Investigating neoplastic progression of ulcerative colitis with label-free comparative proteomics

Abstract

Patients with extensive ulcerative colitis (UC) have an increased risk of colorectal cancer. Although UC patients generally undergo lifelong colonoscopic surveillance to detect dysplasia or cancer in the colon, detection of cancer in this manner is expensive and invasive. An objective biomarker of dysplasia would vastly improve the clinical management of cancer risk in UC patients. In the current study, accurate mass and time methods with ion intensity-based label-free proteomics are applied to profile individual rectal and colon samples from UC patients with dysplasia or cancer (UC progressors) compared to rectal samples from patients that are dysplasia/cancer free (UC nonprogressors) to identify a set of proteins in the rectum mucosa that differentiate the two groups. In addition to the identification of proteins in UC dysplastic colon tissue, we for the first time identified differentially expressed proteins in nondysplastic rectal tissue from UC progressors. This provides a candidate pool of biomarkers for dysplasia/cancer that could be detected in a random nondysplastic rectal biopsy. Mitochondrial proteins, cytoskeletal proteins, RAS superfamily, proteins relating to apoptosis and metabolism were important protein clusters differentially expressed in the nondysplastic and dysplastic tissues of UC progressors, suggesting their importance in the early stages of UC neoplastic progression. Among the differentially expressed proteins, immunohistochemistry analysis confirmed that TRAP1 displayed increased IHC staining in UC progressors, in both dysplastic and nondysplastic tissue, and CPS1 showed a statistically significant difference in IHC staining between the nonprogressor and progressor groups. Furthermore, rectal CPS1 staining could be used to predict dysplasia or cancer in the colon with 87% sensitivity and 45% specificity, demonstrating the feasibility of using surrogate biomarkers in rectal biopsies to predict dysplasia and/or cancer in the colon.

Figure 1

Experimental workflow. 5 samples from each experimental group were processed. Samples were run in replicate on LTQ-Orbitrap. All runs were searched with X!Tandem against a human IPI database; LC-MS feature-finding was performed. AMT database matching assigned peptide IDs to LC-MS features. Replicate run information was combined for each sample (not shown). LC-MS features were aligned across all samples by mass and RT. Peptide intensity ratios were calculated between the P-NEG (yellow) and NP (blue) groups, and between the P-HGD (red) and NP groups; these ratios were combined for each protein (PNEG:NP in green, PHGD:NP in purple). Peptide t-scores were calculated to assess group differences; t-scores were summarized for each protein (purple) and compared with the t-scores from all other peptides (gray). p-values and q-values were calculated for each protein. Candidate proteins chosen based on q-value missing or ≤ 0.1, and ratio ≤ 0.5 or ≥ 2.

Figure 2

Figures 2A and 2B contain separate charts comparing the P-HGD group with the NP group and the P-NEG group with the NP group. A. Histogram of q-values for quantitated proteins with sufficient peptide evidence. B. A scatter plot relating the log of the protein ratio (horizontal axis) to the protein q-value (vertical axis). Blue dots indicate proteins with q-value ≤ 0.1.

Figure 3

A. A scatter plot of peptide log-intensity values for LC-MS features with the same peptide identification and charge state, for one sample. Horizontal axis: log-intensity in replicate run 1. Vertical axis: log-intensity in replicate run 2. B. A scatter plot of protein log-ratios calculated between the P-HGD and NP groups using only the data from the first replicate run from each sample (horizontal axis) vs. the second replicate run (vertical axis). C. As B, but with data from the comparison between P-NEG and NP.

Figure 4

Log-ratio distributions comparing abundance calculated using only replicate 1 vs. abundance calculated using only replicate 2 of each sample, for each of the three experimental groups. For all three sets of replicate experiments, 96% of the proteins display less than 2-fold change between the replicates.

Figure 5

Selected enriched clusters of the differential proteins in UC progressors. Functional enrichments were analyzed by the DAVID online database. A full list of enrichment clusters is presented in supplemental Tables 3 and 4.

Figure 6

IHC analysis of CPS1. A. CPS1 IHC scores in non-progressor non-dysplastic rectal tissues (NP-neg-R), progressor non-dysplastic rectal tissues (P neg-R), and progressor dysplastic or cancerous tissues (P dysplasia/CA). The IHC scores are available in supplemental Table 7. B. ROC analysis of CPS1 staining in rectal tissues of progressors and non-progressors.

Figure 7

IHC analysis of TRAP1 in UC non-progressors and progressors. Moderate to strong staining of TRAP1 was observed in non-dysplastic tissues (B and E) and dysplastic tissues (C and F) from progressors compared to the minimal staining in the non-dysplastic tissues from non-progressors (A and D).