|Home | About | Journals | Submit | Contact Us | Français|
The relationship between smoking and pancreatic cancer (PC) biology, particularly in the context of the heterogeneous microenvironment, remains incompletely defined. We hypothesized that nicotine exposure would lead to the augmentation of paracrine growth factor signaling between tumor-associated stroma (TAS) and PC cells, ultimately resulting in accelerated tumor growth and metastasis.
The effect of tobacco use on overall survival was analyzed using a prospectively maintained database of surgically resected patients with PC. Nicotine exposure was evaluated in vitro using primary patient-derived TAS and PC cells independently and in co-culture. Nicotine administration was then assessed in vivo using a patient-derived PC xenograft model.
Continued smoking was associated with reduced overall survival after surgical resection. In culture, nicotine stimulated hepatocyte growth factor (HGF) secretion in primary patient-derived TAS and nicotine stimulation was required for persistent PC cell c-Met activation in a co-culture model. c-Met activation in this manner led to the induction of inhibitor of differentiation-1 (Id1) in PC cells, previously established as a mediator of growth, invasion and chemoresistance. HGF-induced Id1 expression was abrogated by both epigenetic and pharmacologic c-Met inhibition. In patient-derived PC xenografts, nicotine treatment augmented tumor growth and metastasis; tumor lysates from nicotine-treated mice demonstrated elevated HGF expression by qRT-PCR and phospho-Met levels by ELISA. Similarly, elevated levels of phospho-Met in surgically resected PC specimens correlated with reduced overall survival.
Taken together, these data demonstrate a novel, microenvironment-dependent paracrine signaling mechanism by which nicotine exposure promotes the growth and metastasis of pancreatic cancer.
An estimated 11–32% of pancreatic adenocarcinomas (PC) are attributed to tobacco use, representing an important cause of mortality in the US (1, 2). Yet, the mechanisms linking smoking to the development and progression of PC remain poorly understood (3). Published reports demonstrate that continued smoking is associated with reduced survival in patients with lung and head and neck cancers (4, 5). These results suggest that sustained tobacco use promotes progression of these malignancies; but the effect of continued smoking on the progression of PC is unclear.
Tobacco use yields a multitude of toxins associated with carcinogenesis and tumor progression. Among these, nicotine has been characterized in both the initiation of cancer and progression of disease (6–10) but the effects on continue nicotine exposure on PC or its direct effect on the tumor microenvironment are poorly understood. Rationale for continued investigation of nicotine in cancer is supported by current trends in popular culture as E-cigarettes are promoted as a safer alternative to smoking despite the induction of comparable serum nicotine levels (11).
Mechanistically, our previous work has established an essential role for the transcriptional repressor known as inhibitor of differentiation-1 (Id1). We demonstrated that nicotine induces pancreatic cancer growth, metastasis, and chemoresistance through the induction of Id1 (12). However, this phenomenon was not characterized in the context of the tumor microenvironment which acts to promote tumor growth and chemoresistance through paracrine signaling events between PC cells and stromal elements (13, 14). Given the Src-dependent nature of Id1 induction described in our previous work, we chose to examine a known upstream mediator of Src activation with translational relevance in PC. Specifically, the mesenchymal-epithelial transition factor, c-Met, has emerged as a critical receptor tyrosine kinase (RTK) in cancer development and metastasis (15). For instance, recent investigations have isolated c-Met as a marker of pancreatic cancer stem cells (PCSCs) and pharmacologic inhibition of c-Met reduced the expression of other PCSC markers thereby restoring gemcitabine chemosensitivity (16, 17). In light of these findings, we asked if nicotine exposure could influence the tumor microenvironment through paracrine activation of c-Met signaling and determined whether activated c-Met may contribute to our previously characterized model of Id1 induced chemoresistance.
Our work demonstrates the first report of adverse prognostic effects associated with continued tobacco use in a prospectively maintained cohort of patients with potentially curative, surgically resected PC. Further, incorporation of signaling events between heterogeneous cell types within the PC microenvironment reveals novel, translational insights. We show that nicotine induces patient-derived tumor associated stromal (TAS) cells to secrete HGF, resulting in the stimulation of c-Met and subsequent upregulation of Id1 expression in PC cells. Moreover, HGF-MET signaling induced by nicotine in this manner augments tumor growth and metastasis in a patient-derived PC xenograft model which integrates desmoplastic stromal elements, inherent in our proposed mechanism (18). Collectively, these results are corroborated by quantitative assessments of c-Met activation in resected PC specimens, demonstrating a significant reduction in survival for patients with phospho-Met positive tumors. Taken together, these findings illustrate the translational relevance of c-Met activation in response to nicotine exposure in the tumor microenvironment. The data presented here demonstrate nicotine-induced tumor promoting effects through direct and paracrine signaling events thus providing a global mechanism to the strong association between systemic administration of nicotine and PC tumor progression.
All human PC cell lines were authenticated within 6 months by STR analysis. Human PC cell lines PANC-1, Mia-PaCa-2 and BxPC3 were obtained from American Type Culture Collection (ATCC, Rockville, MD). The L3.6pl pancreatic cancer cell metastatic variant was derived as previously described (19, 20). The selection of L3.6plGemRes gemcitabine-resistant pancreatic cancer cells was conducted as previously described (12, 21). Cells were maintained in culture with Dulbecco’s Modified Eagle’s Medium/F12 (DMEM/F12) with 10% fetal bovine serum (FBS) (Atlanta Biologicals, Atlanta, GA) and 0.6% penicillin/streptomycin and 5% CO2/95% air at 37°C. Patient derived TAS cells were generated from direct culture of gross human pancreatic adenocarcinoma surgical specimens and maintained in Dulbecco's Modified Eagle's Medium/F12 plus 10% fetal bovine serum as previously validated for pancreatic stromal cells (22–27). All primary TAS lines were confirmed to uniformly express high levels of α-smooth muscle actin (>99%) by immunocytochemistry and flow cytometry prior to experimentation (28). TAS lines demonstrated variable glial fibrillary acid protein (GFAP) expression (20–60%) and did not express epithelial surface antigen or the immune cell marker CD45 (28). All experiments with TAS cells were performed on actively replicating cells between passages two and five. Gemcitabine (Eli Lilly, Indianapolis, IN) was suspended in Dulbecco's PBS (D-PBS) and used at concentrations for the half maximal inhibitory concentration (IC50) previously published for PC cell lines (12, 29). Nicotine and recombinant human HGF stimulation were performed on cells that were rendered quiescent by serum-starvation for 48 hours with a physiologic dose of 1 µM nicotine (30) (Sigma Aldrich, St. Louis, MO) or 50 ng/mL recombinant human HGF (Millipore, Billerica, MA). PC cell stimulation by TAS cells was performed by the addition of TAS conditioned media and/or co-culture assays using transwell-Col plates (Costar, Corning, NY) whereby PC cells were plated in a 24 well plate with TAS cells in the insert chambers as indicated. Phospho-Met immunofluorescence was performed at 48 hours on direct co-cultures of TAS and PC cells plated simultaneously in 24-well plates. c-Met kinase inhibition was performed with Crizotinib (Selleck Chemicals, Houston, TX). Cellular viability was quantified by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay (Trevigen, Gaithersburg, MD) according to the manufacturer’s protocol.
Lysates were prepared from PC cell lines, TAS or surgically resected tumor tissues and western blots performed as previously described (12). Lysates were probed with either anti-Id1 antibody (BioCheck, Foster City, CA), anti-Met, anti-phospho-Met and anti-GAPDH (Cell Signaling Technologies, Danvers, MA), or anti-β-actin (Sigma-Aldrich, St. Louis, MO).
Total RNA was harvested from TAS, PC cells or patient tumors and purified using RNeasy Minikit plus DNase I treatment (Qiagen, Valencia, CA). After which RT-PCR was performed using the iScript One-Step RT-PCR Kit with SYBR Green (Bio-Rad, Hercules, CA). RNA was probed for HGF or c-Met (31), AchR subunits or GAPDH (Sigma Aldrich) using human specific primers. Each reaction was performed in a total volume of 50 µL containing 2X SYBR Green, reverse transcriptase (Bio-Rad Laboratories, Hercules, CA), RNA template and 50 nM of each specific primer utilizing the following thermocycler reaction: 50°C for 10 minutes, 95°C for 5 minutes and 40 cycles of amplification (95°C for 10 seconds, 55°C for 30 seconds). All data was normalized to GAPDH expression. β1 nAChR subunit: forward primer, 5′-TCA GAA ATG GGT CCG CCC TG-3′ reverse primer, 5′-TCC TGT TTG AGC CAC ACA TTG GT −3′. α5 nAChR subunit: forward primer, 5′-CTA GGC TGA GGC TGC TGT CCC-3′ reverse primer, 5′-ATG GAG CAC TGA GTG TGA GTC GT-3′. α7 nAChR subunit: forward primer, 5′-TCC CCG GCA AGA GGA GTG AA −3′ reverse primer, 5′-GAG GGC GGA GAT GAG CAC AC −3′. α9 nAChR subunit: forward primer, 5′-ATG CAC CGG CCA TCA CCA AA −3′ reverse primer, 5′-GAT CTC CGC TGT CCA AGG CG −3′.
106 TAS cells or 106 PC cells were plated in a 10 cm dish. After 24 hours of nicotine stimulation, culture media was collected and probed for HGF by ELISA (R&D systems, Minneapolis, MN). For detection of phospho-Met in patient derived tumor lysates, tumors were placed in cell extraction buffer (Invitrogen, Grand Island, NY) with complete Mini protease inhibitor cocktail (Roche, Manheim, Germany) and lysed using the FastPrep-24 system according to the manufacturer’s protocol (MP Biomedicals, Santa Ana, CA). Phospho-Met was determined using the PathScan Phospho-Met (panTyr) Sandwich ELISA (Cell Signaling Technologies, Danvers, MA) according to the manufacturer’s instructions.
PC cells were transfected with control (100 pmol), c-Met (100 pmol), or c-Src (100 pmol) siRNA (Santa Cruz Biotechnology, Santa Cruz, CA) using Oligofectamine reagent (Invitrogen, Grand Island, NY). Pharmacologic inhibition of c-Met phosphorylation was performed using Crizotinib (200 nM). Western blot analysis was used to monitor the expression and phosphorylation levels of c-Met.
Sample processing and immunohistochemical (IHC) staining was performed by the University of Florida’s Molecular Pathology Core Facility. Briefly, patient tumors, xenografts and murine lung specimens were formalin-fixed and paraffin embedded. 5 µm sections were stained with hematoxylin and eosin (H&E). In some tissues, serial 5 µm sections were probed using anti-α5 AchR and anti-α7 AchR (Santa Cruz Biotechnology) or phospho-Met (Cell Signaling Technologies) following antigen retrieval with citrate buffer pH 6.0 for tissue IHC analysis.
Immunofluorescence was performed on cells fixed for 10 minutes using paraformaldehyde followed by a one hour wash with 0.1% Triton X-100 in PBS with 3% bovine serum albumin. Cells were then incubated overnight using a primary antibody to c-Met (Cell Signaling Technologies, Danvers, MA). A goat anti-rabbit secondary antibody conjugated to AF647 was then applied with 4',6-diamidino-2-phenylindole (DAPI) and Phalloidin stains (Life Technologies, Carlsbad, CA).
A review of an institutional review board-approved, prospectively maintained pancreatic cancer database at the University of Florida (UF) was performed. Informed consent was obtained from all patients. Data analyzed included all consecutive patients who underwent surgery for pancreatic adenocarcinoma with continued follow-up from 2000 to 2011 (n = 90). Histologic slides of pancreatic specimens were reviewed by a pathologist specializing in pancreatic cancer (C.L).
All statistical analysis was performed using SPSS version 22.0 (IBM SPSS Statistics for Windows; IBM Corp). All in vitro and animal experimental groups were compared using the independent samples t test. Univariate and multivariate Cox proportional hazards models examined the effect of stage, grade, and smoking status on overall survival. All variables demonstrating an association with overall survival on univariate analysis (P < 0.20) were incorporated into multivariate analysis. A separate prospectively maintained database based of PC specimens obtained immediately upon resection (n = 26) was used to analyze phospho-Met levels in fresh tissue lysates. Patients were dichotomized to phospho-Met positive or negative based on a significant elevation of the intratumoral phospho-Met signal from background on ELISA. Kaplan-Meier survival curves were employed to analyze overall survival and the log-rank test was used to evaluate significance (P < 0.05).
All animal studies were performed with approval from the University of Florida Institutional Animal Care and Use Committee. A 2×2 mm section of a surgically resected primary pancreatic adenocarcinoma or 2 mm core biopsy was implanted subcutaneously into 8-week-old female NOD-SCID IL2 receptor gamma chain knockout (NSG) mice (Jackson Laboratory, Bar Harbor, ME) (n = 20 mice). Mice were anesthetized using inhaled isofluorane during the procedure and administered two doses of buprenorphine immediately and 12 hours postoperatively. On post-operative day 5, mice with visible tumors were equally randomized by size to receive 1 mg/kg nicotine (n = 10) or an equal volume of PBS (n = 10) three times per week via intraperitoneal injections as previously described (12). Tumor dimensions were measured three times per week using calipers. Tumor volumes were calculated using the equation: v = xy2/2 where v is volume, x is tumor length and y is tumor width. Tumors were allowed to reach an endpoint of 2 cm in maximum diameter prior to euthanasia. Primary tumors and lungs were harvested for immunohistochemical analysis.
Continued tobacco use during cancer treatment is associated with reduced survival in both head and neck squamous cell cancers as well as small cell lung cancers (4, 5). However, a comparable effect has not previously been demonstrated for patients battling pancreatic cancer (PC). In order to determine the relationship between continued tobacco exposure after PC resection and survival, we analyzed postoperative survival in ninety consecutive patients with PC who underwent pancreaticoduodenectomy (PD) or distal pancreatectomy (DP) from 2000 to 2011 with curative intent. All patients had continued clinical follow-up at the University of Florida. Patients were categorized into three groups: those who continued to use tobacco (current smokers), had a remote tobacco history (former smokers) and had no history of tobacco use. Univariate analysis was performed using known prognostic clinicopathologic parameters. Positive lymph node ratio, previously established as a prognostic parameter in PC, proved to be a more significant predictor than N stage and was therefore incorporated into multivariate analysis of overall survival (32). All other variables demonstrating established trends with survival were then incorporated into a multivariate Cox regression. Our data demonstrated that poor tumor differentiation (HR 2.03; P = 0.007) and continued tobacco abuse (HR 1.93; P = 0.040) significantly correlated with reduced overall survival (Table 1). Accordingly, quantitative assessment of smoking history in our PC population demonstrated a trend toward significance between estimated pack-years smoked and reduced survival on multivariate analysis (HR 1.01 per pack-year; P = .082, data not shown). Thus, after controlling for tumor stage and grade, it was determined that continued tobacco use is independently associated with reduced overall survival in patients with surgically resected PC (Fig. 1A). Together these data suggest that sustained tobacco use may directly influence the malignant progression of PC.
Tumor associated stromal (TAS) cells have been implicated in the promotion of PC cell growth, invasion, metastasis and chemoresistance via paracrine signaling through multiple growth factor pathways (33, 34). To investigate the role of TAS in chemoresistance, we harvested primary TAS cells from surgically resected PC specimens with an established outgrowth method (28). To determine the effects of cultured TAS on gemcitabine resistance in pancreatic cancer cells, co-culture of TAS outgrowths with the PC cell lines was performed using transwell inserts. PANC-1 and L3.6pl were selected for analysis as they represent both moderately resistant and highly sensitive cell lines to gemcitabine, respectively. As demonstrated in Fig. 1B, 48 hours of gemcitabine treatment reduced the viability of PC cells but co-culture of PC cells with TAS resulted in gemcitabine chemoresistance. We have previously identified the Id1 transcriptional repressor as an essential mediator of nicotine-mediated PC cell chemoresistance (12). TAS-mediated induction of Id1 in PC cells was therefore evaluated. Indeed, TAS conditioned media induced Id1 expression in PC cells in a time-dependent manner in a variety of pancreatic cancer cell lines (Fig. 1C). Thus, paracrine interactions between TAS and PC cells induce a similar chemoresistant phenotype as previously demonstrated with nicotine.
Although c-Met has been established as a pancreatic cancer stem cell marker that contributes to chemoresistance, the responsible signaling events remain poorly defined (35). In order to determine whether paracrine HGF-MET signaling could result in downstream induction of Id1, we first determined the expression of HGF, c-Met and Id1 by both TAS and PC cells. qRT-PCR and ELISA confirmed both expression and secretion of the HGF ligand in patient-derived TAS cultures, while representative PC cell lines displayed low to undetectable levels (Fig. 1D). Conversely, PC cell lines, but not TAS, expressed Id1 and c-Met (Fig. 1E). c-Met expression in PC cells alone was further supported by immunofluorescent microscopy, confirming the expression of c-Met in PC cells and the absence of c-Met in TAS even upon co-culture with PC cells (Fig. 1F). These data support a potential paracrine signaling mechanism whereby HGF secreted by TAS is able to interact with its binding partner c-Met in PC cells to induce Id1-mediated chemoresistance.
We next asked if nicotine could influence the tumor microenvironment by increasing secretion of the HGF ligand in TAS cells and c-Met expression in PC cells. As in Fig. 2A, nicotine treatment stimulated HGF expression and secretion in TAS cells. Additionally, nicotine exposure led to increased c-Met levels in PC cells (Fig. 2B) and this c-Met induction appeared to be occurring at the post-transcriptional level, as qRT-PCR demonstrated no change in c-Met transcription with nicotine treatment (Supplemental Fig. S1). Previous work has identified Src as an early mediator of nicotine stimulation in cancer cells (12, 36). To assess whether the induction of c-Met was similarly dependent on Src expression, Src siRNA treatment was incorporated into nicotine stimulations, demonstrating that nicotine-mediated c-Met induction was indeed Src-dependent in PC cells (Supplemental Fig. S2). We then questioned whether nicotine stimulation alone could lead to persistent c-Met activation in direct TAS/PC cell co-cultures. Indeed, phospho-Met immunofluorescence at 48 hours in co-culture demonstrated c-Met activation only in co-cultures exposed to nicotine, which was abrogated by pharmacologic c-Met inhibition using Crizotinib (Fig. 2C). Importantly, nicotine stimulation did not lead to c-Met activation in PC cells alone (Supplemental Fig. S3).
Given the physiologic response of TAS and PC cells to nicotine treatment, the repertoire of nicotine receptors and their subunits on these cell types was briefly evaluated. Our previous work has already established that PC cells express high levels of the α7 nAchR subunit, which was required for subsequent Id1 induction (12). Interestingly, qRT-PCR assay of the nAchR subunits β3, α5, α7 and α9 on a representative sample of patient-derived TAS cells revealed high expression of the α5 nAchR subunit (Fig. 2D). While the mechanistic role of these subunits in nicotinic signaling within TAS cells remains speculative, these original observations support the presence of nicotinic receptors on TAS cells and therefore support a potential role for nicotine in the pancreatic tumor microenvironment.
We have previously demonstrated the importance of Id1 expression in pancreatic cancer progression and chemoresistance (12). To assess whether Id1 expression could be influenced by TAS cells in a c-Met dependent manner, PC cells were silenced of c-Met followed by stimulation with TAS conditioned media. Inhibition of c-Met expression abrogated the observed induction of Id1 by TAS conditioned media (Fig. 3A) suggesting a c-Met dependent-Id1 signaling axis. To determine whether the Id1 induction observed was specifically due to the HGF ligand and no other potential soluble mediators within the TAS conditioned media, we performed the following experiments. PC cells were stimulated with physiologic levels of HGF (50 ng/mL) alone and Id1 expression evaluated (Fig. 3B). Id1 expression was induced in a similar time-dependent manner as with the addition of TAS conditioned media. This HGF-induction of Id1 was abrogated by siRNA knockdown of c-Met (Fig. 3C) in a manner similar to the results noted in Fig. 3A. Further, Id1 induction in response to HGF (Fig. 3D) or TAS conditioned media (Fig. 3E) was abrogated by the FDA approved targeted c-Met tyrosine kinase inhibitor, Crizotinib. These results demonstrate that PC cell Id1 induction from adjacent TAS signaling is dependent on activation of the HGF-MET cascade. Taken together, we suggest that nicotine may enlist the stromal compartment to induce chemoresistance by augmenting the c-Met signaling pathway in PC cells.
Our previous data demonstrated the effects of nicotine on primary cancer cell growth in an orthotopic pancreatic cancer xenograft model (12), but this model is unable to suggest an effect of nicotine on the more representative stromal rich tumor microenvironment. Therefore, to support our in vitro data, we examined the effects of nicotine on PC and its closely related stromal environment by utilizing a patient-derived PC xenograft model, previously validated to preserve both genetic and epigenetic elements of the original tumor (37). Prior to our in vivo experiments with nicotine, confirmation of the presence of nicotinic receptors in surgically resected PC tissue specimens utilized in our xenograft model was performed. Supporting our in vitro data, α7 subunits localized to PC cells, while α5 displayed a more diffuse pattern, staining the surrounding, desmoplastic stromal elements (Fig. 4A).
Tumor-bearing mice that received physiologic doses of nicotine (1mg/kg) experienced significantly augmented tumor growth as compared to nicotine-free controls (Fig. 4B). Supporting our in vitro results, analysis of xenograft lysates revealed significantly increased HGF expression in tumors from nicotine-treated mice compared to those from controls (Fig. 4C). Next, we determined if nicotine induced activation of c-Met within the tumor microenvironment. Tumor bearing mice treated with nicotine demonstrated consistently elevated levels of phosphorylated c-Met (Fig. 4D, E). Additionally, nicotine administration induced pulmonary metastasis by H&E examination in 5 of 8 (62%) treated animals, as opposed to none of the (0 of 8) controls (Fig. 4F). Hepatic metastasis was not observed in either group. Together the in vivo data support the mechanisms delineated in vitro whereby nicotine augments the HGF-MET signaling cascade by induction of HGF secretion in TAS and c-Met expression in PC cells. These data support a role for nicotine on a more representative intact pancreatic tumor and its supporting microenvironment, strongly correlating with accelerated tumor growth and metastasis in vivo.
Given the relationship between activated c-Met and tumor growth and metastasis in our xenograft model, the prognostic value of activated c-Met levels in resected human PC specimens was evaluated. Specifically, activated c-Met content in fresh tumor lysates from a cohort of twenty-six surgically resected patients with PC was quantitatively evaluated using an ELISA (Fig. 5A). In addition, phospho-Met expression was confirmed by IHC staining on representative samples (Fig. 5A, inset). Here, elevated intratumoral phospho-Met levels correlated with reduced survival when evaluated in a continuous fashion using a Cox proportional hazards model (HR 11.1, P = .049). Kaplan-Meier survival curves were generated upon dichotomization of tumors to phospho-Met positive and negative groups based on significantly elevated phospho-Met signals from background on ELISA. Notably, not one single patient with a positive phospho-Met cancer survived one year postoperatively (Median OS 6.1 vs. 15.2 months; P = .028) (Fig. 5A). Taken together, these results indicate that c-Met activation, implicated by our model of nicotine-mediated PC progression, portends a poor prognosis in PC.
Data presented here provide strong evidence of an association between nicotine exposure and reduced survival in patients with resected PC. Additionally, we demonstrate an equally important role for nicotine on the desmoplastic PC tumor microenvironment. Specifically, in addition to promoting PC growth and chemoresistance directly through the induction of Id1, nicotine concurrently amplifies paracrine HGF-MET signaling in the tumor microenvironment. This mechanism functions through the induction of HGF in TAS cells and subsequent c-Met activation in PC cells, which is both necessary and sufficient for TAS-induced Id1 expression in PC cells (Fig. 5B). Accordingly, induction and upregulation of the HGF-MET axis in patient-derived PC xenografts following nicotine treatment was associated with rapid tumor growth and metastasis. Supporting our preclinical work, when we evaluated our surgically resected patient population, c-Met activation in human PC specimens was a predictor of poor survival.
Our mechanistic findings support the work of others investigating the effects of nicotine in cancer. Chellappan et al. have demonstrated nicotine-induced proliferation and invasion in a variety of cancer types (7, 8, 38). Our previous work built on this foundation, by demonstrating that nicotine-mediated PC growth, metastasis and chemoresistance is Id1-dependent (12). Similarly, our findings regarding the significance of activated c-Met in PC corroborate the work of Simeone et al. identifying c-Met as a putative marker of pancreatic cancer stem cells (17, 35). In addition, c-Met inhibition with such tyrosine kinase inhibitors as Crizotinib has demonstrated success in preclinical work and is currently employed by multiple clinical trials (16), lending further translational relevance to the investigation of the role of HGF-MET signaling in PC. Given the mixed results of c-Met inhibitors in clinical trials thus far, patient selection will be critical to delineate proper application of these therapies (39). Here we present preliminary data supporting a simple clinical tool, a recent history of nicotine exposure, as a part of that selection process.
Here we demonstrate that the mechanism by which nicotine increases c-Met levels in PC cells is Src-dependent. Previous work has identified β-arrestin as a key scaffolding protein allowing nAChR-mediated activation of Src (36). The direct binding of β-arrestin and Src demonstrated in these investigations would suggest a role for Src as an early mediator of the nAChR response. We find that c-Met induction is downstream of Src and occurs at the post-transcriptional level. Control of c-Met levels in this manner has been demonstrated previously through sequestration of the E3-ubiquitin ligase responsible for the ubiquitination of c-Met in the Golgi apparatus (40). However, the precise interplay between Src and c-Met upon nicotine stimulation remains speculative at this point.
Although we report strong correlations between smoking and survival as well as intra-tumoral phospho-Met levels and survival, the link between smoking and c-Met activity in patient tumors has not yet been characterized but is critical to our future work. Importantly, nicotine-mediated upregulation of c-Met occurs regardless of basal c-Met expression levels amongst PC cell lines. However, since there is high variability in observed c-Met expression in human PC specimens, a large number of patients would be required to confirm the relationship between nicotine exposure and c-Met activation within human PC specimens, which we are currently accumulating.
An additional limitation of this work is the lack of pharmacologic inhibition of c-Met in vivo. However, the correlation between nicotine administration, tumor growth, intratumoral HGF expression and phospho-Met levels suggests that nicotine induces HGF-MET signaling within the tumor microenvironment, which is a known promoter of tumor growth and metastasis (34). While the addition of nicotine with c-Met tyrosine kinase inhibition in a patient-derived xenograft model with an intact tumor microenvironment is a future direction, there are published reports demonstrating the importance of the HGF/c-Met cascade in PC in vivo (35, 41–43).
Crizotinib is known to inhibit multiple tyrosine kinase receptors. Thus, despite abrogation with c-Met knockdown, TAS-mediated Id1 induction in PC cells may not be solely dependent on HGF-MET signaling. A recent analysis of Crizotinib activity in PC cell lines implicated ALK inhibition as the primary mechanism of action rather than direct effects on c-Met (44). In this study ALK inhibition contributed to reduced PC cell viability in isolated Crizotinib-treated PC cells. Conversely, others have demonstrated that c-Met represents the therapeutic target in PC and that gemcitabine acts synergistically with Crizotinib in an orthotopic xenograft model (41). Both investigations utilized a xenograft model using established PC cell lines, which lack the human fibrous stromal component responsible for HGF production. Therefore, HGF-MET paracrine signaling is not clinically applicable in the context of isolated PC cells in vitro or injected PC cell xenografts. Our data would therefore suggest that c-Met inhibition may still be a more prominent mechanism mediating the antitumor effects of Crizotinib in an intact tumor microenvironment containing up to 80% HGF-secreting stromal cells by mass, which is supported by multiple recent investigations (41, 45–47).
Accumulating evidence suggests a prominent role for the induction of Id1 as a central event in angiogenesis and tissue regeneration (48, 49). However, the relationship between HGF and Id1 appears to be specific to the environment in question. Recent experiments in hepatoma cell lines suggest that HGF induces cell cycle arrest via the inhibition of Id1 expression in this population (50). On the other hand, our findings are consistent with other literature that demonstrated that HGF-MET-Id1 signaling supports a pro-survival, regenerative mechanism in epithelial cells (15). In addition, investigations into fracture repair have also demonstrated active paracrine HGF stimulation, which leads to Id1 expression and subsequent wound healing (51). Rafii et al examined Id1 in the context of hepatic regeneration, that paradoxically suggested that Id1 expression by endothelial cells leads to HGF secretion which then fuels c-Met activation and proliferation in nearby hepatocytes (52). Alternatively, Apte et al recently demonstrated that HGF secreted from TAS fuels endothelial cell proliferation and tube formation, providing another potential microenvironment-dependent tumor-promoting mechanism (53). Taken together, this literature suggests that functional outcomes of the HGF-MET-Id1 axis are environmentally and cell type dependent.
Finally, we cannot exclude the possibility that non-nicotine components of cigarette smoke may be responsible for reduced survival in patients with PC. Indeed, multiple investigations implicating cigarette smoke in the promotion of cancer cell growth and metastasis describe mechanisms that might not be fully dependent on nicotine (54–56). However, the isolation of nicotine-specific effects in PC is especially relevant given the current social climate regarding electronic cigarettes (E-cigarettes). Widely touted as a healthier alternative to smoking despite the induction of comparable serum nicotine levels, electronic cigarettes remain clinically concerning (10). Further, recent investigations indicate a latency time of approximately ten years between the initiating mutation and the “birth” of the pancreatic cancer initiating cell, with metastatic potential likely requiring an additional five years (57). These data suggest a fifteen-year latency time prior to diagnosis. Given current epidemiologic evidence that 80–85% of patients diagnosed with PC present with advanced lesions (58), we must be aware of such tumor promoting “healthier alternatives” to smoking that might be influencing tumor progression. Taken together with data presented here regarding nicotine-induced tumor promoting effects through direct and paracrine signaling mechanisms, a robust case is established against the systemic administration of nicotine with respect to progression of established as well as potentially undiagnosed malignancies.
In addition to promoting carcinogenesis, sustained tobacco use is associated with accelerated tumor progression and reduced survival in both lung and head and neck cancers. This work first serves to validate this finding in pancreatic cancer. To address the mechanism underlying this phenomenon, we hypothesized that paracrine growth factor signaling within the stroma-rich pancreatic cancer microenvironment may be especially responsive to nicotine exposure. Through the use of human primary cell cultures and patient-derived xenografts, we investigate a novel, pathologic secretory response of the tumor-associated stromal cell to nicotine. Ultimately, this response leads to unrestrained HGF-MET signaling within the tumor microenvironment and the rapid accumulation of metastatic lesions. These results suggest strongly that nicotine administration, whether through tobacco use, replacement therapies or e-cigarettes, continues to promote the progression of pancreatic cancer even after a tumor has been established.
Support: NCI 5T32CA106493-09, Cracchiolo Foundation
Conflicts of interest: None