Integrated profiling of human pancreatic cancer organoids reveals chromatin accessibility features associated with drug sensitivity

Abstract

Chromatin accessibility plays an essential role in controlling cellular identity and the therapeutic response of human cancers. However, the chromatin accessibility landscape and gene regulatory network of pancreatic cancer are largely uncharacterized. Here, we integrate the chromatin accessibility profiles of 84 pancreatic cancer organoid lines with whole-genome sequencing data, transcriptomic sequencing data and the results of drug sensitivity analysis of 283 epigenetic-related chemicals and 5 chemotherapeutic drugs. We identify distinct transcription factors that distinguish molecular subtypes of pancreatic cancer, predict numerous chromatin accessibility peaks associated with gene regulatory networks, discover regulatory noncoding mutations with potential as cancer drivers, and reveal the chromatin accessibility signatures associated with drug sensitivity. These results not only provide the chromatin accessibility atlas of pancreatic cancer but also suggest a systematic approach to comprehensively understand the gene regulatory network of pancreatic cancer in order to advance diagnosis and potential personalized medicine applications.

Conflict of interest statement

The authors declare no competing interests.

© 2022. The Author(s).

Figures

Fig. 1. Establishment and histological characterization of PDPCOs.

a Overview of the study. A total of 84 PDPCO lines, including PDAC, NEN, IPMN and ACC, were established by samples obtained by surgery or EUS-FNA. Original tissues were used for histological examination and WES/panel sequencing. PDPCOs were subjected to engraftment, histological examination, WGS/WES, RNA-seq, ATAC-seq and drug screening. b Representative brightfield image of PDPCOs generated from normal pancreas, PDAC, IPMN, NEN and ACC tissue. Scale bar, 200 μm. Experiments are repeated at least three times with similar results. c, Representative H&E staining of PDAC primary tumors, organoids, and xenografts (CAS-DAC-13, CAS-DAC-29 and CAS-DAC-30). Scale bar, 50 μm. Experiments are repeated at least three times with similar results. d Representative H&E and alcian blue staining of an IPMN primary tissue, organoid and xenograft (CAS-IPMN-1). Scale bar, 50 μm. Experiments are repeated at least three times with similar results. e Representative H&E, SYP, CHGA and Ki67 staining of NEN primary tissues, organoids and xenografts (CAS-NEN-2 and CAS-NEN-4). CAS-NEN-2 was clinically diagnosed as WHO grade G2, while CAS-NEN-4 was diagnosed as NEC. Scale bar, 50 μm. Experiments are repeated at least three times with similar results. PDPCO, patient-derived pancreatic cancer organoid; PDAC, pancreatic ductal adenocarcinoma; NEN, pancreatic neuroendocrine neoplasm; IPMN, intraductal papillary mucinous neoplasm; ACC, acinar cell carcinoma; EUS-FNA, endoscopic ultrasound-guided fine needle aspiration biopsy.

Fig. 2. WGS analysis of genomic mutations and alterations in PDPCOs.

Genomic landscape of 70 exocrine PDPCOs, including PDACs, IPMNs, and ACCs (left), and 4 neuroendocrine PDPCOs (right). The figure shows exonic mutations; noncoding mutations in UTR, upstream/downstream, intronic, ncRNA and intergenic regions; and CNVs. CAS-DAC-23 with WES of blood DNA; CAS-NEN-2 without sequencing of blood DNA. PDPCO patient-derived pancreatic cancer organoid, PDAC pancreatic ductal adenocarcinoma, NEN pancreatic neuroendocrine neoplasm, IPMN intraductal papillary mucinous neoplasm, ACC acinar cell carcinoma, CNV copy number variation.

Fig. 3. Transcriptome analysis of PDPCOs.

a PCA of gene expression for 45 PDPCOs in the discovery cohort: 39 CAS-DACs, 4 CAS-NENs, 1 CAS-IPMN and 1 CAS-ACC. b Heatmap of the four transcriptomic subtypes in the discovery cohort of 40 exocrine PDPCOs (39 CAS-DACs and 1 CAS-IPMN), as revealed by unsupervised NMF clustering. Samples are listed in columns, and signature genes are listed in rows as z-scores. c Identification of biological characteristics in each subtype by previously reported signatures. A P-value < 0.01 was used to determine positive or negative enrichment of published signatures in a subtype (top). Detailed GSEA results are shown at the bottom. d Bar graph showing the results of enrichment analysis of class 4 DEGs by Enrichr. Significance was computed by one-sided hypergeometric test. e Heatmap indicating the expression of glycolysis pathway-related genes across the four subtypes. Samples are listed in rows, and signature genes are listed in columns as z-scores. f External validation of the transcriptomic subtypes in the TCGA PAAD cohort. Samples are listed in rows, and signature genes are listed in columns as z-scores. g Kaplan-Meier survival analysis showing the overall survival characteristics of patients with the four transcriptomic subtypes, as identified in f. P value is determined by using log-rank test. PDPCO, patient-derived pancreatic cancer organoid; PDAC, pancreatic ductal adenocarcinoma; NEN, pancreatic neuroendocrine neoplasm; IPMN, intraductal papillary mucinous neoplasm; ACC, acinar cell carcinoma; NPO, normal pancreatic ductal organoid; PCA, principal component analysis; NMF, nonnegative matrix factorization; GSEA, gene set enrichment analysis; DEG, differentially expressed gene.

Fig. 4. Integrated analysis of RNA-seq and ATAC-seq revealing TF regulons for PDPCO subtypes and putative peak-to-gene links.

a Schematic of the process for generating subtype-specific TF regulons in PDPCOs. Comparison of enriched TF regulons between exocrine and neuroendocrine PDPCOs in b and among the four PDAC transcriptome subtypes in c. The color of each heatmap on the left represents the regulon z-score. The color of each heatmap on the right represents the -log10(P value) of the regulon score. P values associated with one-sided unpaired t test were adjusted for multiple testing using FDR. Representative subtype-specific TF regulons are listed. d Heatmap of 2,257 putative links between ATAC-seq peaks (left) and genes (right) in 41 exocrine PDPCOs (39 PDACs, 1 IPMN, and 1 ACC). Each row represents an individual link between one ATAC-seq peak and one gene. The color of each heatmap represents the z-score for ATAC-seq peak’s accessibility (left) or the z-score for gene expression (right). TF, transcription factor; PDPCO, patient-derived pancreatic cancer organoid; TG, target gene; PDAC, pancreatic ductal adenocarcinoma; FDR, false discovery rate.

Fig. 5. Integrated analysis of noncoding mutations and drug sensitivity in PDPCOs.

a Schematic for identifying potential functional noncoding mutations in 31 exocrine PDPCOs. b Proportions of noncoding mutations located within and outside ATAC peaks. c Dot plot showing fold change in accessibility at ATAC-seq peaks containing noncoding mutations across above 31 exocrine PDPCO. d Box plot showing chromatin accessibility at RIMBP2 putative enhancer and RIMBP2 gene expression across above 31 exocrine PDPCOs. Boxplot show the median (central line), the 25–75% interquartile range (box limits). e Normalized ATAC-seq tracks of RIMBP2 putative enhancer locus in 5 representative samples. The red and gray tracks represent samples with and without mutation of the RIMBP2 putative enhancer, respectively. Black dotted line indicates the position of the mutation, and the predicted enhancer region is highlighted by light blue shading. f Kaplan-Meier analysis of overall survival in the TCGA PAAD cohort stratified by high and low expression of RIMBP2 gene. P value is determined by using log-rank test. g Heatmap of 15,397 putative peak-to-drug links in 39 exocrine PDPCOs. Each row represents an individual link between one ATAC-seq peak and one drug. The color represents the z-score for chromatin accessibility (left) or the z-score for drug AUC (right). Dot plot showing a representative negative correlation in h and a representative positive correlation in i. Significance is computed by Pearson’s correlation coefficients without adjustment. j Normalized ATAC-seq tracks of BAG3 putative enhancer locus in 10 representative samples. The yellow and purple tracks represent samples with and without peaks in BAG3 putative enhancer, respectively. The peak region is highlighted by light blue shading. Comparison between peak high group with peaks in the BAG3 putative enhancer and peak low group without such peaks in the above 10 samples of BAG3 gene expression in k, AUC of 5-FU in l and AUC of PTX in m. Each group contains 5 biologically independent samples. Boxplot show the median (central line), the 25–75% interquartile range (box limits). Significance is computed by two-sided unpaired t test. PDPCO, patient-derived pancreatic cancer organoid; AUC, area under curve; 5-FU, 5-fluorouracil; PTX, paclitaxel.

Fig. 6. Clinically relevant chemosensitivity of PDPCOs with validation in in vivo xenograft models.

a AUCs of 5 chemotherapeutic agents in 39 exocrine PDPCOs (38 CAS-DACs and 1 CAS-IPMN). With respect to each chemotherapeutic agent, the 39 PDPCOs were divided equally into sensitive (blue, AUC lowest 1st-13th), intermediate (yellow, AUC lowest 14th–26th) and resistant (red, AUC lowest 27th–39th) groups. The gray dot indicated CAS-DAC-14. b Diagram of the clinical analysis workflow. c Kaplan-Meier survival analysis showing the recurrence-free survival outcomes of the patients corresponding to the 39 PDPCOs. P value is determined by using log-rank test. Based on the consistency between the clinical adjuvant chemotherapy regimen and chemosensitivity of the matched organoid, three groups were identified: sensitive (n = 10), intermediate (n = 13) and resistant (n = 8). d Representative radiation examination of both the surgical area and liver in the sensitive group (CAS-DAC-24, CE-CT), intermediate group (CAS-DAC-20, CE-MRI), and resistant group (CAS-DAC-22, CE-MRI) at the time of diagnosis and six months post-surgery. Yellow arrow, primary tumor; Blue arrow, hepatic cyst; Red arrow, metastasis. e Drug test of ODX-18 with GEM (n=6), using Vehicle as a control (n=6). Data are presented as mean values + SEM. Statistical Significance was computed by two-sided unpaired t test. Source data are provided as a Source Data file. f Drug test of ODX-20 with GEM (n = 6) and 5-FU (n = 6), using Vehicle as a control (n = 6). Data are presented as mean values + SEM. Statistical Significance was computed by two-sided unpaired t test. Source data are provided as a Source Data file. g Drug test of ODX-36 with GEM (n = 6), 5-FU (n = 6) and PTX (n = 6), using Vehicle as a control (n=6). Data are presented as mean values + SEM. Statistical Significance was computed by two-sided unpaired t test. Source data are provided as a Source Data file. Statistical analysis, ns P ≥ 0.05,*P < 0.05, **P < 0.01, ***P < 0.001. PDPCO, patient-derived pancreatic cancer organoid; AUC, area under curve; GEM, gemcitabine; 5-FU, 5-fluorouracil; PTX, paclitaxel; OXA, oxaliplatin; IRI, irinotecan; CE-CT: contrast-enhanced computed tomography; CE-MRI: contrast-enhanced magnetic resonance imaging.