Desmoplasia and an inflammatory environment are defining features of pancreatic cancer. Unclear is how pancreatic cells that undergo oncogenic transformation can cross-talk with immune cells and how this contributes to the development of pancreatic lesions. Here, we demonstrate that pancreatic acinar cells expressing mutant KRAS can expedite their transformation to a duct-like phenotype by inducing local inflammation. Specifically, we show that KRASG12D induces the expression of intercellular adhesion molecule-1 (ICAM-1), which serves as chemoattractant for macrophages. Infiltrating macrophages amplify the formation of KRASG12D-caused abnormal pancreatic structures by remodeling the extracellular matrix and providing cytokines such as TNF. Depletion of macrophages or treatment with a neutralizing antibody for ICAM-1 in mice expressing oncogenic Kras under an acinar cell–specific promoter resulted in both a decreased formation of abnormal structures and decreased progression of acinar-to-ductal metaplasia to pancreatic intraepithelial neoplastic lesions.
Significance: We here show that oncogenic KRAS in pancreatic acinar cells upregulates the expression of ICAM-1 to attract macrophages. Hence, our results reveal a direct cooperative mechanism between oncogenic Kras mutations and the inflammatory environment to drive the initiation of pancreatic cancer. Cancer Discov; 5(1); 52–63. ©2014 AACR.
This article is highlighted in the In This Issue feature, p. 1
Oncogenic KRAS mutations drive metaplasia of pancreatic acinar cells to a highly proliferative duct-like cell type, which is the precursor for precancerous pancreatic intraepithelial neoplastic (PanIN) lesions (1, 2). Although such KRAS-induced acinar-to-ductal metaplasia (ADM) can be seen as an initial event that leads to development of pancreatic lesions, additional mutations (i.e., loss of the tumor suppressor genes p16INK4A, TP53, and SMAD4) are required for further progression to pancreatic cancer (3). Recent evidence also suggests that alterations in the pancreatic microenvironment are key factors for further development from precancerous lesions to pancreatic ductal adenocarcinoma (PDA). For example, it was demonstrated that caerulein-induced release of digestive enzymes and subsequent pancreatic inflammation in KrasG12D-expressing mice accelerate the development of PDA (4). Similarly, a high-fat diet in KrasG12D-expressing mice dramatically accelerates PDA by causing pancreatic inflammation and macrophage infiltration (5). Consequently, in humans, environmental risk factors for the development of pancreatic cancers are pancreatitis and inducers of inflammatory responses such as obesity and smoking (6, 7).
Although a functional link between inflammation and oncogenic Kras for progression of PDA was demonstrated in the above mouse models, inflammation may not only accelerate KRAS-driven pancreatic cancer but also initiate the ADM process. For example, it was recently shown that depletion of macrophages in a mouse model for acute pancreatitis completely protected acinar cells from undergoing ADM, resulting in a protection of the pancreas from injury (8). Mechanistic insight into how inflammation contributes to changes in the pancreas microenvironment was provided by demonstrating that macrophages can secrete cytokines that drive ADM. These include TNF, RANTES (8), and HB-EGF (9). However, although the contribution of KRAS-initiated signaling pathways that induce ADM is relatively well defined (i.e., MAPK activation, cooperation with EGFR signaling; ref. 10), the signaling cross-talk between KRAS-expressing pancreatic cells and cells of the microenvironment is just beginning to be investigated (8, 9, 11, 12). Specifically, whether KRASG12D expression in acinar cells can initiate signaling pathways that facilitate local inflammation is unknown.
Here, we report that oncogenic KRAS mutations in pancreatic acinar cells induce the expression of intercellular adhesion molecule-1 (ICAM-1). We show that ICAM-1 can act as a chemoattractant for macrophages. Attracted macrophages release matrix-degrading enzymes, including matrix metalloproteinase 9 (MMP9), as well as cytokines such as TNF that synergize with KRAS mutations to drive acinar cell metaplasia. We demonstrate that depleting macrophages or neutralizing ICAM-1 in p48Cre;KrasG12D-expressing mice dampens the development of precancerous lesions. In summary, our data for the first time demonstrate that KrasG12D-expressing acinar cells can expedite their transformation to a duct-like phenotype by inducing local inflammation and macrophage infiltration. Overall, we provide a mechanism of how oncogenic Kras mutations and inflammatory microenvironment function synergistically to drive the earliest abnormal pancreatic structures that precede PDA.
Macrophages Contribute to KRASG12D-Induced Formation and Progression of Pancreatic Lesions
Pancreatitis and macrophage infiltration have been associated with faster progression of pancreatic lesions to tumors (4, 13). To test if macrophages can contribute to the initiation of pancreatic cancer, we utilized the p48Cre;LSL-KrasG12D animal model, in which an oncogenic mutant of Kras is expressed under an acinar cell–specific promoter, and depleted macrophages by treatment with gadolinium (III) chloride (GdCl3). Treatment with GdCl3 (treatment schedule shown in Supplementary Fig. S1A) significantly decreased the presence of macrophages in the KrasG12D-expressing pancreata (Fig. 1A), but had no effect on the presence of CD3-positive T cells or neutrophils (Supplementary Fig. S1B). Although PBS-treated p48Cre;LSL-KrasG12D mice showed ADM and PanIN formation, the formation of these abnormal structures was significantly reduced in the absence of macrophages (Fig. 1A and Supplementary Fig. S1C). Quantitative analysis of pancreata showed that when macrophages were depleted, the overall number of lesions (abnormal structures) was decreased by approximately 50% (Fig. 1B). Division of abnormal structures into regions of isolated ADM, ADM–PanIN transition areas, or PanIN lesions suggested a slower progression of KRAS-caused oncogenesis when macrophages were absent. For example, in the absence of macrophages, isolated ADM areas increased from 25% to 36% and total PanIN lesions decreased from 48% to 38%. Within these PanIN lesions, the occurrence of PanIN1A/1B increased from 77% to 89% and PanIN2/3 lesions decreased from 23% to 11%. Overall, this suggests that macrophages can contribute to the progression of ADM to PanINs. This is supported by an increased occurrence of macrophages in ADM regions, as compared with established PanIN lesions (Fig. 1C).
Mutant KRAS Induces Expression of ICAM-1 in Acinar Cells
We next determined if mutant KRAS in acinar cells can induce the expression of chemoattractants for immune cells. Therefore, we infected primary mouse acinar cells with lentivirus carrying an oncogenic Kras (KrasG12V) mutant or a control virus and then tested for expression of different chemoattractants. These included CXCL11 (chemoattractant for activated T cells), CXCL10 (chemoattractant for T cells, natural killer cells, dendritic cells, and monocytes/macrophages), and ICAM-1 (chemoattractant for leukocytes). Although expression of mutant KRAS had little effect on CXCL10 and CXCL11 expression, the expression of mRNA for Icam1 was increased approximately 100-fold as compared with control cells (Fig. 2A). At the same time, the expression of Egfl7, a molecule previously described as an inhibitor of ICAM-1 expression (14), was significantly decreased. Increased Icam1 mRNA levels translated to increased levels of full-length protein, as judged by Western blotting (Fig. 2B). Interestingly, we also detected slightly increasing levels of soluble ICAM-1 (sICAM-1) in the supernatant of three-dimensional (3D) organoid cultures undergoing KrasG12V-driven ADM. During transdifferentiation of cells, sICAM-1 levels increased from approximately 0.1 +/− 0.044 ng/mL at the day of seeding in 3D culture to 0.36 ± 0.035 ng/mL at day 5 of KrasG12V expression (Fig. 2C). In accordance with KRAS-driven expression of ICAM-1 in acinar cell 3D organoid culture, ICAM-1 expression also was detected in regions of ADM in p48Cre;LSL-KrasG12D mice, but not in control mice (Fig. 2D). This correlated with the presence of macrophages in these regions as determined by F4/80 staining. Regions of ADM also showed an increase of cytokeratin-19, a marker for duct-like cells (Supplementary Fig. S2A). Interestingly, further progressed PanIN lesions in the same samples did show significantly less ICAM-1 expression and less macrophage infiltration in these areas (Supplementary Fig. S2B). Because regions of PanIN and PDA also contain microvessels (15) that could contribute to expression of ICAM-1 in the pancreas in vivo, we performed an additional experiment to demonstrate that KrasG12D-expressing acinar cells indeed upregulate ICAM-1. Therefore, we isolated acinar cells from p48Cre;KrasG12D mice (KrasG12D expression only in acinar cells) or control mice and tested if they expressed ICAM-1. Increased ICAM-1 expression was detected in acinar cells from p48Cre;KrasG12D mice as compared with normal acinar cells, indicating that acinar cells indeed are the source of ICAM-1 (Fig. 2E). Moreover, an immunohistochemical analysis of regions of ADM in p48Cre;KrasG12D mice for endothelial cells (anti-CD31 as a marker) did not indicate significant presence of vessels in these regions, indicating that microvascularization may occur at a later time during tumor development (Supplementary Fig. S3A).
Similar to data obtained with the above genetic mouse model, in human samples, ICAM-1 expression was detected by IHC in regions of ADM, but not in adjacent “normal” tissue or PanIN lesions of types 1A, 1B, or 2 (Fig. 2F and Supplementary Fig. S3B). The exclusive presence in regions of ADM in samples of human origin or mouse genetic models indicates that the expression of ICAM-1 is an initial event that may contribute to the ADM process, but may be less needed during the progression of PanINs.
KRASG12D-Caused Expression of ICAM-1 Serves as a Chemoattractant for Macrophages
ICAM-1, when expressed in endothelial cells, was shown to promote neutrophil adhesion and transendothelial migration (16, 17). We next tested if ICAM-1 also can act as an attractant for macrophages. In an initial experiment, to test if cells that express ICAM-1 can attract macrophages, we generated HeLa cells stably expressing GFP-tagged full-length transmembrane ICAM-1 and cocultured these cells in spatial separation with RAW264.7 macrophage cells stained with a live cell dye. Control cells expressed GFP, but neither ICAM-1 mRNA nor endogenous or ectopic protein was detected (Supplementary Fig. S4A and S4B). After removal of the separation barrier, we observed that expression of GFP-tagged ICAM-1 led to attraction of macrophages (Fig. 3A). We then analyzed the supernatant of ICAM-1–expressing cells for the soluble version of this protein (sICAM-1) by ELISA, and detected sICAM-1 at 2.599 ± 0.422 ng/mL.
Although the above experiment is a proof of principle to show that macrophages can be attracted by cells expressing ICAM-1, two questions remained unanswered. First, can the secreted sICAM-1 contribute to such chemoattraction, and, second, which polarized phenotype of macrophage (M1 or M2) is attracted? We found that RAW264.7 cells migrated toward recombinant sICAM-1 in Transwell assays (Supplementary Fig. S4C). However, this cell line expresses markers for both M1 [i.e., positive for inducible nitric oxide synthase (iNOS)] and M2 (i.e., positive for Ym1) macrophages (Supplementary Fig. S4D), and we were not able to obtain clean polarization to one of these phenotypes after lipopolysaccharide (LPS) + IFNγ or IL4 stimulation (data not shown). Therefore, we decided to use freshly isolated primary mouse macrophages. After treatment with LPS + IFNγ or IL4, we observed ideal M1 or M2 polarization, respectively (Supplementary Fig. S4E and S4F). Freshly isolated mouse primary macrophages as well as fully polarized M1 macrophages migrated toward recombinant sICAM-1 in a dose-dependent manner in Transwell assays. M2-polarized macrophages generally showed higher motility, but were not attracted by sICAM-1 (Fig. 3B and C).
To demonstrate that expression of mutant KRAS in primary acinar cells can induce the attraction of primary macrophages, we isolated acinar cells from LSL-KrasG12D mice and induced KrasG12D expression by infection with Adeno-Cre or control virus (adeno-null). We then determined the migration of freshly isolated or polarized (M1 or M2) primary mouse macrophages toward the KrasG12D-expressing or control acinar cells in a Transwell assay. We found that chemoattraction of freshly isolated or M1-polarized macrophages by acinar cells expressing mutant KRAS was significantly increased, whereas M2-polarized macrophages, although highly motile, were not specifically attracted (Fig. 3D and E). To show that KRASG12D-caused chemoattraction is dependent on ICAM-1 expression, we added an ICAM-1–neutralizing antibody (NAB) to the assay (columns 3 and 4 in Fig. 3D and E)—and this completely blocked KRASG12D-induced chemoattraction of M1 macrophages. In summary, our data show that acinar cells expressing an oncogenic allele of Kras can facilitate the attraction of primary macrophages of the M1 phenotype via ICAM-1 in vitro.
Blocking ICAM-1 with a Neutralizing Antibody Decreases Macrophage Infiltration and KRAS-Caused Abnormal Structures In Vivo
We next tested if blocking macrophage attraction with an ICAM-1 NAB is sufficient for decreasing or blocking KRAS-caused formation of abnormal pancreatic structures in vivo. Therefore, we treated p48Cre;LSL-KrasG12D mice with ICAM-1 NAB over a time period of 11 weeks, starting at an age of 3 weeks (treatment schedule shown in Supplementary Fig. S5A). For controls, mice were treated with either PBS (vehicle) or an isotype-matched antibody (control Ab). At the endpoint (week 14), we found that neutralization of ICAM-1 led not only to a drastic decrease of macrophage infiltration into the pancreas but also to a significant reduction of abnormal lesion structures (Fig. 3F; additional controls and IHC for the PanIN marker claudin-18 are shown in Supplementary Fig. S5B). Results from a basic complete blood count test suggested that administering repeated doses of an anti–ICAM-1 antibody had no effects on generalized immune reactions in mice (Supplementary Fig. S5C). Moreover, control mice treated with ICAM-1 NAB did not show any presence of immune cells in the pancreas. However, in the KrasG12D-expressing pancreas, we clearly show that although macrophage attraction was blocked by an ICAM-1 NAB, Ly6B.2-positive cells (neutrophils) and CD3-positive cells (T cells) were still present (Supplementary Fig. S5B and S5D).
Comparable to the effects we had observed after ablation of macrophages in Fig. 1, we found that neutralization of ICAM-1 led to an increase in regions of isolated ADM (from 23% to 32%) and a decrease of total PanIN lesions from 47% to 36% (Fig. 3G). Within these PanIN lesions, the occurrence of PanIN1A/1B increased from 80% to 89%, and PanIN2/3 lesions decreased from 20% to 11%. Overall, neutralization of ICAM-1 had almost identical effects on the composition of abnormal structures as depletion of macrophages.
Macrophage-Secreted Cytokines and Proteases Contribute to ADM
We recently have shown that in an inflammatory environment in the pancreas, infiltrating macrophages secrete a panel of cytokines, of which some can initiate ADM (8). One major driver of macrophage-induced ADM that we had identified is TNF (8). Because KrasG12D-expressing acinar cells also attract macrophages, we next tested if this correlates with the presence of TNF in regions of ADM. We found that in regions of ADM in p48Cre;KrasG12D mice, the presence of TNF and macrophages correlated and both did not occur when ICAM-1 expression was blocked with the neutralizing antibody (Fig. 4A). Then, to directly determine if TNF can contribute to the KRASG12D-driven ADM process, we performed a 3D explant organoid assay. Therefore, we isolated primary acinar cells from LSL-KrasG12D mice and induced KrasG12D expression by transduction of Cre recombinase with an Adeno-Cre virus (or Adeno-null as control). Then, acinar cells were seeded in 3D collagen culture and treated with TNF or control as indicated (Fig. 4B). At day 5, ADM events were determined. Both KrasG12D mutation and TNF acted synergistically to increase the number of ADM events (ADM event = metaplasia of acinar cells to duct-like cells) as well as the size of duct-like structures formed.
Because macrophages are also known for their ability to secrete proteases that degrade the extracellular matrix, we next tested if regions of ADM showed increased protease activity. Therefore, we first performed an in situ zymography of pancreata of p48Cre;LSL-KrasG12D mice and found increased extracellular matrix (ECM) degradation in ADM and PanIN regions (Fig. 4C). This may be due to increased activities of MMPs as well as other proteases such as members of the A Disintegrin and Metalloproteinase (ADAM) family (8, 10). However, in 3D explant organoid assays, ADM events could be significantly blocked with the MMP inhibitor GM6001, suggesting a major role of MMPs (Supplementary Fig. S6). Because in experimental pancreatitis MMP9 was shown to be a major contributor to the ADM process (8), we next investigated if the appearance of macrophages in regions of ADM correlated with the presence of MMP9. We found MMP9 expression colocalizing with macrophages in regions of ADM in p48Cre;LSL-KrasG12D mice, but not in control mice or in p48Cre;LSL-KrasG12D mice, in which we had depleted macrophages by treatment with GdCl3 (Fig. 4D). Taken together, our data indicate that M1 macrophage-secreted cytokines such as TNF as well as proteases including MMP9 can contribute to mutant KRAS–driven ADM. Figure 4E provides a model of how the interaction of acini and macrophages may occur to promote acinar cell metaplasia to a duct-like phenotype that is believed to be the precursor of PanIN lesions.
Activating KRAS mutations have been long recognized as the drivers of pancreatic intraepithelial lesions (18). When expressed in mice under pancreatic cell–specific (i.e., Pdx1 or Pft1a/p48) promoters, KRASG12D induces ADM and formation of PanINs (1, 19). However, expression of KrasG12D does not lead to all acinar cells undergoing ADM simultaneously, and pancreata of mice show patchy regions of isolated ADM, ADM/PanIN transition areas, as well as progressed lesions (1, 19). A possible explanation is that in order to drive transformation, KrasG12D-expressing cells need to interact with cells of the pancreatic microenvironment, including pancreatic stellate cells or infiltrating immune cells (13, 20, 21). However, direct experimental evidence of such cross-talk for these initial processes was lacking. We here show that attraction of macrophages and microinflammation caused by expression of oncogenic KRAS in acinar cells is a necessary event to drive the formation of precancerous lesions (Fig. 1). Our data suggest that macrophage infiltration predominantly occurs in regions of ADM, but less in the PanIN stage, indicating the importance of macrophage-released factors in the initiation of acinar cell transdifferentiation. Different roles have been demonstrated for M1 and M2 macrophages (22). Both subtypes can be detected in pancreata of p48Cre;LSL-KrasG12D mice (data not shown), but their relative contribution to KRAS-driven ADM at this point is unclear. Because M1 macrophages are attracted by ICAM-1 (Fig. 3) and because we observe the presence of TNF after macrophage attraction (Fig. 4A), we predict that this subtype has a predominant role in driving ADM. However, because M2 macrophages have been shown to activate stellate cells, it is likely that they also contribute via other mechanisms to cross-talk between multiple cell types to drive ADM.
TNF in M1 macrophages is an NF-κB target gene (23). Activation of this transcription factor could be achieved during M1 polarization via IFNγ and LPS (24), which both have been shown to upregulate NF-κB signaling (25, 26). Another possibility is that M1 macrophages, once attracted to the acinar cell clusters, physically interact with ICAM-1 on acinar cells via MUC1, which can also activate NF-κB (27). In addition, NF-κB may also participate in KRAS-induced expression of ICAM-1 in acinar cells. It was shown that ICAM-1 expression is regulated by NF-κB1 (28), and NF-κB1 can be activated by oncogenic KRAS and also amplify RAS activity in pancreatic cancer cells (29, 30).
ICAM-1 is a transmembrane protein that can be converted into a soluble protein (sICAM-1) by shedding (31), and thus can act as a chemoattractant. Acinar cells express several proteinases that could drive this process. It was shown that MMPs can facilitate the formation of sICAM-1 (32). MMP3 has been demonstrated to be upregulated during ADM and to stimulate immune cell infiltration, priming the microenvironment for early tumor development (33). It is possible that MMP3 to some extent is responsible for the generation of sICAM-1. Another proteinase that is upregulated by oncogenic KRAS is ADAM17 (34), which has been reported to produce sICAM-1 (35). Increased circulating levels of sICAM-1 can be detected in pancreatic cystic fluid of patients with pancreatitis, suggesting it as a potential marker for the presence of pancreatic disease (36).
The function of ICAM-1 in the development and progression of PDA so far was not well understood. By showing that it is produced by acinar cells that express oncogenic KRAS, we provide a role for ICAM-1 in the initiation of pancreatic cancer by attracting M1 macrophages. Our data also show that only a fraction of the ICAM-1 produced by acinar cells is soluble. Expression of full-length (transmembrane) ICAM-1 in acinar cells may also have additional functions, such as acting as an antagonist to β2 integrin to loosen cell–matrix connections and enhance ADM. Another possible function is to act as a receptor for M1 macrophage attachment via interaction with the leukocyte adhesion protein LFA-1 (leukocyte function–associated antigen-1) and MUC1. At this point, it is unclear which surface receptor macrophages use to chemodetect ICAM-1, and whether direct contact between both cell types is needed to drive the ADM process. For example, it is also possible that pancreatic stellate cells are involved as an intermediate cell type in this cross-talk (20, 21). Because ICAM-1 and sICAM-1 can have antagonistic effects on the tight junctions (31), they also could act as an autocrine mechanism to affect neighboring cells and to alter the structures of acinar cell clusters.
Although our data suggest a role for ICAM-1 expression by acinar cells as a driving force in tumor-initiating events, the role of ICAM-1 expression in the actual cancer is less clear. In pancreatic cancer cell lines, it was shown that ICAM-1 expression was significantly increased as compared with normal pancreatic cells (37). Moreover, ICAM-1 expression status was linked to the poor prognosis of pancreatic cancer (38). Although our data show that in human tissue ICAM-1 expression mainly occurs in regions of ADM and decreases in regions of PanINs (Fig. 2F), we confirmed that it is re-expressed in human pancreatic cancer (Supplementary Fig. S3C). The re-expression in established adenocarcinoma may be due to increased vascularization (15) and may be unrelated to acinar cell–produced ICAM-1.
Understanding the cross-talk between pancreatic cells expressing oncogenic KRAS mutations and the pancreas microenvironment is of importance to develop targeted strategies for this cancer, including immunotherapy to increase responsiveness of the adaptive immune system (39). Our findings that macrophages are attracted by KrasG12D-expressing acinar cells, and directly contribute to their transdifferentiation by providing inflammatory cytokines or proteases, could be an additional angle to target the microenvironment leading to tumor formation, but also to develop biomarkers for early detection. For example, ICAM-1 levels in the pancreatic juice in combination with markers for advanced PanINs (40) could be predictive markers for the presence of preneoplastic lesions. After prediction of the potential of patients to develop pancreatic cancer, the use of neutralizing antibodies may be a way to prevent progression of precancerous lesions. ICAM-1–blocking antibodies have been tested for different disease models and species in vivo. Most studies in which ICAM-1–blocking antibodies were used to mitigate inflammation or prevent migration of cells to inflammatory lesions have been performed in rats using mouse monoclonal IgG antibodies (41, 42). The ICAM-1–blocking antibody we used is a monoclonal IgG made in hamster, which previously was used for studies to diminish macrophage recruitment to atherosclerotic plaques (43). With our animal studies (Fig. 3F and Supplementary Fig. S5), we provide a proof-of-principle experiment for the use of such blocking antibodies for clinical use in humans. One of these, enlimomab, has already been developed for treatment of ischemic stroke, refractory rheumatoid arthritis, and kidney and liver allograft transplantations (44).
Cell Lines, Antibodies, Viral Constructs, and Reagents
RAW264.7 and HeLa cell lines were obtained from the ATCC. Cells were not further authenticated, but have been passaged in the laboratory for fewer than 6 months. All cell lines were maintained in DMEM (high glucose) with 10% FBS and 100 U/mL penicillin/streptomycin in a 37°C incubator supplemented with 5% CO2. For HeLa cell lines stably expressing GFP or full-length transmembrane GFP-ICAM-1, cells were infected with lentivirus (multiplicity of infection of 5), and 48 hours after infection selected using 5 μg/mL Blasticidin (Invitrogen) for 10 days and then additionally selected by FACS analysis. Anti-CD3, anti-F4/80, anti-iNOS, anti-MMP9, and anti–hICAM-1 (used for IHC of human samples) antibodies were from Abcam, anti–mICAM-1 (used for IHC of mouse samples) was from R&D Systems, anti-Ym1 was from Stemcell Technologies, Ly6B.2 antibody was from AbDSerotec, anti–claudin-18 was from Invitrogen, anti-amylase was from Sigma-Aldrich, and anti-GFP, anti-CD31, and anti–cytokeratin-19 were from Santa Cruz Biotechnology. The neutralizing monoclonal hamster IgG antibody (NAB) for mICAM-1 used in the animal experiments was from Thermo Scientific; an isotype-specific control antibody was from BD Pharmingen. Adeno-Cre and Adeno-null (empty vector, control) adenovirus was purchased from Vector Biolabs. pLenti6.3/V5-GFP-ICAM-1 was generated by cloning human ICAM-1 first into pEGFP-N1 using BamHI and EcoRI sites and the following primers: 5′-GAATTCATGGCTCCCAGCAGC-3′ and 5′-GGATCCAAGGGAGGCGTGGCTTGTG-3′. GFP-ICAM-1 was then amplified by PCR using 5′-GAATTCATGGCTCCCAGCAGC-3′ and 5′-CCGCTCGAGTTACTTGTACAGCTCGTC-3′ as primers, and the fragment was inserted into pLenti6.5/V5-TOPO through TOPO cloning using the TOPO cloning kit from Invitrogen. Recombinant mouse sICAM-1 was purchased from Sino Biological Inc., recombinant mouse TNFα, recombinant mouse IL4, and recombinant mouse IFNγ were from PeproTech. LPS, 4,6-diamidino-2-phenylindole (DAPI), gadolinium chloride hexahydrate, and dexamethasone were from Sigma-Aldrich. DQ gelatin was from Invitrogen and agarose was from Lonza. Soybean trypsin inhibitor, collagenase I, and GM6001 were from EMD Millipore. Rat tail collagen I was from BD Biosciences.
Animals and Treatments
BALB/c mice for isolation of primary pancreatic acinar cells or primary macrophages were purchased from Harlan Laboratories. Ptf1a/p48Cre/+ and LSL-KrasG12D/+ mouse strains and genotyping of mice have been described previously (45–47). To deplete macrophages, mice at 7 weeks of age were intravenously injected with GdCl3 at a dose of 10 mg/kg or PBS solution as control vehicle every 2 days for 1 week. This was repeated at week 10. At week 13, mice were sacrificed and tissues harvested (see time line in Supplementary Fig. S1A). To neutralize ICAM-1 in vivo, 3-week-old mice were intraperitoneally injected with ICAM-1 NAB (or isotype-specific control antibody) at a dose of 2 mg/kg or vehicle every other day for 11 weeks (see time line in Supplementary Fig. S5B). All animal experiments were approved by the Institutional Animal Care and Use Committee of the Mayo Clinic and were performed in accordance with relevant institutional and national guidelines and regulations.
Human Pancreatic Tissue Samples
All patient tissues were obtained in accordance with institutional guidelines and prior institutional review board approval. Tissue microarrays of patient-matched normal pancreas, PanIN, and pancreatic adenocarcinoma were constructed from archival materials in the University of California, Los Angeles (UCLA), Department of Pathology and Laboratory Medicine archives, representing patients who underwent gross resection of tumor at UCLA Medical Center between 1987 and 2005. Tissue microarray construction was supported by a seed grant from the Hirshberg Foundation for Pancreatic Cancer Research.
DAB IHC and Immunofluorescence of Tissues
Slides were deparaffinized (one hour, 60°C), dewaxed in xylene (five times for four minutes), and gradually rehydrated with ethanol (100%, 95%, and 75%, each two times for three minutes). The rehydrated samples were rinsed in water and subjected to antigen retrieval in 10 mmol/L sodium citrate buffer (pH 6.0). Slides were treated with 3% H2O2 (five minutes) to reduce endogenous peroxidase activity, washed with PBS containing 0.5% Tween 20, and blocked with protein block serum-free solution (DAKO) for five minutes at room temperature. Samples were stained with hematoxylin and eosin (H&E), alcian blue. (i) For DAB IHC, anti-F4/80 antibody (1:200), anti-Ly6B.2 (1:3,000), anti-CD3 (1:200), anti–claudin-18 (1:500), anti-CD31 (1:100), anti–hICAM-1 (1:100), or anti–mICAM-1 (1:8,000) was used in Antibody Diluent Background Reducing Solution (DAKO) and visualized using the EnVision Plus Anti-Rabbit Labeled Polymer Kit (DAKO) according to the manufacturer's instructions for the rabbit antibodies. For the rat antibodies, the Rat-on-Mouse Kit was used (Biocare Medical). Images were captured using the ScanScope XT scanner and ImageScope software (Aperio). (ii) Immunofluorescence: sections were subjected to immunofluorescence staining as previously described (8) using anti-amylase (1:300), anti–cytokeratin-19 (1:100), anti-F4/80 (1:200), anti-MMP9 (1:1,000), anti-TNFα (1:200), or anti–ICAM-1 (1:8,000; R&D Systems) in Antibody Diluent Background Reducing solution (DAKO) at 4°C, overnight. After 3 washes with PBS containing 0.05% Tween-20, appropriate Alexa Fluor 488, 594, or 633 labeled secondary antibodies from Invitrogen at a 1:500 dilution were added [room temperature (RT), 1 hour]. DAPI (final concentration 125 μg/mL) was added for 15 minutes after samples were incubated with the secondary antibodies. Fluoromount-G (Southern Biotech) or LabVision PermaFluor (Thermo Scientific) was used as mounting and imaging medium. Images were captured by a fluorescent scanner (ScanScope FL, Aperio) with consistent exposure time and processed using ImageScope software (Aperio).
In Situ Zymography
This method was previously described in detail (48). In brief, frozen pancreas tissue sections (8 μm) were air dried for 10 minutes. A DQ gelatin (0.05 mg/mL)/agarose (1%)/DAPI (1 μg/mL) mixture in PBS was added on top of the tissue section and covered with a coverslip. After gelling for 5 minutes at 4°C, the slides were left at RT for 1 hour, and fluorescence was detected using a fluorescent microscope (IX71; Olympus) and filters for FITC and DAPI.
Isolation and Polarization of Primary Macrophages
Primary murine macrophages were isolated as previously described (8). In brief, mice were intraperitoneally injected with 2 mL of 5% aged thioglycollate solution. At day 5 after injection, peritoneal macrophages were collected through a single injection of 10 mL of RPMI-1640 containing 10% FBS into the peritoneal cavity and subsequent withdrawal. The peritoneal exudate was centrifuged and washed with RPMI-1640 media containing 10% FBS before plating onto tissue culture dishes. After one hour in a 37°C incubator supplemented with 5% CO2, cells were vigorously washed with PBS for three times to remove nonadherent macrophages. To polarize freshly isolated macrophages to an M1 phenotype, 10 ng/mL LPS and 20 ng/mL murine IFNγ were added to the media for 24 hours. To polarize freshly isolated macrophages to an M2 phenotype, 20 ng/mL murine IL4 was added to the media for 24 hours. Polarization for each experiment was controlled by immunofluorescence analysis of cells for iNOS and Ym1 expression (method described in Supplementary Method). Only cell populations that were fully polarized were used for experiments (i.e., M1 = iNOS positive/Ym1 negative; or M2 = iNOS negative/Ym1 positive).
Isolation of Pancreatic Acinar Cells
The procedure to isolate primary pancreatic acinar cells was described in detail previously (8, 49, 50). In brief, the pancreas was removed, washed twice with ice-cold Hank's Balanced Salt Solution (HBSS) media, minced into 1- to 5-mm pieces, and digested with collagenase I (37°C, shaker). The collagen digestion was stopped by the addition of an equal volume of ice-cold HBSS media containing 5% FBS. The digested pancreatic pieces were washed twice with HBSS media containing 5% FBS and pipetted through a 500-μm mesh, and then a 105-μm mesh. The supernatant of this cell suspension containing acinar cells was dropwise added to the top of 20-mL HBSS containing 30% FBS. Acinar cells were then pelleted (1,000 rpm, two minutes, at 4°C) and resuspended in 10-mL Waymouth complete media (1% FBS, 0.1 mg/mL trypsin inhibitor, 1 μg/mL dexamethasone).
3D Organoid Explant Culture of Primary Pancreatic Acinar Cells
This method was described in detail before (8, 49). In short, cell culture plates were coated with collagen I in Waymouth media without supplements. Freshly isolated primary pancreatic acinar cells from wild-type or LSL-KrasG12D mice were added as a mixture with collagen I/Waymouth media on the top of this layer (3D on-top method). Further, Waymouth complete media were added on top of the cell/gel mixture, and were replaced the following day and then every other day. To express proteins using adenovirus, acinar cells were infected with the adenovirus of interest and incubated for three to five hours before embedding in the collagen I/Waymouth media mixture. At day 5, numbers of ducts in three random fields of each sample were determined, and photographs were taken to document structures.
Chemoattraction assays using spatially separated cells in coculture: HeLa cell pools stably expressing GFP or GFP-ICAM-1 were generated by infection with lentivirus harboring GFP or GFP-ICAM-1 followed by selection with puromycin. RAW264.7 cells were in vivo labeled using Vybrant DiI (Invitrogen). Cells were plated in an ibidi removable 2-well silicone culture insert that was placed in a cell culture μ-Dish (ibidi). Twenty-four hours after seeding, the culture inserts were carefully removed. After 24 hours, migration of macrophages toward HeLa-GFP or HeLa-GFP-ICAM-1 cells was assessed by fluorescent microscopy (IX71; Olympus). Chemoattraction assays using Transwell plates: RAW264.7 or primary mouse macrophage cells were in vivo labeled using Vybrant DiO (Invitrogen). Cells (105) were suspended in serum-free DMEM media and placed in the insert of a Transwell plate (5-μm pores; polycarbonate membrane). Dependent on the experiment, either mICAM-1 in serum-free DMEM media or primary pancreatic acinar cells from a LSL-KrasG12D/+ mouse infected with null/control or Cre-expressing adenovirus in serum-free DMEM media were added to the bottom wells of the Transwell plate. Where indicated, the ICAM-1 NAB at a final concentration of 3 μg/mL was added to the bottom wells. Triplicates were used for each condition. After 16 hours, macrophages migrated through the pores were visualized using a fluorescent microscope (IX71; Olympus). Five fields per sample were randomly chosen and counted.
Detection of sICAM-1 Using ELISA
Supernatants of cell culture media were collected, and the concentration of sICAM-1 was determined using human or mouse ICAM-1 ELISA kits from R&D Systems. Assays were performed according to the manufacturer's instructions.
RNA Isolation and Quantitative PCR
Cells were harvested from explant 3D collagen culture by digestion in a 1 mg/mL collagenase solution at 37°C for 30 minutes on a shaker. Cells were washed once with HBSS and twice with PBS, and total RNA isolation was performed using the miRCURY RNA isolation Kit (Exiqon) and the TURBO DNA-free Kit (Ambion) to eliminate residual genomic DNA. The level of mRNA of interest was assessed using a two-step quantitative reverse transcriptase–mediated real-time PCR (qPCR) method. Equal amounts of total RNA were converted to cDNA by the high-capacity cDNA reverse transcriptase kit (Applied Biosystems). qPCR was performed in a 7900HT Fast real-time thermocycler (Applied Biosystems) using the TaqMan Universal PCR Master Mix (Applied Biosystems) with probe/primer sets and the following thermocycler program: 95°C for 20 seconds; 40 cycles of 95°C for one second and 60°C for 20 seconds. All probe/primer sets were purchased from Applied Biosystems (Mm00516023_m1 for mouse ICAM-1, Hs00164932_s1 for human ICAM-1, Mm00618004_m1 for EGFL7, Mm00445235_m1 for CXCL10, Mm00444662_m1 for CXCL11). The amplification data were collected by a Prism 7900 sequence detector and analyzed with Sequence Detection System software (Applied Biosystems). Data were normalized to murine Gapdh, and mRNA abundance was calculated using the ΔΔCT method.
Data are presented as mean ± SD. P values were acquired with the Student t test using Prism (GraphPad Software), and P < 0.05 was considered statistically significant.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: G.-Y. Liou, H. Döppler, P. Storz
Development of methodology: G.-Y. Liou, B. Necela, P. Storz
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): G.-Y. Liou, H. Döppler, B. Necela, D.W. Dawson
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): G.-Y. Liou, H. Döppler, L. Zhang, P. Storz
Writing, review, and/or revision of the manuscript: G.-Y. Liou, H. Döppler, D.W. Dawson, P. Storz
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): B. Edenfield, D.W. Dawson
Study supervision: P. Storz
The authors thank Alicia Fleming for help with the genotyping of mice and Howard C. Crawford for helpful discussion of this article.
This work was supported by the NIH grants CA135102, CA140182, and 50CA102701 (Mayo Clinic SPORE in Pancreatic Cancer; to P. Storz). D.W. Dawson is supported by an American Cancer Society Research Scholars grant (RSG-12-083-01-TBG), the NIH (P01 CA163200), and the Hirshberg Foundation for Pancreatic Cancer Research.
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.