Lipocalin-2 Promotes Pancreatic Ductal Adenocarcinoma by Regulating Inflammation in the Tumor Microenvironment

Pancreatic ductal adenocarcinoma (PDAC) is a devastating disease with a dismal long-term survival (1, 2). It is projected to become the second leading cause of cancer-related deaths in the United States by 2030 due to both an aging population and the obesity epidemic (1, 3–5). Epidemiological data indicate that obesity adversely influences PDAC-related mortality in a dose-dependent manner (2). Moreover, a high BMI is associated with increased risk of PDAC, early age at onset, and decreased overall survival (6). Therefore, it is essential to elucidate the biology linking obesity and PDAC development. Lipocalin-2 (LCN2) is an adipokine elevated in the serum and visceral adipose tissue of obese individuals (7). It regulates adipose tissue inflammation, is a marker associated with obesity, and modulates insulin resistance (8–10). LCN2 is known to be secreted by adipocytes (11, 12), immune (neutrophils and macrophages), and tumor cells (13). Elevated serum LCN2 levels have been reported in ovarian, breast, colorectal, gastric cancer, and pancreas cancer (13–15). In PDAC, LCN2 expression levels increase in early pancreatic intraepithelial neoplasias (PanIN) lesions and increased expression correlates with progression (16–18). LCN2 is highly upregulated in serum and serves as a potential biomarker for patients with PDAC compared with those with chronic pancreatitis and normal healthy individuals (19–21). Efforts to investigate the biological role of LCN2 expression in PDAC have so far been inconclusive. In vitro and in vivo studies, using orthotopic and/or subcutaneous preclinical mouse models, have reported contradictory results with regards to the role of Lcn2 in PDAC growth, invasion, and metastasis (22–24). Therefore, given the mounting evidence linking LCN2 to obesity and inflammation, we hypothesized that LCN2 could link obesity and PDAC development.

Here we investigated the role of Lcn2 in a diet-induced obesity PDAC mouse model in which a genetically engineered mouse model (GEMM) was fed a high fat diet (HFD; ref. 25). We find that Lcn2-null mice expressing pancreas-specific oncogenic KRas developed fewer PanINs, less inflammation, and fibrosis, and displayed increased survival when fed a HFD. Moreover, we find that tumor growth was delayed and survival increased in Lcn2-null mice after injection of mouse PDAC cells expressing Lcn2. In addition, we showed that when human pancreatic cancer stellate cells (PSC), which are important regulators of the PDAC tumor microenvironment, are treated with Lcn2 it induced a proinflammatory response. This response was blocked after expression of the LCN2-specific receptor, solute carrier family 22, member 17 (SLC22A17) was downregulated.

All experiments were performed in compliance with the regulations and ethical guidelines for experimental animal studies of the Institutional Animal Care and Use Committee at the University of Texas MD Anderson Cancer Center (Houston, TX) and/or The Ohio State University (Columbus, OH).

KRas^G12D mice expressing conditional knock-in mutant KRas^GI2D were obtained from the Mouse Models of Human Cancer Consortium Repository (NIH, Bethesda, MD; ref. 26). KRas^G12D mice were bred with Ela-CreERT (CRE) mice described previously (27) mice to generate KRas^G12D/CRE. Lcn2 whole-body knockout (Lcn2^−/−) animals were kindly provided by Dr. Ralph Arlinghaus (MD Anderson Cancer Center; ref. 28) and bred with KRas^G12D/CRE mice to generate (Lcn2^−/−/KRas^G12D/CRE).

Animals at 40 days old were given tamoxifen orally for 3 days. Mice were fed either a CD (test diet DIO 58Y2; Lab Supply) in which 10% of energy was derived from fat (KRas^G12D/CRE, n = 48; Lcn2^−/−/KRas^G12D/CRE, n = 21; Lcn2^+/+, n = 24; Lcn2^−/−, n = 21) or a HFD (test diet DIO 58Y1 van Heek Series; Lab Supply), in which 60% of energy was derived from fat (KRas^G12D/CRE, n = 42; Lcn2^−/−/KRas^G12D/CRE, n = 36; Lcn2^+/+, n = 21; LCN2^−/−, n = 20). Two experimental groups were designed to test the effect of Lcn2 depletion and diet type either short-term after 50 days on diet or a long-term survival study. The pancreas, adipose tissue, and serum were collected to measure serum levels of LCN2 and other secreted molecules as well as for histologic and protein analysis. Body weight (BW) was measured weekly.

C57BL/6 strain mice were injected with cells derived from a pancreatic tumor of a PDX-1-CRE, LSL-KRas^G12D, LSL-Trp53^−/− (KPC) genetically engineered mouse model (GEMM) and transfected with enhanced firefly luciferase (KPC-LUC; refs. 29, 30). KPC-LUC cells in single suspension were prepared and mixed in HBSS and 20% Matrigel (BD Biosciences). A suspension of 0.5 × 10⁶ KPC-LUC cells was carefully implanted as previously described (30). Tumor growth was visualized by injecting mice with d-Luciferin (150 μg/mouse; Caliper Life Sciences) and using the IVIS Imaging System (Caliper Life Sciences; ref. 30). Animals (Lcn2^+/+n = 20 and Lcn2^−/−n = 17) were imaged weekly until euthanasia was required. Bioluminescence was measured using the Living Image software.

Pancreatic tissue was fixed with 10% formaldehyde, embedded in paraffin, sectioned, and stained with hematoxylin and eosin (H&E). Further immunohistochemical staining was performed in pancreatic paraffin sections using the antibodies and conditions listed in Supplementary Table S1. Collagen was detected using a Picro-Sirius Red Stain Kit (Abcam, ab150681).

Cell lines were cultured at 37°C with 5% CO₂ in DMEM with 4.5 g/L glucose and l-glutamine supplemented with 10% (v/v) FBS and tested negative for the presence of mycoplasma. BXPC3, HPAC, CAPAN2, MIAPACA2, and MPANC96 were obtained from ATCC. CAPAN1 cells were kindly provided by Drs. Collisson, Gray and McMahon (University of California and Lawrence Livermore Laboratory, San Francisco, CA). Human pancreatic ductal epithelial (HPDE) and human pancreatic normal epithelial cells (HPNE) were obtained from Dr. Timothy Eberlein (Washington University School of Medicine, St. Louis, MO) and Dr. Tsao (Ontario Cancer Institute, Toronto, Ontario, Canada), respectively. PSC#1 and PSC#2 were obtained from resected human PDAC samples by the outgrowth method and immortalized using lentiviral vectors with human telomerase (hTERT) or SV40 large T antigen (TAg; ref. 31). Primary PSCs were isolated by the outgrowth method from fresh patient resected human PDAC samples tissue and characterized by morphology (32). Cell lines were validated by STR DNA fingerprinting using the Promega 16 High Sensitivity STR Kit (catalog no. DC2100). The STR profiles were compared to online search databases (DSMZ/ATCC/JCRB/RIKEN) of approximately 2,500 known profiles; along with the MD Anderson Characterized Cell Line Core (CCLC) database of approximately 2,600 known profiles. The STR profiles matched known DNA fingerprints or were unique.

RNA was isolated using TRIzol Reagent (Life Technologies). The cDNA was prepared by reverse transcription using the Verso cDNA Synthesis Kit (Thermo Scientific). Gene expression was determined by quantitative PCR using the primer sequences listed in Supplementary Table S2.

Lysates from cultured cells and mouse pancreatic tissue were prepared using RIPA buffer and a cocktail of protease inhibitors (Roche). Conditioned media was prepared by collecting media from cells in 10-mm plates and concentrating the media 10-fold in Ultra-15 centrifugal filter units (EMD Millipore). Protein samples were separated by SDS-PAGE, transferred to a nitrocellulose membrane, blocked with 5% milk in 1× PBS with 1% Tween-20 for 1 hour and incubated with primary antibodies: monoclonal rat anti-LCN2 (R&D Systems) and polyclonal rabbit anti-SLC22A17 (Thermo Scientific) overnight at 4°C. Membranes were incubated with corresponding secondary antibodies and signal was detected by visualizing near-infrared fluorescent signals with the Odyssey imaging system (LI-COR).

Mission short hairpin RNA (shRNA) bacterial glycerol stock plasmid for SLC22A17 with sequence CCGGGCGATTCCTACAGCGAATGATCTCGAGATCATTCGCTGTAGGAATCGCTTTTTG was obtained from Sigma Aldrich. A total of 2.3 μg of the each respective plasmid was cotransfected with the envelope PMD2.G (2 μg) and packaging PsPAX2 (4.7 μg) plasmids (Addgene) into 293FT cells using Lipofectamine 2000 (Life Technologies). PSC#1 were incubated with virus supernatant and polybrene for 24 hours. Stable knockdown cells were selected using 1 μg/mL puromycin. SLC22A17 was also transiently knockdown in PSC#2 using a pool of four siRNAs with the following sequences: (i) CAGCATTATGCTGCTGCCGGA, (ii) AGGGAGCCCATCGGACTTCTA, (iii) CGCACGGTGGCTGATAGTGAA, and (iv) TCGGGAAGACTGAGTAGGGAA. siRNA transfection was performed using 12 μL of HiPerfect transfection reagent (Qiagen) and 12.5 nmol/L of each siSLC22A17 siRNAs for 24 hours.

PSCs with endogenous or knockdown of SLC22A17 were plated at 2 × 10⁵ cells per well in 6-well plates or at 1 × 10⁶ in 10-mm dishes. The next day, cells were washed twice with PBS and serum-starved overnight. Before treatment, cells were again washed twice with PBS. Recombinant human LCN2 (rhLCN2; R&D Systems) was resuspended in 25 mmol/L MES and 150 mmol/L NaCl at pH 6.5. Cells were treated with different concentrations of rhLCN2 at different time points to determine the optimal concentration and treatment time. Most experiments were conducted using 100 or 500 ng/mL rhLCN2 for 24 to 48 hours or as indicated.

Concentrated conditioned media or serum from mice was used to detect the expression of cytokines using the Proteome Profiler human cytokine array panel A/Mouse Adipokine Profiler (R&D Systems). Positive signals were analyzed with the ImageJ software protein array analyzer plug-in (created by Gilles Carpentier, Faculté des Sciences et Technologies, Université Paris, Paris, France) to obtain the pixel density for each spot on the array.

Serum samples were collected as described before (33). Plasma was isolated from peripheral venous blood, which was obtained in sodium heparin tubes from a cohort of treatment naïve patients with metastatic pancreatic cancer under an Institutional review board–approved protocol (34).

Pathologic scores were determined as described previously (25), Briefly, PanIN lesion frequency was evaluated by counting all foci of PanIN lesions. Each affected lobule, irrespective of the number of duct profile, was scored as 1. Pathologic score of the level of inflammation: 1, nonaggregation, occasional scattered leukocytes; 2, intermediate level of leukocytes; 3, aggregates of leukocytes within the parenchyma in a follicle-like formation. Pathologic score of the degree of fibrosis: 1, occasional periductal and scant perilobular fibrosis; 2, intermediate level; 3, definite extensive fibrosis, involving 50% of parenchyma with resultant parenchymal (acinar) atrophy.

Human and mouse LCN2 Quantikine ELISA Kits (R&D Systems) were used to measure LCN2 levels in human serum and plasma as well as mouse serum and fat lysates as per the manufacturer's instructions. All samples were assayed in duplicate and compared with a standard curve to quantitate expression.

Statistical analysis was performed using the Prism 5 software program (GraphPad Software). Results are expressed as the mean ± SEM. A t test or one-way ANOVA was performed on sex-matched littermates. The percentage of survival was analyzed using the Mantel–Cox method. The median survival in days was presented. P levels less the 0.05 were considered significant.

We confirmed previous reports that LCN2 is elevated in sera from PDAC patients (19, 20). LCN2 sera levels were compared among patients with normal pancreas, chronic pancreatitis (a benign inflammatory disease of the pancreas), and PDAC using ELISA (Fig. 1A). Our results showed that serum levels of LCN2 are significantly higher in patients with PDAC (223.2 ng/mL): 1.9-fold increase when compared with those with normal pancreas (115.3 ng/mL) and 2.36-fold increase when compared with those with chronic pancreatitis (94.45 ng/mL). Plasma levels of LCN2 in patients with metastatic PDAC were also analyzed and found to be high (160.6 ng/mL) compared with a previous study that has shown LCN2 plasma levels to be around 111 ng/mL in PDAC patients compared with 67 ng/mL in healthy controls (20). Furthermore, we analyzed mRNA levels of other members of the lipocalin family using microarray data from an earlier study (35) and found that LCN2 is the most upregulated member of the lipocalin family in human PDAC compared with normal pancreas and pancreatitis tissue samples (Fig. 1B).

Because of the reported associations between the expression levels of the adipokine LCN2 in obese individuals and its possible link with cancer (7, 36, 37), we examined if body mass index correlated with plasma levels of LCN2 in our cohort of PDAC patients. No correlation was evident between body mass index (BMI) and plasma levels of LCN2 in PDAC patients (Supplementary Fig. S1). We then examined Lcn2 levels in serum, pancreas, and fat tissue in the obesity-driven PDAC mouse model (25). We found that Lcn2 serum levels in the KRas^G12D/CRE mice increased from 156.7 to 316.6 ng/mL, an approximately 2-fold change, after being fed a HFD. However, we did not observe significant changes in the level of Lcn2 in serum from CRE control mice after feeding a HFD (Fig. 1C). In addition, we observed a significant upregulation of Lcn2 expression in both mRNA (Fig. 1D) and protein (Fig. 1E) levels in the pancreas of KRas^G12D/CRE mice fed a HFD, which were susceptible to PDAC development. We also observed that KRas^G12D/CRE mice express Lcn2 in the adipose tissue lysates (Fig. 1F) and that the stromal vascular fraction from murine adipose tissue expressed Lcn2 (Supplementary Fig. S2).

To explore the role of Lcn2 in obesity, Lcn2 wild-type (Lcn2^+/+) and Lcn2 whole-body knockout mice (Lcn2^−/−) were fed either a control diet (CD) or a HFD over a period of 300 days. During this time, Lcn2^+/+ animals fed a HFD tripled their weight, increasing from approximately 20 to approximately 60 g (Fig. 2A). Although Lcn2^−/− animals fed a HFD gained weight, their average weight was 20 g less than Lcn2^+/+ animals fed the same diet. Lcn2^+/+ animals fed a HFD had a significantly higher weight compared with controls fed a CD after 30 days. In contrast, it took six times longer (180 days), to observe significant weight changes in Lcn2^−/− mice fed a HFD compared with CD (Table 1).

Next, we examined whether Lcn2 depletion protected against weight gain in mice susceptible to PDAC development by diet-induced obesity. For this purpose, Lcn2^−/− mice were bred with KRas^G12D/CRE mice to generate triple transgenic mice lacking Lcn2 but expressing pancreas specific KRas^G12D (Lcn2^−/−/KRas^G12D/CRE). As expected, Lcn2^−/−/KRas^G12D/CRE mice lack serum levels of Lcn2 (Supplementary Fig. S3A) and when fed HFD for >50 days, mice developed less volume of visceral adipose tissue and showed significantly lower changes in BW compared to KRas^G12D/CRE mice (Fig. 2B; Supplementary Fig. S3B). These results indicate that lack of Lcn2 prevents weight gain and obesity in the diet-induced PDAC mouse model.

To evaluate the role of LCN2 in modulating the progression of PDAC by diet-induced obesity, we examined KRas ^G12D/CRE and Lcn2 ^−/−/KRas ^G12D/CRE animals after 50 days on a HFD. We found that Lcn2 ^−/−/KRas ^G12D/CRE mice fed a HFD developed fewer PanIN lesions, less fibrosis, and inflammation compared with KRas ^G12D/CRE controls (Fig. 2C and D). Picrosirius red staining, which stains collagens I and III, revealed less collagen remodeling of the extracellular matrix (ECM) in Lcn2 ^−/−/KRas ^G12D/CRE mice compared with KRas ^G12D/CRE mice (Fig. 2C). KRas ^G12D/CRE animals also exhibited higher levels of alpha smooth muscle actin (αSMA), one of the markers for activated PSCs, which are key regulators of PDAC stroma (Fig.2C; ref. 38). Next, we evaluated the effects of Lcn2 depletion on the inflammatory response in the pancreas of these animals. We assessed the presence of leukocytes by staining for the pan-leukocyte marker CD45. The pancreas of KRas^G12D/CRE animals displayed a higher level of CD45⁺ cells infiltration compared with Lcn2^−/−/KRas^G12D/CRE mice. Furthermore, KRas^G12D/CRE animals had an increased recruitment of F4/80⁺ macrophages (Fig. 2C). Importantly, only 40% of Lcn2^−/−/KRas^G12D/CRE animals developed PanIN 2 lesions compared with a significantly higher rate of 90% of the KRas^G12D/CRE animals. These results suggest that Lcn2 is an important mediator of inflammation, fibrosis, and PanIN development in PDAC.

Next, we evaluated the effects of Lcn2 depletion on the survival of mice with genetic predisposition to develop PDAC. Lcn2 ^−/−/KRas ^G12D/CRE survived longer than KRas ^G12D/CRE animals fed either a CD or a HFD (Fig. 3A). As expected from our previous findings (25), feeding of a HFD significantly reduced the median survival of KRas^G12D/CRE animals (269 to 195.5 days; Table 2). In contrast, Lcn2^−/−/KRas^G12D/CRE animals fed a HFD displayed a significant increase in median survival (249.5 days), compared with KRas^G12D/CRE mice fed a HFD. Moreover, Lcn2 depletion also significantly increased survival of animals fed a CD (296 days). These results indicate that regardless of diet, Lcn2 depletion allows for increased survival in KRas^G12D/CRE mice, which were predisposed to develop PDAC by the expression of oncogenic KRas^G12D in the pancreas. Some Lcn2^−/−/KRas^G12D/CRE animals in this survival cohort did develop PanIN lesions and fibrosis (Fig. 3B). However, KRas^G12D/CRE mice fed a CD or HFD experienced dramatic changes in stromal remodeling at around their median survival age, while Lcn2^−/−/KRas^G12D/CRE mice demonstrated less prominent ECM changes and immune cell infiltration at that age (Fig. 3B and C). These results suggest that Lcn2 has the potential to alter the expression of ECM components as well as immune cell infiltration in PDAC.

Given the important cross-talk between tumor cells and the tumor microenvironment in PDAC and the need to further understand the biological role of LCN2 in cancer, we examined the role of Lcn2 in a syngeneic orthotopic model. KPC-LUC cells were injected into the pancreas of Lcn2^+/+ and Lcn2^−/− mice (Fig. 4A and B and Supplementary Fig. S4). Serum Lcn2 levels significantly increased in Lcn2^+/+ mice after tumor cell implantation while Lcn2 serum levels were undetectable in Lcn2^−/− animals (Fig. 4C). Three weeks after tumor cell implantation, most Lcn2^+/+ animals had developed large tumors visualized by a strong bioluminescent signal that was an average of approximately 15-fold higher than mice lacking Lcn2 (Fig. 4A and B; Supplementary Fig. S5). In addition, the median survival of Lcn2^−/− animals after tumor cell implantation was significantly longer (∼3-fold) than Lcn2^+/+ mice (105 days vs. 33 days, respectively; Fig. 4D). All the Lcn2^+/+ mice became moribund and were humanely euthanized by 104 days after tumor cell implantation, whereas about 50% of Lcn2^−/−animals remained healthy at that time. Surviving Lcn2^−/− animals were sacrificed at 282 days to analyze the pancreas. Consistent with the results from the GEMM experiments, we observed that Lcn2^−/− mice had little to no pancreas ECM remodeling or inflammatory cell infiltration compared with Lcn2^+/+ animals in our orthotopic syngeneic model (Fig. 5A). The Lcn2^−/− animals that develop tumors at a later time (10/17) also had altered stroma and were infiltrated by macrophages and leukocytes. However, the presence of inflammatory, αSMA positive, and proliferative stromal cells (shown by Ki67 staining), was attenuated in the orthotopic tumors of Lcn2^−/− mice compared with those in Lcn2^+/+ mice (Supplementary Fig. S6). We then isolated cells from the peritoneal cavity, a technique that mostly yields macrophages (39). We showed that depletion of Lcn2 does not impair cell invasion of peritoneal macrophages (Supplementary Fig. S6). Moreover, the serum of Lcn2^−/− mice contained significantly lower levels of macrophage colony stimulating factor (M-CSF), an essential chemokine for the differentiation, proliferation, and survival of macrophage precursor cells (Fig. 5B). Further analysis of the serum of these animals also revealed reduced levels of putative PDAC biomarkers such as, intercellular adhesion molecule-1 (ICAM-1), insulin-like growth factor binding protein 1 (IGFBP-1), and tissue inhibitor of metalloproteinases 1 (TIMP-1; Fig. 5B; refs. 40–42;). Subsequent, analysis of bulk pancreas tissue revealed that Lcn2^+/+ animals injected with KPC-LUC cells had significantly higher expression levels of the proinflammatory cytokines IL6, monocyte chemoattractant protein-1 (MCP-1), and IL1B and had higher expression of matrix metallopeptidase 9 (MMP-9; Fig. 5C). These results recapitulate results obtained in our genetic mouse model experiments, supporting the role of Lcn2 in altering the stromal composition and inflammatory response in the PDAC tumor microenvironment.

To further elucidate a mechanism of how LCN2 could mediate the stromal changes observed in the mouse models, we examined the role of LCN2 and the LCN2-specific receptor, SLC22A17, in modulating the production of proinflammatory mediators by PSCs. We showed for the first time that PSCs expressed significantly higher levels of the SLC22A17 compared with PDAC cell lines and normal pancreatic epithelial cells (Fig. 6A). Moreover, we observed that LCN2 is not expressed by PSCs (Fig. 6A and B). In both our genetic and orthotopic syngeneic mouse models, Lcn2-null animals showed a significant reduction in the mRNA expression levels of SLC22A17 in the pancreas tissue (Supplementary Fig. S7).

To determine whether LCN2 may promote a proinflammatory response in PSCs, we first treated PSCs with different doses of exogenous recombinant human LCN2 (rhLCN2; Supplementary Fig. S8) and found that treatment with either 100 or 500 ng/mL rhLCN2 resulted in the induction of the proinflammatory cytokine IL6. We next evaluated a time-course of LCN2-mediated effects and observed a maximal mRNA expression induction of various cytokines after 48 hours of treatment (Supplementary Fig. S8). At this time, exogenous LCN2 induced PSCs to secrete higher levels of the proinflammatory molecules complement component 5a (C5a), IL6, stromal cell-derived factor 1 (SDF1), IL8, and ICAM-1 as assessed by a human cytokine array (Fig. 6C; Supplementary Fig. S8).

To determine whether these effects were LCN2 receptor-mediated, we examined the effects of downregulating the expression levels of the LCN2-specific receptor SLC22A17 in PSCs (Supplementary Fig. S8). LCN2 treatment of PSC#1 cells expressing control shRNA vector increased the mRNA expression levels of the proinflammatory cytokines IL6, MCP-1, IL8, and IL1B, as well as the matrix metalloproteinases MMP-1, MMP-3, and MMP-9, the complement factor C5/C5a, C5aR1, and αSMA (Fig. 6D; Supplementary Fig. S8). However, upon expressing shRNA against SLC22A17 in PSCs, the ability of LCN2 to induce the expression of all these molecules was significantly diminished (Figs. 6D). These results were recapitulated in a second PSC line, PSC#2 using siRNA against SLC22A17 to downregulated expression levels (Supplementary Fig. S8). Thus, these results suggest a previously unidentified role for the LCN2-specific SLC22A17 receptor in PDAC malignancy mediated by the modulation of PSCs in the tumor microenvironment.

Some studies suggest that LCN2 serves as a tumor suppressor, whereas others proposed that it is a tumor promoting factor (22–24). Therefore, the need of using GEMMs to study the role of Lcn2 in PDAC was warranted and addressed in this study. Although it is well known that LCN2 is upregulated in various types of cancer and in obese patients, studies linking LCN2 to both PDAC and obesity are lacking (13). Here we confirmed that LCN2 is the highest overexpressed member of the lipocalin family in PDAC. However, we did not observe any correlation between body mass index (BMI) and plasma levels of LCN2 in patients with metastatic PDAC. Most PDAC patients at time of diagnosis present with advanced disease in which substantial weight loss had already occurred. On the basis of our findings and prior reports on LCN2 as a biomarker of PDAC, we suspect that LCN2 may contribute at earlier stages, such as inflammatory processes that contribute to tumor progression. Thus, possibly due to timing, we were unable to capture any potential relationships in this patient cohort.

Moreover, in this study, we have revealed for the first time a possible mechanism through which Lcn2 depletion attenuated stromal changes, decreased immune cell infiltration and prolonged survival in a PDAC genetic mouse model, and an orthotopic syngeneic mouse model. We also showed that PSCs mediated the LCN2-mediated secretion of proinflammatory cytokines. It has been demonstrated that LCN2 binds to two cell-surface proteins: (i) SLC22A17 and (ii) low-density lipoprotein receptor-related protein (LRP2 or megalin; ref. 13). Both SLC22A17 and LRP2 have been reported to bind and mediate the cellular uptake of LCN2 (43–45). Although SLC22A17 has been shown to bind specifically to LCN2, LRP2 also binds numerous structurally unrelated molecules including aprotinin, vitamin D3, transferrin, and lactoferrin (46, 47). Although there is a limited amount of research published for SLC22A17, recent reports have indicated that its expression is correlated with malignancy. For instance, overexpression of both LCN2 and SLC22A17 has been associated with a worse prognosis in gliomas, hepatocellular, and endometrial cancer (47–49). Moreover, we observed that Lcn2 is also upregulated in the serum after HFD consumption in animals that express pancreas-specific mutant KRas. To the best of our knowledge, these data are the first to link increased expression of Lcn2 to diet-induced obesity in a GEMM of PDAC. In addition, we showed that Lcn2-deficient animals, exhibit attenuated weight gain and adiposity, suggesting a role for LCN2 in obesity preventative efforts.

Lcn2 deficiency also reduced pancreatic stromal fibrosis in animals expressing mutant KRas fed either a CD or a HFD. The ability of Lcn2 to stabilize MMP9 activity has been implicated to promote mammary tumor growth and metastasis (50). Here we showed that MMP9 expression is significantly reduced in Lcn2-null animals, suggesting that MMP9 may play a role in Lcn2-mediated PDAC growth. Furthermore, increased collagen expression has been shown to support ECM and facilitate tumor cell proliferation, migration, and survival (51). Moreover, the expression of αSMA is prominently increased in mutant KRas animals expressing Lcn2, compared to Lcn2-null mice. This observation supports the ability of LCN2 to activate PSCs. These cells are major contributors of ECM proteins, which likely explain the increased levels of collagens in these animals.

In addition, we observed an increase in the numbers of leukocytes and macrophages in the animals expressing Lcn2 and mutant KRas compared with those with deleted Lcn2. Tumor-associated macrophages (TAM) have been shown to infiltrate the PDAC stroma at an early stage and can be either pro- or antitumorigenic (51, 52). Some studies suggest that early presence of TAMs aids in the acinar to ductal metaplasia, a key early process in PDAC development (52). In fact, we observed that, in Lcn2^−/−/KRas^G12D/CRE animals fed a HFD for 50 days, there were lower levels of leukocyte and macrophage infiltration, which correlated with a decreased presence of early PanIN lesions. We also observed attenuated collagen deposition and immune cell infiltration, regardless of diet, in both genetic and syngeneic PDAC mouse models that do not expressed LCN2. These results suggest that Lcn2 expression promotes stromal remodeling and development of PDAC. Lcn2 was dispensable for macrophage migration to sites of inflammation in mice (53), which is consistent with our data showing recruitment of macrophages to inflammatory areas on the Lcn2-null mice. Moreover, we showed that invasion of peritoneal macrophages derived form Lcn2-null mice was not impared.

In this study, we also observed increased PanIN development and reduced median survival of KRas^G12D/CRE mice compared with the Lcn2^−/−/KRas^G12D/CRE mice fed a HFD. It is possible that with the increased adiposity, cells within the adipose tissue are contributing to the secretion of proinflammatory factors, such as LCN2 that could promote PDAC development. LCN2 is known to be expressed by adipose tissue (11, 12) and here we showed expression of LCN2 and the LCN2 receptor SLC22A17 (Supplementary Fig. S9) in adipose tissue-derived from PDAC mice. Moreover, LCN2 expression levels in adipose tissue have been positively correlated with expression of other proinflammatory cytokines (11, 12). Adipose tissue is composed of a variety of cell types, among these are the adipose stromal cells (ASC), which are components of the stromal vascular fraction (SVF). ASCs have been demonstrated to promote tumor growth (54–57) and visceral fat from obese individuals was shown to increase ASCs population and trafficking to the tumor microenvironment (58, 59). Here we showed that cells from the SVF of murine adipose tissue expressed Lcn2 and the Lcn2 receptor SLC22A17. Therefore, there remains a possibility that the adipose tissue via ASCs and/or other cells may be mediating the secretion of proinflammatory molecules such as LCN2 that could contribute to tumor growth. It is also possible that ASCs could be trafficking to areas of PDAC and contributing to PDAC fibrosis for which futures studies are necessary. Other groups have also shown that diet-induced obesity increases inflammation of peripancreatic fat in mouse models of PDAC (60). This further suggests an essential role of adiposity and the risk for PDAC development and complement our current findings. Interestingly, we found that more of the Lcn2^−/−/KRas^G12D/CRE mice treated with caerulein developed inflammation and PanIN lesions compared to those treated with vehicle alone (Supplementary Fig. S3). These data suggest that caerulein-induced inflammation and diet-induced obesity inflammation might be acting via different mechanisms that merit further investigation.

Our results indicate that reexpression of Lcn2 in the pancreas, via orthotopic injection of tumor cells expressing Lcn2, is not sufficient to allow for tumor growth and progression in Lcn2-null animals. This suggests that other sources of Lcn2, perhaps secreted from inflammatory cells and/or adipose tissue as discussed above, are needed to create a suitable tumor microenvironment for disease progression (7, 13). Because Lcn2 has been shown to be expressed by immune cells, it is possible that Lcn2 derived from the immune system could also contribute to PDAC. It has been previously shown that Lcn2-depleted marrow cells prevent the development of BCR-ABL–induced leukemia, indicating the possible role of inflammatory cell–derived Lcn2 in tumor development (61). The cross-talk between tumor cells and stromal components in PDAC is essential for establishing this environment. PSCs are stromal cells that play a key role in secreting molecules such as cytokines and growth factors that aid in the remodeling of the PDAC tumor microenvironment (52, 62, 63). In this study, we show for the first time that pancreatic stromal cells do not synthesize LCN2 but do express LCN2-specific receptors. Moreover, exogenous LCN2 induces PSCs to secrete proinflammatory factors, matrix metalloproteinases, and the complement component C5a. We also observed elevated levels of IL6, MCP-1, IL1B, and MMP-9 in the pancreas of Lcn2-expressing animals after tumor cell implantation. Because MCP-1, IL8, and C5a are potent immune cell chemoattractants, these results provide a possible explanation as to why we observed a lack of immune cell presence in vivo in Lcn2-null mice.

We specifically showed for the first time that the LCN2 receptor SLC22A17 plays a significant role in the LCN2-mediated proinflammatory responses observed in PSCs. We suggest a novel role of LCN2 and its specific receptor SLC22A17 in mediating the synthesis of molecules important for remodeling the PDAC stroma, leading to increased tumor growth and decreased survival. In conclusion, our results showed that Lcn2 depletion delayed weight gain, adiposity, and tumor growth and increased survival in PDAC mouse models. LCN2 also modulated the production of proinflammatory factors in PSCs and reduced the infiltration of macrophages in the stroma of both obesity and nonobesity mouse models of PDAC. Therefore, this study indicates that LCN2 may be an important link between inflammation, obesity, and PDAC.

No potential conflicts of interest were disclosed.

Conception and design: S.B. Gomez-Chou, G.B. Lesinski, T. Bekaii-Saab, M.R. Farren, X. Leng, R.B. Arlinghaus, C.D. Logsdon, Z. Cruz-Monserrate

Development of methodology: S.B. Gomez-Chou, Z. Cruz-Monserrate

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S.B. Gomez-Chou, A.K. Swidnicka-Siergiejko, N. Badi, G.B. Lesinski, M.R. Farren, Y. Liu, D. Deng, R.F. Hwang, H. Wang, Z. Cruz-Monserrate

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S.B. Gomez-Chou, A.K. Swidnicka-Siergiejko, M. Chavez-Tomar, G.B. Lesinski, T. Bekaii-Saab, M.R. Farren, T. Moore, D. Chatterjee, H. Wang, C.D. Logsdon, Z. Cruz-Monserrate

Writing, review, and/or revision of the manuscript: S.B. Gomez-Chou, A.K. Swidnicka-Siergiejko, M. Chavez-Tomar, G.B. Lesinski, T. Bekaii-Saab, M.R. Farren, T.A. Mace, C. Schmidt, R.F. Hwang, T. Moore, H. Wang, X. Leng, R.B. Arlinghaus, C.D. Logsdon, Z. Cruz-Monserrate

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S.B. Gomez-Chou, A.K. Swidnicka-Siergiejko, M. Chavez-Tomar, C. Schmidt, Y. Liu, D. Deng, L. Zhou, C.D. Logsdon, Z. Cruz-Monserrate

Study supervision: G.B. Lesinski, C.D. Logsdon, Z. Cruz-Monserrate

We thank the MDACC Characterized Cell Line Core Facility for the STR DNA fingerprinting, which is funded by the NCI Cancer Center Support Grant.

NIH-DK052067 and The Lockton Endowment (C.D. Logsdon), NIDDK Diversity Supplement DK052067-08 (C.D. Logsdon for Z. Cruz-Monserrate), 2014 Knowledge GAP MDACC (Z. Cruz-Monserrate), Start-up funds from the Comprehensive Cancer Center at The Ohio State University (OSUCCC; Z. Cruz-Monserrate). Also supported in part by the MD Anderson Cancer Center Support Grant CA016672 and by the OSUCCC grant P30 CA16058 National Cancer Institute.

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.