Inhibition of phospholipase D1 reduces pancreatic carcinogenesis in mice partly through a FAK-dependent mechanism

Abstract

Phospholipase D (PLD) plays a critical role in cancer progression. However, its role in pancreatic cancer remains unclear. Thus, we evaluated the role of PLD1, one of two classical isoforms of PLD, in pancreatic carcinogenesis in vivo. The role of PLD1 in tumor growth was evaluated by subcutaneously transplanting human MIA PaCa-2 cells expressing endogenous PLD1 levels (Ctr KD cells) or cells in which PLD1 was knocked down (Pld1 KD cells) into immunodeficient mice. Twenty days post-implantation, tumors that arose from Pld1-KD cells were significantly smaller, compared to controls (Ctr KD). Then, we assessed the role of PLD1 in the tumor microenvironment, by subcutaneously implanting mouse LSL-Kras^G12D/+;Trp53^R172H/+;Pdx-1-Cre (KPC) cells into wild-type (WT) or PLD1 knockout (Pld1^−/−) mice. Compared to WT, tumor growth was attenuated in Pld1^−/− mice by 39%, whereas treatment of Pld1^−/− mice with gemcitabine reduced tumor growth by 79%. When PLD1 was ablated in LSL-Kras^G12D;Ptf1^Cre/+ (KC) mice, no reduction in acinar cell loss was observed, compared to KC mice. Finally, treatment of KC mice with a small molecule inhibitor of PLD1 and PLD2 (FIPI) significantly reduced acinar cell loss and cell proliferation, compared to vehicle-treated mice. Mechanistically, the effect of PLD on tumor growth is mediated, partly, by the FAK pathway. In conclusion, while PLD1 is a critical regulator of pancreatic xenograft and allograft growth, playing an important role at the tumor and at the microenvironment levels, inhibition of PLD1 and PLD2 are necessary to reduce pancreatic carcinogenesis in KC mice, and might represent a novel therapeutic target.

summary:

Genetic and pharmacological inhibition of phospholipase D reduces pancreatic carcinogenesis in mice through a FAK-dependent mechanism.

Introduction

Pancreatic ductal adenocarcinoma (PDA), a complex and lethal cancer with a five-year survival of approximately 13% in the United States, is on track to becoming the second leading cause of cancer-related deaths by 2040 (1). Due to the lack of effective means for an early diagnosis, most patients diagnosed with PDA have locally advanced or metastatic disease and are not candidates for surgical resection. The current standard treatment of care for PDA patients includes clinical use of chemotherapeutic agents, gemcitabine, nab-paclitaxel, and/or FOLFIRINOX. These chemotherapies, however, provide limited benefit in extending patients’ lives and long-term use increases risk of resistance (2,3). The high mortality and limited treatment options create a need to identify new therapeutic targets and strategies to overcome this deadly disease.

Phospholipase D (PLD), a lipid-signaling enzyme, catalyzes the conversion of the membrane phospholipid phosphatidylcholine to choline and phosphatidic acid (PA) (4). PLD functions to maintain structural integrity of intracellular membranes and promotes signal transduction pathways through protein-protein interactions and its catalytic product, PA. Elevation of either PLD1 or PLD2 (two mammalian isoforms of PLD) has been shown to contribute to cancer progression (5–7). The isoform PLD1 is increased in abundance or activity in various human cancers and is linked with proliferative signaling and resistance to cell death. In preclinical models of lung cancer and melanoma, PLD1, but not PLD2, promotes tumor growth and metastasis through roles in the tumor microenvironment (8). PLD1 has been shown to be elevated in clinical samples of PDA and to positively correlate with vascular invasion and poor prognosis (9). However, experimental investigation of the functional role of PLD and associated mechanisms in pancreatic cancer has not yet been pursued.

PLD-generated PA stimulates downstream signaling targets such as MAPK, mTOR, AKT, and focal adhesion kinase (FAK), which in turn modulate cancer processes including cell proliferation, migration, invasion, chemotaxis, apoptosis, and survival (7,10,11). PA can directly associate with AKT and mTOR initiating its activation, and overexpression of both PLD isoforms increased activation of ERK (12,13). Elevated FAK expression or activation enhances tumor malignancy and correlates with poor prognosis in human cancers. PLD activity enhances FAK phosphorylation whereas PLD inhibition impedes adhesion, invasion, and tumor formation (14–18). Furthermore, activation of these signaling pathways induce the expression of pro-survival and anti-apoptotic proteins such as Bcl-2, Bcl-xL, and STAT3 (13,19,20). Thus, PLD as well as MAPK, mTOR, AKT, and FAK pathways are attractive targets in PDA (5,6,12).

In this study, we evaluated the role and mechanism of PLD1 in pancreatic carcinogenesis using genetic and pharmacological approaches in multiple preclinical models. We observed that genetic ablation of PLD1 reduces tumor growth in pancreatic allografts and extends survival in murine models of pancreatic cancer by reducing FAK activation. Moreover, pharmacological inhibition of PLD1 and PLD2 reduced pancreatic carcinogenesis by reducing cell proliferation, through the regulation of FAK signaling in immunocompetent LSL-Kras^G12D;Ptf1^Cre/+(KC) mice. These results indicate that PLD plays an important role in pancreatic carcinogenesis.

Materials and Methods

Cells and cell culture.

Human MIA PaCa-2 were obtained from the ATCC (American Type Culture Collection, Manassas, VA). ATCC characterizes the cell lines using cytogenetic analysis. The cell lines were grown as monolayers in the medium recommended by ATCC and supplemented with 10% (v/v) fetal bovine serum, penicillin (50U/mL) and streptomycin (50 μg/ml), and frozen for future use. Mouse LSL-Kras^G12D/+;Trp53^R172H/+;Pdx-1-Cre (KPC) tumor derived cells were a gift from Dr. David Tuveson, Cold Spring Harbor Laboratory (Cold Spring Harbor, NY). We have not authenticated these cell lines, however all cell lines were characterized by cell morphology and growth rate and passaged in our laboratory less than 6 months after being received. Cells were maintained at 37°C in an incubator with a humidified atmosphere of 5% (v/v) CO₂.

Lentiviral production of PLD1 silenced MIA PaCa-2 cell line.

Stable transfection was performed by lentiviral infection, following the manufacturer’s instructions. Briefly, the lentiviral lysate was transfected in the presence of 8 μg/mL polybrene into the MIA PaCa-2 cancer cells. Transfected cells were selected by incubating them for 2 weeks in growth media containing 0.5 mg/mL neomycin (G418).

Animal experiments.

Six-week-old female T-lymphocyte deficient BALB/c nude mice were purchased from Charles River Laboratories (RRID:MGI:2683685, Wilmington, MA). Mice were given a week to acclimate to the animal facility before being enrolled in the experiments. PLD1-deficient mice (Pld1^−/−) mice were generated in C57BL/6 mice (RRID:IMSR_JAX:000664) as previously described (8). LSL-Kras^G12D; Ptf1^Cre/+(KC) mice in the C57BL/6 background were bred following established procedures described by Hingorani and colleagues (21). KC;Pld1^−/− mice were generated by crossing LSL-Kras^G12D;Ptf1^Cre/+ mice into Pld1^−/− mice generating KC;Pld1^+/− and then crossed again to generate KC;Pld1^−/− mice. KC and KC;Pld1^−/− were bred at the Stony Brook University Animal Facility. All mouse experiments were performed using the protocols approved by the Institutional Animal Care and Use Committees at Stony Brook University and at the University of California, Davis. Rodents were housed in a room with controlled temperature (22–24°C) and humidity (40–60%), maintained on a 12 h light-dark cycle, and fed an ad libitum chow diet (LabDiet 5001, LabDiet, Saint Louis, MO, USA) until grouped for study. Euthanasia was performed by carbon dioxide asphyxiation.

Subcutaneous xenotransplants of BALB/c mice and pancreatic cancer models.

MIA PaCa-2 or MIA PaCa-2 Pld1^−/− cells (1.5 x 10⁶ cells per site) were suspended in 100 μL sterile phosphate-buffered saline (PBS) and subcutaneously injected, bilaterally, in female immune-deficient BALB/c nude mice (six weeks of age) (n= 8 per group). Body weight was determined weekly and tumor size twice weekly. Tumor size was calculated by the formula: [length × width × (length + width/2) × 0.56] in cubic millimeter. Tumors were harvested 31 days after implantation, weighed and stored in formalin or liquid nitrogen for further analysis.

Subcutaneous allografts of PLD1 deficient mice and pancreatic cancer models.

Mouse KPC-tumor derived cells (1 x 10⁶ in 100 μL PBS) were injected bilaterally, subcutaneously in, either, C57BL/6 wild type or Pld1^−/− mice (n = 7-8 per group). Gemcitabine (GEM) (>99% 2′-Deoxy-2′,2′-difluorocytidine; dFdC; Gemzar; LY-188011) from Fisher Scientific (Hampton, NH) was administered in wild type mice at a dose of 100 mg/kg by intraperitoneal injection two times per week. To determine the growth inhibitory effect of GEM in Pld1^−/− mice, mouse KPC-tumor derived cells (1 x 10⁶ in 100 μL PBS) were injected bilaterally, subcutaneously in, either, C57BL/6 wild type or Pld1^−/− mice (n = 7-8 per group). GEM was administered in Pld1^−/− mice at a dose of 100 mg/kg by intraperitoneal injection two times per week. Body weight was determined weekly and tumor size twice weekly. Tumor size was calculated by the formula: [length × width × (length + width/2) × 0.56] in cubic millimeter. Tumors were harvested 14 or 16 days after implantation, weighed, and stored in formalin or liquid nitrogen for further analysis.

Orthotopic allograft of PLD1 deficient mice.

Mouse KPC-tumor derived cells, (3 x 10⁴ or 1 x 10⁵ in 30 μL PBS) were injected into the parenchyma of the pancreas of C57BL/6 wild-type or Pld1^−/− mice (n= 6 per group) with a 27-gauge hypodermic needle.

Genetic ablation of PLD1 in a pancreatic cancer model.

Male and female KC and KC;Pld1^−/− mice were housed and monitored with free access to food and water until eight months of age (n=7-10 per sex per group). The mice were euthanized, and pancreas harvested, weighed, and stored in liquid nitrogen or 10% paraformaldehyde.

Pharmacological inhibition of PLD.

Pharmacological inhibition of PLD with 5-Fluoro-2-indolyl des-chlorohalopemide (FIPI) in animals was performed as described previously (8). Briefly, seven-month old male LSL-Kras^G12D;Ptf1^Cre/+(KC) mice received intraperitoneal injections, twice daily, for 35 days of either vehicle (a solution of 4% DMSO/96% PBS) or FIPI (Cat# 13563; Cayman Chemical Company, Ann Arbor, MI) at a dose of 3 mg/kg body weight in 4% DMSO/96% PBS followed by a second 3 mg/kg body weight dose after 8 hours (n= 8-12 per group). 100 μM of FIPI is required for full inhibition (22) and has 18% bioavailability with a half-life of 5.5 hours (23). 3 mg/kg followed by a second dose of 3 mg/kg should yield approximately 21 hours of full inhibition subsequent to the second dose.

Tissue preparation and histology.

Mice were euthanized, and the pancreas and liver carefully resected. Tissues were immediately fixed with 10% paraformaldehyde for 24 hours. After dehydration, tissues were embedded in paraffin blocks, and five μm paraffin sections were processed routinely, stained with hematoxylin and eosin (H&E) and scored. Acinar cell loss, grade of precancerous lesion and the presence or absence of cancer were assessed on the H&E-stained slides by a pathologist blinded to pancreas sample identity. Scoring: Acinar cell loss was based on the percentage loss across the entire cross-section (24). Mononuclear inflammatory cell infiltration was based on percentage of cross-section containing >50 mononuclear inflammatory cells.

Immunohistochemistry.

IHC analyses were performed on mouse pancreas tissue fixed in 10% paraformaldehyde and washed with 70% ethanol. Five μm paraffin sections from fixed tissues were deparaffinized with xylenes and rehydrated. The paraffin sections were blocked for 1 hour with 1x animal-free serum blocker (Cat# SP-5030-250, Vector Laboratories, Newark, CA) followed by incubation with primary antibodies against PCNA (Cat# 13110, RRID:AB_2636979), p-ERK (Cat# 4370, RRID:AB_2315112), and Bcl-xL (Cat# 2764, RRID:AB_2228008) overnight at 4°C. Sections were washed and incubated with the appropriate biotinylated secondary antibody for 1 hour at room temperature, followed by 1 hour of incubation with streptavidin-peroxidase at room temperature (Cat# PK-4001, Vector Laboratories, Newark, CA). After additional washing, the sections were enhanced using DAB (Cat# SK-4100, Vector Laboratories, Newark, CA) then counterstained with hematoxylin. Finally, sections were dehydrated and fixed with Cytoseal mounting media (Cat# 23-244257, Thermo Scientific, Waltham, MA). Sections were digitally scanned using a microscope (Olympus BX51, Japan) to obtain images at 20x magnification. Scoring: Four to six fields per sample (at magnification 20x) were quantified as the percent positive area using ImageJ Fiji (RRID:SCR_003070).

Western blotting.

Pancreas and tumor samples were lysed with sucrose buffer containing protease and phosphatase inhibitor cocktails. Protein concentration of the samples were determined using the Bradford method. Equal amounts of protein of each sample were loaded in SDS-polyacrylamide gel electrophoresis gels and transferred onto polyvinylidene difluoride (PVDF) membranes. The western blots were performed using primary antibodies against p-FAK (Cat# 8556, RRID:AB_10891442), FAK (Cat# 13009, RRID:AB_2798086), p-AKT (Cat# 4060, RRID:AB_2315049), AKT Cat# 9272, RRID:AB_329827), p-ERK (Cat# 4370, RRID:AB_2315112), ERK (Cat# 4695, RRID:AB_390779), Osteopontin (Cat# AF808, RRID:AB_2194992), Cyclin D1 (Cat# 2922, RRID:AB_2228523), p-STAT3 (Cat# 9145, RRID:AB_2491009), STAT3 (Cat# 30835, RRID:AB_2798995), p-STAT5 (Cat# 9351, RRID:AB_2315225), STAT5 (Cat# 94205, RRID:AB_2737403), p-4E-BP1 (Cat# 2855, RRID:AB_560835), 4E-BP1 (Cat# 9452, RRID:AB_331692), p-p38 (Cat# 4511, RRID:AB_2139682), p38 (Cat# 8690, RRID:AB_10999090), PLD1 (Cat# 3832, RRID:AB_2172256), Bcl-xL (Cat# 2764, RRID:AB_2228008), Bcl-2 (Cat# 3498, RRID:AB_1903907), β-tubulin (Cat# 2128, RRID:AB_823664), β-actin (Cat# A5316, RRID:AB_476743) or Vinculin (Cat# 13901, RRID:AB_2728768). The blots were developed with secondary antibodies conjugated with HRP (Cat# 7074, RRID:AB_2099233) or biotinylated (Cat# 14708, RRID:AB_2798581) and streptavidin (Cat# 3999, RRID:AB_10830897) antibodies. The conjugates were visualized and quantified by chemiluminescence detection, and the densitometry analysis was performed using the ImageJ Fiji (RRID:SCR_003070).

RNA isolation and real-time quantitative polymerase chain reaction (RT-qPCR):

For PCR studies, RNA was isolated from RNAlater stabilization solution (Cat #AM7020; Invitrogen, Waltham, MA) preserved pancreas samples via RNeasy Mini Kit (Cat #74104; Qiagen, Redwood City, CA) following the manufacturers’ instructions. cDNA was produced for pancreas using high-capacity cDNA Reverse Transcriptase (Cat# 4368813; Applied Biosystems, Waltham, MA). mRNA relative expression of pancreas samples was assessed using primers synthesized by Invitrogen for IL-6 – Forward: CTGCAAGAGACTTCCATCCAG, Reverse: AGTGGTATAGACAGGTCTGTTGG. The housekeeping gene, glyceraldehyde-3-phosphate dehydrogenase, GAPDH, normalized the Ct values, and gene expression was determined using the 2^−ΔΔCt method (25).

Evaluation of liver-kidney function parameters and inflammatory markers.

Blood samples were collected via cardiac puncture and serum was isolated after centrifugation at 3000xg for 10 minutes at room temperature. Inflammation related biomarkers were assayed using V-PLEX Proinflammatory Panel 1 kit (Cat# K152A0H-1, Meso Scale Discovery, Rockville, MD). Clinical chemistry analysis to assess liver and kidney enzymes was performed by the UC Davis Comparative Pathology Laboratory.

Statistical analysis.

The data, obtained from at least three independent experiments, were expressed as the mean ± SD. Statistical evaluation was performed by unpaired t-test followed by post-hoc analysis using Bonferroni test for multiple comparisons to evaluate differences between two selected groups. One-factor or two-factor Analysis of variance (ANOVA) followed by Tukey’s multiple comparison test was conducted to evaluate differences among multiple groups. Analyses were performed by GraphPad Prism (RRID:SCR_002798). Two-sided p<0.05 was regarded as statistically significant. The level of significance is indicated in the figure legends.

Results

PLD1 downregulation in pancreatic cancer cells reduces tumor growth in vivo

To start exploring the role of PLD1 in regulation pancreatic tumor growth, human MIA PaCa-2 pancreatic cancer cells expressing endogenous PLD1 levels [Control knock-down (Ctr KD)] or PLD1 knock-down (Pld1 KD) MIA PaCa-2 cells (Figure 1A) were subcutaneously transplanted into immunodeficient nude mice and tumor growth monitored over time. Beginning on day 20 post-implantation, a significant reduction in tumor volume was observed in tumors that arose from Pld1-KD cells compared to the Ctr KD tumors (Figure 1B). While the rate of the Pld1 KD tumor growth appeared to increase after day 20, it is common with KD cells for variants no longer expressing the KD construct to be selected for when there is negative pressure. We do not know how much of the PLD1 knock-down was preserved at day 31. No significant difference in mouse body weight was observed throughout the experimental period (Figure 1C).

Figure 1: PLD1 downregulation in pancreatic cancer cells reduces tumor growth in vivo.

(A) Immunoblot of PLD1 in MIA PaCa-2 cells expressing endogenous PLD1 expression (Ctr KD) or PLD1 knock-down (Pld1 KD) MIA PaCa-2 cells. β-Tubulin is shown as a loading control. (B) Tumor volume progression over time for Ctr KD and Pld1 KD subcutaneous xenografts in female BALB/c nude mice. (C) Body weight at sacrifice, n= 6-10 per group. (D and E) Immunoblots of Bcl-xL, Bcl-2, p-ERK, ERK, p-AKT, AKT, p-STAT3, STAT3 and p-4E-BP1, and 4E-BP1 in total tumor homogenates collected from transplanted tumor xenografts, *p<0.05. n= 6-9 per group. Values are presented as mean ± SD.

To explore potential cellular mechanisms underlying the inhibition of tumor progression by genetic ablation of PLD1, we performed western blot analysis in total homogenates obtained from the Ctr KD and Pld1 KD tumors at day 31. Bcl-xL levels were reduced by 40% in Pld1 KD tumors compared to Ctr KD tumors with Bcl-2 levels showing a similar trend but not reaching significance (Figure 1D). Pld1 KD tumors showed reduced AKT phosphorylation compared to Ctr KD tumors. (Figure 1E). In contrast, significant differences in ERK, STAT3 or 4E-BP-1 phosphorylation were not observed between the groups.

Pancreatic tumor growth is impaired in mice lacking PLD1

Next, to characterize whether PLD1 plays a role in the microenvironment to promote pancreatic cancer growth, we subcutaneously implanted mouse pancreatic cancer cells derived from pancreatic tumors of LSL-Kras^G12D;Trp53^R172H;Pdx1-Cre (KPC) mice into C57BL/6 wild-type (WT) or C57BL/6 PLD1 knockout (Pld1^−/−) mice. Fourteen days after tumor implantation, the growth of KPC tumors was attenuated in Pld1^−/− mice by 33% (Figure 2A). Tumor weight at sacrifice was also significantly reduced in Pld1^−/− mice compared to WT mice (Figure 2B). Mechanistically, we explored the expression levels of various pathways related to tumor growth that can be activated by PLD1 (7,10,11). The levels of p-FAK and p-STAT3 were significantly reduced in tumors implanted in Pld1^−/− mice compared to those implanted in WT mice (Figure 2C). Moreover, the expression levels of Bcl-xL were also reduced in Pld1^−/− mice compared to WT mice (Figure 2D). In contrast, no significant differences were observed in osteopontin, cyclin D1, or in ERK, AKT, STAT5, p38, or 4E-BP1 phosphorylation between the groups (Figure 2D).

Figure 2: Pancreatic tumor growth is impaired in mice lacking PLD1.

(A) Tumor volume progression over time for C57BL/6 (WT) and Pld1^−/− mice subcutaneously transplanted with mouse KPC cells. (B) Tumor weight at sacrifice, *p<0.05. n= 6-10 per group. (C and D) Immunoblots for Bcl-xL, Cyclin D1, Osteopontin, p-ERK, ERK, p-STAT5, STAT5, p-p38, p38, p-4E-BP1, 4E-BP1, p-FAK, FAK, p-STAT3 and STAT3 in total tumor homogenates collected from transplanted WT and Pld1^−/− mice, *p<0.01. n= 6-9 per group. (E) Kaplan-Meier survival curve of WT (black) and Pld1^−/− mice (grey) injected with 30,000 KPC cells, n= 5-6 per group. (F) Kaplan-Meier survival curve of WT (black) and Pld1^−/− mice (grey) injected with 100,000 KPC cells, n= 9 per group. Values are presented as mean ± SD.

Then, to further assess whether PLD1 is important in tumor growth, we conducted a survival study implanting KPC cells orthotopically in the pancreas of WT and Pld1^−/− mice. Pld1^−/− mice orthotopically implanted with 30,000 KPC cells survived longer than WT mice (Figure 2E), however, this survival benefit was lost when a larger number of cells were implanted (100,000 cells) (Figure 2F).

We next compared the growth inhibitory effect observed in Pld1^−/− mice to that of gemcitabine (GEM), a current first-line chemotherapeutic drug used in pancreatic cancer patients (26). To evaluate these effects, we subcutaneously injected KPC cells in WT and Pld1^−/− mice. After 7 days post-implantation the mice were treated intraperitoneally with either PBS or with 100 mg/kg GEM two times per week. While tumor growth in Pld1^−/− mice was reduced by 36%, treatment with GEM had a stronger effect, reducing tumor growth by 55% in WT mice, without showing signs of toxicity (Figure 3A and B). Upon mechanistic analysis, Bcl-xL was similarly reduced in Pld1^−/− mice and WT mice treated with GEM when compared to vehicle-treated WT mice (Figure 3C). No significant differences were observed in osteopontin, cyclin D1, as well as in ERK, AKT, STAT5, p38, or 4E-BP1 phosphorylation between the groups (Figure 3C).

Figure 3: Pancreatic tumor growth inhibition of gemcitabine is comparable to that of mice lacking PLD1.

(A) Tumor volume progression over time of subcutaneous xenograft of mouse KPC cells in C57BL/6 (WT), Pld1^−/− mice, and C57BL/6 treated with 100mg/kg gemcitabine 2x/week (GEM). (B) Body weight at five and twelve days of treatment. (C) Immunoblots for p-ERK, ERK, p-STAT3, STAT3, p-AKT, AKT, p-4E-BP1, 4E-BP1, and Bcl-xL, *p<0.05, ***p<0.0005. n= 8 per group. Values are presented as mean ± SD. (D) Tumor volume progression over time of subcutaneous xenograft of mouse KPC cells in C57BL/6 (WT), Pld1^−/− mice, and Pld1^−/− mice treated with 100mg/kg GEM 2x/week. *p<0.05 vs Pld1^−/− mice.

To evaluate whether GEM could enhance the growth inhibitory effect observed in Pld1^−/− mice, we subcutaneously injected KPC cells in WT and Pld1^−/− mice. After 5 days post-implantation, a cohort of Pld1^−/− mice were treated intraperitoneally with, either, PBS or with 100 mg/kg GEM two times per week. Compared to WT mice, while tumor growth in Pld1^−/− mice was reduced by 39%, treatment of Pld1^−/− mice with GEM had a significantly stronger effect, reducing tumor growth by 79% compared to WT mice and also being significant when comparing it to Pld1^/− mice (Figure 3D).

PLD1 ablation does not impair pancreatic carcinogenesis in KC mice.

To investigate the role of PLD1 in pancreatic carcinogenesis in an immunocompetent mouse model, we used a well-established mouse model of early-stage PDA by conditionally expressing an endogenous mutant Kras^LSL-G12D allele in developing pancreatic tissues through the use of pancreas-specific Cre recombinase alleles (21). The LSL-Kras^G12D;Ptf1^Cre/+ (KC) mice are immunocompetent and contain premalignant epithelial lesions that mirror human pancreatic intraepithelial neoplasia (PanIN) (21,27). We crossed the KC mice to the Pld1^−/− mice (both are in the C57BL/6 background) to generate KC;Pld1^−/− mice (8) (Figure 4A).

Figure 4: PLD1 ablation does not impair pancreatic carcinogenesis in mice with active Kras.

(A) PCR analysis of wild-type (+/+) and Pld1 (−/−) genomic data (left) and immunoblot of PLD1 in LSL-Kras^G12D;Ptf1^Cre/+ (KC) and LSL-Kras^G12D;Ptf1^Cre/+;Pld1^−/− (KC;Pld1^−/−) mice (right). (B) Pancreas weight of male and female KC and KC;Pld1^−/− mice at sacrifice, *p<0.05, **p<0.005. n=7-10 per sex per group (C) H&E and immunostaining of PCNA, p-ERK, and Bcl-xL were performed on pancreas sections. Representative images (20x magnification) are shown. Scale bar: 100 μm. Acinar cell loss and mononuclear inflammatory cell infiltration results are expressed as percent area as described in Materials and Methods. Immunostaining results are expressed as percent of PCNA(+), p-ERK(+), or Bcl-xL(+) cells ± SD, *p<0.05, **p<0.001. n=6-10 per sex/group (D) Immunoblots of p-FAK, FAK, p-STAT3, STAT3, p-ERK, ERK, p-AKT, AKT, p-4E-BP1, and 4E-BP1, ***p<0.0005. n= 4-7 per sex per group. Values are presented as mean ± SD.

Cohorts of male and female KC and KC;Pld1^−/− mice were fed a standard chow diet and were euthanized at eight months of age to evaluate the degree of progression (e.g., from pre-invasive to invasive PDA) and the degree of recruitment of the tumor microenvironment. At euthanasia, pancreata from KC;Pld1^−/− mice were significantly smaller than those obtained from KC mice (Figure 4B). Histological analysis did not reveal significant differences in the degree of acinar cell loss or infiltration of mononuclear inflammatory cells in the pancreata between KC and KC;Pld1^−/− mice (Figure 4C) nor were proliferation levels, assessed by PCNA, different (Figure 4C). However, an increase in p-ERK and Bcl-xL levels was observed in the pancreata of female, but not male, KC;Pld1^−/− mice, compared to female KC mice (Figure 4C). In contrast, there were no significant differences in phosphorylation levels of FAK, STAT3, ERK and AKT between KC and KC;Pld1^−/− mice (Figure 4D). We observed that p-4E-BP1 levels were increased in the pancreata of male KC;Pld1^−/− mice compared to male KC mice and female KC;Pld1^−/− (Figure 4D).

Pharmacological inhibition of PLD reduces pancreatic carcinogenesis in mice with active Kras

Next, we determined if pharmacological inhibition of PLD could reduce PDA progression, by using FIPI, a small molecule inhibitor of PLD (22). For this purpose, cohorts of seven-month-old KC mice were treated with FIPI or vehicle-control (VC) intraperitoneally twice daily for 35 days to provide continuous inhibition of PLD.

We initially assessed the safety profile of KC mice treated with FIPI. No significant differences in body weight or liver weight was observed between VC and FIPI groups throughout the treatment (Supplemental Figure 1A and B). Histological analysis of H&E-stained liver sections revealed no significant differences in inflammation between groups (Supplemental Figure 1C). To expand further, liver and kidney function biomarkers were assessed in the serum. No significant differences in the levels of multiple liver enzymes (alkaline phosphatase, alanine aminotransferase, and aspartate aminotransferase) and kidney markers were observed in either VC of FIPI groups (Supplemental Figure 1D).

At endpoint, histological analysis of the pancreas indicated a significant decrease in acinar cell loss and infiltration of mononuclear inflammatory cells in mice treated with FIPI, compared to VC group (Figure 5B). In addition, we observed premalignant lesions analogous to low-grade PanINs, but no progression to cancer at this stage. Furthermore, although no differences were observed in pancreas weight between VC and FIPI treated mice (Figure 5A), FIPI demonstrated a 57% decrease in pancreatic proliferation, as measured immunohistochemically by PCNA (Figure 5B).

Figure 5: Pharmacological inhibition of PLD reduces pancreatic carcinogenesis in KC mice.

(A) Pancreas weight of vehicle control and FIPI treated mice at sacrifice. (B) H&E and immunostaining of PCNA and p-ERK were performed on pancreas sections. Representative images (20x magnification) are shown. Scale bar: 100 μm. Acinar cell loss and mononuclear inflammatory cell infiltration results are expressed as percent area as described in Materials and Methods. Yellow lines delineate areas of acinar cell loss. Immunostaining results are expressed as percent of PCNA(+) or p-ERK(+) cells ± SD. (C) Immunoblots of p-FAK, FAK, p-STAT3, STAT3, p-ERK, ERK, p-AKT, AKT, p-4E-BP1, and 4E-BP1, *p<0.05, **p<0.005. n=8-13 per group. Values are presented as mean ± SD.

Mechanistically, FIPI significantly decreased the phosphorylation levels of ERK in mice treated with FIPI when compared to VC (Figure 5B), and reduced FAK phosphorylation levels (Figure 5C). Interestingly, we observed that FIPI increased STAT3 phosphorylation when compared to VC mice (Figure 5C). No differences were observed in Bcl-xL levels or phosphorylation levels of AKT and 4E-BP1 (Figure 5C).

Finally, to further characterize the effects of FIPI on KC mice, we assessed the levels of various inflammatory cytokines in the serum of the mice at endpoint. While no significant differences were observed in TNF-α, KC/GRO, IFN-γ, or IL-10 among the groups, a significant increase in IL-6 levels were observed in the FIPI treated mice when compared to VC (Supplemental Figure 2A). When evaluated at the level of the pancreas, no significant differences in IL-6 gene expression were observed between the pancreata of FIPI and VC mice. (Supplemental Figure 2B).

Discussion

Given that PDA is an increasingly common cause of cancer mortality, there is a pressing need for the identification of new molecular targets against this deadly disease. Using genetic ablation and pharmacological inhibition of PLD, PLD1 was identified as a key player in pancreatic carcinogenesis. In addition, our findings highlight that FAK is a key downstream signaling pathway modulated by PLD.

PLD1 has been shown to be elevated in clinical samples of PDA, and positively correlates with vascular invasion and poor prognosis (9). Our studies demonstrate that the inhibition of PLD1 in pancreatic cancer cells leads to a reduced growth in tumor allografts. When pancreatic cancer cells were grown in PLD1-deficient mice, a reduction in tumor growth progression was observed, despite normal expression of PLD1 in the tumor cells, and this effect was enhanced when combined with GEM. An intriguing observation is that GEM attenuated tumor growth by 79% in Pld1^−/− mice, whereas it only reduced tumor growth by 55% in WT mice. These findings suggest the possibility that inhibiting PLD1 could represent a novel combination strategy for GEM. However, further efficacy studies in other preclinical models of PDA are needed to validate our preliminary findings.

When we extended the experimental paradigm to genetically-modified immunocompetent mice, PLD1 genetic ablation was unable to prevent pancreatic carcinogenesis. A possible explanation for the conflicting findings is that PLD ablation occurred simultaneously in both tumor and microenvironment in the KC mice or PLD1 deficiency affects the proliferation rates but not the underlying oncogenic process.

Targeting both PLD1 and PLD2 could represent a better strategy for PDA treatment. When KC mice were treated with FIPI, which inhibits both PLD1 and PLD2, acinar cell loss was effectively reduced by 35% with no toxic effects. Consistent with our findings, FIPI has been shown to effectively reduce lung tumor growth and metastasis, as well inhibit breast cancer growth (8,28). Overall, these findings highlight the promise of inhibiting PLD as a strategy for reducing tumor growth.

A key finding in our study was the observation that PLD inhibition affected FAK activation. FAK is a central hub for multiple signaling pathways, affecting a series of cellular biological behaviors such as proliferation, survival, adhesion, migration, invasion, and metastasis in various tumor types, including pancreatic tumors (17), and several inhibitors of FAK are being tested clinically. Moreover, elevated FAK expression can enhance tumor malignancy through driving an immunosuppressive microenvironment causing resistance to immunotherapy (14–17). Furthermore, FAK plays an important role in cell migration, chemoresistance, and modulating the fibrotic tumor microenvironment in PDA (29). The finding that genetic ablation of PLD in the microenvironment reduced FAK activation in KPC tumors points to the importance of PLD not only at the tumor level but highlights the possibility of cross signaling communication driving tumor growth. The decrease in FAK activation observed through inhibiting PLD in allograft models as well as in immunocompetent KC mice raises the possibility, although more research is warranted, that targeting PLD could affect immune microenvironment in PDA.

Another key aspect of inhibiting PLD was the regulation of Bcl-xL. Bcl-xL, overexpressed in PDA (30,31), has been identified as a key mediator of gemcitabine resistance. Bcl-xL contributes to the evasion of apoptosis, making cancer cells less susceptible to gemcitabine-induced cell death (32). We observed that PLD1 inhibition reduces Bcl-xL levels in pancreatic xenograft and allograft models at the tumor and the microenvironment levels. This reduction in Bcl-xL expression might explain in part, the reduction in tumor growth associated with the ablation of PLD1. In contrast, pharmacological inhibition of PLD with FIPI failed to affect Bcl-xL levels, whereas genetic ablation of PLD1 increased Bcl-xL levels in the pancreata of female KC mice. This apparent discrepancy warrants additional research to decipher the exact role of Bcl-xL in tumor growth in the context of PLD inhibition.

An interesting observation was the increase in serum IL-6 levels following FIPI treatment. When evaluated at the level of the pancreas, no significant increase was observed between FIPI and vehicle-treated controls. IL-6 is an inflammatory mediator that is produced by stimuli such as acute phase responses and immune reactions (33). These findings suggest that inhibition of PLD1 and PLD2 might play a role in immune reactions of an acute phase response, however, further research is needed to better characterize the role of PLD and immune responses in the context of PDA.

In summary, our findings identify PLD1 as a critical regulator of pancreatic xenograft and allograft growth, playing an important role at the tumor and at the microenvironment levels. Moreover, inhibition of both isoforms, PLD1 and PLD2, is necessary to reduce pancreatic carcinogenesis in immunocompetent KC mice. The mechanisms underlying the effect of PLD inhibition on pancreatic carcinogenesis are potentially mediated by the FAK pathway; an interesting finding worthy of further research. Furthermore, additional studies are warranted to evaluate whether PLD inhibitors enhance the efficacy of current standard-of-care for pancreatic cancer.

Supplementary Material

Download all supplementary files included with this article

Data Availability:

The data underlying this article are available in the article and in its online supplementary material.