Critical Roles for Macrophages in Islet Angiogenesis and Maintenance During Pancreatic Degeneration

The maintenance and genotyping of E2f1, E2f2, Rag2, GFP transgenic, and Csf1r mutant mice has been previously described (9,17,19). Balb/c mice were purchased from Jackson Labs (Bar Harbor, ME). For splenocyte transfer, 5 × 10⁶ GFP transgenic splenocytes were transferred by subocular injection. For hematopoietic transplants, 1 × 10⁷ nucleated BM cells from 1- to 2-month-old donor mice or 1 × 10⁵ nucleated fetal liver cells from e12.5 embryos were transplanted via subocular injection into irradiated DKO or wild-type (WT) mice. DKO mice were given a half-lethal dose (450 Rad), which is sufficient to result in almost 100% donor contribution to reconstitute hematopoiesis, since these mice exhibit severely defective hematopoiesis (19). Mice transplanted for caerulein experiments were given a lethal dose (900-rad) of X-ray irradiation. Transplants resulted in essentially complete replacement of hematopoiesis by the donor transplant as judged by PCR analyses of Csf1r, E2f, or Rag2 genotype for peripheral blood. Csf1r^+/+ and Csf1r^+/− fetal liver donors for transplants into DKO mice were in the C57Bl/6 background (10 generations backcrossed). Csfr^+/+ and Csf1r^−/− BM donors for transplants into DKO mice were in a mixed 129:C57Bl/6 background. Recipient DKO mice were H-2^b/b and in the C57Bl/6 background (more than four generations). The Csfr mutation in 129:C57Bl/6 was bred two generations into Balb/c (with H-2^d/d genotype confirmed) for transplants into Balb/c recipients for caerulein experiments. Collection of peripheral blood to measure nonfasting blood glucose has been described (17).

Caerulein treatment was performed, with slight modifications, as previously described (21). Briefly, Balb/c mice reconstituted with control or Csf1r mutant BM (6–8 weeks after bone marrow transplant [BMT]) were injected with a combined dose of 64 μg caerulein (American Peptide, Sunnyvale, CA) administered in eight injections a day (1 h apart) over 2 days. The University of Colorado Denver Animal Care and Use Committee approved all mouse experiments.

Tissue fixation, histology, immunohistochemistry, and immunofluorescence were performed as previously described (21). Antibodies used were guinea pig α-insulin (1:1,000; Dako A0564), mouse α-glucagon (1:1,000; Sigma G2654), rabbit α-amylase (1:1,000; Sigma A8273), mouse α-bromodeoxyuridine (BrdU) (1:100, Dako M0744), rat α-MECA-32 (1:100; BD Pharmingen 550563), mouse α-vascular endothelial growth factor (VEGF):VEGF receptor (VEGFR) complex (1:10; East Coast Bio CD301), rat α-F4/80 (1:100, Serotec MCA97GA), and rabbit α-MMP9 (1:100; Abcam ab16306). Secondary antibodies were purchased from Jackson Immunoresearch and used at a concentration of 1:100. Images were obtained with a Leica DM5000 microscope and an Evolution MP Color camera and were processed using ImagePro software from Media Cybernetics. Morphometric determination of islet, islet vascularization, and VEGF:VEGFR area and mass was performed using ImagePro software to quantitate areas bound by a particular fluorescent antibody together with visual inspection to eliminate artificially stained areas. A minimum of three randomly chosen areas, at least 1 mm apart, for each tissue was analyzed per mouse.

Pancreata were mechanically dispersed and single cell suspension was washed in PBS containing 0.5% BSA. Pelleted cells were stained with combinations of the following antibodies (all at 1:200 dilutions): phycoerythrin (PE) linked anti-Ter119, phycoerythrin-cyanine (PE-Cy7)–anti-Gr1, fluorescein-linked–anti-Mac1 (PharMingen), and allophycocyanin-linked–anti-F4/80 (eBioscience) for 30 min at room temperature. For interleukin (IL)-10R staining, phycoerythrin-linked IL-10R or IgG₁ isotype controls were used (Pharmingen), and Mac1⁺Gr1^lowIL-10R⁺ macrophages were quantified. Staining with antibodies against Ter119 provided a negative gate to eliminate erythrocytes and autofluorescent cells. Cells were washed once in 1 ml 0.5% BSA-PBS/1% anti-Fc antibody and resuspended in 400 μl PBS for analysis using a Cytomics FC 500 flow cytometer and CXP software (Beckman Coulter).

Pancreata from day 3 PBS (n = 4), day 1 caerulein (n = 3), and day 3 caerulein (n = 3) injected mice (the “n” represent biological replicates) were purified using Trizol and the RNeasy mini kits (Qiagen, Valencia, CA). Total RNA (2–3 mg) was used to generate cRNA probes. Preparation of cRNA, hybridization, and scanning of Mouse Expression Array 430 2.0 microarrays were performed according to the manufacturer's protocol (Affymetrix, Santa Clara, CA). Images were processed into intensity data and normalized by global scaling to a target intensity of 500 using the Affymetrix GeneChip Operation Software v1.1.1 that contains the High-Resolution Scanning Update. The resulting data were imported into the Partek analysis program for principal components analyses.

Statistics were calculated using GraphPad Prizm 4 statistical analysis software. Kaplan-Meier survival curves were used to calculate statistical significance for prevention of diabetes. Mann-Whitney U test was used to determine statistical significance of BrdU incorporations, given large variance in islet size. Statistical significance in all other experiments was calculated using the two-tailed Student's t test. For all figures, * indicates a P value from 0.01 to 0.05, ** indicates a P value from 0.001 to 0.01, and *** indicates a P value below 0.001. Standard error is depicted in all graphs.

To determine the BM-derived cell type important for islet protection during pancreatitis, we used BM from mice genetically deficient for different leukocyte subsets. As we previously reported (17), while all untransplanted DKO mice become diabetic, transplantation of WT BM into DKO mice is sufficient to prevent pancreatitis-associated islet loss and diabetes. Transplantation of DKO mice with Rag2^−/− BM was as effective as WT BM in preventing diabetes, demonstrating that mature T- and B-cells are not necessary for prevention of diabetes in DKO mice (Supplementary Fig. 1A) (available online at http://dx.doi.org/10.2337/db07-1577). Transplantation of the CD45⁺ hematopoietic BM fraction rescued DKO mice from diabetes development, while the CD45^neg fraction completely failed to rescue, demonstrating that restoring WT hematopoiesis, and not the mesenchymal lineage, is required to prevent diabetes development (Supplementary Fig. 1B).

DKO mice are deficient in production of macrophages (Fig. 1A). Before receiving BMT, very few macrophages were present in pancreata of DKO or WT mice. But 6 weeks after transplantation with WT BM, macrophage infiltration was evident in DKO but not WT pancreata (Fig. 1B). Furthermore, injection of GFP transgenic splenocytes demonstrated substantial migration of macrophages to DKO, but not WT, pancreata (Fig. 1C).

To determine the importance of macrophages in diabetes development, DKO mice were transplanted with Csf1r^−/− BM or fetal liver cells, which are deficient for production of macrophages (9). While reconstitution of hematopoiesis with WT hematopoietic progenitors efficiently prevented diabetes development in DKO mice, DKO mice reconstituted with Csf1r^−/− hematopoietic progenitors developed diabetes with kinetics indistinguishable from untransplanted DKO mice (Fig. 1D and Supplementary Fig. 1A). DKO mice reconstituted with WT BM, but not Csf1r^−/− BM, exhibited substantial macrophage infiltration into the pancreas (Fig. 1B and E). Importantly, while transplantation with WT BM leads to islet maintenance in DKO mice, Csf1r^−/− BMT recipients failed to maintain islet mass (Fig. 2A–C). The degeneration of the exocrine pancreas was not alleviated by either type of transplant (Figs. 1F and 2B) (17). These data demonstrate that macrophages preferentially migrate to degenerating pancreata and are essential for maintaining islet mass and preventing diabetes during pancreatitis in DKO mice.

We considered that macrophages may reduce the release of damaging enzymes during pancreatitis, thus protecting islets. Activated trypsin in the pancreas is a hallmark of pancreatitis in humans and is believed to be a triggering event (22). While DKO pancreata possessed a substantial amount of activated trypsin (undetectable in control pancreata), transplantation with WT BM or Csf1r^−/− BM failed to affect the presence of activated trypsin in the pancreas (Supplementary Fig. 2), further arguing that macrophages do not prevent islet loss by alleviating exocrine degeneration.

Macrophages have been shown to promote angiogenesis (11). Furthermore, β-cells produce a high amount of VEGF-A and the VEGF receptor 2 (VEGFR2) is preferentially expressed within islet vasculature (23). The activation and release of matrix bound VEGF by matrix metalloproteinase (MMP)-9 expressing hematopoietic cells has been shown to be required to induce angiogenesis in tumors (24). We asked whether macrophages were similarly promoting angiogenesis in islets during pancreatitis. Islet vasculature is decreased in DKO pancreata (Fig. 2D and E), presumably as the result of chronic pancreatitis. Importantly, islet vascularization in DKO mice transplanted with WT BM mirrored that in WT islets, while DKO mice transplanted with Csf1r^−/− BM had low levels of vascularization similar to islets in untransplanted DKO mice (Fig. 2D and E). Moreover, DKO mice transplanted with WT BM exhibited more MMP9⁺ cells within the pancreas than DKO mice transplanted with Csf1r^−/− BM (Fig. 2F and G). Many of these MMP9⁺ cells also expressed the macrophage marker F4/80 (inset in Fig. 2G). Note that despite reduced macrophage production in DKO mice, degenerating DKO pancreata still show some MMP9⁺ cell infiltration, which is significantly augmented by WT BMT.

Since increased islet vasculature and MMP9-expressing macrophages were present in pancreata of DKO mice after WT BMT, we measured angiogenesis by using the GV39M antibody, which recognizes VEGF-A bound to VEGFR2 (25), which we will refer to as “VEGF:R.” We observed a large increase in VEGF:R within islets of DKO mice that were transplanted with WT BM (Fig. 3A and B), and ∼75% of islet vasculature colocalized with VEGF:R (Fig. 3A and C). Interestingly, neither in islets from WT mice nor from DKO mice transplanted with Csf1r^−/− BM was there appreciable VEGF bound to its receptor, and most islet vasculature did not colocalize with VEGF:R (Fig. 3A–C). Thus, pancreatitis leads to macrophage-dependent VEGF release from the extracellular matrix, which then associates with receptors expressed on islet vasculature. As expected, the islet vasculature in healthy pancreata was largely unbound by VEGF, consistent with minimal angiogenesis in an adult tissue with low cell turnover.

Finally, we measured the number of islet cells progressing through S-phase by BrdU incorporation. While DKO mice transplanted with Csf1r^−/− BM had very few islet cells that incorporated BrdU, transplantation of DKO mice with WT BM resulted in a significant increase in islet cells incorporating BrdU (Fig. 3D and E). The replication of β-cells could be a secondary effect of increased angiogenesis and/or could represent a direct trophic effect of macrophages toward β-cells. In all, these data indicate that macrophages are required during chronic exocrine degeneration to promote islet angiogenesis and islet maintenance and prevent diabetes development.

While the phenotype seen in DKO mice is similar in some aspects to chronic pancreatitis in humans, there are also clear dissimilarities. As for the human disease, DKO mice do have degeneration of the exocrine pancreas, pancreatic fibrosis, fat deposits, activation of trypsin, and, after restoration of WT hematopoiesis, migration of macrophages to the degenerative organ. However, polyploidization of exocrine tissue has not been described as a cause of pancreatitis in humans. We thus sought to test our findings in another rodent model of pancreatitis to further validate roles for macrophages in islet protection in the face of exocrine degeneration.

To extend our studies to another model of pancreatitis, we explored the importance of macrophages in protecting islets during caerulein-induced pancreatitis. Caerulein is a peptide analog of the pancreatic secretagogue cholecystokinin. High alcohol intake increases serum cholecystokinin levels and potentiates premature activation of zymogens (26). Caerulein is frequently used to model acute pancreatitis in mice (21), and acute pancreatitis is frequently associated with hyperglycemia (usually transient) in humans (27). Treatment of mice with caerulein causes hypersecretion of exocrine enzymes into the pancreas and exocrine destruction, with full recovery in about a week (21). Upon treatment of WT mice with caerulein, macrophages rapidly migrated to the degenerative pancreas starting 1 day after treatment (Fig. 4A–C and E–G), leaving by ∼5 days post-caerulein (Fig. 4D and H). Migration of lymphocytes to the pancreas is delayed relative to macrophage recruitment (Fig. 4I–P). Finally, caerulein treatment increased pancreatic CSF1 levels, suggesting a possible mechanism for macrophage recruitment (Fig. 4Q).

We analyzed global gene expression changes in pancreata associated with caerulein treatment. Pancreatic RNA from PBS- and caerulein-injected mice were harvested 1 and 3 days after treatment. Genes were clustered into different A and B groups for increased and decreased (respectively) expression in the caerulein groups relative to PBS (Fig. 5A). That gene expression analyses were performed on whole pancreata complicates our interpretations. Nonetheless, we can extract genes whose expression is known to exhibit some restriction to macrophages, revealing an extensive macrophage signature in the caerulein-treated pancreata. Observed expression patterns for chemokines, cytokines, growth factors, and cell surface receptors are similar to those seen with TAMs and M2 (type 2) macrophages in other studies (28–31) (Fig. 5B, Supplementary Fig. 3, and Supplementary Table 1). The expression of arginase1, mannose receptors, and IL-10 and its receptor, previously associated with TAMs and M2 macrophages, is increased during caerulein-induced pancreatitis and is known to promote and influence trophic macrophage functions (32). Moreover, expression of CSF1, CSF1R, and MMP9 mRNAs were all greatly increased in pancreata after treatment with caerulein, with the latter two presumably reflecting macrophage infiltration during pancreatitis. We confirmed that IL-10 receptor expression on infiltrating macrophages is increased after caerulein treatment (Fig. 6A–C). We also confirmed that pancreatitis results in CSF1R-dependent infiltration of MMP9⁺ cells, which cluster around islets (Fig. 6D–E). Finally, the expression of over 22 different chemokines was increased after caerulein treatment (Fig. 5B). Included in this group were CCL2, CCL3, CCL5, and CCL8, all of which have been implicated in migration of TAMs to tumors (33).

Given that macrophages migrate to the pancreas after caerulein treatment and the postcaerulein expression profile indicative of TAMs, we asked whether macrophages are necessary for maintaining islets after caerulein treatment. WT mice were transplanted with WT or Csf1r^−/− BM, and 8 weeks after BMT, the mice were treated with PBS or caerulein. Pancreatic macrophage infiltration was observed after caerulein treatment in mice transplanted with WT BM, but not with recipients of Csf1r^−/− BMT (Fig. 7A and B). After caerulein treatment, islet mass was decreased in both groups of mice. However, by day 5 after caerulein treatment, islet mass in the recipients of WT BMT is stabilized, while that of the Csf1r^−/− BMT recipients exhibited further decline (Fig. 7C). By day 10, islet mass in caerulein-treated WT and Csf1r^−/− recipients had returned to basal levels. Consistent with their reduced maintenance of β-cells, day 5 caerulein-treated Csf1r^−/− BMT recipients were significantly delayed in the clearance of a glucose bolus compared with WT BMT recipients (Supplementary Fig. 4B). Importantly, while macrophages are needed to maintain islet mass, there is no difference in the caerulein-induced destruction and subsequent regeneration of pancreatic exocrine tissue, as determined both by histology and analyses of serum amylase levels (Supplementary Fig. 4A; data not shown), indicating that macrophages are needed for islet maintenance but not exocrine regeneration after caerulein treatment.

Islet vascularization was similar between PBS-treated Csf1r^−/− or WT BM recipients, indicating that macrophages are not essential for maintaining islet vasculature under steady-state conditions (Fig. 7D and E). However, while caerulein-treated WT BM recipients had normal levels of islet vascularization, caerulein-treated Csf1r^−/− BM recipients had a significant decrease in islet endothelial cells per islet area. Thus, the loss of islet vasculature seems to be outpacing β-cell loss. Very low levels of VEGF:R were observed within islets of PBS-treated mice. Interestingly, while both caerulein-treated WT and Csf1r^−/− recipients had increased levels of VEGF:R, consistent with pancreatitis-induced vasculature remodeling, macrophage deficiency severely attenuated this increase in intra-islet VEGF:R (Fig. 7F and G). These data demonstrate that macrophages are needed to promote angiogenesis in islets after caerulein-induced pancreatitis.

Pancreatitis frequently leads to insulin-dependent diabetes, which is generally believed to be the inevitable consequence of exocrine destruction. Our results reveal an endogenous mechanism by which islets are normally protected during exocrine degeneration. While inflammatory cells are normally thought to instigate β-cell loss, we describe a novel role for macrophages in promoting islet angiogenesis and maintaining β-cells in two mechanistically distinct models of pancreatitis. As therapeutic approaches are available to manipulate macrophage production and function, promotion of trophic macrophages could be an approach to protect islets and prevent diabetes associated with pancreatic degeneration.

Tumor angiogenesis is dependent on macrophage-mediated remodeling of vasculature, and MMP9 expression by myeloid cells is needed for VEGF-induced angiogenesis (24,34–36). Our data demonstrate that MMP9-expressing macrophages may play a similar critical role in islet angiogenesis during pancreatitis, since the infiltration of MMP9⁺ cells, VEGF activation, and islet angiogenesis during exocrine degeneration is highly deficient in recipients of Csf1r^−/− BM. While F4/80-expressing macrophages are not observed directly proximal to or within islets in the DKO model after WT BMT, a requirement for macrophages nonetheless is evident in islet maintenance, suggesting that macrophages may not need to be continuously within islets to confer protection.

An intimate relationship exists between β-cells and vascular endothelial cells, and pancreas development depends on the presence of vasculature (37,38). The endocrine pancreas receives a disproportionate amount of vasculature (39), and vascular endothelial cells produce basement membrane proteins that promote β-cell proliferation and function (40). We propose a model whereby exocrine degeneration damages islets and their vasculature and results in CSF1-mediated recruitment of macrophages to the pancreas. Expression of MMPs by these macrophages then contributes to activation of VEGF, angiogenesis, and consequent islet cell proliferation and survival, which is important to maintain normal glucose homeostasis during pancreatitis. However, at present, we do not know if VEGF and/or MMP9 are necessary for increased angiogenesis and islet maintenance during pancreatitis.

Our studies also suggest differential requirements for macrophages in protecting islets under different contexts of pancreatic degeneration. During the chronic pancreatitis-like syndrome in DKO mice, functioning macrophages are absolutely required to prevent islet loss and diabetes. However, while macrophages are important to maintain islet vasculature and β-cells during caerulein-induced acute pancreatitis, macrophage deficiency does not result in diabetes, and islets eventually recover. Thus, macrophages and macrophage-initiated angiogenesis appear necessary to maintain islet vasculature and islets during damage, but other mechanisms can restore islet mass after the damaging insult is gone. This damage-dependent role is also consistent with the observation that CSF1-dependent macrophages are dispensable under steady-state conditions in an adult for islet vasculature and islet maintenance, consistent with the paucity of macrophages in healthy adult pancreata.

There are additional clues in the literature suggesting roles for myeloid lineage cells in islet maintenance. Transplanted mouse islets lose their vasculature, but granulocyte/macrophage-CSF injection led to increased vascularization, which was attributed to angioblast mobilization (41). Furthermore, transgenic expression of granulocyte/macrophage-CSF in islets recruits leukocytes (mostly expressing F4/80) to the pancreas and reduces the penetrance of streptozotocin-induced diabetes, although streptozotocin-mediated islet damage was not noticeably altered (42).

Most current estimates propose that only ∼1% of diabetes cases are caused by pancreatitis (43). However, in several large studies, exocrine insufficiency (significantly lower fecal elastase) was measured in >50% of type 1 diabetic patients and ∼35% of type 2 diabetic patients (44–46). While the accuracy of the fecal elastase test for diagnosing pancreatitis is controversial, the majority of diabetic patients exhibiting very low fecal elastase levels experienced steatorrhea, which reflects severely depleted exocrine function (47). Furthermore, a large fraction of type 1 and type 2 diabetic patients were shown to have hallmarks of pancreatitis by endoscopy (48). Pancreatitis also appears much more common in the general public than appreciated, since around 5–13% of nondiabetic subjects and about twice as many diabetic subjects exhibited chronic pancreatitis upon autopsy (3). While many reports have argued that diabetes may lead to exocrine degeneration and dysfunction (such as by autoimmune targeting of exocrine cells or decreased trophic action of insulin on exocrine tissue), Hardt and colleagues (3,45) postulate that diabetes may be secondary to pancreatitis in far more cases than currently appreciated. Still, cause and effect relationships are not clear from these studies. Nonetheless, several observations do suggest that links between pancreatic exocrine degeneration, macrophage infiltration, and islet status during type 1 and type 2 diabetes deserve further study, including 1) pancreatitis can clearly cause islet loss (1,2), 2) increased numbers of islet-associated macrophages are present in type 2 diabetic people and rodents (49), 3) known roles for macrophages in tissue support (6), and 4) studies presented here that demonstrate macrophage-dependent angiogenesis and islet maintenance. Trophic macrophages may be essential for protecting islets in the face of exocrine degeneration, and while speculative, the loss of trophic macrophage protection or conversion of macrophages to a more destructive phenotype may contribute to islet loss in late-stage pancreatitis or type 2 diabetes.

J.D. was supported by grants from the National Institutes of Health (NIH RO1 CA77314 and R21 DK063299), J.S.T. by a predoctoral training grant (5T32 GM08730), H.P. by The Finnish Medical Foundation and the Cancer Society of Southern Finland, J.J. by a grant from the National Institutes of Health (NIH) Beta Cell Biology Consortium (DK61248), E.R.S. by NIH CA32551 and 5P30-CA13330, and X.-M.D. by an ASH Fellow Scholar Award and an LLS Special Fellow Award.

We would like to thank Lori Sussel and Peter Henson for advice, Sayan Nandi for performing the CSF1 radioimmunoassay, Maria Veronica Albertoni for helping to generate the array data, and Alecksandr Kutchma for array analyses. We also thank K. Helm and C. Childs of the Cancer Center Flow Cytometry Core, supported by grant P30 CA 46934.