What this is
- Pancreatic ductal adenocarcinoma (PDAC) is a highly lethal cancer with poor prognosis.
- This research identifies 263 () linked to PDAC progression.
- MMP14 and COL12A1 are proposed as potential based on their association with patient survival.
Essence
- The study identifies MMP14 and COL12A1 as significant in pancreatic ductal adenocarcinoma (PDAC), correlating their upregulation with poor patient survival.
Key takeaways
- 263 were identified from three microarray datasets, with 167 upregulated and 96 downregulated in PDAC tissues compared to non-tumor tissues.
- Survival analysis indicated that high expression levels of MMP14 and COL12A1 correlated with poor prognosis in PDAC patients (p < 0.05).
- The study's pathway analysis revealed that the hub genes are primarily involved in cell adhesion and extracellular matrix interactions, critical for PDAC progression.
Caveats
- The findings rely on bioinformatics analysis and require further validation in clinical settings to confirm the prognostic value of MMP14 and COL12A1.
- The study does not establish direct causation between gene expression and PDAC progression; further research is needed to clarify these relationships.
Definitions
- Differentially Expressed Genes (DEGs): Genes that show statistically significant differences in expression levels between two or more conditions, such as cancerous vs. non-cancerous tissues.
- Prognostic Biomarkers: Biological markers that can predict the likely outcome or progression of a disease, aiding in patient management and treatment decisions.
AI simplified
Introduction
Pancreatic ductal adenocarcinoma (PDAC) is the most common malignant tumor of the pancreas and is a lethal malignancy with poor prognosis, which is in part due to its rapid progression and the lack of diagnostic and therapeutic targets. In 2018, pancreatic cancer (PC) ranked 11th among the most common cancers, with 458,918 new cases and 432,242 deaths due to PC worldwide (). Recent work suggests that alcohol is a risk factor for PC (), while both genetic and environmental factors also play a role in the development and progression of PC (). [Bray et al., 2018] [Go, Gukovskaya & Pandol, 2005] [Piepoli et al., 2006]
Understanding genetic alterations in the context of biological pathways can help identify specific novel biomarkers of PDAC. Previous studies identified several cancer-associated genes implicated in PDAC, including(),(), and(). It is widely accepted that the formation of stroma contributes to tumor proliferation, invasion, and metastasis (). Particularly pathognomonic for PDAC is a stromal reaction that occurs during tumor progression and extensively involves fibroblasts and the extracellular matrix (ECM) (). Nevertheless, the precise etiology and pathogenetic mechanism of PDAC remain unclear. KRAS MYC CDKN2A [Waters & Der, 2018] [Witkiewicz et al., 2015] [Sikdar et al., 2018] [Von Ahrens et al., 2017] [Mahadevan & Von Hoff, 2007]
Microarray technology provides high-throughput methods for quantitatively measuring the expression levels of thousands of genes simultaneously, and microarray-based gene expression profiling can filter differentially expressed genes (DEGs) and biological pathways linked to various malignant tumors. Therefore, microarray techniques are promising and efficient ways to identify candidate biomarkers involved in the pathogenesis of PDAC. The purpose of our study was to determine significant DEGs and pathways implicated in PDAC by integrated bioinformatics analysis and to provide novel insights into the progression, diagnosis, and therapeutic targets of PDAC.
Materials & Methods
Screening database
The Gene Expression Omnibus (GEO:) is a public repository of high-throughput gene expression genomics datasets (). In this study, we downloaded three microarray datasets, namely,,, and, from the NCBI-GEO database. The array data inconsist of 45 matching pairs corresponding to PDAC and adjacent non-tumor tissues (;).includes data for 118 whole-tumor tissue and 13 control samples ().incorporates data for 8 normal pancreatic and 25 PDAC tissues (). Altogether, data for 188 PDAC tissues and 66 non-tumor tissues were available. GSE28735 GSE62165 GSE91035 GSE28735 GSE62165 GSE91035 https://www.ncbi.nlm.nih.gov/geo/↗ [Clough & Barrett, 2016] [Zhang et al., 2013] [Zhang et al., 2012] [Janky et al., 2016] [Sutaria et al., 2017]
Screening of DEGs
GEO2R () is an online analysis tool that is based on the R programming language and can be used to identify DEGs that differentiate between cancer and normal samples in a GEO series (). Using GEO2R, we analyzed DEGs that differentiate between PDAC and non-tumor tissue samples. An adjusted-value of <0.05 and |logFC| > 1 were employed as the cutoff criteria representing a significant difference. Using a data processing standard, we filtered DEGs via the Venn diagram tool at. A total of 263 DEGs were selected, which consisted of 167 upregulated genes and 96 downregulated genes. https://www.ncbi.nlm.nih.gov/geo/geo2r/↗ http://bioinformatics.psb.ugent.be/webtools/Venn/↗ [Yao & Liu, 2018] p
Establishment of the protein–protein interaction (PPI) network
The Search Tool for the Retrieval of Interacting Genes (STRING:) is an online application that can be used to assess DEG-encoded proteins and protein–protein interaction (PPI) networks (). A combined score of >0.4 was set as the threshold. http://string-db.org/↗ [Szklarczyk et al., 2015]
Cytoscape software v3.2.1 () was utilized to visualize the PPI network, which established a new way to find potential key candidate genes and core proteins. We utilized cluster analysis via the Molecular Complex Detection (MCODE) plugin with degree cutoff = 2, node score cutoff = 0.2, k-core = 2, and max depth = 100, which detected significant modules in the PPI network. To identify the hub genes, we also utilized the CytoHubba plugin, which provided a novel method of exploring significant nodes in PPI networks. These tools yield new insights into normal cellular processes, the underlying mechanisms of disease pathology, and clinical treatment. [Shannon et al., 2003]
Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis of DEGs
The Gene Ontology (GO) is used to perform enrichment analysis, which covers the cellular component (CC), biological process (BP), and molecular function (MF), of the selected genes (). The Kyoto Encyclopedia of Genes and Genomes (KEGG) is a database that helps to illustrate the functionalities and pathways of the selected genes (). The Database for Annotation, Visualization, and Integrated Discovery (DAVID:) is a public online bioinformatics database () that contains information on biological functional annotations for genes and proteins. The cutoff criteria were selected on the basis of< 0.05. We performed enrichment of the GO terms and KEGG pathways for the candidate DEGs using DAVID. [Young et al., 2010] [Altermann & Klaenhammer, 2005] [Dennis Jr et al., 2003] http://david.ncifcrf.gov/↗ p
Survival analysis of the candidate genes and validation of DEGs using TCGA and GTEx databases
Based on data for 9,736 tumors and 8,587 normal samples from The Cancer Genome Atlas (TCGA) database and the Genotype–Tissue Expression (GTEx) database, the Gene Expression Profiling Interactive Analysis tool (GEPIA:) is used to perform functions such as survival analysis, the detection of similar genes, and correlation analysis to clarify the relationships between diseases and DEGs (). http://gepia.cancer-pku.cn/↗ [Tang et al., 2017]
The GEPIA was also utilized for validating and visualizing the selected DEGs using TCGA and GTEx databases (). [Tang et al., 2017]
Validation of expression of candidate gene-encoded proteins
The expression of proteins encoded by the PDAC candidate genes was validated using the Human Protein Atlas (HPA:) website on the basis of spatial proteomics data and quantitative transcriptomics data (RNA-Seq) obtained from immunohistochemical analysis of tissue microarrays. https://www.proteinatlas.org/↗
Results
Identification of DEGs
A total of 263 DEGs were identified from,, and. There were 167 upregulated genes and 96 downregulated genes in PDAC tissues in comparison with non-tumor tissues () (). GSE28735 GSE62165 GSE91035 Fig. 1 Table 1
Venn diagram. Identification of differentially expressed genes (DEGs) from,, and. The different colored areas represent the different datasets, and a total of 263 DEGs were common to all three datasets. GSE28735 GSE62165 GSE91035
| DEGs | Gene names |
|---|---|
| Upregulated | ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,XDHRTKN2PTPRRADAM12STYK1TPX2PADI1HEPHCEACAM6ITGA3COL1A1ANLNFNDC1PCDH7SLC6A6TRIM29PXDNEDNRALTBP1MFAP5PLA2R1FN1KRT17PGM2L1IFI27ASAP2LAMB3TNFAIP6HOXB5OAS1NTMCOL5A2OSBPL3TMPRSS4ANTXR1SDR16C5OLR1NT5ECTSKSULF2MXRA5APOL1CDH11AREGMALLS100A16BGNLAMA3COL8A1IGFBP5MMP12ADAMTS6SLC2A1CD109ECT2KIF23MMP11CDH3LMO7CCL18ATP2C2POSTNMMP14ADAM28SRPX2CEACAM5TMC5OAS2MUC17GABRPCOMPSYTL2GPX8RUNX2DLGAP5KRT19VCANMKI67SULF1LAMC2GCNT3NMUMUC13CEACAM1ETV1COL12A1AGR2ST6GALNAC1SLC44A4PLAUS100PSERPINB5FOXQ1TGM2ITGB4DCBLD2TRIM31RAI14NRP2SGIP1CST1ARNTL2LEF1MYOFANO1S100A14DDX60KYNUCAPGCCL20MATN3NPR3GPRC5ANOX4IL1RAPACSL5HPGDGREM1SCELFBN1IGFL2SLC6A14KRT6ADHRS9ANGPT2MST1RCOL3A1TMEM45BEDIL3ASPMFAPINPP4BLOXL2NQO1CYP2C18IFI44LHK2EFNB2AEBP1SLC16A3CORINTHBS2BCAS1DSG3DKK1RHBDL2COL17A1TSPAN1FERMT1CXCL5COL6A3COL10A1ACTA2PLAC8AHNAK2MLPHFBXO32TGFBIKCNN4CLDN18FGD6MTMR11FXYD3MBOAT2SEMA3CDPYSL3CENPF |
| Downregulated | ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,EPB41L4BGSTA2KIAA1324CELA3AACADLCELSLC39A5LONRF2SLC3A1NRG4MT1GPROX1G6PC2C5EGFFAM3BAQP8CLPSSLC17A4CPB1GP2PDK4RBPJLPDIA2PM20D1CTRCIAPPPLA2G1BERP27CELA2BGRPRREG1AKIF1AGUCA1CCTRLSYCNCHRM3TMED6ALBKCNJ16REG3ASLC4A4AOX1SERPINA5CELA2ASPINK1FAM129AFAM150BSLC16A12F11CPA2SV2BBNIP3C2CD4BSLC1A2REG1BSCGNPAK3PRSS3GRB14REG3GDCDC2F8GPHA2EPHX2PNLIPRP2SLC7A2CPA1PRKAR2BONECUT1BACE1NUCB2HOMER2CXCL12SLC43A1GNMTNR5A2ALDH1A1IL22RA1BEX1ANPEPCFTRFLRT2LMO3FGL1NRCAMFABP4PNLIPRP1KLK1SERPINI2GATMDPP10C6SLC16A10PRSS1PAH |
Establishment of the PPI network
Using the STRING application and Cytoscape software, 225 nodes and 803 edges were mapped in the PPI network (). In association with these nodes, the whole PPI network was analyzed using the MCODE plugin, and one significant module was identified with average MCODE score = 8.6, nodes = 11, and edges = 43 (). This significant module comprised 11 DEGs, namely,,,,,,,,,,, and. From the PPI network, the top 20 hub genes were filtered by the CytoHubba plugin using the maximal clique centrality method. Their order of sequence was as follows:,,,,,,,,,,,,,,,,,,, and(). Via data mining, we found that the significant module and hub genes mainly consisted of upregulated genes. Fig. 2A Fig. 2B Fig. 2C COL6A3 COL3A1 VCAN COL5A2 COL12A1 THBS2 FBN1 POSTN LTBP1 MMP14 CDH11 FN1 COL1A1 COL3A1 BGN POSTN FBN1 COL5A2 COL12A1 THBS2 COL6A3 VCAN CDH11 MMP14 LTBP1 IGFBP5 ALB CXCL12 FAP MATN3 COL8A1
Protein–protein interaction (PPI) network of DEGs. (A) PPI network of 263 DEGs in PDAC tissues. Red nodes represent upregulated genes, whereas blue nodes represent downregulated genes. (B) Significant module identified from PPI network via the Molecular Complex Detection plugin. This module consisted of upregulated genes. (C) Top 20 hub genes filtered using CytoHubba plugin. Nodes shown in darker colors were found to have higher significance. Red represents the highest significance, followed by orange, whereas yellow represents the lowest significance.
GO and KEGG pathway analysis of DEGs
Functional and pathway enrichment analyses were accomplished using DAVID. GO analysis showed that the most significant module was mainly enriched in cell adhesion, extracellular matrix structural constituent, and proteinaceous extracellular matrix () (). Moreover, the 20 hub genes were mainly enriched in cell adhesion, endodermal cell differentiation, proteinaceous extracellular matrix, and calcium ion binding () (). In addition, KEGG pathway enrichment analysis demonstrated that the DEGs in the most significant module were enriched in ECM–receptor interaction () () and the hub genes were mainly enriched in ECM–receptor interaction, focal adhesion, protein digestion and absorption, and PI3K-Akt signaling pathway () (). (If< 0.0001, the corresponding term was considered to be enriched.) Fig. 3 Table 2 Fig. 4 Table 3 Fig. 3 Table 2 Fig. 4 Table 3 p
Results of Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses of the most significant module. The blue color represents biological process (BP), the gray color represents molecular function (MF), the green color represents cellular component (CC), and the orange color represents KEGG pathways.
Results of GO and KEGG pathway analyses of 20 hub genes. The blue color represents BP, the green color represents CC, the yellow color represents MF, and the orange color represents KEGG pathways.
| Pathway ID | Pathway description | Count in gene set | -valuep | FDR | DEGs |
|---|---|---|---|---|---|
| GO:0007155 | Cell adhesion | 5 | 1.37E−06 | 0.001121514 | COL6A3, COL12A1, VCAN, POSTN, THBS2 |
| GO:0001501 | Skeletal system development | 3 | 3.80E−04 | 0.310820683 | FBN1, VCAN, COL5A2 |
| GO:0030199 | Collagen fibril organization | 2 | 0.017003019 | 13.1195401 | COL5A2, CDH11 |
| GO:0005201 | Extracellular matrix structural constituent | 4 | 2.39E−06 | 0.001487027 | COL3A1, FBN1, VCAN, COL5A2 |
| GO:0005509 | Calcium ion binding | 4 | 0.003312243 | 2.045566581 | FBN1, VCAN, THBS2, CDH11 |
| GO:0005578 | Proteinaceous extracellular matrix | 6 | 1.81E−08 | 1.29E−05 | COL3A1, COL6A3, FBN1, COL12A1, VCAN, POSTN |
| GO:0005581 | Collagen trimer | 3 | 3.07E−04 | 0.217600344 | COL3A1, COL6A3, COL12A1 |
| GO:0031012 | Extracellular matrix | 3 | 0.001912518 | 1.350246354 | FBN1, COL12A1, THBS2 |
| GO:0005615 | Extracellular space | 4 | 0.008881226 | 6.138544445 | COL6A3, FBN1, COL12A1, POSTN |
| GO:0005604 | Basement membrane | 2 | 0.032936583 | 21.16653102 | FBN1, THBS2 |
| gga04512 | ECM–receptor interaction | 4 | 4.51E−05 | 0.019602271 | COL3A1, COL6A3, THBS2, COL5A2 |
| gga04510 | Focal adhesion | 4 | 7.12E−04 | 0.309146908 | COL3A1, COL6A3, THBS2, COL5A2 |
| Pathway ID | Pathway description | Count in gene set | -valuep | FDR | DEGs |
|---|---|---|---|---|---|
| GO:0007155 | Cell adhesion | 6 | 3.07E−07 | 3.44E−04 | COL12A1, POSTN, VCAN, COL8A1, THBS2, FN1 |
| GO:0035987 | Endodermal cell differentiation | 4 | 6.17E−06 | 0.006927836 | COL12A1, MMP14, COL8A1, FN1 |
| GO:0018149 | Peptide cross-linking | 2 | 0.02230059 | 22.35847168 | COL3A1, FN1 |
| GO:0043588 | Skin development | 2 | 0.035457234 | 33.30840338 | COL3A1, COL5A2 |
| GO:0030199 | Collagen fibril organization | 2 | 0.035457234 | 33.30840338 | COL3A1, COL5A2 |
| GO:0042060 | Wound healing | 2 | 0.039805932 | 36.60570116 | COL3A1, FN1 |
| GO:0001501 | Skeletal system development | 2 | 0.048448486 | 42.72191792 | FBN1, COL5A2 |
| GO:0005578 | Proteinaceous extracellular matrix | 7 | 1.25E−08 | 1.13E−05 | MATN3, BGN, FBN1, COL12A1, POSTN, COL1A1, FN1 |
| GO:0031012 | Extracellular matrix | 4 | 1.37E−04 | 0.123843274 | COL12A1, VCAN, MMP14, COL8A1 |
| GO:0005581 | Collagen trimer | 3 | 7.87E−04 | 0.70897958 | COL12A1, COL1A1, COL8A1 |
| GO:0005615 | Extracellular space | 6 | 0.001248571 | 1.122732142 | ALB, FAP, COL3A1, FBN1, COL12A1, VCAN |
| GO:0005604 | Basement membrane | 3 | 0.001310663 | 1.178271997 | FBN1, COL8A1, THBS2 |
| GO:0001527 | Microfibril | 2 | 0.008113224 | 7.097635089 | LTBP1, FBN1 |
| GO:0070062 | Extracellular exosome | 7 | 0.033863185 | 26.75315613 | BGN, ALB, FBN1, COL12A1, COL8A1, FN1, CDH11 |
| GO:0005509 | Calcium ion binding | 7 | 9.57E−06 | 0.007173793 | MMP14, MATN3, LTBP1, FBN1, VCAN, THBS2, CDH11 |
| GO:0005201 | Extracellular matrix structural constituent | 3 | 4.84E−04 | 0.362368483 | COL3A1, FBN1, COL5A2 |
| GO:0008201 | Heparin binding | 3 | 0.003684116 | 2.728227814 | POSTN, THBS2, FN1 |
| ocu04512 | ECM–receptor interaction | 6 | 5.00E−08 | 4.15E−05 | COL3A1, COL6A3, COL1A1, COL5A2, THBS2, FN1 |
| ocu04510 | Focal adhesion | 6 | 3.64E−06 | 0.003011517 | COL3A1, COL6A3, COL1A1, COL5A2, THBS2, FN1 |
| ocu04974 | Protein digestion and absorption | 5 | 8.69E−06 | 0.007197854 | COL3A1, COL6A3, COL12A1, COL1A1, COL5A2 |
| ocu04151 | PI3K-Akt signaling pathway | 6 | 3.26E−05 | 0.026969277 | COL3A1, COL6A3, COL1A1, COL5A2, THBS2, FN1 |
| ocu05146 | Amoebiasis | 4 | 3.54E−04 | 0.292970736 | COL3A1, COL1A1, COL5A2, FN1 |
| ocu04611 | Platelet activation | 3 | 0.012199259 | 9.669074187 | COL3A1, COL1A1, COL5A2 |
Overall survival analysis of the top 20 hub genes
Patient survival analysis performed via the GEPIA using TCGA and GTEx databases demonstrated that the high expression levels ofandwere correlated with an unfavorable prognosis in PDAC patients (< 0.05) (). The overall survival analysis showed that the other hub genes had no statistically significant correlations (> 0.05). COL12A1 MMP14 p p Fig. 5
Overall survival analysis. Overall survival curves for (A)and (B)expression in PDAC patients in comparison with a high-risk group and a low-risk group. A value of< 0.05 was regarded as statistically significant. TPM, transcripts per million; HR, hazards ratio. COL12A1 MMP14 p
Validation of DEGs using TCGA and GTEx databases
To ensure the reliability of the identification of the top 20 hub genes, we validated these via the GEPIA using TCGA and GTEx databases. Boxplots of the hub genes associated with PDAC were downloaded from the GEPIA. The results demonstrated that,,,,,,,,,,,,,,,,, andwere significantly overexpressed in PDAC tissues in comparison with normal pancreatic tissues, whereaswas underexpressed in PDAC tissues (< 0.05) ().was expressed in PDAC tissues, but with no statistically significant difference in expression (> 0.05). FN1 COL1A1 COL3A1 BGN POSTN FBN1 COL5A2 COL12A1 THBS2 COL6A3 VCAN CDH11 MMP14 LTBP1 IGFBP5 FAP MATN3 COL8A1 ALB p CXCL12 p Fig. 6
Validation of DEGs using The Cancer Genome Atlas and Genotype–Tissue Expression databases. The boxplots were downloaded from the Gene Expression Profiling Interactive Analysis tool and are arranged in the following order: (A), (B), (C), (D), (E), (F), (G), (H), (I), (J), (K), (L), (M), (N), (O), (P), (Q), (R), and (S). A value of< 0.05 was regarded as statistically significant. The-axes represent the expression in terms of log(TPM + 1). The red boxes represent the expression levels of DEGs in PAAD tissues, whereas the gray boxes represent the expression levels of DEGs in normal tissues. PAAD, pancreatic adenocarcinoma. FN1 COL1A1 COL3A1 BGN POSTN FBN1 COL5A2 COL12A1 THBS2 COL6A3 VCAN CDH11 MMP14 LTBP1 IGFBP5 FAP MATN3 COL8A1 ALB p Y 2
Validation of expression of candidate gene-encoded proteins
We obtained the expression levels of proteins encoded by the 20 hub genes associated with PDAC from the HPA website. No data for proteins encoded by,, andare reported on the HPA website, and expression profiles of the other 17 genes in PDAC clinical specimens are shown in. COL5A2 IGFBP5 MATN3 Fig. 7
The protein expressions of,,,,,,,,, andwere upregulated in PDAC tissues in comparison with normal tissues, with onlybeing downregulated in PDAC tissues.,,, andwere not expressed in either PDAC tissues or normal tissues, andandwere overexpressed in both cancer and normal tissues. FN1 MMP14 COL12A1 COL3A1 COL1A1 POSTN VCAN LTBP1 FBN1 FAP ALB COL6A3 COL8A1 CDH11 CXCL12 BGN THBS2
Expression of 20 candidate DEGs in human pancreatic cancer specimens. The immunohistochemical data were obtained from the Human Protein Atlas. Except for,, and, expression profiles of the other 17 genes in PDAC clinical specimens are shown. Staining demonstrated that the protein expressions of (A), (B), (C), (D), (I), (L), (M), (N), (O), and (Q)were higher in PDAC tissues than in normal pancreatic tissues, with only (F)being downregulated in PDAC tissues. (E), (H), (J), and (K)were not expressed, whereas (G)and (P)were overexpressed in both PDAC tissues and normal tissues. COL5A2 IGFBP5 MATN3 FN1 MMP14 COL12A1 COL3A1 COL1A1 FAP FBN1 LTBP1 POSTN VCAN ALB COL6A3 CDH11 COL8A1 CXCL12 BGN THBS2
Discussion
Our study was based on GEO datasets, namely,,, and. The main findings deduced from the studies used to compilewere that dipeptidase 1 and a unique set of free fatty acids played roles in the development, progression, and prognosis of PC and might be potential targets in PDAC (;). The study that was used to compilefound that hepatocyte nuclear factor (HNF)-1and HNF-1seem to be good candidates as tumor suppressors in PDAC (). Another paper, which was used to compile, concluded that an increase in the expression of the processed transcript ofwas associated with a poor prognosis in PDAC (). GSE28735 GSE62165 GSE91035 GSE28735 GSE62165 GSE91035 [Zhang et al., 2012] [Zhang et al., 2013] [Janky et al., 2016] [Sutaria et al., 2017] α β HNRNPU
In our study, GO analysis showed that the most significantly enriched BP, CC, and MF terms among the 20 hub genes were cell adhesion, proteinaceous extracellular matrix, and calcium ion binding, respectively. Cell adhesion is the attachment of a cell either to another cell or to an underlying substrate. The proteinaceous extracellular matrix provides structural support and biochemical or biomechanical cues for cells or tissues and is a structure located external to one or more cells. The ECM is a crucial factor in both promoting the progression of PDAC and inhibiting the delivery of antitumor therapy (). [Weniger, Honselmann & Liss, 2018]
According to the analysis of the MF terms among the hub genes,,,,,,, andwere jointly involved in calcium ion binding, which is defined as selective and non-covalent interactions with calcium ions (Ca). Cais a ubiquitous and versatile second messenger involved in the regulation of numerous cellular functions, including gene transcription, vesicular trafficking, and cytoskeletal rearrangements (). Caand Ca-regulating proteins contribute to a large number of processes that are key to cancer cells, including proliferation, invasion, and cell death (;). A high serum Calevel is associated with a poor prognosis in PDAC (), and cytosolic Caoverload triggers apoptotic death pathways (). It is thus reasonable that the seven abovementioned genes might regulate calcium ion binding and affect the development of PDAC. Furthermore, our study suggests thatis a promising biomarker for survival in PDAC. Considering that Cacannot be produced in cells but undergoes flux between intracellular calcium storage, cytosolic calcium signals, and the extracellular calcium pool (), it would be reasonable to hypothesize that the overexpression ofinfluences calcium ion storage and thus might cause disorders of calcium homeostasis and hence contribute to an unfavorable prognosis in PDAC patients. MMP14 THBS2 CDH11 FBN1 LTBP1 MATN3 VCAN MMP14 MMP14 2+ 2+ 2+ 2+ 2+ 2+ 2+ [Nunes-Hasler, Kaba & Demaurex, 2020] [Monteith, Prevarskaya & Roberts-Thomson, 2017] [Prevarskaya, Skryma & Shuba, 2011] [Dong et al., 2014] [Brini & Carafoli, 2009] [Yang et al., 2020]
Matrix metalloproteinases (MMPs) are a family of calcium- and zinc-dependent membrane-anchored or secreted endopeptidases that are overexpressed in various diseases, including breast cancer (). MMP14 is located in neoplastic epithelium. It is speculated that the overexpression ofalone may be sufficient to induce the development of PDAC (). Moreover,is overexpressed in the epithelium in invasive PC (;), and MMP14, as an endopeptidase, can degrade various components of the ECM such as collagen, which possibly leads to metastasis of tumors (). Type I collagen can induce the expression ofandin pancreatic ductal epithelial cells (), andencodes the major component of type I collagen. The expression ofin PDAC cells stimulates pancreatic stellate cells (PSCs) and enhances the production of type I collagen by increasing transforming growth factor-signaling (). Ottaviano et al. found that fibrosis and the expression ofin tumor specimens increased in comparison with those in normal pancreatic tissue (). These findings suggest the key role of interactions betweenand type I collagen in the progression of PDAC and supportas a potential target for inhibiting fibrosis, preventing metastasis, and treating PDAC. [Min et al., 2015] [Shields et al., 2012] [Iacobuzio-Donahue et al., 2002] [Shields et al., 2012] [Golubkov et al., 2010] [Ottaviano et al., 2006] [Krantz et al., 2011] [Ottaviano et al., 2006] MMP14 MMP14 MMP14 TGF- β1 COL1A1 MMP14 β MMP14 MMP14 MMP14
The KEGG pathway analysis showed that six hub genes, namely,,,,,, and, were significantly associated with ECM–receptor interactions, focal adhesion, and the phosphatidylinositol-3-kinase–protein kinase B (PI3K-Akt) signaling pathway. In addition, collagen-encoding genes, including,, and, were also enriched in protein digestion and absorption and platelet activation. COL1A1 COL3A1 COL5A2 COL6A3 FN1 THBS2 COL1A1 COL3A1 COL5A2
ECM–receptor interactions play important roles in the processes of cell shedding, adhesion, degradation, migration, differentiation, hyperplasia, and apoptosis (). PSCs secrete several ECM proteins, including collagen, fibronectin, fibulin-2, and laminin, as well as hyaluronan (). Moreover,andwere significantly downregulated in PC (<0.0001) after treatment with gemcitabine in combination with EC359 (). The geneencodes the pro-alpha 1 chain of type I collagen, which is closely associated with.was found to encode a major structural component of hollow organs such as large blood vessels, the uterus and bowel, and tissues that must withstand stretching (). As an important molecule,is associated with remodeling of the ECM and is differentially expressed between in situ ductal carcinoma and invasive ductal carcinoma (). The alpha 3 chain of type VI collagen is mainly present in the desmoplastic stroma in PDAC, with large deposits between the sites of stromal fatty infiltration and around the malignant ducts (), and the circulating form of this protein has potential clinical significance in the diagnosis of pancreatic malignancy ().encodes a collagen-associated protein that has been identified as a potential biomarker of an unfavorable prognosis in PDAC ().appears in the early stages of PDAC and hence has great potential for the diagnosis of PDAC, with 98% specificity (). [Bao et al., 2019] [Hall et al., 2019] [Hall et al., 2019] [Kuivaniemi & Tromp, 2019] [Vargas et al., 2012] [Arafat et al., 2011] [Kang et al., 2014] [Hu et al., 2018] [Kim et al., 2017] COL1A1 COL3A1 p COL1A1 MMP14 COL3A1 COL5A2 FN1 THBS2
At points of ECM–cell contact, specialized structures are formed, which are termed focal adhesions. Some components of focal adhesions contribute to cell migration in PDAC and participate in structural links between the actin cytoskeleton and membrane receptors, whereas others are signaling molecules (). [Manoli et al., 2019]
The PI3K-Akt signaling pathway regulates fundamental cellular functions, including transcription, translation, proliferation, growth, and survival. Accumulating evidence has implied that the PI3K-Akt signaling pathway promotes malignant processes of PDAC cells, including proliferation, angiogenesis, metastasis, suppression of apoptosis, and chemoresistance, and targeting the PI3K-Akt signaling pathway has been a potential therapeutic strategy for the treatment of PC (). [Ebrahimi et al., 2017]
In PC, both exocrine and endocrine functions are abnormal, which profoundly influences the secretion of proteases, and hence protein digestion and absorption is a prominent metabolic change (). Platelet activation facilitates the P-selectin- and integrin-dependent accumulation of cancer cell microparticles and promotes tumor growth and metastasis (). However, the effect of collagen-mediated platelet activation on the progression of PDAC needs further investigation. [Gilliland et al., 2017] [Mezouar et al., 2015]
Collagens are centrally involved in the formation of fibrillar and microfibrillar networks of the ECM and basement membranes, as well as other structures of the ECM (). We further found that the collagen family is closely associated with PDAC. Interestingly, Wang and Li also found that the collagen family andhave an influence on PC via data mining using a different gene set () (). As we have done, they suggested that, together with,, and, may be key molecules in the development and progression of PDAC owing to their involvement in ECM–receptor interactions and focal adhesion pathways. These DEGs were also identified in our study. Furthermore, we found thatandare probably also key DEGs that influence PDAC, which differs from the results of Wang and Li. Although the specific relationship betweenand PDAC has not been reported, our findings also suggest thatis a potential prognostic biomarker in patients with PDAC. GSE15471 [Gelse, Poschl & Aigner, 2003] [Wang & Li, 2015] FN1 FN1 COL1A1 COL3A1 COL5A2 COL12A1 COL6A3 COL12A1 COL12A1
We also found thatandappear to be involved in the progression of PDAC.encodes a structural component of the microfibrils of the ECM that have diameters of 10–12 nm, which impart both regulatory and structural properties to load-bearing connective tissues (). The silencing ofinhibits the proliferative, migratory, and invasive activities of gastric cancer cells, whereas upregulation of the expression ofhas the opposite effect ().encodes a macromolecular component of the subendothelium (). It is suggested thatmay be associated with malignant processes of hepatocarcinoma () and the progression and prognosis of human colon adenocarcinoma (). FBN1 COL8A1 FBN1 FBN1 FBN1 COL8A1 COL8A1 [Lee et al., 2004] [Yang, Zhao & Chen, 2017] [Xu et al., 2001] [Zhao et al., 2009] [Shang et al., 2018]
Conclusions
In conclusion, we screened the top 20 hub genes (,,,,,,,,,,,,,,,,,,, and) and the related enriched functions or pathways, which regulate the progression and metastatic invasion of PDAC, as well as overall survival. The results demonstrate that the upregulation ofandin PDAC is closely associated with poor overall survival, that these might be a potential combination of prognostic biomarkers in patients with PDAC, and thatandmight be biomarkers of PDAC. In brief, our study increases the understanding of the potential critical genes and related pathways that participate in the pathogenesis of PDAC. FN1 COL1A1 COL3A1 BGN POSTN FBN1 COL5A2 COL12A1 THBS2 COL6A3 VCAN CDH11 MMP14 LTBP1 IGFBP5 ALB CXCL12 FAP MATN3 COL8A1 MMP14 COL12A1 FBN1 COL8A1