Protein Folding Stability Profiling of Colorectal Cancer Chemoresistance Identifies Functionally Relevant Biomarkers

Abstract

Reported here is the application of three protein folding stability profiling techniques (including the stability of proteins from rates of oxidation, thermal protein profiling, and limited proteolysis approaches) to identify differentially stabilized proteins in six patient-derived colorectal cancer (CRC) cell lines with different oxaliplatin sensitivities and eight CRC patient-derived xenografts (PDXs) derived from two of the patient derived cell lines with different oxaliplatin sensitivities. Compared to conventional protein expression level analyses, which were also performed here, the stability profiling techniques identified both unique and novel proteins and cellular components that differentiated the sensitive and resistant samples including 36 proteins that were differentially stabilized in at least two techniques in both the cell line and PDX studies of oxaliplatin resistance. These 36 differentially stabilized proteins included 10 proteins previously connected to cancer chemoresistance. Two differentially stabilized proteins, fatty acid synthase and elongation factor 2, were functionally validated in vitro and found to be druggable protein targets with biological functions that can be modulated to improve the efficacy of CRC chemotherapy. These results add to our understanding of CRC oxaliplatin resistance, suggest biomarker candidates for predicting oxaliplatin sensitivity in CRC, and inform new strategies for overcoming chemoresistance in CRC.

Keywords: chemical denaturation; elongation factor 2; fatty acid synthase; limited proteolysis; mass spectrometry; oxaliplatin; thermal denaturation.

Figure 1.

(A) Oxaliplatin sensitivity for a panel of lab-cultured colorectal cancer cell lines. The cell lines highlighted in lavender and pink are the oxaliplatin sensitive and resistant cell lines used in the protein expression and protein stability profiling studies performed in this work. (B) Chemical structure of oxaliplatin. (C) Schematic representation of the workflow used for the protein stability profiling experiments performed on the oxaliplatin resistant and sensitive CRC cell lines and PDX mouse models using the SPROX, TPP, and LiP techniques.

Figure 2.

Volcano plots generated in the SPROX, TPP, LiP, and expression level analyses performed on the CRC Cell lines (A-D, respectively) and PDX mouse models (E-H, respectively) in this work. (A-H) Red dots indicate the peptides (SPROX and LiP) or proteins (TPP) selected as differentially stabilized in the oxaliplatin resistant and sensitive cell lines and PDXs using a p-value of 0.01 which is indicated by the dotted, horizontal lines. The black dots indicate the proteins (or peptides) not selected as hits.

Figure 3.

Venn diagrams showing the overlapping protein stability hits in the (A) CRC cell line and (B) PDX mouse model analyses using the SPROX, TPP, and LiP techniques. The Venn diagrams in (C) and (D) show the overlap between all the protein stability hits and the protein expression level hits in the CRC cell line and PDX mouse model analyses, respectively. The light-brown circles in (C) and (D) represent the subset of proteins identified as hits in at least two of the protein stability profiling techniques. The number of proteins in each area is labeled in the figure.

Figure 4.

A. Cell sensitivity assay on RKO with oxaliplatin, cerulenin, and A-484954 revealed IC50 values of 1.8 μM, >100 μM, and 5.8 μM, respectively. B. Synergy analysis revealed synergy between oxaliplatin with both cerulenin and A-484954.

Figure 5.

Over-representation analysis of the GO cellular component terms enriched in differentially (A) expressed and (B) differentially stabilized protein hits. Protein network graphs showing the relationship between enriched GO terms in the (C) expression level and (D) stability profiling analyses. The color of the dots indicates the p-value for the enrichment, and the size of the dots indicates the number of the hits that are correlated with the corresponding GO term.