1 From the Interdepartmental Nutrition Program and Department of Foods and Nutrition, Purdue University, West Lafayette, IN.
2 Presented at the conference "Vitamin D and Health in the 21st Century: Bone and Beyond," held in Bethesda, MD, October 9–10, 2003. 3 Supported by a grant from the NIH (grant DK54111). 4 Address correspondence to JC Fleet, 700 West State Street, Purdue University, West Lafayette, IN 47907-2059. E-mail: fleetj{at}cfs.purdue.edu.
ABSTRACT
Although we have learned a great deal about vitamin D metabolism and function since it first became apparent that this factor was required for bone health, there are still many gaps in our understanding, at both the basic science (eg, the molecular actions and targets of vitamin D) and applied (eg, what "adequate" vitamin D status means) levels. For example, although the identification of extrarenal 1-hydroxylase activity suggests that autocrine/paracrine actions of 1,25-dihydroxyvitamin D complement the classic endocrine actions of the hormone, the practical implications of this finding are only now being explored. In addition, studies showed that 1,25-dihydroxyvitamin D rapidly activates signal transduction pathways in addition to the classic transcriptional activation pathways that require the vitamin D receptor. These new modes of vitamin D action may be crucial to our understanding of both the traditional calcium-regulating actions of vitamin D and the anticancer actions of this essential mediator. Recent developments in genomics and proteomics have provided new opportunities for us to identify molecular targets of vitamin D action. Cancer researchers have demonstrated that these methods have utility for identifying useful biomarkers of disease states. Can these approaches be used to help clarify what constitutes optimal serum concentrations of 25-hydroxyvitamin D? I present an overview of how proteomic and genomic evaluations of cells, animals, and human subjects have been and can be used to improve our understanding of vitamin D biological processes and the role of vitamin D in health.
Key Words: Proteomics • genomics • biomarker
INTRODUCTION
Traditionally, vitamin D metabolism has been viewed as an endocrine system that responds to changes in serum calcium concentrations (1). Low dietary calcium intake is reflected by a decrease in serum calcium concentrations, which is in turn a signal for the increased production and release of parathyroid hormone (PTH). Among its functions, PTH stimulates renal 1-hydroxylase activity, leading to increased conversion of 25-hydroxyvitamin D3 [25(OH)D3] to 1,25-dihydroxyvitamin D3 [1,25(OH)2D3]. Elevated serum 1,25(OH)2D3 concentrations then stimulate the expression of vitamin D-responsive genes within the primary target tissues that control calcium homeostasis (ie, TRPV6 and calbindin D9k in intestine, osteocalcin and RANKL in bone, and TRPV5 and calbindin D28k in kidney) through activation of the vitamin D receptor (VDR). Although this system is clearly functional and biologically important during periods of calcium stress, it may not be sufficient to explain all of the biological actions of vitamin D.
As several recent reports and other reviews in this conference demonstrated, improved bone health and cancer chemoprevention may be more closely related to changes in serum 25(OH)D3 concentrations than to serum 1,25(OH)2D3 concentrations. For example, increases in serum 25(OH)D3 concentrations were associated with both maximal suppression of PTH (a proresorptive agent) (2–4) and increased efficiency of calcium absorption (5–8). There is also evidence that regular, high-dose, vitamin D supplementation decreased factures rates for common osteoporotic sites (9). This finding and other data suggest that local production of 1,25(OH)2D3, rather than endocrine signaling attributable to renal production, is critical for optimal bone health and cancer prevention (10–12). In addition, as we have come to understand more about the details of the molecular mechanisms controlling VDR function (13, 14), researchers have identified 1,25(OH)2D3 as an activator of various signal transduction pathways leading to the stimulation of protein kinases such as Src kinase, protein kinase C, protein kinase A, and the mitogen-activated protein kinases (15). These examples suggest that strict adherence to the accepted concepts of vitamin D biological mechanisms and actions are not likely to explain fully the health benefits of vitamin D.
In light of these and other recent findings, I contend that the field of vitamin D research would be well served by the use of several new technologies in an attempt to better describe the full range of vitamin D actions. The advances in technologies that permit whole-transcriptome analysis and large-scale proteomic analysis permit us to conduct unbiased evaluations of physiologic states and responses to treatments and interventions. In this review, I briefly demonstrate several instances in which genomic or proteomic analysis has expanded our understanding of vitamin D actions, I present a paradigm for future discovery-based research, and I demonstrate how these and other tools might be used to better define optimal vitamin D status.
GENOMIC AND PROTEOMIC APPROACHES: WHAT DO THEY OFFER TO THE FIELD OF VITAMIN D RESEARCH?
General approach
In the past decade, advances in the areas of genomics and proteomics created a revolution in science. These approaches are reviewed in detail elsewhere (16, 17) but are summarized briefly in Figure 1. Our traditional approach to understanding biological processes has been reductionist. We look for proteins that modulate functions of cells and tissues, and we study the regulation of their production and activity. There is no question that this approach has been very fruitful and will continue to be so. However, while this approach focuses our attention on testable hypotheses, it also provides virtual blinders, ie, we look only for the things we have already been studying. Moving beyond this approach can lead to dramatic advances in our understanding of scientific areas. For example, studies of the known proteins involved in iron metabolism failed to identify the cause of the iron-overload disease hemochromatosis. Only after the gene that is mutated in hemochromatosis was identified through a large-scale sequencing effort (genomics) did researchers find that the mutated gene encoded a protein with features similar to the major histocompatibility complex proteins, rather than a protein that was previously thought to be involved in iron metabolism (18). Similarly, studies have revealed that bone mass and metabolism can be regulated by genes that we would classically associate with obesity, such as leptin (19) and LDL receptor-related protein 5 (20). "Omic" approaches give us the opportunity to see beyond our expectations and discover new gene targets for vitamin D actions and functions.
View larger version (45K):
FIGURE 1.. Comparison of reductionist and genomic/globalist approaches to elucidating biological processes.
Studies identifying potential, new, vitamin D-regulated targets with genomics
Several transcript-profiling studies with microarrays have been conducted to elucidate the biological role of 1,25(OH)2D3. Because of the high cost of microarray-based gene expression profiling, most of the studies reported in the literature used very restricted conditions [eg, a single dose and time of treatment with 1,25(OH)2D3] and did not include replicates. This limits the utility of the approach and decreases our confidence in the results. However, more carefully controlled experiments were conducted and addressed some of the experimental design points noted in Figure 2. Figure 2A demonstrates that, depending on the time point examined after 1,25(OH)2D3 treatment and the time course of primary responses to the treatment, the changes in transcript levels could include both primary responses (eg, those resulting from direct, VDR-mediated interactions with gene promoters) and downstream responses that are indirect. Detection of a change in expression does not prove that a gene is a direct vitamin D target gene.
View larger version (33K):
FIGURE 2.. Experimental issues relevant to the use of genomic approaches to study vitamin D biological processes. A: Temporal relationship between vitamin D treatment and the molecular response at the level of the cell. B: Use of multiple independent perturbations to confirm the molecular actions of vitamin D. 1 OHase, 1-hydroxylase.
This phenomenon of differential patterns of responses to vitamin D is evident in the work of Lin et al (21). Those authors examined the time course of the response to 100 nmol/L 1,25(OH)2D3 or the vitamin D analog EB1089 in squamous cell carcinoma cells (SCC25) and identified 152 genes as being regulated by vitamin D (89 up and 63 down), with the use of the Affymetrix FL array (Affymetrix, Santa Clara, CA) and a 2.5-fold cutoff for determining a meaningful change in expression. Clustering was performed on the basis of the pattern of expression or the functional classification of the transcripts. Figure 3 shows the diversity of the vitamin D responses in these cells. Even within the genes with documented, functional, vitamin D response elements (VDREs) in their promoters, there was heterogeneity in responses. For example, the CYP24 transcript levels were rapidly increased by vitamin D treatment (significantly increased in 1 h), whereas osteopontin transcript levels increased more slowly in response to treatment (maximal expression at 12 h). This suggests that similar VDREs are differentially regulated, depending on the promoter context (22), but it also demonstrates the difficulty of discerning a direct transcriptional response solely on the basis of a time course, ie, even later responses can be attributable to direct effects. Another interesting finding from that study was that the vitamin D-induced responses in the transcript profile were much more diverse than might have been predicted previously. For example, several transcripts coding for proteins involved in protection from oxidative stress were gradually up-regulated by the vitamin D analog, including glucose-6-phosphate dehydrogenase (generating NADPH), glutathione peroxidase, and selenoprotein P. In addition, the thioredoxin reductase transcript was increased by 1 h after treatment, with peak induction by 6 h. Rapid suppression of transcripts for a variety of signaling peptides (eg, PTH-related protein and galanin) and induction of intracellular cell signaling proteins (eg, Cox-2, phosphoinositide-3-kinase, and p85 subunit) were also observed after treatment. It is not clear which of these responses is primary; none of these genes was previously shown to be regulated by vitamin D or to contain a functional VDRE. However, because 1,25(OH)2D3 promotes cellular differentiation, the up-regulation of some transcripts may represent a vitamin D-induced shift to a more differentiated phenotype. In any case, these data suggest that the traditional approach of examining only the expression of genes controlling cell cycle proteins in an attempt to explain the prodifferentiating action of vitamin D may provide limited information regarding the biological mechanisms of 1,25(OH)2D3 actions in proliferating or cancer cells.
View larger version (20K):
FIGURE 3.. Summary of patterns observed in response to treatment of squamous carcinoma cells with calcitriol or the vitamin D analog EB1089. The symbols represent the 10 distinct groups of up-regulated (5 groups) (black symbols) and down-regulated (5 groups) (white symbols) transcripts. These data demonstrate that there is considerable diversity in the temporal response of transcripts to vitamin D treatment. Adapted from Lin et al (21).
To identify direct vitamin D actions, scientists often conduct multiple complementary experiments and compare the results for consistency. In Figure 2B, 3 complementary mouse experiments are illustrated. Treatment of normal mice with 1,25(OH)2D3 would be expected to up-regulate or down-regulate specific genes in a target tissue, whereas the examination of target tissue transcript profiles in mice that lack essential components of the vitamin D signaling system (eg, VDR or 1-hydroxylase) would be expected to demonstrate opposite effects on vitamin D target genes [eg, a gene that is activated with 1,25(OH)2D3 injection would be down-regulated in VDR- or 1-hydroxylase-null mice]. A preliminary attempt at this approach with a small group of animals was recently reported by Li et al (23). By comparing the gene expression profile changes that occurred in kidney after 1,25(OH)2D3 injection with those that resulted from loss of the VDR (wild-type mice compared with VDR knockout mice), those authors identified 95 genes for which the response attributable to vitamin D injection was the opposite of the response attributable to loss of the VDR. Twenty-eight of those transcripts (including 1-hydroxylase mRNA) were up-regulated in VDR-null mice and down-regulated in vitamin D-treated mice, whereas 67 of the transcripts (including 24-hydroxylase mRNA) were down-regulated in VDR-null mice and up-regulated in vitamin D-treated mice. Like the study by Lin et al (21), this study identified many potential vitamin D target genes. Although neither of these studies definitively identified new targets, they narrowed the list of candidates considerably and provided clear guidance for investigators who wish to conduct careful reductionist experiments involving these genes.
Studies identifying new protein complexes necessary for vitamin D actions
Very few studies have taken a proteomics approach to the examination of vitamin D actions in cells. One notable study was recently conducted with keratinocytes by Oda et al (24). Previously, Rachez et al (25) developed an in vitro assay to assess the complex of proteins that associates with the VDR during vitamin D-mediated gene transcription. By using a VDR ligand-binding domain–glutathione S-transferase fusion protein, they were able to identify the VDR-interacting protein (DRIP) complex, a complex of 16 proteins that is essential for vitamin D-mediated gene transcription because of its ability to recruit RNA polymerase II to vitamin D-responsive genes (14, 25). Oda et al (24) used this approach to identify the proteins associated with the VDR in nuclear extracts from proliferating and differentiated keratinocytes. Although they identified a similar complex of proteins interacting with the VDR in proliferating keratinocytes (eg, DRIP complex members and RXR), they found that key members of the complex had changed in differentiated keratinocytes. Proteomic analysis of the proteins associated with the VDR in proliferating and differentiated nuclear extracts showed that at least 5 members of the DRIP complex were lost with differentiation but 2 new proteins, SRC-2 and SRC-3, became prominent members of the complex. That study suggested that the complex mediating VDR-mediated gene expression might not be uniform across vitamin D target tissues or even within the cells of a tissue at different stages of their life spans. This could account for the observed diversity of sensitivity of various cell types/tissues to 1,25(OH)2D3 treatment or the ability of a vitamin D analog to work in one tissue but not another.
At least one other line of vitamin D research might also be improved with a proteomics approach. Specifically, it is now clear that vitamin D stimulates rapid activation of signal transduction pathways, eg, it activates several protein kinases, leading to phosphorylation of various proteins (15). The final targets of the kinases activated by 1,25(OH)2D3 have not yet been identified but it is clear that these targets of phosphorylation are critical for understanding the biological importance of these rapid vitamin D actions. It is apparent that new technologies developed for assessment of the phosphoproteome (26) are likely to identify not only the signal transduction pathways used by 1,25(OH)2D3 but also the terminal proteins whose biological functions are either activated or inhibited through stimulation of these phosphorylation cascades.
Identification of functional biomarkers of vitamin D actions
A major issue in the area of vitamin D research is defining the amount of vitamin D needed for optimal health. Although there are some concerns about the reliability and cross-comparison of various assays for measuring 25(OH)D3 concentrations as an index of vitamin D status (27), the main issue is how 25(OH)D3 concentrations relate to function, ie, bone biological processes related to the risk of osteoporosis or epithelial cell biological processes related to cancer risk. The data from several studies suggest that the cutoff value for adequate serum 25(OH)D3 concentrations may be higher than simply the concentration that prevents rickets (2–4). This idea is based on the association of high vitamin D concentrations with suppression of serum PTH concentrations, a measurement that is being used as a surrogate marker of bone resorption (with the assumption that maximal suppression of PTH is necessary for maximal protection of bone). However, is the PTH concentration an appropriate marker for bone resorption? Or would linking serum 25(OH)D3 concentrations to a more relevant functional endpoint or using a biomarker that is correlated directly with a functional endpoint be better for defining optimal vitamin D status? And what about cancer risk? Should we assume that the serum 25(OH)D3 concentration that is optimal for the protection of bone is optimal for the prevention of cancer? In a perfect world, scientists would be able to directly relate vitamin D status to factors such as fracture incidence, changes in bone density with time, or the development of prostate, breast, or colon cancer. However, such studies would require very large populations and a long study period (as well as being ethically questionable). An alternative might be to correlate vitamin D status with an intermediate endpoint; for bone health, for example, active bone resorption could be measured for a smaller, more controlled, study population. Unfortunately, the means to study active bone resorption (and early functional indices of cancer risk) are currently inadequate for this task. Bone density changes require long periods for reliable observation, and current serum markers are highly variable and thus less reliable. This indicates that new biomarkers of relevant functional endpoints important for bone health and vitamin D biological processes are needed.
Cancer researchers are leading the way toward using the techniques of genomics and proteomics to provide diagnostic markers. For example, Sorlie et al (28) used gene expression profiling of breast tissue biopsies to define the discriminating diagnostic signatures for 6 distinct classes of breast cancer, which suggests that a molecular signature could be used instead of a more subjective histologic evaluation. Although that study and other cancer-profiling studies made great use of gene expression profiles for tumor biopsies as a classification system, the need for a tissue biopsy is a serious limitation to the use of gene expression profiling as a general screening tool for healthy people. Effective biomarkers for assessing healthy people are likely to come from readily available, minimally invasive sampling of blood, blood cells, serum, or urine. The use of serum proteomic profiles, such as that used for ovarian cancer diagnosis, as reported by Petricoin et al (29), may be a more fruitful approach for issues related to nutrient status and health.
In the "omic" approach to identifying and defining biomarkers of disease or of physiologic deficits, the basic idea is to compare the profiles of individuals in 2 well-defined groups (eg, vitamin D-replete and vitamin D-deficient subjects) and then define the changes in the profiles that are correlated best with changes in the condition of interest. For continuous variables such as vitamin D status, the markers may be continuous or they may exhibit breakpoints. Although this approach may provide us with assessment parameters that may be easier to measure than vitamin D metabolite concentrations, without functional correlates this would be no more informative than serum 25(OH)D3 concentrations. Functional assessment could be included to strengthen the relationship; this general scheme is presented in Figure 4. For example, bone resorption could be measured directly through assessment of the release of calcium from bone. Ongoing work in Dr. Connie Weaver's laboratory at Purdue University suggests that a new technique for labeling bones in vivo with small amounts of the natural isotope 41Ca may provide an objective accurate measure of bone resorption (30). 41Ca is a long-lived isotope of calcium that can be produced inexpensively through neutron activation. Because miniscule amounts of 41Ca can be measured with atomic mass spectroscopy, radiologically benign amounts of the isotope can be administered to human subjects to label their bones. After 41Ca has been cleared from soft tissues (100 d), the appearance of 41Ca in the serum or urine is a direct reflection of calcium lost from bone. Because of the long half-life of 41Ca and the relatively low turnover of bone, subjects can be examined repeatedly for >10 y (permitting assessment of multiple treatments or multiple levels of a treatment for the same subject). With the exception of the long study period, this is consistent with work that was previously conducted with animals and high concentrations of 45Ca or tetracycline (31).
View larger version (22K):
FIGURE 4.. Scheme for the identification and cross-validation of serum biomarkers related to vitamin D (VD) status and bone resorption.
A discordance in the relationship between the effects of vitamin D status on bone resorption and on the serum proteome could have important physiologic implications. For example, if serum proteins changed as vitamin D status was reduced but this was not accompanied by coordinate changes in bone resorption (41Ca release), then this could be an indication that the vitamin D biomarker is related more to cancer risk (or one of the other potential functions now being proposed for vitamin D, such as diabetes mellitus risk). This could prove to be the basis for future work on the vitamin D–non-bone disease connections.
CONCLUSIONS
The field of vitamin D biology has a long history. However, we are now being confronted with both basic questions related to vitamin D actions and applied issues regarding how to use our fundamental understanding of vitamin D biological processes. Advances in chromatographic techniques have permitted us to better understand vitamin D metabolism, and the revolution in molecular biology has helped us explain the mechanistic basis for vitamin D actions on cells. In this review, I have attempted to explain how the new approaches of genomics and proteomics have the potential to expand our understanding of vitamin D biological processes and the role of vitamin D in human health.
REFERENCES
【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
From the Departments of Cardiac and Vascular Sciences (M.M., U.M., X.Y., L.L., S.F., Y.H., Q.X.) and Basic Medical Sciences (Y.-L.C., H.T., J.R.G.), St George’s, University of London, UK.
Correspondence to Dr Manuel Mayr, Department of Cardiac and Vascular Sciences, St George’s University of London, UK, Cranmer Terrace, London SW17 0RE, UK. E-mail m.mayr@sgul.ac.uk
Abstract
Objective— Proteomics and metabolomics are emerging technologies to study molecular mechanisms of diseases. We applied these techniques to identify protein and metabolite changes in vessels of apolipoprotein E–/– mice on normal chow diet.
Methods and Results— Using 2-dimensional gel electrophoresis and mass spectrometry, we identified 79 protein species that were altered during various stages of atherogenesis. Immunoglobulin deposition, redox imbalance, and impaired energy metabolism preceded lesion formation in apolipoprotein E–/– mice. Oxidative stress in the vasculature was reflected by the oxidation status of 1-Cys peroxiredoxin and correlated to the extent of lesion formation in 12-month-old apolipoprotein E–/– mice. Nuclear magnetic resonance spectroscopy revealed a decline in alanine and a depletion of the adenosine nucleotide pool in vessels of 10-week-old apolipoprotein E–/– mice. Attenuation of lesion formation was associated with alterations of NADPH generating malic enzyme, which provides reducing equivalents for lipid synthesis and glutathione recycling, and successful replenishment of the vascular energy pool.
Conclusion— Our study provides the most comprehensive dataset of protein and metabolite changes during atherogenesis published so far and highlights potential associations of immune-inflammatory responses, oxidative stress, and energy metabolism.
Our study is a first attempt to show how changes in the proteome and the metabolome are reciprocally connected during atherogenesis and provides evidence that attenuated lesion formation in apolipoprotein E–/– mice is associated with reduced oxidative stress and successful recovery of the vascular energy pool.
Key Words: animal model ? apolipoprotein E ? atherosclerosis ? metabolomics ? oxidative stress ? proteomics
Introduction
The generation of apolipoprotein E-deficient (apolipoprotein E–/–) mice1,2 has been one of the most critical advancements in the elucidation of factors affecting atherogenesis. It is currently the most popular murine model in cardiovascular research and has revealed important insights into atherosclerosis. But despite a decade of research, there is still a need for sophisticated experimental techniques to obtain a more comprehensive understanding of the complex pathophysiology.3 Previous studies have revealed apolipoprotein E-related alterations in the transcriptome.4 However, simple deduction of protein expression from mRNA transcript analysis is insufficient5 and, importantly, provides no information on post-translational modifications, which are known to be instrumental in many human diseases.
We recently analyzed the proteomic profile of mouse arterial smooth muscle that was markedly influenced by mutational ablation of the protein kinase C delta gene.6 Our proteomic findings were translated into a functional context by combining proteomics with metabolomic techniques, under in vivo7,8 as well as in vitro conditions.6 This new research strategy allows us to decipher dynamic alterations of cellular proteins and metabolites revealing multiple facets of a single pathogenesis.9
In vascular research, proteomics and metabolomics are still in their infancies. Human umbilical cord endothelial cells and arterial and saphenous vein medial smooth muscle have been scantily characterized, but most attempts to apply proteomic techniques to human atheroma were jeopardized by the accumulation of serum proteins and the genetic heterogeneity of human samples, as summarized in a recent review article.9 Thus, we decided to use a mouse model, which offers the opportunity to analyze protein changes during various stages of atherogenesis under well-defined laboratory conditions and in animals with identical genetic background facilitating proteomic comparisons by limiting biological variation.
Materials and Methods
Mice
All procedures were performed according to protocols approved by the Institutional Committee for Use and Care of Laboratory Animals. Apolipoprotein E–/– mice on a C57BL/6 background were purchased from Jackson Laboratories (West Grove, Pa) and maintained in our laboratory. Mice were fed a normal chow diet containing 4.5% fat by weight (0.02% cholesterol) and kept on a light/dark (12/12-hour) cycle at 22°C, receiving food and water ad libitum.
Assessment of Lesion Formation
Aortas from 10-week-old, 12-month-old, and 18-month-old mice were dissected from the brachiocephalic trunk to the iliac bifurcation. Macroscopically, no lesions were observed on the aortic surface of 10-week-old apolipoprotein E–/– mice. Aortic lesions in 12-month-old mice were quantified by estimating the lesion-covered area on the aortic surface (as percent of surface area) and were classified as light (<10%), medium (10% to 30%), and severe (>30%). Disease severity was further verified by oil red O staining of the aortic root (measured as averaged lesion per cross-section of the aortic sinus). The frequency of light, moderate and severe lesions in 12-month-old apolipoprotein E–/– mice on normal chow diet was 40%, 50%, and 10%, respectively. At 18 months, most apolipoprotein E–/– mice had severe lesions in their arteries.
Proteomic and Metabolomic Analysis
For proteomic and metabolomic analyses, aortas were rinsed thoroughly with cold phosphate-buffered saline to remove any blood components and frozen immediately in liquid nitrogen. Aortas from both sexes were used in all experiments. A detailed methodology is available online (http://atvb.ahajournals.org). Protocols can be downloaded from our website (www.vascular-proteomics.com).
Standard Methods
Western blotting, immunohistochemistry, and enzymatic assays are available online at http://atvb.ahajournals.org.
Statistical Analysis
Statistical analysis was performed using the analysis of variance and Student t test. Pairwise comparisons between metabolites were performed using the Bonferroni/Dunn test. Results were given as means±SE. P<0.05 was considered significant.
Results
Proteomic Analysis
To analyze changes in the proteome, we created protein profiles of aortas by 2-dimensional (2D) gel electrophoresis. Aortas were derived from 10-week-old and 12-month-old apolipoprotein E+/+ and apolipoprotein E–/– mice. Average gels were obtained from at least 4 animals per group. A direct overlay is presented in Figure 1. Using a broad range pH gradient (pH 3 to 10 NL), 2-D gels comprised 1500 protein features. Differentially expressed spots are highlighted in color (orange and blue indicate an increase in aortas of apolipoprotein E+/+ and apolipoprotein E–/– mice, respectively). Numbered spots were excised and subject to in-gel tryptic digestion. Protein identifications as obtained by mass spectrometry are listed in Table I (available online at http://atvb.ahajournals.org). For spots marked with an asterisk (*), further proof of identification was obtained by tandem mass spectrometry (n=38; Table II, available online at http://atvb.ahajournals.org). Representative spectra are shown online.
Figure 1. 2D map of proteins expressed in aortas of 10-week-old apolipoprotein E+/+ and apolipoprotein E–/– mice. Protein extracts were separated on a pH 3 to 10NL IPG strip, followed by a 12% SDS polyacrylamide gel. Spots were detected by silver staining. The figure represents a direct overlay of average gels from apolipoprotein E+/+ and apolipoprotein E–/– vessels. Average gels were created from 4 single gels (total n=8). Differentially expressed spots are highlighted in color (orange and blue for apolipoprotein E+/+ and apolipoprotein E–/– vessels, respectively). Proteins identified by mass spectrometry are marked with numbers and listed in Table I.
To quantitatively monitor protein changes during atherogenesis, aortas of 12-month-old apolipoprotein E–/– mice were classified according to their atherosclerotic surface area in vessels with light (<10%), medium (10% to 30%), and severe atherosclerosis (>30%). Confirmation was provided by oil red O staining (Figure 2A). Cholesterol levels were significantly increased in all subgroups of chow-fed apolipoprotein E–/– mice compared with wild-type controls (P<0.001 ANOVA), but no correlation was observed between cholesterol levels and disease severity in 12-month-old apolipoprotein E–/– mice (449±49 and 421±127 mg/dL in animals with light and severe lesions, respectively), which is in line with previous reports.10 Quantitative data on protein changes during disease progression are summarized in Table III (available online at http://atvb.ahajournals.org).
Figure 2. Atherosclerotic lesions in apolipoprotein E–/– mice. A, Representative photographs of oil red O-stained sections from aortic roots, indicating lesions (red color) of 10-week-old apolipoprotein E+/+ and apolipoprotein E–/– mice and 12-month-old apolipoprotein E–/– mice with light, medium, and severe disease. Original magnification x100. B and C, Western blots probed with antibodies to albumin, immunoglobulins, and apolipoprotein A1. Note that albumin undergoes extensive fragmentation in advanced lesions (C) and that immunoglobulins were barely detectable in apolipoprotein E+/+ aortas, but abundant in apolipoprotein E–/– vessels (B).
Accumulation of Serum Components
As expected, macrophage markers (MAC2, CapG) increased, whereas SMC markers (SM22) decreased, and serum proteins accumulated with lesion progression, including fibrinogen, transferrin, and hemopexin. Interestingly, immunoglobulin deposits were barely detectable in apolipoprotein E+/+ mice, but abundant even in aortas of young apolipoprotein E–/– mice (Figure 2B), forming 2 charge trains of molecular masses 25 700 and 50 800 Da with isoelectric point values of 7.8 to 5.8 on 2D gels. Whereas further immunoglobulin deposition occurred during lesion progression, albumin was subject to extensive fragmentation within advanced atherosclerotic plaques (Figure 2C). Apolipoprotein A1 (apoA1), the major protein fraction of high-density lipoprotein, whose protective anti-oxidative role in the cardiovascular system is well-established, was significantly reduced in aortas of apolipoprotein E–/– mice (Figure 2B).
Increased Oxidative Stress
Besides revealing differences in protein expression, 2D gel electrophoresis has the potential to display differences in posttranslational modifications. Redox-active cysteines constitute the main antioxidative component of peroxiredoxins.11 This protein family represents a special type of peroxidase as the protein is the reducing substrate itself; on oxidative stress, the cysteine in the active site is either oxidized to cysteine sulfenic acid or overoxidized to cysteine sulfinic acid. Whereas the first modification is DTT-sensitive and therefore undetectable in 2-D gels, the latter modification is DTT-resistant and results in a charge shift toward a more acidic isoelectric point.12 Thus, peroxiredoxins are often encountered as doublet spots in 2-D and the ratio of oxidized to reduced protein is a reliable surrogate marker for oxidative stress.11,12
1-Cys peroxiredoxin (1-Cys prx), a novel antioxidant conferring protection against oxidative membrane damage,13 was almost exclusively present as a reduced (basic) isoform in apolipoprotein E+/+ aortas, whereas oxidation of 1-Cys prx was detectable in vessels of young apolipoprotein E–/– mice, resulting in decreased abundance of the reduced isoform (Figure 3A). Consequently, the ratio of oxidized to reduced 1-Cys prx was 15-times higher in young apolipoprotein E–/– mice compared with wild-type controls (0.58±0.18 versus 0.04±0.03; P<0.05). Surprisingly, it temporarily normalized in vessels harboring light lesions (0.10±0.06 versus 0.04±0.03, nonsignificant); however, overall, there appeared to be a linear relationship between the extent of oxidation of 1-Cys prx and the extent of lesion formation in aortas of 12-month-old apolipoprotein E–/– mice (Figure 3A).
Figure 3. Oxidative stress in apolipoprotein E–/– aortas. The spot pair corresponding to 1-Cys peroxiredoxin (1-Cys prx) is marked with an arrow (A). Numbers correspond to protein identities in Table I. Quantitative changes in expression of the oxidized and reduced form of 1-Cys prx during atherogenesis are shown below. Note that 1-Cys prx is predominantly present as reduced protein in apolipoprotein E+/+ vessels but is oxidized in apolipoprotein E–/– vessels. Expression pattern of malic enzyme supernatant (MOD-1) (B). *Significant difference from wild-type controls, P<0.05, ** P<0.01.
Antioxidant Defense
1-Cys prx is able to reduce peroxidized membrane phospholipids by using glutathione (GSH) as a reductant.13 Under oxidative stress, GSH is oxidized to GSSG and subsequently reduced by GSH reductase through the coupling reaction of NADPH to NADP. Strikingly, GSH reductase activity was found to be increased in aortas of young apolipoprotein E–/– mice (78.4±12.4 IU/L versus 37.9±1.4 IU/L; n=3; P<0.05) and the oxidation state of 1-Cys prx in 12-month-old apolipoprotein E–/– mice correlated to the expression pattern of the cytosolic isoform of malic enzyme (MOD-1), which generates cytosolic NADPH,14 providing reducing equivalents for lipid synthesis and GSH recycling (Figure 3B). Aortas harboring only light lesions demonstrated a prominent change in MOD-1 (Figure 3B) associated with decreased oxidation of 1-Cys prx (Figure 3A), lower levels of the oxidative stress-induced protein HO-1 (Figure 4A and 4B), and a trend to higher GSH concentrations compared with age-matched vessels with advanced disease (42±0.9 versus 31±0.8 nmol/g wet weight; n=3; P=0.10). In contrast, upregulation of antioxidant proteins was only detectable in advanced, but not early, stages of disease (Figure 4A). This is consistent with previously published mRNA data, reporting decreased antioxidant transcription in aortas of apolipoprotein E–/– mice at the onset of lesion formation.15
Figure 4. Antioxidants in apolipoprotein E–/– aortas. Western blot analysis to determine expression levels of antioxidant proteins in aortic tissues, including heme oxygenase-1 (HO-1), superoxide dismutase-1 (SOD-1), catalase 1, and peroxiredoxin 1 (A). Note that HO-1 expression is higher in aortas with advanced atherosclerosis compared with age-matched vessels harboring only light lesions (B).
Enzymatic Changes
Among the differentially expressed proteins were several glycolytic enzymes, including triose phosphate isomerase, transketolase, glyceraldeyde-3-phosphate dehydrogenase, enolase, and phosphoglycerate mutase, as well as all 3 subunits of the pyruvate dehydrogenase complex, which accomplishes the irreversible step from glycolysis to the trichloroacetic acid (TCA) cycle by converting pyruvate to acetyl-coenzyme A (CoA). Changes of enzymes involved in glucose metabolism were accompanied by a reduction of cytoplasmic malate dehydrogenase, which is involved in the transfer of cytosolic NADH into mitochondria. Concomitantly, short chain-specific acyl-CoA dehydrogenases, responsible for the degradation of short chain fatty acids to acetyl-CoA, were differentially expressed in aortas of young apolipoprotein E–/– mice and medium chain-specific acyl-CoA dehydrogenases became upregulated in vessels of old apolipoprotein E–/–mice (Table III).
Metabolite Changes
To clarify the metabolic net effect of these enzymatic changes, we measured vascular metabolites by high-resolution nuclear magnetic resonance spectroscopy. A representative proton magnetic resonance spectrum of an aortic extract is shown in Figure 5. Quantitative data are included as Table IV (available online at http://atvb.ahajournals.org), whereas Figure 6 shows a histogram displaying the relative metabolite ratios for apolipoprotein E–/– aortas derived from 10-week-old and 18-month-old apolipoprotein E–/– mice compared with wild-type controls. Decreased concentrations of alanine, a transamination product of pyruvate, were associated with a reduction of the adenosine nucleotide pool in aortas of young apolipoprotein E–/– mice and a coordinated but nonsignificant decline of other energy metabolites, such as total creatine and succinate, the oxidation of which is directly associated with respiratory chain reactions. The ratio of alanine to pyruvate was significantly decreased in young apolipoprotein E–/– mice compared with wild-type controls (1.7±0.8 versus 7.5±1.6; P=0.002) and remained reduced in old apolipoprotein E–/– mice (4.1±1.7 versus 7.5±1.6, P=0.019, respectively), but the adenosine nucleotide and creatine pool normalized. The metabolic profiles also revealed a significant increase in choline in aortas of old apolipoprotein E–/– mice. Interestingly, concentrations of trimethylamine oxide, a breakdown product of choline, were significantly higher in male than female aortas, suggesting a gender-specific difference in choline metabolism, which was independent of the apolipoprotein E genotype (inset in Figure 6). An additional comparison of metabolic profiles obtained from sex-matched aortas of 12-month-old apolipoprotein E–/– mice (n=3 in each group, 2 males, 1 female) revealed that aortas harboring only light lesions had a 1.7- and 1.9-fold increase in adenosine nucleotides (P=0.028) and succinate (P=0.060), respectively, but only half the glucose concentration (P=0.109) compared with aortas with severe disease.
Figure 5. Nuclear magnetic resonance spectra of a murine aorta derived from 18-month-old apolipoprotein E–/– mice. Within the aliphatic region (–0.05 to 4.2 parts per million) of the nuclear magnetic resonance spectra, resonances have been assigned to lactate (Lac), alanine (Ala), pyruvate (Pyr), acetate (Acet), succinate (Succ), carnitine (Car), choline (Cho), phosphocholine (PC), taurine (Tau), scyllo-inositol (Scy-ino), glycolic acid (Glyco), trimethylamine oxide (TMAO), glutamate (Glu), creatine (Cr), phosphocreatine (PCr). ADP+ATP and formate are showing in the aromatic region of the spectra (6.0 to 9.0 parts per million, see inset).
Figure 6. Comparison of metabolites in apolipoprotein E+/+ and apolipoprotein E–/– aortas. Relative changes of metabolites in aortas derived from 10-week-old (gray bars) and 18-month-old (black bars) apolipoprotein E–/– aortas compared with apolipoprotein E+/+ aortas (reference line). Abbreviations for metabolites are explained in the legend to Figure 5. Data are provided in Table IV. The inset highlights a gender difference for TMAO concentrations in murine aortas. *Significant difference with Bonferroni/Dunn, P<0.017.
Discussion
Our study provides evidence that immune activation, oxidative stress, and energetic impairment are among the earliest alterations in hyperlipidemic animals.
Inflammation
Immunoglobulin deposition within the vessel wall of apolipoprotein E–/– mice is close to peak levels even before lesion formation initiates and cannot be accounted for by impaired endothelial barrier function, because other serum components, such as albumin and fibrinogen, did not accumulate in vessels without overt atherosclerosis. In murine models, antibodies recognizing oxidized phospholipids correlate closely with lesion progression and regression16–18 and a class shift from IgG2a to IgG1, indicative for a switch of the T-cell response from Th1 to Th2, has been observed for circulating oxidized low-density lipoprotein antibodies in apolipoprotein E–/– mice.19 Similarly, we observed a preponderance of IgG1 within atherosclerotic lesions and mass spectrometry data obtained from the variable region of accumulated immunoglobulins suggest that at least some are directed against phosphocholine (gi30720232, anti-phosphocholine immunoglobulin heavy chain variable region [mus musculus], sequence coverage 33%). However, further studies will be required to allow for a more detailed characterization.
Oxidative Stress
Oxidative stress, the local imbalance between the ubiquitous formation of reactive oxygen species (ROS) and the equally ubiquitous antioxidant defenses, is thought to play an important role in vascular injury and atherogenesis.20–22 The complexity of the antioxidant defense have made it difficult to assess their impact on atherosclerosis as it is likely that knockout of individual ROS-generating or ROS-scavenging enzymes are compensated for by synergistic ones.23 Because the pathogenetic outcome is determined by the balance between pro-oxidants and antioxidants, measurements of individual enzymes at a single time are unlikely to shed much light and a more comprehensive approach is needed.23
Our proteomic data support the role of oxidative stress in atherogenesis: First, oxidation of 1-Cys prx, a reliable in vivo marker of oxidative stress,11,12 was significantly elevated in young apolipoprotein E–/– mice compared with wild-type controls. Second, the oxidation state of 1-Cys prx correlated to lesion size in aortas of 12-month-old mice indicating that reduced oxidative stress might attenuate lesion progression in apolipoprotein E–/– mice. Third, the observed reduction of oxidative stress in vessels with light lesions was not a result of increased expression of antioxidants, because protein levels of catalase 1, SOD-1, and peroxiredoxin 1 were similar to those in young apolipoprotein E–/– mice. The oxidation status of 1-Cys prx, however, showed a striking correlation to proteomic changes of malic enzyme supernatant (MOD-1). As demonstrated previously, such changes in the protein pattern are likely to reflect alterations in enzymatic activity.6–9 The soluble form of malic enzyme is 1 of 3 enzymes, apart from glucose 6-phosphate dehydrogenase and cytoplasmic isocitrate dehydrogenase, that can generate cytosolic NADPH,14 providing reducing equivalents for lipid synthesis, as well as for glutathione and thioredoxin reductase, which are of paramount importance in maintaining the reducing intracellular environment.24,25 The importance of thiol-based defense mechanisms in hyperlipidemia is supported by the increase in glutathione reductase activity in vessels of young apolipoprotein E–/– mice. This initial glutathione defense appears to be overwhelmed in advanced stages of disease as indicated by a rebound in the oxidation of 1-Cys prx, increased expression of HO-1, and lower glutathione levels compared with aortas harboring light lesions. Thus, upregulation of antioxidant proteins appears to be the last resort, once other counter-regulatory mechanisms cannot provide sufficient reducing equivalents to antagonize ROS, rather than the first attempt to confine oxidative stress. These findings would be consistent with previous studies reporting a weak glutathione-related enzymatic antioxidant shield in human atheroma.26
Despite a conjunct upregulation of antioxidant proteins in advanced stages of disease, deleterious consequences of reactive oxygen species became apparent, such as proteolysis of oxidatively damaged proteins.27 Albumin is degraded 50-times faster on oxidation,27,28 providing a possible explanation for its extensive fragmentation. Similarly, enzymes known to be susceptible to free radical-mediated inactivation such as aconitase and the Rieske protein of ubiquinol cytochrome C reductase, which contain iron-sulfur centers, a coordination complex with cysteine sulfurs of proteins, were altered in advanced stages of atherosclerosis.29,30 Such interactions are of potential importance, as damage to such complexes results in release of free iron and subsequent formation of hydroxyl radical, a highly reactive oxygen species, which perpetuates the vicious cycle of oxidative stress.31
Energy Metabolism
Metabolic disturbances are likely to be a key factor in both the initiation and progression of atherosclerosis. The upregulation of acyl-CoA dehydrogenases, the decrease in alanine, a transamination product of pyruvate, and the downregulation of cytoplasmic malate dehydrogenase, responsible for transferring cytosolic NADH produced during glycolysis into mitochondria,32 suggest that vascular cells might respond to hyperlipidemia by metabolizing lipids instead of glucose. Increased fatty acid oxidation would exert a negative feedback on the activity of the pyruvate dehydrogenase complex32 slowing down glucose metabolism, the main source of energy for the vasculature.33 Moreover, when excess fatty acids reach mitochondria, there is even a risk of uncoupling oxidation from phosphorylation with oxygen wastage.32 Insufficient phosphorylation of energy metabolites will cause their degradation and tissue depletion, providing a possible explanation for the observed reduction of the adenosine nucleotide pool in young apolipoprotein E–/– mice. Notably, breakdown products of adenosine are xanthine and hypoxanthine, both substrates for the xanthine oxidase, which is a powerful enzyme in the generation of ROS.34 It is noteworthy that the depletion of vascular energy metabolites coincided with increased oxidation of 1-Cys prx in young apolipoprotein E–/– mice, whereas attenuated lesion formation in 12-month-old apolipoprotein E –/– mice was associated with reduced oxidative stress and successful recovery of the adenosine nucleotide pool possibly serviced by increased glucose use. Supporting our findings are previous observations that insulin supplementation reduces lesion formation and oxidative stress in apolipoprotein E–/– mice35. In contrast, overexpression of the uncoupling protein 1 results in mitochondrial dysfunction and promotes atherosclerosis by depleting energy stores and increasing superoxide production.36 Thus, there is evidence that inefficient glucose and energy metabolism may contribute to oxidative stress and vascular disease in hyperlipidemic mice.
Study Limitation
A main obstacle for applying proteomic analysis to vascular pathology is the heterogeneous cellular composition of atherosclerotic plaques. Whereas smooth muscle cells dominate proteomic profiles of normal vessels, advanced lesions contain large numbers of monocyte-derived macrophages. Overall, the proteomic profiles were remarkably consistent in young and old apolipoprotein E–/– mice: only 2 macrophage proteins, namely CapG, which accounts for 0.6% of total macrophage proteins, and MAC-2, which is also abundant in activated macrophages, showed a significant increase in advanced stages of disease, indicating that the concentration of other macrophage proteins in aortic extracts was not high enough to allow detection on 2D gels. Thus, the proteomic and metabolic profiles remain dominated by vascular smooth muscle cells, facilitating data interpretation.
Finally, we should point out that our proteomic analysis revealed differential expression of several signaling proteins, but the vascular function of some proteins, eg, dihydropyrimidinase-like proteins 2 and 3, which are regulators of neuronal development and axonal outgrowth,37 is currently unknown. For others, there is evidence for their involvement in atherosclerosis: downregulation of Ras suppressor protein 1, an endogenous inhibitor of the Ras signaling pathway, during lesion progression, is consistent with a study reporting attenuation of lesion formation in apolipoprotein E–/– by inhibiting the Ras signaling pathway;38 the functional relevance of 14-3-3 gamma, an inhibitor of the protein kinase C signaling pathway, is supported by our findings that deficiency for protein kinase C delta accelerates neointima formation in a mouse model of vein graft arteriosclerosis.39
Summary
Our study is a first attempt to show how changes in the proteome and the metabolome are reciprocally connected during atherogenesis and provides evidence that attenuated lesion formation in apolipoprotein–/– mice is associated with reduced oxidative stress and successful recovery of the vascular energy pool.
Acknowledgments
The use of the facilities of the Medical Biomics Centre at St. George’s, University of London, is gratefully acknowledged.
This work was supported by grants from the British Heart Foundation and the Oak Foundation.
References
Zhang SH, Reddick RL, Piedrahita JA, Maeda N. Spontaneous hypercholesterolemia and arterial lesions in mice lacking apolipoprotein E. Science. 1992; 258: 468–471.
Plump AS, Smith JD, Hayek T, Aalto-Setala K, Walsh A, Verstuyft JG, Rubin EM, Breslow JL. Severe hypercholesterolemia and atherosclerosis in apolipoprotein E-deficient mice created by homologous recombination in ES cells. Cell. 1992; 71: 343–353.
Meir KS, Leitersdorf E. Atherosclerosis in the apolipoprotein E-deficient mouse: a decade of progress. Arterioscler Thromb Vasc Biol. 2004; 24: 1006–1014.
Wuttge DM, Sirsjo A, Eriksson P, Stemme S. Gene expression in atherosclerotic lesion of Apolipoprotein E deficient mice. Mol Med. 2001; 7: 383–392.
Gygi SP, Rochon Y, Franza BR, Aebersold R. Correlation between protein and mRNA abundance in yeast. Mol Cell Biol. 1999; 19: 1720–1730.
Mayr M, Siow R, Chung YL, Mayr U, Griffiths JR, Xu Q. Proteomic and metabolomic analysis of vascular smooth muscle cells: role of PKCdelta. Circ Res. 2004; 94: e87–e96.
Mayr M, Metzler B, Chung YL, McGregor E, Mayr U, Troy H, Hu Y, Leitges M, Pachinger O, Griffiths JR, Dunn MJ, Xu Q. Ischemic preconditioning exaggerates cardiac damage in PKC- null mice. Am J Physiol Heart Circ Physiol. 2004; 287: H946–H956.
Mayr M, Chung YL, Mayr U, McGregor E, Troy H, Baier G, Leitges M, Dunn MJ, Griffiths JR, Xu Q. Loss of PKC- alters cardiac metabolism. Am J Physiol Heart Circ Physiol. 2004; 287: H937–H945.
Mayr M, Mayr U, Chung YL, Yin X, Griffiths JR, Xu Q. Vascular proteomics: Linking proteomic and metabolomic changes. Proteomics. 2004; 4: 3751–3761.
Tangirala RK, Rubin EM, Palinski W. Quantitation of atherosclerosis in murine models: correlation between lesions in the aortic origin and in the entire aorta, and differences in the extent of lesions between sexes in LDL receptor-deficient and apolipoprotein E-deficient mice. J Lipid Res. 1995; 36: 2320–2328.
Rabilloud T, Heller M, Gasnier F, Luche S, Rey C, Aebersold R, Benahmed M, Louisot P, Lunardi J. Proteomics analysis of cellular response to oxidative stress. Evidence for in vivo overoxidation of peroxiredoxins at their active site. J Biol Chem. 2002; 277: 19396–19401.
Wagner E, Luche S, Penna L, Chevallet M, Van Dorsselaer A, Leize-Wagner E, Rabilloud T. A method for detection of overoxidation of cysteines: peroxiredoxins are oxidized in vivo at the active-site cysteine during oxidative stress. Biochem J. 2002; 366: 777–785.
Manevich Y, Sweitzer T, Pak JH, Feinstein SI, Muzykantov V, Fisher AB. 1-Cys peroxiredoxin overexpression protects cells against phospholipid peroxidation-mediated membrane damage. Proc Natl Acad Sci U S A. 2002; 99: 11599–11604.
Lee SM, Koh HJ, Park DC, Song BJ, Huh TL, Park JW. Cytosolic NADP(+)-dependent isocitrate dehydrogenase status modulates oxidative damage to cells. Free Radic Biol Med. 2002; 32: 1185–1196.
t Hoen PA, Van der Lans CA, Van Eck M, Bijsterbosch MK, Van Berkel TJ, Twisk J. Aorta of Apolipoprotein E-deficient mice responds to atherogenic stimuli by a prelesional increase and subsequent decrease in the expression of antioxidant enzymes. Circ Res. 2003; 93: 262–269.
Palinski W, Ord VA, Plump AS, Breslow JL, Steinberg D, Witztum JL. Apolipoprotein E-deficient mice are a model of lipoprotein oxidation in atherogenesis. Demonstration of oxidation-specific epitopes in lesions and high titers of autoantibodies to malondialdehyde-lysine in serum. Arterioscler Thromb. 1994; 14: 605–616.
Shaw PX, Horkko S, Chang MK, Curtiss LK, Palinski W, Silverman GJ, Witztum JL. Natural antibodies with the T15 idiotype may act in atherosclerosis, apoptotic clearance, and protective immunity. J Clin Invest. 2000; 105: 1731–1740.
Palinski W, Tangirala RK, Miller E, Young SG, Witztum JL. Increased autoantibody titers against epitopes of oxidized LDL in LDL receptor-deficient mice with increased atherosclerosis. Arterioscler Thromb Vasc Biol. 1995; 15: 1569–1576.
Zhou X, Paulsson G, Stemme S, Hansson GK. Hypercholesterolemia is associated with a T helper (Th) 1/Th2 switch of the autoimmune response in atherosclerotic apo E-knockout mice. J Clin Invest. 1998; 101: 1717–1725.
Touyz RM. Reactive oxygen species, vascular oxidative stress, and redox signaling in hypertension: what is the clinical significance? Hypertension. 2004; 44: 248–252.
Mayr M, Hu Y, Hainaut H, Xu Q. Mechanical stress-induced DNA damage and rac-p38MAPK signal pathways mediate p53-dependent apoptosis in vascular smooth muscle cells. FASEB J. 2002; 16: 1423–1425.
Khatri JJ, Johnson C, Magid R, Lessner SM, Laude KM, Dikalov SI, Harrison DG, Sung HJ, Rong Y, Galis ZS. Vascular oxidant stress enhances progression and angiogenesis of experimental atheroma. Circulation. 2004; 109: 520–525.
Palinski W. United they go: conjunct regulation of aortic antioxidant enzymes during atherogenesis. Circ Res. 2003; 93: 183–185.
Powell LA, Nally SM, McMaster D, Catherwood MA, Trimble ER. Restoration of glutathione levels in vascular smooth muscle cells exposed to high glucose conditions. Free Radic Biol Med. 2001; 31: 1149–1155.
Okuda M, Inoue N, Azumi H, Seno T, Sumi Y, Hirata K, Kawashima S, Hayashi Y, Itoh H, Yodoi J, Yokoyama M. Expression of glutaredoxin in human coronary arteries: its potential role in antioxidant protection against atherosclerosis. Arterioscler Thromb Vasc Biol. 2001; 21: 1483–1487.
Lapenna D, de Gioia S, Ciofani G, Mezzetti A, Ucchino S, Calafiore AM, Napolitano AM, Di Ilio C, Cuccurullo F. Glutathione-related antioxidant defenses in human atherosclerotic plaques. Circulation. 1998; 97: 1930–1934.
Davies KJ. Protein damage and degradation by oxygen radicals. I. General aspects. J Biol Chem. 1987; 262: 9895–9901.
Davies KJ, Goldberg AL. Proteins damaged by oxygen radicals are rapidly degraded in extracts of red blood cells. J Biol Chem. 1987; 262: 8227–8234.
Grune T, Blasig IE, Sitte N, Roloff B, Haseloff R, Davies KJ. Peroxynitrite increases the degradation of aconitase and other cellular proteins by proteasome. J Biol Chem. 1998; 273: 10857–10862.
Zhang Y, Marcillat O, Giulivi C, Ernster L, Davies KJ. The oxidative inactivation of mitochondrial electron transport chain components and ATPase. J Biol Chem. 1990; 265: 16330–16336.
Liochev SL. The role of iron-sulfur clusters in in vivo hydroxyl radical production. Free Radic Res. 1996; 25: 369–384.
Opie LH. Heart physiology: from cell to circulation. Philadelphia: Lippincott Williams & Wilkins; 2004.
Lynch RM, Paul RJ. Compartmentation of glycolytic and glycogenolytic metabolism in vascular smooth muscle. Science. 1983; 222: 1344–1346.
Cai H, Harrison DG. Endothelial dysfunction in cardiovascular diseases: the role of oxidant stress. Circ Res. 2000; 87: 840–844.
Shamir R, Shehadeh N, Rosenblat M, Eshach-Adiv O, Coleman R, Kaplan M, Hamoud S, Lischinsky S, Hayek T. Oral insulin supplementation attenuates atherosclerosis progression in apolipoprotein E-deficient mice. Arterioscler Thromb Vasc Biol. 2003; 23: 104–110.
Bernal-Mizrachi C, Gates AC, Weng S, Imamura T, Knutsen RH, DeSantis P, Coleman T, Townsend RR, Muglia LJ, Semenkovich CF. Vascular respiratory uncoupling increases blood pressure and atherosclerosis. Nature. 2005; 435: 502–506.
Byk T, Ozon S, Sobel A. The Ulip family phosphoproteins-common and specific properties. Eur J Biochem. 1998; 254: 14–24.
Indolfi C, Avvedimento EV, Rapacciuolo A, Di Lorenzo E, Esposito G, Stabile E, Feliciello A, Mele E, Giuliano P, Condorelli G, et al. Inhibition of cellular ras prevents smooth muscle cell proliferation after vascular injury in vivo. Nat Med. 1995; 1: 541–545.
Leitges M, Mayr M, Braun U, Mayr U, Li C, Pfister G, Ghaffari-Tabrizi N, Baier G, Hu Y, Xu Q. Exacerbated vein graft arteriosclerosis in protein kinase Cdelta-null mice. J Clin Invest. 2001; 108: 1505–1512.
Proteome Systems, Ltd
Department of Respiratory Medicine, Royal Prince Alfred Hospital, Sydney, Australia
Department of Pediatrics, University of California, San Francisco, San Francisco, California
ABSTRACT
Rationale: Recurrent pulmonary exacerbations are associated with progressive lung disease in cystic fibrosis (CF). Current definitions of an exacerbation, although not precisely defined, include new/worsening symptoms, declining lung function, and/or changing radiologic appearance. Early diagnosis of exacerbations by rapid noninvasive means should expedite therapeutic intervention, thereby minimizing lung damage.
Objectives: To identify biomarkers of lung exacerbation for point-of-care monitoring of CF lung disease progression.
Methods: Saline-induced sputum was collected from adults with CF with an exacerbation and requiring hospitalization (FEV1 < 60%), a subset of these adults at hospital discharge, children with stable CF and preserved lung function (FEV1 > 70%), and control subjects (FEV1 > 80%). Sputum was arrayed by two-dimensional electrophoresis and differentially expressed proteins were identified by proteomic analysis.
Measurements and Main Results: Sputum profiles from adults with CF with an exacerbation were characterized by extensive proteolytic degradation and influx of inflammation-related proteins, with some adults with CF approaching a "healthy" protein profile after hospitalization. Two children with CF showed profiles and biomarker expression resembling those of adults with an exacerbation. Levels of differentially expressed myeloperoxidase, cleaved 1-antitrypsin, IgG degradation, interleukin-8, and total protein concentration, together with their correlation to FEV1, were statistically significant. Statistical correlation analyses indicated that changes in myeloperoxidase expression and IgG degradation were the strongest predictors of FEV1.
Conclusions: We identified extensive protein degradation and differentially expressed proteins as biomarkers of inflammation relating to pulmonary exacerbations. Prediction of exacerbation onset and more precise evaluation of the extent of resolution with treatment could be achieved by including biomarkers in standard assessment.
Key Words: 1-antitrypsin exacerbation immunoglobulin inflammation myeloperoxidase
Cystic fibrosis (CF) is one of the most common lethal genetic diseases in white individuals, with a carrier rate of approximately 4 to 5% and an incidence of approximately 1 in 2,500. CF is caused by mutations in the gene encoding the CF transmembrane conductance regulator (CFTR) (1). Symptoms include salty-tasting skin; persistent coughing, sometimes with phlegm; wheezing; shortness of breath; malnutrition; abdominal pain; and frequent, bulky, greasy stools (1). Symptoms vary between individuals, partly due to more than 1,000 known mutations of the CFTR gene. The F508 mutation accounts for approximately 70% of the CFTR mutations in white populations (2, 3).
Progressive lung disease is the major cause of morbidity and mortality in patients with CF (1). Airways become colonized with bacteria, particularly Pseudomonas aeruginosa and Staphylococcus aureus, while recurrent pulmonary infections and inflammation result in submucosal gland hypertrophy and excessive mucous secretion in the lungs (4, 5). Impaired mucociliary clearance and plugging of small airways cause progressive bronchiectasis, ultimately resulting in respiratory failure (6, 7). Lung disease starts as early as the first few months of life and is difficult to detect without invasive techniques such as flexible bronchoscopy and bronchoalveolar lavage (8). Although CF can be diagnosed in newborns by genetic screening (9), therapy is directed by evaluation, which includes review of symptoms, lung function, and to a lesser extent radiologic changes, and is therefore likely to lag behind the occurrence of established lung pathology.
Proteins are the ultimate product of gene expression and the development of prognostic tests and drugs for CF will occur through a greater understanding of the proteins and their interactions within the lung environment. Proteomics provides the ability to characterize proteins and their post-translational modifications and offers a greater understanding of the physiology of the lung environment. For complex solutions, such as sputum, two-dimensional (2-D) electrophoresis represents a key technology of choice for arraying and characterizing constituent proteins, and has been used to help characterize protein expression in bronchoalveolar lavage fluid (BALF) from individuals with CF (10, 11). This study, which represents the first proteomic study to address differential sputum proteomes in the context of subjects with CF versus control individuals, aims to identify protein biomarkers that are indicative of an acute pulmonary exacerbation. Sputum protein profiles from subjects with CF with an exacerbation have been compared with those from a subgroup of the same subjects with CF after hospital treatment, with clinically stable children with CF with preserved lung function, and with control subjects to further elucidate changes in protein profiles and expression as markers of disease progression.
Understanding changes in protein expression with pulmonary disease will permit development of clinical assays for rapid, noninvasive analysis of fluids such as blood, sputum, or saliva. This will help elicit early intervention of severe airway infection and/or acute inflammatory responses and help dictate short- and long-term therapy for CF lung disease. Minimizing cumulative pulmonary deterioration from the recurring cycle of infection and inflammation will ultimately help prolong the length and improve the quality of life for an individual with CF (5, 12).
METHODS
Clinical Samples
Saline-induced sputum from human subjects was collected according to previously described methods (13, 14) from healthy control adults (18–40 yr of age; n = 20) and children (8–14 yr of age; n = 5) with forced expiratory volume in 1 s (FEV1, %pred) greater than 80%; adults with CF and an acute pulmonary exacerbation (15, 16), FEV1 less than 60%, and requiring hospitalization (n = 20); a subset of these adults with CF also at the time of hospital discharge (n = 13); and children with CF without clinical evidence of an exacerbation (n = 7), FEV1 greater than 70% (Table 1). Sputum was immediately placed on ice and then solubilized within 1 h of collection. Sputum was solubilized for 1 h at 4°C, using methods similar to those previously described (13, 14). A complete ethylenediaminetetraacetic acid–free protease inhibitor cocktail tablet (Roche Molecular Biochemicals, Mannheim, Germany) was added to sputum samples to prevent proteolytic degradation during solubilization. Cell and bacterial debris was subsequently removed by centrifugation at 2,000 x g (10 min at 4°C) and 0.2-μm pore size filtration before proteomic analysis. An identical clinical protocol was used for collection of samples from adults and children. On-site training ensured standardization of procedures at all sites of sample preparation. Sputum was qualified using a criterion of a squamous cell count less than 80% (14). Subjects with CF were excluded from this study if they had any other coexisting acute or chronic illnesses. Institutional human research ethics committees approved human subject recruitment and research involving human samples for these studies. Written consent was obtained from all subjects (or their legal guardians) participating in this study. These studies were conducted in accordance with the World Medical Association Declaration of Helsinki regarding ethical principles for medical research involving human subjects.
Sample Preparation and 2-D Gel Electrophoresis
Chemicals were obtained from Sigma-Aldrich (St. Louis, MO) unless specified otherwise. After liquefaction, sputum proteins were resuspended in ProteomIQ CHAPS resuspension reagent (Proteome Systems, Inc., Woburn, MA) with 40 mM TRIS and reduced and alkylated with 5 mM tributylphosphine and 10 mM acrylamide for 1 h at room temperature. Samples were then desalted to remove TRIS and subsequently analyzed by 2-D electrophoresis (2-DE), using 11-cm IPG strips, pH 4–7, with either 6–15 or 14% polyacrylamide GelChIP gels (Proteome Systems Ltd, Sydney, Australia) using IsoelectrIQ2 and ElectrophoretIQ3 devices with a ProteomIQ platform (Proteome Systems Ltd) as previously described (17). Proteins (300 μg/gel) were visualized by staining gels with both SYPRO Ruby (Molecular Probes/Invitrogen, Eugene, OR) and Coomassie Brilliant Blue G-250. Differential protein expression was determined with ImagepIQ (Proteome Systems Ltd). Protein spots present in the majority of the CF adult (exacerbation) gels and whose expression levels showed distinct up- or downregulation when compared with respective gels at discharge and control subjects were selected for further analysis. Gel pieces were excised, destained, digested with trypsin, and desalted with Xcise (Proteome Systems Ltd and Shimadzu-Biotech, Kyoto, Japan) with a ProteomIQ Xcise in-gel digest kit (Proteome Systems, Inc.).
Mass Spectrometry
Protein digests were analyzed with an Axima-CFR matrix-assisted laser desorption/ionization time of flight (MALDI-TOF) mass spectrometer (Kratos, Manchester, UK) as previously described (17). All solutions used for MALDI-TOF mass spectrometry analysis were from a ProteomIQ Xcise in-gel digest kit (Proteome Systems, Inc.). BioinformatIQ (Proteome Systems Ltd) was used for data analysis and tracking. Protein identifications were confirmed by nanoflow liquid chromatography–mass spectrometry with an LCQ DECA ion trap mass spectrometer (Thermo Electron Corp., Waltham, MA; see the online supplement for further detail).
ELISA and Western Blotting
ELISA.
Myeloperoxidase levels in liquefied sputum were measured by sandwich ELISA using the 2C7 anti-human myeloperoxidase monoclonal antibody (Serotec, Oxford, UK). A rabbit anti-human myeloperoxidase polyclonal antibody (Chemicon International, Inc., Temecula, CA) and a sheep anti-rabbit horseradish peroxidase–conjugated antibody (Chemicon International) were used for detection. Interleukin-8 (IL-8) levels were measured with an IL-8 EASIA kit (BioSource, Nivelles, Belgium). See the online supplement for further detail.
Western blotting.
Subsequent to 1-DE, sputum was analyzed by Western blotting with a sheep anti-mouse horseradish peroxidase–conjugated antibody (Chemicon International) to measure IgG expression (see the online supplement for further detail).
Statistical Analyses
Incidences and spot volumes of each protein in control and CF groups, determined by image analysis, were compared by Fisher's exact test and Student t test, respectively. The Student t test was based on two-way analysis of variance, which included terms for disease status, age (adult vs. child), and their interaction. Strong control of the family-wise type I error rate for each term (age, disease status, and interaction) was maintained by Holm's adjustment (18), applied to all the protein spot volumes. This procedure is a stochastically dominant modification of the Bonferroni procedure for testing each hypothesis at a p value of 0.05/K, where K is the number of protein spots analyzed. Statistical analyses were performed with S-Plus 6 software (Insightful Corporation, Seattle, WA).
Differences between biomarker concentration and disease groups.
Data were analyzed with a linear model, applied separately for children and for adults. The linear model for children was a fixed effects linear model, with a term representing disease status. The mixed model was fitted with the residual maximum likelihood algorithm (19) as implemented in the R package nlme (20) (see the online supplement for further detail).
Relationships between biomarker concentration and lung function.
The relationship of each biomarker with FEV1 was assessed by two statistical tests.
Ignoring Other Markers: The effect of each marker was tested by looking at the reduction in sums of squares obtained when the marker is added to a model containing only the age effect.
Eliminating Other Markers: The effect of each marker was tested by looking at the reduction in sums of squares obtained when the marker is added to a model containing age and all the other markers. In addition, Akaike's information criterion was used in backward elimination, to select variables for inclusion in a final model (see the online supplement for further detail).
RESULTS
2-D Profiling of Sputum
Extensive differences in sputum protein profiles between adults with CF with an exacerbation and control subjects were observed (Figure 1). Destruction of whole proteins resulting from proteolytic degradation in sputum from adults with CF with an exacerbation was evidenced by the relative increase in low-molecular-weight protein. Disappearance of trains of protein spots was consistently observed in these 2-D profiles compared with control subjects. An influx of neutrophil-derived proteins, identified by mass spectrometry, into the sputum of subjects with CF with an exacerbation was also apparent (Figures 1C–1F). The number of protein spots observed in the 2-D profiles was different for each of the subject groups analyzed. The average number of spots for each group (mean ± SD) was as follows: adults with CF with exacerbation, 478 ± 124; adults with CF at discharge, 507 ± 101; control adults, 338 ± 64; children with CF, 441 ± 75; control children, 382 ± 51.
Sputum from 13 of the 20 adults with CF was also collected at hospital discharge, approximately 2 wk after admission. Sputum profiles from four of these discharged adults with CF, particularly subject 11 (Figure 1D), resembled those of control subjects (Table 1). Profiles from eight discharged adults with CF did not show clear evidence of proteomic improvement despite an increase in FEV1 (Table 1). Interestingly, CF subject 39 had an improvement in FEV1 from 66.9 to 82.7% at discharge, yet displayed minimal change in sputum profile (Figures 1E and 1F); the remaining inflammation-derived proteins, which will continue to cause tissue damage, suggests poor lung recovery from the recent exacerbation. Adult CF subject 37 was diagnosed with a viral infection 3 d before discharge. This was reflected by a decline in respiratory status, determined by FEV1 and sputum profiling, which showed increased levels of protein degradation and neutrophil-derived proteins.
Clinical criteria were used to define children with CF with stable disease and preserved lung function as a second control group for the adults with CF with an exacerbation. Despite all children with CF having clinical measures, particularly FEV1, similar to those of control subjects (Table 1), CF subjects 64 and 69 presented sputum profiles clearly indicating signs of inflammation as observed for adults with CF with an exacerbation (Figures 1I–1J). Thus, although FEV1 measurements indicated healthy respiratory capacity, proteomic data indicated early signs of inflammation and/or infection. Interestingly, CF Child 64 was clinically diagnosed as having a flare of allergic bronchopulmonary aspergillosis 96 d after proteomic analysis. An elevation of total IgE associated with a drop in lung function, new infiltrates, and acute respiratory signs were found at the time of the flare. Child CF subject 69 was clinically diagnosed 49 d later with a flare of S. aureus infection. The other children with CF remained stable during these time frames.
Biomarker Identification
We identified a number of neutrophil-derived and inflammation-associated proteins that were differentially expressed between adults with CF with an exacerbation and control subjects at statistically significant levels (p 0.005), using 2-DE–based proteomics. These differences were also observed in child CF subjects 64 and 69 relative to control subjects. We have characterized three of these proteins, myeloperoxidase, 1-antitrypsin, and IgG, for this study.
Differential expression of myeloperoxidase in sputum between CF and control subjects was confirmed by ELISA (Figure 2A). Mean absorbance (± SD) for myeloperoxidase levels in adults with CF with an exacerbation was 2.74 ± 0.22 AU compared with 0.37 ± 0.10 AU in control adults. Adult CF subjects 11, 44, and 46 showed a significant decrease in expression of myeloperoxidase after hospitalization, whereas other adults with CF showed minimal change. These patterns closely support clinical data and observations made by 2-D profiling of sputum (Table 1). In contrast, discharged adult CF subject 39, with clinical recovery from an exacerbation, showed no change whereas CF child subjects 64 and 69, meeting the definition of stable disease, showed relative increased levels of myeloperoxidase (Figure 2A). As a known marker of inflammation (21), IL-8 concentrations in sputum were measured by ELISA as a comparison (Figure 2B). Expression patterns of myeloperoxidase versus IL-8 were found to be similar for the respective children. For the adults, other than CF subjects 39 and 41, there was a clear drop in IL-8 for subjects with CF at discharge relative to the exacerbation time point. Interestingly, the change in expression of IL-8 for CF subjects 12, 20, and 46 was greater for IL-8 than for myeloperoxidase, yet for CF subject 44 the change in myeloperoxidase expression was greater at discharge. For CF subject 41, IL-8 concentration was close to that of control subjects, despite 2-DE profiles and myeloperoxidase levels clearly indicating extensive inflammation. IL-8 concentration in sputum from CF subject 39 more closely mirrored the profile observed by 2-DE, which worsened at discharge, despite an improvement in FEV1. These data suggest that these proteins may be markers of different, albeit related, stages of the inflammation process.
Protein spots corresponding to putative proteolytic cleavage products of 1-antitrypsin were observed in adults with CF with an exacerbation and in CF Child subjects 64 and 69, as defined by a decrease in both pI (0.5 units) and molecular mass (3 kD; Figure 3A). Measurement of noncleaved 1-antitrypsin showed a large disease status effect (CF vs. control) for adults and children (p = 0.007); however, there was a clear lack of age (adult vs. child) by CF/control status interaction (Figure 3B). In contrast, measurement of cleaved 1-antitrypsin or the ratio of noncleaved to cleaved 1-antitrypsin showed a clear CF versus control effect (both p < 0.001) as well as an age by CF/control status interaction that was also statistically significant (p < 0.002 and p < 0.022, respectively; Figures 3C and 3D). Adult and child control subjects had similar levels of cleaved 1-antitrypsin, but adults with CF with an exacerbation had much higher levels than did children with CF. Normalized gel spot volumes of cleaved 1-antitrypsin or the ratio of noncleaved to cleaved 1-antitrypsin also distinguished between adult and child CF cohorts (p < 0.002 and p < 0.006, respectively; Figures 3C and 3D). Clearly the ratio response is driven by the cleaved 1-antitrypsin response.
Degradation of Immunoglobulin in Children with CF
One-dimensional gel analysis of sputum was performed to determine the IgG profiles between CF and control adults and children. Western blotting and MALDI-TOF mass spectrometry analyses identified numerous IgG-1 heavy-chain fragments in sputum, molecular mass about 25 to 45 kD, from all adults with CF with an exacerbation but only full-length chains in adult control subjects (Figure 4A). The patterns of degradation, quantified through calculation of a degradation ratio, closely mirrored the 2-D profiles at exacerbation and discharge described in Figure 1, where IgG degradation is clearly distinguishable from the lack thereof in control subjects. Subjects 11, 41, and 44 show a clearance of this degradation pattern at discharge. Similar analyses of sputum from the children revealed that the two children with CF with proteomic profiles similar to those of adults with an exacerbation (i.e., subjects 64 and 69) also expressed similar fragments of IgG heavy chain (Figure 4B). Sputum from all other children with CF and control children contained only full-length IgG heavy and light chains.
Statistical Analyses: Relationships between Disease Groups, Biomarker Expression, and Lung Function
Differences between disease groups.
Combined analysis of expression of total protein concentration, myeloperoxidase, IL-8, IgG degradation, and cleaved 1-antitrypsin all showed statistically significant differences between disease groups for the adults (Table 2). For each biomarker, values were smallest in control groups and largest in the CF exacerbation group. The values for adults with CF at discharge were in between those of the control and CF exacerbation groups. For FEV1, again the adults showed marked statistically significant differences between disease groups, with FEV1 values for subjects with CF experiencing an exacerbation being the worst (smallest) and values for control subjects being the best (highest; Table 2). Differences for children, in biomarker concentrations or FEV1, were not statistically significant, although they were almost so for IL-8 (Table 2).
Relationships between biomarker concentration and lung function.
Tests for the relationship of each biomarker, including total sputum protein concentration, with FEV1, both ignoring and eliminating the effects of the other markers, are shown in Table 3A. It is clear that each of the biomarkers is statistically significantly associated with FEV1 (each of the biomarkers has statistically significant F values, ignoring the other markers). However, because of the strong correlation between biomarkers, no biomarker is statistically significant when added to a model that includes the other biomarkers. The implication is that at least one of the biomarkers may be used to predict FEV1, but that the full set of biomarkers is not required to achieve this. When biomarkers were selected with Akaike's information criterion, only two proteins were included in the model: log myeloperoxidase optical density and IgG degradation. The other biomarkers (IL-8 and total protein concentration) were dropped. That is, the use of Akaike's information criterion suggested that myeloperoxidase concentration and IgG degradation have a stronger relationship with FEV1 than does adult/child status, IL-8, or total protein concentration.
When a dataset containing matched 1-antitrypsin data was further analyzed, each of the biomarkers was again statistically significant when considered while ignoring the effects of the others, but none were statistically significant when considered while eliminating the effect of the other markers (Table 3B). The strongest marginal relationship was between FEV1 and the ratio of noncleaved to cleaved 1-antitrypsin. However, backward elimination based on Akaike's information criterion for the data in Table 3B resulted in a model containing only log myeloperoxidase optical density. This is apparently in contradiction to the latter results, in which the strongest single relationship was between FEV1 and the ratio of noncleaved to cleaved 1-antitrypsin. This is an example of the backward elimination process failing to find the best overall solution for a given number of variables and is typical of the problems of interpreting multiple strongly correlated predictor variables. These results can best be interpreted as suggesting a strong relationship between both myeloperoxidase and the ratio of noncleaved to cleaved 1-antitrypsin and FEV1, but with little to choose between the markers.
DISCUSSION
Proteomic-based approaches have been used to discover biomarkers of disease in a wide range of diseases, including cancer (22), neurologic disorders (23), aging (24), heart disease (25), and lung disorders (26). In particular, 2-DE is a powerful proteomic tool with which to visualize modified forms of proteins and to compare proteomic profiles for diseased and nondiseased states (27). For CF, a number of proteomic studies have focused on analysis of protein expression in BALF (28–30). Altered expression of surfactant proteins SP-A and SP-D in BALF has been demonstrated by 2-DE and high-resolution mass spectrometry studies (10, 11, 31). For development of biomarker screening tests for respiratory diseases such as CF, particularly for point of care, sputum is a more amenable sample compared with BALF given it can be easily collected by noninvasive means. Sputum collected by saline induction has been shown to be a valuable tool for sampling and analysis of the contents of the lower airway (32, 33). In this study, saline-induced sputum profiles from adults with CF with an exacerbation and from children with CF with stable disease and preserved lung function were compared with profiles from adult and child control subjects. Furthermore, we have highlighted examples of protein isoforms that are differentially expressed between these different subject groups.
Cumulative lung damage in CF results from continual cycles of infection and inflammation that occur throughout an individual's lifetime (12), and is particularly characterized by a marked increase and persistent influx of neutrophils into the airways with consequent release of noxious mediators such as reactive oxygen species and proteolytic enzymes (34). At present, no quantitative test for point-of-care monitoring of CF lung disease is commercially available.
Despite the striking differences in 2-DE sputum protein profiles between the different subject groups, it could be speculated that such differences, particularly between adults with CF and children with CF, may be artifacts of sampling handling and processing times at different study sites. We believe this to be unlikely given the same protocols were standardized across all sites. Although there may have been slight variations in the time between sputum collection and the start of sample solubilization, albeit within 1 h, samples were always immediately placed on ice on collection. This would have minimized any further proteolytic degradation to an extent far less than what would have already occurred in the actual lung environment.
There is no "gold standard" for defining a pulmonary exacerbation. Diagnosis is based on a number of variables, including subjective measures of symptoms and clinical history (15, 16, 35). Measurement of FEV1 is widely used to monitor changes in respiratory condition in CF (15, 16, 35). Although decreasing or increasing FEV1 values help define a pulmonary exacerbation or recovery therefrom, respectively, we have shown by proteomic profiling of sputum that there are discrepancies between results when using proteomics and measurements of spirometry.
These studies demonstrate the feasibility of using biomarkers to monitor the dynamic biological processes in the lung that impact on tissue quality and ultimately respiratory function. Sputum profiles from adults with CF with an exacerbation demonstrated considerable protein expression differences after hospitalization and from control subjects. Here we have presented data for myeloperoxidase, 1-antitrypsin, IgG degradation, and total protein concentration in comparison with IL-8, as biomarkers of lung exacerbation and examples of protein modifications that could be used in a prognostic/diagnostic test.
Although we have shown that increasing levels of myeloperoxidase, a protein involved in the inflammatory response by breaking down peroxide, are indicative of an exacerbation, decreasing levels are suggestive of improving pulmonary status. High levels of myeloperoxidase remained in certain adults with CF after hospitalization, suggesting insufficient clearance of inflammation. Adult CF subject 39, for example, presented a sputum profile at discharge indicating the possibility for further inflammation-induced tissue damage, despite a 23.6% increase in FEV1 after hospitalization. This highlights the utility of biomarkers for helping to determine the length of drug treatment times and for regular monitoring for exacerbation and inflammation, particularly given current limitations in defining a pulmonary exacerbation (15, 16).
Comparative measures of IL-8, the major neutrophil chemoattractant peptide and a previously reported marker of inflammation in CF (33, 36, 37), show slightly different patterns relative to myeloperoxidase expression. For adults with CF, the differential expression at the exacerbation and discharge time points versus that of myeloperoxidase was greater for the majority of subjects with CF. The more notable discrepancies in expression of myeloperoxidase versus IL-8 were for subjects 12, 39, 44, and in particular subject 41, in whom minimal IL-8 was detected. One can only speculate as to reasons for these differences. It is unlikely that protein half-life accounts for these differences as both proteins have estimated half-lives of 30 h (http://us.expasy.org/tools/protparam.html). It is possible that IL-8 and myeloperoxidase are biomarkers of slightly different stages of inflammation and/or exacerbation. IL-8 is more likely an indicator of early stage inflammation and exacerbation, given its role in recruitment of neutrophils, whereas myeloperoxidase possibly represents a biomarker of the underlying neutrophil influx as a consequence of inflammation at exacerbation. Further studies would be required to resolve this. In contrast to findings by Wolter and coworkers (38), but in agreement with others (33, 36, 37), our findings also suggest IL-8 as a biomarker of exacerbation (17). Nevertheless, our statistical analyses demonstrated that myeloperoxidase, and IgG degradation, were stronger predictors of FEV1 than was IL-8.
A number of potential biomarkers of exacerbation in CF have been previously elucidated, particularly in association with inflammation and oxidative stress (32, 33, 39–48). Nevertheless, none of these markers have been developed into a rapid point-of-care test for patients with CF, often because of a paucity of statistical correlation (38). Myeloperoxidase has been reported as a marker for oxidative stress in inflammation (41, 49–53). With inflammation being a major cause of pulmonary deterioration (5, 34), and myeloperoxidase levels fluctuating as a measurable outcome of inflammation, myeloperoxidase represents an appealing biomarker of CF lung disease progression, which may be used in conjunction with FEV1 measurements.
Cleavage of 1-antitrypsin can be equally effective in monitoring pulmonary status in subjects with CF. 1-Antitrypsin, a major protease inhibitor in the respiratory tract, is known to be cleaved by site-specific neutrophil-derived and bacterial proteases, including P. aeruginosa elastase (54). Expression of cleaved 1-antitrypsin can distinguish between control subjects and subjects with CF and severity of disease (i.e., adults with CF with an exacerbation versus clinically stable children with CF). Many CF markers receive criticism given that patient-to-patient variability and day-to-day fluctuations in absolute biomarker expression can compromise their prognostic value. The relationship between subject status and age effects still held after analysis of the ratio of cleaved to noncleaved forms of 1-antitrypsin, confirming that ratio measurements will help control for these variations. Like the other markers, including total protein concentration, we demonstrated a statistically significant correlation between 1-antitrypsin and FEV1, more so than IL-8; however, it was not as strong a predictor of FEV1 as myeloperoxidase or IgG degradation.
Although there are a number of proteomic studies that have started to analyze BALF (10, 11, 28, 31), this is the first proteomic study to address changes in protein expression in sputum, particularly in the context of CF at different time points relative to healthy subjects. Studies of BALF in CF have demonstrated significant differences in protein patterns between healthy control and clinically stable subjects with CF, with the latter showing a predominance of low molecular weight proteins as we have observed in sputum (11). In agreement with our findings, studies of 2-D protein patterns of BALF, including analysis of proteolysis of SP-A, have also demonstrated how changes in protein patterns can be used to monitor molecular changes as a consequence of therapeutic intervention (10).
Longitudinal monitoring of IgG degradation could have important prognostic value for pulmonary status in chronically infected subjects with CF. Several human pathogens, including P. aeruginosa, Haemophilus influenzae, and various species of Streptococcus, encode proteases that cleave the heavy chains of IgG and IgA (55–58). These proteases are important bacterial virulence factors, allowing these pathogens to evade host defense mechanisms. Bacterially triggered degradation of immunoglobulins is thought to promote colonization and invasiveness at mucosal surfaces (55–58). Of the seven children with CF analyzed, IgG degradation was observed in sputum only from subjects 64 and 69, who both presented 2-DE profiles similar to those of adults with CF with an exacerbation. Degradation was observed in all the adults with CF with an exacerbation with degradation patterns closely relating to the proteomic patterns observed by 2-DE, with adult CF subjects 11, 41, and 44 clearly showing a reduced level of degradation at discharge. Degradation of IgA was also observed in sputum from all adults with CF with an exacerbation but not in any control subjects analyzed (data not shown). Although not extensively characterized in this study, the degradation of the immunoglobulin heavy chain is most likely due to the effective proteolytic activity of the bacterial proteases, concerning which a high degree of specificity for immunoglobulin molecules has been previously described (55, 56, 58).
Abnormalities in expression of IgG heavy chains have been associated with other inflammatory diseases such as chronic arthritis and rheumatoid arthritis, as well as heavy-chain diseases and amyloidosis (59–63). An impaired clearance of IgG4 has also been reported with inflammatory bowel disease (62). Without knowledge of the cleavage sites, quantitation of these fragments in an immunodiagnostic test would be challenging. Nevertheless, as we have demonstrated, an electrophoresis-based assay could be developed as an alternative test for quantifying fragmentation patterns relating to lung disease, particularly given the strong correlation between IgG degradation and FEV1.
Child CF subjects 64 and 69 had clinical symptoms and FEV1 values within the normal range comparable to those of the other children with CF with stable disease and control subjects, yet their sputum protein profiles resembled those of adults with an exacerbation, with evidence of inflammation and protein degradation. Despite all clinical signs indicating a nonexacerbated state, without signs of acute inflammation, these two children with CF had higher levels of myeloperoxidase and IL-8, increased levels of cleaved 1-antitrypsin, and IgG degradation, all of which indicated propensity for near-term exacerbation. Downstream diagnosis of flares of infection in only these two children with CF is interesting from a prognostic perspective; however, proof of causality, combined with a larger sample set and further longitudinal monitoring, is required.
Multiplexing these and/or other disease markers (33) for simultaneously monitoring infection, inflammation, and lung degradation might better predict exacerbation, thereby permitting early medical intervention before clinical symptoms develop, consequently minimizing tissue damage and lung pathology. Longitudinal studies, both in children and adults, will now be required to further address the prognostic utility of these markers. It is foreseeable that a prognostic test that incorporates these markers will also be of use for monitoring lung condition in other respiratory diseases, such as chronic obstructive pulmonary disorder, emphysema, chronic bronchitis, primary ciliary dyskinesia, and asthma, and may also assist as a useful or robust outcome measure in clinical trials.
Acknowledgments
The authors thank their colleagues Dr. Alamgir Khan, Cameron Hill, Eva Wex, Gemma Williams, Dr. Jonathan Arthur, Mai Loan Nguyen, Mathew Traini, and Dr. Matthew McKay (Proteome Systems Ltd, Sydney, NSW, Australia) for technical assistance; Dr. John Fahy, Hofer Wong, and Jane Liu (Moffitt Hospital, University of California, San Francisco, San Francisco, CA) for collection and processing of child samples; Carmel Moriarty (Royal Prince Alfred Hospital, Sydney, NSW, Australia) for assistance in compiling clinical data for the adult subjects; Dr. Melissa Ashlock (CFFT, Bethesda, MD), Dr. Preston Campbell III, Dr. Christopher Penland, and Dr. Robert Beall (CFF, Bethesda, MD), Dr. Bonnie Ramsey, (TDN Coordinating Centre, Seattle, WA), as well as project steering committee members Dr. Harvey Pollard (USUHS, Bethesda, MD), Dr. Ronald Gibson (Children's Hospital and Regional Medical Centre, Seattle, WA), and Dr. Richard Moss (Stanford University, Palo Alto, CA), for expertise in CF and guidance throughout the course of the project; and Dr. Mervyn Thomas (Emphron Informatics Pty Ltd, Chapel Hill, Queensland, Australia) for statistical analyses.
FOOTNOTES
Supported by Cystic Fibrosis Foundation Therapeutics (Bethesda, MD).
This article has an online supplement, which is accessible from this issue's table of contents at www.atsjournals.org
Originally Published in Press as DOI: 10.1164/rccm.200409-1215OC on September 15, 2005
Conflict of Interest Statement: A.J.S. is an employee of Proteome Systems Ltd. and has share options in this company. He is an inventor on provisional patents (pending) that have been filed around data presented in this manuscript. R.A.L. is an employee of Proteome Systems Ltd. and an inventor on provisional patents (pending) that have been filed around data presented in this manuscript. S.S.P. is an employee of Proteome Systems Ltd. and an inventor on provisional patents (pending) that have been filed around data presented in this manuscript. L.T.S. is an employee of Proteome Systems Ltd. and has share options in this company. S.K.P. is an employee of Proteome Systems Ltd. and an inventor on a provisional patent (pending) that has been filed around data presented in this manuscript. M.R. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. P.T.B. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. D.W.N. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. J.L.H. is an employee of Proteome Systems Ltd. and has shares in this company.
REFERENCES
Welsh MJ, Ramsey BW, Accurso F, Cutting GR. Cystic fibrosis. In: Scriver CR, Beaudet AL, Sly WS, Valle D, Childs B, Kinzler KW, Vogelstein B, editors. The metabolic and molecular bases of inherited disease, 8th ed. New York: McGraw-Hill; 2001. pp. 5121–5188.
Cystic Fibrosis Genetic Analysis Consortium. Worldwide survey of the F508 mutation: report from the Cystic Fibrosis Genetic Analysis Consortium. Am J Hum Genet 1990;47:354–359.
Cystic Fibrosis Genetic Analysis Consortium. Population variation of common cystic fibrosis mutations. Hum Mutat 1994;4:167–177.
Gibson RL, Burns JL, Ramsey BW. Pathophysiology and management of pulmonary infections in cystic fibrosis. Am J Respir Crit Care Med 2003;168:918–951.
Ramsey BW. Management of pulmonary disease in patients with cystic fibrosis. N Engl J Med 1996;335:179–188.
Davies JC. Pseudomonas aeruginosa in cystic fibrosis: pathogenesis and persistence. Paediatr Respir Rev 2002;3:128–134.
Donaldson SH, Boucher RC. Update on pathogenesis of cystic fibrosis lung disease. Curr Opin Pulm Med 2003;9:486–491.
Armstrong DS, Grimwood K, Carzino R, Carlin JB, Olinsky A, Phelan PD. Lower respiratory infection and inflammation in infants with newly diagnosed cystic fibrosis. BMJ 1995;310:1571–1572.
Wagener JS, Sontag MK, Accurso FJ. Newborn screening for cystic fibrosis. Curr Opin Pediatr 2003;15:309–315.
Griese M, von Bredow C, Birrer P. Reduced proteolysis of surfactant protein A and changes of the bronchoalveolar lavage fluid proteome by inhaled 1-protease inhibitor in cystic fibrosis. Electrophoresis 2001;22:165–171.
von Bredow C, Birrer P, Griese M. Surfactant protein A and other bronchoalveolar lavage fluid proteins are altered in cystic fibrosis. Eur Respir J 2001;17:716–722.
Chmiel JF, Berger M, Konstan MW. The role of inflammation in the pathophysiology of CF lung disease. Clin Rev Allergy Immunol 2002;23:5–27.
Gershman NH, Liu H, Wong HH, Liu JT, Fahy JV. Fractional analysis of sequential induced sputum samples during sputum induction: evidence that different lung compartments are sampled at different time points. J Allergy Clin Immunol 1999;104:322–328.
Sagel SD, Kapsner R, Osberg I, Sontag MK, Accurso FJ. Airway inflammation in children with cystic fibrosis and healthy children assessed by sputum induction. Am J Respir Crit Care Med 2001;164:1425–1431.
Dakin C, Henry RL, Field P, Morton J. Defining an exacerbation of pulmonary disease in cystic fibrosis. Pediatr Pulmonol 2001;31:436–442.
Rosenfeld M, Emerson J, Williams-Warren J, Pepe M, Smith A, Montgomery AB, Ramsey B. Defining a pulmonary exacerbation in cystic fibrosis. J Pediatr 2001;139:359–365.
Sloane AJ, Duff JL, Wilson NL, Gandhi PS, Hill CJ, Hopwood FG, Smith PE, Thomas ML, Cole RA, Packer NH, et al. High throughput peptide mass fingerprinting and protein macroarray analysis using chemical printing strategies. Mol Cell Proteomics 2002;1:490–499.
Holm S. A simple sequentially rejective multiple test procedure. Scand J Stat 1979;6:65–70.
Pinheiro JC, Bates DM. Mixed effects models in S and S-Plus, 1st ed. New York: Springer-Verlag; 2000.
R Development Core Team. R: a language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing; 2003.
Konstan MW, Berger M. Current understanding of the inflammatory process in cystic fibrosis: onset and etiology. Pediatr Pulmonol 1997;24:137–142.
Wulfkuhle JD, Paweletz CP, Steeg PS, Petricoin EF III, Liotta L. Proteomic approaches to the diagnosis, treatment, and monitoring of cancer. Adv Exp Med Biol 2003;532:59–68.
Rohlff C, Southan C. Proteomic approaches to central nervous system disorders. Curr Opin Mol Ther 2002;4:251–258.
Robinson LJ, Karlsson NG, Weiss AS, Packer NH. Proteomic analysis of the genetic premature aging disease Hutchinson-Gilford progeria syndrome reveals differential protein expression and glycosylation. J Proteome Res 2003;2:556–557.
You SA, Archacki SR, Angheloiu G, Moravec CS, Rao S, Kinter M, Topol EJ, Wang Q. Proteomic approach to coronary atherosclerosis shows ferritin light chain as a significant marker: evidence consistent with iron hypothesis in atherosclerosis. Physiol Genomics 2003;13:25–30.
Sepper R, Prikk K. Proteomics: is it an approach to understand the progression of chronic lung disorders J Proteome Res 2004;3:277–281.
Herbert B. Advances in protein solubilisation for two-dimensional electrophoresis. Electrophoresis 1999;20:660–663.
Neumann M, von Bredow C, Ratjen F, Griese M. Bronchoalveolar lavage protein patterns in children with malignancies, immunosuppression, fever and pulmonary infiltrates. Proteomics 2002;2:683–689.
Plymoth A, Lofdahl CG, Ekberg-Jansson A, Dahlback M, Lindberg H, Fehniger TE, Marko-Varga G. Human bronchoalveolar lavage: biofluid analysis with special emphasis on sample preparation. Proteomics 2003;3:962–972.
Noel-Georis I, Bernard A, Falmagne P, Wattiez R. Proteomics as the tool to search for lung disease markers in bronchoalveolar lavage. Dis Markers 2001;17:271–284.
Bai Y, Galetskiy D, Damoc E, Paschen C, Liu Z, Griese M, Liu S, Przybylski M. High resolution mass spectrometric alveolar proteomics: identification of surfactant protein SP-A and SP-D modifications in proteinosis and cystic fibrosis patients. Proteomics 2004;4:2300–2309.
Henig NR, Tonelli MR, Pier MV, Burns JL, Aitken ML. Sputum induction as a research tool for sampling the airways of subjects with cystic fibrosis. Thorax 2001;56:306–311.
Ordonez CL, Henig NR, Mayer-Hamblett N, Accurso FJ, Burns JL, Chmiel JF, Daines CL, Gibson RL, McNamara S, Retsch-Bogart GZ, et al. Inflammatory and microbiologic markers in induced sputum after intravenous antibiotics in cystic fibrosis. Am J Respir Crit Care Med 2003;168:1471–1475.
De Rose V. Mechanisms and markers of airway inflammation in cystic fibrosis. Eur Respir J 2002;19:333–340.
Rabin HR, Butler SM, Wohl ME, Geller DE, Colin AA, Schidlow DV, Johnson CA, Konstan MW, Regelmann WE. Pulmonary exacerbations in cystic fibrosis. Pediatr Pulmonol 2004;37:400–406.
Karpati F, Hjelte FL, Wretlind B. TNF- and IL-8 in consecutive sputum samples from cystic fibrosis patients during antibiotic treatment. Scand J Infect Dis 2000;32:75–79.
Sagel SD, Sontag MK, Wagener JS, Kapsner RK, Osberg I, Accurso FJ. Induced sputum inflammatory measures correlate with lung function in children with cystic fibrosis. J Pediatr 2002;141:811–817.
Wolter JM, Rodwell RL, Bowler SD, McCormack JG. Cytokines and inflammatory mediators do not indicate acute infection in cystic fibrosis. Clin Diagn Lab Immunol 1999;6:260–265.
Ordonez CL, Kartashov AI, Wohl ME. Variability of markers of inflammation and infection in induced sputum in children with cystic fibrosis. J Pediatr 2004;145:689–692.
Abou-Hatab K, Nixon LS, O'Mahony MS, Newsway V, Patel S, Shale DJ, Woodhouse KW. Plasma esterases in cystic fibrosis: the impact of a respiratory exacerbation and its treatment. Eur J Clin Pharmacol 1999;54:937–941.
Koller DY, Gotz M, Wojnarowski C, Eichler I. Relationship between disease severity and inflammatory markers in cystic fibrosis. Arch Dis Child 1996;75:498–501.
McGrath LT, Mallon P, Dowey L, Silke B, McClean E, McDonnell M, Devine A, Copeland S, Elborn S. Oxidative stress during acute respiratory exacerbations in cystic fibrosis. Thorax 1999;54:518–523.
Montuschi P, Kharitonov SA, Ciabattoni G, Corradi M, van Rensen L, Geddes DM, Hodson ME, Barnes PJ. Exhaled 8-isoprostane as a new non-invasive biomarker of oxidative stress in cystic fibrosis. Thorax 2000;55:205–209.
Valletta EA, Rigo A, Bonazzi L, Zanolla L, Mastella G. Modification of some markers of inflammation during treatment for acute respiratory exacerbation in cystic fibrosis. Acta Paediatr 1992;81:227–230.
Wilmott RW, Frenzke M, Kociela V, Peng L. Plasma interleukin-1 and , tumor necrosis factor-, and lipopolysaccharide concentrations during pulmonary exacerbations of cystic fibrosis. Pediatr Pulmonol 1994;18:21–27.
Antuni JD, Kharitonov SA, Hughes D, Hodson ME, Barnes PJ. Increase in exhaled carbon monoxide during exacerbations of cystic fibrosis. Thorax 2000;55:138–142.
Carpagnano GE, Barnes PJ, Geddes DM, Hodson ME, Kharitonov SA. Increased leukotriene B4 and interleukin-6 in exhaled breath condensate in cystic fibrosis. Am J Respir Crit Care Med 2003;167:1109–1112.
Paredi P, Shah PL, Montuschi P, Sullivan P, Hodson ME, Kharitonov SA, Barnes PJ. Increased carbon monoxide in exhaled air of patients with cystic fibrosis. Thorax 1999;54:917–920.
Van DV, Nguyen MN, Shigenaga MK, Eiserich JP, Marelich GP, Cross CE. Myeloperoxidase and protein oxidation in cystic fibrosis. Am J Physiol Lung Cell Mol Physiol 2000;279:L537–L546.
Worlitzsch D, Herberth G, Ulrich M, Doring G. Catalase, myeloperoxidase and hydrogen peroxide in cystic fibrosis. Eur Respir J 1998;11:377–383.
Stockley RA, Bayley D, Hill SL, Hill AT, Crooks S, Campbell EJ. Assessment of airway neutrophils by sputum colour: correlation with airways inflammation. Thorax 2001;56:366–372.
Meyer KC. Neutrophils, myeloperoxidase, and bronchiectasis in cystic fibrosis: green is not good. J Lab Clin Med 2004;144:124–126.
Regelmann WE, Siefferman CM, Herron JM, Elliott GR, Clawson CC, Gray BH. Sputum peroxidase activity correlates with the severity of lung disease in cystic fibrosis. Pediatr Pulmonol 1995;19:1–9.
Sponer M, Nick HP, Schnebli HP. Different susceptibility of elastase inhibitors to inactivation by proteinases from Staphylococcus aureus and Pseudomonas aeruginosa. Biol Chem Hoppe Seyler 1991;372:963–970.
Batten MR, Senior BW, Kilian M, Woof JM. Amino acid sequence requirements in the hinge of human immunoglobulin A1 (IgA1) for cleavage by streptococcal IgA1 proteases. Infect Immun 2003;71:1462–1469.
Chintalacharuvu KR, Chuang PD, Dragoman A, Fernandez CZ, Qiu J, Plaut AG, Trinh KR, Gala FA, Morrison SL. Cleavage of the human immunoglobulin A1 (IgA1) hinge region by IgA1 proteases requires structures in the Fc region of IgA. Infect Immun 2003;71:2563–2570.
Heck LW, Alarcon PG, Kulhavy RM, Morihara K, Russell MW, Mestecky JF. Degradation of IgA proteins by Pseudomonas aeruginosa elastase. J Immunol 1990;144:2253–2257.
Pawel-Rammingen U, Johansson BP, Bjorck L. IdeS, a novel streptococcal cysteine proteinase with unique specificity for immunoglobulin G. EMBO J 2002;21:1607–1615.
Eulitz M, Weiss DT, Solomon A. Immunoglobulin heavy-chain–associated amyloidosis. Proc Natl Acad Sci USA 1990;87:6542–6546.
Husby G, Blichfeldt P, Brinch L, Brandtzaeg P, Mellbye OJ, Sletten K, Stenstad T. Chronic arthritis and heavy chain disease: coincidence or pathogenic link Scand J Rheumatol 1998;27:257–264.
Husby G. Is there a pathogenic link between gamma heavy chain disease and chronic arthritis Curr Opin Rheumatol 2000;12:65–70.
Monteleone G, Cristina G, Parrello T, Morano S, Biancone L, Pietravalle P, Sagratella E, Doldo P, Luzza F, Di Mario U, et al. Altered IgG4 renal clearance in patients with inflammatory bowel diseases: evidence for a subclinical impairment of protein charge renal selectivity. Nephrol Dial Transplant 2000;15:498–501.
Takatani T, Morita K, Takaoka N, Tatsumi M, Okuno Y, Masutani T, Murakawa K, Fukui A, Tsukaguchi N, Okamoto Y. heavy chain disease screening showing a discrepancy between electrophoretic and nephelometric determinations of serum globulin concentration. Ann Clin Biochem 2002;39:531–533.
Division of Allergy, Pulmonary, and Critical Care Medicine, Department of Medicine
Departments of Biostatistics and Pathology
Mass Spectrometry Research Center
Division of Hematology and Oncology, Departments of Surgery and Cancer Biology, Vanderbilt-Ingram Comprehensive Cancer Center, Vanderbilt University School of Medicine
Veterans Affairs Medical Center, Nashville, Tennessee
Department of Medicine and Pathology, University of Colorado Health Science Center, Denver, Colorado
ABSTRACT
Purpose: A proteomics approach is warranted to further elucidate the molecular steps involved in lung tumor development. We asked whether we could classify preinvasive lesions of airway epithelium according to their proteomic profile.
Experimental Design: We obtained matrix-assisted laser desorption/ionization time-of-flight mass spectrometry profiles from 10-μm sections of fresh-frozen tissue samples: 25 normal lung, 29 normal bronchial epithelium, and 20 preinvasive and 36 invasive lung tumor tissue samples from 53 patients. Proteomic profiles were calibrated, binned, and normalized before analysis. We performed class comparison, class prediction, and supervised hierarchic cluster analysis. We tested a set of discriminatory features obtained in a previously published dataset to classify this independent set of normal, preinvasive, and invasive lung tissues.
Results: We found a specific proteomic profile that allows an overall predictive accuracy of over 90% of normal, preinvasive, and invasive lung tissues. The proteomic profiles of these tissues were distinct from each other within a disease continuum. We trained our prediction model in a previously published dataset and tested it in a new blinded test set to reach an overall 74% accuracy in classifying tumors from normal tissues.
Conclusions: We found specific patterns of protein expression of the airway epithelium that accurately classify bronchial and alveolar tissue with normal histology from preinvasive bronchial lesions and from invasive lung cancer. Although further study is needed to validate this approach and to identify biomarkers of tumor development, this is a first step toward a new proteomic characterization of the human model of lung cancer tumorigenesis.
Key Words: early detection mass spectrometry preneoplasia profiling
The discovery of preinvasive lesions in the high-risk population places patients at increased risk of developing lung cancer (1–3). Although the natural history of these preinvasive lesions is poorly understood, increasing evidence suggests that, in the absence of treatment, high-grade preinvasive lesions, most commonly found in patients with a prior history or with concomitant cancer, will develop into invasive carcinoma in 30 to 50% of the cases, whereas the vast majority of low-grade preinvasive lesions remain stable or regress on follow-up (4–7). In addition, nonstepwise progression of preinvasive lesions makes prediction of lung cancer progression based on histologic grade unreliable. Therefore, the need for molecular biomarkers predictive of tumor progression is becoming more evident (8–10).
In a recent report, we used conventional matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI MS) to generate protein profiles that accurately classify and predict histologic subgroups of lung cancer (11). Using biostatistical methods to select differentially expressed peaks (MS signals) and after the development of a class prediction model (12), 82 discriminatory signals were found to classify normal lung from lung cancer tissue samples in a derivation and validation study design with excellent accuracy. Other recent studies indicated the importance of using protein expression profiles as a diagnostic or prognostic biomarker for patients with early-stage lung cancer using two-dimensional polyacrylamide gel electrophoresis analysis (13, 14), protein arrays (15), or surface-enhanced laser desorption/ionization MS (16). Although new strategies are needed for the early detection of lung cancer, clinical proteomics applied directly to the airway epithelium may provide a novel approach for the selection of biomarkers.
Our primary goal is to establish an MALDI MS–based proteomics approach to obtain specific patterns of protein expression of the airway epithelium at different stages of tumor development. We hypothesize that airway epithelium (alveolar and bronchial) modify its proteomic expression profile during tumor development. In this first report, we demonstrated that proteomic profiles obtained by MALDI MS allow the classification of normal alveolar or normal bronchial epithelium, preinvasive lesions, and invasive lung tumors and may provide new insights to the biology of these preinvasive lesions of unclear clinical significance. Some of the results of these studies have been previously reported in abstract form (17).
METHODS
Patients and Tissue Samples
A total of 53 patients from Vanderbilt University Medical Center (VUMC) participated in this study between January 2002 and November 2003. A total of 110 samples were obtained from bronchoscopic or surgical procedures. Eighteen of 29 normal bronchial epithelium samples and 18 of 20 preinvasive tissues were obtained at the time of bronchoscopy, and all 25 normal alveolar and all 36 lung tumor tissues were obtained at the time of surgery. Normal alveolar specimens were obtained from at least 2 cm away from the tumor. Endobronchial biopsies were obtained by white-light bronchoscopy and laser-induced fluorescence endoscopy from VUMC and University of Colorado Health Science Center (UCHSC) between May 2002 and December 2003. Biopsies were obtained from patients with lung cancer or at risk of lung cancer from predetermined as well as suspicious sites. Of the 10 patients without concomitant lung cancer, five were recruited from the UCHSC lung cancer cohort (with sputum atypia), one patient was a donor of bronchial epithelium as a motor vehicle accident victim, and four patients were recruited from VUMC on the basis of clinical suspicion of lung cancer. Squamous metaplasia and mild dysplasia were grouped as low-grade preinvasive lesion, whereas moderate dysplasia, severe dysplasia, and carcinoma in situ were grouped as high-grade preinvasive lesion. Bronchial epithelial lesions were graded according to the World Health Organization nomenclature (18). The project was approved by the local institutional review board, and informed consent was obtained for all individuals.
Preparation of Tissue Samples
Tissue sections were cut into 10-μm sections and stored in a –140°C freezer until use. Only a minimum amount of optimal cutting temperature (OCT) medium was used to immobilize the tissues, avoiding embedding the area to be sectioned for MALDI MS. Sections were directly transferred onto a gold-coated stainless steel sample plate (PE Biosystems, Foster City, CA) and a glass slide. Because of their small size, bronchial biopsies were embedded into OCT medium. To remove OCT medium from bronchial biopsy sections, water wash was performed at room temperature. The sections were dried for 45 min in a dessicator. An adjacent section was stained with hematoxylin and eosin as a guide, and the area to be analyzed by MALDI MS was precisely marked by one of the present authors (A.L.G.), the lung pathologist at Vanderbilt University, who was trained by another pathologist at UCHSC (W.A.F.). Samples from UCHSC were examined by W.A.F. Matrix solution (sinapinic acid, 35 mg/dl in water:acetonitrile:trifluoroacetic acid, 500/500/1, vol/vol/vol) was deposited before MS analysis. Matrix deposition was performed with a fine capillary needle under microscopic guidance.
Proteomic Analysis
Protein expression profiles were obtained as described previously (11, 19–21). Briefly, each spectrum was acquired over the surface of the matrix spot. In this analysis, signals in the range between 2,000 and 20,000 mass-to-charge ratios were considered. Each spectrum underwent smoothening, baseline correction, and internal calibration, with peaks from internal hemoglobin and chains using Data Explorer software (Applied Biosystems, Foster City, CA). Peak list obtained from the spectra was normalized and binned with a custom algorithm written by the group at VUMC, resulting in 1,261 bins in total, and then submitted for statistical analyses. Representative spectra from each histologic subtype are shown in Figure 1. Reproducibility and variability issues of spectra acquisition were addressed in previous reports (11, 22, 23) and confirmed in this dataset. The intrasample variability for the top 100 peaks was less than 30% of the overall variability (intra- and intersample variability).
Alignment, Binning, and Normalization of the MS Profiles
Alignment of MALDI MS peaks across multiple spectra was accomplished by use of a computing algorithm that simultaneously determines an optimal set of " bins" for categorizing each peak as a specific protein (24). Optimal bins are identified as those that maximize the number of single peaks within each bin across the spectra while minimizing the number of bins that have multiple peaks within a spectrum. We have demonstrated that this is an effective approach in a recent proteomics study of lung cancer (11). Spectra were normalized to each other according to the maximum sum of total ion current before binning. The maximum sum was divided by the sum of each spectrum to obtain the normalization factor for each spectrum. This factor was then used to multiply each data point for each spectrum.
Statistical Analysis
The primary objective of this study was to identify a set of proteins expressed differentially among study groups. The statistical analyses were focused on the following steps:
Selection of MS features that were differentially expressed between the study groups was performed. The selection was based on the Kruskal-Wallis test, Fisher's exact test (dichotomize the expression level as present or not), the permutation t test, significance analysis of microarrays (25), weighted gene analysis (26), and the modified information score method (27). The cutoff points for each method were p < 0.0005, p < 0.0005, p < 0.0005, 2, 2, and 0, respectively. Because there is no standard statistical method for analyzing the proteomic data, the selection of the "winner" markers was based on these six different statistical methods. A feature was included on the final list if it met at least two of these six criteria.
The "winner" features that have reasonable predictive power were examined. The statistical class-prediction model based on the selected features was applied to determine whether the proteomic patterns could be used to classify normal from preinvasive and invasive tissue samples. The weighted flexible compound covariate method (WFCCM) (11, 12, 28, 29) was used in the class-prediction model based on the selected features to determine whether the proteomic patterns could be used to classify tissue samples among groups. We estimated the misclassification rate using the leave-one-out cross-validation class-prediction method based on the WFCCM.
The prediction model generated from the training set was applied to the blinded test cohort, and the samples were classified based on the closeness of the distance from the two classes.
The agglomerative hierarchic clustering algorithm (28) and multidimensional scaling (MDS) method were applied to investigate the pattern among the statistically significant discriminatory features as well as disease status using M. Eisen's software (30) and MATLAB (MathWorks Inc., Natick, MA), respectively. MDS can be considered to be an alternative to factor analysis. In general, the goal of the analysis is to detect meaningful underlying dimensions that allow the researcher to explain observed similarities or dissimilarities (distances) between the investigated objects. Multiple comparisons in this study were adjusted by using the false discovery rate (FDR). FDR is a new approach to the multiple comparisons problem. Instead of controlling the chance of any false positives (as in Bonferroni's adjustment), the FDR of a set of predictions is the expected percentage of false predictions in the set of markers.
RESULTS
Study Population
To obtain specific patterns of protein expression of the airway epithelium at different stages of tumor development, we obtained MALDI MS profiles on 110 tissue samples from a total of 53 individuals, 43 of whom were patients with concomitant lung cancer. The patients' characteristics are presented in Table 1. The nature of the samples obtained in patients with specific tumor histology is provided in Table 2. Because of the heterogeneous nature of lung cancer and because the nature of the cell type of origin of these tumors remains a subject of controversy, both alveolar epithelium and bronchial epithelium were included as controls for lung tumors examined. Some patients provided more than one type of tissue sample as shown in Table 3. These samples included 25 histologically normal lung (alveolar space), 29 histologically normal bronchial epithelium, 20 preinvasive tumor tissue samples (13 low-grade and 7 high-grade), and 36 invasive tumor tissue samples.
Preinvasive Lesions Differ from Normal Airway Epithelium and from Invasive Lung Cancer
For each histologic subgroup, we obtained a proteomic profile that distinguishes from other lesions with predictive accuracy between 83 and 100%. Results are summarized in Table 4. Examples of such profiles are provided in Figure 1. Classification and misclassification rates were calculated by use of leave-one-out cross-validation class prediction method based on our covariate method of analysis. To avoid the possibility of overfitting, we reported only one set of "winner" features (based on the predetermination cut-off of the p values), which was applied to the prediction model. Normal bronchial epithelium proteomic profiles clearly differ from those of invasive cancer and all other classes. Preinvasive lesions provided different proteomic patterns from normal bronchial epithelium or invasive cancer. Importantly, and despite our relatively small sample size, we found patterns characteristic of either low- or high-grade lesions based on the different expression of 14 features (Figure 2A). We performed the same comparisons using agglomerative hierarchic cluster analysis with very similar results. For example, hierarchic cluster analysis of high-grade preinvasive lesion versus invasive tumor is shown in Figure 2B. The potential issue of multiple comparisons in high-dimensional data analysis is adjusted by the FDR (see STATISTICAL ANALYSIS).
Distinguishing Normal from Preinvasive Lesions from Cancer within a Continuum
In an effort to distinguish proteomic profiles of normal, preinvasive, and invasive tumors in a continuum, we analyzed the profiles obtained from normal bronchial epithelium, preinvasive bronchial epithelium, and squamous carcinomas of the lung. Results of supervised clustering analysis (Figure 3A) and of multidimensional scaling analysis of these three groups (Figure 3B) show distinct clustering of the groups in a continuum. Biomarkers found in preinvasive and in invasive lesions but not in normal airway epithelium of subjects with or at risk for lung cancer would be of particular interest in the development of markers for early detection. We searched for such features in our dataset and, although some specific features are completely absent from normal bronchial tissues, they were found in 0 to 15% of low-grade preinvasive lesions, in 14 to 43% of high-grade lesions, and in 6 to 24% of the squamous invasive tumor tissues; none of these candidate biomarkers reached statistical significance set at p < 0.0005.
Proteomic Profiles of Preinvasive Lesions Resemble Those of Normal Epithelium
After training our prediction model to detect features that discriminate between normal bronchial epithelium and invasive tumor, we asked how the model would classify low- and high-grade lesions. Proteomic profiles obtained from 12 of 13 low-grade and all high-grade preinvasive lesions were classified by 54 discriminatory features as normal airway epithelium.
Validation of Prediction Model in Normal and Invasive Lung Cancer
Finally, we trained our prediction model in our previously published dataset of 14 normal lung tissue samples and 66 tumors (11) and tested the discriminatory profile in our new and independent test set of 36 tumors, 25 alveolar lung tissues, and 29 bronchial specimens. The two datasets were binned and normalized together. We reached an overall 74% rate of accuracy in classifying tumors from normal tissues. Specifically, we correctly classified 18 of 25 of the alveolar lung tissues, 23 of 29 of the normal bronchial specimens, and 26 of 36 tumors.
DISCUSSION
We report for the first time proteomic expression profiles specific to different stages of lung tumor development. We detected MALDI MS signals that were discriminatory and predictive of alveolar and bronchial epithelium, low- and high-grade preinvasive lesions, and invasive lung tumors with overall 90% accuracy. When proteomic profiles of preinvasive lesions were tested using patterns that distinguish between normal airway and invasive tumors obtained in a training set, the majority of preinvasive lesions were classified as normal-appearing epithelium. We also demonstrated in a training and test paradigm that a set of discriminatory features obtained in an independent, previously published set of normal and tumor samples correctly classified 74% of our samples as tumor or normal epithelium.
Although the risk of developing lung cancer increases with the presence of preinvasive lesions, the molecular determinants predicting the irreversible progression to lung cancer have not been identified. Previous studies have addressed the genetic basis for classification of preinvasive lesions in a model of tumor development (1). Yet, none of these biomarkers are predictive of progression to invasive cancer. The proteomic analysis of tumor development might provide a new understanding of the pathologic states a given tumor may undergo before acquiring its invasive phenotype. Our previous study, which defined profiles that allow classification of lung tumors from normal lung and from tumors associated with lymph node involvement (11), clearly showed the strength of tissue-based proteomics as an analytic tool.
This study provides evidence for a specific phenotype of the airway epithelium as it progresses from normal to preinvasive and to invasive cancer. The findings of the supervised cluster and MDS analyses, although they do not prove true progression, suggest a continuum between proteomic patterns of lesions as they progress toward invasive lesions. Yet, we did not find significant MALDI MS features robust enough to pass our conservative statistical approach that would be present in preinvasive and invasive lesions but not in normal bronchial epithelium. Several considerations may explain this lack of specific markers of tumor development. The study design (cross-sectional as opposed to longitudinal), the sample size, and the analysis of only a fraction of the proteome offered by the MALDI MS approach need to be taken into consideration. Biological variability may also prevent us from finding markers of tumor development because only a small percentage of preinvasive lesions are known to transform into cancer (7). The samples examined may not have harbored these features. A prospective study following the natural history of a large number of these lesions may answer this question.
This report is our first step toward the selection of proteomic profiles predictive of lung cancer development. We found that this tissue-based proteomics approach allows us not only to recapitulate the classification of preinvasive lesions on the basis of histologic grade but to provide biological information about these lesions. Indeed, when testing proteomic predictors of lung cancer versus normal tissue to classify profiles from preinvasive lesions (as being closer to normal or cancer), both low- and high-grade lesions clustered with normal epithelium. These findings suggest that only a subset of preinvasive lesions (yet to be identified in a prospective study) may in fact progress to an invasive phenotype, and histology by itself may not be the ideal surrogate marker of tumor progression. This observation is in agreement with higher regression and low progression of preinvasive lesions shown in earlier studies (7, 31, 32).
With the current and independent set of tissues, we were able to reproduce similar results to those obtained in our previous report (11). Despite the inclusion of normal bronchial samples in the test set that were not represented in the training set (i.e., those of a separate operator [S.M.J.R.]) and of differences in the data processing (see METHODS), we were able to correctly classify 74% of the samples as normal epithelium or as invasive tumor. These results further validate our proteomic strategy.
In our previous report (12), only alveolar tissues were representative of the control group. In the current study, because 18 of 20 preinvasive tissues were obtained from bronchial biopsies, we included a series of bronchial tissue as controls in addition to alveolar tissue. The difference between profiles of normal alveolar epithelium and bronchial epithelium is interesting and important for future research related to cellular origin of squamous carcinoma and adenocarcinoma.
Limitations to this study include considerations about patient selection, methodology, sample size, and protein identification. First, most of our preinvasive lesions and our normal-appearing epithelium were obtained from patients with concomitant lung cancer. The natural history of these preinvasive lesions may be different from those of patients without concomitant lung cancer. Also, we lack a true biological "normal" epithelium for reference because the majority of our patients were known to have lung cancer and were current or ex-smokers. Second, technical considerations may have affected the outcome of the analysis. The tissues analyzed were obtained from different sources (surgically resected specimen vs. endobronchial biopsies) and processed slightly differently (without or with OCT embedding). Although OCT medium prevents optimal ionization of proteins (33), the necessary removal of water-soluble OCT could introduce a difference in methodology, yet across all histologic subtypes and therefore unlikely to affect the analysis. In addition, the resolution of the MS, daily variation of the vacuum, and laser intensity of the instrument itself and its overall performance are variables that have been addressed before (11). Third, a relatively small sample size of preinvasive lesions requires confirmation in a larger dataset. Nevertheless, our data strongly illustrate the strength of proteomic profiles to discriminate pathologic states along the various steps of tumor development. And fourth, we report the predictive value of protein profiles and not proteins or peptides. The identification of the proteins or peptides making these specific signatures, though critical, is beyond the scope of this study because we first need to validate them in an independent test set. Ultimately, selection and identification of proteins predictive of what preinvasive lesion is likely to develop to lung cancer is a critical goal of future investigation.
In conclusion, the original selection of these discriminatory profiles by MALDI MS is a critical first step in the identification of proteomic expression patterns along lung cancer development. Ultimately, this MALDI MS–based approach in tissue may provide characterization of the molecular determinants leading preinvasive lesions to an invasive phenotype.
Acknowledgments
The authors thank the individuals who provided their informed consent and participated in the study. They thank Lynne Fenner, Blake Mann, Darienne Adkins, Heather Templeton, Candace Murphy, and Anthony Frazier for their assistance in consenting individuals, in obtaining clinical data elements and biological samples. They also thank Hans Rudolf Aerni and Lisa Manier for their expertise in proteomic analysis.
FOOTNOTES
Supported by the Damon Runyon Cancer Research Foundation (P.P.M. is a Damon Runyon-Lilly Clinical Investigator, CI 19-03). Also supported in part by the Lung SPORE P50 CA 90949 to D.P.C. and by the Flight Attendant Medical Research Institute to P.P.M.
Originally Published in Press as DOI: 10.1164/rccm.200502-274OC on September 22, 2005
Conflict of Interest Statement: None of the authors have a financial relationship with a commercial entity that has an interest in the subject of this manuscript.
REFERENCES
Hirsch FR, Franklin WA, Gazdar AF, Bunn PA Jr. Early detection of lung cancer: clinical perspectives of recent advances in biology and radiology. Clin Cancer Res 2001;7:5–22.
Franklin WA. Diagnosis of lung cancer: pathology of invasive and preinvasive neoplasia. Chest 2000;117(4, Suppl 1):80S–89S.
Prindiville SA, Byers T, Hirsch FR, Franklin WA, Miller YE, Vu KO, Wolf HJ, Baron AE, Shroyer KR, Zeng C, et al. Sputum cytological atypia as a predictor of incident lung cancer in a cohort of heavy smokers with airflow obstruction. Cancer Epidemiol Biomarkers Prev 2003;12:987–993.
Risse EK, Vooijs GP, van't Hof MA. Diagnostic significance of "severe dysplasia" in sputum cytology. Acta Cytol 1988;32(5):629–634.
Saccomanno G, Archer VE, Auerbach O, Saunders RP, Brennan LM. Development of carcinoma of the lung as reflected in exfoliated cells. Cancer 1974;33:256–270.
Mulshine JL, Cuttitta F, Tockman MS, De Luca LM. Lung cancer evolution to preinvasive management. Clin Chest Med 2002;23:37–48.
Breuer RH, Pasic A, Smit EF, van Vliet E, Vonk Noordegraaf A, Risse EJ, Postmus PE, Sutedja TG. The natural course of preneoplastic lesions in bronchial epithelium. Clin Cancer Res 2005;11:537–543.
Wistuba II. Histologic evaluation of bronchial squamous lesions: any role in lung cancer risk assessment Clin Cancer Res 2005;11:1358–1360.
Chanin TD, Merrick DT, Franklin WA, Hirsch FR. Recent developments in biomarkers for the early detection of lung cancer: perspectives based on publications 2003 to present. Curr Opin Pulm Med 2004;10:242–247.
Massion PP, Taflan PM, Shyr Y, Rahman SM, Yildiz P, Shakthour B, Edgerton ME, Ninan M, Andersen JJ, Gonzalez AL. Early involvement of the phosphatidylinositol 3-kinase/Akt pathway in lung cancer progression. Am J Respir Crit Care Med 2004;170:1088–1094.
Yanagisawa K, Shyr Y, Xu BJ, Massion PP, Larsen PH, White BC, Roberts JR, Edgerton M, Gonzalez A, Nadaf S, et al. Proteomic patterns of tumour subsets in non-small-cell lung cancer. Lancet 2003;362:433–439.
Shyr Y, Kim K. Weighted flexible compound covariate method for classifying microarray data. In: Berrar D, editor. A practical approach to microarray data analysis. New York: Kluwer Academic; 2003. pp. 186–200.
Chen G, Gharib TG, Wang H, Huang CC, Kuick R, Thomas DG, Shedden KA, Misek DE, Taylor JM, Giordano TJ, et al. Protein profiles associated with survival in lung adenocarcinoma. Proc Natl Acad Sci USA 2003;100:13537–13542.
Li C, Chen Z, Xiao Z, Wu X, Zhan X, Zhang X, Li M, Li J, Feng X, Liang S, et al. Comparative proteomics analysis of human lung squamous carcinoma. Biochem Biophys Res Commun 2003;309:253–260.
Valle RP, Chavany C, Zhukov TA, Jendoubi M. New approaches for biomarker discovery in lung cancer. Expert Rev Mol Diagn 2003;3:55–67.
Zhukov TA, Johanson RA, Cantor AB, Clark RA, Tockman MS. Discovery of distinct protein profiles specific for lung tumors and pre-malignant lung lesions by SELDI mass spectrometry. Lung Cancer 2003;40:267–279.
Rahman S, Shyr Y, Yildiz P, Gonzalez A, Li H, Zhang X, Chaurand P, Yanagisawa K, Slovis B, Miller RD, et al. Proteomic patterns of preinvasive bronchial lesions in patients with lung cancer . Am J Respir Crit Care Med 2004;169:A57.
Travis W, Colby TV, Corrin B, Shimosato Y, Travis WD, Brambilla E. In: Sabin LH, editor. Histological typing of lung and pleural tumours (International Histological Classification of Tumours), 3rd ed. Berlin: Springer-Verlag; 1999.
Chaurand P, Schwartz SA, Caprioli RM. Imaging mass spectrometry: a new tool to investigate the spatial organization of peptides and proteins in mammalian tissue sections. Curr Opin Chem Biol 2002;6:676–681.
Stoeckli M, Chaurand P, Hallahan DE, Caprioli RM. Imaging mass spectrometry: a new technology for the analysis of protein expression in mammalian tissues. Nat Med 2001;7:493–496.
Xu BJ, Caprioli RM, Sanders ME, Jensen RA. Direct analysis of laser capture microdissected cells by MALDI mass spectrometry. J Am Soc Mass Spectrom 2002;13:1292–1297.
Schwartz SA, Weil RJ, Johnson MD, Toms SA, Caprioli RM. Protein profiling in brain tumors using mass spectrometry: feasibility of a new technique for the analysis of protein expression. Clin Cancer Res 2004;10:981–987.
Chaurand P, Sanders ME, Jensen RA, Caprioli RM. Proteomics in diagnostic pathology: profiling and imaging proteins directly in tissue sections. Am J Pathol 2004;165:1057–1068.
Moore J, Parker J. Methods of microarray data analysis. Boston: Kluwer Academic; 2001.
Tusher VG, Tibshirani R, Chu G. Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci USA 2001;98:5116–5121.
Hedenfalk I, Duggan D, Chen Y, Radmacher M, Bittner M, Simon R, Meltzer P, Gusterson B, Esteller M, Kallioniemi OP, et al. Gene-expression profiles in hereditary breast cancer. N Engl J Med 2001;344:539–548.
Ben'Dor AFN, Yakhini Z. Scoring genes for relevance. Tech Report AGL-2000-13, Agilent Labs, Agilent Technologies, 2000. Available from: http://www.labs.agilent.com/resources/techreports.html (accessed 2001).
Yamagata N, Shyr Y, Yanagisawa K, Edgerton M, Dang TP, Gonzalez A, Nadaf S, Larsen P, Roberts JR, Nesbitt JC, et al. A training-testing approach to the molecular classification of resected non-small cell lung cancer. Clin Cancer Res 2003;9:4695–4704.
Tukey JW. Tightening the clinical trial. Control Clin Trials 1993;14:266–285.
Eisen MB, Spellman PT, Brown PO, Botstein D. Cluster analysis and display of genome-wide expression patterns. Proc Natl Acad Sci USA 1998;95:14863–14868.
Bota S, Auliac JB, Paris C, Metayer J, Sesboue R, Nouvet G, Thiberville L. Follow-up of bronchial precancerous lesions and carcinoma in situ using fluorescence endoscopy. Am J Respir Crit Care Med 2001;164:1688–1693.
Banerjee AK, Rabbitts PH, George PJ. Are all high-grade preinvasive lesions premalignant, and should they all be treated Am J Respir Crit Care Med 2002;165(10):1452–1453. [Author reply, 1453.]
Schwartz SA, Reyzer ML, Caprioli RM. Direct tissue analysis using matrix-assisted laser desorption/ionization mass spectrometry: practical aspects of sample preparation. J Mass Spectrom 2003;38:699– 708.
