Proteomic analysis-based discovery of a novel biomarker that differentiates intestinal Behçet’s disease from Crohn’s disease

Intestinal Behçet’s disease (BD) and Crohn’s disease (CD) present similar manifestations, but there are no specific diagnostic tests to differentiate them. We used a proteomic approach to discover novel diagnostic biomarkers specific to intestinal BD. Colon mucosa tissue samples were obtained from patients with intestinal BD or CD using colonoscopy-guided biopsy of the affected bowel. Peptides from seven intestinal BD and seven CD patients were extracted and labeled using tandem mass tag (TMT) reagents. The labeled peptides were identified and quantified using liquid chromatography-tandem mass spectrometry (LC–MS/MS). The proteins were further validated using immunohistochemical (IHC) analysis with tissue samples and an ELISA test with serum samples from 20 intestinal BD and 20 CD patients. Using TMT/LC–MS/MS-based proteomic quantification, we identified 39 proteins differentially expressed between intestinal BD and CD. Beta-2 glycoprotein 1 (APOH) and maltase-glucoamylase (MGAM) showed higher intensity in the IHC staining of intestinal BD tissues than in CD tissues. The serum MGAM level was higher in intestinal BD patients. Proteomic analysis revealed that some proteins were differentially expressed in patients with intestinal BD compared with those with CD. Differential MGAM expression in intestinal BD suggests its role as a potential novel diagnostic biomarker.


Intestinal Behçet's disease (BD) and Crohn's disease (CD) present similar manifestations, but
there are no specific diagnostic tests to differentiate them. We used a proteomic approach to discover novel diagnostic biomarkers specific to intestinal BD. Colon mucosa tissue samples were obtained from patients with intestinal BD or CD using colonoscopy-guided biopsy of the affected bowel. Peptides from seven intestinal BD and seven CD patients were extracted and labeled using tandem mass tag (TMT) reagents. The labeled peptides were identified and quantified using liquid chromatography-tandem mass spectrometry (LC-MS/MS). The proteins were further validated using immunohistochemical (IHC) analysis with tissue samples and an ELISA test with serum samples from 20 intestinal BD and 20 CD patients. Using TMT/LC-MS/MS-based proteomic quantification, we identified 39 proteins differentially expressed between intestinal BD and CD. Beta-2 glycoprotein 1 (APOH) and maltase-glucoamylase (MGAM) showed higher intensity in the IHC staining of intestinal BD tissues than in CD tissues. The serum MGAM level was higher in intestinal BD patients. Proteomic analysis revealed that some proteins were differentially expressed in patients with intestinal BD compared with those with CD. Differential MGAM expression in intestinal BD suggests its role as a potential novel diagnostic biomarker. Methods Study population and sample collection. In total, 47 patients with intestinal BD and 47 patients with CD were recruited from the IBD Clinic of Severance Hospital, Seoul, Korea. Intestinal BD and CD were diagnosed according to clinical, histological, endoscopic, and radiological criteria 16,17 . Exclusion criteria included indeterminate colitis, ulcerative colitis, intestinal tuberculosis, a history of malignancy, or insufficient available medical records. The intestinal BD ulcer type was defined according to the Korean Inflammatory Bowel Disease study group, and the CD phenotype was defined according to the Montreal Classification 18,19 . The severity of intestinal BD was determined using the Disease Activity Index for Intestinal Behçet's Disease (DAIBD) score, and CD severity was determined using the Crohn's Disease Activity Index (CDAI) score 20,21 . The intestinal mucosa tissue samples were obtained from patients with intestinal BD or CD using colonoscopy-guided biopsy of the affected bowel. The tissue and serum samples were preserved at −80 °C.

Abbreviations
Informed consent was obtained from all individuals enrolled in this study. This study was approved by the Institutional Review Board of Yonsei University College of Medicine (IRB No: 2012-0039-030) and was conducted in accordance with the Declaration of Helsinki.
TMT sample processing and protein quantitation. For proteomic analysis, the colon mucosa tissue samples were lysed and labeled with tandem mass tag (TMT, Thermo Scientific, San Jose, CA, USA) according to the manufacturer's instructions. The labeled peptide samples were pooled into a new vial and dried using SpeedVac (Thermo Scientific). The following processes, including strong cation exchange fraction and liquid chromatography (LC)-mass spectrometry (MS) and database searching, were performed by Poochon Scientific (Frederick, MD) as described previously 22  www.nature.com/scientificreports/ plex) fractionation was performed using an Agilent AdvanceBio Column and Agilent UHPLC 1290 system (Agilent, Santa Clara, CA). LC/MS/MS analysis was performed using a Thermo Scientific Q-Exactive hybrid Quadrupole-Orbitrap Mass Spectrometer and Thermo Dionex UltiMate 3000 RSLCnano System (Thermo Scientific). Raw MS data files were searched against the human protein sequence databases obtained from the NCBI website using Proteome Discoverer 1.4 software (Thermo Scientific) based on the SEQUEST and percolator algorithms. The false positive discovery rate (FDR) was set at 5%. The resulting Proteome Discoverer Report from Poochon Scientific contained all assembled proteins with peptide sequences and peptide spectrum match counts (PSM#) and TMT-tag-based quantification ratios.
Immunohistochemistry (IHC). Formalin-fixed paraffin-embedded colonic biopsy sections from patients with intestinal BD (n = 20) and those with CD (n = 20) were stained with 39 antibodies and counterstained with hematoxylin-eosin. Antibodies against APOH (HPA003732) and MGAM (HPA002270) were purchased from Atlas Antibodies AB (Bromma, Sweden). Detailed antibody information, antibody dilution factors, and antigen retrieval methods are provided in Supplementary Table I. IHC analysis was performed as previously described 23 . Primary antibodies (1:1000 dilution for anti-APOH and 1:2000 dilution for anti-MGAM) were applied overnight at 4 °C. Staining was visualized using an Olympus BX43 microscope with the Olympus CellSens Entry software (Hamburg, Germany). We used a semi-quantitative grading method to assess protein expression, as described previously 24 . Briefly, 100 × magnification was used to grade all fields for each sample, and the staining intensity was scored from 1 to 3. Staining extent was scored from 1 to 4. The scores were multiplied together, and the final scores were classified as follows: 1-3, weak; 4-8, moderate; and 9-12, strong staining. Fisher's exact test was used to assess the immunochemical scores for protein expression. In this particular test, values of p less than 0.2 were determined as a threshold for moving to the next validation test.

Enzyme-linked immunosorbent assay (ELISA). Serum samples from patients with BD (n = 20) and
those with CD (n = 20)-constituting an independent cohort of patients distinct from those providing samples for IHC staining-were analyzed using ELISA. The serum was diluted 1:50,000 for APOH and 1:200 for MGAM. ELISA was performed according to the manufacturer's instructions using the APOH ELISA kit (KA0982, Abnova, Taipei, Taiwan) and the MGA ELISA kit (MBS2021345, MyBioSource, San Diego, CA).

Data analysis and statistical methods.
A heat map was generated using the "limma" package for R 25 .
The fold changes of proteins and p values were calculated using linear regression in "limma. " The heat map was drawn using the heatmap.2 function in the "gplots" package for R (https:// CRAN.R-proje ct. org/ packa ge= gplots), and colors were changed using the "RColorBrewer" package for R (1.1-2. https:// CRAN.R-proje ct. org/ packa ge= RColo rBrew er). Volcano plots were plotted using GraphPad Prism 8 software. Data were expressed as the mean ± SEM. Student's t-test was used for the statistical analysis of serum ELISA data. Statistical significance was set at p < 0.05.  Table 1. The mean age of the patients with intestinal BD and CD was 51. 0 ± 11.4 years and 36. 1 ± 16.0 years (p < 0.001), respectively, which were comparable to those obtained in a previous study 7 . A total of 42.6% of the patients with intestinal BD and 74.5% of the patients with CD were men. The mean disease duration of intestinal BD and CD was 7.1 ± 6.0 years and 8.0 ± 5.5 years, respectively. Oral ulcers, genital ulcers, skin manifestations, and joint manifestations were more common in patients with intestinal BD and perianal lesions were more common in those with CD. Intestinal BD ulcers were located in the ileocecal valve (74.5%), ascending colon (10.6%), and postoperative anastomosis site (14.6%). The intestinal BD ulcers occurred with the following frequency: solitary (48.9%), two (19.1%), and multiple (31.9%). The CD showed localization in the ileal (14.9%), colonic (2.1%), and ileocolic (83.0%) regions and could be characterized as non-stricturing and non-penetrating (63.0%), stricturing (15.2%), and penetrating (21.7%). The number of patients with a DAIBD score ≥ 40 was 35 (74.5%) and that of patients with a CDAI score ≥ 150 was 24 (51.1%).

Discovery of biomarker candidates using TMT/LC-MS/MS-based proteomic approach.
The colon mucosa tissue samples of patients with intestinal BD (n = 7) and those with CD (n = 7) were analyzed using a TMT/LC-MS/MS-based proteomic approach to identify intestinal BD-specific markers (Fig. 1). A total of 3,266 proteins were quantitatively identified, and at least two identified peptides were detected for each protein ( Fig. 2A). Logistic regression analysis showed that 39 proteins were significantly different between intestinal BD and CD patients (p < 0.05) (Fig. 2B). Among them, 34 proteins were overexpressed, and 5 proteins were underexpressed in patients with intestinal BD compared to patients with CD ( Table 2). The proteins were classified by cellular compartment (CC) based on gene ontology (GO) enrichment analysis using a functional annotation tool (DAVID Bioinformatics Resources, version 6.8) (Supplementary Table II www.nature.com/scientificreports/ a large portion of colon mucosal proteomes were composed of secretion-related proteins (Fig. 2C). Notably, APOH and MGAM were included in all six secretion-related CC categories in the GO analysis.
IHC validation of biomarker candidates. IHC analysis was performed to validate 39 candidate biomarkers. A total of 40 colonic biopsy sections from patients with intestinal BD (n = 20) and CD (n = 20) were stained with each antibody and analyzed using a semi-quantitative grading method as described in the Methods section. Seven candidate proteins were selected based on their differential expression between intestinal BD and CD: maltase-glucoamylase (MGAM), beta-2 glycoprotein 1 (APOH), plasminogen (PLG), pro-interleukin 16 (IL16), serine/arginine-rich splicing factor 3 (SRSF3), clusterin (CLU), and serine/threonine-protein phosphatase 4 catalytic subunit (PPP4C). Fisher's exact test showed that APOH (p = 0.039) and MGAM (p = 0.192) levels were consistently higher in intestinal BD than in CD, suggesting that these two proteins might be distinctive biomarkers of intestinal BD distinguishing it from CD. There were no significant differences in the expression of PLG (p = 0.480), IL16 (p = 0.563), SRSF3 (p = 0.591), CLU (p = 1.000), or PPP4C (p = 0.450) between intestinal BD and CD. IHC analysis of the colon mucosa tissue obtained from patients with intestinal BD showed that stronger APOH immunoreactivity was detected in a portion of the lamina propria (Fig. 3A, lower panels), and the MGAM immunoreactive signal was exclusively present in the brush border membrane of the epithelium (Fig. 3B, lower panels), consistent with previous findings 26 .
Serum ELISA validation of biomarker candidates. Since APOH and MGAM were included in all six secretion-related CC categories of GO analysis (Fig. 2C), we hypothesized that these proteins could also be detected in the blood. This notion was supported by several biochemical studies that identified APOH as a component of circulating plasma lipoproteins 27 , and the fact that MGAM is found in brush border membrane vesicles 28 . Both proteins were also recently detected in exosomes using in-depth proteomic analyses 29 . Although there was no difference in APOH concentration between the two groups (Fig. 4A), serum concentrations of MGAM were statistically higher in patients with intestinal BD compared to those with CD (p < 0.05), when the two candidate biomarkers were further tested with serum ELISA in an independent validation cohort (intestinal BD, n = 20; CD, n = 20) (Fig. 4B). These results suggest that MGAM can be a specific, diagnostic biomarker of intestinal BD. www.nature.com/scientificreports/ Analysis of the MGAM-related pathway in the colon mucosa proteome. The DAVID functional annotation tool was used to identify the biological pathways of the colon mucosa proteome of intestinal BD and CD patients, in which MGAM could be involved. The Kyoto Encyclopedia of Genes and Genomes (KEGG) demonstrated that 85 pathways were significantly enriched (p < 0.05) (Fig. 5A). Among them, MGAM was involved in three enriched pathways: metabolic pathways (hsa01100, 467 proteins), galactose metabolism (hsa00052, 19 proteins), and starch and sucrose metabolism (hsa00500, 17 proteins). Unexpectedly, APOH was not related to any of the 85 enriched pathways.
To determine whether MGAM and APOH share biological functions, 3,266 proteins were divided into three GO annotation categories: cellular compartment (CC), molecular function (MF), and biological process (BP). Next, the categories containing MGAM or APOH were selected and compared. There were six CC categories, but no MF or BP categories, correlating with both MGAM and APOH as represented in Figs. 2C and 5B-D. This suggests that the physiological roles of MGAM and APOH might differ in the colon mucosa, but both are assumed to be secreted.

Predicted upstream regulators and disease-related functions of MGAM in intestinal BD and CD.
Ingenuity pathway analysis (IPA) was performed to predict canonical pathways, upstream regulators, and disease-related functions to gain more insight into the pathophysiological role of MGAM in intestinal BD and CD. As shown in Supplementary Fig. 1A, the only canonical pathway related to MGAM was enriched in "Glycogen degradation III. " Although absolute z scores (CD, − 0.302; BD, 0.302) were lower than the cut-off (z-score > 2), the pathway significantly differed between intestinal BD and CD (p < 0.000001). The pathway was predicted to be inhibited in CD and activated in intestinal BD (Supplementary Fig. 1B). Upstream regulator analysis revealed that the dexamethasone-related pathway was inactivated in CD compared to its status in intestinal BD (Fig. 6A), but the medication history of corticosteroids (31.9%) was similar between the two patient groups participating in this study (Table 1). This result is consistent with previous observations that patients with CD require corticosteroid therapy more often than patients with intestinal BD 7 . Comparison of the diseases and functional analyses of the two patient groups revealed that the activation of degranulation and the inactivation of genitourinary carcinogenesis were conserved, and the former is predicted to be more upregulated in intestinal BD than in CD, and the latter is more downregulated in CD than in intestinal BD (Fig. 6B).

Discussion
Accurate diagnosis of intestinal BD and CD is important for establishing proper treatment plans and predicting disease prognosis 30 . Clinical, laboratory, and endoscopic approaches have limitations in differentiating between these two diseases. Gastrointestinal and systemic symptoms, elevated inflammatory markers, and endoscopic findings, such as asymmetric deep ulcers in the ileocecal valve, are often shared between the two diseases in clinical practice 31 . We quantified 3,266 proteins from the colon mucosa tissue samples, identified 39 novel diagnostic biomarkers, and validated the MGAM protein as a novel diagnostic biomarker using patient serum samples. To the best of our knowledge, this study is the first to use proteomics to identify a diagnostic marker that can differentiate intestinal BD from CD.
Quantitative proteomic analysis using isobaric chemical labeling, including super-stable isotope labeling with amino acids in cell culture, isobaric tags for relative and absolute quantitation, or TMT, is emerging as a highly effective approach with good quantification performance and reproducibility for profiling new biomarkers in numerous diseases 32 . These relatively new proteomic techniques enable the discovery of diagnostic biomarkers by providing methods not only for peptide identification, but also for the quantification of biological samples. Thus far, they have been applied to IBD 33 or intestinal TB 24 , but not to intestinal BD. We identified 39 potential novel diagnostic biomarkers using quantitative proteomic analysis, and validated MGAM as a biomarker for differentiating intestinal BD from CD.
MGAM is involved in carbohydrate digestion in the small intestine. MGAM deficiency has been reported in congenital diarrheal diseases 34 . MGAM and sucrase-isomaltase (SI) have identical exon structures. They are anchored in the small intestinal mucosal brush border and hydrolyze substrates to glucose 35 . Romach et al. reported a trinitrobenzene sulfonic acid-induced colitis rat model that showed a loss of SI expression and Here, the tissue expression and serum concentrations of MGAM were lower in patients with CD patients than in those with intestinal BD. Our IPA analysis also predicted that the MGAM-related "glycogen degradation III" pathway was inactivated in CD ( Supplementary Fig. 1A,B). Consistent with our findings, the involvement of the small intestine is relatively common in CD than in intestinal BD, resulting in digestive problems and nutrient malabsorption. Thus, differential MGAM expression may be related to different clinical manifestations between intestinal BD and CD, which should be further validated. MGAM has been shown to be important in neutrophil biology but not in lymphocytes 38,39 . We identified MGAM in the colon mucosa tissue samples, but MGAM was also detected in the serum samples. The detected serum MGAM may be secreted from the gastrointestinal tract or contained in neutrophils. All five degranulationrelated pathways (degranulation, degranulation of cells, degranulation of phagocytes, degranulation of neutrophils, and degranulation of granulocytes) enriched in the disease or functional annotations with high activation z-score indicated that MGAM might be involved in the degranulation process in which mast cell activation is essential. Although little is known regarding the degranulation in intestinal BD, degranulation signaling in IBD has been implicated in the regulation of inflammatory responses in the gastrointestinal tract, where the largest    www.nature.com/scientificreports/ population of mast cells in the body resides [40][41][42][43] . Higher MGAM levels were detected in intestinal BD patients than in CD patients, and this was related to innate immunity pathogenesis. Further studies are required to understand how MGAM influences immune cell degranulation and the resulting gut inflammation. Moreover, it would be valuable to determine whether MGAM levels correlate with the disease activity of intestinal BD.
Here, MGAM was shown to differentiate BD from CD with an area under the curve of 0.805 (95% confidence interval, 0.665-0.945), 85% sensitivity, and 70% specificity at a cut-off of 150 ng/mL (Fig. 7). However, further studies with a larger sample size are warranted to validate our data on the role of MGAM in chronic enterocolitis and to understand its molecular mechanism in depth. In addition, a significant proportion of our CD cohort population had perianal involvement, which distinguishes CD from intestinal BD. Perianal lesions in CD occur more frequently in Korea than in Western countries, but the ideal cohort population for the development of diagnostic biomarkers should present common or resembling clinical, endoscopic, radiologic, and histologic features of the two diseases. Future studies using such a cohort will be needed to validate MGAM as a distinguishing diagnostic biomarker. Finally, the measurement of MGAM levels in a variety of inflammatory diseases such as ulcerative colitis and intestinal tuberculosis could be worthwhile.
In summary, we used TMT-based proteomic quantification to identify 39 candidate proteins that were differentially expressed between intestinal BD and CD. Then, we selected APOH and MGAM proteins as possible biomarkers for intestinal BD based on the results of IHC staining and semi-quantitative grading. Finally, we suggest that the levels of MGAM protein in patient serum can potentially be used to differentially diagnose intestinal BD from CD.