons et al compared transcriptional profiles from endothelial cells isolated from both sides of the aortic valve leaflet and found side dependent differences in gene expression that may be implicated in valvular pathogenesis. 21 Our comparison system builds on these studies by enabling the profiling of both these cell types and their responses to shear stress using cells isolated from the same normal animals in the same statistically motivated experimental design. The transcriptional profiles coupled with the ontological profiles suggest that, like vascular endothelial cells, unidirectional shear stress inhibits oxidative and inflammatory responses of valvular endothelial cells. Our comparison group 2 (PAECs shear versus static) had many of the same genes significantly regulated as other published vascular microarray studies, including upregulation of antioxidants CYP1A1, PRDX1, and SOD2 and downregulation of a proinflammatory mediator BMP-4. 17-19 In addition, shear stress also protects both endothelial cell types from chondro/osteogenic differentiation, which is in accordance with the ventricular endothelial protection from this differentiation suggested by the study by Simmons et al. 21 We also found novel differences between the endothelial cell types including the increased expression of chondrogenic factors in PAVECs but osteogenic factors in PAECs, The expression trends of genes implicated in each of these functional areas were confirmed at the transcript and protein level, and conforming expression patterns were found in vivo in adult tissue. These data point to important valvular endothelial functions that may not be mimicked by vascular endothelium and suggest that valvular endothelium acts as a distinct organ system.
The similar regulation of oxidative and inflammatory genes by PAVECs and PAECs by shear stress suggests that hemodynamics may play a similarly important role in the pathogenesis of valvular diseases as it does in the vasculature. The hemodynamics experienced by the ventricular and aortic surfaces of aortic leaflets are distinctly different, with changes in flow direction on the aortic surface similar to those seen in vascular bifurcations. 22 Mapping of the TIE1 promoter in mice localized to regions of vasculature that were atheroprone, including at bifurcations and on the aortic surface of the aortic valve, 23 suggesting that both of these regions may be similarly susceptible to inflammatory disease. Early valvular lesions are primarily localized to the aortic (fibrosa) surface and characterized by endothelial expression of adhesion receptors such as VCAM1, intercellular adhesion molecule, and E-selectin. 10,11 The more abundant expression of proinflammatory genes by PAECs compared with PAVECs (and less abundant expression of antiinflammatory genes by PAECs) suggests that aortic valve endothelial cells may be more intrinsically antiinflammatory than aortic endothelial cells.
More advanced lesions on both aortic valve leaflets and aortas are characterized by calcification of underlying tissue. Although calcification of valvular interstitial cells and vascular smooth muscle cells is mediated by apoptosis and enhanced with transforming growth factor-ß stimulation, 24,25 the role of endothelial cells in this process is not completely understood. Increasing evidence suggests that endothelial cells may play a critical role in regulating this process. Endothelial cells in calcification-susceptible regions express calcification stimulating factors such as BMP-4 and reduced expression of inhibitory factors such as osteoprotegrin and osteopontin. 21,26 We show shear stress downregulates initiators of calcification such as BMP-4 in vascular and valvular endothelial cells, as well as other genes, suggesting that hemodynamics may also mediate these events. We also discovered the almost exclusive expression of CAD11 in PAVECs and POSTN in PAECs, and both of these genes were significantly downregulated by steady shear stress. CAD11 is a member of the cadherin adhesion receptor family and is expressed in a variety of mesenchymal cells. 27 Transfection of embryonic stem cells with CAD11 directly induces differentiation to chondrogenic and osteogenic phenotypes mediated by cell-cell contacts. 28 Laminar shear inhibition of CAD11 may be critical for inhibition of calcification in valvular tissue. POSTN is a fibrous extracellular matrix protein with repeating fasciclin domains and 4 known isoforms, each of which appears to be a positive regulator of osteogenesis in preosteoblastic cells. 29 Interestingly, POSTN is initially expressed in the mesenchyme of developing valve cushions, yet its expression is reduced in adult leaflets in comparison to the aorta by birth. 30 We found 2 POSTN isoforms expressed in both endothelial cells, the 77-kDa isoform conforming to the microarray data ( Figure 2; Figure II). We detected POSTN only in the aortic endothelium in vivo but not in the aortic valve endothelium. POSTN is dramatically increased during vascular injury and dilated cardiomyopathy and enhanced by BMP stimulation, 31,32 suggesting its role in pathogenic cardiovascular remodeling. The downregulation of POSTN by shear stress in PAECs suggests that hemodynamics may also play a role in the regulation of POSTN in a calcific resistive manner in vascular tissue. It was also interesting to note that the preponderance of calcification genes expressed by PAVECs were chondrogenic, whereas osteogenic in PAECs. There have been rare occurrences of complete transformation of aortic valves into cartilage in humans, 33,34 and cartilaginous tissue has been found in explants of bioprosthetic valves. 35 The authors of these reports suggested that the formation of this cartilage was associated with tissue stress levels and represented a repair of ossified tissue. The interaction between valvular endothelial cells and interstitial cells leading to this transformation are unclear but suggest a unique result of pathogenic hemodynamic stimuli in valvular tissue.
In summary, transcriptional profile comparisons of valvular and vascular endothelial cells in different hemodynamic environments suggest that these cell types are distinctly different with respect to important biological functions but respond similarly to unidirectional shear stress to maintain a mature quiescent phenotype. The results of these studies raise important questions about endothelial phenotypes and the role hemodynamics play in regulating them, and the presented data provide a rich foundation from which more detailed investigation can progress.
Acknowledgments
This research was funded by the American Heart Association Southeast Affiliate (J.B., predoctoral fellowship award 0315103B; G.S., postdoctoral fellowship), the Georgia Tech/Emory Center for the Engineering of Living Tissues, NSF Grant EEC-9731643 (R.N.), and the grant HL71014 from the National Institutes of Health (H.J.). We gratefully acknowledge the assistance of Guoshen Wang and Wenli Wang and the rest of the staff at the Cardiovascular Research Institute of the Morehouse School of Medicine in processing the microarrays.
【参考文献】
Otto CM, Lind BK, Kitzman DW, Gersh BJ, Siscovick DS. Association of aortic-valve sclerosis with cardiovascular mortality and morbidity in the elderly. N Engl J Med. 1999; 341: 142-147.
Hsu SY, Hsieh IC, Chang SH, Wen MS, Hung KC. Aortic valve sclerosis is an echocardiographic indicator of significant coronary disease in patients undergoing diagnostic coronary angiography. Int J Clin Pract. 2005; 59: 72-77.
Baxley WA. Aortic valve disease. Curr Opin Cardiol. 1994; 9: 152-157.
Otto CM, Kuusisto J, Reichenbach DD, Gown AM, O?Brien KD. Characterization of the early lesion of ?degenerative? valvular aortic stenosis. Histological and immunohistochemical studies. Circulation. 1994; 90: 844-853.
Harasaki H, Hanano H, Tanaka J, Tokunaga K, Torisu M. Surface structure of the human cardiac valve. A comparative study of normal and diseased valves. J Cardiovasc Surg (Torino). 1978; 19: 281-290.
Tanaka K, Sata M, Fukuda D, Suematsu Y, Motomura N, Takamoto S, Hirata Y, Nagai R. Age-associated aortic stenosis in apolipoprotein E-deficient mice. J Am Coll Cardiol. 2005; 46: 134-141.
Rajamannan NM, Subramaniam M, Rickard D, Stock SR, Donovan J, Springett M, Orszulak T, Fullerton DA, Tajik AJ, Bonow RO, Spelsberg T. Human aortic valve calcification is associated with an osteoblast phenotype. Circulation. 2003; 107: 2181-2184.
Lee YS, Chou YY. Pathogenetic mechanism of senile calcific aortic stenosis: the role of apoptosis. Chin Med J (Engl). 1998; 111: 934-939.
Chappell DC, Varner SE, Nerem RM, Medford RM, Alexander RW. Oscillatory shear stress stimulates adhesion molecule expression in cultured human endothelium. Circ Res. 1998; 82: 532-539.
Ghaisas NK, Foley JB, O?Briain DS, Crean P, Kelleher D, Walsh M. Adhesion molecules in nonrheumatic aortic valve disease: endothelial expression, serum levels and effects of valve replacement. J Am Coll Cardiol. 2000; 36: 2257-2262.
Muller AM, Cronen C, Kupferwasser LI, Oelert H, Muller KM, Kirkpatrick CJ. Expression of endothelial cell adhesion molecules on heart valves: upregulation in degeneration as well as acute endocarditis. J Pathol. 2000; 191: 54-60. <a href="/cgi/external_ref?access_num=10.1002/(SICI)1096-9896(200005)191:1
Butcher JT, Penrod AM, Garcia AJ, Nerem RM. Unique morphology and focal adhesion development of valvular endothelial cells in static and fluid flow environments. Arterioscler Thromb Vasc Biol. 2004; 24: 1429-1434.
Agilent G2566AA Feature Extraction Software User Manual. Palo Alto, Calif: Agilent Technologies; 2001.
Zhao Y, Chen BP, Miao H, Yuan S, Li YS, Hu Y, Rocke DM, Chien S. Improved significance test for DNA microarray data: temporal effects of shear stress on endothelial genes. Physiol Genomics. 2002; 12: 1-11.
Tusher VG, Tibshirani R, Chu G. Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci U S A. 2001; 98: 5116-5121.
Zeeberg BR, Feng W, Wang G, Wang MD, Fojo AT, Sunshine M, Narasimhan S, Kane DW, Reinhold WC, Lababidi S, Bussey KJ, Riss J, Barrett JC, Weinstein JN. GoMiner: a resource for biological interpretation of genomic and proteomic data. Genome Biol. 2003; 4: R28.
Sorescu GP, Sykes M, Weiss D, Platt MO, Saha A, Hwang J, Boyd N, Boo YC, Vega JD, Taylor WR, Jo H. Bone morphogenic protein 4 produced in endothelial cells by oscillatory shear stress stimulates an inflammatory response. J Biol Chem. 2003; 278: 31128-31135.
McCormick SM, Eskin SG, McIntire LV, Teng CL, Lu CM, Russell CG, Chittur KK. DNA microarray reveals changes in gene expression of shear stressed human umbilical vein endothelial cells. Proc Natl Acad Sci U S A. 2001; 98: 8955-8960.
Chen BP, Li YS, Zhao Y, Chen KD, Li S, Lao J, Yuan S, Shyy JY, Chien S. DNA microarray analysis of gene expression in endothelial cells in response to 24-h shear stress. Physiol Genomics. 2001; 7: 55-63.
Farivar RS, Cohn LH, Soltesz EG, Mihaljevic T, Rawn JD, Byrne JG. Transcriptional profiling and growth kinetics of endothelium reveals differences between cells derived from porcine aorta versus aortic valve. Eur J Cardiothorac Surg. 2003; 24: 527-534.
Simmons CA, Grant GR, Manduchi E, Davies PF. Spatial heterogeneity of endothelial phenotypes correlates with side-specific vulnerability to calcification in normal porcine aortic valves. Circ Res. 2005; 96: 792-799.
Nicosia MA, Cochran RP, Einstein DR, Rutland CJ, Kunzelman KS. A coupled fluid-structure finite element model of the aortic valve and root. J Heart Valve Dis. 2003; 12: 781-789.
Porat RM, Grunewald M, Globerman A, Itin A, Barshtein G, Alhonen L, Alitalo K, Keshet E. Specific induction of tie1 promoter by disturbed flow in atherosclerosis-prone vascular niches and flow-obstructing pathologies. Circ Res. 2004; 94: 394-401.
Jian B, Narula N, Li QY, Mohler ER III, Levy RJ. Progression of aortic valve stenosis: TGF-beta1 is present in calcified aortic valve cusps and promotes aortic valve interstitial cell calcification via apoptosis. Ann Thorac Surg. 2003; 75: 457-465.
Simionescu A, Philips K, Vyavahare N. Elastin-derived peptides and TGF-beta1 induce osteogenic responses in smooth muscle cells. Biochem Biophys Res Commun. 2005; 334: 524-532
Seipelt RG, Backer CL, Mavroudis C, Stellmach V, Cornwell M, Seipelt IM, Schoendube FA, Crawford SE. Osteopontin expression and adventitial angiogenesis induced by local vascular endothelial growth factor 165 reduces experimental aortic calcification. J Thorac Cardiovasc Surg. 2005; 129: 773-781.
Simonneau L, Kitagawa M, Suzuki S, Thiery JP. Cadherin 11 expression marks the mesenchymal phenotype: towards new functions for cadherins? Cell Adhes Commun. 1995; 3: 115-130.
Kii I, Amizuka N, Shimomura J, Saga Y, Kudo A. Cell-cell interaction mediated by cadherin-11 directly regulates the differentiation of mesenchymal cells into the cells of the osteo-lineage and the chondro-lineage. J Bone Miner Res. 2004; 19: 1840-1849.
Litvin J, Selim AH, Montgomery MO, Lehmann K, Rico MC, Devlin H, Bednarik DP, Safadi FF. Expression and function of periostin-isoforms in bone. J Cell Biochem. 2004; 92: 1044-1061.
Norris RA, Kern CB, Wessels A, Moralez EI, Markwald RR, Mjaatvedt CH. Identification and detection of the periostin gene in cardiac development. Anat Rec A Discov Mol Cell Evol Biol. 2004; 281: 1227-1233.
Lindner V, Wang Q, Conley BA, Friesel RE, Vary CP. Vascular injury induces expression of periostin: implications for vascular cell differentiation and migration. Arterioscler Thromb Vasc Biol. 2005; 25: 77-83.
Katsuragi N, Morishita R, Nakamura N, Ochiai T, Taniyama Y, Hasegawa Y, Kawashima K, Kaneda Y, Ogihara T, Sugimura K. Periostin as a novel factor responsible for ventricular dilation. Circulation. 2004; 110: 1806-1813.
Groom DA, Starke WR. Cartilaginous metaplasia in calcific aortic valve disease. Am J Clin Pathol. 1990; 93: 809-812.
Seemayer TA, Thelmo WL, Morin J. Cartilaginous transformation of the aortic valve. Am J Clin Pathol. 1973; 60: 616-620.
Arbustini E, Jones M, Ferrans VJ. Formation of cartilage in bioprosthetic cardiac valves implanted in sheep: a morphologic study. Am J Cardiol. 1983; 52: 632-636.
作者单位：Jonathan T. Butcher; Sarah Tressel; Tiffany Johnson; Debi Turner; George Sorescu; Hanjoong Jo; Robert M. NeremFrom the Petit Institute for Bioengineering and Bioscience (J.T.B., T.J., R.M.N.) and Woodruff School of Mechanical Engineering (J.T.B., R.M.N.), Georgia Institute of Technology, Atlanta; Co
【摘要】
Bladder cancer transformation and immortalization require the inactivation of key regulatory genes, including TP53. Genotyping of a large cohort of bladder cancer patients (n = 256) using the TP53 GeneChip showed mutations in 103 cases (40.2%), the majority of them mapping to the DNA-binding core domain. TP53 mutation status was significantly associated with tumor stage (P = 0.0001) and overall survival for patients with advanced disease (P = 0.01). Transcript profiling using oligonucleotide arrays was performed on a subset of these cases (n = 46). Supervised analyses identified genes differentially expressed between invasive bladder tumors with wild-type (n = 24) and mutated TP53 (n = 22). Pathway analyses of top-ranked genes supported the central role of TP53 in the functional network of such gene patterns. A proteomic strategy using reverse phase arrays with protein extracts of bladder cancer cell lines validated the association of identified differentially expressed genes, such as gelsolin, to TP53 status. Immunohistochemistry on tissue microarrays (n = 294) revealed that gelsolin was associated with tumor stage and overall survival, correlating positively with TP53 status in a subset of these patients. This study further reveals that TP53 mutations are frequent events in bladder cancer progression and identified gelsolin related to TP53 status, tumor staging, and clinical outcome by independent high-throughput strategies.
--------------------------------------------------------------------------------
The nuclear protein Tp53 plays an essential role in the regulation of cell cycle and apoptosis, contributing to transformation and malignancy.1 Tp53 is a DNA-binding protein containing transcription, DNA binding, and oligomerization activation domains, functioning as a tumor suppressor.2,3 Mutants of TP53 that frequently occur in a number of different human cancers, including bladder cancer, fail to bind the consensus DNA-binding site and hence cause the loss of tumor suppressor activity.4 Alterations of the TP53 gene occurs both as germline mutations, such as in cancer-prone families with Li-Fraumeni syndrome, or somatic mutations in diverse human malignancies.5
TP53 is one of the proteins better characterized in cancer research with reported targets, regulators, and binding proteins. For example, targets regulated by TP53 include cell-cycle genes, such as p21, and anti-apoptotic genes, such as bax. Regulators of TP53 include ataxia telangiectasia mutated (ATM) and Chk2, whereas Abl1 and the adenomatous polyposis gene (APC) are among known binding TP53 proteins.6-8 However, little is known of the differential gene expression patterns of human tumors presenting wild-type TP53 compared with those with a mutant protein. With the advent of microarray technologies, characterization of TP53 sequences and gene expression profiles associated with TP53 status are available in a high-throughput manner. Bladder cancer is one of the tumors in which TP53 is altered with a high frequency, mutation rates being 40% in advanced stages of the disease.9-12 The present study was designed to identify targets that would differentiate patients presenting advanced disease with wild-type versus mutant TP53 (Figure 1 ). Gelsolin was selected as one on the genes located to chromosome 9q33, a frequently mutated locus in bladder cancer.9,13 Two proteomic approaches were used to evaluate the link of gelsolin with tumor progression and TP53 status. Immunohistochemical analyses on tissue arrays containing well-annotated bladder tumors and known TP53 status served to associate the expression of gelsolin with TP53, tumor stage, and survival. The differential expression of gelsolin among several bladder cancer cell lines of known TP53 alterations was evaluated by custom-made reverse phase arrays.
Figure 1. Experimental design. TP53 genotyping was performed on 256 bladder tumors using the TP53 sequencing arrays. Gene expression analyses using the U133A array were performed in a subset of 46 bladder tissues to identify targets differentially expressed in bladder cancer regarding their TP53 status (TP53 wild type, n = 24; and TP53 mutated, n = 22). Supervised methods identified 149 probes differentially expressed between those cases with either wild-type or mutated TP53. Two types of validation studies of the association of molecular profiles with TP53 status were performed. Immunohistochemical patterns were analyzed on tissue arrays containing 294 tumors, a subset of them of known TP53 and clinical outcome status. Proteomic reverse phase arrays were also performed on protein extracts of bladder cancer cell lines of known TP53 status.
【关键词】 proteomic profiles association gelsolin progression
Materials and Methods
TP53 Sequencing Analyses
DNA Extraction and Tissue Samples
Total DNA was extracted using a nonorganic method (Oncor, Gaithersburg, MD). Macrodissection of OCT-embedded tissue blocks was performed to ensure a minimum of 75% tumor cells.13 DNA quality was evaluated based on 260/280 ratios of absorbances. Specimens were collected under institutional review board approval. These tumors comprised 10 pTa, 32 pT1, 22 pT2, 175 pT3, and 15 pT4 specimens from patients with bladder cancer.
TP53 oligonucleotide array assay (GeneChip p53; Affymetrix, Santa Clara, CA). Purified DNA (100 ng) was subjected to multiplex-polymerase chain reactions (PCRs) amplifying exons 2 to11 simultaneously, using reagents supplied by the manufacturer (Affymetrix). Apart from the DNA, each PCR reaction contained 10 U of AmpliTaq Gold, PCR buffer II, 2.5 mmol/L MgCl2, 5 µl of the primer set, and 0.2 mmol/L each dNTP. The reaction was performed in a final volume of 100 µl. The PCR profile consisted of an initial heating at 95??C for 10 minutes, followed by 35 cycles of 95??C for 30 seconds, 60??C for 30 seconds, and 72??C for 45 seconds, with a final extension step at 72??C for 10 minutes. Forty-five µl of the PCR product was then fragmented by the addition of 0.25 U of fragmentation reagent (DNase I in 10 mmol/L Tris-HCl, pH 7.5, 10 mmol/L CaCl2, 10 mmol/L MgCl2, and 500 ml/L glycerol) along with 2.5 U of calf intestine alkaline phosphatase, 0.4 mmol/L ethylenediaminetetraacetic acid, and 0.5 mol/L Tris-acetate, and incubation at 25??C for 15 minutes, followed by heat inactivation at 95??C for 10 minutes. For labeling, 50 µl of the fragmented DNA was incubated at 37??C for 45 minutes with 10 µmol/L fluorescein-N6-dideoxy-ATP, 25 U of terminal transferase, and TdTase buffer in a total volume of 100 µl, followed by heat inactivation at 95??C for 10 minutes. The sample was hybridized to the chip in a volume of 0.5 ml containing 6x sodium chloride/sodium phosphate/EDTA (SSPE) buffer, 0.5 ml/L Triton X-100, 1 mg of acetylated bovine serum albumin, 2 nmol/L control oligonucleotide, and the labeled DNA sample. Hybridization was done in an oven with constant agitation at 45??C for 30 minutes. The chip was then washed on the wash station four times with 3x SSPE containing 0.05 ml/L Triton X-100. After washing, GeneChips were read using a confocal laser scanner, and data were aligned and analyzed. A reference from the control DNA supplied was also analyzed. This reference belonged to the same PCR round and was measured on the same batch of chips.14,15
Gene Profiling Using U133A GeneChips
Tissue Samples and RNA Extraction
Tumors belonging to patients with invasive bladder cancer (pT2+) were obtained by cystectomy or cystoprostatectomy at Memorial Sloan-Kettering Cancer Center. Specimens were collected under institutional review board approval of this institution. Macrodissection of OCT-embedded tissue blocks was performed to ensure a minimum of 75% tumor cells. Because of the high heterogeneity of muscle-invasive bladder tumors, this conservative cutoff of 75% would guarantee that tumor subpopulations would be representative enough to identify targets associated with TP53 status in cancer cells. Total RNA was extracted using TRIzol (Life Technologies, Rockville, MD) and purification with RNeasy columns (Qiagen, Valencia, CA). RNA quality was evaluated based on 260/280 ratios of absorbances and by gel analysis using an Agilent 2100 BioAnalyzer (Agilent Technologies, Palo Alto, CA).13 Selection of cases for oligonucleotide arrays focused on balancing numbers of cases with wild-type (n = 24) and mutant TP53 (n = 22), covering the most frequent TP53 mutations in the DNA-binding core domain in cases displaying all advanced disease (PT2+).
Labeling and Hybridization
Complementary DNA of the analyzed specimens was synthesized from 1.5 µg of total RNA using a T7-promoter- tagged oligo-dT primer. RNA target was synthesized by in vitro transcription and labeled with biotinylated nucleotides (Enzo Biochem, Farmingdale, NY). Labeled target was hybridized on GeneChip test 3 arrays (Affymetrix) to assess the quality of the sample before hybridizing onto the human genome U133A arrays including 22,283 probes representing known genes and expressed sequence tags (Affymetrix), as previously reported.13
GeneChip Analysis
Scanned image files were visually inspected for artifacts and analyzed using Affymetrix Microarray Suite 5.0 (MAS 5.0). Expression values of each array were multiplicatively scaled to have an average expression of 500 at least across the central 95% of all genes on the array. Signal was used as the primary measure of expression level, and detection was retained as a complementary measure.13
Immunohistochemical Analyses
Cell Lines, Tissue Arrays, and Immunohistochemistry
Cytospins of bladder cancer cell lines were obtained after centrifugation at low speed, 800 rpm, for 5 minutes.16 Four different bladder cancer microarrays were constructed in the Division of Molecular Pathology and used in this study. These arrays included a total of 294 primary transitional cell carcinomas (TCCs) of the bladder, belonging to patients recruited at Memorial Sloan-Kettering Cancer Center under institutional review board-approved protocols. A total of 93 non-muscle-invasive and 201 invasive TCC tumors were analyzed in these microarrays. These tumors corresponded to 34 grade 1, 69 grade 2, and 191 grade 3 lesions. One of these tissue microarrays comprised a cohort of four non-muscle-invasive lesions and 91 invasive tumors with annotated follow-up and known status of TP53. This array allowed clinical outcome assessment and evaluation of the associations of novel markers with TP53. Protein expression patterns of gelsolin were assessed at the microanatomical level on these tissue microarrays by immunohistochemistry using standard avidin-biotin immunoperoxidase procedures. Western blot assays were performed to address the specificity of the antibodies under study. We used a mouse monoclonal antibody against TP53 (1801) at 1:500 dilution (Calbiochem, San Diego, CA) and gelsolin at 1/1000 (Sigma, St. Louis, MO) on formalin-fixed/paraffin-embedded sections. The avidin-biotin immunoperoxidase technique was the immunohistochemical method applied. For specific epitopes on paraffin sections, we used antigen retrieval methods (0.01% citric acid for 15 minutes under microwave treatment) before incubation with primary antibodies or antiserum overnight at 4??C. Secondary antibodies were biotinylated horse anti-mouse or goat anti-rabbit antibodies (Vector Laboratories, Peterborough, UK), which were used at 1:500 or 1:1000 dilution, respectively. Diaminobenzidine was used as the final chromogen and hematoxylin as the nuclear counterstain. Two independent pathologists (C.C.-C. and N.B.), blinded to the TP53 or clinical status of the samples, reviewed immunohistochemical stainings.
Statistical Analysis
All TCCs (n = 294) were used for the analysis of association among gelsolin with clinicopathological variables and the expression patterns of TP53. The consensus value of the representative cores from each tumor sample arrayed was used for statistical analyses. The association of the expression of the selected targets with histopathological stage and tumor grade was evaluated using the nonparametric Wilcoxon-Mann-Whitney and Kruskall-Wallis tests. There is no consensus on the cutoffs of the immunohistochemical expression of the other markers, and thus they were analyzed as continuous variables.17 Survival analyses were performed taking the cutoffs of 20% for TP53 and 5% for gelsolin.
The associations of the markers identified in the DNA microarray analysis to outcome were also evaluated at the protein level using a subset of 95 TCCs of the bladder cases for which follow up was available. Overall-survival time was defined as the years elapsed between transurethral resection or cystectomy and death from disease (or the last follow-up date). Patients who were alive at the last follow-up or lost to follow-up were censored. For survival analysis, the association of marker expression levels with overall survival was analyzed using the Wald test, and the log-rank test was used to examine their relationship when different cutoffs were applied.17 Survival curves were plotted using the standard Kaplan-Meier methodology. Associations among gelsolin with TP53 were analyzed using Kendall??s b-test.17 Statistical analyses were performed using the SPSS statistical package (version 10.0).
Reverse-Phase Arrays
Bladder Cancer Cell Lines
Nine bladder cancer cell lines were obtained from the American Type Culture Collection (Rockville, MD), grown, and collected under standard tissue culture protocols as previously reported.16 These cell lines were derived from TCCs of the bladder of early stage (RT4), low grade (5637), invasive (T24, J82, UM-UC-3, HT-1376, and HT-1197), and metastatic bladder tumors (TCCSUP), as well as a squamous cell carcinoma cell line (ScaBER). Bladder cancer cell lines were wild type for TP53 (RT4) or presented mutations in TP53 at the following exons: 4 (UM-UC-3, ScaBER), 5 (T24), 7 (HT-1376), 8 (5637 and J82), 10 (TCCSUP), and 11 (HT-1197).16
Protein Lysate Preparation
The bladder cancer cell lines were cultured, and protein extracts were prepared from them as previously described.16 In brief, cells were collected by scraping and washed three times with ice-cold phosphate-buffered saline. The resulting pellets were lysed in buffer containing 9 mol/L urea (Sigma), 4% 3--1-propanesulfonate (CHAPS; Calbiochem), 2%, pH 8.0 to 10.5, Pharmalyte (Amersham Pharmacia Biotech, Piscataway, NJ), and 65 mmol/L dithiothreitol (Amersham Pharmacia Biotech). After lysis, the samples were centrifuged briefly, and the supernatants were stored at C80??C.
Protein Lysate Array Design and Production
Arrays were prepared on nitrocellulose-coated glass slides (FAST Slides; Schleicher & Schuell, Keene, NH) by using a pin-in-ring format GMS 417 arrayer (Affymetrix) with four 500-µm-diameter pins. Because the samples were viscous, the pin-in-ring format was used to avoid problems because of clogging of quills. Five twofold serial dilutions were made from each lysate. Four 384-well microtiter plates (Genetix, New Milton, UK) were used to array 180 spots (plus eight spatial registration marks for use in image processing) on a 21 x 35-mm area of nitrocellulose membrane. The first dilution (fourfold) was made with buffer containing 5 mol/L urea, 2% Pharmalyte, pH 8 to 10.5, and 65 mmol/L dithiothreitol. The remaining dilutions were then made with buffer containing 6 mol/L urea, 1% CHAPS, 2% Pharmalyte, pH 8 to 10.5, and 65 mmol/L dithiothreitol. Hence, only the lysate concentration changed along each dilution series. The urea concentration was thus kept at 6 mol/L and the CHAPS concentration at 2%, to keep proteins in their denatured forms. To avoid evaporation in the microtiter plate during spotting, humidity in the array chamber was kept at 70 to 90% with a Vicks ultrasonic humidifier (Kaz, Hudson, NY).18
Detection of Specific and Total Protein on Microarrays
Each array was incubated with a specific primary antibody, which was detected by using the catalyzed signal amplification system (DAKO, Carpinteria, CA). Briefly, each slide was washed manually with deionized water to remove urea. Then, in an Autostainer universal staining system (DAKO), it was blocked with I-block (Tropix, Bedford, MA) and incubated with primary and secondary antibodies. Also in the Autostainer, it was then incubated with streptavidin-biotin complex, biotinyl tyramide (for amplification) for 15 minutes, streptavidin-peroxidase for 15 minutes, and 3,3'-diaminobenzidine tetrahydrochloride chromogen for 5 minutes. Between steps, the slide was washed with catalyzed signal amplification buffer. The signal was scanned with a Perfection 1200S scanner (Epson America, Long Beach, CA) with 256-shade gray scale at 600 dots per inch. For detection of total protein, arrays were stained with SYPRO ruby protein blot stain (Molecular Probes, Eugene, OR) and scanned with a FluorImager SI (Amersham Pharmacia Biotech) at 100-µm resolution. Gelsolin expression was quantified at a 1/1000 dilution using a mouse monoclonal antibody (Sigma), whereas mutated TP53 was measured using a mouse monoclonal at 1/500 (Calbiochem, Darmstadt, Germany). Spot images were converted to raw pixel values by a modified version of the P-SCAN (Peak Quantification with Statistical Comparative Analysis) software.18,19
Western Blotting
Murine monoclonal antibodies were screened for specificity by Western blotting with 20 µg of lysate protein per lane. Western blotting of gelsolin was performed at a 1/500 dilution using a mouse monoclonal antibody (Sigma). The running buffer contained 62.5 mmol/L Tris-HCl, pH 6.8, 2% sodium dodecyl sulfate, 10% glycerol, and 2.5% 2-mercaptoethanol. We used a 4 to 15% sodium dodecyl sulfate-polyacrylamide linear gradient gel (Tris?
【参考文献】
Levine AJ: The p53 tumor-suppressor gene. N Engl J Med 1992, 326:1350-1352
Levine AJ: P53, the cellular gatekeeper for growth and division. Cell 1997, 88:323-331
Sengupta S, Harris CC: p53: traffic cop at the crossroads of DNA repair and recombination. Nat Rev Mol Cell Biol 2005, 6:44-55
Wolff EM, Liang G, Jones PA: Mechanisms of disease: genetic and epigenetic alterations that drive bladder cancer. Nat Clin Pract Urol 2005, 2:502-510
Soussi T, Lozano G: p53 mutation heterogeneity in cancer. Biochem Biophys Res Commun 2005, 331:834-842
Cordon-Cardo C, Dalbagni G, Saez GT, Oliva MR, Zhang ZF, Rosai J, Reuter VE, Pellicer A: p53 mutations in human bladder cancer: genotypic versus phenotypic patterns. Int J Cancer 1994, 56:347-353
Markl IDC, Jones PA: Presence and location of TP53 mutation determines pattern of CDKN2A/ARF pathway inactivation in bladder cancer. Cancer Res 1998, 58:5348-5353
Zhao R, Gish K, Murphy M, Yin Y, Notterman D, Hoffman WH, Tom E, Mack DH, Levine AJ: Analysis of p53-regulated gene expression patterns using oligonucleotide arrays. Genes Dev 2000, 14:981-993
Dalbagni G, Presti J, Reuter V, Fair WR, Cordon-Cardo C: Genetic alterations in bladder cancer. Lancet 1993, 342:469-471
Lianes P, Orlow I, Zhang ZF, Oliva MR, Sarkis AS, Reuter VE, Cordon-Cardo C: Altered patterns of MDM2 and TP53 expression in human bladder cancer. J Natl Cancer Inst 1994, 86:1325-1330
Cordon-Cardo C, Sheinfeld J, Dalbagni G: Genetic studies and molecular markers of bladder cancer. Semin Surg Oncol 1997, 13:319-327
Cordon-Cardo C: p53 and RB: simple interesting correlates or tumor markers of critical predictive nature? J Clin Oncol 2004, 22:975-977
Sanchez-Carbayo M, Socci ND, Lozano J, Saint F, Cordon-Cardo C: Defining molecular profiles of poor outcome in patients with invasive bladder cancer using oligonucleotide microarrays. J Clin Oncol 2006, 24:778-789
Wikman FP, Lu ML, Thykjaer T, Olesen SH, Andersen LD, Cordon-Cardo C, Orntoft TF: Evaluation of the performance of a p53 sequencing microarray chip using 140 previously sequenced bladder tumor samples. Clin Chem 2000, 46:1555-1561
Lu ML, Wikman F, Orntoft TF, Charytonowicz E, Rabbani F, Zhang Z, Dalbagni G, Pohar KS, Yu G, Cordon-Cardo C: Impact of alterations affecting the p53 pathway in bladder cancer on clinical outcome, assessed by conventional and array-based methods. Clin Cancer Res 2002, 8:171-179
Sanchez-Carbayo M, Socci ND, Charytonowicz E, Lu M, Prystowsky M, Childs G, Cordon-Cardo C: Molecular profiling of bladder cancer using cDNA microarrays: defining histogenesis and biological phenotypes. Cancer Res 2002, 62:6973-6980
Dawson-Saunders B, Trapp RG: Basic and Clinical Biostatistics, ed 2. 1994 Appleton and Lange, Norwalk
Carlisle AJ, Prabhu VV, Elkahloun A, Hudson J, Trent JM, Linehan WM, Williams ED, Emmert-Buck MR, Liotta LA, Munson PJ, Krizman DB: Development of a prostate cDNA microarray and statistical gene expression analysis package. Mol Carcinog 2000, 28:12-22
Nishizuka S, Charboneau L, Young L, Major S, Reinhold WC, Waltham M, Kouros-Mehr H, Bussey KJ, Lee JK, Espina V, Munson PJ, Petricoin E, III, Liotta LA, Weinstein JN: Proteomic profiling of the NCI-60 cancer cell lines using new high-density reverse-phase lysate microarrays. Proc Natl Acad Sci USA 2003, 100:14229-14234
Tanaka M, Mullauer L, Ogiso Y, Fujita H, Moriya S, Furuuchi K, Harabayashi T, Shinohara N, Koyanagi T, Kuzumaki N: Gelsolin: a candidate for suppressor of human bladder cancer. Cancer Res 1995, 55:3228-3232
Sakai N, Ohtsu M, Fujita H, Koike T, Kuzumaki N: Enhancement of G2 checkpoint function by gelsolin transfection in human cancer cells. Exp Cell Res 1999, 251:224-233
Celis A, Rasmussen HH, Celis P, Basse B, Lauridsen JB, Ratz G, Hein B, Ostergaard M, Wolf H, Orntoft T, Celis JE: Short-term culturing of low-grade superficial bladder transitional cell carcinomas leads to changes in the expression levels of several proteins involved in key cellular activities. Electrophoresis 1999, 20:355-361
Rao J, Seligson D, Visapaa H, Horvath S, Eeva M, Michel K, Pantuck A, Belldegrun A, Palotie A: Tissue microarray analysis of cytoskeletal actin-associated biomarkers gelsolin and E-cadherin in urothelial carcinoma. Cancer 2002, 95:1247-1257
Spinardi L, Rietdorf J, Nitsch L, Bono M, Tacchetti C, Way M, Marchisio PC: A dynamic podosome-like structure of epithelial cells. Exp Cell Res 2004, 295:360-374
作者单位：From the Tumor Markers Group,* Spanish National Cancer Center, the Division of Molecular Pathology, the Computational Biology Center, and the Department of Urology,¶ Memorial Sloan-Kettering Cancer Center, New York, New York; and the Center for Applied Proteomics and Molecular Medicine, George
Center for Gene Therapy, Tulane University Health Sciences Center, Center for Infectious Diseases and Department of Tropical Medicine
Tulane University Health Sciences Center, New Orleans
Tulane National Primate Research Center, Covington, Louisiana
We used human microarrays to examine gene expression in a rhesus monkey model of human Plasmodium vivax malaria (P. cynomolgi in Macaca mulatta). Whole-blood cells were collected for extraction of RNA before infection, during both the initial liver phase of infection and bloodstream infection, and during the course of 2 bloodstream relapses. Clustering analysis showed that similarities in gene expression were greater at similar stages of the protocol for the 2 different monkeys than for the same monkey at different stages of the protocol. Interestingly, a large number of genes involved in RNA processing showed distinct down-regulation during the initial liver phase of infection. When only up-regulated genes were examined, there was evidence of an increasing number of "defense response" genes as the infection evolved but not of "cytoskeleton" genes (P .001). These results demonstrate the value of microarrays for studying the response of the primate transcriptome to malaria infection; they suggest that the host response is modulated by groups of genes.
High-density oligonucleotide microarrays make it possible to examine the mRNA transcripts for most genes simultaneously (the transcriptome) in a way that has not been possible previously [1, 2]. This capacity is of potential value for examining the host response to a broad spectrum of diseases, including infectious diseases such as malaria. However, most of the microarray results reported previously have focused on gene expression by an invading microorganism or by malignant cells within a tumor. Previous studies have typically compared gene expression by microorganisms at 1 or 2 time points [37] or by malignant cells at 1 time point [810]. The recent study by Sexton et al. is arguably one of the few studies to approach malaria infection by use of microarray analysis of the host, albeit in a murine model [11].
We infected 2 rhesus monkeys with sporozoites of Plasmodium cynomolgi, a monkey malaria parasite, and followed the changes in levels of gene expression of the host cells during the course of a malaria infection, including relapses. A major challenge in performing the present study was that microarrays for nonhuman primates are not yet available. Several investigators have shown that microarrays designed for humans (Affymetrix) can be used to follow changes in gene expression by nonhuman primates [1215]. Cross-species mismatches may result in underestimation of the abundance of a transcript; however, relative changes in signal during the course of an infection should be reliable (R. Norgren, personal communication).
One rationale for using a monkey model is that such models permit controlled infection by sporozoite inoculation at defined times with a specific isolate, which is not possible with humans, who acquire malaria by natural transmission. Using a monkey model, we can also allow infections to continue through several relapses, which is not ethical in human studies. Confounding factors such as nutritional status and coinfections are not a consideration in monkey studies.
P. cynomolgi, the putative ancestor of P. vivax [16], readily infects humans and is used frequently as a model of human P. vivax infection, which is widespread in Asia, Latin America, the Middle East, North Africa, and the South Pacific. Both parasites have 48-h asexual erythrocytic cycles in their respective hosts [16] and produce relapses from persistent liver stages (hypnozoites) [17]. Phylogenetic trees based on small-subunit rRNA group the 2 parasites closely together [18], primers based on P. vivax sequences readily amplify homologous sequences from P. cynomolgi (J. Alger, personal communication), and proteins from P. cynomolgi are fundamentally similar in peptide sequence and function to those from P. vivax (e.g., merozoite surface protein 1) [18, 19]. For these reasons, P. cynomolgi infection of the rhesus monkey has been used as a model of human P. vivax infection to study the biology of the hypnozoite stage responsible for relapse [17], to test molecular markers for relapse, and to test for drug efficacy [20].
The present study demonstrates that microarrays can be used effectively to study the host gene expression in nonhuman primates in response to infection. This strategy should also be useful for examination of host responses in other nonhuman primate models of human disease and in humans with infectious diseases such as malaria. We show that similar patterns of expression have genes with similar biological processes in both of the recipient monkeys in the present study and that 1 of the patterns with the most-significant functions consists of RNA processing genes that were down-regulated during the initial liver phase of infection.
MATERIALS AND METHODS
Monkey infections.
A splenectomized donor monkey (Macaca mulatta) was inoculated intravenously (iv) with 106 asexual parasites of P. cynomolgi bastianellii. After the first detection of gametocytes on blood film (Giemsa-stained thick and thin), 300 female Anopheles stephensi mosquitoes/day were allowed to feed on the monkey, for 3 days. The mosquitoes were maintained for 2 weeks at 27°C, and infective-stage sporozoites were harvested, suspended in medium 199 with 10% rhesus monkey serum, and injected iv into the recipient monkeys.
The 2 recipient monkeys, with spleens intact, received 106 infective sporozoites iv via the saphenous vein. Parasitemias were monitored as above, and the monkeys were treated with chloroquine (7 mg base/kg/day intramuscularly for 5 days) and with primaquine (4 mg base/kg/day orally for 7 days) after the second relapse. One of the recipient monkeys (CP80) was given interferon (IFN) (5 g/kg/day subcutaneously for 3 days) beginning on day 1 of parasitemia (before sample 3 was obtained), in an effort to increase systemic (blood) levels of tumor necrosis factor (TNF). The effect of treatment with IFN- on levels of TNF was monitored by measuring blood levels of TNF by use of an ELISA-based kit (Biosource International), using both blood levels of TNF before treatment and those in the untreated monkey (CL61) as controls.
RNA samples.
Blood samples were obtained before infection (baseline; sample 1), during the initial liver phase of infection (8 days after iv infection; sample 2), when the parasitemia reached 10% (1 × 106 parasites/L) during the first bloodstream infection (sample 3), and during the first and second relapses (samples 4 and 5, respectively) (figure 1). Total RNA was isolated from whole blood by use of the PAXGene kit (Qiagen) and was purified by use of the RNAeasy Mini kit (Qiagen).
Microarray.
Eight micrograms of total RNA was used to synthesize double-stranded cDNA (Superscript Choice System; GIBCO BRL Life Technologies). After synthesis, the cDNA was purified by phenol/chloroform extraction (Phase Lock Gel; Eppendorf Scientific) and concentrated by ethanol precipitation. In vitro transcription was used to produce biotin-labeled cRNA (BioArray HighYield RNA Transcription Labeling Kit; Enzo Diagnostics). The biotinylated cRNA was cleaned (RNAeasy Mini Kit; Qiagen), fragmented, and hybridized on microarray chips (HG-U133A; Affymetrix) containing 22,215 probes representing 15,003 genes. After they had been washed, individual microarray chips were stained with streptavidin-phycoerythrin (Molecular Probes), amplified by use of biotinylated anti-streptavidin (Vector Laboratories), and scanned for fluorescence (GeneArray Scanner; Hewlett Packard), by use of Microarray Suite 5.0 software (MAS 5.0; Affymetrix)
The scanned images, together with absolute calls for each gene (present [P], marginal, or absent), were transferred to the dChip program (version 1.3+; available at: http://biosun1.harvard.edu/complab/dchip/) [21]. Chips were normalized against an array with a median overall signal intensity (SI) value of 172. Expression values were calculated on the basis of both perfect matches and mismatches, and negative values were assigned a value of 1. Differentially expressed genes were obtained in both experiments (monkeys) separately by searching for genes that (1) were scored P in at least 1 sample and (2) had a coefficient of variation (CV; SD of SIs divided by mean SI across all time points) >0.5. This criterion (rather than a CV >0.3) was chosen because it reduced the number of genes required for clustering to a more manageable number. After Affymetrix control genes and redundant genes were removed, the number of genes was reduced from 15,003 to 3278 (CL61) and 3532 (CP80).
Hierarchical clustering and gene ontologies (GOs).
Before hierarchical clustering, the dChip program was used to standardize the SIs for each gene by linearly adjusting their values across all time points to a mean of zero, with an SD of 1. The program was also used to perform hierarchical clustering of the samples.
Eighteen distinct patterns of gene expression were identified from the hierarchical clustering picture, representing 99% of the differentially expressed genes. The dChip program was used to calculate P values for each GO term by use of an exact hypergeometric distribution, to compare the frequencies of individual GO terms within the pattern with the frequencies of those terms on the entire microarray (P .01 was considered to be significant) [22]. GO terms provide information on cellular components, molecular function, and biological processes (9518 of the 15,003 genes on the chip have 1 GO term).
Data were searched for genes with the "transcription factor activity" GO term (n = 634). A gene was retained for further analysis if it (1) was scored P in at least 1 of the 10 samples from the 2 monkeys and (2) had significant variation in its expression across the samples (CV, >0.3). After Affymetrix control genes and redundant genes were removed, the number of transcription factors for hierarchical clustering was reduced to 208.
Next, the data were searched for genes with either a "defense response" (n = 632) or a "cytoskeleton" (n = 640) GO term. These genes were then used to perform pairwise comparisons between all the samples and the baseline sample within an experiment (monkey). Genes retained for further analysis met 3 criteria: (1) they were up-regulated to a value at least 2-fold greater than the baseline level, (2) they were scored P in at least 1 sample, and (3) their SIs were at least 30 points higher than the baseline level.
Reverse-transcriptase polymerase chain reaction (RT-PCR).
Previously isolated RNA from samples 1 and 3, from both monkeys, were used to perform RT-PCR, in accordance with the manufacturer's instructions (iScript cDNA Synthesis Kit and iQ Supermix; BioRad), and the products were separated by an agarose gel electrophoresis. The following primer pairs were used in the RT-PCR: -actin (100 bp), 5-AGAAAATCTGGCACCACACC-3 and 5-GGGGTGTTGAAGGTCTCAAA-3; TNF (155 bp), 5-AACCTCCTCTCTGCCATCAA-3 and 5-TCGAGATAGTCGGGCAGATT-3; a disintegrin and metalloproteinase domain 17 (ADAM17) (144 bp), 5-GGTGGTGGATGGTAAAAACG-3 and 5-GCCCCATCTGTGTTGATTCT-3; v-rel reticuloendotheliosis viral oncogene homolog B (RELB) (110 bp), 5-ATCTGCTTCCAGGCCTCATA-3 and 5-CGCAGCTCTGATGTGTTTGT-3; and signal transducer and activator of transcription 3 (STAT3) (123 bp), 5-CTGGCCTTTGGTGTTGAAAT-3 and 5-CTCTGCCCAGCCTTACTCAC-3.
RESULTS
Time course of P. cynomolgi infection.
After the iv inoculation of infectious sporozoites (after sample 1 was obtained on day 0), parasites entered hepatocytes in the liver within minutes, where they matured into preerythrocytic schizonts or became dormant hypnozoites within 8 days (sample 2). Between days 10 and 12, merozoites released from mature preerythrocytic schizonts in the liver entered the bloodstream, where they produced detectable parasitemias, which peaked on day 14 (sample 3). After treatment with chloroquine to clear asexual parasites from the bloodstream, the first relapse occurred on day 27 (sample 4). After an additional course of chloroquine to clear the first relapse, blood smears remained negative until the second relapse, on day 46, which was treated with chloroquine beginning on day 47 (sample 5). After treatment with primaquine to eradicate hypnozoites remaining in the liver, there were no further relapses. Blood samples for extraction of RNA were collected on days 0, 8, 14, 27, and 47 (figure 1). Neither monkey showed detectable amounts of circulating TNF- during the initial peak parasitemia.
Hierarchical clustering of samples and genes.
Samples were clustered by use of differentially expressed genes (3278 for CL61 and 3532 for CP80), and the results are presented in dendrograms in which the lengths of the branches are proportional to differences in gene expression between samples from the same recipient monkey (figure 3A) or between samples from both monkeys (figure 3B). As shown in the first dendrogram (figure 3A), gene expression in the baseline sample (sample 1) was the most different. Conversely, the most similar gene expression was observed in samples from the first and the second relapses (samples 4 and 5, respectively).
When samples from both monkeys were clustered together (4350 genes), the corresponding samples from the 2 recipient monkeys clustered pairwise. Thus, the second dendrogram (figure 3B) demonstrates that similarities in gene expression were greater at similar stages of the protocol for the 2 different monkeys than for the same monkey at different stages of the protocol. The single exception to this generalization was the second relapse in monkey CL61 (sample 5), during which gene expression was different from that in sample 5 from CP80 (figure 3B).
Next, differentially expressed genes were clustered hierarchically, and, when the results were inspected visually, 18 (CL61) and 19 (CP80) distinct patterns of expression, with narrow confidence intervals, were identified (figure 4). Patterns revealed marked changes in gene expression, including the down-regulation of multiple genes during the initial liver phase of infection (sample 2).
From gene expression to gene function.
GO terms assigned to genes clarify the cellular location, molecular function, and biological processes related to the protein product of each gene. GO terms in a pattern were considered to be significant if P .01. The most-significant GO terms in each pattern are shown in tables 1 (CL61) and 2 (CP80). The patterns with the greatest significance were 16 (CL61) and H (CP80). These patterns were for genes that, compared with the baseline level, were markedly down-regulated during the initial liver phase of infection and thereafter (figure 4). The genes in these patterns were involved in nucleic acid binding, RNA splicing, and related functions (tables 1 and 2). Other patterns with highly significant GO terms were 10 (CL61) and L (CP80). The genes in these patterns had their highest levels of expression during the second relapse (figure 3). Not surprisingly, many of these genes had defense and immune response functions (tables 1 and 2). Defense and immunity genes were also significantly represented in patterns 9 (CL61) and K and N (CP80). The genes in these patterns had their highest levels of expression during either the initial peak parasitemia or the subsequent relapses (figure 4).
RT-PCR.
RT-PCR assays were used to confirm the expression patterns of TNF, ADAM17, RELB, and STAT3. TNF is a cytokine that is important for the defense response, and ADAM17 functions as a TNF-converting enzyme, whereas RELB and STAT3 are transcription factors. The results demonstrate that the expression patterns obtained by use of RT-PCR are similar to the patterns obtained by use of the microarray (figure 7).
DISCUSSION
As demonstrated by the present study and by previous studies [1215], human microarrays can be used to study gene expression in nonhuman primates. However, some differences are inevitable when chips designed for another species are used. For instance, some investigators have suggested that up to 40% of genes in nonhuman primates may not be detected by human microarrays [23] and that the percentage of undetected genes may vary unpredictably across the genome (R. Norgren, personal communication). The percentage of genes detected (called P) in the present study (monkey RNA hybridized to human arrays) ranged from 16% to 36%, compared with 40%50% detected in human studies (human RNA hybridized to human arrays) by use of the same microarrays and software (J.Y. and J. Manges, unpublished observations).
Although interspecies differences are important, the most challenging aspect of gene expression studies is the question of how to move from the overwhelming amount of data generated by microarrays on the entire transcriptome to groups of genes, individual genes, and biological function. In the present study, we have addressed these questions by developing a stepwise protocol, which began by grouping genes on the basis of their expression over time (hierarchical clustering of samples and genes). We then used GO terms and the expression of transcription factors to aid in the conceptual transition from the transcriptome to groups of genes, individual genes, and biological function.
The most striking result of the hierarchical clustering studies is the complex response, on the level of gene expression, of the host to malaria (P. cynomolgi infection). With a single exception (sample 5), the gene expression patterns in the recipient monkeys were clustered primarily according to the stage of the infection (figure 3B). This suggests that the study of gene expression by use of microarrays should permit the development and testing of hypotheses about the response of the host transcriptome to infection.
Graphic representation of gene expression (figure 4) demonstrated that the response to infection was remarkably heterogeneous. Although previous studies (before microarrays were available) invariably emphasized the up-regulation of individual genes, the results presented here indicate that the host response to infection is a complex mix of both up- and down-regulation of groups of genes.
Subsequently, at times of bloodstream infection by asexual parasites (samples 35), one of the interesting patterns observed was the up-regulation of defense-response and immune-response genes in both monkeys (pattern 10, CL61; pattern L, CP80). These patterns are consistent with the development of immune responses to the parasite by the host during the course of bloodstream infection. However, conclusions drawn from these results are observational results (descriptive), rather than analytical (not hypothesis based).
To address this limitation, we compared the expression of genes with the defense response GO term with that of genes with the cytoskeleton GO term. The results of this comparison (figure 6) demonstrate an increase in the number of up-regulated defense response genes (P .001), but not of the cytoskeleton genes (P = .8 and P = .2, 2 test for trend), during P. cynomolgi infection. These results are consistent with the hypothesis that the number of up-regulated genes related to the host defense increases in response to malaria infection and also increases during the course of 2 relapses.
Because transcription factors may be responsible for the coordinated up- and down-regulation of groups of genes, we examined the expression of genes encoding transcription factors, to provide an alternative perspective on the role of gene expression in the response to infection (figure 5). Transcription factors up-regulated during the initial liver phase of infection included genes involved in NF-B activity (RELB), a thyroid hormone receptor (THRA), and a gene involved in pituitary organogenesis and motor neuron development (LHX3). Transcription factors down-regulated during the initial liver phase of infection included the gene for enolase (ENO1), the enhancer binding protein that helps regulate the inflammatory response (CEBPB), and STAT3, which is involved in cytokine release. Transcription factors up-regulated at the time of the initial peak parasitemia included a gene involved in transforming growth factor signaling (MADH4), a gene that encodes a nuclear protein with regulatory functions (HOXC6), and a gene that encodes a nuclear factor (NFE2L3) that regulates erythroid-specific genes.
In the present study, we used hierarchical clustering of microarray data to evaluate the effect of infection on the expression of groups of genes. Also, the expression of transcription factors was used to provide a second perspective on coordination of the host response, and, finally, GO terms were used to link gene expression to biological function (i.e., to test whether groups of defense response genes are up-regulated in response to malaria).
These results provide a logical point of reference for further studies of malaria in nonhuman primates, with the potential to compare the effects of relatively benign infections such as P. cynomolgi with those of more-severe infections, to evaluate the role of circulating TNF and cytokine receptor levels (A.C.R., J.Y., and F.B.C., unpublished data), and to develop strategies for the study of humans with P. vivax and P. falciparum infections.
Acknowledgments
We thank the late Chris Kirijan, for veterinary support; Justin Manges, for technical assistance; and Leena Ala-Kokko, James Colborn, Mark James, Robert Norgren, Jr., Bruce Bunnell, and Darwin J. Prockop, for their thoughtful and constructive reviews of the manuscript.
References
1. Lockhart DJ, Dong H, Byrne MC, et al. Expression monitoring by hybridization to high-density oligonucleotide arrays. Nat Biotechnol 1996; 14:167580. First citation in article
2. Duggan DJ, Bittner M, Chen Y, Meltzer P, Trent JM. Expression profiling using cDNA microarrays. Nat Genet 1999; 21:104. First citation in article
3. Rathod PK, Ganesan K, Hayward RE, Bozdech Z, DeRisi JL. DNA microarrays for malaria. Trends Parasitol 2002; 18:3945. First citation in article
4. Bozdech Z, Zhu J, Joachimiak MP, Cohen FE, Pulliam B, DeRisi JL. Expression profiling of the schizont and trophozoite stages of Plasmodium falciparum with a long-oligonucleotide microarray. Genome Biol 2003; 4:R9. First citation in article
5. Florens L, Washburn MP, Raine JD, et al. A proteomic view of the Plasmodium falciparum life cycle. Nature 2002; 419:5206. First citation in article
6. Kappe SH, Gardner MJ, Brown SM, et al. Exploring the transcriptome of the malaria sporozoite stage. Proc Natl Acad Sci USA 2001; 98:9895900. First citation in article
7. Le Roch KG, Zhou Y, Blair PL, et al. Discovery of gene function by expression profiling of the malaria parasite life cycle. Science 2003; 301:15038. First citation in article
8. Hasegawa S, Furukawa Y, Li M, et al. Genome-wide analysis of gene expression in intestinal-type gastric cancers using a complementary DNA microarray representing 23,040 genes. Cancer Res 2002; 62:70127. First citation in article
9. Kim B, Bang S, Lee S, et al. Expression profiling and subtype-specific expression of stomach cancer. Cancer Res 2003; 63:824855. First citation in article
10. Ntzani EE, Ioannidis JP. Predictive ability of DNA microarrays for cancer outcomes and correlates: an empirical assessment. Lancet 2003; 362:143944. First citation in article
11. Sexton AC, Good RT, Hansen DS, et al. Transcriptional profiling reveals suppressed erythropoiesis, up-regulated glycolysis, and interferon-associated responses in murine malaria. J Infect Dis 2004; 189:124556. First citation in article
12. Bigger CB, Brasky KM, Lanford RE. DNA microarray analysis of chimpanzee liver during acute resolving hepatitis C virus infection. J Virol 2001; 75:705966. First citation in article
13. Enard W, Khaitovich P, Klose J, et al. Intra- and interspecific variation in primate gene expression patterns. Science 2002; 296:3403. First citation in article
14. George MD, Sankaran S, Reay E, Gelli AC, Dandekar S. High-throughput gene expression profiling indicates dysregulation of intestinal cell cycle mediators and growth factors during primary simian immunodeficiency virus infection. Virology 2003; 312:8494. First citation in article
15. Marvanova M, Menager J, Bezard E, Bontrop RE, Pradier L, Wong G. Microarray analysis of nonhuman primates: validation of experimental models in neurological disorders. FASEB J 2003; 17:92931. First citation in article
16. Coatney GR, Collins WE, Warren McW, Contacos PG. The primate malarias. Washington, DC: US Government Printing Office, 1971. First citation in article
17. Cogswell FB. The hypnozoite and relapse in primate malaria. Clin Microbiol Rev 1992; 5:2635. First citation in article
18. Waters AP, Higgins DG, McCutchan TF. Evolutionary relatedness of some primate models of Plasmodium. Mol Biol Evol 1993; 10:91423. First citation in article
19. Barnwell JW, Galinski MR, DeSimone SG, Perler F, Ingravallo P. Plasmodium vivax, P. cynomolgi, and P. knowlesi: identification of homologue proteins associated with the surface of merozoites. Exp Parasitol 1999; 91:23849. First citation in article
20. Schmidt LH, Fradkin R, Genther CS, Rossan RN, Squires W, Hughes HB. Plasmodium cynomolgi infections in the rhesus monkey. Am J Trop Med Hyg 1982; 31:609703. First citation in article
21. Li C, Wong WH. Model-based analysis of oligonucleotide arrays: expression index computation and outlier detection. Proc Natl Acad Sci USA 2001; 98:316. First citation in article
22. Ashburner M, Ball CA, Blake JA, et al. Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat Genet 2000; 25:259. First citation in article
23. Chismar JD, Mondala T, Fox HS, et al. Analysis of result variability from high-density oligonucleotide arrays comparing same-species and cross-species hybridizations. Biotechniques 2002; 33:5168, 520, 522. First citation in article
24. Hoffman SL, Isenbarger D, Long GW, et al. Sporozoite vaccine induces genetically restricted T-cell elimination of malaria from hepatocytes. Science 1989; 244:107881. First citation in article
From the University Clinic of Neurology, Clinical Department of Clinical Neurology (W.L., W. Lang, S.G.), and Department of Emergency Medicine (M.M.), University of Vienna Medical School, Vienna, Austria.
| Abstract |
|---|
| Top Abstract Introduction Patients and Methods Results Discussion Appendix 1 References |
|---|
Methods— We determined the frequency of total cholesterol (TC) and low-density lipoprotein cholesterol measurements and the use of statins in a multicenter prospective cohort study of 1743 patients with acute ischemic stroke or transient ischemic attack (TIA). Using multivariate logistic regression analysis, we determined the influence of several clinical variables on lipid measurements and the prescription of statins at hospital discharge.
Results— TC was measured in 90% and low-density lipoprotein cholesterol was measured in 48% of the patients. Differences between the centers accounted for most of the observed variability in a multivariate model. Statin prescription also varied widely between the centers. The prescription of a statin at discharge was most strongly associated with statin intake before the event and with increasing TC levels; elderly patients received statins less often. Coronary artery disease, peripheral artery disease, and other manifestations of atherosclerosis were not independently associated with the use of statins; 68% of the patients with manifest atherosclerosis and TC levels >200 mg/dL were discharged without a statin.
Conclusions— The determination of serum lipid profiles varies widely between different centers. Statins are highly underused in patients with recent ischemic stroke or TIA, particularly in those in whom statins are indicated according to existing recommendations (eg, patients with additional coronary artery disease and hypercholesterolemia). Currently, international guidelines concerning the use of statins are not adequately implemented in clinical practice in patients with stroke or TIA.
Key Words: lipids • prevention • statins • stroke
| Introduction |
|---|
| Top Abstract Introduction Patients and Methods Results Discussion Appendix 1 References |
|---|
In patients with CAD, the rate of lipid evaluation is low, and statins are underused.14–20 There is no comparable information for patients with recent ischemic stroke or TIA, even though myocardial infarction is the main cause of death in survivors of ischemic stroke and in patients who suffered a TIA.21–23
The aim of this cohort study was to assess (1) the proportion of patients in whom lipid profiles were determined during hospitalization after an acute ischemic stroke or TIA and (2) the use of statins.
| Patients and Methods |
|---|
| Top Abstract Introduction Patients and Methods Results Discussion Appendix 1 References |
|---|
For the present cohort study, data from 1743 patients with acute ischemic stroke or TIA for whom clinical data were available at the time of analysis and who were admitted between October 1998 and June 2001 were analyzed. Patients with hemorrhagic stroke and patients who died before discharge were excluded. Recruitment of patients to the registry is an ongoing process, and we arbitrarily closed the database for this study on June 2001. At that time, 3069 patients were admitted because of suspected ischemic or hemorrhagic stroke, and complete data for 2171 patients were entered into the database. Patients for whom detailed clinical information was still missing (n=898) were excluded from the analysis. These excluded patients were comparable in terms of age (67.4±14.6 and 66.6±18.5 years [mean±SD], P=0.5) and sex (47% and 45% female, P=0.4). We estimate that 15% of these would not meet our inclusion criteria because of a nonischemic cerebrovascular event (intracerebral hemorrhage, subarachnoid hemorrhage, sinus vein thrombosis) or posthoc verification of a noncerebrovascular diagnosis (epileptic seizure, hypertensive crisis, etc). Data are entered into the database in batches, and we are not aware of a selection mechanism other than order of admission.
Statistical Analysis
Univariate comparison of continuous variables was performed with the unpaired t test or Mann-Whitney U test as appropriate. Binary and categorical data were analyzed with 2 statistics.
To determine the influence of clinical variables simultaneously on the prescription of statins (yes versus no), we applied multivariate logistic regression. We included all variables that were at least weakly associated with the use of statins (P<0.2 in univariate analyses). The Nagelkerke pseudo-R2 was used to assess the variability explained by each model. The Hosmer-Lemeshow 2 test was used to assess the model fit. We assumed that treatment strategies may be determined partly by "local culture." Therefore, we also investigated the potential impact of a cluster effect with regard to hospital (center) on determinants for the prescription of statins. We used a random-effects logit model.26
The following variables were treated as dependent variables: (1) determination of total cholesterol (TC) levels (yes versus no), (2) determination of LDL-C levels (yes versus no), and (3) prescription of a statin at hospital discharge or transfer (yes versus no). The few patients (2%) treated with fibrates were classified as not receiving statins.
The following parameters were included in the analyses as independent variables: age (<55, 55 to 64, 65 to 74, 75 to 84, 85 years); sex; stroke severity at 1 week according to the Rankin Scale (0 to 1, 2 to 3, 4 to 5); cause (large-vessel disease [ipsilateral carotid stenosis 70%, presumable local thrombosis of a large intracranial vessel, arterioarterial embolism from aortic plaques/thrombi], small-vessel disease [clinical lacunar syndrome and no lesion or subcortical lesion <1.5 cm on CT or MRI], cardioembolic [high-risk source of cardiac embolism27], or no determined etiology); history of hypertension as reported by patient or relative or documented in previous medical records (yes versus no); history of diabetes as reported by patient or relative (yes versus no); current cigarette smoking (yes versus no); previous stroke (yes versus no); clinically manifest CAD (yes versus no); clinically manifest peripheral artery disease (PAD); chronic or paroxysmal atrial fibrillation (yes versus no); index event under lipid-lowering drug therapy; index event under antiplatelet drug (yes versus no); index event under oral anticoagulation; and TC level (200, 201 to 220, 221 to 240, 241 to 260, 261 to 280, 281 to 300, >300 mg/100 mL [5.17, 5.19 to 5.69, 5.71 to 6.21, 6.23 to 6.72, 6.75 to 7.24, 7.27 to 7.76, >7.76 mmol/L]). Any question answered with "unknown" was classified as "no."
| Results |
|---|
| Top Abstract Introduction Patients and Methods Results Discussion Appendix 1 References |
|---|
|
|
LDL-C levels were measured in 835 (48%) of the patients. Again, department was the only factor significantly associated with LDL-C determination.
Statin Treatment at Hospital Discharge
Most patients (68%) with clinically relevant atherosclerosis (CAD, PAD, symptomatic carotid stenosis, plaques of the aortic arch >4 mm) and TC levels >200 mg/dL (>5.17 mmol/L) were not treated with a statin (Table 2). Almost 40% of the patients with manifest atherosclerosis and TC levels >300 mg/dL (7.76 mmol/L) were discharged without a statin.
|
|
Overall, 1342 patients (77%) received antiplatelet agents and another 505 patients (29%) received oral anticoagulants or heparin at discharge. We found that 1081 of all 1743 patients (62%) and 909 of those 1109 patients with known hypertension (82%) were treated with antihypertensive agents at discharge.
Determinants of Statin Treatment at Hospital Discharge
Statin prescription varied widely between departments for both the total study population and patients with manifest atherosclerosis and TC levels >200 mg/dL (P<0.001 and P=0.027, respectively; Table 1). Not surprisingly, statin treatment was used less often when TC and LDL-C levels were not measured (OR, 0.4; 95% CI, 0.2 to 0.7; P=0.002; and OR, 0.6; 95% CI, 0.4 to 0.8; P<0.001 adjusted for history of hypertension, history of diabetes, cigarette smoking, CAD, PAD, previous intake of a lipid-lowering drug, previous intake of antiaggregants, atrial fibrillation, stroke cause, Rankin Scale at 1 week, and department).
Other clinical factors associated with statin treatment at discharge are shown in Table 3. The factor most strongly associated with the prescription of statins, as expected, was the intake of a lipid-lowering medication before the index event (Table 4). TC levels were strongly and almost linearly associated with the use of statins. Elderly patients were less often treated with statins. The variability between the departments remained highly significant. We found a trend for less frequent use of statins in patients with severe stroke. CAD, PAD, and large-artery disease were not associated with the use of statins in this model. The model explained 37% of the variability (Hosmer-Lemeshow 2=7.65; df=8; P=0.47). When a cluster effect of the treating department was taken into account, the effect sizes remained largely unchanged compared with the estimates in Table 4 (data not shown).
|
|
|
|
| Discussion |
|---|
| Top Abstract Introduction Patients and Methods Results Discussion Appendix 1 References |
|---|
To the best of our knowledge, this is the first study to investigate factors influencing the evaluation and management of lipid disorders in a large cohort of patients with acute stroke or TIA. Previous studies have reported low rates of lipid screening in patients with acute myocardial infarction.15,28
The value of educational programs in hospitals for the improvement of lipid evaluations could be demonstrated,28 and we are trying to establish common guidelines for lipid evaluation as a consequence of our results.
Overall, 23% of our patients and 32% of those with clinically relevant atherosclerosis and cholesterol levels >200 mg/dL received a statin at discharge. This rate is comparable to those reported in a meta-analysis of patients with hyperlipidemia and CAD, which found that only 37% of 622 patients with CAD and hyperlipidemia were treated with statins.17 A very recent investigation of a large national sample of patients with acute myocardial infarction found an overall frequency of the use of statins of 32%.29 Another investigation found an overall rate of statin use of 30% at discharge in 17 732 German patients with acute myocardial infarction.30 In patients with PAD, a recent study described that 35% of all patients (and only 30% of those with lower-extremity disease) were on a lipid-lowering agent,31 whereas yet another investigation found that only 5% of their patients with critical lower-extremity vascular disease received lipid-lowering treatment.32 There is also evidence that patients with PAD receive lipid-lowering treatment less often than patients with CAD.33
We found a very high variability for the use of statins between the participating departments. Furthermore, there was a highly significant association between lipid evaluation and statin treatment. Small area variation for the use of lipid-lowering drugs between hospitals has so far been addressed in only 1 study, which found a significantly higher rate of use of lipid-lowering drugs in teaching hospitals (39.4%) compared with nonteaching hospitals (30.3%).29 Lack of adherence to clinical guidelines on the use of lipid-lowering drugs has been described previously.15,20 Our findings underscore the need for interventions on a local basis, eg, the establishment of local guidelines and hospital education programs.
The recently published British Heart Protection Study found a strong reduction in vascular events with simvastatin compared with placebo in a mixed population of vascular patients, including patients after cerebrovascular events.6 The risk reduction was significant not only in patients with hypercholesterolemia but also in patients with normal or low-normal cholesterol levels. Patients with previous cerebrovascular disease had a risk reduction with statins similar to that of the other patient groups in this study. Considering the high annual risk of myocardial infarction, stroke, or vascular death in patients with a recent ischemic stroke or TIA, it might therefore be reasonable to advocate the use of statins in most patients with stroke or TIA. However, until now, no guidelines or recommendations have existed for the general use of statins in patients with ischemic stroke or TIA. At first glance, this might explain the low overall frequency of statin use and the large variability between different centers. But even in patients with a given indication for statin use according to international guidelines, almost two thirds are left untreated, and a large variability remains between different centers according to our data. Assigning a 5-year risk of a major vascular event or vascular death of 30% to these high-risk patients, which is a conservative estimate, and assuming that treatment with statins would reduce this risk by 20% means that 40 serious vascular events and deaths per 1000 patients would not be prevented in the current clinical setting.
Study Limitations
We collected data on the determination of lipid levels and initiation of statin therapy during hospitalization, but we did not determine the frequency of lipid evaluation and therapy after discharge. Failure to implement screening and treatment in the outpatient setting, however, has been documented,14–17 and it has been suggested that these measures be implemented during the predischarge phase of hospitalization.34
Our investigation was limited to patients treated in neurological departments. Patients with stroke admitted to general medical departments may receive a more appropriate lipid-lowering treatment. This is not very likely because the data from patients with myocardial infarction15,17,20,29 were collected in medical departments. The frequency of lipid evaluation and the use of statins in our study compare favorably.
Finally, 29% of our sample were not analyzed because of logistical reasons caused by the ongoing nature of the registry; of these subjects, 15% would not meet the inclusion criteria. Those not included were comparable in terms of age and sex distribution to those included, and we are not aware of any selection mechanism other than order of admission and batch size. Thus, we believe that selection bias is not a problem in our study.
Study Strengths
Randomized, controlled trials are necessary to establish the effectiveness of healthcare interventions. Large-scale observational studies such as our stroke registry are the next necessary step in evaluating healthcare interventions in real-life settings. The multicenter approach incorporates clinical practices of primary, secondary, and tertiary care hospitals serving a large urban population. Although geographic variation is to be expected, we believe that our results are generalizable to a Western medical setting.
Conclusions
There is an urgent need to optimize lipid evaluation and treatment with statins in patients with stroke or TIA, particularly in high-risk patients for whom guidelines already exist but are not adequately followed.
| Acknowledgments |
|---|
| Appendix 1 |
|---|
| Top Abstract Introduction Patients and Methods Results Discussion Appendix 1 References |
|---|
Follow-Up Examinations and Data Documentation
A. Albrecht, J. Arrich, S. Bonelli, D. Brenner, R. Crevenna, D. Doppelbauer, V. Dorda, J. Ferrari, J. Frühauf, S. Greisenegger, R. Hosner, M. Jankovic, J. Klinger, J. Königseder, C. Kunaver, A. Lengyel, I. Lobentanz, A. Maj, B. Mann, S. Milovic, M. Mirafzal, A. Pourkarami, M. Reumann, T. Scholze, D. Strozyk, and S. Tentschert, all from the University Clinic of Neurology, University of Vienna.
Biostatistics and Epidemiology
M. Müllner, University Clinic of Emergency Medicine, University of Vienna.
Cooperating Centers
Departments of Radiology of the following hospitals: Krankenanstalt Rudolfsstiftung, Kaiser-Franz-Josef-Spital, Krankenhaus Lainz, Neurological Hospital Rosenhügel, Wilhelminenspital, and Donauspital.
Received May 14, 2002; revision received July 18, 2002; accepted July 23, 2002.
| References |
|---|
| Top Abstract Introduction Patients and Methods Results Discussion Appendix 1 References |
|---|
