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.
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
The overwhelming majority of patients with pancreatic ductal adenocarcinoma (PDAC) present with metastases, and nearly all will succumb to disease within 1 year of diagnosis (1). A limited understanding of the biological processes underlying metastasis has hampered the development of customized therapeutic interventions, resulting in pancreatic cancer evolving into the second leading cause of cancer-related deaths (2). In recent years, genomic and transcriptomic characterization of PDAC has revealed that this is not a singular disease but rather is comprised of subtypes with distinct molecular aberrations and characteristics (3–5). However, defining pathways that are activated within these subtypes to drive metastasis and implementing targeted therapies to disrupt these pathways remain elusive goals.
Transcriptomic profiling of PDAC has yielded its classification into two major subtypes—(i) classical/progenitor and (ii) squamous/quasi-mesenchymal/basal-like—with the latter exhibiting higher tumor grade, enhanced epithelial-to-mesenchymal transition (EMT)/invasion, and decreased survival (6). The squamous subtype has been defined further by lineage-specific markers—namely, TP63, PTHLH (parathyroid hormone like hormone, also known as PTHrP), KRT5, and KRT6A. The transcription factor tumor protein p63 (TP63) may constitute a master regulator of the other squamous lineage–specific genes (7). Whole-genome sequencing has established that the squamous subtype is annotated by KRAS genomic amplification (8), which is significantly enriched in metastatic tumors relative to primary tumors (9). Interestingly, co-amplification occurs in nearby genes within the KRAS amplicon, including PTHLH. Our own work, using genetically engineered mouse models of PDAC, has established that Pthlh is one of the top five most upregulated genes in a highly metastatic subset of cells that have undergone EMT (10). We therefore hypothesized that copy-number gains in the KRAS–PTHLH loci drive tumors toward a highly invasive squamous subtype that is marked by mesenchymal characteristics, potentially due to enhanced PTHLH expression and function.
PTHrP was described initially as a secreted factor produced by tumors that leads to paraneoplastic hypercalcemia (11). It has since been demonstrated to be involved in multiple cellular processes, including regulation of intracellular calcium, proliferation/hypertrophy, and differentiation (reviewed in refs. 12 and 13). These diverse functions are likely due to the processing of the mature PTHrP protein into at least three active peptides that are thought to act independently through autocrine, paracrine, and intracrine routes (12, 13). PTHrP isoforms are produced and secreted by malignant cells, including moieties with intact N-terminus as well as mid-region fragments. The most widely studied N-terminal peptide, PTHrP(1–34), binds to the parathyroid hormone receptor 1 (PTHR1) in an autocrine/paracrine manner to activate downstream effectors such as extracellular signal-regulated kinase (ERK), protein kinase A, protein kinase C, AKT, and cyclin D1, as well as RUNX and cAMP-responsive element binding protein transcription factors (14–18). Upstream drivers of Pthlh expression include members of the transforming growth factor beta (TGFβ) superfamily that signal through the SMAD transcription factors (19, 20), the RAS family through the erythroblast transformation-specific (ETS) transcription factors (21–24), and mitogenic stimuli such as epidermal growth factor and insulin-like growth factor 1 (23–25). Additionally, recent evidence has revealed that TP63 is a major driver of PTHLH expression in the squamous subtype of PDAC (7). PTHrP has been described as oncogenic and fostering bone metastasis in lung, breast, head and neck, lymphoma, and colon cancers (14, 18, 24, 26–29). Paradoxically, some studies have concluded that PTHLH has a tumor-suppressive role in breast cancer and skeletal metastases (30, 31), suggesting an incomplete understanding of the role that PTHrP plays during tumorigenesis. Furthermore, how PTHLH functions in vivo to modulate pancreatic cancer progression and metastasis has yet to be explored.
In the current study, we establish Pthlh as a bona fide oncogene and a novel metastasis driver in pancreatic cancer. To that end, we use genetic deletion and complementary pharmacologic inhibition of PTHrP to demonstrate that its loss significantly delays primary tumor growth, reduces metastatic colonization, and increases overall survival in mice. We nominate anti-PTHrP therapy as a viable candidate for the transition from preclinical models to clinical trials in PDAC.
PTHLH Is Amplified in a Subset of Patients with Pancreatic Cancer
To better understand the extent to which known KRAS amplifications in PDAC also encompass PTHLH (which are separated by 2.7 Mb on human chromosome 12 and by 2.0 Mb on murine chromosome 6; Supplementary Fig. S1A), we first evaluated all copy-number alterations for KRAS and PTHLH in patients with PDAC from The Cancer Genome Atlas (TCGA). A Spearman correlation coefficient of R = 0.97 (P = 8.96e−107) was observed between KRAS and PTHLH copy number (Fig. 1A). Additionally, treatment of two independent murine PDAC tumor cell lines with MEK1/2 (U0126) and ERK1/2 (SCH772984) inhibitors decreased Pthlh expression (Supplementary Fig. S1B and S1C), demonstrating that Pthlh is transcriptionally activated by RAS signaling. Further analysis of TCGA indicated that ∼25% of patients exhibited PTHLH copy-number gains (Fig. 1B). PTHLH was shown, in an independent patient cohort (32), to have copy-number gains in nearly one third of PDAC patients (Fig. 1B). A pan-cancer analysis of the Cancer Cell Line Encyclopedia (CCLE) database showed similar PTHLH and KRAS copy-number gains in other cancers (Supplementary Fig. S1D). Of all CCLE cell lines, pancreatic cancers harbored the second highest PTHLH amplifications, behind only bile duct carcinomas (Fig. 1C; Supplementary Fig. S1E). We demonstrated that copy number gains/amplifications functionally led to increased PTHLH mRNA expression in both the TCGA and CCLE datasets (Supplementary Fig. S1F and S1G). Additionally, we confirmed PTHLH amplification in human PDAC cell lines through fluorescence in situ hybridization (FISH), revealing multiple copies of PTHLH per cell, whereas the normal human pancreatic ductal epithelial cell line had the expected two copies per cell (Fig. 1D). Importantly, Kaplan–Meier survival analysis of the TCGA–Pancreatic Adenocarcinoma cohort showed that patients with PTHLH copy-number gains correlated with decreased survival relative to those with fewer than two copies (Fig. 1E). Patients with more than two copies of PTHLH died 3 months earlier than those with one or two copies (12.1 months and 15.3 months, respectively; Supplementary Fig. S1H). Additionally, increased PTHLH mRNA expression in the same cohort correlated with reduced survival (Fig. 1F). Finally, immunohistochemical (IHC) staining of tumor microarrays of patients with PDAC showed that high PTHLH protein expression correlated with decreased survival (Fig. 1G and H). Collectively, these data show that amplification of the KRAS locus in PDAC has concurrent PTHLH co-amplification, thereby setting the stage to pursue mechanistic studies of how PTHrP might be driving PDAC progression and/or metastasis.
PTHLH Is Enriched in Patients with Metastatic Pancreatic Cancer
PDAC samples in the TCGA are primarily nonmetastatic (American Joint Committee on Cancer TMN score of M0) or metastatic burden not classified (Mx), thereby making it difficult to evaluate for any potential correlation between PTHrP expression and proclivity for metastasis. The recently launched COMPASS trial (NCT02750657), which recruited patients with advanced-stage PDAC for genomic analysis, has provided a prism through which to view the genomic landscape of patients with stage IV PDAC (8). Whole-genome sequencing and transcriptomic RNA-sequencing (RNA-seq) profiling of laser capture microdissected tumor epithelia indicated that PTHLH is one of the top 20 genes used to classify a basal-like (analogous to squamous/quasimesenchymal, discussed above) transcriptional signature, along with the squamous lineage genes TP63, KRT5, and KRT6A (8). Furthermore, copy-number analysis in this cohort demonstrates that PTHLH gains correlate with decreased survival (Fig. 2A). Additionally, nearly one third of patients with metastatic PDAC lesions show PTHLH copy-number gains (Fig. 2B). Finally, PTHLH gains were strongly enriched in the basal-like signature (48% of basal-like patients exhibited PTHLH copy-number gains) relative to the classical signature (Fig. 2C).
These analyses show that PTHrP may be a potential driver of the invasive and highly metastatic basal subtype. To investigate this further in independent cohorts, we evaluated PTHLH mRNA expression in the publicly available dataset from Bailey and colleagues (3), which revealed that PTHLH mRNA was significantly enriched in the squamous subset relative to other subtypes (Fig. 2D). Additionally, PTHLH expression was highest in the quasi-mesenchymal subset of the PDAC dataset established by Collisson and colleagues (ref. 4; Fig. 2E). PTHLH was higher in the basal subset relative to the classical dataset generated by Moffitt and colleagues (ref. 5; Fig. 2F). Publicly available chromatin immunoprecipitation sequencing of TP63 in the squamous subtype cell line BxPC3 (7) revealed that TP63 binds to the promoter of PTHLH (Supplementary Fig. S2A). Thus, PTHrP is consistently highly expressed in the squamous/quasi-mesenchymal/basal-like subsets identified by others as being the most aggressive and least differentiated tumor subtype in PDAC.
Next, we found that human squamous subtype PDAC cell lines (DanG, BxPC3, and CFPACI; refs. 7, 33) harbored >100-fold higher PTHLH mRNA expression than PDAC cell lines of the nonsquamous subtypes (Fig. 2G). Analysis of RNA-seq from murine PDAC tumor cell lines classified according to an EMT subtype [complete EMT (cEMT) vs. partial EMT (pEMT); ref. 34] segregates them into squamous and classical subtypes. Pthlh expression is increased significantly in squamous/cEMT cell lines (Fig. 2H). Finally, we have demonstrated previously that P120CTN loss in the pancreata of mice results in a highly aggressive and metastatic phenotype (10). These highly mesenchymal and metastatic KPCY-p120ctn−/− tumor cells have >50-fold increased Pthlh expression relative to KPCY-p120ctn+/+ tumor cells (Fig. 2I). In fact, differential gene expression analysis of these cells indicates that Pthlh was the second-most upregulated gene in the highly metastatic KPCY-p120ctn−/− cells (P = 5.49e–19; Fig. 2I). Taken together, these findings show that PTHrP is highly expressed in the most aggressive and metastatic subsets of human and mouse PDAC cells.
Pthlh Is Induced During Tumorigenesis, and Its Loss Retards Tumor Growth
To determine the stage in which PTHrP expression is induced during PDAC progression, we evaluated PTHrP IHC of pancreatic tissues from wild-type (WT), LSL-KrasG12D; Pdx1-Cre (KC) and LSL-KrasG12D; p53R172H; Pdx1-Cre (KPC) mice. WT mice exhibited very little PTHrP expression except in islet cells, consistent with previous work showing a role in beta cell proliferation (refs. 35, 36; Fig. 3A). Pancreatic intraepithelial neoplasia (PanIN) cells and PDAC cells were positive for PTHrP expression (Fig. 3A), consistent with a role for RAS/mitogen-activated protein kinase (MAPK)/ETS signaling as one driver of Pthlh expression. Quantification of PTHrP staining in pancreatic tissue with PanIN lesions (KC mice; 3–4 months old), nonmetastatic primary PDACs (KPC; 3–7 months old), and metastatic primary PDACs (KPC; 3–7 months old) showed a trend toward increasing PTHrP levels with disease progression (Supplementary Fig. S3A). Western blot analysis of lineage-labeled yellow fluorescent protein (YFP)–positive tumor cells isolated from KPC-Rosa26LSL-YFP (KPCY) pancreata demonstrated high PTHrP expression in the malignant epithelia relative to normal ductal cells isolated from WT (Pdx1-Cre; LSL-Rosa26YFP/YFP) mice (Fig. 3B). YFP/PTHrP coimmunofluorescence of metastatic liver lesions showed that YFP+ tumor cells maintained PTHrP expression (Fig. 3C). Additionally, analysis of the publicly available transcriptomic datasets GSE90824 (37) and GSE144561 (38) showed that metastatic tumors had higher PTHrP expression relative to primary tumors (Supplementary Fig. S3B) and that PTHrP expression was highest in circulating tumor cells from metastatic PDAC mice (Supplementary Fig. S3C).
Primary PDAC burden was reduced significantly following pancreatic orthotopic injection with KPCY-shPthlh knockdown cell lines (Fig. 3D and E; knockdown was confirmed in Supplementary Fig. S3D and S3E). KPCY-shPthlh showed a statistically significant, albeit modest, decrease in proliferation in vitro (Supplementary Fig. S3F) in concordance with the in vivo phenotype. Fluorescence-activated cell sorting (FACS) analysis of livers from mice with shPthlh primary tumors showed a marked reduction in YFP+ tumor cells (Fig. 3F), consistent with a reduced capacity for metastatic colonization upon Pthlh loss. To rule out the possibility that KPCY-shPthlh tumors are less metastatic due to their smaller primary tumors, we repeated the orthotopic injection of KPCY-shPthlh tumor cells and let them grow until they reached a primary tumor burden similar to shNT (5–7 weeks in shPthlh vs. 3 weeks in shNT; see Supplementary Fig. S3G). Strikingly, these long-term KPCY-shPthlh orthotopic tumors continued to show no evidence of metastasis (Supplementary Fig. S3H). Reduced primary tumor and metastatic burden were observed also upon CRISPR-mediated knockout of Pthlh in an independent KPC-derived tumor cell line (Supplementary Fig. S3I–S3K). These loss-of-function experiments suggest a potential oncogenic role for PTHrP. To establish further the proproliferative function of PTHrP, we used lentiviral-mediated Pthlh overexpression plasmids in either WT murine three-dimensional (3-D) organoids (which do not express Pthlh) or KPCY tumor cells (enhancing the endogenous expression of Pthlh). Reverse-transcription polymerase chain reaction (RT-PCR) confirmed Pthlh overexpression, which leads to enhanced proliferation in both WT 3-D organoids and KPCY tumor cells (Supplementary Fig. S3L–S3O).
Pthlh Deletion Reduces Tumor Burden, Abolishes Metastasis, and Extends Survival in KPC Mice
The encouraging reduction of primary and metastatic tumor burden observed in Pthlh knockdown and knockout experiments in pancreatic orthotopic transplant models suggested that PTHrP may have an oncogenic and prometastatic role in PDAC. To test this hypothesis in an autochthonous model, we bred the PthlhLoxP conditional deletion allele into KPCY mice (Fig. 4A). Of note, we were not able to combine homozygous PthlhLoxP/LoxP alleles with the LSL-KrasG12D knock-in allele, likely due to the <2.5-Mb genomic distance that separates the endogenous Kras and Pthlh loci on mouse chromosome 6. More than 200 pups were genotyped, and no LSL-KrasG12D; PthlhLoxP/LoxP animals were obtained; therefore, we proceeded with LSL-KrasG12D; p53R172H; Pdx1-Cre; LSL-Rosa26YFP/YFP; PthlhLoxP/wt (designated as KPCY-PthlhHET) mice with heterozygous Pthlh knockout for further analysis. To validate the fidelity of our model, we confirmed a reduction in PTHrP expression in the ductal epithelia of KPCY-PthlhHET PDAC tumors relative to KPCY mice (Fig. 4B). We performed FACS of YFP+ tumor cells from KPCY-PthlhHET and KPCY pancreata followed by RT-PCR to confirm Pthlh deletion in vivo (Supplementary Fig. S4A). In parallel experiments, we generated Pdx1-Cre; LSL-Rosa26YFP/YFP; PthlhLoxP/wt (CY-PthlhHET) animals and performed histopathologic analysis of the tissues, demonstrating that the pancreas developed normally after Pthlh heterozygous deletion (Supplementary Fig. S4B), with normal acinar, ductal, and endocrine compartments. Furthermore, we generated LSL-KrasG12D; Pdx1-Cre; LSL-Rosa26YFP/YFP; PthlhLoxP/wt (KCY-PthlhHET) mice and harvested pancreatic tissues at 3 months of age, which showed similar levels of acinar-to-ductal metaplasia (ADM) and PanIN lesions relative to KCY control animals (Supplementary Fig. S4C), indicating no difference in tumor initiation.
We observed a dramatic increase in overall survival of KPCY-PthlhHET mice (P = 3.0e-4), whose median survival was 192 days compared with 111 days in KPCY mice (Fig. 4C). This survival advantage of 81 days (representing a 73% increase) is one of the strongest survival benefits observed in the aggressive KPC model to our knowledge. These results are even more impressive taking into consideration that we utilized an ∼50% reduction in Pthlh expression by using KPCY-PthlhHET mice. More surprisingly, despite KPCY-PthlhHET mice being 3 months older, only one of the 16 KPCY-PthlhHET mice (6.25%) showed gross metastasis by YFP analysis under a dissection microscope (Fig. 4D). This is in stark contrast to the nearly 45% of KPCY mice with gross YFP+ metastases (Fig. 4D). KPCY-PthlhHET mice likely lived longer due to the combined effects of having significantly decreased primary tumor burden (Fig. 4E) and decreased metastases to the liver and lungs (Fig. 4F and G). Representative YFP images of gross pancreata, liver, and lung demonstrate the reduced primary tumor and metastatic burden observed in KPCY-PthlhHET mice relative to controls (Fig. 4H). It is possible that KPCY-PthlhHET mice may harbor fewer metastases due to their decreased primary tumor load. To address this possibility, we reexamined the metastatic burden in an independent cohort of KPCY control animals with a primary tumor load matched to our KPCY-PthlhHET cohort (i.e., a restricted analysis of our historical control KPCY tumors to include only mice with tumor burden similar to that of KPCY-PthlhHET mice). This new analysis demonstrated that KPCY-PthlhHET mice continued to show reduced metastasis compared to KPCY mice with size-matched primary tumor load (Supplementary Fig. S4D).
Anti-PTHrP Monoclonal Neutralizing Antibody Reduces Primary Tumor Load and Metastatic Outgrowth
We next sought to test the therapeutic efficacy of inhibiting PTHrP in vivo. We engrafted KPCY tumor cells by orthotopic injection into the pancreas of C57BL/6 mice, allowed them to establish for 7 days, and then began treatment with either anti-PTHrP monoclonal neutralizing antibody or control anti-immunoglobulin G (IgG) antibody. After 3 weeks of treatment, anti-PTHrP–treated tumors were significantly smaller (Fig. 5A and B). Their rate of tumor growth (Fig. 5C) and their endpoint masses (Fig. 5D) were significantly less. None of the anti-PTHrP–treated mice showed gross YFP+ metastases in contrast to approximately one third of anti-IgG–treated mice with overt metastases, primarily to the liver (Fig. 5E). This experiment was repeated with a second independent KPCY cell line, which showed significantly reduced tumor growth (Supplementary Fig. S5A), smaller primary tumors (Supplementary Fig. S5B and S5C), and no metastases in anti-PTHrP–treated mice (Supplementary Fig. S5D). To test the ability of the anti-PTHrP antibody to block metastatic outgrowth, we performed tail vein injections of KPCY cells, allowed them to engraft for 5 days, and then began anti-PTHrP treatment. After 3 weeks of growth, anti-PTHrP–treated mice had reduced metastatic burden as compared to anti-IgG–treated controls (Fig. 5F and G). YFP staining and quantification of the percentage of lung tissue occupied by YFP+ tumor cells confirmed that anti-PTHrP treatment reduced metastatic outgrowth (Fig. 5G). This tail vein metastasis experiment was repeated with a second independent KPCY cell line, showing the same reduced metastatic growth (Supplementary Fig. S5E and S5F).
Pthlh Deletion or Pharmacologic Inhibition Decreases the Proliferative Capacity of Pancreatic Tumor Cells
To determine if the reduction in overall tumor burden upon decreased PTHrP expression was due to a cell- autonomous mechanism, we stained tissues from KPCY-PthlhHET mice with Ki-67, a marker for proliferation. We observed reduced Ki-67 expression in YFP+ tumor cells from KPCY-PthlhHET mice relative to KPCY mice (Fig. 6A and B). Additionally, delaminating YFP+ tumor cells in KPCY mice maintained their Ki-67 expression, which was not typical in KPCY-PthlhHET tissues (Fig. 6A). PTHrP is a secreted ligand that binds to the G-protein–coupled receptor PTH1R in either an autocrine or paracrine fashion, and we performed Western blot analysis to determine if PDAC tumor cells express PTH1R, confirming such a status (Fig. 6C). Stimulation of the receptor with recombinant PTHrP(1–34) enhanced growth (Supplementary Fig. S6A). The recombinant mid-region PTHrP(67–86) and C-terminal PTHrP(107–111) peptides, which do not signal through PTH1R receptor, did not enhance growth (Supplementary Fig. S6A), suggesting that PTHrP(1–34) signaling through PTH1R is the predominant, if not exclusive, pathway in promoting proliferation of KPCY cells. We next isolated conditioned media from shNT and shPthlh cell lines and performed media swapping experiments. To that end, shNT-conditioned media (containing secreted PTHrP) rescued the reduced growth phenotype of both shPthlh clones (Supplementary Fig. S6B). Conversely, treatment of shNT cells with shPthlh-conditioned media (lacking secreted PTHrP) reduced growth (Supplementary Fig. S6C). Finally, stimulation of shPthlh clones with PTHrP(1–34) enhanced proliferation to shNT levels (Supplementary Fig. S6D). Collectively, these data support the premise that an autocrine signaling network exists whereby pancreatic tumor cells secrete PTHrP that binds to PTH1R on adjacent tumor cells, thereby stimulating growth. We cannot exclude the possibility of paracrine signaling to nontumor cells, the topic of future investigation.
Treatment of a panel of KPCY cell lines with an anti-PTHrP antibody showed a dose-dependent growth reduction (Fig. 6D). Treatment with the same anti-PTHrP antibody over time appeared to block cellular proliferation (Fig. 6E). Anti-PTHrP treatment of established human PDAC cell lines showed similar growth-reducing capabilities (Fig. 6F). Finally, we harvested 3-D organoid cultures from patients with PDAC and treated them with anti-PTHrP antibody, which led to greatly reduced viability (Fig. 6G and H), even beyond that of established cell lines. These results support the conclusion that, in part, secretion of PTHrP by tumor cells stimulates proliferation of neighboring tumor cells and provides an explanation for the reduced tumor burden upon genetic reduction or pharmacologic inhibition of PTHrP in vivo.
Pthlh Loss Reduces EMT and Metastatic Competency
We next evaluated the differentiation status of primary PDAC tumors from KPCY-PthlhHET and age-matched KPCY control mice, which revealed that a significant portion of KPCY-PthlhHET mice harbored only PanIN lesions and no evidence of PDAC (Fig. 7A). In addition, the percentage of KPCY-PthlhHET mice that did progress to PDAC was greatly reduced relative to KPCY control mice in this age-matched cohort (Fig. 7A). Of the KPCY-PthlhHET mice that harbored PDAC, the percentage of animals with evidence of poorly differentiated tumor cells was reduced (Fig. 7B). Additionally, the percentage of the pancreata designated as poorly differentiated in these mice was reduced greatly, with <2% of the entire pancreas being classified as poorly differentiated in KPCY-PthlhHET mice (Fig. 7C). This reduction in the poorly differentiated status in KPCY-PthlhHET tissues is illustrated best by costaining for YFP with E-cadherin (ECAD). YFP+/ECAD− cells that have left their epithelial clusters and undergone EMT are abundant in KPCY tissues and absent from KPCY-PthlhHET tissues (Fig. 7D). Quantification of the delaminating cells undergoing EMT demonstrated a dramatic approximately 10-fold reduction in KPCY-PthlhHET (Fig. 7E).
RNA-seq and gene set enrichment analysis (GSEA) of KPCY-shPthlh tumor cells indicated that EMT and associated pathways were highly enriched in control cells relative to those without Pthlh (Fig. 7F). In the latter, the top genes lost upon Pthlh knockdown include Smad3/6/7, Tgfb1, Tgfbr1/2, Bmp1/2, Bmpr2, Spp1, and Pthlh itself (Fig. 7G; Supplementary Fig. S7A–S7C). This analysis suggests that cells with reduced PTHrP are putatively unable to undergo EMT. To test this, we treated cells with TGFβ, a potent EMT inducer. KPCY cells, which express high PTHrP levels, were able to efficiently undergo EMT as measured by ECAD loss by FACS analysis (Fig. 7H and I). Additionally, the downregulation of ECAD mRNA (Cdh1 expression) after TGFβ-induced EMT in KPCY cells was associated with a concomitant upregulation of Pthlh (Supplementary Fig. S7D). Furthermore, stimulation with TGFβ enhanced the already high levels of secreted PTHrP present in the media; note that this immunoradiometric assay (IRMA) detects only human PTHrP, thus necessitating the use of the squamous PDAC subtype cell line BxPC3 (Supplementary Fig. S7E). Remarkably, shPthlh cell lines were refractory to TGFβ-stimulated EMT (Fig. 7H and I). We additionally performed immunofluorescence for ECAD to show that control KPCY cells efficiently undergo EMT and completely lose their ECAD after TGFβ treatment, whereas shPthlh cells maintain their ECAD expression after TGFβ stimulation (Supplementary Fig. S7F). The reduction in ECAD in control KPCY cells coincides with a change to mesenchymal morphology, whereas shPthlh cells maintain their epithelial morphology after TGFβ treatment (Supplementary Fig. S7F). To determine if the lack of EMT observed in shPthlh cell lines is functionally important for metastasis, we performed a tail vein metastasis assay, which revealed that cells with reduced PTHrP had a markedly reduced capacity to form lung metastases (Fig. 7J and K). This was repeated in a second independent KPCY shPthlh knockdown cell line, and reduced lung metastatic burden was observed in a similar fashion (Supplementary Fig. S7G and S7H). Treatment with anti-PTHrP antibody in a transwell migration assay reduced the in vitro invasive capacity of KPCY tumor cells (Fig. 7L).
RNA-seq analysis suggested that one downstream mechanism governing PTHrP-driven metastases may be through the known metastasis regulating gene Spp1, which was the second highest expressed transcript in shNT cells relative to shPthlh cells (Supplementary Table S1). Western blot analysis revealed a decrease in SPP1 (osteopontin) protein levels in Pthlh knockdown cells (Supplementary Fig. S7I). To determine if the reduced metastasis phenotype observed in shPthlh cells could be restored, we performed lentiviral-mediated stable overexpression of Spp1 (Spp1 OE). Western blot analysis confirmed efficient overexpression in Spp1 OE cells relative to empty vector (EV; see Supplementary Fig. S7J). Tail vein injection of KPCY-shPthlh cells expressing EV or Spp1 OE partially rescued the metastatic ability in vivo (Supplementary Fig. S7K and S7L). Additionally, Spp1 OE enhanced the in vitro migratory capacity of KPCY-shPthlh tumor cells in a transwell migration assay (Supplementary Fig. S7M).
We have demonstrated that a subset of human patients with PDAC show PTHLH copy-number gains as part of the KRAS amplicon. Increased PTHLH mRNA expression correlates strongly with the squamous/quasi-mesenchymal/basal-like subtypes of PDAC, and PTHLH copy-number gains are enriched in metastatic patients. We functionally demonstrated that genetic Pthlh loss, in either genetically engineered KPCY mice or orthotopic injection models, reduces both primary and metastatic tumor burden, leading to enhanced overall survival in mice. Anti-PTHrP monoclonal antibody therapy reduces tumor cell proliferation and overall tumor burden. Finally, we have proposed a mechanism for reduced metastatic competency whereby tumor cells without PTHrP are unable to undergo EMT and thus cannot initiate the metastatic cascade (Supplementary Fig. S8).
In the context of tumor metastasis biology, PTHrP has been studied as a potent inducer of osteolytic bone metastases through a process termed the “vicious cycle.” The abundance of TGFβ in the bone microenvironment stimulates local PTHrP production by tumor cells, which activates PTH1R signaling on osteoblasts, beginning a process favoring the osteoclast lineage through the RANKL/RANK axis (reviewed in ref. 39). This paracrine signaling causes osteoclasts to degrade bone (more permissive for metastatic growth), thereby releasing more TGFβ superfamily members, which induces the tumor cell to produce even more PTHrP (40). In this regard, PTHrP is functioning in a manner similar to the highly related parathyroid hormone, a critical protein involved in bone homeostasis. This parathyroid hormone–like effect on bone metastasis is fundamentally different from the cell-autonomous mechanisms of metastasis we propose here, where PTHrP signaling is a necessary component of EMT within tumor cells, thereby driving invasion and dissemination to distant sites. We further demonstrate that PTHrP loss maintains cells primarily in an epithelial state, resulting in a nearly complete arrest of the macrometastatic process.
Provocatively, in KPCY-PthlhHET mice, survival was prolonged with nearly complete elimination of metastases in the otherwise highly aggressive KPCY autochthonous PDAC mouse model. Our work provides a foundation for the application of anti-PTHrP therapy in clinical trials of patients with PDAC. Our loss-of-function experiments in complementary mouse models and human 3-D organoids demonstrate that PTHrP deletion/inhibition results in less-aggressive disease, with reduced primary and metastatic tumor growth. Metastatic progression is a complex multistep process consisting of invasion through the basement membrane, intravasation into adjacent vasculature, survival in circulation, extravasation into the metastatic tissue, and successful colonization at the secondary site (41). Our results indicate that PTHrP inhibition inhibits cells from entering the metastatic cascade by hindering their ability to undergo EMT and invade. The inability of PTHrP-deficient cells to undergo EMT maintains them in organized epithelial clusters, thus preventing invasion. The striking change from predominantly poorly differentiated PDAC in KPCY mice to well-differentiated tumors in KPCY-PthlhHET mice holds promise for the ability of long-term anti-PTHrP treatment to alter differentiation status in patients with PDAC. Additionally, we have shown that PTHrP loss reduces cellular proliferation, which would provide a two-pronged attack for anti-PTHrP therapy—reduced invasion through inhibition of EMT and decreased proliferation in tumor cells that colonize the metastatic site.
Others have demonstrated that the metastatic cascade in pancreatic cancer is a highly choreographed interplay between proliferation and metastatic competency, modulated by the transcription factor Runx3 (42). PTHrP has been shown to be an upstream driver of other Runx family members, namely Runx1 and Runx2 (15, 43, 44), which may themselves induce EMT (15). Additionally, our differential gene expression analysis indicates that expression of many RUNX3-regulated transcripts (42) is also enhanced in PTHrP-expressing cells. Most notably, RUNX3 controls the metastatic switch to promote metastasis through osteopontin (encoded by Spp1) and Col6a1 (42), which we show are both highly enriched in Pthlh-intact cells relative to Pthlh knockdown cells (Supplementary Table S1). Spp1, in particular, is the second highest expressed transcript in control cells relative to Pthlh knockdown, with nearly complete loss in shPthlh cells. We demonstrate that PTHrP exerts its prometastatic effects, in part, by driving Spp1 expression, and that reexpression of Spp1 in cells lacking Pthlh increases their metastatic proclivity. Future studies will look to determine the full extent of cross-talk between PTHrP and RUNX signaling to find new therapeutically actionable nodes within the pathways that may be important for metastasis.
PTHrP, a marker of the squamous PDAC subtype, endows cells with the capacity for invasion and metastasis, which can be inhibited by an anti-PTHrP monoclonal antibody treatment. In a broader sense, we have uncovered a targeted therapeutic vulnerability that emerges from amplification of a “passenger” oncogene and offer a new paradigm to look at genes in close proximity to other commonly amplified oncogenes. PTHLH is co-amplified as part of the KRAS amplicon in patients with PDAC. This collateral amplification (i.e., amplification of a supposed passenger gene that ends up having its own oncogenic functions) is analogous to collateral passenger deletions of genes that are lost when established tumor suppressors and their surrounding genes are deleted (45–47). The collateral amplification confers a bona fide change in phenotype that can be attributed to the amplified “passenger” gene (PTHLH) in addition to KRAS. This genomic amplification of PTHLH does not preclude other methods of enhancing PTHrP transcription in the squamous and mesenchymal states, through factors such as TP63/ETS/SMAD. In fact, our observations indicate that the increase in PTHrP expression in PDAC is likely a complex interplay of multiple pathways. For example, it is tempting to speculate that a single patient with PDAC may possess multiple means of enhancing PTHrP levels. There may be PTHLH copy-number gains as part of the KRAS amplicon, the latter of which activates MAPK/ETS signaling, leading to further increased PTHrP expression. In addition, this amplicon correlates with the squamous subtype, thus activating a transcriptional program through TP63, which itself augments PTHrP expression further. On a cellular basis, individual cells undergoing EMT would have higher PTHLH expression. Collectively, our work nominates PTHrP as a therapeutic avenue in patients with PDAC that has the potential to limit both primary and metastatic tumor growth.
LSL-KrasG12D; Pdx1-Cre (KC) and LSL-KrasG12D; P53R172H or LoxP; Pdx1-Cre (KPCY) mice have been described previously (48). The PthlhLoxP mice were a generous gift from Dr. Andrew Karaplis (McGill University; ref. 49). For survival analysis, mice were palpated and examined for evidence of morbidity twice per week and sacrificed when moribund. Both male and female mice were used for survival analysis. The percentage of metastatic mice was determined based on gross YFP positivity in the liver, lungs, and other metastatic sites. Female C57BL/6J (stock no. 000664) or outbred athymic nude J:NU (stock no. 007850) for tumor cell injection experiments were obtained from The Jackson Laboratory. Mice treated with anti-PTHrP monoclonal antibody (200 μg; Richard Kremer Laboratory, McGill University Health Centre, and material transfer agreement with Biochrom Pharma, Inc.) received intraperitoneal injections three times per week every 48 hours day for the indicated time. All vertebrate animals were maintained and experiments were conducted in compliance with the NIH guidelines for animal research and approved by the University of Pennsylvania Institutional Animal Care and Use Committee.
Cell Lines and Culture Conditions
Murine PDAC cell lines were derived from KPCY mice that were backcrossed onto the C57BL/6J strain or KPC tumors of mixed genetic background (50). All human cell lines were originally obtained from the American Type Culture Collection. All cell lines were cultured in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin–streptomycin and maintained in standard culture conditions. Cell lines were regularly tested for Mycoplasma. The following doses were used for in vitro treatment: U0126, 10 μmol/L; SCH772983, 2 μmol/L; recombinant PTHrP(1–34), 100 nmol/L; recombinant PTHrP(67–86), 100 nmol/L; recombinant PTHrP(107–111), 100 nmol/L; and TGFβ, 10 ng/mL. For conditioned media experiments, complete DMEM media (10% FBS, 1% penicillin–streptomycin) were collected after 48 hours of culture and combined with fresh complete DMEM at a 1:1 ratio for subsequent treatments.
Immunohistochemistry and Immunofluorescence Staining
Dissected mouse pancreata were fixed in zinc formalin, processed, embedded in paraffin, sectioned, and mounted. IHC sections were stained using a Bond RX autostainer (Leica). Immunofluorescent sections were deparaffinized and rehydrated and antigen retrieval was performed, followed by overnight incubation with the specific primary antibodies at 4°C. Primary antibodies used were PTHrP (AV33885, Sigma-Aldrich), Ki-67 (ab16667, Abcam), ECAD (Clone M108, Takara), and green fluorescent protein (Abcam), which recognizes YFP. Images were acquired using an Olympus IX71 inverted multicolor fluorescent microscope equipped with a DP71 camera, and quantification was performed using ImageJ software.
All pathologic analysis was performed by the lab of Anirban Maitra (The University of Texas MD Anderson Cancer Center) in accordance with the consensus report and recommendations for pathologic analysis of genetically engineered mouse models of pancreatic exocrine cancer (51). For tissue grading, the percentages of normal pancreas tissue, ADM, PanIN, well-differentiated carcinoma, poorly differentiated carcinoma, sarcomoid carcinoma, and necrosis were analyzed on every slide in a blinded fashion. For ADM and PanIN quantification, the size of pancreas tissue was measured and the total numbers of PanINs and ADMs were counted. For PTHrP IHC H score analysis, 0 indicates negative staining; 1, weak staining; 2, moderate staining; and 3, strong staining. The H score was calculated as [3 × (percentage of strong staining)] + [2 × (percentage of moderate staining)] + [1 × (percentage of weak staining)], giving a range of 0 to 300.
Human PDAC Specimens
Human 3-D organoid samples were obtained from patients at the University of Pennsylvania and Columbia University under institutional review board approval. The PDAC tissue microarrays were obtained from patients who underwent radical pancreatectomy at the Department of General Surgery, Chiba University Hospital, Japan, from June 2013 to December 2015. All patients were histologically diagnosed with primary invasive PDAC. The study protocol was approved by the Ethics Committees of Chiba University, Graduate School of Medicine, and written informed consent was obtained from each patient before surgery. Formalin-fixed and paraffin-embedded samples were stained with anti-PTHrP antibody (AV33885, Sigma-Aldrich) using standard immunohistochemistry techniques. The staining intensities of PTHrP expression were evaluated independently by two investigators and scored as follows: low expression, 0% to 50% tumor cells stained positive; high expression, more than 50% tumor cells stained positive.
Cells were treated with colcemid for 60 minutes prior to harvest to arrest cells at metaphase. Trypsinized and pelleted cells were resuspended in a 75-mmol/L KCl hypotonic solution and incubated at 37°C for 20 minutes. Cells were fixed three times in a 3:1 ice-cold methanol:acetic acid fixative solution at −20°C for 5 minutes. After the final fixative solution wash, cell suspensions were dropped onto slides from a height of 10 cm. Slides were dehydrated through 70%/85%/100% ethanol washes and air dried. While the slides dried, the probe mixture was prepared by mixing a 1:5 dilution of probe to FISH probe buffer (PTHLH-CHR12-20, Empire Genomics). The probe mixture was applied to the air-dried slides, and the coverslips were placed on top prior to denaturing the slides on a slide warmer at 73°C for 2 minutes. The slides were next placed in a humidified chamber and incubated overnight at 37°C. After incubation, the coverslips were removed, and the slides were washed in Wash Solution 1 (WS1; Empire Genomics) wash buffer at 73°C for 2 minutes. Slides were transferred to Wash Solution 2 (WS2; Empire Genomics) at room temperature for 2 minutes. After washes, the slides were removed, cleaned, and allowed to air dry. The slides were stained with Vectashield Antifade Mounting Medium with 4′,6-diamidino-2-phenylindole (DAPI; H-1200-100, Vector Laboratories), coverslipped, and stored for imaging.
Western Blot Analysis
Cells were washed with ice-cold phosphate-buffered saline (PBS) and lysed in radioimmunoprecipitation assay buffer. Equal amounts of protein were run in reducing conditions on sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) gels and transferred to Immoblion-FL polyvinylidene fluoride membrane (IPFL00010, EMD Millipore). Blocking was performed in Odyssey Blocking Buffer (Li-Cor) for 1 hour at room temperature. After blocking, membranes were incubated in primary antibody diluted in Odyssey Blocking Buffer overnight at 4°C. After PBS-T washes, membranes were incubated with secondary antibody diluted in Odyssey Blocking Buffer at room temperature in the dark for 1 hour. Following PBS-T washes, membranes were imaged on a Li-Cor Odyssey Imaging System.
shRNA and CRISPR
For shRNA experiments, cells were transduced with Mission pLKO.1-puro Lentiviral particles with non-target (shNT; 5′ CCGGGATACCTAACTCAGGAAACCACTCGAGTGGTTTCCTGAGTTAGGTATCTTTTTG 3′), shPthlh #1 (5′ CCGGGATACCTAACTCAGGAAACCACTCGAGTGGTTTCCTGAGTTAGGTATCTTTTTTG 3′), or shPthlh #2 (5′ CCGGCCAATTATTCCTGTCACTGTTCTCGAGAACAGTGACAGGAATAATTGGTTTTTTG 3′) and stable colonies selected with puromycin (Sigma-Aldrich). For CRISPR experiments, cells were transduced with pLentiCRISPR V2 lentiviral particles with Pthlh guide RNA (5′ GCCCCTCCACCGAACGCCCG 3′) and stable colonies selected with puromycin (Sigma-Aldrich).
For Pthlh overexpression, cells were transduced with pLenti-C-Myc-DDK-P2A-Puro Lentiviral Particles with EV (PS100092, OriGene) or Pthlh ORF (MR201519L3, OriGene). For Spp1 overexpression, cells were transduced with pCDH-EF1-FHC Lentiviral Particles with empty vector (64874, Addgene) or Spp1 ORF cloned into the NheI-BamHI sites.
Conditioned media were harvested from BxPC3 cells 48 and 72 hours after TGFβ treatment (10 ng/mL) and frozen for downstream PTHrP 2-site IRMA analysis. Active PTHrP IRMA (DSL8100, Beckman Coulter) was performed as per the manufacturer's instructions. Briefly, 200 μL of calibrator or sample was added to antibody-coated tubes, followed by 100 μL of iodide-125 tracer and gentle vortexing. Samples were incubated overnight at room temperature on a shaker and then discarded. Tubes were washed three times with deionized water, and signal was measured using a PerkinElmer 2470 Wizard automatic gamma counter. PTHrP concentration (pg/mL) was calculated based off of a standard curve.
Orthotopic injection was performed as previously described (52). Briefly, after anesthesia and sterile preparation of the abdomen, an incision was made in the upper left quadrant, and the pancreas was exteriorized on a sterile field. Then, 1.0 to 2.5 × 105 cells in 100 μL sterile DMEM were injected into the tail of the pancreas via an insulin syringe. KPCY-derived tumor cells were injected into C57BL/6J mice (000664, The Jackson Laboratory) and mixed-background KPC-derived tumor cells into athymic nude J:NU mice (002019, The Jackson Laboratory). A cotton swab was held over the injection site to ensure that no cells leaked into the peritoneal cavity. Successful injection was verified by the appearance of a liquid bleb at the injection site. Afterward, the pancreas was placed back into the peritoneum and the incision was closed with 4-0 coated sutures. Tumors, lungs, and livers were harvested at the indicated time points.
Tail Vein Injection Metastasis Assay
An insulin syringe was used to inject 1.0 to 2.5 × 105 KPCY cells in 100 μL sterile DMEM into the tail veins of C57BL/6J mice (000664, The Jackson Laboratory). Lungs were harvested at the indicated time points. Tail vein injected lungs were stained for YFP or DAPI and imaged on a Keyence microscope, and stitched images were then generated to quantify the percentage of YFP+ area relative to the entire lung.
For ECAD flow cytometry, cells were dissociated into single cells with Hank's Enzyme Free Cell Dissociation Solution (S-004-C, EMD Millipore). Cells were stained with anti-ECAD (147308, BioLegend) or isotype control (400418, BioLegend) in staining solution on ice for 15 minutes. Cells were washed three times in staining solution, filtered through a 70-μm strainer, and then stained with DAPI prior to flow cytometric analysis. This assay measured only surface ECAD levels, as we did not permeabilize the cells.
For FACS of YFP+ cells, tumors were washed, minced, and digested with Collagenase Type V (2 mg/mL; C9263, Sigma-Aldrich) for 20 to 30 minutes at 37°C and vortexed every 5 minutes. Digested tissue was filtered through a 70-μm filter into a conical tube, and complete DMEM (10% FBS, 1% penicillin–streptomycin) was added. Cells were pelleted at 300 g for 5 minutes and resuspended in cold sorting buffer (1× Hank's Balanced Salt Solution with 25 mmol/L HEPES, 5 mmol/L MgCl2, 17.5 mmol/L d-glucose, 1× glutamax, 1 mmol/L sodium pyruvate, 25 μg/mL DNase) with DAPI. YFP+ cells were collected via FACS, and RNA was isolated for downstream analysis.
Cell Viability Assays
The WST-1 Cell Proliferation Assay (ab65475, Abcam) was performed per the manufacturer's instructions. Briefly, 2.0 × 103 to 5.0 × 103 cells were plated on 96-well plates and grown in standard culture conditions. Media were replaced with 100 μL fresh media containing 10 μL WST solution and incubated for 2 hours at 37°C. The plate was shaken, and absorbance was read at 420 to 480 nm and with a reference wavelength of 650 nm. Final absorbance was obtained by subtracting the 650-nm reference wavelength from the 420- to 480-nm read. The CellTiter-Glo 3D Cell Viability Assay (G9681, Promega) was performed per the manufacturer's instructions. Briefly, a 1:1 dilution of 3-D organoid culture media to CellTiter-Glo 3D reagent was added to organoid cultures and incubated for 30 minutes at 37°C, after which 100 μL of the 1:1 mixture was added to a 96-well plate, and luminescence was read in triplicate.
Human pancreatic cancer patient–derived 3-D organoid cultures were grown as previously described (53). Corning Matrigel-embedded organoids were maintained in culture media containing A 83-01 (2939, Tocris), B27 Supplement (17504044, Thermo Fisher Scientific), hEGF (AF-100-15, PeproTech), hFGF-10 (100-26, PeproTech), hGastrin (3006, Tocris), mNoggin (250-38, PeproTech), N-acetylcysteine (A9165-5G, Sigma-Aldrich), nicotinamide (N0636-100G, Sigma-Aldrich), Y-27632 dihydrochloride (Y0503-5 mg, Sigma-Aldrich), R-Spondin1–conditioned medium, and Wnt3a-conditioned medium.
Matrigel Invasion Assay
Transwell inserts (3464, Corning) were coated with Matrigel and placed in 24-well plates above complete DMEM. Cells were seeded with serum-free DMEM on top of the Matrigel layer. Cells were incubated under normal conditions for 24 hours in the presence of hydroxyurea. Transwell inserts were removed, washed three times, and fixed in 4% paraformaldehyde for 15 minutes. Fixed Matrigel inserts were washed three times, stained with DAPI, and imaged.
RNA-seq and GSEA
RNA samples were extracted using the QIAGEN RNeasy kit following the manufacturer's instructions. RNA was sent out to a commercial company, Novogene, for library preparation and high-throughput sequencing using Illumina sequencers to generate paired-end results. Raw counts of gene transcripts were obtained using an alignment-independent tool, Salmon (https://combine-lab.github.io/salmon/), using standard settings. The raw count matrix was subsequently imported into Rstudio (R 3.5) and used as input file for DESeq2 analysis (https://bioconductor.org/packages/release/bioc/html/DESeq2.html) with default settings from online software instruction for normalization and differential gene expression analysis. Salmon was used to normalize and quantitate gene expression in transcripts per million through quasi-alignment. Differentially expressed genes were used as input for principal component analysis, GSEA (https://www.gsea-msigdb.org/gsea/index.jsp), and motif analysis using HOMER (http://homer.ucsd.edu/homer/ngs/index.html). Detailed scripts and parameters used for each step of analysis can be provided by reasonable request to the authors.
Data and Code Availability
All sequencing data have been deposited in the Gene Expression Omnibus under the series GSE154661.
A. Maitra reports other support from Cosmos Wisdom Biotechnology and Thrive Earlier Detection outside the submitted work. R. Kremer reports grants from Canadian Institutes of Health Research and U.S. Department of Defense during the conduct of the study; nonfinancial support from Biochrom Pharma outside the submitted work; and the following patents: US7897139 (PTHrP, its isoforms and antagonist thereto in the diagnosis and treatment of disease), US8501929B2 (PTHrP, its isoforms and antagonist thereto in the diagnosis and treatment of disease), US9057055 (PTHrP, its isoforms and antagonist thereto in the diagnosis and treatment of disease), US959153 (A method of inhibiting cancer in mammals by administering an antibody to PTHrP), IP 08783388.5 and IP 1466/2190990 (PTHrP, its isoforms and antagonist thereto in the diagnosis and treatment of disease), and IP 200880112153.8 (PTHrP, its isoforms and antagonist thereto in the diagnosis and treatment of disease). B.Z. Stanger reports grants from National Cancer Institute–National Institutes of Health and Boehringer Ingelheim and personal fees from iTeos Therapeutics outside the submitted work. No potential conflicts of interest were disclosed by the other authors.
J.R. Pitarresi: Conceptualization, data curation, formal analysis, funding acquisition, investigation, writing–original draft, writing–review and editing. R.J. Norgard: Data curation, formal analysis, writing–review and editing. A.M. Chiarella: Conceptualization, resources, data curation, supervision, funding acquisition, project administration, writing–review and editing. K. Suzuki: Data curation, writing–review and editing. B. Bakir: Data curation. V. Sahu: Data curation, writing–review and editing. J. Li: Data curation, visualization, writing–review and editing. J. Zhao: Data curation. B. Marchand: Data curation, visualization, writing–review and editing. M.D. Wengyn: Data curation, writing–review and editing. A. Hsieh: Data curation, writing–review and editing. I.-K. Kim: Data curation. A. Zhang: Data curation. K. Sellin: Data curation. V. Lee: Data curation. S. Takano: Data curation. Y. Miyahara: Data curation. M. Ohtsuka: Data curation. A. Maitra: Data curation, formal analysis. F. Notta: Data curation, formal analysis, supervision. R. Kremer: Resources, data curation, formal analysis, supervision, funding acquisition, investigation, writing–review and editing. B.Z. Stanger: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, investigation, methodology, writing–original draft, project administration, writing–review and editing. A.K. Rustgi: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, investigation, methodology, writing–original draft, project administration, writing–review and editing.
This work was supported by an American Gastroenterology Association Bern Schwartz Research Scholar Award in Pancreatic Cancer (J.R. Pitarresi); National Cancer Institute (NCI)–National Institutes of Health (NIH) National Research Service Award (F32CA221094 to J.R. Pitarresi); NIH Loan Repayment Program (J.R. Pitarresi); Hopper-Belmont Foundation Inspiration Award (J.R. Pitarresi); NCI–NIH grant (P30CA01369644 to A.K. Rustgi); National Institute of Diabetes and Digestive and Kidney Diseases–NIH grant (R01DK060694 to A.K. Rustgi); NCI–NIH grant (R01CA229803 to B.Z. Stanger); U.S. Department of Defense contract (W81XWH-15-1-0723 to R. Kremer); Canadian Institutes of Health Research grant (MOP142287); and American Cancer Society–Fairfield County Comedy Against Cancer Postdoctoral Fellowship (PF-19-227-01-CMS to A.M. Chiarella). The authors thank the following core facilities through the Herbert Irving Comprehensive Cancer Center: Molecular Pathology, Genetically Engineered Mouse Models, and Flow Cytometry. They thank Iok In Christine Chio (Columbia University) for the human PDAC organoids, Karine Sellin (McGill University Health Centre) for the production of anti-PTHrP monoclonal antibodies, and Mary Ann Crissey (University of Pennsylvania) for mouse colony maintenance.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.