Pancreatic ductal adenocarcinoma (PDAC) is a leading cause of cancer-related deaths worldwide, with an exceedingly low 5-year survival rate. PDAC tumors are characterized by an extensive desmoplastic stromal response and hypovascularity, suggesting that tumor hypoxia could regulate PDAC initiation and/or progression. Using a well-defined, autochthonous KrasG12D-driven murine model, as well as human tumors, we demonstrate that hypoxia and stabilization of hypoxia-inducible factor 1α (HIF1α), a principal mediator of hypoxic adaptation, emerge early during preinvasive stages of PDAC. Surprisingly, pancreas-specific Hif1a deletion drastically accelerated KrasG12D-driven pancreatic neoplasia and was accompanied by significant increases in intrapancreatic B lymphocytes, featuring prominent influx of a rare “B1b” B-cell subtype. Finally, treatment of HIF1α-deficient mice with B cell–depleting αCD20 monoclonal antibodies inhibited progression of pancreatic intraepithelial neoplasia (PanIN). Our data reveal a previously unrecognized role for B cells in promoting pancreatic tumorigenesis and implicate HIF1α as a critical regulator of PDAC development.
Significance: We show here that pancreas-specific Hif1a deletion promotes PDAC initiation, coincident with increased intrapancreatic accumulation of B cells, and that B-cell depletion suppresses pancreatic tumorigenesis. We therefore demonstrate a protective role for HIF1α in pancreatic cancer initiation and uncover a previously unrecognized function of B cells. Cancer Discov; 6(3); 256–69. ©2015 AACR.
See related commentary by Roghanian et al., p. 230.
See related article by Pylayeva-Gupta et al., p. 247.
See related article by Gunderson et al., p. 270.
This article is highlighted in the In This Issue feature, p. 217
Pancreatic ductal adenocarcinoma (PDAC) is one of the most lethal human cancers, with a 5-year survival rate of 6% (1). PDAC arises predominantly through the sequential progression of pancreatic intraepithelial neoplasia (PanIN; ref. 2). Activating KRAS mutations are found in more than 90% of PDAC cases, from the earliest stage PanIN-1 lesions to invasive adenocarcinoma (3, 4). Pancreas-specific expression of mutated Kras in mice results in PanIN formation that eventually progresses to PDAC (5), supporting the prominent role of oncogenic KRAS in its genesis. In addition to the unusually high prevalence of KRAS mutations, a distinguishing feature of PDAC is an extensive inflammatory, desmoplastic stromal reaction contributing to a hypovascular and hypoxic microenvironment (6, 7).
Hypoxia, a condition of insufficient oxygen (O2) availability, is a critical feature of the tumor microenvironment. To cope with hypoxic stress, cells activate numerous adaptive responses, including a large transcriptional program primarily coordinated by hypoxia-inducible factors (HIF; ref. 8). At normal levels of O2, HIF1α is rapidly degraded via the ubiquitin-proteasome pathway, whereas hypoxia results in its reversible accumulation. Once stabilized, HIF1α transactivates a wide range of genes involved in the regulation of metabolism, angiogenesis, cell survival, and inflammation (9, 10). Elevated HIF1α expression is associated with increased patient mortality in many cancer types, including breast and colorectal cancers. However, in certain malignancies, HIF1α accumulation actually correlates with lower cancer stage or decreased patient mortality, implicating opposing, context-dependent functions for HIF1α (11).
Though hypoxia and consequent HIF1α stabilization in human PDAC have been observed (12, 13), previously published reports have provided inconsistent results on correlations between HIF1α expression and clinical outcomes in PDAC (13–15). Therefore, the precise role of HIF1α in PDAC pathogenesis is not fully understood. Moreover, previous studies to date have relied on RNA interference of HIF1α in in vitro assays or xenograft tumor models (15, 16), which exclude the complexity of the tumor microenvironment, including dynamic changes of O2 pressure and immune cell responses.
In the present study, we sought to identify the functional role of HIF1α in pancreatic tumorigenesis using the well-characterized p48-Cre;LSL-KrasG12D autochthonous mouse model of PDAC. We demonstrate that hypoxia and HIF1α stabilization occur at very early stages of disease in both humans and mice, and that pancreatic epithelial Hif1a deletion surprisingly accelerates pancreatic carcinogenesis, concomitant with increased infiltration of B lymphocytes. B cells are an important, albeit understudied, leukocyte subset in tumor microenvironments, and have recently been shown to affect squamous cell carcinoma and prostate cancer (17–21). Our findings now identify B cells as a key immune component of pancreatic cancer and provide insights into the interplay between O2 homeostasis and immune responses during pancreatic tumorigenesis.
Hypoxia and HIF1α Accumulation Occur during Early Pancreatic Neoplasia
To investigate the role of HIF1α in pancreatic tumorigenesis, we first examined the status of hypoxia and HIF1α expression in this disease. Although human and murine PDAC have been shown to be hypoxic (12, 22, 23), it is not known at what point in pancreatic tumorigenesis hypoxic microenvironments arise. Because a prominent immune infiltration is observed even around the lowest-grade PanINs (24) and inflammation is often associated with tissue hypoxia (25), we determined whether pancreata experience hypoxia during preinvasive stages by utilizing a mouse model of PDAC. p48-Cre;LSL-KrasG12D mice (henceforth KrasG12D) express a constitutively active form of mutated Kras in the pancreas and develop PanINs that progress through all the histologic stages described for the human disease (5). At age 2 months, KrasG12D mice retain normal pancreatic tissue structure and display only sporadic PanINs (5). Hypoxyprobe, an indicator of pO2 levels ≤ 1%, and HIF1α protein were barely detectable in wild-type (WT) samples, whereas pronounced increases in Hypoxyprobe+ cells and HIF1α expression were observed in pancreatic tissue from 2-month-old KrasG12D mice (Fig. 1A). Notably, strong Hypoxyprobe and HIF1α immunostainings were mostly restricted to PanINs and fibroinflammatory areas (Fig. 1A). Moreover, hypoxia and HIF1α accumulation persisted throughout development of invasive carcinoma (Fig. 1A). Of note, 18 of 24 (75%) human PDAC patient samples were also positive for HIF1α staining, and HIF1α expression was detected in PanINs as well as in invasive PDAC lesions (Fig. 1B), suggesting a pathophysiologic relevance of HIF1α modulation during pancreatic oncogenesis. Thus, hypoxic microenvironments emerge early during pancreatic tumorigenesis, implying that O2 limitation and consequent HIF1α stabilization might be involved in disease evolution.
Hif1a Deletion Promotes KrasG12D-Induced Pancreatic Tumorigenesis
To determine whether HIF1α is functionally relevant during pancreatic tumorigenesis, we first generated pancreatic epithelium-specific HIF1α-deficient mice by crossing p48-Cre animals (26) with Hif1afl/fl alleles (27), resulting in the genotype p48-Cre;Hif1afl/fl. Lineage-tracing studies previously demonstrated uniform expression of Cre recombinase throughout the pancreas in p48-Cre mice (26), and PCR analysis verified efficient Cre-dependent recombination of Hif1afl/fl loci in pancreata (Supplementary Fig. S1A). Although global Hif1a deletion results in embryonic lethality (28, 29), p48-Cre;Hif1afl/fl mice were born at expected Mendelian ratios with histologically normal pancreatic tissue (Fig. 2A). Glucose tolerance tests and monitoring of body and pancreas weight showed no difference between mutant and control samples (Supplementary Fig. S1B–S1D), indicating that HIF1α is largely dispensable for the proper development of mouse pancreas.
We then generated p48-Cre;LSL-KrasG12D;Hif1afl/fl mice (henceforth KrasG12D;Hif1aKO), which allow simultaneous activation of mutant Kras and Hif1a deletion exclusively in pancreatic epithelium. Efficient Hif1a deletion in KrasG12D;Hif1aKO pancreata was confirmed by PCR and quantitative RT-PCR analyses (Supplementary Fig. S2A and S2B). KrasG12D;Hif1aKO mice were born at the expected frequency, but histologic analysis (and Alcian blue staining for mucins) of pancreatic tissue from 2-month-old KrasG12D;Hif1aKO mice revealed a significant impact of HIF1α deficiency on neoplastic progression. KrasG12D control mice demonstrated predominantly focal low-grade PanINs surrounded by abundant normal pancreas area (Fig. 2B and C). In striking contrast, pancreata from KrasG12D;Hif1aKO mice displayed a marked enhancement of both PanIN number and grade (Fig. 2B and C), and the acinar parenchyma was largely replaced by an intense stromal reaction (Fig. 2D). Acceleration of PanIN progression was also observed in 5-month-old KrasG12D;Hif1aKO mice (Supplementary Fig. S2C). These data clearly demonstrate that loss of HIF1α actually facilitates the initiation and neoplastic progression of KrasG12D-driven PanINs, and implicate HIF1α accumulation during PDAC evolution as a protective mechanism against hypoxic stress to maintain tissue homeostasis and integrity, and limit tumor initiation.
Consistent with PanIN progression upon Hif1a deletion, KrasG12D;Hif1aKO pancreata exhibited pronounced increases in cell proliferation, based on Ki67 immunostaining (Fig. 2E). Apoptosis was a rare event in both KrasG12D and KrasG12D;Hif1aKO samples (Fig. 2E). Of note, although HIF1α is a well-known promoter of angiogenesis (29–31), pancreatic epithelial Hif1a deletion affected neither vascular density nor Vegfa expression during early pancreatic neoplasia (Fig. 2E and F). This might be due to preferential regulation of Vegfa expression through the related HIF2α isoform rather than HIF1α, as previously shown in liver hemangiomas and teratomas (32, 33). Both KrasG12D and KrasG12D;Hif1aKO pancreatic tissues stained positively for Hypoxyprobe and HIF2α, with both being more pronounced in KrasG12D;Hif1aKO pancreata compared with KrasG12D counterparts, consistent with prominent increases in PanINs/fibroinflammatory areas in KrasG12D;Hif1aKO pancreatic tissues (Supplementary Fig. S2D). Taken together, our data suggest that HIF1α loss results in enhanced cell proliferation during KrasG12D-driven PanIN progression, and that HIF2α cannot compensate for its activity in this context.
To determine whether HIF1α influences later stages of PDAC progression, we aged a large cohort of KrasG12D and KrasG12D;Hif1aKO mice. Despite the profound differences in PanIN progression between the KrasG12D and KrasG12D;Hif1aKO animals, there were no significant differences in survival time, tumor incidence, tumor weight, and metastasis rates (Supplementary Fig. S3A–S3C). Compared with KrasG12D tumors, KrasG12D;Hif1aKO tumors exhibited decreased expression of Ldha and Pgk1, well-known HIF1α target genes critical for tumor metabolism (Supplementary Fig. S3D). Thus, inability of KrasG12D;Hif1aKO tumors to induce HIF1α in response to intratumoral hypoxia might ultimately decelerate tumor progression by hindering metabolic reprogramming, and eventually negate the pro-neoplastic effects of Hif1a deletion on PDAC initiation (see below for further discussion on this topic). Interestingly, as observed in early pancreatic neoplasia, Vegfa levels were also not significantly different between KrasG12D and KrasG12D;Hif1aKO tumors (Supplementary Fig. S3D). Although there was no difference in overall survival between the KrasG12D and KrasG12D;Hif1aKO groups, it should be noted that a significantly higher fraction of KrasG12D;Hif1aKO tumors contained areas of poorly differentiated or undifferentiated histology compared with KrasG12D tumors (7%, 1 of 14 for KrasG12D versus 47%, 9 of 19 for KrasG12D;Hif1aKO, P = 0.02; Supplementary Fig. S3E), implicating HIF1α deficiency in restraining PDAC differentiation.
Elimination of HIF1α Promotes Intrapancreatic Accumulation of B Cells during Early Pancreatic Neoplasia
Given the compelling evidence for a strong link between inflammation and PDAC development (34), as well as our finding that Hif1a deletion results in a marked increase in CD45+ immune cell infiltration (Figs. 2D and 3A), we set out to characterize immunomodulation by HIF1α during KrasG12D-induced pancreatic oncogenesis. Flow cytometry analysis of pancreatic tissue for the identification of leukocyte subsets revealed that both the percentage and absolute number of B cells were dramatically increased in KrasG12D;Hif1aKO pancreata compared with their KrasG12D counterparts (Fig. 3B and C). Absolute numbers of other leukocyte populations, including those of macrophages, Gr1+CD11b+ cells [often called myeloid-derived suppressor cells (MDSC)], dendritic cells, and T cells, were also increased (Supplementary Fig. S4A); however, their percentages among CD45+ immune cells were not significantly changed (Fig. 3B). By immunohistochemistry, B cells were evident in close proximity to PanIN lesions in KrasG12D;Hif1aKO mice, confirming a role for HIF1α in opposing intrapancreatic B-cell recruitment (Fig. 3D). As expected, fibroinflammatory areas with infiltrated B cells were positive for Hypoxyprobe staining (Supplementary Fig. S4B). Significantly, human pancreatic cancers (17 of 24 samples) were also associated with intrapancreatic B-cell infiltration (Fig. 3E), supporting the potential importance of B cells as regulators of pancreatic tumorigenesis. Interestingly, B-cell numbers in KrasG12D pancreata did not increase, but rather decreased during the course of carcinogenesis (Supplementary Fig. S4C), suggesting that increased B-cell accumulation in KrasG12D;Hif1aKO pancreata compared with tissue from KrasG12D mice is not merely a reflection of disease progression by Hif1a deletion.
Mice bearing PDAC have been reported to develop splenomegaly (35). We noted that KrasG12D;Hif1aKO mice develop enlarged spleens (Fig. 3F and G), consistent with advanced disease status of KrasG12D;Hif1aKO pancreata. KrasG12D;Hif1aKO mice also exhibited an increase in the absolute number of splenic CD19+ B cells (Fig. 3H). These data suggest that HIF1α deficiency during KrasG12D-initiated pancreatic tumorigenesis elicits a systemic immune reaction.
Characterization of B-cell Subsets in KrasG12D and KrasG12D;Hif1aKO Pancreata
B-cell subset characterization has not been previously reported in the context of pancreatic tumor development; we therefore evaluated the distribution of B-cell subpopulations in pancreata from WT, KrasG12D, and KrasG12D;Hif1aKO mice by flow cytometry. Mature peripheral B cells can be divided into two main subsets, i.e., conventional B2 (CD19+CD43−CD5−) and B1 (CD19+CD43+IgMhi) cells, which can be further subdivided into B1a (CD19+CD43+IgMhiCD5+) and B1b (CD19+CD43+IgMhiCD5−) cells based on CD5 expression (Supplementary Fig. S5A; refs. 36, 37). As shown in Fig. 4A, B2 cells were the prevalent B-cell subset in mice of all three genotypes. However, KrasG12D pancreata showed diminished frequencies of B2 cells and elevated frequencies of B1 cells compared with WT pancreata (Fig. 4A and B). Furthermore, decreases in B2 and concomitant increases in B1 cell frequencies appeared to be even greater in KrasG12D;Hif1aKO pancreata in comparison to both WT and KrasG12D tissues (Fig. 4A and B). Interestingly, the significant elevation in B1 cell proportion was mostly due to an increase in B1b cells, whereas changes in B1a frequencies were minimal (Fig. 4B). Of note, a similar pattern of alterations in the relative proportions of B1 and B2 lineages was observed in the spleen (Supplementary Fig. S5B). These findings suggest that the composition of B1 versus B2 subsets is altered during KrasG12D-driven pancreatic tumorigenesis, and that Hif1a deletion cooperates with oncogenic Kras to modulate the balance between B1 and B2 cells.
Regulatory B cells (Breg), a rare CD19hiCD1dhiCD5+ B-cell subset, have been implicated in suppression of antitumor immunity (38). Importantly, both KrasG12D and KrasG12D;Hif1aKO pancreata exhibited higher frequencies of Bregs compared with WT pancreata (Fig. 4C). However, no significant differences were detected in Breg frequencies between KrasG12D and KrasG12D;Hif1aKO pancreata (Fig. 4C), indicating that the expansion of Bregs is regulated by Kras activation, but not HIF1α status.
B-cell Depletion Restrains Progression of PanINs and Development of Invasive Carcinoma
Tumor-promoting effects of B cells have been reported in certain malignancies, such as squamous cell carcinoma and prostate cancer (17–21). Because enhanced PanIN progression upon Hif1a deletion is accompanied by increased B-cell infiltration to sites of neoplasia, we evaluated the function of B cells during pancreatic tumorigenesis and significance of B-cell recruitment in the context of HIF1α deficiency. We selectively depleted B cells in vivo by administering a CD20-specific monoclonal antibody (αCD20 mAb) to KrasG12D and KrasG12D;Hif1aKO mice. Greater than 95% of all B cells in the pancreas, spleen, and peripheral blood were depleted in both KrasG12D and KrasG12D;Hif1aKO mice following αCD20 mAb injection (Fig. 5A–C). Importantly, treatment with αCD20 mAbs reduced the number of PanIN-3 in both KrasG12D and KrasG12D;Hif1aKO mice (Fig. 5D and E), and significantly decreased the fraction of KrasG12D;Hif1aKO animals with microinvasive lesions (Fig. 5F and Supplementary Fig. S6). Interestingly, isotype control IgG2a itself accelerated PanIN progression as determined by earlier detection of PanIN-2 and PanIN-3 in IgG2a-treated, 12-week-old KrasG12D mice (Fig. 5E), as opposed to such advanced lesions being largely undetected in untreated, 5-month-old KrasG12D mice (Supplementary Fig. S2C). A positive effect of IgG on PanIN progression would be consistent with promotion of squamous carcinoma by immunoglobulins, as previously shown (17, 18, 20). B-cell depletion with αCD20 mAbs was associated with a significant increase in the percentage of intrapancreatic T cells as well as a decrease in CD45+ immune infiltration; however, percentages of myeloid populations (macrophages, MDSCs, and dendritic cells) were not affected by B-cell depletion (Supplementary Fig. S7). Alterations of regulatory T cell (Treg) differentiation and macrophage polarization have been proposed as protumorigenic functions of B cells (18, 20, 39, 40); however, B-cell depletion did not change the frequency of Tregs (Foxp3+CD25+CD4+), or the relative proportions of M1-like (CD86+F4/80+CD11b+) and M2-like (CD206+F4/80+CD11b+) macrophages in this context (Supplementary Fig. S7). Of note, IgG2a-treated KrasG12D;Hif1aKO mice exhibited lower frequencies of intrapancreatic T cells and M2-like macrophages compared with IgG2a-treated KrasG12D mice (Supplementary Fig. S7), raising the possibility that these immune subsets could also be affected by epithelial HIF1α expression. Together, these findings strongly suggest that B cells are critical immune components of pancreatic cancer and that HIF1α deficiency accelerates pancreatic tumorigenesis, at least in part, by promoting B-cell recruitment.
HIF1` Deficiency Results in Increased Secretion of the B-cell Chemoattractant CXCL13
In order to determine the mechanisms and factors through which HIF1α ablation facilitates recruitment of B cells into the pancreas, we first assessed the production of CXCL13, a key B-cell chemoattractant, in WT, KrasG12D, and KrasG12D;Hif1aKO pancreata by immunohistochemical staining of tissue sections and ELISA analysis of tissue supernatants. As shown in Fig. 6A and B, CXCL13 protein levels were elevated in KrasG12D;Hif1aKO pancreata relative to WT and KrasG12D samples. Cxcl13 mRNA was also increased in KrasG12D;Hif1aKO pancreatic tissues (Fig. 6C). It should be noted that CXCL13 secretion measured by ELISA was comparable between WT and KrasG12D pancreata, whereas relative to WT, KrasG12D samples exhibited slight but detectable increases in Cxcl13 mRNA (assessed by quantitative RT-PCR) as well as CXCL13 protein (based on immunohistochemical staining; Fig. 6A–C). Although this implies differential regulation of CXCL13 expression and secretion by oncogenic KRAS, it may simply be the result of variable assay sensitivities instead (ELISA versus quantitative RT-PCR or immunohistochemistry). Importantly, human PDAC patient samples (19 of 24) were positive for CXCL13 protein accumulation (Fig. 6D). On the other hand, CXCL13 levels in mouse PDAC cell cultures were not affected by hypoxia or Hif1a knockdown, suggesting that CXCL13 regulation by HIF1α in pancreatic tissues occurs predominantly through a non–cell-autonomous mechanism (Supplementary Fig. S8A). CCL19, CCL20, CCL21, and CXCL12 are also known to influence B-cell migration (41, 42), and all four chemokines were overexpressed in KrasG12D;Hif1aKO pancreata (Fig. 6E). An array analysis with pancreatic tissue supernatants showed no significant reproducible changes in production of 25 other cytokines and chemokines upon HIF1α elimination, implicating preferential modulation of B cell–recruiting chemokines by HIF1α ablation (Supplementary Fig. S8B). Collectively, these results suggest that Hif1a deletion promotes pancreatic B-cell infiltration by indirectly upregulating the expression of chemokines involved in B-cell trafficking.
Hypoxia is a general property of solid tumor microenvironments and influences numerous aspects of tumor biology and response to therapy (11, 43). Pancreatic cancer in particular is defined by drastic alterations of the microenvironment (44); however, the effects of hypoxia on pancreatic tumorigenesis in vivo have remained largely unknown. Using a Kras-driven mouse model of PDAC and human patient samples, we found that hypoxia and accumulation of HIF1α, a master regulator of the hypoxic transcriptional response, occur much earlier in disease progression than previously appreciated. Hif1a deletion unexpectedly led to accelerated KrasG12D-initiated neoplasia, accompanied by elevated intrapancreatic recruitment of B cells. Recent studies have observed tumor-promoting effects of B cells in certain cancers (17–21); however, no functional studies of B cells have been reported so far with regard to PDAC. This led us to ask whether B cells contribute to pancreatic tumorigenesis. Indeed, B-cell depletion in autochthonous PDAC mouse models revealed that B cells promote PanIN progression, and that HIF1α regulates pancreatic oncogenesis at least in part through opposing B-cell recruitment. Consistent with our results, two accompanying reports (45, 46) utilizing alternative murine models and/or stages of PDAC from the ones used here also concluded that B cells clearly promote pancreatic tumorigenesis. Moreover, Gunderson and colleagues identified B cells as one of the predominant leukocyte populations in human PDAC (46). These complementary data carry important clinical implications for targeting B cells as part of overall pancreatic cancer treatment.
Our study shows that pancreatic epithelial HIF1α affects B-cell subset distribution, as well as B-cell infiltration, in the context of KrasG12D-induced pancreatic neoplasia. Epithelial expression of HIF1α has not been previously linked to differentiation and migration of B cells. We observed that pancreatic Hif1a deletion elevates expression of multiple chemokines involved in B-cell migration within the tumor microenvironment. Of note, CXCL13, one of the chemokines upregulated by HIF1α deficiency, has previously been reported to recruit B cells into prostate tumors (19). Moreover, Pylayeva-Gupta and colleagues (45) showed that CXCL13 neutralization reduces pancreatic B-cell infiltration, suggesting that CXCL13 represents a crucial pancreatic B-cell chemoattractant. At present, the precise mechanism for chemokine overexpression upon Hif1a deletion remains unknown and is likely to involve complex interactions between epithelial and stromal cells, which warrant further study. We also found that HIF1α deficiency accompanies an elevated proportion of B1 cells during KrasG12D-driven pancreatic tumorigenesis. B1 cells are considered innate immune cells that produce the majority of natural antibodies against a broad spectrum of infections (47). However, very little is known about B1 cells in the context of oncogenesis. Our finding that PanIN progression coincides with a selective increase in B1 cell frequency implicates protumorigenic functions for them. The effects of B1 cells on pancreatic tumorigenesis and mechanisms responsible for elevated proportions of B1 lymphocytes, especially B1b cells, in the pancreas remain to be determined. Interestingly, lymphoid Hif1a deletion results in the appearance of abnormal peritoneal, B1-like cells (48), raising the intriguing possibility that regulation of B1-to-B2 ratios by HIF1α in lymphoid and epithelial cells might share a common mechanism. It is noteworthy that Pylayeva-Gupta and colleagues (45) found that IL35-producing CD1dhiCD5+ B cells are also critical for orthotopic early-stage pancreatic neoplastic growth, and Gunderson and colleagues (46) observed a correlation between decreased IgMloCD23+CD5− follicular and IgMloCD23− memory B cells and improved outcome in orthotopic late-stage tumors upon BTK inhibitor/Gemcitabine treatment. These results, along with increased numbers of CD43+IgMhiCD5− B1b cells detected in KrasG12D;Hif1aKO PanIN lesions, suggest that multiple B-cell subsets are important in PDAC and may differ depending on tumor stage, genetic profile, drug treatment, and hypoxic adaptation.
The dynamic interplay between cells and their microenvironment evolves during the multistep development of human tumors (49, 50). The severity and duration of hypoxic stress that neoplastic and stromal cells experience are progressively altered over the course of cancer progression (11). Dramatic stromal expansion and accelerated tumorigenesis by epithelial Hif1a deletion may not be too surprising, when one considers HIF1α as a protein that maintains tissue homeostasis in response to hypoxic stress. Failure to activate HIF1α signaling as O2 levels decline in neoplastic development could hasten disruption of systemic homeostatic responses, especially B-cell homeostasis in the context of PDAC initiation, and consequently exacerbate tumorigenesis. On the other hand, in full-blown PDAC with limiting O2 and nutrients, we speculate that HIF1α may confer survival and growth advantages to pancreatic cancer cells by facilitating metabolic reprogramming. This might explain HIF1α overexpression found in human PDAC and counterbalance protective effects of HIF1α on tumor initiation in murine PDAC. It is noteworthy that pancreatic deletion of the Hif2a isoform has been shown to inhibit PanIN progression to advanced lesions (51), revealing distinct functions of HIFα isoforms during PDAC pathogenesis. Even though Hif1a deletion in KrasG12D-driven PDAC did not change overall mouse survival, we noted that KrasG12D;Hif1aKO tumors more frequently contain areas of poorly differentiated or undifferentiated histology in comparison with KrasG12D tumors, implicating HIF1α as an important determinant of tumor differentiation status. Given that subtypes of PDAC show distinct sensitivities to various drugs (52, 53), hypoxia and HIF1α may provide utility in optimizing subtype-specific therapies. Further studies based on a dual-recombinase system enabling sequential genetic manipulation of the pancreas (54) are warranted to precisely delineate the differential effects of HIF1α on the initiation and progression to end-stage PDAC.
In summary, we discovered an unanticipated functional facet of HIF1α—the capacity to indirectly modulate B cells in the context of KrasG12D-initiated pancreatic neoplasia. HIF1α deficiency accelerates PDAC initiation through enhanced B-cell infiltration. We show that B-cell depletion suppresses pancreatic tumorigenesis and suggest that B cells are crucial pro-neoplastic elements in pancreatic cancer. Our findings reinforce the significance of microenvironmental regulation of tumorigenesis and offer insights into potential efficacy of therapies for pancreatic cancer, targeting aberrant immune responses and O2 homeostasis.
All animal protocols were reviewed and approved by the Institutional Animal Care and Use Committee of the University of Pennsylvania (Philadelphia, PA). The Hif1afl/fl, LSL-KrasG12D, and p48-Cre strains have been previously described (26, 27, 55). Animals were on a mixed B6;129 genetic background. Animals were randomly included in the experiments according to genotyping results. No noninclusion or exclusion parameters were used in our studies. Unless otherwise stated, all experimental mice were a mix of male and female. The numbers of animals used per experiment are stated in the figure legends. No sample size calculations were performed. The researchers were not blinded to the experimental groups during in vivo treatments; however, data assessment was conducted in a blinded fashion. Mice received intraperitoneal injections with isotype control mouse IgG2a (C1.18.4, #BE0085, BioXcell) or mouse IgG2a anti-mouse CD20 mAb (5D2, a kind gift from Genentech) at 10 mg/kg at 2 weeks old and 5 mg/kg thereafter at 2-week interval until 10 weeks of age.
Histology and Immunohistochemistry
Pancreata were fixed in 4% paraformaldehyde/PBS (4°C, overnight) and processed for paraffin embedding. For histology, deparaffinized sections (5 μm) were stained with Harris hematoxylin and eosin (both from Sigma-Aldrich) followed by alcohol dehydration series and mounting. For Alcian blue staining, rehydrated paraffin sections were stained with 1% Alcian blue 8GX in 3% acetic acid (pH 2.5) for 30 minutes and counterstained with Nuclear Fast Red (Sigma-Aldrich). Collagen was stained using the Masson's Trichrome Kit (Sigma-Aldrich), and nuclei were counterstained with Weigert's iron hematoxylin (Sigma-Aldrich).
For immunohistochemistry, slides (5 μm) were deparaffinized, rehydrated, quenched in 0.6% hydrogen peroxide/methanol for 15 minutes, and boiled for 20 minutes in 10 mmol/L sodium citrate (pH 6.0) for antigen retrieval. Sections were blocked with 5% serum/1% BSA/0.5% Tween-20 for 1 hour. Slides were incubated with primary antibodies diluted in blocking buffer overnight at 4°C. Following the primary antibody, slides were incubated with biotinylated secondary antibodies followed by ABC solution and developed with 3,3′-diaminobenzidine (all from Vector Laboratories). Slides were counterstained with hematoxylin, dehydrated, and mounted with Permount (Fisher). Slides were examined on a Leica DM5000B microscope. The following antibodies were used: rat anti-B220 (RA3-6B2, #557390, 1:400; BD Biosciences), rat anti-CD19 (1D3, #553783, 1:200; BD Biosciences), rabbit anti-CD31 (#ab28364, 1:200; Abcam), rat anti-CD45 (30-F11, #550539, 1:200; BD Biosciences), rabbit anti-cleaved caspase-3 (#9661, 1:200; Cell Signaling Technology), goat anti-CXCL13 (#AF470, 1:50; R&D Systems), rabbit anti-HIF1α (#ab2185, 1:5,000; Abcam), rabbit anti-HIF2α (#ab199, 1:2,000; Abcam), and rabbit anti-Ki67 (#VP-K451, 1:4,000; Vector Laboratories).
Human Pancreas Specimen
Human pancreatic tissue sections (7 normal and 24 PDAC samples) were obtained from the Cooperative Human Tissue Network and handled at the University of Pennsylvania with the approval of its institutional review board committees. All tissues were collected with the donor being informed and giving consent, and anonymized prior to being provided to the investigators. Immunohistochemical analysis was performed as described above. The following antibodies were used: mouse anti-human CD20 (FB1, #555677, 1:1,000; BD Biosciences), goat anti-human CXCL13 (#AF801, 1:50; R&D Systems), and mouse anti-human HIF1α (HA111, #NB100-296, 1:250; Novus).
Histologic assessment of pancreatic tissue sections was blinded in order to ensure that the pathologists (E.L. Buza and A.C. Durham) who assessed the tissues did not know which treatment or genotype each sample belonged to. For quantification of PanIN lesions, 10 high power fields (HPF) in which pancreas tissue covered at least 90% of the entire field were analyzed per animal. At least 7 mice per genotype per treatment were analyzed, and data were represented as the average number of lesions per HPF.
Mice received injection intraperitoneally with 60 mg/kg of Hypoxyprobe (pimonidazole hydrochloride; Hypoxyprobe, Inc.) and were sacrificed 1.5 to 2 hours later. To detect the formation of pimonidazole adducts, pancreatic tissue sections (paraffin-embedded) were immunostained with Hypoxyprobe-1-MAb1 (Hypoxyprobe-1 Plus kit; Hypoxyprobe, Inc.) following the manufacturer's instructions.
Single-cell suspensions from the tissues were prepared as follows: spleens were mechanically disrupted, suspended in RPMI-1640 supplemented with 10% FBS and 2 mmol/L l-glutamine (FACS media), passed through a 70-μm strainer and treated with ACK Lysing Buffer (Lonza). Pancreata were minced into small fragments and incubated in collagenase solution (1 mg/mL collagenase V in RPMI-1640) at 37°C for 40 minutes. Dissociated cells were passed through a 70-μm cell strainer and washed twice in FACS media. Blood samples were obtained via retro-orbital bleeding and treated with ACK Lysing Buffer.
Cells were stained in PBS/0.5% FBS/2 mmol/L EDTA with the following fluorochrome-conjugated antibodies: PE-conjugated anti-CD1d (1B1, #553846, 1:100), V450-conjugated anti-CD3 (500A2, #560801, 1:100), APC-Cy7–conjugated anti-CD4 (GK1.5, #552051, 1:200), APC-conjugated anti-CD5 (53-7.3, #561895, 1:50), PE-Cy7-conjugated anti-CD8 (53-6.7, #552877, 1:200), APC-conjugated anti-CD11b (M1/70, #553312, 1:200), APC-Cy7–conjugated anti-CD11b (M1/70, #561039, 1:50), V450-conjugated anti-CD11c (HL3, #560521, 1:100), APC-conjugated anti-CD19 (1D3, #550992, 1:200), PE-Cy7–conjugated anti-CD43 (S7, #562866, 1:100), PE-conjugated anti-CD45 (30-F11, #553081, 1:100), PE-Cy7–conjugated anti-CD45 (30-F11, #552848, 1:100), V500-conjugated anti-CD45 (30-F11, #561487, 1:100), APC-Cy7–conjugated anti-Gr1 (RB6-8C5, #557661, 1:200; from BD Biosciences); APC-conjugated anti-CD86 (GL1, #17-0862, 1:100), PE-conjugated anti-F4/80 (BM8, #12-4801, 1:100; from eBioscience); Pacific Blue–conjugate anti-CD19 (6D5, #A14905, 1:100; from Molecular Probes); FITC-conjugated anti-IgM (#115-097-020, 1:100; from Jackson ImmunoResearch); FITC-conjugated anti-CD206 (MR5D3, #MCA2235FT, 1:50; from AbD Serotec). The viability marker 7-aminoactinomycin D (7-AAD) was from BD Biosciences.
Staining for intracellular Foxp3 was performed using the Mouse Regulatory T Cell Staining Kit (FJK-16; eBioscience). Dead cells were excluded by staining with Fixable Viability Stain 450 (BD Biosciences). Flow cytometry was performed on a FACSCanto flow cytometer (BD Biosciences), and data were analyzed using FlowJo software.
Fresh Tissue Supernatant Collection and ELISA
Pancreata were harvested, washed in PBS, weighed, and minced into small fragments with sterile scissors. Tissue supernatant was harvested by adding complete media (400 μL/100 mg tissue) and compressing tissue fragments with the plunger of a 3-mL syringe. The tissue solution was centrifuged at 18,000 × g for 7 minutes at 4°C, and the supernatant was carefully collected. CXCL13 protein levels in tissue supernatant were quantified by mouse CXCL13 Quantikine ELISA kit (R&D Systems). Milliplex Map Mouse Cytokine/Chemokine Magnetic Bead Panel–Immunology Multiplex Assay (#MCYTOMAG-70K-PMX; Millipore) was used to quantify cytokines and chemokines in tissue supernatant.
Total RNA was isolated from pancreatic tissues using the RNeasy Mini Kit (Qiagen). cDNA was synthesized using a High-Capacity RNA-to-cDNA Master Mix (Applied Biosystems). PCR reactions were performed using TaqMan Universal PCR regents mixed with indicated cDNAs and TaqMan primers in a ViiA7 Real-Time PCR system (Applied Biosystems). Expression levels were normalized by 18S rRNA.
The 4662 PDAC cell line was derived from single-cell suspensions of PDAC tissue from Pdx1-Cre;LSL-KrasG12D;Trp53LSL-R172H/+ mice as previously described (56), and its genotype was verified. Cells were tested by using the Infectious Microbe PCR Amplification Test (IMPACT) and authenticated by the Research Animal Diagnostic Laboratory (RADIL) at the University of Missouri. Cells were cultured in DMEM plus 10% FBS.
Western Blot Analysis
Cells were lysed with 10 mmol/L Tris at pH 7.5, 150 mmol/L NaCl, 5 mmol/L EDTA, 0.1% SDS, and protease/phosphatase inhibitor cocktail (Cell Signaling). Cell lysates were separated by SDS-PAGE, transferred to nitrocellulose membranes, blotted with primary antibodies overnight at 4°C, and detected using horseradish peroxidase–conjugated secondary antibodies (Abcam, Cell Signaling Technology) followed by exposure to enhanced chemiluminescence reagents (Pierce). The following antibodies were used: rabbit anti-HIF1α (#10006421, 1:500; Cayman), goat anti-CXCL13 (#AF470, 1:2000; R&D Systems), and mouse anti-Histone H3 (#3638, 1:1000; Cell Signaling Technology).
Data are presented as the means with their respective standard errors or as statistical scatter plots. A Mann–Whitney test was performed for all statistical analyses unless otherwise specified using GraphPad Prism. P value of <0.05 was considered statistically significant.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: K.E. Lee, L.J. Bayne, D. Allman, M.C. Simon
Development of methodology: K.E. Lee, L.J. Bayne, E.L. Buza, A.C. Durham, D. Allman, M.C. Simon
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K.E. Lee, M. Spata, L.J. Bayne, R.H. Vonderheide, M.C. Simon
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K.E. Lee, L.J. Bayne, A.C. Durham, D. Allman, R.H. Vonderheide, M.C. Simon
Writing, review, and/or revision of the manuscript: K.E. Lee, R.H. Vonderheide, M.C. Simon
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K.E. Lee, M. Spata, L.J. Bayne, M.C. Simon
Study supervision: K.E. Lee, M.C. Simon
This work was supported by the Howard Hughes Medical Institute, a Pancreatic Cancer Action Network-AACR Innovative Grant (13-60-25-SIMO, to M.C. Simon), and NIH grant R01 CA169123 (to R.H. Vonderheide).
The authors are grateful to Genentech for providing αCD20 mAbs, and to T. Jacks, D.A. Tuveson, C.V. Wright, and R.S. Johnson for providing mice. They thank A.B. Troxel for advice on statistical analyses, L.P. Richman for technical support, and the members of the Simon laboratory for comments and discussions.
Note: Supplementary data for this article are available at Cancer Discovery Online (http://cancerdiscovery.aacrjournals.org/).
- Received July 6, 2015.
- Revision received December 18, 2015.
- Accepted December 22, 2015.
- ©2015 American Association for Cancer Research.