Mass Spectrometry-Based Workflow for the Identification and Quantification of Alternative and Canonical Proteins in Pancreatic Cancer Cells

Abstract

The identification of small proteins and proteins produced from unannotated open reading frames (called alternative proteins or AltProts) has changed our vision of the proteome and has attracted more and more attention from the scientific community. Despite several studies investigating particular AltProts in diseases and demonstrating their importance in such context, we are still missing data on their expression and functions in many pathologies. Among these, pancreatic ductal adenocarcinoma (PDAC) is a particularly relevant case to study alternative proteins. Indeed, late detection of this disease, notably due to the lack of reliable biomarkers of early-stage PDAC, and the fact that tumors rapidly develop resistance to most of the treatments used in the clinics warrant the exploration of new repertoires of molecules. In the present article, we aim to investigate the alternative proteome of pancreatic cancer cell lines as a first attempt to decipher the expression of AltProts in PDAC. Thanks to a combined data-dependent and data-independent acquisition mass spectrometry workflow, we were able to identify tryptic peptides matching 113 AltProts in a panel of 6 cell lines. In addition, we identified AltProts differentially expressed between pancreatic cancer cell lines and other cells (HeLa and HEK293T). Finally, mining the TCGA and Gtex databases showed that the corresponding transcripts encoding several AltProts we identified are differentially expressed between PDAC tumors and normal tissues and are correlated with the patient's survival.

Keywords: alternative proteins; data independent acquisition; microproteins; pancreatic ductal adenocarcinoma; proteomics; short open reading frame-encoded peptides.

Figure 1

Mass spectrometry-based workflow to identify alternative proteins in pancreatic cancer cell lines. Proteins were extracted from pancreatic or HeLa and HEK293T cell lines, digested with trypsin and most of the resulting peptides were fractionated using a high pH reverse phase. The resulting fractionated peptides were analyzed in data-dependent acquisition (DDA) mode, and the generated raw data were processed using Proline with the OpenProt database to identify alternative and canonical proteins. Peptide spectrum matches were manually validated, and a curated spectral library was created. In parallel, unfractionated peptides were analyzed in data-independent acquisition (DIA) mode, and the generated raw data were processed using DIA-NN and the curated spectral library created using DDA data.

Figure 2

Distribution of amino acid length, isoelectric point, and source of production of the alternative proteins identified. (A) Distribution of the amino acid length of the AltProts identified in this study. (B) Repartition of AltProts depending on their isoelectric point. (C) Proportion of alternative proteins produced from ORFs from novel isoforms, ORFs located in the 5′UTR (uORF), coding sequence (intORF), or 3′UTR (dORF) of mRNAs, from non-annotated ORFs on long non–coding RNAs (lncRNA) or miscellaneous RNA (misc_RNA). The nomenclature of the different classes of ORFs was adapted from [29].

Figure 3

Identification of protein domains and short linear motifs in alternative proteins. (A) Proportion of AltProts with at least one predicted disordered region predicted by InterPro. (B) Proportion of AltProts with at least one known protein domain identified by InterPro. (C) Frequency of protein domains and number of proteins bearing a particular protein domain in the AltProts identified. (D) Counts of the different classes of SLiMs identified from the AltProts sequence. SLiM classes are targeting sites for subcellular localization (TRG), post-translational modification sites (MOD), ligand-binding sites (LIG), docking sites (DOC), degradation sites (DEG), and proteolytic cleavage sites (CLV).

Figure 4

Quantitative analysis of the data-independent acquisition of the alternative proteins between pancreatic cancer cells and HeLa and HEK293T. (A) Volcano plot representing the log 2 ratio (pancreatic cell lines/other cell lines) for each AltProt quantified in the DIA analysis and the corresponding Welch’s t-test p-value (−Log₁₀ transformed). The blue dots represent the AltProts more abundant in pancreatic cancer cell lines (p-value < 0.05 and Log₂ fold change > 1), red dots represent the AltProts more abundant in HeLa and HEK293T cells (other cell lines) (p-value < 0.05 and Log₂ fold change < 1), and gray dots represent AltProts not differentially expressed between pancreatic cancer cell lines and other cell lines. (B–D) Chart displaying the protein intensity measured for the AltProts II_629105 (A), IP_720068 (B), and IP_2374496 (C) across all the replicates of all the cell lines used in this study.

Figure 5

Several RNAs coding for alternative proteins are differentially expressed between tumors and normal tissues and their level correlates with PDAC patient’s survival. (A–C) Survival analysis of patients expressing high and low levels of RNAs encoding the IP_653613 (A), IP_747506 (B), and IP_150823 (C) AltProts. (D–F) Expression levels of the RNAs encoding the IP_653613 (A), IP_747506 (B), and IP_150823 (C) AltProts in PDAC (PAAD) and corresponding normal pancreatic tissues (based on TCGA and Gtex data). The red * is displayed for p-values < 0.01.

Figure 6

Comparison of the proteome from HeLa and HEK293T versus pancreatic cancer cell lines. (A) Volcano plot representing the log₂ ratio (Pancreatic cancer cell lines/Other cell lines) for each protein quantified and the corresponding −Log₁₀ p-value. The blue, red, and gray dots represent the proteins more abundant in the pancreatic cancer cell lines (p-value < 0.01 and Log₂ fold change > 1), more abundant in the other cell lines (HeLa and HEK293T) (p-value < 0.01 and Log₂ fold change < −1), or not differentially expressed, respectively. (B,C) Graphs displaying the pathways identified in a genome ontology (GO) analysis from the proteins more abundant in HeLa and HEK293T (B) or more abundant in pancreatic cancer cell lines (C). The number of genes (Log₂ transformed gene count) and the false discovery rates (FDR, −log₁₀ transformed) are displayed in the figures. Higher values indicate a higher number of genes identified from a pathway and a better FDR value.

Figure 7

Several canonical proteins are more expressed in cancer pancreatic cell lines and their corresponding mRNAs are differentially expressed between tumors and normal tissues and their level correlates with PDAC patient’s survival. (A,B) Chart displaying the protein intensity measured for the NT5E (A) and RHPN2 (B) proteins. (C,D) Expression levels of the mRNAs encoding the NT5E (C) and RHPN2 (D) proteins in PDAC (PAAD) and corresponding normal pancreatic tissues (based on TCGA and Gtex data). The red * is displayed for p-values < 0.01. (E,F) Survival analysis of PDAC patients expressing high and low levels of mRNAs encoding the NT5E (E) and RHPN2 (F) proteins.