PTHrP Drives Pancreatic Cancer Growth and Metastasis and Reveals a New Therapeutic Vulnerability

Abstract

Pancreatic cancer metastasis is a leading cause of cancer-related deaths, yet very little is understood regarding the underlying biology. As a result, targeted therapies to inhibit metastasis are lacking. Here, we report that the parathyroid hormone-related protein (PTHrP encoded by PTHLH) is frequently amplified as part of the KRAS amplicon in patients with pancreatic cancer. PTHrP upregulation drives the growth of both primary and metastatic tumors in mice and is highly enriched in pancreatic ductal adenocarcinoma metastases. Loss of PTHrP-either genetically or pharmacologically-dramatically reduces tumor burden, eliminates metastasis, and enhances overall survival. These effects are mediated in part through a reduction in epithelial-to-mesenchymal transition, which reduces the ability of tumor cells to initiate metastatic cascade. Spp1, which encodes osteopontin, is revealed to be a downstream effector of PTHrP. Our results establish a new paradigm in pancreatic cancer whereby PTHrP is a driver of disease progression and emerges as a novel therapeutic vulnerability. SIGNIFICANCE: Pancreatic cancer often presents with metastases, yet no strategies exist to pharmacologically inhibit this process. Herein, we establish the oncogenic and prometastatic roles of PTHLH, a novel amplified gene in pancreatic ductal adenocarcinoma. We demonstrate that blocking PTHrP activity reduces primary tumor growth, prevents metastasis, and prolongs survival in mice.This article is highlighted in the In This Issue feature, p. 1601.

Figure 1

(A) PTHLH and KRAS Log2 copy-number values from TCGA-PAAD cBioportal. Each circle indicates one patient. (B) Oncoprints of copy number gains and amplifications for PTHLH or KRAS from TCGA-PAAD or UTSW datasets. (C) Box and whisker plot of PTHLH copy number Log2 Ratio distribution of common cancer cell lines in the CCLE database. Number of samples per cohort is indicated in parentheses. (D) Representative FISH images of nuclei stained for PTHLH (red) and DAPI (blue) in human pancreatic cancer and normal ductal epithelial cell lines. Each red spot is one copy of PTHLH DNA. Yellow dotted lines indicate individual nuclei. Violin plots depict the number of PTHLH copies per nuclei in the indicated cell lines. Scale bars = 5μm. Statistical analysis by Mann-Whitney U test with significance indicated (*, p<0.05; N=50–100 nuclei per cell line). (E) Kaplan-Meier overall survival plots stratified by PTHLH copy number gains (red) or PTHLH diploid (blue) status from TCGA-PAAD. Logrank p=9.0×10⁻³. (F) Kaplan-Meier overall survival plots stratified by high PTHLH (red) or low PTHLH (blue) mRNA expression from TCGA-PAAD. Logrank p=2.1×10⁻³. (G-H) (G) Kaplan-Meier overall survival plots stratified by high PTHrP (red) or low PTHrP (blue) protein expression from a PDAC patient TMA. Logrank p=3.6×10⁻². (H) Representative IHC images of low PTHrP or high PTHrP staining in PDAC patient tissue. Scale bars = 100μm.

Figure 2

(A) Kaplan-Meier overall survival plots stratified by PTHLH copy number gains (red) or PTHLH diploid (blue) from the COMPASS trial cohort of PDAC patients. Logrank p=3.1×10⁻². (B) Percentage of patients in the COMPASS trial that have PTHLH copy number gains segregated into non-metastatic and metastatic patients. (C) Percentage of patients in the COMPASS trial that have PTHLH copy number gains segregated into Classical and Basal-like PDAC subtypes. (D-F) Box and whisker plots showing quantification of PTHLH mRNA expression in PDAC subtypes identified by (D) Bailey et al. [3], (E) Collisson et al. [4], and (F) Moffitt et al. [5] Statistical analysis by Student’s unpaired t-test with significance indicated (*, p<0.05). (G) PTHLH mRNA expression (RPKM) in a panel of PDAC CCLE cell lines that are Squamous (blue) or Progenitor (yellow) subtypes. (H) Box and whisker plot showing quantification of PTHLH mRNA expression in EMT subtypes identified by Aiello et al [34]. Statistical analysis by Student’s unpaired t-test with significance indicated (*, p<0.05). (I) Volcano plot of differentially expressed genes in p120ctn^−/− versus p120ctn^+/+ identified by Reichert et al [10]. Green dots passed p-value <0.05 and fold change >2 cut-offs. Pthlh is enlarged and highlighted.

Figure 3

(A) Representative IHC images of PTHrP staining in PanIN (from KC mice) or PDAC tumors (from KPC mice). Scale bars = 100μm. (B) Western blots of PTHrP and ACTIN in cell lines isolated from WT pancreatic ductal epithelial cells (Pdx1-Cre; LSL-Rosa26^YFP/YFP) or KPCY pancreatic cancer cells. (C) Representative IF images of metastatic liver lesions from KPCY mice stained with YFP (green), PTHrP (red), and DAPI (blue). Scale bars = 100μm. (D-F) (D) Representative brightfield dissection scope images of orthotopic injection tumors from KPCY-shNT, KPCY-shPthlh #1, and KPCY-shPthlh #2 cell lines 3-weeks post-implantation. (E) Quantification of pancreatic tumor burden normalized to body mass after orthotopic injection. (F) Quantification of the percentage of YFP+ metastatic tumor cells present in FACS analyzed livers in mice with implanted orthotopic primary tumors. Statistical analysis by Student’s unpaired t-test with significance indicated (*, p<0.05; error bars indicate SD; n=5 mice/group).

Figure 4

(A) Schematic drawing showing pancreas epithelial cell-specific heterozygous deletion of Pthlh in the KPCY mouse pancreatic cancer model. (B) Representative IHC images of PTHrP staining in pancreas from KPCY or KPCY-Pthlh^HET mice. Scale bars = 100μm. (C-H) (C) Kaplan-Meier overall survival plots of KPCY or KPCY-Pthlh^HET mice cohorts. Logrank p=3.0×10⁻⁴. (D) Quantification of the percentage of mice with gross YFP+ metastases. (E) Quantification of endpoint pancreatic tumor mass normalized to body mass. (F) Quantification of endpoint liver mass normalized to body mass. (G) Quantification of endpoint lung mass normalized to body mass. (H) Representative brightfield and YFP images of pancreas, liver, and lungs at endpoint. Statistical analysis by Mann-Whitney U test (E) or Student’s unpaired t-test (F-G) with significance indicated (*, p<0.05; error bars indicate SD; n=41 mice and n=16 for KPCY or KPCY-Pthlh^HET cohorts, respectively).

Figure 5

(A-E) (A) Quantification of tumor volume over time from ultrasound imaging of orthotopic implanted tumors. Arrow indicates the induction of anti-PTHrP (or anti-IgG) therapy. (B) Representative brightfield and YFP images of orthotopic pancreas tumors after treatment with the indicated therapy. (C) Waterfall plots of percent change in tumor volume relative to baseline. Each bar indicates an independent animal. (D) Quantification of pancreatic tumor mass normalized to body mass. (E) Quantification of the percentage of mice with gross YFP+ metastases. Statistical analysis by ANOVA (A) or Student’s unpaired t-test (D) with significance indicated (*, p<0.05; error bars indicate SD; n=6 mice/group). (F-G) (F) Representative brightfield and YFP images of metastatic lung tumors 14-days post-tail vein injection and treatment with the indicated therapy beginning at day 5. Scale bars = 100μm. Quantification of lung mass normalized to body mass. (G) IF images of whole lungs after tail vein injection and treatment with the indicated therapy, stained with YFP (green) and DAPI (blue). Scale bars = 1mm. Quantification of percent YFP+ area relative to total lung area. Statistical analysis by Student’s unpaired t-test with significance indicated (*, p<0.05; error bars indicate SD; n=6 mice/group).

Figure 6

(A-B) (A) Representative IF images of primary pancreatic tumors stained with YFP (green), Ki-67 (red), and DAPI (blue). White arrows indicate YFP+/Ki-67+ tumor cells in epithelial clusters; white arrowheads indicate YFP+/Ki-67+ delaminating tumor cells that have begun to metastasize. Scale bars = 100μm. (B) Quantification of YFP+/Ki-67+ proliferating tumor cells per high powered field (HPF). Statistical analysis by Student’s unpaired t-test with significance indicated (*, p<0.05; error bars indicate SD; n=6 mice/group). (C) Western blots of PTH1R and ACTIN in cell lines isolated from WT pancreatic ductal epithelial cells (Pdx1-Cre; LSL-Rosa26^YFP/YFP) or KPCY pancreatic cancer cells. (D-E) (D) Cell viability assay measuring Optical Density (OD) after WST-1 cell proliferation reagent addition in a panel of murine KPCY PDAC cell lines treated with the indicated doses of anti-PTHrP monoclonal neutralizing antibody therapy for 48 hours. (E) Cell viability assay measuring OD after WST-1 cell proliferation reagent addition in 2838 and 6419 murine KPCY PDAC cell lines treated with 10μg/ml of anti-PTHrP monoclonal neutralizing antibody therapy over time. Statistical analysis by ANOVA with significance indicated (*, p<0.05; error bars indicate SEM; n=3 cultures/cell line). (F) Cell viability assay measuring OD after WST-1 cell proliferation reagent addition in human PDAC cell lines treated with the indicated doses of anti-PTHrP therapy for 48 hours. (error bars indicate SEM; n=3 cultures/cell line). (G-H) (G) Representative brightfield images of human pancreatic cancer patient-derived 3D organoid cultures after 48 hours of treatment. (H) Cell viability assay measuring luminescence after Cell Titer Glow 3D cell proliferation reagent addition in a panel of human pancreatic cancer patient-derived 3D organoid cultures treated with the indicated doses of anti-PTHrP therapy for 48 hours. (error bars indicate SEM; n=2 cultures/cell line).

Figure 7

(A-C) (A) Classification of the highest-grade lesions present in 5–7-months old KPCY or KPCY-Pthlh^HET mice. (B) Quantification of the percentage of mice with poorly-differentiated PDAC present in 5–7-months old KPCY or KPCY-Pthlh^HET mice. (C) Quantification of the percentage of pancreatic tissue per mouse designated poorly-differentiated in 5–7-months old KPCY or KPCY-Pthlh^HET mice. (D-E) (D) Representative IF images of primary pancreatic tumors stained with YFP (green), ECAD (red), and DAPI (blue). Scale bars = 100μm. (E) Quantification of YFP+/ECAD- delaminating tumor cells that have begun to metastasize. Statistical analysis by Student’s unpaired t-test with significance indicated (*, p<0.05; error bars indicate SEM; n=6 mice/group). (F-G) (F) Top gene set enrichment analyses (GSEA) plots by normalized enrichment score (NES) from the Hallmark (left) or GO (right) collections, enriched in shNT (blue) versus shPthlh (red) cells. (G) GSEA plots of the Hallmark EMT and TGF-β gene sets. The top genes lost in shPthlh and enriched in shNT cells are indicated in blue. NES, nominal p value, and FDR are indicated. (H-I) (H) Representative ECAD flow histograms for shNT and shPthlh cells after treatment with 10 ng/ml TGF-β (or vehicle) for 48 hours. Isotype control stained cells shown in yellow, vehicle treated cells in red, and TGF-β treated in blue. (I) Quantification of the percentage of ECAD low cells post-TGF-β treatment. Statistical analysis by Student’s unpaired t-test with significance indicated (*, p<0.05; error bars indicate SEM; n=3 cultures/group). (J-K) (J) Representative brightfield and YFP images of metastatic lung tumors 3-weeks post-tail vein injection. Scale bars = 100μm. Quantification of lung mass normalized to body mass. (K) IF images of whole lungs after tail vein injection, stained with YFP (green) and DAPI (blue). Scale bars = 1mm. Quantification of percent YFP+ area relative to total lung area. Statistical analysis by Mann-Whitney U test (J) or Student’s unpaired t-test (K) with significance indicated (*, p<0.05; error bars indicate SD; n=5 mice/group). (L) Representative IF images (left) and quantification (right) of DAPI-stained (blue) KPCY cells after 48 hours in a matrigel transwell cell invasion assay. Statistical analysis by Student’s unpaired t-test with significance indicated (*, p<0.05; error bars indicate SD; n=3 cultures/group).