Unstimulated monocytes of at-risk/type 1 diabetic humans and macrophages of the NOD mouse have markedly elevated autocrine GM-CSF production and persistent STAT5 phosphorylation. We analyzed the relationship between GM-CSF production and persistent STAT5 phosphorylation in NOD macrophages using reciprocal congenic mouse strains containing either diabetes-susceptible NOD (B6.NODC11), or diabetes-resistant C57L (NOD.LC11) loci on chromosome 11. These intervals contain the gene for GM-CSF (Csf2; 53.8 Mb) and those for STAT3, STAT5A, and STAT5B (Stat3, Stat5a, and Stat5b; 100.4–100.6 Mb). High GM-CSF production and persistent STAT5 phosphorylation in unactivated NOD macrophages can be linked to a region (44.9–55.7 Mb) containing the Csf2 gene, but not the Stat3/5a/5b genes. This locus, provisionally called Idd4.3, is upstream of the previously described Idd4.1 and Idd4.2 loci. Idd4.3 encodes an abundance of cytokine genes that use STAT5 in their macrophage activation signaling and contributes ∼50% of the NOD.LC11 resistance to diabetes.
Defective monocyte-macrophage differentiation and activation contribute to immunopathogenesis of type 1 diabetes (T1D)4 in humans and NOD mice (1, 2, 3, 4, 5). Through their roles as APCs, cytokine factories, and phagocytic effector cells, autoimmune (AI) monocytes and macrophages have a great potential for promoting immunopathology. In addition, monocyte and macrophage production of prostanoids promotes a proinflammatory microenvironment within the immune system conducive to the loss of tolerance regulation by other APC, such as dendritic cells (1, 2, 6).
GM-CSF is a key regulatory cytokine in myeloid cell differentiation and in monocyte and macrophage activation. Like many cytokines involved in hemopoiesis, GM-CSF uses Jak2 activation of the signal transduction/transcriptional regulator proteins, STAT5A and STAT5B, to influence gene regulation (6, 7, 8, 9, 10, 11, 12, 13, 14). In immature myeloid cells and unactivated monocytes, GM-CSF can induce signaling through both full-length STAT5A (94–96K) and STAT5B (94–92k) isoforms as well as through post-translationally modified truncated isoforms (77K and 80K) (8, 9, 15, 16, 17). During cytokine-induced differentiation, truncated STAT5 isoforms can act as repressors of gene transcription in immature/unactivated cells, whereas full-length STAT5 isoforms induced in mature/activated cells act as gene transcription activators (7, 8, 15, 16, 17).
The genes encoding STAT5A, STAT5B, and GM-CSF (Csf2), are all found on chromosome 11 in mice, with Csf2 found nearer the centromeric end at 53.85 Mb, and Stat5a/Stat5b located more toward the telomeric end at 100.53, and 100.55 Mb, respectively (Fig. 1) (18, 19). These regions are contained within the previously mapped diabetes susceptibility Idd locus (Idd4). Idd4 was originally mapped to NOD chromosome 11 after an outcross to C57BL/10 and contained an interval starting near Csf2 and extending ∼52 Mb downstream toward the telomere. Using congenic strains derived from introgression of C57BL/6 alleles onto the NOD genetic background, this locus has been further resolved into at least two distinct regions: Idd4.2, which confers decreased diabetes susceptibility and delayed onset, and Idd4.1, which is associated with the initiation of insulitis (18, 19, 20, 21, 22).
We have shown that monocytes of humans with T1D and autoimmune thyroid diseases as well as NOD mouse monocytes and macrophages have three distinct, but possibly interrelated, phenotypes that can contribute to chronic inflammatory response: 1) aberrantly high expression of the key prostanoid synthesis enzyme, PGS2/cyclooxygenase 2 (COX2) (2, 3); 2) a marked elevation of autocrine GM-CSF production (23, 24); and 3) persistent STAT5 phosphorylation (24). In LPS-activated myeloid cells, GM-CSF is an activator of PGS2/COX2 expression, but GM-CSF overproduction alone does not account for the high, IL-10-resistant PGS2/COX2 expression seen in AI monocytes and NOD macrophages (23, 24). Although the endogenous GM-CSF production of AI monocytes and macrophages can stimulate high levels of STAT5 phosphorylation, the persistence of phosphorylated STAT5 and its subsequent effects on STAT5 DNA binding characteristics in these cells can become independent of GM-CSF and Jak activity after induction (24). However, the resistance of STAT5 phosphorylation to suppression by IL-10 and its potential effects on IL-10-resistant PGS2/COX2 expression in monocytes and macrophages require at least a brief exposure to GM-CSF (24).
These findings have led us to use NOD chromosome 11 congenic mouse strains to analyze the contribution of chromosome 11 components to the GM-CSF and STAT5 dysfunction phenotypes in autoimmune diabetes. Congenic strains of mice have been a very useful tool for the study of genetic factor contributions to multifactorial diseases, such as T1D and systemic lupus erythematosus (21, 22, 25). By breeding defined chromosomal intervals onto the background of either susceptible or resistant strains, we can analyze the relative contributions of genes within these intervals to autoimmune phenotypes and pathology.
In these studies we used two previously described complementary congenic mouse strains: the diabetes-resistant NOD.LC11 (derived from the NOD.LC11 strain originally designated NOD.DR3 (21)) that carries only a C57L chromosome 11 interval on a NOD genetic background and the peri-insulitis-prone B6.NOD.C11, which has a C57BL/6 genetic background with a NOD chromosome 11 interval (22). Both congenic strains encompass the entire Idd4 interval and Stat5a/Stat5b/Stat3 complex (21, 22). In addition, we analyzed recombinant subcongenic strains, NOD.LC11f and NOD.LC11b, derived from the NOD.LC11, which segregates the confirmed Idd4 intervals away from the Stat5a/Stat5b region (Fig. 1). From our analysis of GM-CSF and STAT5 phenotypes in macrophages from these congenic strains, we can exclude the downstream Stat5a/Stat5b-containing region and the upstream Il12b (encoding IL-12p40 subunit)-containing regions (26, 27) from contributing to these myeloid cell phenotypes. Our data suggest that a smaller upstream region (44.9–55.7 Mb) containing the Csf2 gene loci contributes to both the NOD GM-CSF and STAT5 phenotypes as well as a significant portion of the NOD.LC11 resistance to diabetes.
Materials and Methods
Congenic mouse strains
Fig. 1 gives a schematic representation of the regions on chromosome 11 contributed by NOD, C57BL/6, and C57L parental strains in the NOD.LC11 (21) and B6.NODC11 (22) congenic strains used in this study.
The NOD background congenic strains (NOD.LC11 and its subcongenic derivatives, NOD.LC11b and NOD.LC11f) were developed and maintained in the University of Virginia specific pathogen-free colony. NOD.LC11f mice carry a C57L interval covering the Csf2 gene for GM-CSF and at least part of the Idd4 interval, but excluding the Stat3/Stat5a/Stat5b gene region. NOD.LC11b C57L region excludes all the telomeric regions downstream of the Csf2 gene. These strains develop diabetes at a significantly decreased incidence from the parental NOD strain. The B6.NODC11 strain does not develop diabetes, but does develop peri-insulitis (22). All the NOD.LC11 strain intervals exclude Il12b, the gene encoding IL-12p40 protein (26, 27). NOD, C57BL/6, and C57L mice were purchased from The Jackson Laboratory, and the in-house bred B6.NODC11 congenic mice were housed at the University of Florida specific pathogen-free colony. Both facilities are American Association for Assessment of Laboratory Animal Care inspected, and animals were handled in accordance with institutional animal care and use committee-approved protocols at their respective institutions. Diabetes incidence in the NOD.LC11 strains was measured as previously described (21).
Peritoneal macrophage collection and culture for GM-CSF production analysis
Peritoneal macrophages were collected from 8- to 10-wk-old NOD, C57BL/6, and congenic strain female mice by peritoneal lavage in accordance with University of Florida institutional animal care and use committee-approved protocols. Macrophages were pooled from one to six animals per strain for each experiment. Peritoneal macrophages were analyzed for STAT5 phosphorylation as freshly isolated cells, then adherence purified and cultured overnight in RPMI 1640 medium supplemented with 10% FBS and 1% penicillin-streptomycin-amphiterin mix (Mediatech-Cellgro) at 37°C in 5% CO2. Cell-free culture supernatants were analyzed by ELISA (BD Pharmingen) or Luminex (Upstate Biotechnology) for GM-CSF production.
Flow cytometric analysis for STAT5 phosphorylation
Intracellular flow cytometric analysis for STAT5 and tyrosine-phosphorylated STAT5 was performed as previously described (4, 5, 22), with the exception that an FITC-conjugated anti-tyrosine-phosphorylated STAT5 (anti-STAT5-Ptyr; Upstate Biotechnology) was used in place of the PE conjugate in some experiments. BD Biosciences FACSCalibur or FACSort flow cytometers and CellQuest software were used to determine stained cell percentages.
Deconvolution microscopy for STAT5 phosphorylation
After flow cytometric analysis, anti-STAT5Ptyr-labeled cells were analyzed by deconvolution microscopy. Analysis of immunofluorescence was performed as previously described (22) using a Delta Vision deconvolution microscopy system with an Olympus OMT inverted microscope (Applied Precision). Images were processed with 15 reiterations of the deconvolution algorithm and presented as three-dimensional reconstructions of 10–36 optical slices of 0.2-μm thicknesses.
Diabetes incidence of the NOD.LC11 mice and its recombinants were compared with that of NOD.LtJ controls using a two-tailed Fisher exact test. Phosphorylated STAT5 flow cytometric data and GM-CSF concentration data generated by ELISA and Luminex analysis were analyzed by one-way ANOVA. Pairwise comparisons of these data were performed using Student’s t test or Mann-Whitney U test analyses. The results of these analyses are listed with appropriate p values and statistical parameters (n, mean, and SEM) on figures, in figure legends, and/or in Results.
Chromosome 11 contributes to NOD diabetes susceptibility
NOD.LC11 mice are highly resistant to diabetes (21), although B6.NODC11 mice develop peri-insulitis (22), confirming that chromosome 11 loci are involved in NOD diabetes susceptibility. Both the NOD.LC11f and the NOD.LCllb congenic intervals carry C57L DNA, which encodes a cluster of cytokine genes, including Csf2 (GM-CSF) that are upstream of the previously defined Idd4.1/4.2 loci. In addition, the NOD.LCllf strain carries C57L DNA that extends from the cytokine cluster down to overlap the Idd4.1 and Idd4.2 intervals and has a diabetes incidence of 19% (Fig. 1 and Table I). In NOD.LC11b mice, whose C57L interval contains this cluster, but excludes the Idd4.1/Idd4.2 region, the cumulative incidence of diabetes by 30 wk of age was 27%, suggesting that the Csf2 gene region represents a previously undefined diabetes susceptibility locus (Fig. 1 and Table I). We provisionally designate the region defined by the NOD.LC11b mouse strain as Idd4.3. These data strongly suggest that the major protective effects of chromosome 11 C57L alleles are determined by Idd4.3, with relatively little additional protection afforded by C57L alleles at Idd4.1 and Idd4.2. Furthermore, the contrast with the near-complete protection of the NOD.LC11 parental strain indicates that additional loci downstream of Idd4.1–4.3 are also involved in regulating disease penetrance.
|Strain .||NOD .||NOD.LC11 .||NOD.LC11f .||NOD.LC11b .|
|Age of onseta||146||155||172||170|
|Strain .||NOD .||NOD.LC11 .||NOD.LC11f .||NOD.LC11b .|
|Age of onseta||146||155||172||170|
Average days of age at onset and (age range in days).
Compared to NOD diabetes incidence in two-tailed Fisher exact test.
Chromosome 11 and GM-CSF production
We previously reported high GM-CSF production in NOD macrophages under conditions where C57BL/6 control macrophages have moderate to low GM-CSF production (22). To assess the genetic contribution of chromosome 11 loci to these phenotypes, we first examined peritoneal macrophages of two complementary strains of congenic mice, B6.NODC11 and NOD.LC11. In NOD.LC11 mice, macrophage GM-CSF production was reversed to levels at or below that of C57L and C57BL/6 controls, whereas NOD or higher levels of GM-CSF production were seen in B6.NODC11 macrophages (Fig. 2). These results demonstrate that the GM-CSF overproduction phenotype of the NOD is controlled by chromosome 11. GM-CSF production by macrophages of the NOD.LC11b and NOD.LC11f subcongenic strain mice was significantly lower than that of NOD macrophages and equal to or less than that of NOD.LC11, C57BL/6 and C57L macrophages. Therefore, the shared Idd4.3 C57L interval in these subcongenic strains contributes to the NOD GM-CSF phenotype.
Chromosome 11 and persistent STAT5 phosphorylation in NOD macrophages
Similarly, we examined uninduced STAT5 phosphorylation in the C57BL/6, NOD, NOD.LC11, NOD.LC11b, NOD.LC11f, and B6.NODC11 strains. We found that, as seen with the GM-CSF production phenotype, NOD.LC11 macrophages as well as those of its subcongenic strains had STAT5 phosphorylation levels within the range seen in C57BL/6 control macrophages. However, NOD, C57BL/6, NOD.LC11, and NOD.LC11f macrophage STAT5 phosphorylation levels were all significantly higher than those seen in C57L. Only the STAT5 phosphorylation levels of NOD.LC11b strain macrophages were not significantly higher than that of the C57L control and were significantly lower than that of the NOD. In contrast, B6.NODC11 macrophages retain the NOD high STAT5 phosphorylation phenotype (Fig. 3). These data suggest that the Idd4.3 region on the chromosome 11 is associated with high STAT5 phosphorylation seen in NOD macrophages and excludes contributions to this phenotype by the Idd4.1/Idd4.2 region. Furthermore, these data exclude the more telomeric region that contains the Stat5a/Stat5b genes and the centromeric region containing Il12b as regulators of the STAT5 phenotype.
Role of chromosome 11 regions in the etiology of STAT5 dysfunction in NOD macrophages
Davoodi-Semiromi et al. (28) recently reported a single base pair polymorphism unique to the NOD mouse within the DNA binding domain of Stat5b DNA cloned from IL-2 stimulated whole spleen. Although we cannot rule out the potential for the Stat5a/Stat5b region to enhance or augment the effects of the Idd4.3 region in other cell types, our analysis of NOD chromosome 11 congenic mice strongly suggests that the Stat5a/Stat5b/Stat3-containing region of this chromosome is not involved in either GM-CSF or STAT5 phenotypes of NOD macrophages.
In addition, the overlap of the B6.NODC11 interval and that of the NOD.LC11b suggest that the proposed quantitative Idd4 contribution of the chromosome 11 region encoding the gene for Il12b (26, 27) is not contributing to these myeloid cell phenotypes or the diabetes susceptibility imparted by the Idd4.3 locus. Additional sequence mapping and congenic strain development are underway to confirm these predictions.
Relating STAT5 dysfunction with GM-CSF in autoimmune monocytes and macrophages
In our NOD and parallel human monocyte studies to date, we have sought to find a link between immunopathology and STAT5/GM-CSF autoimmune phenotypes. These findings in NOD congenic mice indicate that components of chromosome 11 may link GM-CSF overproduction with persistent STAT5 phosphorylation. These findings coupled with the GM-CSF-dependent STAT5 IL-10 resistance and isoform-specific dysfunctions we previously noted in AI monocytes and macrophages (22) suggest that dysregulation of differentiation- and activation-specific gene expression by GM-CSF and other STAT5-inducing cytokines is associated with chromosome 11 encoded components in autoimmune myeloid cells. The coincidence of these complex phenotypes with the diabetes susceptibility conveyed by region Idd4.3 suggests the potential for a connection between their dysfunction and diabetes incidence in the NOD mouse. We are continuing to narrow the congenic interval on chromosome 11 to elucidate the contribution of myeloid cell cytokine signaling dysfunction to diabetes.
We thank the members of the University of Florida Diabetes Research Group and laboratory personnel of Dr. M. McDuffie (University of Virginia) for their support of this project. We gratefully acknowledge the help of Drs. A. Davoodi-Semiromi, J. M. Crawford, and C. Ketcham for their editorial review and intellectual support.
The authors have no financial conflict of interest.
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.
This work was supported by National Institutes of Health Grants KO1DK02947 and R21DK60802 (from the National Institute of Diabetes and Digestive and Kidney Diseases; to S.A.L.) and AIPO1AI42288 (to M.C.S.).
Abbreviations used in this paper: T1D, type 1 diabetes; AI, autoimmune; COX2, cyclooxygenase 2; Idd4, insulin-dependent diabetes region 4; Ptyr, phosphotyrosine.