The Inhibitory G Protein α-Subunit, Gα_z, Promotes Type 1 Diabetes-Like Pathophysiology in NOD Mice

Abstract

The α-subunit of the heterotrimeric G_z protein, Gα_z, promotes β-cell death and inhibits β-cell replication when pancreatic islets are challenged by stressors. Thus, we hypothesized that loss of Gα_z protein would preserve functional β-cell mass in the nonobese diabetic (NOD) model, protecting from overt diabetes. We saw that protection from diabetes was robust and durable up to 35 weeks of age in Gα_z knockout mice. By 17 weeks of age, Gα_z-null NOD mice had significantly higher diabetes-free survival than wild-type littermates. Islets from these mice had reduced markers of proinflammatory immune cell infiltration on both the histological and transcript levels and secreted more insulin in response to glucose. Further analyses of pancreas sections revealed significantly fewer terminal deoxynucleotidyltransferase-mediated dUTP nick end labeling (TUNEL)-positive β-cells in Gα_z-null islets despite similar immune infiltration in control mice. Islets from Gα_z-null mice also exhibited a higher percentage of Ki-67–positive β-cells, a measure of proliferation, even in the presence of immune infiltration. Finally, β-cell–specific Gα_z-null mice phenocopy whole-body Gα_z-null mice in their protection from developing hyperglycemia after streptozotocin administration, supporting a β-cell–centric role for Gα_z in diabetes pathophysiology. We propose that Gα_z plays a key role in β-cell signaling that becomes dysfunctional in the type 1 diabetes setting, accelerating the death of β-cells, which promotes further accumulation of immune cells in the pancreatic islets, and inhibiting a restorative proliferative response.

β-cell Gα_z contributes to β-cell death and dysfunction and blocks replicative capacity in the immune-mediated nonobese diabetic (NOD) mouse model, promoting T1DM-like pathophysiology.

Type 1 diabetes mellitus (T1DM) is a result of insulin deficiency arising from immune-mediated β-cell destruction. In the US population alone, it is expected that the prevalence will increase more than 144% by 2050, which means that T1DM will be diagnosed in more than 500,000 individuals in the next 35 years (1, 2).

T1DM pathogenesis targets the β-cells in the islets of Langerhans, which are the sole producers of insulin and are necessary to maintain normoglycemia. Although the precise series of events that result in T1DM onset has not been fully elucidated, factors such as genetic susceptibility due to single nucleotide polymorphisms in the various major histocompatibility complex (MHC) loci, viral infection, endoplasmic reticulum stress, and neonatal β-cell apoptosis have all been reported to trigger T1DM development (3–7). Apoptosis of β-cells due to one or more of these initiating events can result in the release of cell debris and can facilitate immune cell exposure to β-cell antigens. Islet inflammation results in both nonspecific and targeted β-cell death, including macrophage secretion of cytokines that results in activation of nuclear factor κB (NF-κB)–mediated apoptotic pathways and granzyme- and perforin-mediated β-cell death facilitated by antigen-specific cluster of differentiation 8 (CD8)–positive T-cells (8). T1DM presents only after β-cell mass drops below the threshold necessary to maintain euglycemia (9).

We previously showed that activation of the inhibitory G protein–coupled E prostanoid receptor EP₃ is highly induced in diabetic islets and plays a negative role in glucose- and hormone-stimulated insulin secretion (10, 11). We also showed that in pancreatic β-cells, EP₃ is specifically coupled to the inhibitory G protein G_z, whose catalytic α-subunit, when activated, inhibits adenylate cyclase, reducing cyclic adenosine monophosphate production and blunting glucose-stimulated insulin secretion (GSIS) (10, 12–14). Nevertheless, cyclic adenosine monophosphate also has known effects on β-cell replication and survival. We showed that mice lacking Gα_z are protected from developing glucose intolerance after high-fat diet feeding (15) and are resistant to multiple low-dose streptozotocin (MLD-STZ)–induced diabetes when concurrently treated with a low dose of the GLP-1 mimetic exendin-4 (Ex4) (16). Protection from high-fat diet–induced glucose intolerance appeared to be due primarily to increased β-cell replication, resulting in augmented functional β-cell mass, whereas protection from MLD-STZ resulted from both increased β-cell replication and decreased β-cell apoptosis, leading to preserved functional β-cell mass (15, 16). In addition, active Gα_z was able to augment β-cell death as stimulated by the proinflammatory cytokine interleukin 1β (IL-1β) (16), suggesting that it plays a critical role in the pathophysiology of immune-mediated β-cell destruction.

The nonobese diabetic (NOD) mouse is a well-accepted T1DM model that mimics the insulin-dependent diabetes mellitus seen in humans (17, 18). Although both male and female NOD mice present with insulitis, 70% to 80% of female NOD mice typically become hyperglycemic between 12 and 16 weeks of age as an initial wave of immune infiltration stimulates β-cell death, ultimately decreasing functional β-cell mass (19). Because our previous data supported the finding that loss of Gα_z can affect β-cell proliferation, function, and demise in the context of β-cell stressors, we hypothesized that genetic deficiency of Gα_z could also alter T1DM progression. Thus, we investigated the consequences of the loss of Gα_z in NOD mice.

Research Design and Methods

Mouse breeding and phenotyping

C57BL/6N mice containing a genomic insertion of a pGKneoR cassette at 160 base pairs downstream of the translation start site of the Gα_z gene (gene symbol: Gnaz) have been described and characterized previously (15). These mice were backcrossed onto the NOD background for at least 10 generations and were single nucleotide polymorphism–genotyped by Jackson Laboratories to ensure that the background was 100% NOD, save for the genomic regions surrounding the cassette inserted into the Gnaz gene. Breeding colonies were housed in a limited-access, pathogen-free facility where all cages, enrichment, and water were sterilized before use (University of Wisconsin-Madison, Biotron) on a 12-hour light/12-hour dark cycle with ad libitum access to water and irradiated breeder chow (Teklad 2919). Gα_z-null and wild-type (WT) NOD mice were generated by heterozygous matings to produce littermate pairs.

Mice with LoxP sites flanking the first protein-coding exon (exon 3) of the Gnaz gene (Gα_z-flox/flox) were generated from embryonic stem cells obtained from the European Conditional Mouse Mutagenesis Program (Gnaz^{tm1a(EUCOMM)Hmgu}). These embryonic stem cells contain the Gnaz gene interrupted by a β-galactosidase/neomycin resistance gene-trap cassette inserted between exons 2 and 3. Karyotyping was performed to ensure >80% euploidy of the cells before injection into C57 Albino embryos. Male progeny displaying high percentages of coat/eye chimerism were crossed back to female C57 Albino mice; progeny carrying a founding F1 germline mutation were identified by coat/eye color and confirmed by genotyping. F1 germline mice were bred with Flp “deleter” mice (Jackson Laboratories) to excise the β-galactosidase/neomycin resistance gene-trap cassette, leaving two LoxP sites surrounding exon 3 of Gnaz that do not affect protein expression. These mice were fully backcrossed into the C57BL/6J background.

To generate the β-cell–specific Gα_z knockout (KO) and control mice, we bred Gα_z-flox/flox mice with RIP-Cre^Herr mice [“InsPr-Cre” in (20)], also fully backcrossed into the C57BL/6J background. This Cre line has low hypothalamic recombination and does not express the human growth hormone mini-gene (21, 22). Experimental animals were identified by genotyping for the WT of floxed Gnaz gene and for the presence of the Cre gene. WT Cre⁺ mice were used as controls.

Upon weaning, mice were housed five or fewer per cage of mixed genotypes with ad libitum access to lower-fat chow in one of two facilities: the Madison VA Animal Resource Facility (LabDiet 5001, nonirradiated; 17-week study) or investigator-accessible housing in the Biotron [(Teklad 2920X, irradiated) 35-week study, T-cell analysis, and female oral glucose tolerance test (OGTT)]. All procedures were performed according to approved protocols in accordance with the principles and guidelines established by the University of Wisconsin and Madison VA institutional animal care and use committees.

Blood glucose measurements were taken weekly from 4 weeks of age up to 35 weeks of age using a blood glucose meter (AlphaTRAK) and rat/mouse–specific test strips. At the end of the study, mice were subjected to collagenase perfusion of the pancreas to isolate pancreatic islets for in vitro analysis or pancreatic dissection for either cryofixation or paraffin embedding. A separate cohort of mice from the 17-week study was euthanized at the 4- to 5-week time point, and pancreata were collected for sectioning.

MLD-STZ induction of diabetes

MLD-STZ (Sigma; #S01230) [50 mg/kg of body weight (BW)] induction of hyperglycemia and treatment with 10 μg/kg BW Ex4 (Sigma-Aldrich; #E7144) was conducted as previously described (16).

Mouse islet isolation and GSIS assay

Mice were euthanized using 2,2,2-tribromoethanol (Sigma; #T48402) anesthesia followed by cervical dislocation. Intact pancreatic islets were isolated from mice using a collagenase digestion protocol (23). On the day of isolation, islets were picked into 100 µL of islet medium (RPMI 1640; Gibco; #11879020) containing 11.1 mmol/L glucose (Fisher Scientific; #D16), 10% heat-inactivated fetal bovine serum (Sigma-Aldrich; #12306C), and 1% Hepes (Sigma; #H4034) and penicillin/streptomycin (Gibco; #15070-063)) in each well of a 96-well V-bottom tissue culture-coated plate (Corning Life Sciences; #3894) according to a protocol optimized in our laboratory (24). Insulin enzyme-linked immunosorbent assays were performed essentially as described elsewhere (11).

Glucose tolerance testing

Animals were fasted for 4 to 6 hours before glucose tolerance testing. Glucose was given using an oral gavage at a dose of 2 g/kg BW. Blood glucose measurements were taken immediately before gavage and at 5, 15, 30, 60, and 120 minutes following gavage. For female mice, blood was collected using a lateral tail nick at baseline and 5 minutes after gavage. Oral glucose tolerance testing was performed at 16 and 24 weeks of age for male and female mice, respectively.

Whole pancreas staining

For all slide staining assays, 10-μm serial sections were cut on positively charged slides, with 18 sections per stop position (three per slide) and three stop positions per pancreas separated by at least 200 μm. Immune infiltration of the β-cell was determined using standard eosin (Polysciences; #17269) and hematoxylin (Sigma; #GHS280) staining. Following staining, quantification of immune infiltration was accomplished using a numerical scoring system, ranking the extent of immune infiltration from no immune cells present (0) to an islet that was completely infiltrated with immune cells (4) (Supplemental Fig. 1 ^{(27.7MB, docx)} ). A single observer who was blind to genotype identifiers analyzed all tissue sections. For islet infiltration scoring, every islet in a single, 10-μm section was scored per mouse. Staining and analysis techniques used for determining β-cell fractional area and β-cell death (as shown by TUNEL positivity) were performed as previously described (16).

The immunohistochemical staining protocol described previously was modified for optimization of Gα_z analysis (16). The primary modification to the immunostaining protocol was use of the rabbit anti-Gα_z primary antibody (Abcam; #AB150434), diluted 1:400 in antibody diluent with background-reducing components (Dako; #S3022). Every islet in a single 10-μm section was scored, and three distinct sections were averaged per mouse to give one biological replicate.

Quantitative real-time polymerase chain reaction

Islets from WT and Gα_z-KO mice were isolated and collected for isolation of whole RNA using the Qiagen RNeasy^® Kit (Qiagen; #74106) according to the manufacturer’s instructions. Complementary DNA (cDNA) was then generated with random hexamers (High-Capacity cDNA Reverse Transcription Kit, Applied Biosystems^®; #4368813), and quantitative reverse transcription polymerase chain reaction (qRT-PCR) was performed using SYBR green (Roche; #04913914001) according to previously published methods (25). Quantification of messenger RNA (mRNA) expression of IL-1β (gene symbol: IL1β), interferon γ (IFNγ), tumor necrosis factor α (TNFα), nuclear factor kappa-light-chain-enhance of activated β-cell (NFκB), C-X-C motif chemokine 10 (CXCL10), chemokine (C-C motif) ligand 2 and 4 (CCL2, CCL4), and cluster of differentiation 3 and 8 (CD3, CD8) was performed by normalization of cycle times within each sample to those of β-actin. Primer sequences can be found in Table 1.

Table 1.

qRT-PCR Primer Sequences and Properties

Gene (M. Mus.)	Orientation	Sequence (5′ → 3′)	Tm	% GC
NF-κB	F	TCCTGATCCAGACAAAAACTGG	54.87	45.45
	R	ATCCCGGAGTTCATCTCATAGT	54.97	45.45
IL-1β	F	CCACCTTTTGACAGTGATGAGA	54.81	45.45
	R	GAGATTTGAAGCTGGATGCTCT	54.98	45.45
IFNγ	F	TGAAAGACAATCAGGCCATCAG	62.18	45.45
	R	GTGGGTTGTTGACCTCAAACT	62.36	47.62
TNFα	F	CTCACACTCACAAACCACCAAG	63.15	50
	R	GTAGACAAGGTACAACCCATCG	62.03	50
CXCL10	F	GGCCATAGGGAAGCTTGAAAT	62.21	47.62
	R	CATCGTGGCAATGATCTCAACA	62.63	45.45
CCL2	F	TCATTCACCAGCAAGATGATCC	61.99	45.45
	R	GCTTGGTGACAAAAACTACAGC	61.82	45.45
CCL4	F	GCTGTGGTATTCCTGACCAAAA	62.38	45.45
	R	CAGTTCAACTCCAAGTCACTCA	61.66	45.45
CD3	F	CCAGTCAAGAGCTTCAGACAAG	62.46	50
	R	TCAGTTGGTTTCCTTGGAGATG	61.85	45.45
CD8	F	CCTCAACCTGCTTTGTCTACAG	62.50	50
	R	GCAAGAGTGGCTGAATGTAGT	61.95	47.62
Gαz	F	AAGGAACATCAGGGCCAGAG	56.40	55
	R	CAGCACTACGGTTCAGCAAC	55.38	50

Abbreviations: F, forward; %GC, percent guanosine/cytosine content; R, reverse; T_m, melting temperature.

Intracellular cytokine staining

Splenocytes were harvested from 17- to 19-week-old female NOD or Gα_z-KO mice and cultured for 3 days with 1 μg/mL plate-bound anti-CD3ε (clone 145-2C11) and 1 μg/mL soluble anti-CD28 (clone 37.51) (BD Biosciences; #553057 and #553294) in complete RPMI media with 10% fetal bovine serum. Suspensions of single cells were incubated with GolgiStop (BD Biosciences; 51-2092KZ) for 4 hours before staining. Cells were stained for surface markers, fixed and permeabilized with Cytofix/Cytoperm Plus reagents (BD Biosciences; #51-2090KZ and #51-2091KZ), stained for intracellular cytokines, and analyzed with the BD LSR II flow cytometer. Fluorescent antibodies for CD3ε (clone 145-2C11), CD8α (clone 53-6.7), TNFα (clone MP6-XT22), and IFNγ (clone XMG1.2) were purchased from BD Biosciences (#563565, #561092, #561041, and #563376, respectively). The fluorescent antibody for CD4 (clone RM4-5) was purchased from eBioscience (#11004282). Ghost Dye viability reagent was purchased from Tonbo Biosciences (#130865). Flow cytometry data were analyzed with FlowJo software.

Statistical analysis

Data are expressed as mean ± standard error of the mean unless otherwise noted. Data were compared by one- or two-way analysis of variance or Student t test as appropriate and as described in the figure legends. A P value <0.05 was considered statistically significant. Statistical analyses were performed with GraphPad Prism version 6 (GraphPad Software, San Diego, CA).

Results

Diabetes protection afforded by loss of Gα_z is robust and durable

The robustness of the NOD phenotype has been reported as variable and dependent on environmental conditions (17). To determine the natural history of diabetes development in our NOD colony as well as any protection afforded by loss of Gα_z, we tracked blood glucose levels of female mice weekly from 4 to 35 weeks of age. Two waves of diabetes development were apparent in the WT mice [Fig. 1(a)], with ∼45% of the mice becoming diabetic by 23 weeks of age and then a lag of 7 weeks before an additional 25% of mice became diabetic [Fig. 1(b)]. None of the 11 Gα_z-KO females became hyperglycemic. Analysis of the Kaplan-Meier survival data revealed a strong protection from diabetes development, with WT females being 9.644 times more likely to develop diabetes than the Gα_z-KO females [Fig. 1(b)]. Body weights did not differ between the genotypes until ∼25 weeks of age, when approximately half of the WT mice were diabetic, consistent with a wasting phenotype [Fig. 1(c)]. Therefore, we concluded that the diabetes protection phenotype in the Gα_z-KO females on NOD background was robust and durable, at least through 8 to 9 months of age.

Figure 1.

The diabetes protection phenotype of female (♀) Gα_z KO NOD mice is robust and durable. (a) Timeline of diabetes development in individual Gα_z-KO or WT NOD female mice. (b) Kaplan-Meier survival curve and (c) body weights of the animals shown in (a). (d) Mean blood glucose levels and (e) diabetes-free survival of female Gα_z-KO NOD mice and WT NOD controls housed in a facility separate from those shown in (a). (f) The 4- to 6-hour fasting plasma glucose levels of Gα_z-KO NOD and WT NOD female mice. (g) Plasma blood glucose values, (h) area under the curve, and (i) plasma insulin levels during OGTTs of 24-week-old Gα_z-KO NOD and WT NOD female mice. (j) Change in plasma insulin levels at 5 minutes shown as fold-change from baseline. (d, g, i) Data were compared by two-way analysis of variance with the Tukey post hoc test. **P < 0.01; ****P < 0.0001. (d) n = 20 or 21 per group and (g, i) n = 6 per group. (f, j) Data were compared by Student t test; n = 20 or 21 per group and n = 6 per group, respectively. All error bars represent standard error of the mean. NS, not significant; OR, odds ratio.

To determine the molecular and cellular signaling events mediating the early development of diabetes, we tracked blood glucose levels from 11 to 17 weeks of age in a second cohort of female Gα_z-KO and WT NOD mice. Gα_z-KO NOD mice had significantly lower random-fed blood glucose levels than WT controls throughout the 17 weeks of the study, with a divergence between the genotypes beginning around week 13 [Fig. 1(d)]; this is within the expected timeline for initial diabetes development in female NOD mice (17, 18). When expressed as diabetes-free survival (blood glucose level <250 mg/dL), Gα_z-KO female mice had 90% disease-free survival over the 17-week period, which was nearly double that observed in WT NOD females (50%) [Fig. 1(e)]. At week 17, the 4- to 6-hour fasting plasma glucose levels were significantly lower in Gα_z-KO females than in WT animals, who overall had fasting hyperglycemia [Fig. 1(f)].

Next, WT and Gα_z-KO NOD females were given an oral glucose challenge. The mean blood glucose levels of Gα_z-KO NOD females was lower at each time point during the glucose challenge and was statistically lower at the 30-minute time point [Fig.1(g)]. The area under the curve was also lower in Gα_z-KO NOD mice, although this was not statistically significant [Fig. 1(h)]. The mean levels of plasma insulin were also higher in the Gα_z-KO NOD mice 5 minutes after the glucose challenge, whether represented as ng/mL or as the fold-change vs the level at time = 0, although these differences were also not statistically significant [Fig. 1(i) and 1(j), respectively].

Loss of Gα_z protects from diabetes by affecting multiple molecular and cellular events

Following our conclusion that loss of Gα_z led to increased protection from diabetes and improved glucose tolerance in the NOD mice, we evaluated whether this protection resulted from a decrease in immune infiltration and maintenance of functional β-cell mass. The extent of insulitis was analyzed using the scoring system described in the Methods (see Supplemental Fig. 1 ^{(27.7MB, docx)} for scoring examples). Islets from female Gα_z-KO NOD mice exhibited a reduced inflammatory profile compared with those of WT controls [Fig. 2(a)]. When shown as a mean insulitis score, islets from Gα_z-KO mice exhibited a 30% reduction in overall mean insulitis compared with those of WT females [Fig. 2(b)]. This reduction in islet immune infiltration correlated with a fourfold increase in β-cell fractional area as determined by immunohistochemistry [Fig. 2(c)]. Representative images of insulin and hematoxylin-stained pancreas sections from female WT NOD and Gα_z-KO NOD mice are included for visualization of quantified differences [Fig. 2(d) and 2(e), respectively].

Figure 2.

Female (♀) Gα_z-KO NOD mice have decreased islet insulitis and increased insulin-positive pancreas area at 17 weeks of age. Pancreata were collected from female WT and Gα_z-KO mice at 17 weeks of age and subjected to hematoxylin and eosin staining to determine islet inflammation or insulin immunohistochemistry with hematoxylin counterstain to determine β-cell fractional area. (a) Percentage of islets in each of the islet inflammation scoring categories. (b) Mean islet immune infiltration score for each biological replicate. (c) Quantification of β-cell fractional area calculated from all islets counted on three sections of each mouse pancreas separated by at least 200 μm. Representative pancreas sections from WT and Gα_z-KO mice are shown in panels (d) and (e), respectively. (f) Quantification of insulin⁺ and Ki67⁺ cells in 17-week-old female mice calculated by counting all islets in a pancreas section and averaging two sections separated by at least 200 μm per mouse. Representative pancreas sections from WT and Gα_z-KO mice are shown in panels (g) and (h), respectively. White arrows indicate insulin⁺ Ki67⁺ cells. In all images, the scale bar represents 200 μm. (b) Data were compared by Student t test; n = 21 per group. (c) n = 7 per group. (f) n = 6 per group.

To further investigate the increased β-cell area seen in Gα_z-KO NOD females, pancreas sections were immunofluorescently labeled to identify nuclear Ki-67, a marker of actively replicating cells. Gα_z-KO NOD female islets exhibited a dramatic increase in insulin⁺, Ki-67⁺ cells compared with those of WT NOD females, suggesting that increased β-cell replication was in part responsible for enlarged β-cell area [Fig. 2(f–h)].

Finally, we repeated the 17-week observational study with male NOD mice, for which the prevalence of outright hyperglycemia was lower, to determine whether they exhibited any protective phenotype with regard to insulitis and/or glucose tolerance. We found male Gα_z-KO mice had slightly lower mean random-fed blood glucose levels than WT controls at weeks 16 and 17, but this was not statistically significant (Supplemental Fig. 2A) ^{(27.7MB, docx)} . None of the male Gα_z-KO mice became hyperglycemic by 17 weeks of age, whereas two of 23 male WT mice (8.6%) became diabetic (Supplemental Fig. 2B) ^{(27.7MB, docx)} . Interestingly, though, insulitis scores mimicked those observed in female mice, with the Gα_z-KO animals being relatively more protected from severe insulitis, as plotted by percentage of islets in each scoring category (Supplemental Fig. 2C) ^{(27.7MB, docx)} or mean insulitis score for each animal (Supplemental Fig. 2D) ^{(27.7MB, docx)} . Male Gα_z-KO mice trended toward lower peak blood glucose levels during OGTTs, a lower glucose area under the curve, and a higher fasting blood glucose level, although none of these measurements were statistically different from those of WT mice (Supplemental Fig. 2E–G) ^{(27.7MB, docx)} . Next, to directly measure β-cell function, we harvested islets from three nondiabetic mice of each genotype, matched as closely as possible for fasting blood glucose levels, and performed in vitro GSIS assays. In response to 16.7 mM of stimulatory glucose, islets from Gα_z-KO mice secreted >45% more insulin than WT islets, shown as a percentage of total islet insulin (Supplemental Fig. 2H) ^{(27.7MB, docx)} . Interestingly, however, the insulin content of Gα_z-KO islets was also significantly increased (by about 45%) (Supplemental Fig. 2I) ^{(27.7MB, docx)} .

Loss of Gα_z improves NOD islet inflammatory profile

Islets were isolated from both male and female WT and Gα_z-KO mice at 17 weeks of age. Following isolation, cDNA samples were prepared, and we investigated the mRNA expression of various immune cell-surface markers and a panel of chemokines and cytokines. We found that the transcripts for CD3 (a marker of T-cell lineage) and CD8 (exclusive to cytotoxic T-cells) were significantly reduced in both Gα_z-KO males and females compared with respective WT controls [Fig. 3(a)]. Interestingly, a significant decrease was also observed in both CD3 and CD8 transcripts when Gα_z-KO NOD male islets were compared with Gα_z-KO NOD female islets. We investigated the levels of various chemoattractant molecules, including CCL2, also known as monocyte chemoattractant protein 1; CCL4, also known as macrophage inflammatory protein 1 β; and CXCL10, also known as interferon γ–induced protein 10. We found that CCL2 transcript was increased only in islets from Gα_z-KO NOD males [Fig. 3(b)]. CCL4 transcript levels were lower in all groups compared with those in WT NOD females [Fig. 3(b)]. Interestingly, although CXCL10 was lower in Gα_z-KO NOD female islets, it was increased in both WT and Gα_z-KO NOD male islets [Fig. 3(b)]. Similar to the expression pattern seen in CD3 and CD8, we found that IFNγ was significantly lower in Gα_z-KO islets than in WT NOD islets, relative to sex [Fig. 3(c)]. IL-1β, a key inflammatory player, and NF-κB, a proapoptotic factor, were both decreased only at the transcript level in male NOD islets compared with female islets, with respect to genotype [Fig. 3(c)]. Islet transcript levels of TNFα were higher in male NOD islets than in those of NOD females, relative to genotype [Fig. 3(c)].

Figure 3.

Transcripts of important inflammatory players are different between islets of WT and Gα_z-KO animals, regardless of sex. T-cell markers Islets were isolated from both male and female Gα_z-KO and WT NOD animals and subjected to mRNA expression analysis for (a) T-cell markers, (b) chemoattractant molecules, and (c) cytokines known to be important in T1DM pathology. Data were compared by two-way analysis of variance with Tukey post hoc test, with the exception of CD8, which was compared using Student t test; N = 3 to 5 per group. (a) P < 0.05 vs WT NOD, respective to sex; (b) P < 0.05 vs female, respective to genotype. Ct, cycle threshold; ND, not detectable.

Loss of Gα_z protects NOD β-cells from apoptosis upon initial immune infiltration

To gain more insight into the mechanisms by which loss of Gα_z protects female NOD mice from becoming hyperglycemic, pancreata from 4-week-old female mice were collected and analyzed using our established insulitis scoring system. Although there was a small but statistically significant reduction in the mean immune infiltration score [Fig. 4(a)], both WT and Gα_z-KO female mice had a majority of islets in grades 2 to 4, indicating partial to complete islet immune infiltration [Fig. 4(b)]. These results suggest Gα_z does not play a role in initial immune cell infiltration in the NOD model. Next, the percentage of TUNEL⁺ β-cells, a marker for β-cell apoptosis, was quantified on pancreas sections from 4-week-old female WT and Gα_z-KO mice costained with insulin and 4′,6-diamidino-2-phenylindole (DAPI). Gα_z-KO pancreas sections showed a 30% reduction in the percentage of TUNEL⁺, insulin⁺ cells, indicative of a reduction in β-cell apoptosis [Fig. 4(c)]. As expected, TUNEL⁺ β-cells tended to surround the invading immune infiltrate, which can be observed as clusters of small DAPI-positive cells in representative sections [Fig. 4(d) and 4(e)].

Figure 4.

Loss of Gα_z did not dramatically alter insulitis by 4 weeks of age, but it did reduce β-cell death. Pancreata were collected from female (♀) WT and Gα_z-KO mice at 4 weeks of age and subjected to hematoxylin and eosin staining. Insulitis was quantified using the previously described scoring system. (a) Percentage of islets in each of the scoring categories and (b) mean islet immune infiltration score for each biological replicate. (c–e) Additional pancreata were collected at 4 weeks of age and subjected to immunofluorescence analysis for insulin staining (red) and TUNEL assay (green), with a DAPI nuclear counterstain (blue). (c) Quantification of TUNEL⁺ β-cell nuclei calculated from all islets counted on three sections of each mouse pancreas separated by at least 200 μm. Representative TUNEL-stained pancreas sections from (d) WT mice and (e) Gα_z-KO mice are shown, with white arrows indicating TUNEL⁺ β-cell nuclei. In both images, the scale bar represents 100 μm. Data were compared by Student t test. (a, b) n = 7 per group; (c) n = 5 per group. All error bars represent standard error of the mean.

The protective effect of Gα_z loss on hyperglycemia is β-cell specific

Gα_z has a relatively limited expression pattern, with protein expression being confirmed only in the brain, pancreatic islet, and platelets (12, 26–28). Nevertheless, we wanted to confirm that the effects on islet immune infiltration and β-cell replication, function, and death were due to Gα_z expression being lost from the islet cells and not the invading immune cells. Pancreata from WT and Gα_z-KO female mice were immunohistochemically stained for Gα_z, confirming that Gα_z was specifically expressed in the islet cells (localized around the plasma membrane as expected for a membrane-associated protein) and not in the exocrine pancreas or invading immune infiltrate [Fig. 5(a)]. A Gα_z-KO islet was provided as a negative control for antibody specificity [Fig. 5(b)].

Figure 5.

Gα_z is expressed in islets but not in the islet-infiltrating immune cells. Representative islet images from both (a) WT and (b) Gα_z-KO NOD pancreata that were subjected to immunohistochemical analysis for Gα_z protein expression (brown) and counterstained with hematoxylin (purple).

To confirm a lack of expression/effect of Gα_z in T-cells, we isolated RNA from CD4⁺ and CD8⁺ splenic T-cells and performed qRT-PCR analyses. Gα_z levels were not significantly different between the genotypes and were significantly lower than known Gα_z-expressing tissues, including islet and brain [Fig. 6(a)], and were essentially identical to a known negative control for Gα_z expression, the kidney. Islet and brain Gα_z-null samples did express an mRNA construct amplifiable by the Gnaz qRT-PCR primers used, but because the DNA cassette used to generate the KO interfered with the translation start site, production of Gα_z protein was completely absent in these tissues in the KO mouse (10, 29). We then investigated whether T-cell cytokine production was altered between WT and KO mice. Splenic CD4⁺ and CD8⁺ cells did not show any significant difference in the production of type 1 T helper (Th1) cell cytokines TNFα and IFNγ, suggesting that protection from hyperglycemia observed in Gα_z mice may be due to loss of Gα_z in the β-cells and not the immune cells [Fig. 6(b–d)].

Figure 6.

Gα_z mRNA is not expressed in splenic T-cells and does not significantly alter their type 1 T helper cell response. Splenic T-cells were harvested from female (♀) WT or Gα_z-KO mice at 17 to 19 weeks of age and separated into CD4⁺ and CD8⁺ populations by flow cytometry. (a) Gating strategy to obtain CD4⁺ and CD8⁺ T-cell populations. (b) Expression of Gnaz transcript in CD4⁺ and CD8⁺ T-cells, islets, brain, and kidney samples from WT NOD and Gα_z-KO animals. (c) Representative scatter plots for intracellular cytokine staining of TNFα and IFNγ in CD4⁺ and CD8⁺ T-cell populations stimulated with CD3ε and CD28 antibodies for 3 days. (d) Quantification of percent positive cells from data displayed in (c). Data were compared by Student t test; n = 3 to 5 per group. All error bars represent standard error of the mean. NS, not significant.

Finally, we used the MLD-STZ procedure, which includes an immune component but is initiated by β-cell destruction, to specifically test the role of β-cell Gα_z in β-cell survival to insult. First, we bred C57BL/6J mice containing exon 3 of Gα_z flanked by LoxP sites with mice expressing Cre under the control of the rat insulin promoter to generate experimental animals [Fig. 7(a)]. Animals were distinguished by the presence of one or two bands upon polymerase chain reaction amplification of genomic DNA [Fig. 7(b)]. Genotyping polymerase chain reaction primers was designed to detect the deleted allele; any animals showing this pattern were not used for experiments (data not shown). Isolation of tissues known to express Gα_z (islets, hypothalamus, and platelets) confirmed the β-cell–specificity of the Gα_z deletion in Gα_z-flox/flox;Cre-positive animals [Fig. 7(c)].

Figure 7.

Validation of β-cell–specific deletion of Gα_z-null and protection from streptozotocin-induced diabetes. (a) Breeding strategy to generate the β-cell–specific Gα_z-null mouse. Green triangles indicate LoxP sites. Blue boxes represent the Gα_z protein gene product that is disrupted by Cre-mediated recombination. F/F indicates the “floxed” (i.e., recombined) allele. (b) Representative genotyping results to identify experimental animals. (c) Gα_z protein expression in multiple tissues demonstrates islet-specific loss of Gα_z protein. Hypothalamus and platelets are positive controls for Gα_z expression, whereas the kidney serves as a negative control. Alpha-tubulin (α-tub.) or beta-actin (β-act.) was used as a loading control. (d) Blood glucose levels after MLD-STZ induction show protection from severe hyperglycemia in the Gα_z-F/F, Cre⁺ animals concurrently treated with 10 μg/kg exendin-4 (Ex4). Data were compared by two-way analysis of variance. All error bars represent the standard error of the mean. *P < 0.05 vs WT^Cre+; n = 4 or 5 per group.

Experimental animals were subjected to MLD-STZ induction with and without the addition of 10 μg/kg/d Ex4, which we previously demonstrated was sufficient to protect full-body Gα_z-null C57BL/6N mice from severe hyperglycemia induced by streptozotocin (16). Indeed, although WT mice treated with Ex4 or mice lacking β-cell Gα_z trended toward lower blood glucose values during the 6-week experimental time course, only the combination of β-cell–specific Gα_z loss and Ex4 treatment was able to protect mice from developing severe hyperglycemia, phenocopying our results from the full-body KO [Fig. 7(c)] (16) and implicating β-cell Gα_z as an initiating factor in β-cell loss in T1DM.

Discussion

The NOD mouse is a well-accepted model of T1DM because it replicates the key pathogenic characteristics of the human disease (17, 18). Female NOD mice develop early insulitis that typically progresses to overt diabetes. Furthermore, although the development of overt hyperglycemia in male NOD animals is of lower penetrance, they exhibit other important pathogenic aspects of T1DM, including β-cell dysfunction and insulitis. We previously showed that loss of Gα_z protects mice from both diet-induced glucose intolerance, a model of type 2 diabetes mellitus, and streptozotocin-mediated β-cell insult, a chemically induced model of T1DM (10, 15, 16). In both models, there were key improvements in β-cell replication, survival, or both, lending credence to testing the impact of the Gα_z-KO mutation in a physiologically relevant T1DM model.

We previously showed Gα_z-KO C57BL/6N mice are protected from developing severe hyperglycemia after MLD-STZ induction of diabetes only when coadministered a subtherapeutic dose of the GLP-1 mimetic Ex4 (16). In the NOD model, Gα_z loss alone was sufficient to protect from the development of severe hyperglycemia. These results are not necessarily discordant because MLD-STZ induces a rapid, immune-mediated loss of functional β-cell mass with recovery dependent not only on β-cell survival but also on the mounting of a β-cell replication response. In the NOD model, we propose that Gα_z is primarily playing a role in the initial apoptotic event, which then induces further migration of immune cells to the islet. When uncontrolled, this immune infiltration propagates continued cell death, as apoptotic and necrotic cells can be potent activators of dendritic cells and other antigen-presenting cells (30). This cycle causes the rapid immune-mediated destruction of pancreatic β-cells, resulting in T1DM pathogenesis. Gα_z-KO mice still have immune infiltration into the islets (Fig. 2; Supplemental Fig. 2 ^{(27.7MB, docx)} ); therefore, we do not suspect that loss of Gα_z is playing a role in the initial immune onslaught. Rather, we propose that in the absence of Gα_z, the progression of insulitis is reduced or halted and β-cell survival and function are maintained. This hypothesis is supported by the decrease in islet immune infiltration scores from 4 to 5 weeks to 17 weeks of age in Gα_z-null NOD mice [compare Fig. 4(a) and 4(b) to Fig. 2(a) and 2(b)].

T1DM is foremost an autoimmune, inflammatory condition; thus, the primary and often most successful therapeutic targets focus on these mediators and pathways. It is well accepted that T1DM pathogenesis is T-cell–mediated; therefore, we were interested specifically in CD3⁺, a marker of T-cell lineage, and CD8⁺, a cell surface marker specific to cytotoxic T-cells (31, 32), both of which were reduced in Gα_z-KO NOD islets [Fig. 3(a)]. Furthermore, IFNγ is secreted specifically by Th1 cells and is known to bind to receptors directly on β-cells that can induce apoptotic pathways via STAT1 and NF-κB (33). Because T1DM is a Th1 cell–mediated disease, high IFNγ and low interleukin-4 values are correlated with pathogenic β-cell insulitis (34). Although we do not see a difference in islet mRNA expression of NF-κB between WT and Gα_z-KO mice [in fact, it is slightly increased in KO mice; Fig. 3(c)], IFNγ transcript expression is reduced eightfold in Gα_z-KO NOD islets, regardless of sex [Fig. 3(c)]. In mice at 17 weeks of age, when islets were collected for cytokine/chemokine expression profiling, the initiation of apoptotic events had long since passed, which could explain why we did not see a change in NF-κB expression [Fig. 3(c)]. Reduction in expression of and likely local levels of IFNγ would lead to decreased expression of MHC II on antigen-presenting cells, such as macrophages and dendritic cells present in and around the islet. A reduction in MHC II expression would indicate a reduced capacity for MHC II to present antigen epitopes to other adaptive immune cells (35), thereby reducing the further activation of β-cell–specific inflammatory events. Reduced expression of IFNγ is likely necessary to prevent insulitis and maintain β-cell survival and function because of its ability to influence the cyclic apoptotic nature through multiple pathways.

Interestingly—and to our surprise—we found that TNFα was dramatically upregulated in islets of male NOD mice, regardless of genotype [Fig. 3(c)]. There is some evidence that TNFα offers a protective role in β-cells at later stages of insulitis (36–38). We believe these findings could offer some insight into the divergence in hyperglycemic development and β-cell pathogenesis seen between male and female NOD mice and should be a topic of further investigation.

The effects of inflammatory mediators in the pathogenesis of T1DM continue to be thoroughly investigated because of the direct and profound ties to the disease (39–42). However, cytokines and their local and systemic effects are highly influenced by the presence or absence of other cytokines. In vivo, cytokines are present in a milieu and are not likely to be found in isolation (43). It is the mixture of these proteins that leads to inflammatory phenotypes of chemoattraction and cytotoxicity. In addition, these molecules can have broad actions owing to a wide distribution of receptors making their therapeutic potential rather limited (44–46). Our finding that loss of Gα_z mitigates some of these key inflammatory events is exciting. Gα_z has an attractive therapeutic potential, in that targeting a single protein results in alterations in other key pathways that themselves are difficult to target. A role for Gα_z in natural killer cells has been previously suggested (46–49), although these results have not been replicated.

In our study, we confirmed that Gα_z is not expressed in the infiltrating immune cells, whereby its expression is clearly apparent around the islet cell membranes, and therefore propose that the changes seen in the islet cytokine and chemokine mRNA profiles are a result of the absence of Gα_z acting in the β-cell to mediate apoptosis and paracrine signaling. We previously showed that overexpression of both WT and constitutively active Gα_z in the INS-1 (832/3) rat insulinoma cell line increased cytotoxicity, decreased viability, and increased caspase cleavage, particularly in response to IL-1β, indicating a role for Gα_z in promoting stress-induced β-cell death pathways (16). In support of this, Gα_z-KO C57BL/6N islets are resistant to IL-1β–induced upregulation of mRNA encoding C/EBP homologous protein, Tribbles homologue 3, and Caspase 9, functional regulators of β-cell death (16). Our MLD-STZ study with the β-cell–specific Gα_z-null C57BL/6J mouse is fully consistent with a β-cell–centric effect of Gα_z loss in diabetes protection [Fig. 7(d)]. We expect that Gα_z is acting in a similar manner in the NOD mouse islets but were unable to test the intrinsic susceptibility to β-cell death with isolated islets in this model because of the rapid insulitis that occurred, which is statistically different between genotypes even at 4 weeks of age (Fig. 4).

Altered β-cell function and survival can potentially affect the function, activation, and/or migration of immune cells. Consistent with this, we observed a lesser degree of insulitis in Gα_z-KO mice most likely because of reduced antigen presentation and/or activation of immune cells. However, our initial assessment of T-cell function/cytokine production did not show a significant difference between KO and WT mice. More detailed immunophenotyping of T-cells and/or other immune cells that are critically implicated in T1DM is necessary to identify the contribution of these cells to the phenotype. A β-cell–specific KO of Gα_z in the NOD background will confirm whether the protection afforded the Gα_z-KO NOD mice is β-cell autonomous.

In summary, Gα_z is a crucial mediator of T1DM pathogenesis in a well-accepted mouse model of the disease and as a single modulator is able to exert action on multiple facets of the disease. Combined, these characteristics provide a relevant rationale for specifically targeting the action of Gα_z to promote β-cell function and survival in early T1DM.

Appendix.

Antibody Table

Peptide/Protein Target	Name of Antibody	Manufacturer, Catalog #, and/or Name of Individual Providing the Antibody	Species Raised in; Monoclonal or Polyclonal	Dilution Used	RRID
Gαz	Rabbit anti-Gαz	Abcam, AB150434	Rabbit; monoclonal	1:400	AB_150434
Insulin	GP anti-insulin	Dako, A056401-2	Guinea pig; polyclonal	1:500	AB_2617169
CD28	Anti-mouse CD28	BD Biosciences, 553295	Hamster anti-mouse; monoclonal	1 μg/mL	AB_394764
CD3ε	Anti-mouse CD3ε	BD Biosciences, 553058	Hamster anti-mouse; monoclonal	1 μg/mL	AB_394591
CD8α	Anti-mouse CD8α	BD Biosciences, 561093	Rat anti-mouse; monoclonal		AB_561093
TNFα	Anti-mouse TNF	BD Biosciences, 554418	Rat anti-mouse; monoclonal		AB_395379
INFγ	Anti-mouse IFNγ	BD Biosciences, 557735	Rat anti-mouse; monoclonal		AB_396843
CD4	Anti-mouse CD4	eBioscience, 56-0042-80	Rat anti-mouse; monoclonal		AB_494002

Abbreviation: RRID, Research Resource Identifier.