In this study samples from both inflamed and non-inflamed tissues from each patient were analyzed. The histological assessment of the biopsies confirmed that the patient biopsy from the left colonic flexure could be assessed as non-inflamed and did not show any signs of sub-inflammation. This is the first study to assess a relatively high number of UC patients using this approach and in spite of large individual heterogeneity we found statistical support for tissue specific levels of many proteins being associated with disease exacerbation. Recently, proteomic studies have contributed with promising candidate markers being correlated with disease phenotypes, either in IBD patients in general, specifically within UC patients or between UC, CD and control persons [20
In this study, the spots characterized by a larger than twofold difference in protein level were classified as the most promising markers. These spots together with the additional spots identified included proteins already known to be associated with inflammation states of IBD (e.g. triosephosphate isomerase, alpha enolase, selenium binding protein, superoxide dismutase, thioredoxin-dependent peroxide reductase, Hsp60, Hsp70) [19
], as well as novel proteins (e.g. glycerol-3-phosphate-dehydrogenase, isocitrate dehydrogenase, L-lactate dehydrogenase B-chain, inorganic pyrophosphatase, enoyl-CoA hydratase, peroxiredoxin-4, peroxiredoxin-6), where the association with UC and inflammation needs replication. Integration of excised biopsies from control persons was used to strengthen our conclusions, since protein spots showing different protein levels between RE and LF colon mucosa of control persons could be excluded as merely position specific markers, and not disease markers. Furthermore, there is a strong support for regarding our significant spots as true positives since only nine false positives are expected out of 222 spots (q
0.041), and only 0.02 spot out of the 39 highly significant (q
The proteomic profile of the inflamed tissue was very distinct with 37% of the annotated protein spots having significantly different protein levels when comparing inflamed and non-inflamed patient mucosa. Focusing on the spots found to have at least more than twofold difference in protein level between patient biopsies, 70% (14 spots out of 20 possible) were successfully identified. Further identification of additional protein spots with strong statistical support strengthened the overall understanding of how inflammation changes the proteomic signature of UC patient colon biopsies. We found that, apart from the highly abundant proteins like serum albumin and cytoplasmic actin, excised spots could primarily be assigned to proteins involved in stress response mechanisms and energy metabolism. Different levels of these proteins could infer that acute inflammation in UC patients in particular impair and affect regulation of these biological processes.
Different protein profiles of several glycolytic enzymes (triosephosphate isomerase, glycerol-3-phosphate-dehydrogenase, alpha enolase) and other proteins involved in energy generation (isocitrate dehydrogenase, L-lactate dehydrogenase B-chain, inorganic pyrophosphatase, enoyl-CoA hydratase) could indicate inflammation associated alterations in energy metabolism in UC patients. This could be due to malfunction of the utilization of n-butyrate, which is the preferred fuel for colonocytes in the distal colon [36
]. The observed lowered enoyl-CoA hydratase in inflamed tissue infer impaired fatty acid oxidation and the energy-deficiency is further strengthened by change in the glycolytic pathway reflected as a higher expression of glycolytic enzymes in inflamed mucosa. In accordance with the present findings, Hsieh et al. [22
] observed a down-regulation of triosephosphate isomerase in UC-diseased colon mucosa, while Nanni et al. [35
] observed an up-regulation of the same enzyme in intestinal epithelial cells from CD patients. In addition, a changed regulation of alpha-enolase has been associated with IBD [37
], which has been suggested to reflect anaerobic glycolysis in UC patients with inflamed pouch [34
]. However, changes in the level of alpha-enolase could also reflect non-glycolytic mechanisms, since the protein is also induced as a heat shock protein under hypoxic stress [38
]. Induced expression of heat shock proteins due to cellular stress would probably be expected in IBD patients [39
] and is further reflected in the lower expression of 60 and 70
kDa heat shock proteins observed in inflamed tissue.
Inflammation in IBD patients generally increases the level of reactive oxygen metabolites resulting in oxidative stress due to an imbalance between antioxidants and reactive oxygen [42
]. Different levels of several antioxidant proteins (selenium binding protein, superoxide dismutase, thioredoxin-dependent peroxide reductase, peroxiredoxin-4 and peroxiredoxin-6) suggest an increased level of oxidative stress in UC patients, and earlier studies have specifically documented a regulation of selenium-binding protein and superperoxide dismutase in relation to patients with IBD [26
]. Impaired oxidative metabolism and affected antioxidant defenses thus seem to play an important role in the pathogenesis of IBD diseases, and changed expression of involved protein is most likely related to a stimulated cell activity (e.g. neutrophils) due to severe inflammation.
Generally, the control persons demonstrated a high variability among individuals reflecting an individual protein signature rather than a strong position specific proteomic profile of biopsies excised from the same bowel location in different individuals. However, some proteins were still found to be significantly differently expressed between RE and LF tissues, indicating that these could be assigned as position markers. Two of the spots found to represent very different protein levels within the UC patients fell into this category (α-enolase (spot 10) and hydroxymethyl glutaryl-CoA synthase). These proteins could falsely have been ascribed as good candidates for disease markers if the control persons were not taken into account. Regardless of the need for control persons in terms of validation of markers as actual disease markers, and not just position markers, internal controls within the same patient seems to be a robust and more simplistic approach to implement. Here, the control persons were healthy, while in other studies, controls have been patients with e.g. colorectal carcinoma [19
], which could complicate the interpretation of results, since UC and colorectal carcinoma are closely linked conditions, and even tumor free tissue from cancer patients could deviate from normal tissue [47
]. We have not addressed the individual variation in protein level between patients, which could be substantial [19
], but instead focused on protein spots having robust signatures across patients.
Overall, 2-DGE analysis has improved in relation to reproducibility, repeatability and better image analysis, and in this study the limitation related to PMF using MALDI-TOF MS technology, was due to the quantity of the protein spots and sensitivity of the instrument, which played a major role for successful protein identification. Despite its flaws, MS-based proteomic techniques are efficient in providing new insights into pathogenesis of diseases not only through identification of involved proteins but also for potential effects of different isoforms, like post-translational modifications [48
]. We observed several individual proteins being assigned to multiple spots (e.g. alpha-enolase), but the underlying modifications were not studied further. Some proteins with horizontal patterns could for instance, indicate phosphorylations, changing the charge of proteins as observed in the train of protein spots assigned to selenium binding protein.
Protein extraction from whole colonic biopsies as conducted here means that a long and tedious procedure for isolation of e.g. intestinal epithelial cells is avoided. On the other hand the proteomic profile from diverse cell types of both intestinal and immune origin could potentially result in larger variability between the samples and a limited control of cell specific signatures, with a risk of some cell types overruling important expression patterns of others [8
]. Essentially, whole biopsies from IBD patients contain multiple populations of cells, including inflammatory cells, and thereby proteins identified could potentially reflect other mechanisms resulting from i.e. cell death or serum. To avoid introduction of sources of errors whole biopsies is preferred for clinical use. Our study thus resembles the clinical setting in this respect. Precautions in relation to collection, processing and storage should be taken seriously for minimizing factors that potentially could alter the molecular composition of the tissue material [49
]. Here, the biopsies were stored immediately on dry ice after collection and subsequently on −80°C, which in combination with gentle handlings throughout all analyses ensure a high quality of the collected material.
The patients in the UC cohort included in this study received different medical treatments, which potential could affect their protein profiles. Based on the PCA plot, the variation among the patient group treated with 5-ASA was, however, not different from the remaining patients, and therefore it was not evident that the treatment had a strong effect on the proteomic signature of the analyzed biopsies.
In conclusion, we have generated insights into the underlying mechanisms of active UC. The overall proteomic signature of inflamed colon mucosa was strong. Such assessment of biopsies from the active site of the disease reveal protein markers associated with inflamed tissue, and could be an important entry point for the discovery of new and improved non-invasive markers [24
]. Thus, future studies further addressing the marker proteins found are essential for evaluation of disease specificity and clinical relevance.