Background: Tumor tissue represents a mixture of tumor cells, stroma and other cellular components. To minimize potential effect of tumor heterogeneity during proteomic analysis, several approaches such as LMD and coring technique have been used to selectively isolate tumor cells from other cellular components. Although these techniques can provide a relatively pure quantity of tumor specimens, the potential effect of sampling techniques on proteomic analysis is still largely unknown. In this study, we analyzed proteomic signature of tumor cells obtained by different sampling techniques from pancreatic cancer (PCA).
Method: Four pairs of PCA and tumor-matched normal pancreatic tissue, and four non-paired PCA samples were included. Tumor cells were harvested using three techniques (LMD, coring and bulk section). Peptides were separated on a Dionex Ultimate 3000 RSLC nano system (Thermo Scientific). Data was analyzed by the MS-PyCloud proteomics pipeline, using the MS-GF+ search engine to search against a concatenated target-decoy database.
Findings: A median number of 7388, 8202, 8213 proteins were identified in LMD, coring and bulk specimens, respectively. However, differential proteomic profiles were identified among samples. Differentially expressed proteins in tumors vs matched normal from both coring and bulk samples had a similar profile, and the upregulated-tumor proteins were more consistent with TCGA mRNA expression. Protein profiles, particularly the phosphoproteome in LMD samples, revealed a different pattern by clustering analysis. For top 150 upregulated-tumor proteins based on TCGA mRNA expression, we found 25 (17%), 75 (50%), and 51 (33%) proteins were upregulated in LMD, coring, and bulk samples.
Conclusion: Heterogeneity of tumor tissue is an important issue for proteomics research. Tumor cells can be isolated using LMD and coring technique with high purity. However, the data from LMD samples demonstrates a unique proteome profile, and with a fewer protein identification number.