Proteomics and Metabolomics in Inflammatory Bowel Disease

Yunki Yau; Rupert W Leong; Ming Zeng; Valerie C Wasinger

Disclosures

J Gastroenterol Hepatol. 2013;28(7):1076-1086. 

In This Article

Where We Have Been

In scrutinizing the impact of proteomics and metabolomics in IBD, it is helpful to briefly narrate the recent history of disease marker exploration in the field before and after high-throughput capabilities (Fig. 1). This is by no means exhaustive, but provides an overview of the general course that has shaped some of the climate of clinical IBD today.

Figure 1.

Timeline of select notable findings in inflammatory bowel disease (IBD) biomarker research. (3). Indium-111 granulocyte scanning in faeces measures IBD inflammation.25 (7). C-reactive protein (CRP) measures IBD inflammation.26 (8). Platelet and erythrocyte sedimentation rate (ESR) count differentiates IBD and infective diarrhea.27 (9). Interleukin (IL)-12 is found to be increased in IBD colon.28 (11). α4β7 integrin changes are found in IBD T-lymphocytes.29 (16) IL-23 is found to be elevated in Crohn's disease (CD) colon.30

Arrival of omics technologies assured the eventual complete archival of all biochemical entities—the challenge has always been in deciphering which pieces of information are relevant to the condition in question. These pieces of information may be differentially measurable—representing disease risk, progression, or treatment-induced change, otherwise known as biomarkers.[31]

The seed in the search for laboratory-based IBD biomarkers was likely sown in 1936, when Bargen and Barker observed thromboembolic complications in UC.[32] Thrombotic elements were subsequently reaffirmed and analyzed in the IBDs in the following decades,[33–35] and in 1966, thrombocytosis was possibly the first serological index proposed for IBD activity (1 in Fig. 1).[33,36]

The concept of an autoimmunological basis to IBD was also first introduced in the sixties by Broberger and colleagues, who probed UC serum with antigens derived from various endogenous tissues to entice antibody reactivity.[37,38] Regional ileitis/enteritis became widely known as CD at this time, and was thought to be a hypo-immunological condition (differing from UC).[39,40] Multiple investigators looked to characterize immunoglobulin turnover in CD by quantifying specific markers in serum and feces, with mixed results.[41–43] The enteropathogenic Escherichia Coli (E. coli) was also discovered in the context of IBD using antibodies at the end of this decade (2 in Fig. 1).[44]

The seventies came around, and radioimmunoassays were being widely used to measure carcinoembryonic antigen as a potential early detection marker for carcinoma and UC disease activity, and beta-2-microglobulin as an indicator of lymphocyte activation during CD inflammation.[45–50] Elsewhere, lymphocytotoxins and antilymphocyte antibodies were being characterized in IBD sera by diethylaminoethyl cellulose (DEAE) ion exchange chromatography and immunoabsorption columns in an effort to understand lymphocyte regulation in IBD.[51–53]

The first documented application of mass spectrometry (MS) in IBD occurred in 1982, when an absolute targeted quantification of small molecules was carried out by Nishida and colleagues using gas chromatography/mass spectrometry (GC/MS) with an internal standard calibration curve and stable isotope labeling to describe bile acid circulation impairment in CD patients after ingestion of deuterium labeled chenodeoxycholic acid.[54]

Several years later, interleukin (IL)-1 was implicated in CD by cytotoxicity assays with murine fibroblast targets[55,56] and serendipitously attracted curiosity to the related cytokine tumor necrosis factor-alpha (TNF-α) (5 in Fig. 1).[57]

By the end of the decade, refined cytotoxicity experiments linked TNF-α with colonic epithelial cell death in IBD and enzyme-linked immunosorbent assays (ELISAs) were used extensively to validate TNF-α in various human media as a biomarker of disease activity.[57–61] Humanized monoclonal antibodies also first appeared around this time.[62]

Immuno-based photometric techniques were widely used in the nineties with significant returns: ELISAs and other immunofluorescent techniques were used to establish a significant number of the IBD biomarkers we know today, including fecal lactoferrin (10 in Fig. 1), calprotectin (14 in Fig. 1), calgranulin C (S100A12) (15 and 17 in Fig. 1), anti-saccharomyces antibodies (ASCA) (12 in Fig. 1), perinuclear antineutrophil cytoplasmic antibody (pANCA) (6 and 12 in Fig. 1), anti-outer membrane porin C (Anti-OmpC), and anti-flagellin (Anti-Cbir1), among others (13 in Fig. 1).[63–80]

With the development of protein and metabolite repositories for proteomics and metabolomics experiments, the 2000s saw a steady influx of functional and absolute hypothesis-free protein and metabolite profiling studies in IBD, starting with the aforementioned Barcelo-Batllori group.[23]

Proteomics in IBD

Spanning across Switzerland, Japan, and Germany, Barcelo-Batllori et al. profiled the human epithelial cell proteome in vitro before and after exposure to IL-γ, IL-1β, and IL-6, using a combination method of two-dimensional polyacrylamide gel electrophoresis (2D-PAGE), matrix-assisted laser desorption/ionization–time of flight (MALDI-TOF) MS, and Western blotting.[23] They found an overabundance of the enzyme indoleamine-2,3-dioxygenase in IBD compared with controls, and hypothesized an involvement of the kynurenine pathway of tryptophan metabolism in the IBDs.[23]

Two years later, Hardwidge and colleagues utilized an LC–tandem MS (MS/MS) method to measure the response of human intestinal epithelial cells to E. Coli, and to validate the pathogen's mode of action in a proof of concept study.[81] They accounted significant changes in actin-related proteins before and after infection.[81]

In 2006, multiple independent groups made use of the 2D-PAGE/MALDI-TOF MS peptide mass fingerprinting methodology in IBD proteomics. In Taiwan, Hsieh et al. employed a similar workflow with MS/MS to profile and sequence proteins in the colon mucosa of active and nonactive UC, indeterminate colitis, and healthy controls, and found a host of downregulated mitochondrial proteins that suggested colonocyte mitochondrial dysfunction.[82] Weichart et al. meanwhile, demonstrated the multifaceted proinflammatory program regulated by nucleotide-binding oligomerization domain-containing protein 2 (NOD2) through cross-referencing protein expression between bearers of the NOD2 gene and those of previously found CD-associated NOD2 variants before and after muramyl dipeptide simulation.[6, 7, 83] Weichart's work was eminent in illustrating the significant cellular changes before and after activation of the highly noted NOD2 receptor. In Germany, Shkoda et al. described the proteome of epithelial cells purified from CD, UC, and colon cancer intestinal tissue and used Western blotting as a validation tool.[84] Shkoda and colleagues reported a host of differentially expressed proteins between study groups involved in signal transduction, stress response, and cellular homeostasis.[84]

Meuwis et al. published the first serum proteomic study of IBD in 2007, using surface-enhanced laser desorption ionization (SELDI)-TOF MS for the initial proteome scan, followed by extensive validation of proteins of interest using MALDI MS/MS, Western blotting, and ELISA assay.[85] The Belgium-based group compared serum protein profiles between CD, UC, nonspecific inflammatory, and healthy controls, and validated four biomarker candidates, although the authors contend that all are known proteins of acute inflammation.[85]

In the following year, Meuwis and colleagues followed up their study with a functional proteomics experiment—again using SELDI-TOF MS—to record the serum proteome of CD patients before and after infliximab treatment, and compare patients who responded and did not respond to therapy.[86] The researchers validated their previous platelet factor 4 biomarker candidate as being significantly higher in abundance in infliximab nonresponders compared with responders.[86]

An Italian group of investigators recently presented two novel technical contributions to proteomics-based biomarker discovery studies in IBD. Firstly, Nanni et al. introduced a solid-phase bulk protein extraction protocol that included carbon-18 reverse phase, strong anion-exchange, and metal ion affinity LC techniques for maximizing protein yield from blood serum in 2007,[87] and in 2009, Nanni and colleagues demonstrated the use of a label-free proteome comparison strategy that did not require isotopic labeling reagents (thus saving considerable cost in high-throughput experiments with many samples) that had not previously been employed in IBD research.[88]

Most recently, several investigators have applied proteomic techniques in resourceful and innovative methodologies. M'Koma and colleagues profiled the proteomes of Crohn's colitis (CC) and UC colonic mucosal and submucosal tissues with MALDI-MS, comparing histologically indicated inflamed and uninflamed sample areas both within and across CC and UC.[89] They found five unknown molecular species (identified by their m/z property) that significantly differed between the two colitides, highlighting the potential for MS-based biomarkers to aid diagnostic accuracy in clinically ambiguous cases.[89] In New Zealand, Cooney et al. conducted an extensive study correlating colonic transcriptomic and proteomic profiles of IL-10 deficient mice administered with various polyunsaturated fatty acid diets, delineating possible functional pathways of fatty acids n-3 eicosapentaenoic and n-6 arachidonic acid in attenuating inflammation.[90] In June 2011, Presley et al. defined a "metaproteome"—the protein expression on the mucosal-luminal interface of the intestine—that would provide a unique medium describing the interactions between host and the resident luminal organisms.[91] The authors employed a novel saline-lavage technique to extract this habitat (without interference from intestinal layer contents that a biopsy sample would enclose) and deployed SELDI-TOF MS to report the protein species present.[91] In this work, Presley et al. correlated bacterial phylotypes with specific immunological protein features, potentially disclosing important host–microbe interactions in IBD pathogenesis.[91]

Metabolomics in IBD

Application of metabolomics in IBD began as noted, in 2007, when Marchesi and colleagues utilized 1H NMR spectroscopy to examine fecal extracts from IBD patients and healthy controls.[24] The investigators found a more marked difference in the fecal metabolomes of CD patients and controls than when UC was compared with controls—possibly an indicator of the extent of inflammation and disease of Crohn's.[24]

Each successive year that followed saw global metabolite profiling experiments using NMR, with many studies characterizing the metabolomes of various tissues in mouse models of IBD.[92–94]

In 2011, a Japanese team employed GC/MS for the first time in IBD research, examining the metabolomes of a mouse model of colitis and human UC in separate studies.[95,96] In their animal study, Shiomi et al. analyzed serum and colon tissue of dextran sulfate sodium (DSS)-induced colitis mice, finding lower abundances of tricarboxylic acid (TCA) cycle metabolites and glutamine, tryptophan, tyrosine, asparagine, and glycine in the serum of colitis mice compared with controls.[96] In particular, Shiomi et al. found glutamine abundance to be positively correlated with inflammation, and proceeded to investigate whether glutamine supplementation would alleviate DSS-induced colitis.[96] Their hypothesis proved true, with results indicating that glutamine reduced colon tissue lesions in a dose-dependent fashion.[96] This important study demonstrates the potential for omics workflows to uncover novel medical tools.

Subsequently, the group published their GC/MS-based metabolite profiling in human UC, where they focused their effort on low molecular weight metabolites in the range of 35–600 mass/charge ratio (m/z), with an interest in amino acids and TCA cycle metabolites.[95] In this study, they once again found select TCA cycle metabolites to be decreased in disease when compared with control (lesion tissue vs normal tissue in UC), and reported decreased serum levels of glutamine in IBD compared with healthy controls.[95]

Recently, Baur et al. combined the advantages of NMR and LC-MS in a detailed study that delineated the intestinal tissue metabolome of a TNF-impaired ileitis mouse model that mimics CD.[97] With 1H NMR, the investigators characterized the metabolic differences in four different intestinal sections (distal jejunum, distal ileum, proximal colon, and distal colon) between inflamed and healthy mice, and employed a highly specific LC-MS methodology called selected reaction monitoring to quantify a panel of 63 inflammatory markers of interest in healthy and inflamed distal ileum tissue, finding wide-ranging alterations in metabolism of cholesterols, triglycerides, and phospholipids between disease and health.[97]

Almost simultaneously, independent researchers in USA and Canada performed targeted metabolite profiling using 1H NMR on urine collected from IBD patients, and on IBD serum, plasma, and urine, respectively.[98,99] Both groups reported only nominally differentiated metabolic profiles in CD and UC.[98,99]

Clinical Implications

Proteomics and metabolomics are fledging bioanalytical fields and many investigators have been inspired to quickly demonstrate the prodigious resolving power that these technologies are capable of. As bioanalytical methods advanced, biomarkers were developed that significantly affected IBD management; the monitoring of thiopurine metabolites to predict hepatotoxicity in thiopurine analogue therapy,[100] fecal calprotectin as a noninvasive measure of intestinal inflammation,[72] and ASCA and pANCA as complimentary tools in differentiating CD and UC,[85] among others[14] (Fig. 1). Nevertheless, the grievance seems to be that such tools, and novel definitive tools, have not come as a result of high-throughput, hypothesis-free "omics" methodologies.[12–14,101]

processing....