Background:
Proteomic analysis of formalin-fixed paraffin embedded tissues is challenging due to low sample recovery and poor LC-MS workflow compatibility. We developed an improved sample preparation workflow and data-independent acquisition strategy for quantitative analysis of colorectal cancer (CRC) FFPE specimens. We applied this approach to investigate molecular features associated with lymph node metastasis in minimally invasive tumours (i.e. T1-2 depth of invasion).
Methods:
Twenty Stage1 and 3a CRC FFPE sections (5x5 µm per sample) were macro-dissected, lysed, and digested according to the SP3 protocol. A pooled sample was used to acquire a DIA gas-phase fractionated chromatogram library. Wide window DIA data were acquired using an overlapping window scheme on a QExactive HFX mass spectrometer with 140min LC time using a 40cm x 75µm self-packed pulled column. For gene expression analysis, samples were analysed by Nanostring using the Cancer Progression kit profiling 770 genes.
Results:
We identified over 6000 proteins and quantified over 4000 proteins across all samples. Using unsupervised clustering methods, we were able to clearly distinguish the cancer Stages based on their proteomic profile. 237 proteins were significantly regulated (q-value <0.05, FC>2). GO-Enrichment analysis showed that in Stage 3a samples (glycoso)aminoglycan metabolic processes (e.g. Biglycan) as well proteins required for Epithelial to Mesenchymal transition were upregulated. Gene expression data for EMT related genes supported the proteomic findings. Two genes (TPM2 and CNN1) that were found significant in both proteomic and Nanostring data and showed upregulation in Stage 3a were also found to be significant for overall survival for patients when queried against the TCGA data for CRC.
Conclusion:
Protein expression phenotype from FFPE tissues of minimally invasive CRC tumours reveals features of EMT associated with lymph node metastasis.