The non-obese diabetic (NOD) mouse is a widely used animal model for the study of human diabetes. Before the start of lymphocytic insulitis, DC accumulation around islets of Langerhans is a hallmark for autoimmune diabetes development in this model. Previous experiments indicated that an inflammatory influx of these DCs in the pancreas is less plausible. Here, we investigated whether the pancreas contains DC precursors and whether these precursors contribute to DC accumulation in the NOD pancreas. Fetal pancreases of NOD and control mice were isolated followed by FACS using ER-MP58, Ly6G, CD11b and Ly6C. Sorted fetal pancreatic ER-MP58+ cells were cultured with GM-CSF and tested for DC markers and antigen processing. CFSE labeling and Ki-67 staining were used to determine cell proliferation in cultures and tissues. Ly6Chi and Ly6Clow precursors were present in fetal pancreases of NOD and control mice. These precursors developed into CD11c+MHCII+CD86+ DCs capable of processing DQ-OVA. ER-MP58+ cells in the embryonic and pre-diabetic NOD pancreas had a higher proliferation capacity. Our observations support a novel concept that pre-diabetic DC accumulation in the NOD pancreas is due to aberrant enhanced proliferation of local precursors, rather than to aberrant “inflammatory infiltration” from the circulation.
The non-obese diabetic (NOD) mouse is used as a spontaneous model to study the development of type 1 diabetes 1. Lymphocytes accumulate around and in the islets of Langerhans in NOD mice from around 6 weeks of age onwards, which results in the destruction of β-cells followed by a decrease in insulin production leading to diabetes. Prior to T- and B-cell accumulation the number of DCs increases in the pancreas and concentrates around the islets (from the age of 5 weeks onwards) 2, 3. DCs are potent APCs capable of stimulating both naïve and memory T cells 4. The observation that DCs are the first immune cells to increase in number in the NOD pancreas points to a crucial role for DCs in the initiation of the islet autoimmune reaction. Such a role was recently proven by the demonstration that a temporal depletion of DCs totally abrogated the development of insulitis and diabetes in the NOD mouse model 5.
Early studies have shown that BM precursors give rise to monocytes in blood, which circulate for a few days before they migrate into tissue where they develop into different types of DCs and macrophages. Blood monocytes can be subdivided into at least two subsets based on their Ly6C expression: classical and nonclassical monocytes. The classical monocytes, which are Ly6Chi, are selectively recruited to inflamed tissues and lymph nodes and differentiate into inflammatory DCs 6. The nonclassical monocytes, which are Ly6Clow, patrol the endothelium of the blood vessels and are required for rapid tissue invasion at the site of an infection 7. Ly6Clow monocytes are considered CD11c−, but some studies have reported the expression of CD11c on these cells 6, 8. Both types of monocytes are F4/80+ and CD86−6.
Data are accumulating on the presence of local tissue precursors for DCs and macrophages and the contribution of these precursors to DC and macrophage accumulation under pathological conditions. In organs, such as the skin and brain, local precursors for macrophages and Langerhans cells have been detected 9–11. We earlier described the presence of local precursors for macrophages in the fetal pancreas of C57BL/6 mice 12. However, little is known about the origin of the DCs that accumulate in the pre-diabetic NOD pancreas and the factors driving this accumulation. It is generally assumed that these cells are inflammatory in nature and infiltrate from the circulation. However, previous studies from our group suggest that the early accumulation of DCs in the pre-diabetic NOD pancreas cannot only be explained by a massive influx of DCs and DC precursors from the blood. First, pro-inflammatory chemokines that normally attract monocytic cells (CCL2 and CCL3) could not be detected in the pancreas at the time of DC accumulation 13. Second, DCs and monocytes of NOD mice have an impaired migration towards pro-inflammatory chemokines in vivo and in vitro 13, although the contribution of other chemokines cannot be excluded. Finally, the depletion of phagocytic cells with clodronate resulted in a late re-appearance of DCs in the NOD pancreas (28 days after depletion), while monocytes and DCs had already re-appeared in the blood and spleen 4 days after depletion. This late re-appearance suggests that pancreatic DCs are not only replenished from the circulation 14.
We therefore hypothesized that local precursors for DCs are present in the pancreas and that an enhanced proliferation and differentiation of these cells is responsible for the enhanced accumulation of pancreatic DCs initiating the islet autoimmune reaction.
In this study, the presence of local pancreatic precursors for DCs, their proliferative capacity and the actual generation of DCs from these pancreatic precursors was investigated in the fetal pancreas and the pre-diabetic pancreas of NOD and control mice.
Myeloid precursor populations in the fetal pancreas
The presence of precursors for DCs in the fetal pancreas was studied using the myeloid progenitor marker ER-MP58. ER-MP58 has previously been described by our laboratory as a marker for all myeloid progenitor cells in BM 15. A double staining with ER-MP58 and insulin was performed on the E15.5 pancreas of C57BL/6 and NOD/LTj mice using immunofluorescence (Fig. 1). The results showed that ER-MP58+ cells were present in and around the insulin positive islets of Langerhans in the E15.5 pancreas.
To investigate the phenotype of this myeloid precursor in the pancreas a FACS staining was performed on fetal pancreas cells and compared with blood monocytes (4 weeks) from C57BL/6 and NOD/LTj mice. In the blood, SSClow cells were gated on ER-MP58+Ly6G− cells. These cells were subdivided into two populations: CD11bhiLy6Chi (classical) and CD11bhiLy6Clow (non-classical) monocytes (Fig. 2A). In the fetal pancreas two precursor populations were present with a similar phenotype as blood monocytes. Due to a genetic abnormality of the Ly6C gene in NOD mice the expression of Ly6C is present, but significantly lower than in control mice 16.
The phenotype of the two monocyte populations was further characterized using Ab against CD11c, F4/80 and CD86. In blood, Ly6Chi monocytes were CD11clowF4/80+CD86low in both C57BL/6 and NOD mice (Fig. 2B). Ly6Clow blood monocytes expressed CD11c. Two CD11c+ cell populations were observed: CD11clow and CD11chi. The Ly6Clow blood monocyte population of NOD mice had more CD11chi cells than in C57BL/6 mice. Ly6Clow blood monocytes were F4/80+CD86low in both strains. In the fetal pancreas Ly6Chi cells were CD11c−F4/80+CD86− in C57BL/6 and NOD mice. In the fetal pancreas Ly6Clow cells were F4/80+CD86− and expressed CD11c, although not that high as the Ly6Clow blood monocytes. No differences were observed between C57BL/6 and NOD fetal pancreas. Thus, in the fetal pancreas two myeloid precursor populations (Ly6Chi and Ly6Clow) were present. These cells showed a similar expression of F4/80 as blood monocytes, but had a lower CD11c expression on Ly6Clow cells and lacked CD86.
Isolated myeloid precursors from fetal pancreas differentiate into functional DCs
To show that ER-MP58+ cells in the fetal pancreas are able to develop into CD11c+ DCs, ER-MP58+ cells were isolated by cell sorting followed by culture with GM-CSF. After culture for 8 days the generated cells displayed a typical DC appearance with dendrites (Fig. 3A). More than 40% of these cells expressed CD11c and expressed MHCII and the co-stimulatory molecule CD86 (Fig. 3B). The absolute number of generated CD11c+ cells from cultured pancreatic ER-MP58+ cells was significantly higher in NOD than in C57BL/6 (Fig. 3C). The generated CD11c+ cells from NOD and C57BL/6 were able to quench DQ-OVA showing the capability to process antigens (Fig. 3D). No significant difference in the DQ-OVA expression was detected between NOD and C57BL/6.
Increased proliferation of myeloid precursors in NOD fetal pancreas
A property of precursors is their proliferative capacity; therefore the proliferation of precursors in the fetal pancreas was analyzed by flow cytometry using Ki-67. In NOD fetal pancreas the number of Ly6ChiKi-67+ cells was significantly higher than in C57BL/6 (2.5-fold). No difference was found in the number of Ly6ClowKi-67+ cells between NOD and C57BL/6 (data not shown).
To determine the proliferative capacity of ER-MP58+ cells in culture we used CFSE labeling. ER-MP58+ cells from the fetal pancreas, fetal liver, adult BM and blood were labeled and cultured with GM-CSF. Microscopic evaluation on day 4 of the GM-CSF culture of ER-MP58+ cells from the NOD fetal pancreas revealed increased cell numbers compared to C57BL/6 and BALB/c cultures (Fig. 4A). After 2 days culture the CFSE signal on half of the ER-MP58+ cells from the NOD fetal pancreas was decreased, in contrast to the C57BL/6 (Supporting Information Fig. 1A). After 5 and 8 days culture the CFSE signal of ER-MP58+ cells from the NOD fetal pancreas was dramatically decreased in line with a high proliferative activity (Fig. 4B and Supporting Information Fig. 1A). No such a decrease was detected in C57BL/6. Although a decrease of the CFSE signal was detected in the BALB/c fetal pancreas, the decrease was less compared with NOD. In the fetal liver as well as in the adult BM the majority of ER-MP58+ cells showed a low CFSE signal, with no differences between NOD and controls. The number of CFSElow cells in the culture of ER-MP58+ cells from the NOD fetal pancreas was significantly higher compared with controls. Cells with at least 5 divisions were counted as CFSElow cells (Fig. 4C).
As monocytes in the peripheral blood also express ER-MP58 these cells were analyzed for their proliferative capacity too. The CFSE signal of day 8 cultures of ER-MP58+ cells from the blood was not decreased, showing that ER-MP58+ peripheral blood monocytes were not able to proliferate after GM-CSF stimulation (Supporting Information Fig. 1B). In conclusion, myeloid precursors in the NOD fetal pancreas have a specific proliferation abnormality.
Increased proliferation of ER-MP58 cells in pre-diabetic NOD pancreas
DCs are the first cells that start to accumulate around the islets in the pancreas at 5 weeks of age in the pre-diabetic NOD mice. To investigate if this DC accumulation is preceded by an increased proliferation of local pancreatic precursors the pre-diabetic pancreas was studied for ER-MP58+Ki-67+ cells by immunofluorescence and FACS analysis. To assess if the proliferation abnormality in the NOD pancreas is a general phenomenon of the genetic background of these mice, the non-obese resistant mouse (NOR) was included as an extra control. In the NOD pancreas of 5 weeks of age the number of ER-MP58+Ki-67+ cells was significantly higher compared to C57BL/6 and NOR (Fig. 5A and B). This was confirmed by FACS analysis of the pancreas of 5–week-old NOD, NOR and C57BL/6 mice (Supporting Information Fig. 2 and 5C). No significant difference in the total number of ER-MP58+ cells between NOD, NOR and C57BL/6 was detected (data not shown). Thus, proliferating myeloid precursors are present before the DC accumulation in the NOD pre-diabetic pancreas and this is not due to the genetic background of this mouse.
We here show that ER-MP58+Ly6G−CD11bhiLy6Chi and ER-MP58+Ly6G−CD11bhiLy6Clow precursors for myeloid DCs are present in the pancreas of C57BL/6 and NOD mice from embryonic (E15.5) age onwards. After sorting and culture in GM-CSF, these precursors have the potential to develop into CD11c+MHCII+CD86+ DCs capable of processing antigens. Although the number of precursors is not increased in the NOD mouse pancreas, the cells have a higher proliferative capacity in the embryonic as well as in the pre-diabetic NOD pancreas. This abnormality was specific for the pancreas and did not occur in blood, liver and BM.
It is assumed that the autoimmune process in the NOD mouse starts with DC accumulation around 5 weeks of age. However, the presence of abnormal DC precursors in the fetal and pre-diabetic pancreas of NOD mice indicates that the autoimmune process in the NOD mouse starts much earlier. Several studies showed aberrancies already in the pre-diabetic NOD mice. An increased level of the extracellular matrix protein fibronectin was found in the early postnatal NOD pancreas, and is associated with an enhanced accumulation of macrophages and altered islet morphology 17. In the early neonatal pancreas of NOD mice abnormalities in DC and macrophage populations were described 18.
ER-MP58 is a marker which is present on all myeloid progenitors. However, some non-myeloid cells can express this marker at low levels 15. Isolated ER-MP58+ cells from the pancreas were used in cultures with GM-CSF and developed into DCs. Only cells of the myeloid lineage will respond to this growth factor 19.
BM cells from NOD mice have previously been shown by several groups to have reduced responses to GM-CSF 20, 21. In contrast, myeloid precursors from NOD fetal pancreas showed an increased response to GM-CSF compared with C57BL/6. These cells had an increased proliferation and produced more DCs, suggesting a proliferation and/or apoptotic defect in myeloid precursors in the NOD fetal pancreas and indicating towards an intrinsic abnormality of these cells. Interestingly, it has been described that NOD myeloid cells have a high GM-CSF expression 22. This suggests that if the pancreatic precursors exhibit this phenotype as well, an autocrine loop driven by GM-CSF might contribute to the abnormal expansion and differentiation of the local pancreas DC precursors in the NOD mouse. However, a contribution of additional signals from the pancreatic tissue itself might explain why at specific ages waves of DC accumulation have been observed.
Our observations on the presence of abnormal local precursors in the NOD pancreas are suggestive for a new concept on the role of local pancreatic DC precursors in the development of diabetes. This proposed model differs from current paradigms of acute inflammation, where Ly6Chi monocytes are recruited from the circulation to a site of pre-autoimmune injury to become DCs 23–25. In our concept inflammation and organ-specific autoimmunity use different routes for accumulation of DCs in target organs-to-be and suggest that the accumulating DCs in the NOD pancreas are different from the well-characterized TNF/iNOS-producing DCs (TIP-DCs) that are recruited from the peripheral blood to sites of inflammation.
A large body of research has been carried out on the development of DCs in various lymphoid tissues from BM precursors. The macrophage and DC precursor (MDP) for lymphoid tissue conventional DCs (cDCs), pDCs and monocytes is characterized as a cell expressing Lin−c-kithiCD115+CX3CR1+Flt3+8, 26. Another distinct progenitor is called the common DC precursor (CDP) (Lin−c-kitlowCD115+Flt3+) and is restricted to produce cDCs and pDCs, but not monocytes 27, 28. Preliminary data showed that ER-MP58+ cells do not express Flt3 and do not produce pDCs when cultured in the presence of Flt3 in the fetal and pre-diabetic pancreas. This suggests that our pancreas DC precursor is distinct from the MDP or CDP. We therefore assume that the local pancreatic precursor has a unique phenotype different from peripheral blood monocytes and precursors for cDCs in the BM.
Our study has limitations. One could argue that the local precursors are not present in the “pancreas-anlage” itself, but in the vicinity of this tissue in specialized blood-forming tissues, like the aorta-gonad-mesonephros (AGM) and the fetal liver. In this study the preparation method excludes these organs, which strongly argues in favor of a presence of the precursors in the fetal pancreas itself.
Second, the local pancreatic precursor could simply represent early seeded monocytes in the tissues. Indeed, the local pancreas DC precursor has a similar phenotype as blood monocytes, except for the lower CD11c expression on the Ly6Clow cells and is expressing ER-MP58, which is a marker for both myeloid precursors in the BM and peripheral blood monocytes 15. Upon GM-CSF stimulation the local ER-MP58+ cells isolated from fetal pancreas displayed a high proliferative activity. Such a proliferation was not observed in cultures of ER-MP58+ monocytes isolated from NOD peripheral blood. It is known that blood monocytes are nondividing cells 24. These data, the presence of ERMP58+ cells in the pancreas from embryonic live onwards and the observation of Ki-67+ER-MP58+ cells in the pre-diabetic pancreas support our conclusion that this ER-MP58+ cell is a myeloid precursor cell distinct from a peripheral blood monocyte. However, the possibility that migrating blood monocytes are modified by the microenvironment of the pancreas and obtain a proliferative capacity cannot be excluded completely.
The proliferation/differentiation aberrancies of local NOD pancreatic DC precursors described here are very similar to the aberrancies previously found by us in DC precursors of the BM in the animal models of type 1 diabetes 29. DC precursors in BM of NOD mice and BB-DP rats also show proliferation/differentiation abnormalities and from these precursors abnormal “steady state” DCs arise with a spontaneous high pro-inflammatory set point 29, 30. These abnormal DCs have a high level of NF-kB and a high acid phosphatase, high IL-12 and low IL-10 expression 31–34. These DCs are incapable of sufficiently sustaining the proliferation of Treg-cell populations in the NOD mouse and BB-DP rat 35, 36. It has been shown that correction of these DC abnormalities prevents the development of autoimmune diabetes 37, 38. It is tempting to speculate that the locally generated DCs in the pancreas of NOD mice show a similar pro-inflammatory set point as their BM correlates and cannot sustain Treg cells sufficiently.
Materials and methods
C57BL/6 mice were obtained from Charles River Laboratories (Maastricht, The Netherlands), BALB/c mice from Harlan (Horst, The Netherlands) and NOR/LTj mice from the Jackson Laboratory (Bar Harbor, ME, USA). NOD/LTj mice were bred in our own facility under specified pathogen-free conditions. Breedings were done from the age of 8 weeks and older. The appearance of the vaginal plug was noted as E0.5. Pregnant mice were sacrificed and embryos dissected at embryonic age of E15.5. BM cells were isolated from the femora from mice of 8 weeks. All mice were female and were supplied with water and standard chow ad libitum. Experimental procedures were approved by the Erasmus University Animal Ethical Committee.
Preparation of cell suspensions
Embryonic (E15.5) pancreas (pooled) and liver were isolated and micro-dissected from the stomach and digested with Collagenase Type 1 (1 mg/mL), hyaluronidase (2 mg/mL) (both Sigma Aldrich, St. Louis, MO, USA) and DNAse I (0.3 mg/mL) (Roche Diagnostics, Almere, The Netherlands) for 10 min at 37°C. Embryonic pancreas and liver cells were flushed through a 70 μm filter and washed. Pancreases of 5-week-old mice were isolated after a cardiac perfusion and cut into small pieces and digested with Collagenase Type 1, hyaluronidase and DNAse I for 40 min at 37°C. Cells were flushed through a 70 μm filter and washed. Blood of 4 week old mice was collected in EDTA tubes using a heartpunction. Erythrocytes were lysed with NHCL2 buffer and washed. Single-cell suspensions of BM were prepared as described previously 39. All cells were resuspended in PBS containing 0.1% BSA and were ready for flow cytometry staining.
Single-cell suspensions from pancreas (E15.5 and 5 wk) were labeled with mAbs. Antibodies used were ER-MP58-biotin (own culture), Ly6C-FITC (Abcam, Cambridge, UK), Ly6G-Pacific Blue (Biolegend, Uithoorn, The Netherlands), CD11b-allophycocyanin-Cy7, CD86-PE (both Becton Dickinson, San Diego, CA, USA), CD11c-allophycocyanin, CD11c-PE, CD11c-PE-Cy7, CD86-Pacific Orange, F4/80-PE-Cy5 (all eBiosciences, San Diego, CA, USA), MHC class II-PE (C57BL/6, clone M5/114, Becton Dickinson) and MHC class II-biotin (NOD clone 10.2.16, own culture). Afterwards cells were washed and incubated with streptavidin-allophycocyanin (Becton Dickinson). To detect proliferation, the cells were fixed in 2% paraformaldehyde, and permeabilized using 0.5% saponin. Subsequently, cells were incubated with Ki-67-FITC (Becton Dickinson) diluted in 0.5% saponin, washed and resuspended in 0.1% BSA. Cells suspensions were analyzed using a FACS Canto HTSII (Becton Dickinson) flow cytometer and FACS Diva and Flowjo software.
Antigen processing was determined by measurement of the fluorescence upon proteolytic degradation of the self-quenched conjugate DQ-Ovalbumin 40. Briefly, cells were resuspended in PBS with 2% FCS and 100 μg/mL DQ-Ovalbumin (Molecular Probes, Breda, The Netherlands) and incubated for 30 min at 37°C. Cells were washed and incubated with CD11c-allophycocyanin and analyzed by flow cytometry.
Cell sort experiments
Cells from the embryonic pancreas (pooled), liver and adult BM were incubated with ER-MP58-biotin (own culture) and afterwards with streptavidin-allophycocyanin (Becton Dickinson). ER-MP58+ cells were sorted with FACSAria (Becton Dickinson). Subsequently, ER-MP58+ cells were cultured for 8 days on 0.5% gelatin-coated wells (96-well plate) in RPMI 1640 medium supplemented with 10% FCS, 50 μM β-mercaptoethanol and 50 ng/mL GM-CSF (MT Diagnostics, Etten-Leur, The Netherlands). Finally cells were harvested with 2 mM EDTA.
To monitor the proliferation capability, cells from the embryonic pancreas (pooled), liver, adult BM and blood were labeled with 5 μM carboxyfluorescein succinimidyl ester (CFSE) (Sigma Aldrich) and incubated for 10 min at 37°C. Cells were washed and incubated with ER-MP58-biotin and afterwards with streptavidin-allophycocyanin. ER-MP58+ cells were sorted with FACSAria and were cultured for 8 days with 50 ng/mL GM-CSF. Cells were harvested with 2 mM EDTA.
Cryostat sections (6 μm) of E15.5 pancreases from C57BL/6 and NOD/LTj mice were prepared and fixed with cold methanol and acetone. Slides were incubated with guinea pig-anti-insulin (DAKO, Glostrup, Denmark) and rat-anti-ER-MP58 followed by rabbit-anti-guinea pig-FITC (Abcam) and goat-anti-rat-TexasRed (Southern Biotechnology Associates, Birmingham, AL, USA). Finally, slides were mounted in Vectashield with DAPI (Vector Laboratories, Burlingame, CA, USA).
Cryostat sections (6 μm) of 5-wk-old pancreases from C57BL/6, NOR/LTj and NOD/LTj mice were prepared and fixed with cold methanol and acetone. Slides were incubated with Ki-67-FITC and rat-anti-ER-MP58 followed by goat-anti-rat-TexasRed. Finally, slides were mounted in Vectashield with DAPI.
Data were analyzed by Mann–Whitney U test for unpaired data. All analyses were carried out using SPSS software (SPSS, Chicago, IL, USA) and considered statistically significant if p<0.05.
The authors thank Pieter Leenen for his expert advice and the Juvenile Diabetes Research Foundation for supporting this study
Conflicts of interest: The authors declare no financial or commercial conflict of interest.