S100A10, a novel biomarker in pancreatic ductal adenocarcinoma

Pancreatic cancer is arguably the deadliest cancer type. The efficacy of current therapies is often hindered by the inability to predict patient outcome. As such, the development of tools for early detection and risk prediction is key for improving outcome and quality of life. Here, we introduce the plasminogen receptor S100A10 as a novel predictive biomarker and a driver of pancreatic tumor growth and invasion. We demonstrated that S100A10 mRNA and protein are overexpressed in human pancreatic tumors compared to normal ducts and nonductal stroma. S100A10 mRNA and methylation status were predictive of overall survival and recurrence‐free survival across multiple patient cohorts. S100A10 expression was driven by promoter methylation and the oncogene KRAS. S100A10 knockdown reduced surface plasminogen activation, invasiveness, and in vivo growth of pancreatic cancer cell lines. These findings delineate the clinical and functional contribution of S100A10 as a biomarker in pancreatic cancer.

H-score = 0 x percentage contribution of negative pixels 1 x percentage contribution of low positive pixels 2 x percentage contribution of positive pixels 3 x percentage contribution of highly positive pixels Supplemental Table 1. Calculation scheme of the H-score. The score represents both the intensity and number of DAB-positive pixels in stained sample.

Median cut-off
Optimal cut-off Ternary cut-off Supplemental Figure 4. The three cut-offs of S100A10 mRNA. S100A10 REVs of the TCGA PDAC cohort follow a relatively normal Gaussian distribution. The three cut-off system is based on the median expression value (A), optimal expression value (D), or a ternary expression classifier (CG). The median cutoff is based on the median REV. In the case of even patient number, the median REV was considered high or low based on closeness of expression. Optimal cut-offs were determined using the cut off finder database based on the lowest p-value and highest hazard ratio possible. Source: (http://molpath.charite.de/cutoff/) Budczies et al. (2012), PLoS ONE 7(12): e51862. The Ternary cut-off was derived from bin frequency of REVs. This identified two cut-off REVs resulting in three expression groups. Kaplan Meier analyses of overall (B, E, H) and recurrence-free survival (C, F, I) based on the median cut-off (B, C), the optimal cutoff (E, F) and the ternary cut-off (H, I).  0  2000  4000  6000  8000  10000  12000  14000  16000  18000  20000  22000  24000  26000  28000  30000  32000  34000  36000  38000  40000 Patient frequency S100A10 mRNA, RNA-Seq V2 (RSEM)   Table 2. Higher S100A10 mRNA, higher copy number and low-methylation scores correlate with lower short-term survival. The percentages represent the percentages of patients alive or recurrence-free at one, three and five year time point during this study. These percentages also represent the likelihoods of being alive or recurrence-free after one, three and five years if the TCGA cohort is similar to the entire population PDAC cohort.  Figure 5. Correlation of S100A10 mRNA expression, linear copy number and copy number status with overall and recurrence-free survival. Pearson correlation analysis of S100A10 mRNA (expression values normalized to average) with (A) relative linear copy number and (B) copy number status. Kaplan Meier analysis of overall survival of TCGA PDAC patients in relation to S100A10 copy number score based on an optimal cut-off of (C) OS and (D) RFS. Kaplan Meier analysis of (E) OS and (F) RFS based on copy number status of S100A10. Gain and amplification are based on the Cbioportal definition where gain represents a low-level increase in copy number while amplification represents a highlevel of increase.
Supplemental  Multiple comparisons of S100A10 methylation survival functions were performed on the TCGA and ICGC patient cohorts. P-values were adjusted to the Bonferroni-corrected threshold. Adjusted p-value is p-value/K = 0.017 where K=3 and represents the number of comparisons made.
Supplemental Figure 6. The β values of probes that were not differentially-methylated and/or did not negatively correlate with S100A10 mRNA expression. For normal vs. tumor comparisons, the raw data was extracted from MethHC (http://methhc.mbc.nctu.edu.tw/php/index.php), described by Huang et al.  The 377-nucleotide promoter region of S100A10 used for pyrosequencing. The sequence highlights the sequenced CpG sites as well the location of HM450 methylation probes (as highlighted). The beginning of exon1 is underlined. (C) Promoter CpG island analysis using EMBOSS CpGplot tool from the EMBL-EBI database: (https://www.ebi.ac.uk/Tools/seqstats/emboss_cpgplot/). The CpG island criteria set by Takai and Jones (2002) were used. These include: 1) minimum length of an island is 500bp. 2) Minimum observed/expected is the minimum average observed to expected ratio of C plus G to CpG in a set of 10 windows that are required before a CpG island is reported. The threshold value is 0.65. 3) Minimum percentage is minimum average percentage of G plus C a set of 10 windows that are required before a CpG island is reported. The threshold value is 0.55. Supplemental Figure 13. RT-qPCR of several genes in scramble control and S100A10-shRNA 1 Panc-1 tumors. These genes were not significantly altered by S100A10 depletion in Panc-1 tumors.  Figure 14. Schematic representation of KRAS G12D -and methylation-mediated regulation of S100A10-dependent plasminogen activation. Oncogenic KRAS induces S100A10 upregulation which in turn contributes to plasminogen activation and plasminogen-dependent invasion. The expression of S100A10 was also driven by DNA methylation of its promoter region. A heterotetrameric complex is formed of two annexin A2 subunits and 2 subunits of S100A10 (dimer). KRAS is also capable of upregulating uPA and uPAR whose localization is induced by S100A10 binding to plasminogen. The latter is activated into plasmin which cleaves extracellular matrix (ECM) proteins and destabilizes its structure allowing pancreatic cancer cell advancement.

NORMAL CELL
Annexin A2 Growth factor/receptor complex KRAS S100A10 Hypermethylation S100A10 ECM KRAS G12D Hypomethylation ECM Plasminogen uPAR Plasmin uPA S100A10 Invasion Supplemental Table 11. List of primer sequences used in RT-qPCR and pyrosequencing as well as dsDNA oligo used for S100A10 shRNA. b represents biotinylated primers. "Seq" is used for the pyrosequencing step along with the biotinylated reverse primer.