Proteogenomic analysis reveals Arp 2/3 complex as a common molecular mechanism in high risk pancreatic cysts and pancreatic cancer

doi:10.21203/rs.3.rs-4612255/v1

Download PDF

Article

Proteogenomic analysis reveals Arp 2/3 complex as a common molecular mechanism in high risk pancreatic cysts and pancreatic cancer

https://doi.org/10.21203/rs.3.rs-4612255/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Pancreatic cysts, particularly intraductal papillary mucinous neoplasms (IPMNs), pose a potential risk for progressing to pancreatic cancer (PC). This study investigates the genetic architecture of benign pancreatic cysts and its potential connection to PC using genome-wide association studies (GWAS). The discovery GWAS identified significant genetic variants associated with benign cysts, specifically the rs142409042 variant near the OPCML gene. A pairwise GWAS comparing PC to benign cysts revealed the rs7190458 variant near the BCAR1 and CTRB1 genes. Further analysis with identified GWAS genes highlighted the Actin Related Protein (Arp) 2/3 complex as a potentially important molecular mechanism connecting benign cysts and PC. The Arp2/3 complex-associated genes were significantly upregulated in PC, suggesting their role in the malignant transformation of pancreatic cysts. Differential expression of these genes was observed across various cell types in PC, indicating their involvement in the tumor microenvironment. These findings suggest that the Arp2/3 complex-associated genes can serve as potential biomarkers for predicting the malignant transformation of pancreatic cysts, opening new avenues for targeted therapies and early detection strategies.

Health sciences/Medical research/Biomarkers/Predictive markers

Biological sciences/Cancer/Cancer genetics

Pancreatic cysts are fluid-filled mucus within the pancreas, categorized as benign or malignant. The majority of the cysts are benign, but a subset has the potential to progress to pancreatic cancer (PC) ¹. PC is associated with a highly lethal malignant transformation and a poor prognosis. In fact, only about 0.25% of pancreatic cysts identified incidentally are cancerous, and less than 0.25% of these benign cysts transform into cancer within a year of their discovery ².

The transformation of pancreatic cysts from intraductal papillary mucinous neoplasms (IPMNs) to PC is often characterized by a sequence of dysplastic stages. It starts from low-grade dysplasia (LGD), advances to high-grade dysplasia (HGD), invasive IPMN, and ultimately leads to PC ³. LGD represents an early, less severe form of pre-cancerous alteration, characterized by minimal cellular abnormalities. As the dysplasia progresses to HGD, the abnormalities become more pronounced, indicating a higher risk of transitioning to an invasive cancer. This progression highlights the critical need to monitor and understand the molecular and genetic architecture changes occurring within IPMNs. The presence of HGD is a strong predictor of the potential for malignancy and an indicator for surgical intervention ³.

Identifying and understanding the changes in biological mechanisms from LGD to HGD and further into invasive IPMN and PC is essential for early prediction and management of potential malignant transformations. Many studies were conducted to find biomarkers for high-risk IPMN, but to date, none have been incorporated into routine clinical application due to their limited prognostic efficiency ^4–7. Another significant challenge in identifying biomarkers is the intralesional heterogeneity within IPMNs and PC, which can be explored by analyzing the different cell types present in these lesions. Recent studies have begun to shed light that HGD IPMN may share molecular mechanisms with PC ⁸. This connection suggests that the molecular pathways involved in the malignant transformation of pancreatic cysts might also play a role in the development and progression of PC. Genetic alterations ⁹ and cellular abnormalities observed in HGD IPMN show similarity to those seen in PC ¹⁰.

Genome-wide association studies (GWAS) have revolutionized the understanding of complex traits by identifying genetic variants associated with various diseases. GWAS could provide valuable insights into the genetic basis of various conditions and elucidate the genetic architecture to help in understanding disease mechanisms. Unfortunately, the genetic architecture of pancreatic cysts remains poorly understood. Previous studies have largely focused on genetic alterations in PC rather than on the benign precursors, pancreatic cysts. Genetic variants identified by GWAS can be used to elucidate the underlying disease mechanisms using mapped genes, and construct polygenic risk scores (PRS) to reflect an individual's risk for a specific disease, further as an indicator of malignant transformation^11–19.

In our study, we aimed to define the genetic makeup of benign cysts and find potential biomarkers in IPMN that share similar regulatory directions in PC at cellular levels. For this, we performed genome-wide association studies, and systematically evaluated the concordance of the changes in protein/mRNA expression in PC at bulk and single-cell levels (Fig. 1). GWAS-genes and commonly up and down-regulated proteins/genes in PC revealed that Actin Related Protein (Arp) 2/3 complex-associated genes/proteins were significantly upregulated in PC and high-risk pancreatic cysts.

Genetic architecture of pancreatic cyst

To our knowledge the genetic architecture of pancreatic cysts has not yet been characterized. The main goals of this study were to determine 1) the genetic architecture of benign pancreatic cysts using GWAS data, and 2) if this was associated with PC.

First, a discovery GWAS was performed to identify genetic associations in individuals with benign pancreatic cysts compared to healthy controls (707 individuals with benign pancreatic cysts (cases) and 14,000 healthy controls of European origin from the UK Biobank (UKBB). Additionally, a pairwise GWAS was performed to determine the genetic associations in individuals with PC compared to those with benign pancreatic cysts (584 individuals diagnosed with PC (cases) versus 707 individuals with benign pancreatic cysts (controls), supplementary Table S1). Individuals diagnosed with malignant cysts were excluded from both the discovery and pairwise GWAS analyses.

Analyses of the discovery GWAS identified rs142409042 genetic variant near the opioid binding protein/cell adhesion molecule like (OPCML) gene on chromosome 11 (Fig. 2A) as having genome-wide significance association (p < 10^− 08) with benign pancreatic cysts versus healthy controls (odd ratio OR (95% confidence interval (CI) = 2.583 (1.835–3.636)). Similarly, rs7190458 genetic variant near the BCAR1 scaffold protein, Cas family member (BCAR1) and chymotrypsinogen B1 (CTRB1) on chromosome 16 (Fig. 2B) showed a genome-wide significant association with PC versus benign pancreatic cysts (OR (95%CI) = 0.152 (0.094–0.244)). Notably, there was no overlap of genome-wide significant SNPs p < 10^− 08 between the discovery GWAS (benign pancreatic cysts versus healthy controls) and pairwise GWAS (PC versus benign pancreatic cysts) (Fig. 2, supplementary Table S2, S3).

Polygenic risk score (PRS) for prediction of the risk of developing benign pancreatic cyst

We next aimed to develop a PRS to predict the risk of benign pancreatic cysts. We refer to this PRS as benign pancreatic cyst PRS. Genetic variants identified in the discovery GWAS (benign pancreatic cyst versus healthy controls) were used to calculate PRS for each individual across the four groups: healthy controls, benign pancreatic cysts, PC, PC with cyst. To construct the PRS, 32 genetic variants from the discovery GWAS (p < 10^− 05) were weighted by their effect sizes. PRS calculation was performed for each individual, and percentiles were established based on the PRS of the healthy control group. Each individual from the four groups was assigned a percentile based on her or his PRS score. The logistic regression model and ANOVA were performed to compare benign pancreatic cysts, PC with cyst and PC with healthy controls. PRS showed a significant association with the risk to develop benign pancreatic cysts versus healthy controls (OR (95%CI) = 2.159 (2.017–2.311), supplementary Table S4). 65% of individuals with benign pancreatic cysts were represented in the upper percentile (Fig. 2C). In the lower percentile, the PC with cyst and PC were represented (ANOVA p < 2 × 10^− 16). A similar distribution pattern within the percentiles was noticed when analyzing females and males separately in supplementary Figure S1 and Table S4. This is consistent with the benign pancreatic cyst PRS being in a different range from PC and PC with cyst. Thus, benign pancreatic cyst PRS constructed from discovery GWAS were associated with benign cysts but not with PC or PC with cyst (Fig. 2C).

However, since some pancreatic cysts become malignant, we next investigated whether the genes associated with benign cysts interacted with or overlapped with the genes or gene products associated with PC. For this purpose, we mapped the discovery and pairwise GWAS genetic variants to the corresponding genes. For the next section, discovery GWAS mapped genes are referred to as GWAS-genes. These discovery GWAS-genes were used to test for overlap or molecular interactions with differentially expressed genes/proteins (DEGs/DEPs, respectively) from bulk and single cell transcriptomics or proteomics data. Interactions between GWAS-genes and DEGs/DEPs were sought based on the protein-protein interaction network based on STRING database.

Arp2/3 complex as a common molecular mechanism between pancreatic cancer and pancreatic cyst

We next tested if there was any association between the GWAS-genes and mRNAs/proteins in PC. To define such mRNAs/proteins, we analyzed pancreatic tumor tissues and normal tissues from The Cancer Genome Atlas (TCGA) and Clinical Proteomic Tumor Analysis Consortium (CPTAC) to identify DEGs and DEPs, respectively (Tumor versus normal, adjusted p value (p-adjusted) < 0.05, absolute log fold change > 0.25). We identified 2,169 DEGs/DEPs that were consistently upregulated or downregulated in tumors across TCGA and CPTAC. Of them, six genes were the GWAS-genes, namely NTM, ARPC3, CDCP1, FEZ2, FXYD2, and MYO1D. All the corresponding mRNAs and proteins were upregulated in tumors, except FXYD2 (supplementary Figure S2). We further hypothesized that the protein products of these six GWAS-genes interacted with these similarly upregulated or downregulated DEGs/DEPs within the tissue at a local level, potentially driving progression from pancreatic cyst to PC.

To test this hypothesis, we performed a PPI network analysis for six GWAS-genes based on the STRING database. We utilized PPIs with a confidence score (> 0.95) to exclude spurious interactions random (a score of 1 represents the highest possible confidence in the STRING database). Additionally, we also considered co-expression of the proteins based on the STRING database to identify the co-expressed interaction partners. We found that only one GWAS-gene protein product, ARPC3, was highly connected in PPI network (confidence score > 0.95, Fig. 3A, supplementary Table S5). We further identified ARPC3 protein as primary neighbors, directly interacted with ARPC2, ARPC1B, ARPC4, ARPC5, and ACTR3 (confidence score > 0.95 based on PPI and co-expression) in PC. These DEPs were part of the (Arp 2/3 complex. This complex, in turn, interacted with the effector downstream proteins including ACTG1 and ACTB DEPs with the high possible confidence score (confidence score > 0.95) (Fig. 3B). The Arp2/3 complex genes mentioned above as well as their downstream genes, referred to Arp2/3 complex-associated genes, were linked with the actin-related pathways, including regulation of actin dynamics for phagocytic cup formation ²⁰.

Next, we hypothesized that Arp2/3 complex-associated genes have vital roles in the development of PC. So, we further investigated the expression patterns of their corresponding proteins in PC at both bulk level and the expression patterns of their corresponding mRNA in PC at cellular level.

Arp 2/3 complex-associated genes differed in gene expression and may be promising biomarkers in PC

To investigate the regulation of Arp2/3 complex-associated genes/proteins in PC, we analyzed the DEGs and DEPs of PC versus normal from TCGA and CPTAC. All eight genes and proteins associated with the Arp2/3 complex were upregulated significantly in PC patients of both TCGA and CPTAC datasets (absolute log fold change > 0.25 and p-adjusted < 0.05, Fig. 4A, supplementary Table S6). Upregulation of the Arp2/3 complex which is related to actin cytoskeleton and cellular movement indicated the importance of this complex in PC.

To examine if this pattern was generalizable, we analyzed protein data from 191 PC tumor and 90 paired tumor-adjacent tissues²¹. The analysis showed that all of the eight proteins were significantly upregulated in the PC patients (absolute log fold change > 0.25, p-adjusted < 0.05, Fig. 4B).

Following this, we extended our analysis to single-cell RNA sequencing (scRNA-seq) data to determine the presence of Arp2/3 complex-associated genes in specific cell types within the PC tumor microenvironment. We analyzed scRNA-seq data from 24 primary pancreatic tumors and 11 normal pancreases, subjecting 41,148 cells from tumor tissues and 15,444 cells from normal tissues to quality control and clustering. This process identified nine distinct cell types, each featuring 272 to 1,982 DEGs between tumor and normal tissues. Among these DEGs, genes ACTB, ARPC2, ARPC3, ARPC4, ARPC5, and ACTG1 of the Arp2/3 complex-associated genes were notably upregulated across multiple cell types, including acinar, fibroblast, endocrine, and epithelial cells (Fig. 4A).

Taken together, these analyses supported that the Arp2/3 complex-associated genes may play an important role in the progression from pancreatic cysts to PC at cellular levels.

Arp2/3 complex-associated genes as potential biomarkers for malignant transformation in pancreatic cysts

To systematically assess Arp2/3 complex-associated genes as potential biomarkers for malignant transformation from low-risk PC cysts to high-risk PC cysts, we explore whether the similar expression pattern of these genes was present in different cell types in HGD IPMN. We analyzed available scRNA-seq data from pancreatic cyst biopsies with HGD IPMN, LGD IPMN and primary pancreatic tumors¹⁰. After quality filtering, we analyzed 7,986 cells for cell typing and subsequent analyses. We identified two cell types: epithelial cell and macrophages (supplementary Figure S3). Due to low cell number/counts, we only focused on epithelial cells (n = 4,892 cells). Within this subset, we identified 327 DEGs for HGD IPMN (HGD versus LGD, absolute log fold change > 0.25, p-adjusted < 0.05) and 882 DEGs for PC (PC versus LGD) (supplementary Tables S7, S8). In epithelial cells, six genes associated with Arp2/3 were identified as DEGs in both PC and HGD IPMN when compared to LGD IPMN (Fig. 4C). These genes included ARPC3, ACTB, ACTR3, ARPC2, ARPC5, and ACTG1. In addition, ARPC4 and ARPC1B were found to be differentially expressed in PC epithelial cells compared to LGD IPMN (Fig. 4C, supplementary Table S8).

The further pathway enrichment analyses revealed that Arp2/3-associated genes, including ARPC3, ACTB, ACTR3, ARPC2, ARPC5, and ACTG1, were significantly enriched (p-adjusted < 0.05) in integrin signaling and signaling by Rho family GTPases in both epithelial cells of PC and HGD IPMN (Fig. 4D). These pathways, which are crucial for cellular adhesion and motility, are potentially relevant to malignant transformation in pancreatic cysts^22,23.

Our study highlighted novel insights into the genetic architecture of benign pancreatic cysts and PC. Progression of benign pancreatic cysts to PC is an important step for understanding the molecular mechanisms of cancer progression. By using GWAS and subsequent proteogenomic analyses, we have identified distinct genetic variants associated with benign pancreatic cysts and PC, pointing towards common molecular mechanisms involved in high-risk pancreatic cysts and PC.

This study provides the first comprehensive analysis of the genetic architecture of benign pancreatic cysts, utilizing GWAS data. Our discovery GWAS revealed a significant association between genetic variants near the OPCML in individuals diagnosed with benign cysts versus healthy controls. OPCML gene is known to have tumor inhibitory effects in many cancers. This finding suggests a potential genetic predisposition that the benign pancreatic cysts are protected from developing to malignant by this tumor suppressor mechanism^24–27. On the other hand, in our pairwise GWAS, genetic variants near the well-known PC-associated gene BCAR1 on chromosome 16 was significantly associated with PC compared to benign pancreatic cysts²⁸. The lack of overlap in genome-wide significant SNPs between the discovery and pairwise GWAS suggests distinct genetic pathways for benign pancreatic cysts and PC. This distinction is essential for understanding the molecular mechanisms underlying the malignant transformation in benign pancreatic cysts and also suggests why all cysts do not turn to PC.

We also developed a PRS based on 32 genetic variants from the discovery GWAS to predict the risk of developing benign pancreatic cysts. Our analysis demonstrated that 65% individuals diagnosed with benign pancreatic cyst had higher PRS falling within the upper percentiles of the PRS distribution. In contrast, individuals diagnosed with PC did not present in the upper percentile as benign pancreatic cyst. This could suggest that our constructed PRS based on the genetic architecture of benign pancreatic cyst reflects the specific genetic signature for benign cysts so it can be used to successfully distinguish benign cyst from PC. Another explanation, that the difference in the genetic architecture underlying both diseases, pancreatic cyst and PC. This finding showed that the potential of PRS as a predictive tool for benign pancreatic cysts.

We were interested to explore further how the genetic architecture could reveal the mechanisms underlying the transformation from benign pancreatic cyst to malignant. We annotated the genetic variants from our discovery GWAS that were genome-wide significant (p < 5 × 10^− 08) as well as the suggestive significance level (p < 10^− 05) and performed PPI. Our hypothesis was that the stronger PPI interaction with GWAS-gene protein products and PC differentially expressed proteins could reflect that mechanism underlying the transformation of benign pancreatic cysts to malignant PC.

Our PPI network analysis identified the Actin Arp 2/3 complex as a key molecular mechanism linking benign cysts and PC. The ARPC3 GWAS gene, a component of the Arp2/3 complex, was significantly upregulated in PC. The corresponding protein of ARPC3 showed extensive interaction with other Arp2/3 complex proteins, suggesting its critical role in the malignant transformation of pancreatic cysts. Arp2/3 complex-associated genes were shown to be regulated together for cell motility and actin cytoskeleton ^29,30. Further analysis of the Arp2/3 complex-associated genes revealed significant upregulation of these genes in both bulk and scRNA-seq data from PC. The consistent upregulation across various cell types, including acinar, fibroblast, endocrine, and epithelial cells, indicates the possible important role of the Arp2/3 complex in the tumor microenvironment. The Arp2/3 complex, known for its role in cell migration and invasion, has been studied in PC research. For example, silencing of the ARP2/3 complex subunits typically resulted in reduced cell migration capacity in PC ^29,30. Another study suggested that inhibition of Arp2/3 complex affects tumor microtubule assembly and cellular connections in PC ²⁹. Arp2/3 complex played important roles in PC initiation by promoting cancer progression and changing in actin dynamics ³¹. In their study, Zhao et al. demonstrated that the Arp2/3 complex influenced early pancreatic carcinogenesis and played important roles in PC initiation by promoting cancer progression and changing in actin dynamics ³¹. Consistently, our pathway enrichment analysis revealed that Arp2/3-complex associated genes were linked to integrin signaling and Rho family GTPase activities, which are crucial for cellular adhesion and motility, thereby influencing malignant transformation of pancreatic cysts and cancer progression. This suggested that the Arp2/3 complex-associated genes could serve as potential biomarkers for the malignant transformation of pancreatic cysts. This complex was also shown to influence KrasG12D-mediated acinar-to-ductal metaplasia and early pancreatic carcinogenesis ³¹.

To further explore the clinical relevance of these findings, we performed survival analyses on PC patients of TCGA. These analyses revealed that PC patients with higher expression of the ARP2/3 complex-associated genes had significantly shorter overall survival than those with lower expression levels with a hazard ratio of 1.68 (95%CI, 1.04–2.72, P = 0.033) (supplementary Figure S4). This suggests that the ARP2/3 complex-associated genes not only plays a role in the malignant transformation of pancreatic cysts but also impacts patient prognosis, making it a potential target for therapeutic interventions. The identification of the Arp2/3 complex-associated genes as a key player in the malignant transformation process hints its potential use for targeted therapies and early detection strategies ²⁹.

Our study has several limitations. Firstly, genetic data for benign pancreatic cyst were only obtained from UKBB and there is no other genetic data from independent cohorts to replicate our GWAS findings. Secondly, the lack of sufficient bulk and single-cell omics data of pancreatic cysts to compare with PC/healthy control to further validate the findings or extend the analysis.

Our study highlights several important areas for future research. Firstly, validation of our findings in larger cohorts is essential, along with an exploration of the functional roles of these genetic variants in the progression of pancreatic cysts to PC. Secondly, refining PRS models by incorporating additional genetic and environmental factors could enhance the accuracy of predicting benign pancreatic cysts and their potential transformation into malignancy. Investigating the intricate interactions between genetic variants and other molecular pathways involved in both pancreatic cysts and PC will provide a deeper understanding of their complex genetic landscape.

In conclusion, our study provides novel insights into the genetic architecture of benign pancreatic cysts and their potential link to PC. The identification of significant genetic variants and the role of the Arp2/3 complex in malignant transformation showed the importance of genetic screening and targeted research in improving the diagnosis and treatment of pancreatic cysts. These findings may help future studies to explore targeted therapies and early intervention strategies for individuals at high risk of developing pancreatic cancer.

UKBB participant stratifications and data analysis

Participants of this study were a part of the UKBB dataset, a large prospective cohort study consisting of more than 500,000 participants recruited in the United Kingdom ³². Full details of the UKBB study can be found on the UKBB website (https://biobank.ndph.ox.ac.uk/showcase/). UKBB received ethical approval from the National Information Governance Board for Health and Social Care and the National Health Service, Northwest Multi-Center Research Ethics Committee ³³. All participants gave informed consent through electronic signatures before enrolment in the study. This research has been conducted under approved UKB Project ID 102162. The data was downloaded, and thus the end of the follow up of the individuals was the 3rd of May 2023. The specific data fields used in this analysis were imputed genomic data, date of recruitment, age, sex, date of diagnosis, and Diagnostic codes ICD9 and ICD10.

The ICD10 codes was used to assess the diagnosis of the patients. We identified patient groups based on ICD10 codes in the hospital inpatient data (UKBB datasets field: 41270), which is curated from UKBB as provided. In the analysis, we included pancreatic cancer (ICD code C25 and Pancreatic cyst (ICD code K86.2). We only selected patients of European origin. For each group of patients, we identified a group of healthy controls, which were defined as all patients without any disease code ICD9 or ICD 10 codes. We used the MatchIt package ³⁴ to match the healthy control cases with the prevalent and incident cases based on age and sex. The match was performed by the nearest neighbor method using Euclidean distance as a measure of similarity. We used exact matching for sex. Patient characteristics are in supplementary Table S1.

We selected PC patients with cysts who have both ICD code C25 and K86.2 in the hospital inpatient data fields. Genetic data was downloaded from UKBB. The genotyping and imputation (and quality control) were performed by the UKBB ³². Genome-wide data were available from the UKBB v3 imputed data in BGEN v1.2 format.

Genome wide association analysis (GWAS)

Two case-control GWASs (Discovery GWAS and Pairwise GWAS) were performed using 90 million imputed autosomal genetic variants; single nucleotide polymorphisms (SNPs) from UKBB genetic data^32,35. Details about the quality control and imputation of the genotyped genetic variants can be found in the UK Biobank_genotyping_QC_documentation³⁶. All 707 individuals diagnosed with benign pancreatic cysts and 14,000 healthy controls were included in the discovery GWAS. 584 individuals diagnosed with (PC (cases) versus 707 individuals with pancreatic cysts (controls) were included in the pairwise GWAS to determine the differential genetic variants that are associated with the development of PC versus pancreatic cysts. Individuals diagnosed with pancreatic cysts and PC were excluded from both GWASs. SNPTEST v.2.5.2 was used to perform the GWAS analysis using the frequentist association analysis score method and adjusting for genetic principal components³⁷. GWAS associations results were filtered based on minor allele frequency MAF (MAF > 0.01), Hardy–Weinberg equilibrium (P > 5×10^− 7) and imputation info scores (INFO > 0.4). Results visualization was done in R v.4.3.1 as Manhattan plots and Q–Q plots using the qqman and plyr packages ³⁸.

Annotation of GWAS genes

National Center for Biotechnology Information (NCBI) SNP database, Variant Effector Predictor (VEP) and Functional mapping and annotation (FUMA version 1.5.2) were used to map the GWAS significant genetic variants (p < 10^− 08) and suggestive GWAS genetic variants (p < 10^− 05) to genes for both the discovery and pairwise GWAS. For the differential expressed genes and Protein-protein interaction section, we used only genes of the discovery GWAS which were referred to as GWAS-genes (supplementary Tables S2, S3, S9) ^39,40.

Construction of polygenic risk score (PRS)

PRS were constructed from top genetic variants of the discovery GWAS (p < 10^− 05) as described in Mansour Aly et al⁴¹. Genetic variants weighted by effect-size were used to calculate individual scores for four groups of individuals: benign pancreatic cyst, malignant pancreatic cyst, pancreatic cancer, and healthy individuals. Percentiles were calculated for PRS based on healthy controls; P_1 (< 20%), P_2 (20–40%), P_3 (40–60%), P_4 (60–80%) and P_5 (> 80%). Logistic regression models were performed for association analysis of the PRS using the scaled scores and adjusting for genetic principal components in R v4.3.1. All reported p values are two-sided.

Protein-protein interaction (PPI) network construction and subsetting

PPI network was performed using STRING v.12.0 ⁴² database with the list of commonly up and downregulated DEGs/DEPs of TCGA and CPTAC. PPI and co-expression data were downloaded from the string data base on 12th February 2024. GWAS-genes and 2,169 common DEGs/DEPs were used to construct a PPI network. To ensure the highest probability of protein-protein interaction and minimize random interactions, we selected interactions with a confidence score greater than 0.95 ^42,43. The reason for selecting the highest confidence score > 0.95 was to ensure the highest probability of PPI over randomly observing an interaction. Primary neighbors of GWAS genes were selected for the sub-setting of the network. Subset networks were selected if a GWAS gene was connected with any other protein in the network with a confidence score > 0.95 and by co-expression. The network generation, filtering, and figure generation were done by the R package Igraph 2.0.2 ⁴⁴.

TCGA, CPTAC and PC proteomics (Jiang et al., 2024) data analysis

Gene count (STAR-count) matrices for 140 pancreatic tumor tissues and 76 normal pancreatic tissues (67 normal adjacent tissues and 9 normal ductal tissues) were downloaded by r package TCGAbiolink ⁴⁵ for differential expression analysis. Samples exhibiting low read counts or poor-quality metrics (50% of genes have zero counts in a sample) were excluded to ensure data integrity. Normalization of raw counts was performed using the variance-stabilizing transformation (VST) function from the DESeq2 (V 1.44.0) package in R by adjusting sequencing depth. Quality control checks, including principal component analysis (PCA) and hierarchical clustering, were conducted to identify and remove outliers and batch effects. Count data was modeled using a negative binomial distribution and differential expression was assessed by comparing tumor samples to normal samples. Genes with an p-adjusted (false discovery rate) < 0.05 and absolute log fold change > 0.25 were considered significantly differentially expressed (DEGs).

Proteomics data were obtained from the CPTAC, which included 140 tumor tissues and 76 normal tissues (67 normal adjacent tissues and 9 normal ductal tissues) of pancreas. Wilcoxon rank-sum test was used which is a non-parametric approach suitable for data without a normal distribution assumption to identify DEPs between tumor and normal tissues. DEPs were selected based on a threshold of fold change greater than 0.25 and an p-adjusted of less than 0.05, to account for multiple testing corrections and minimize false discovery rates. We used R version 4.1.0 for all the statistical analysis.

Normalized protein expression matrix of 191 PC tumor patients compared to 90 paired normal࿼-adjacent tissues were downloaded from the study by Jiang et al., 2024 ²¹. Wilcoxon rank-sum test was used to identify if Arp2/3 associated proteins were differentially expressed in PC tumor patients compared to paired tumor-adjacent tissues ⁴⁶. Statistical analyses were conducted using R version 4.1.0.

scRNA-Seq data processing of PC tumors and control samples

We analyzed single-cell RNA-seq data from 24 pancreatic ductal adenocarcinoma (PC) tumors and 11 control pancreases, including samples from non-pancreatic tumor patients and non-malignant pancreatic tumor patients, obtained from GEO (GSA: CRA001160)⁴⁷. The samples had similar gender distributions but differed in age (PC: mean 59.54 ± 8.67; controls: mean 47 ± 12.34; p = 0.008). Using Seurat v4.3.0, we pre-processed the data, filtering out cells based on mitochondrial content, read counts, and gene coverage. For quality control, cells with > 10% mitochondrial gene expression, < 200 detected genes were excluded. This resulted in a dataset comprising 41,148 cells from PC samples and 15,544 cells from control samples for further analysis ⁴⁸. We normalized each sample, identified 2000 variable features, and integrated the data using Seurat's anchor-based methods ⁴⁹. After scaling and PCA, we identified⁴⁷. The difference of gene expression between PC and normal cells was assessed using MAST with FindMarkers function, focusing on genes with an p-adjusted < 0.05 and an absolute log fold change > 0.25 to identify significant changes indicative of up- or downregulation in tumor tissues.

scRNA-seq data processing of pancreatic cyst (HGD IPMN, LGD IPMN) and PC

We analyzed single-cell RNA-seq data from 2 high risk IPMN (HGD IPMN), 2 low risk IPMN (LGD IPMN) and 2 PC patients¹⁰. scRNA-seq count matrix were available on request ¹⁰. Data preprocessing involved filtering cells based on mitochondrial content, read counts, and gene coverage, resulting in 17,847 cells from HGD IPMN, LGD IPMN and PC. For further analysis, we removed 1,344 cells that contained < 10% mitochondrial genes and < 200 genes. The dataset was processed in Python3 using Scanpy V2.0⁵⁰ for numerical operations, including normalization, log-transformation, and scaling. Each sample was normalized, and batch effects were removed by ComBat command in Scanpy. Cell clusters and their marker genes were identified through PCA and clustering algorithms. For cell types, similar marker genes were used ¹⁰ as well scRNA-seq of PC mentioned above. For differential gene expression analysis, MAST package⁵¹ in R, integrating an R connector rpy2 package within Python3 was used to facilitate the same DE analysis by MAST. DEGs were calculated between HGD IPMN vs LGD IPMN and PC vs LGD IPMN. Genes were marked as DEGs with a p-adjusted < 0.05 and an absolute log fold change > 0.25 to highlight significant up- or downregulation in HGD IPMN and PC compared to LGD IPMN.

To identify significant pathways of DEGs calculated by scRNA-seq data, we utilized the Molecular Signatures Database (MSigDB)⁵² R package. We selected relevant gene sets, including hallmark and curated gene sets (C2). The GSEA algorithm evaluated the enrichment of these gene sets in DEGs, providing enrich pathways and gene list in the pathways. The adjusted p-values were calculated using the Benjamini-Hochberg (BH) method. We selected significant pathway if p-adjusted value was < 0.05.

Ethics approval and consent to participate

UK Biobank has approval from the Northwest Multi-centre Research Ethics Committee (MREC) as a Research Tissue Bank (RTB) approval. This approval means that researchers do not require separate ethical clearance and can operate under the RTB approval (there are certain exceptions to this which are set out in the Access Procedures, such as re-contact applications).

Consent for publication

Not appliable

Data availability

Raw data from the UK Biobank is available for approved researchers through the UK Biobank data-access protocol. STAR-count TCGA data is available in GDC data portal (https://portal.gdc.cancer.gov/projects/TCGA-PAAD). CPTAC data were downloaded from (https://proteomic.datacommons.cancer.gov/pdc/). PC scRNA-seq data were available under Accession number GSA: CRA001160. scRNA-seq data were available on request ¹⁰^. Any processed data (except UK Biobank) are available on request to [email protected]

Code availability

All codes used in this manuscript are available at https://github.com/SDTC-CPMed/PC_cyst.

Competing interests

MB is the scientific founder of Mavatar, Inc. The other authors declare that they have no competing interests.

Funding

Open access funding provided by Karolinska Institute. This work was supported by funding from Swedish Cancer Society CAN 2017/411 and Swedish Cancer Foundation 230759PT03H.

Authors' contributions

AKM.F.M. , D.M.A. and M.B. had a primary role in analysis and interpretation of the data, AKM.F.M wrote the main manuscript text and prepared figures 1, 3 and 4. D.M.A. prepared figure 2. M.S., Y.Z., and O.S. contributed to those analyses. X.L. supervised these studies. All authors contributed to the writing of the manuscript. All authors reviewed and approved the manuscript.

Chhoda, A. et al. Benefit of Extended Surveillance of Low-Risk Pancreatic Cysts After 5-Year Stability: A Systematic Review and Meta-Analysis. Clinical Gastroenterology and Hepatology 21, 1430-1446 (2023). https://doi.org:https://doi.org/10.1016/j.cgh.2022.04.025
Distler, M. et al. Pathohistological subtype predicts survival in patients with intraductal papillary mucinous neoplasm (IPMN) of the pancreas. Ann Surg 258, 324-330 (2013). https://doi.org:10.1097/SLA.0b013e318287ab73
Salvia, R., Burelli, A., Perri, G. & Marchegiani, G. State-of-the-art surgical treatment of IPMNs. Langenbecks Arch Surg 406, 2633-2642 (2021). https://doi.org:10.1007/s00423-021-02349-9
Maker, A. V. et al. Cyst Fluid Biosignature to Predict Intraductal Papillary Mucinous Neoplasms of the Pancreas with High Malignant Potential. J Am Coll Surg 228, 721-729 (2019). https://doi.org:10.1016/j.jamcollsurg.2019.02.040
Maker, A. V. et al. Pancreatic cyst fluid and serum mucin levels predict dysplasia in intraductal papillary mucinous neoplasms of the pancreas. Ann Surg Oncol 18, 199-206 (2011). https://doi.org:10.1245/s10434-010-1225-7
Maker, A. V. et al. Cyst fluid biomarkers for intraductal papillary mucinous neoplasms of the pancreas: a critical review from the international expert meeting on pancreatic branch-duct-intraductal papillary mucinous neoplasms. J Am Coll Surg 220, 243-253 (2015). https://doi.org:10.1016/j.jamcollsurg.2014.11.001
Yip-Schneider, M. T., Carr, R. A., Wu, H. & Schmidt, C. M. Prostaglandin E(2): A Pancreatic Fluid Biomarker of Intraductal Papillary Mucinous Neoplasm Dysplasia. J Am Coll Surg 225, 481-487 (2017). https://doi.org:10.1016/j.jamcollsurg.2017.07.521
Iyer, M. K. et al. Digital spatial profiling of intraductal papillary mucinous neoplasms: Toward a molecular framework for risk stratification. Sci Adv 9, eade4582 (2023). https://doi.org:10.1126/sciadv.ade4582
Noe, M. et al. Genomic characterization of malignant progression in neoplastic pancreatic cysts. Nat Commun 11, 4085 (2020). https://doi.org:10.1038/s41467-020-17917-8
Bernard, V. et al. Single-Cell Transcriptomics of Pancreatic Cancer Precursors Demonstrates Epithelial and Microenvironmental Heterogeneity as an Early Event in Neoplastic Progression. Clin Cancer Res 25, 2194-2205 (2019). https://doi.org:10.1158/1078-0432.CCR-18-1955
Tam, V. et al. Benefits and limitations of genome-wide association studies. Nat Rev Genet 20, 467-484 (2019). https://doi.org:10.1038/s41576-019-0127-1
Plomin, R. & von Stumm, S. The new genetics of intelligence. Nat Rev Genet 19, 148-159 (2018). https://doi.org:10.1038/nrg.2017.104
Schaid, D. J., Chen, W. & Larson, N. B. From genome-wide associations to candidate causal variants by statistical fine-mapping. Nat Rev Genet 19, 491-504 (2018). https://doi.org:10.1038/s41576-018-0016-z
Torkamani, A., Wineinger, N. E. & Topol, E. J. The personal and clinical utility of polygenic risk scores. Nat Rev Genet 19, 581-590 (2018). https://doi.org:10.1038/s41576-018-0018-x
Yang, T. L. et al. A road map for understanding molecular and genetic determinants of osteoporosis. Nat Rev Endocrinol 16, 91-103 (2020). https://doi.org:10.1038/s41574-019-0282-7
Reay, W. R. & Cairns, M. J. Advancing the use of genome-wide association studies for drug repurposing. Nat Rev Genet 22, 658-671 (2021). https://doi.org:10.1038/s41576-021-00387-z
Przybyla, L. & Gilbert, L. A. A new era in functional genomics screens. Nat Rev Genet 23, 89-103 (2022). https://doi.org:10.1038/s41576-021-00409-w
Yang, X., Kar, S., Antoniou, A. C. & Pharoah, P. D. P. Polygenic scores in cancer. Nat Rev Cancer 23, 619-630 (2023). https://doi.org:10.1038/s41568-023-00599-x
Kachuri, L. et al. Principles and methods for transferring polygenic risk scores across global populations. Nat Rev Genet 25, 8-25 (2024). https://doi.org:10.1038/s41576-023-00637-2
Mullins, R. D., Heuser, J. A. & Pollard, T. D. The interaction of Arp2/3 complex with actin: nucleation, high affinity pointed end capping, and formation of branching networks of filaments. Proc Natl Acad Sci U S A 95, 6181-6186 (1998). https://doi.org:10.1073/pnas.95.11.6181
Jiang, L. et al. Prospective observational study on biomarkers of response in pancreatic ductal adenocarcinoma. Nat Med 30, 749-761 (2024). https://doi.org:10.1038/s41591-023-02790-x
Brannon, A., 3rd et al. Beta 1 integrin signaling mediates pancreatic ductal adenocarcinoma resistance to MEK inhibition. Sci Rep 10, 11133 (2020). https://doi.org:10.1038/s41598-020-67814-9
Kimmelman, A. C. et al. Genomic alterations link Rho family of GTPases to the highly invasive phenotype of pancreas cancer. Proc Natl Acad Sci U S A 105, 19372-19377 (2008). https://doi.org:10.1073/pnas.0809966105
Khamko, R., Daduang, J., Settasatian, C. & Limpaiboon, T. OPCML Exerts Antitumor Effects in Cholangiocarcinoma via AXL/STAT3 Inactivation and Rho GTPase Down-regulation. Cancer Genomics Proteomics 18, 771-780 (2021). https://doi.org:10.21873/cgp.20296
Lian, B. et al. Expression and promoter methylation status of OPCML and its functions in the inhibition of cell proliferation, migration, and invasion in breast cancer. Breast Cancer 28, 448-458 (2021). https://doi.org:10.1007/s12282-020-01179-9
Antony, J., Zanini, E., Birtley, J. R., Gabra, H. & Recchi, C. Emerging roles for the GPI-anchored tumor suppressor OPCML in cancers. Cancer Gene Ther 28, 18-26 (2021). https://doi.org:10.1038/s41417-020-0187-6
Simovic, I., Castano-Rodriguez, N. & Kaakoush, N. O. OPCML: A Promising Biomarker and Therapeutic Avenue. Trends Cancer 5, 463-466 (2019). https://doi.org:10.1016/j.trecan.2019.06.002
Costamagna, A. et al. Docking Protein p130Cas Regulates Acinar to Ductal Metaplasia During Pancreatic Adenocarcinoma Development and Pancreatitis. Gastroenterology 162, 1242-1255 e1211 (2022). https://doi.org:10.1053/j.gastro.2021.12.242
Latario, C. J. et al. Tumor microtubes connect pancreatic cancer cells in an Arp2/3 complex-dependent manner. Mol Biol Cell 31, 1259-1272 (2020). https://doi.org:10.1091/mbc.E19-11-0605
Rauhala, H. E., Teppo, S., Niemela, S. & Kallioniemi, A. Silencing of the ARP2/3 complex disturbs pancreatic cancer cell migration. Anticancer Res 33, 45-52 (2013).
Zhao, Y. et al. mTORC1 and mTORC2 Converge on the Arp2/3 Complex to Promote Kras(G12D)-Induced Acinar-to-Ductal Metaplasia and Early Pancreatic Carcinogenesis. Gastroenterology 160, 1755-1770 e1717 (2021). https://doi.org:10.1053/j.gastro.2020.12.061
Bycroft, C. et al. The UK Biobank resource with deep phenotyping and genomic data. Nature 562, 203-209 (2018). https://doi.org:10.1038/s41586-018-0579-z
Littlejohns, T. J. et al. The UK Biobank imaging enhancement of 100,000 participants: rationale, data collection, management and future directions. Nat Commun 11, 2624 (2020). https://doi.org:10.1038/s41467-020-15948-9
Ho, D., Imai, K., King, G. & Stuart, E. A. MatchIt: Nonparametric Preprocessing for Parametric Causal Inference. Journal of Statistical Software 42, 1 - 28 (2011). https://doi.org:10.18637/jss.v042.i08
Sudlow, C. et al. UK biobank: an open access resource for identifying the causes of a wide range of complex diseases of middle and old age. PLoS Med 12, e1001779 (2015). https://doi.org:10.1371/journal.pmed.1001779
Genotyping and quality control of UK Biobank, a large-scale, extensively phenotyped prospective resource Information for researchers v1.2 Oct 2015.
Marchini, J., Howie, B., Myers, S., McVean, G. & Donnelly, P. A new multipoint method for genome-wide association studies by imputation of genotypes. Nature Genetics 39, 906-913 (2007). https://doi.org:10.1038/ng2088
Turner, J. o. O. S. S., 3(25), 731, https://doi.org/10.21105/joss.00731. qqman: an R package for visualizing GWAS results using Q-Q and manhattan plots. Journal of Open Source Software, 3(25), 731 (2018). https://doi.org:https://doi.org/10.21105/joss.00731
Watanabe, K., Taskesen, E., van Bochoven, A. & Posthuma, D. Functional mapping and annotation of genetic associations with FUMA. Nature Communications 8, 1826 (2017). https://doi.org:10.1038/s41467-017-01261-5
Watanabe, K., Umićević Mirkov, M., de Leeuw, C. A., van den Heuvel, M. P. & Posthuma, D. Genetic mapping of cell type specificity for complex traits. Nature Communications 10, 3222 (2019). https://doi.org:10.1038/s41467-019-11181-1
Mansour Aly, D. et al. Genome-wide association analyses highlight etiological differences underlying newly defined subtypes of diabetes. Nature Genetics 53, 1534-1542 (2021). https://doi.org:10.1038/s41588-021-00948-2
Szklarczyk, D. et al. The STRING database in 2023: protein-protein association networks and functional enrichment analyses for any sequenced genome of interest. Nucleic Acids Res 51, D638-D646 (2023). https://doi.org:10.1093/nar/gkac1000
Bozhilova, L. V., Whitmore, A. V., Wray, J., Reinert, G. & Deane, C. M. Measuring rank robustness in scored protein interaction networks. BMC Bioinformatics 20, 446 (2019). https://doi.org:10.1186/s12859-019-3036-6
Csárdi, G. & Nepusz, T.
Colaprico, A. et al. TCGAbiolinks: an R/Bioconductor package for integrative analysis of TCGA data. Nucleic Acids Res 44, e71 (2016). https://doi.org:10.1093/nar/gkv1507
Jiang, L. et al. Prospective observational study on biomarkers of response in pancreatic ductal adenocarcinoma. Nat Med (2024). https://doi.org:10.1038/s41591-023-02790-x
Peng, J. et al. Single-cell RNA-seq highlights intra-tumoral heterogeneity and malignant progression in pancreatic ductal adenocarcinoma. Cell Res 29, 725-738 (2019). https://doi.org:10.1038/s41422-019-0195-y
Stuart, T. et al. Comprehensive Integration of Single-Cell Data. Cell 177, 1888-1902.e1821 (2019). https://doi.org:10.1016/j.cell.2019.05.031
Stuart, T. et al. Comprehensive Integration of Single-Cell Data. Cell 177, 1888-1902 e1821 (2019). https://doi.org:10.1016/j.cell.2019.05.031
Wolf, F. A., Angerer, P. & Theis, F. J. SCANPY: large-scale single-cell gene expression data analysis. Genome Biology 19, 15 (2018). https://doi.org:10.1186/s13059-017-1382-0
Finak, G. et al. MAST: a flexible statistical framework for assessing transcriptional changes and characterizing heterogeneity in single-cell RNA sequencing data. Genome Biology 16, 278 (2015). https://doi.org:10.1186/s13059-015-0844-5
Liberzon, A. et al. Molecular signatures database (MSigDB) 3.0. Bioinformatics 27, 1739-1740 (2011). https://doi.org:10.1093/bioinformatics/btr260

No competing interests reported.

SupplementaryTables.xlsx
Supplementary Tables S1-S9
SupplemetaryFigures.docx
Supplementary Figure S1-S4

Download PDF

Reviewers agreed at journal
09 Sep, 2024
Reviewers agreed at journal
09 Sep, 2024
Reviews received at journal
12 Aug, 2024
Reviews received at journal
12 Aug, 2024
Reviewers agreed at journal
12 Aug, 2024
Reviewers agreed at journal
12 Aug, 2024
Reviewers invited by journal
12 Aug, 2024
Editor assigned by journal
12 Aug, 2024
Editor invited by journal
11 Jul, 2024
Submission checks completed at journal
08 Jul, 2024
First submitted to journal
20 Jun, 2024

You are reading this latest preprint version

Proteogenomic analysis reveals Arp 2/3 complex as a common molecular mechanism in high risk pancreatic cysts and pancreatic cancer

Status:

Version 1

Abstract

Figures

Introduction

Results

Genetic architecture of pancreatic cyst

Polygenic risk score (PRS) for prediction of the risk of developing benign pancreatic cyst

Arp2/3 complex as a common molecular mechanism between pancreatic cancer and pancreatic cyst

Arp 2/3 complex-associated genes differed in gene expression and may be promising biomarkers in PC

Arp2/3 complex-associated genes as potential biomarkers for malignant transformation in pancreatic cysts

Discussion

Methods

UKBB participant stratifications and data analysis

Genome wide association analysis (GWAS)

Annotation of GWAS genes

Construction of polygenic risk score (PRS)

TCGA, CPTAC and PC proteomics (Jiang et al., 2024) data analysis

scRNA-Seq data processing of PC tumors and control samples

scRNA-seq data processing of pancreatic cyst (HGD IPMN, LGD IPMN) and PC

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1