Prolactin Promotes Fibrosis and Pancreatic Cancer Progression

Pancreatic ductal adenocarcinoma (PDAC) is an almost uniformly fatal disease, arising from mutant KRAS-expressing precursors following a prolonged period of local inflammation and stress (1). A characteristic feature of PDAC is its intense stromal content, accounting for up to 90% of the tumor volume, and is composed of cancer-associated fibroblasts, pancreatic stellate cells, vascular endothelium, and immune cells, including tumor-associated macrophages (TAM), one of the most abundant myeloid-derived cell types in the stroma (2).

Macrophages are monocyte-derived myeloid cells that develop from a common myeloid progenitor cell residing within the bone marrow (BM; ref. 3). Upon tissue damage or in pathologic conditions, inflammatory monocytes are recruited across the endothelium, where they differentiate into macrophages (3). Another subset of macrophages, called tissue-resident macrophages, originally derived from primitive hematopoiesis in the embryonic yolk sac are deposited early in development (4). These embryonic-derived resident macrophages persist into adulthood, and self-maintain independently of the BM (4, 5). Recently, it was reported that in addition to the BM-derived macrophages, embryonic-derived tissue-resident macrophages contributed significantly to the TAM population in PDAC (6). This newly identified TAM subpopulation can expand in situ, displaying a profibrotic expression profile and facilitating PDAC progression (6). Macrophages exist across an M1–M2 polarization state, in which M1 cells initiate and sustain inflammation through production of high levels of proinflammatory cytokines, reactive nitrogen species, and oxygen intermediates, whereas the more heterogeneous M2 cells are characterized by alternative mechanisms of arginine metabolism, exhibit a different chemokine expression profile, and are associated with wound healing or smoldering chronic inflammation (7).

Expansion and progression of noninvasive pancreatic precursor lesions, as well as pancreatic cancer, relies heavily on soluble factors produced by individual cell populations within the stroma. Together, these factors exert effects on a wide range of cell types (8). For instance, although IL6, IL17, TGFβ, and Notch pathway ligands primarily influence the malignant epithelium, other factors such as GM-CSF are implicated in the recruitment and expansion of both immunosuppressive elements and antigen cross-presenting dendritic cells (9–14). The stroma also displays an abundant and rigid extracellular matrix, which promotes increases in tumor interstitial fluid, induces vascular compression, and impedes access of therapeutic agents (15, 16). Depleting stroma could in theory enhance drug delivery while disrupting stroma–cancer cell interactions (13–15, 17, 18). However, targeting stroma alone may actually favor a more aggressive form of PDAC (17, 18), thus arguing for therapeutic approaches that specifically target the cancer cells.

Prolactin (PRL) is a polypeptide hormone that is predominantly synthesized in, and secreted from, lactotroph cells of the anterior pituitary gland (19). Although it is best known as the hormone that elicits lactation in mammals, it is present in males and females in all vertebrates and in evolutionary terms, is an ancient hormone (19, 20). Prolactin has many functions that extend beyond reproduction and lactation, with approximately 300 identifiable biological actions in vertebrates (21). Binding of prolactin to its membrane-associated receptor, the prolactin-receptor (PRLR), induces several intracellular oncogenic signaling cascades including JAK/STAT signaling that stimulate proliferation of cancer cells and promote tumor growth (19, 21, 22). Although PRL-induced JAK2/STAT5 activation has been considered the hallmark of PRLR signaling, multiple intracellular kinases, such as Src-Family Kinases, focal adhesion kinase (FAK), MAPK, and PI3K, are also induced by PRL (22, 23). Altered JAK/STAT activity in pancreatic cancer cells stimulates cell proliferation and malignant transformation, inhibits apoptosis, and may be responsible for promoting resistance to therapy (24). Mounting evidence suggests that blocking JAK/STAT signaling pathways inhibits pancreatic cancer growth (25–27). Nevertheless, despite the oncogenic activity of the PRL/PRLR axis in general, and the involvement of JAK/STAT in PDAC in particular, the role of PRL and its cognate receptor in pancreatic cancer remains poorly defined.

In this study, we investigated the role of PRL/PRLR axis in pancreatic intraepithelial neoplasia (PanIN) and PDAC progression to determine the mechanism by which PRL promotes pancreatic tumorigenesis, and to identify the source of prolactin within the tumor microenvironment.

Human pancreatic adenocarcinoma specimens were collected from patients who underwent surgical procedure. The studies were conducted in accordance with the ethical guidelines of the Declaration of Helsinki and were approved by the University of Pittsburgh Institutional Review Board. All surgically removed pancreatic tissues were collected in accordance with the University of Pittsburgh's Institutional Review Board–approved protocols and used for analysis under written-informed consent from the patients.

Mice used in these studies were maintained according to protocols approved by the University of Pittsburgh Institutional Animal Care and Use Committee. The Ptf1aCre (28), Kras^G12D (29), and the prolactin^−/− strains (30) were obtained from the Mutant Mouse Regional resource Centers, National Cancer Institute Repository, and the Jackson laboratories, respectively.

Tissue processing and immunostaining were conducted as previously described (31). Images were acquired on a Zeiss Imager Z1 microscope with a Zeiss AxioCam driven by Zeiss AxioVision Rel.4.7 software (Zeiss). The antibodies used in this study are listed in Supplementary Table S1.

Generation of bone marrow–derived (BMD) macrophages as well as isolation of primary pancreatic macrophages was performed as described previously (32).

Recombinant human prolactin was obtained from R&D Systems Inc. Cells were serum-starved overnight and then cultured with 10 nmol/L hrPRL for up to 60 minutes.

Mice were s.c. injected once a day with 100 μg Metoclopramide for 3 weeks, as described elsewhere (33).

PANC-1 and BxPC-3 cells were purchased from the ATCC in 2015, 2017, and 2019. Cell identity was verified by the ATCC. Cells used for the experimental study were passaged within 10 to 20 passages after reviving from the frozen vials. Cell lines were screened at early and late passages for Mycoplasma. PANC-1 cells were cultured in a complete DMEM media (ATCC), and BxPC-3 cells were cultured in RPMI 1640 media (ATCC), containing 10% FBS and 1% penicillin–streptomycin solution.

PRLR expression was silenced with commercially obtained shRNA-containing lentiviral particles (Origene). Control shRNA (TR30021V) and shRNA targeting PRLR (TL310169V) were used to silence PRLR expression. The constructs were introduced into the cells by lentiviral transduction (200 μL of lentiviral shRNA supernatant), supplemented with 8 μg/mL Polybrene (Sigma-Aldrich). Cells were then replenished with fresh medium with 20 nmol/L PRL.

Cells were seeded in 96-wells plates at a density of 1 × 10² cells in 100 μL per well. The CellTiter-Glo Luminescent Cell Viability Assay (Promega) was used as follows. Cells were allowed to grow for 72 hours at 37°C and then mixed 1:1 with CellTiter-Glo reagent in a clear-bottom white 96-wells plate and incubated in dark for 5 minutes to stabilize the signal. The luciferase signal, which is directly proportional to the amount of metabolically active cells in the sample, was measured on a Glomax luminometer (Promega).

Results in bar graphs are expressed as mean values ± SEM. The number of independently performed experiments is mentioned in the respective figure legends. To compare the experimental groups, the Student t test or one-way ANOVA followed by Bonferroni correction was performed using GraphPad Prism 5 software (Graphpad Software Inc.). Differences with P < 0.05 were considered statistically significant.

Given the oncogenic nature of the PRL/PRLR axis, we first studied the expression of PRLR in specimens obtained from pancreatic cancer patients undergoing operative extirpation. PRLR could not be detected in normal human pancreas (Fig. 1A), or in early neoplastic lesions; however, it was expressed in high-grade PanINs (Fig. 1B). PRLR was expressed in all invasive cancer samples studied (15/15), although the extent of its expression by cancer cells varied within the cohort (Fig. 1C and D). A recent integrated genomic analysis has categorized PDAC into four distinct classes, namely aberrantly differentiated endocrine exocrine (ADEX), pancreatic progenitor-like, immunogenic, and squamous (34). Among these classes, the squamous subtype is an independent prognostic factor and is associated with poor outcomes (34). We found lower PRLR expression levels in the squamous class compared with the other subtypes (Fig. 1E). Accordingly, patients with low PRLR expression survived less than patients with higher PRLR expression (Fig. 1F). Interestingly, closer examination revealed that in addition to the malignant epithelium, PRLR is expressed in a subset of stromal cells (Fig. 1G). Similarly, PRLR could be detected in mPanINs and a subset of surrounding stromal cells as well as PDACs in Ptf1aCre;Kras^G12D mice (abbreviated KC mice; Fig. 1H and I). The presence of PRLR in both PanINs and PDAC suggests that PRL/PRLR axis may play a role in the disease progression.

As the first step to study the PRL/PRLR axis in vitro, we examined two different pancreatic carcinoma cell lines, which either expressed PRLR (PANC-1), or lacked PRLR expression (BxPC-3; Fig. 2A). Moreover, we found that activation or inactivation of the pathway by recombinant prolactin or sh-RNA targeting PRLR in PANNC-1 cells resulted in increase or decline in cell proliferation, respectively (Fig. 2B). Furthermore, regardless of PRLR expression, known PRLR downstream signaling molecules STAT3 and STAT5 (23) were phosphorylated in both cell lines (Fig. 2C). Next, we studied the ability of PRL/PRLR intracellular signaling through activation of JAK2 signaling cascade. Stimulation of PANC-1 cells with human recombinant prolactin resulted in a time-dependent increase in pJAK2, and pSTAT3, whereas surprisingly the level of pSTAT5 remained unchanged (Fig. 2D). To further confirm this finding, we transfected Panc-1 cells with a 6xSTAT5-Luc reporter construct (35), and examined the prolactin-induced STAT5 transcriptional activity. As shown in Fig. 2E, prolactin treatment did not increase 6xSTAT5-Luc promoter activity, indicating that STAT5 activation in human pancreatic cancer cells is not dependent on PRL/PRLR signaling. Noteworthy, we could not detect any changes in p-STAT3 or p-STAT5 levels in BxPC3 cells following prolactin treatment (Supplementary Fig. S1). These results confirm the data presented in Fig. 1 that pancreatic cancer cells express PRLR in a heterogeneous manner. Moreover, the PRL/PRLR axis activates STAT3 through a JAK2 signaling cascade. Finally, phosphorylation of STAT5 in these cells is likely to be independent of the PRL/PRLR pathway.

As shown in Fig. 1, in addition to being present in PDAC cells, PRLR could be detected in PanINs, and also in the surrounding stromal cells. Pregnancy and lactation are two conditions associated with high systemic prolactin levels. To test the ability of PanINs to respond to prolactin, 6- to 8-week-old KC females were mated and subsequently sacrificed 3 weeks post-partum (at the time of weaning). This approach exposed PanINs to high levels of prolactin for almost 6 weeks, which resulted in an intense stromal response, as well as a higher incidence of both acinar-to-ductal metaplasia (ADM) and PanIN1 lesions, although still in the context of intact lobule architecture (Fig. 3A–C).

Next, 6- to 8-week-old KC females were again mated, but instead they were sacrificed either 1 month after weaning, or alternatively they were subjected to a second round of pregnancy and lactation (Fig. 3D–G). Here, the KC dams were all sacrificed at 16 to 17 weeks of age. Compared with control 4-month-old unmated KC mice (Fig. 3D), both experimental cohorts showed a significant decrease in the number of normal ducts, and an increase in PanIN1 lesions (Fig. 3E–G). Furthermore, a robust desmoplastic response, secondary loss of the lobular acinar parenchyma, as well as increased duct ectasia could be observed in these mice (Fig. 3E). Mice with two back-to-back pregnancies also showed significant higher numbers of ADMs and PanIN2s (Fig. 3F and G).

Pregnancy is associated with numerous hormonal changes, thus the observed acceleration in PanIN progression as the result of pregnancy and lactation could be the combined effect of many hormones. Prolactin secretion is inhibited by dopamine; therefore, metoclopramide, a dopamine antagonist, has been used to stimulate prolactin secretion (19, 33). To exclude the potential influence of other hormones, we next studied PanIN progression in response to metoclopramide treatment. To do so, 4-week-old male KC mice were treated with metoclopramide for 3 weeks as described elsewhere (33), and were sacrificed immediately after the treatment. As expected, we detected increased serum prolactin levels in mice treated with metoclopramide (Fig. 3H). In the control untreated KC mice, except for only scattered early lesions, the pancreas looked morphologically normal (Fig. 3I). In contrast, the metoclopramide-treated mice displayed a higher incidence of both ADMs and PanIN1 lesions (Fig. 3J and K). Together, these findings suggest that elevated systemic prolactin levels promote PanIN progression.

Binding to PRLR is not specific for prolactin, as placental lactogen in pregnant females also has some affinity for this receptor (23). Thus, the ability of PanINs to respond to high systemic prolactin levels does not necessarily mean that prolactin is part of disease progression. To dissect the potential role of prolactin in pancreatic tumorigenesis in vivo, we crossed the KC mice with prolactin-deficient mice (Prl^−/− mice) to generate a Ptf1aCre;Kras^G12D;Prl^−/− transgenic line (abbreviated KC;Prl^−/−). In Prl^−/− mice mammary gland, development is impaired; however, they have a normal hematopoietic system (30). More importantly, the Prl^−/− mice show normal pancreatic gross morphology and differentiation (Supplementary Fig. S2). As in the KC mice (Fig. 4A), although we could find areas with desmoplastic reaction and neoplastic lesions throughout the 6-month-old KC; Prl^−/− pancreas (Fig. 4B), many lobes displayed generally normal gross morphology (Fig. 4C). In the areas containing lesions, although the number of ADMs was similar to the age-matched KC mice, there was a decrease in PanIN1 and PanIN2 lesions (Fig. 4A–D). It should be noted that throughout this study, KC mice showed classic stepwise PanIN formation and in some cases (2/5) transition to PDAC (Figs. 3A and D, 4A, E, F). Interestingly, no further PanIN progression or PDAC formation could be detected in 1-year-old KC;Prl^−/− mice (Fig. 4G and H). Considering that Prl^−/− female mice are infertile (30), it was not feasible to study the effect of pregnancy on PanINs in KC;Prl^−/− mice. Nevertheless, these findings suggest that prolactin is dispensable for initiation of PanINs but is required for PanIN progression.

Our in vitro studies identified STAT3 as downstream effector of PRL/PRLR axis in PDAC (Figs. 2 and 3). To gain insight into the mechanism by which prolactin contributes to PanIN progression, we analyzed the status of pSTAT3 in 6-month-old KC, and KC;Prl^−/− mice, as well as in 4-month-old KC dams with two previous pregnancies (Fig. 5A–D). In the pancreas, pSTAT3 is present in metaplastic ducts, a subset of premalignant lesions as well as in F4/80⁺ macrophages (11, 12, 36, 37). In accordance with these reports, in control KC mice, we could detect pSTAT3 in ADMs, PanINs (60%), and a subset of surrounding stromal cells (Fig. 5A). In KC;Prl^−/−pancreas, although total Stat3 could be detected in PanINs (Supplementary Fig. S3A), the number of pSTAT3-positive PanINs had declined significantly (Fig. 5B). Moreover, pStat3 was absent in stromal cells, whereas it could still be detected in ADMs (Fig. 5B, inset). In mice with previous pregnancies, the number of pSTAT3-positive cells increased both in the stromal compartment and the lesions (Fig. 5C and D).

In addition to induction of Jak2/Stat pathway, prolactin may also promote phosphorylation of FAK through activation of Src (23). Recent studies have shown that high FAK1 activity in PDAC is associated with extensive collagen deposition and fibrosis (38). Given the absence of desmoplastic reaction in relatively large areas in the KC;Prl^−/−pancreas, and increased fibrosis as the result of pregnancy, we next compared the pFAK1 status, the collagen deposition, as well as the abundance of hyaluronic acid (HA) in these settings. In control KC mice, pFAK1 could be detected in ADMs, PanINs (80%), and also in some stromal cells (Fig. 5E). Although prolactin deficiency did not alter total FAK1 (Supplementary Fig. S3B) or pFAK1 in ADMs, pFAK1 was barely detectable in PanINs and the stroma (Fig. 5F). In contrast, previous pregnancy led to a more prominent pFAK1 presence within the stromal cells, and a significant increase in the epithelial pFAK1 levels compared with the age-matched control KC mice (Fig. 5F–H. Furthermore, we found that KC;Prl^−/−mice had reduced levels of fibrosis as seen by Sirius Red (Fig. 5I–L) and collagen staining (Fig. 5M–P). Interestingly, the difference in collagen deposition appeared more prominent in the area surrounding the lesions (Fig. 5J–L, insets). The abundance of HA on the other hand, confirmed by staining with HA-binding protein, was similar in KC and KC;Prl^−/−mice (Supplementary Fig. S3C–S3F). Together, these findings indicate that prolactin influences both the neoplastic lesions and the surrounding stroma.

Pancreatic cancer is somewhat more common in men than in women. Pancreatic cancer patients also have normal blood prolactin levels (39). Therefore, it is unlikely that high systemic prolactin levels would be driving progression of the disease. Thus, the source of prolactin may be found within the pancreas. We have previously reported the role of macrophages during pancreatic regeneration (32). In search of the putative cell type responsible for prolactin production in the pancreas, we examined F4/80 and prolactin expression in KC pancreas and found prolactin-positive macrophages surrounding the PanIN lesions (Fig. 6A and B).

The high mobility group box 1 (HMGB1) protein regulates DNA damage responses within the nucleus (40), and serves to promote access to various transcriptional complexes including hormone/nuclear receptor complexes (41). However, under stressful situations, such as in pancreatitis or PDAC, HMGB1 is released into the cytosol where it promotes autophagy and mitochondrial quality control as well as into the extracellular space, enhancing the inflammatory response (42). Given the abundance of macrophages in proximity to PanINs, and also the ability of extracellular HMGB1 to bind to TLR2, TLR4, and the receptor for advanced glycation end products in macrophages (43), we speculated that HMGB1 might induce prolactin expression in macrophages. To test this hypothesis, conditioned media (CM) collected from activated wild-type or HMGB1-deficient platelets were added to the BMD macrophage cultures. Macrophages cultured in the absence of CM did not express prolactin, whereas we could detect prolactin expression in those cultured with CM obtained from wild-type platelets (Fig. 6C). Interestingly, HMGB1-deficient CM failed to stimulate similar prolactin expression in macrophages (Fig. 6C). To confirm our in vitro study, we next examined PdxCre;Kras^G12D;Hmgb1^fl/fl mice (KCH mice; ref. 44) and found reduced prolactin expressed by F4/80⁺ cells (Fig. 6D). These data imply that macrophages present in the tumor environment express prolactin, and that this prolactin expression can be stimulated by extracellular HMGB1.

The presence of PRLR⁺ cells in the stroma (Figs. 1 and 4) in conjunction with the observed effect of prolactin on stromal pFAK1 and collagen deposition (Fig. 5) suggests that in addition to the neoplastic epithelium, prolactin may also affect stromal cells directly. Because embryonic-derived resident macrophages are implicated in PDAC fibrogenesis (6), we hypothesized that the resident macrophages with profibrotic activities may express PRLR. To test this hypothesis, we first generated macrophages from the BM in vitro, and evaluated PRLR expression. As shown in Fig. 6E, the monocyte-derived F4/80⁺/CD11b⁺ macrophages do not express PRLR. Next, we isolated primary macrophages from the adult pancreas. F4/80⁺/CD11b⁺/CD11c⁺/CD206⁺ cells are embryonic-deposited–resident macrophages in the pancreas (45). In the uninjured pancreas, we found that the vast majority of macrophages were CD11c⁺/CD206⁺ resident macrophages, whereas around 10% displayed pure M1 feature (Fig. 6F). Notably, at the baseline, macrophages did not express PRLR (Fig. 6F). To explore the impact of tissue damage and inflammation on CD11c⁺/CD206⁺/PRLR⁺ macrophages, we analyzed the status of these cells in wild-type animals following caerulein treatment (Fig. 6G) or KC mice (Fig. 6H). As anticipated, the overall number of F4/80⁺/CD11b⁺ cells were higher in these two conditions compared with the uninjured pancreas (Fig. 6G–I). However, we observed a shift from a M1/M2 to a M1 phenotype within the macrophage population (Fig. 6G–I). Moreover, a subpopulation of PRLR⁺ macrophages could be found exclusively within the CD11c⁺/CD206⁺ macrophages (Fig. 6G–I). Interestingly, in KC mice, despite an overall decline in the number of CD11c⁺/CD206⁺ M1/M2 cells, the ratio of PRLR-expressing cells among resident macrophages had increased (Fig. 6H and I).

Together, our findings suggest the PRLR-expressing resident macrophages appear to have a more prominent presence in the pancreatic tumor microenvironment compared with noninjured or caerulein-treated pancreas.

Prolactin is produced by many tissues outside the pituitary, including the breast and prostate (23, 46). These findings suggest that prolactin participates in tumor growth, via a paracrine or autocrine mechanism (23, 46). Here, we report that in the pancreas, prolactin is expressed by TAMs. Thus, it appears that prolactin functions in a paracrine manner during individual stages of pancreatic tumorigenesis. Although it is well established that the pituitary prolactin secretion is regulated by the inhibitory action of dopamine (33, 47), the mechanisms behind the extrapituitary prolactin expression or release remain poorly understood. In experimental pancreatitis, damaged/dead acinar cells or activated macrophages release HMGB1 into the extracellular space, where it triggers an innate immune response (42). Given that we observed reduced prolactin in macrophages in KCH mice in which HMGB1 is conditionally knocked out in the pancreatic epithelium, it is likely that the main source for HMGB1 in this context may be damaged/dead neoplastic cells. Regardless of the origin, HMGB1-mediated prolactin expression in the macrophages may be part of the nascent immune response.

The physiologic functions of prolactin are facilitated through its cognate receptor, which we found to be expressed in premalignant and malignant lesions in the human and murine KC pancreas. However, PRLR was expressed as early as in PanIN1s in mice, whereas in patient samples, we first detected PRLR only in more advanced PanINs. The difference in the onset of PRLR expression may reflect the differences in species.

In the current study, we provide evidence that prolactin plays a key role in PanIN progression. Our conclusion is based on gain-of-function (pregnancy or metoclopramide treatment) and loss-of-function (KC;Prl^−/− mice) approaches, where we found an interesting relationship between PanIN acceleration and prolactin levels. Cytokines favor the expansion and development of noninvasive pancreatic precursor lesions (8, 9, 15). Accordingly, we could observe an overall increase in the number of pSTAT3 cells as the result of high systemic prolactin. Moreover, in KC;Prl^−/− mice, pSTAT3 declined both in PanINs and stromal cells. Interestingly, the status of pSTAT3 was unchanged in ADMs, suggesting that the ADM to PanIN transition coincides with a switch from a prolactin-independent to a prolactin-dependent mechanism for STAT3 activation. IL6 has been shown to stimulate STAT3 activity in PanINs and myeloid cells (10, 11, 36). Thus, although inactivation of Stat3 in KC mice prevents ADM to PanIN transformation (11), in KC;Prl^−/− pancreas, PanINs are initially formed, but they fail to advance further.

An intriguing finding in this study was the effect of prolactin on fibrogenesis and collagen deposition. We have found that the number of pSTAT3⁺ and pFAK⁺ stromal cells correlate with prolactin levels. Furthermore, we have identified a subpopulation of embryonic-derived pancreatic resident macrophages that express PRLR during tumorigenesis. These findings indicate that prolactin (i) regulates STAT3 and FAK1 phosphorylation in PanINs and a subset of stromal cells, and (ii) induces profibrotic activities in resident TAMs. Given the oncogenic nature of the PRL/PRLR pathway, we were surprised to find that patients with high PRLR expression displayed better survival than those with lower PRLR levels. However, this can be partially explained by the fact that the PRLR low-expressing cells belong to the squamous class, which is generally associated with poor survival (34). Moreover, the extent of PRLR-expressing resident macrophages that can respond to prolactin may also contribute to the overall survival of the patients.

In summary, we propose that extracellular HMGB1, either passively released by damaged/dying neoplastic cells, or actively secreted by TAMs, stimulates prolactin expression by macrophages (Fig. 7A). The macrophage-derived prolactin may promote fibrosis through PRLR-expressing resident macrophages. In addition, prolactin binds to its cognate receptor on PanIN cells, where it maintains FAK1 and STAT3 activity. Because collagen density or stiffness could also result in FAK1 activation in normal or malignant cell types including the pancreas (38, 48), it is possible that the effect of prolactin on pFAK1 in PanINs is the result of the overall desmoplastic status of the environment. The canonical PRL/PRLR signaling may result in phosphorylation of cSrc or JAK2, which in turn can directly activate FAK1 and STAT5, respectively (23). JAK2 can also activate STAT3, although not directly, but through JAK1 as an intermediate (Fig. 7B; ref. 23). Our study shows that STAT5 is active in pancreatic cancer, but its activity is not dependent on PRL/PRLR, as the current literature would suggest. Instead, prolactin appears to regulate STAT3 activity (Fig. 7C). This finding is clinically relevant, as it would suggest that inactivation of STAT3 in pancreatic cancer patients could potentially include IL6 in conjunction with prolactin as therapeutic targets.

We have demonstrated that the source of prolactin in pancreatic tumorigenesis is locally derived within the pancreas. Nevertheless, given the relatively higher systemic prolactin in females, one would expect a higher PDAC incidence among women compared with men. It has been suggested that it may take up to 20 years from the time of overt tumor initiation to the time of a patient's death (49). The average age for PDAC diagnosis in both sexes is in the mid-70s, which implies that the average age for tumor initiation should be above 50 years. Clearly, in individuals younger than 50 years of age, prolactin concentrations are higher in women than men (47, 50); however, postmenopausal women (typically over 50 years old) exhibit a 40% decrease in prolactin secretion, whereas men above age 50 display an 18% increase (47, 51). Thus, the differences in systemic prolactin between sexes decline around the putative time of pancreatic tumor initiation.

Prolactin is not a carcinogen, but it appears to be required for the normal progression of the disease. Nonetheless, high levels of prolactin in conditions such as obesity (52, 53) may accelerate the oncogenic events. Together, these findings indicate that prolactin signaling is a key mediator of the cross-talk between the stroma and the neoplastic epithelium. Further studies are required to determine whether it may represent an intriguing and novel target for therapy in this disease.

No potential conflicts of interest were disclosed.

Conception and design: M. Tandon, A. Criscimanna, M.T. Lotze, H. Zeh, G.K. Gittes, F. Esni

Development of methodology: A. Criscimanna, M. Socorro, Z. Cruz-Monserrate, M.T. Lotze

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): G.M. Coudriet, A. Criscimanna, M. Socorro, Z. Cruz-Monserrate, A.V. Biankin

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Tandon, G.M. Coudriet, A. Criscimanna, M. Socorro, M. Eliliwi, A.D. Singhi, P. Bailey, M.T. Lotze, J. Hu, F. Esni

Writing, review, and/or revision of the manuscript: M. Tandon, Z. Cruz-Monserrate, M.T. Lotze, H. Zeh, V. Goffin, A.V. Biankin, F. Esni

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M. Eliliwi, V. Goffin, A.V. Biankin

Study supervision: M.T. Lotze, F. Esni

This study was supported by NIH/NIDDK grant DK101413 (to F. Esni); National Pancreas Foundation (to F. Esni); The Children's Hospital of Pittsburgh of UPMC (to F. Esni); Cancer Research UK C29717/A17263, C29717/A18484, C596/A18076, C596/A20921, and A23526 (to A.V. Biankin); Wellcome Trust Senior Investigator Award 103721/Z/14/Z (to A.V. Biankin); Pancreatic Cancer UK Future Research Leaders Fund FLF2015_04_Glasgow (to A.V. Biankin); Scottish Genome Partnership SEHHD-CSO 1175759/2158447 (to A.V. Biankin); MRC/EPSRC Glasgow Molecular Pathology Node (to A.V. Biankin); and The Howat Foundation (to A.V. Biankin). We acknowledge the Comparative Pathology and Mouse Phenotyping Shared Resource (CPMPSR) of The Ohio State University Comprehensive Cancer Center for excellent immunohistochemical (Alan Flechtner and Florinda Jaynes) support. The work was supported in part by Cancer Center Support Grant P30 CA016058, National Cancer Institute, Bethesda, Maryland. We would like to thank Joshua Michel for helping us with the flow cytometry analyses.

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.