RANKL/RANK is required for cytokine-induced beta cell death; osteoprotegerin, a RANKL inhibitor, reverses rodent type 1 diabetes

Abstract

Treatment for type 1 diabetes (T1D) requires stimulation of functional β cell regeneration and survival under stress. Previously, we showed that inhibition of the RANKL/RANK [receptor activator of nuclear factor kappa Β (NF-κB) ligand] pathway, by osteoprotegerin and the anti-osteoporotic drug denosumab, induces rodent and human β cell proliferation. We demonstrate that the RANK pathway mediates cytokine-induced rodent and human β cell death through RANK-TRAF6 interaction and induction of NF-κB activation. Osteoprotegerin and denosumab protected β cells against this cytotoxicity. In human immune cells, osteoprotegerin and denosumab reduce proinflammatory cytokines in activated T-cells by inhibiting RANKL-induced activation of monocytes. In vivo, osteoprotegerin reversed recent-onset T1D in nonobese diabetic/Ltj mice, reduced insulitis, improved glucose homeostasis, and increased plasma insulin, β cell proliferation, and mass in these mice. Serum from T1D subjects induced human β cell death and dysfunction, but not α cell death. Osteoprotegerin and denosumab reduced T1D serum-induced β cell cytotoxicity and dysfunction. Inhibiting RANKL/RANK could have therapeutic potential.

Fig. 1.. OPG reduces β cell death, proinflammatory signaling pathway activation, and β cell dysfunction induced by cytokines in mouse islets.

Mouse islet cells treated without (Ctrl; C) or with cytokines (CTK), and vehicle (Veh; V) or mouse OPG (100 ng/ml; O) for 16 to 24 hours (A) were stained for insulin (red), TUNEL (green), and 4′,6-diamidino-2-phenylindole (DAPI) (blue), as represented in the confocal images; and (B) quantified for percent TUNEL-positive β cells (n = 6); or (C) were stained for insulin (red), annexin V (green), and DAPI (blue); white arrows indicate annexin V/insulin double-positive cells and (D) quantified for percent annexin V–positive β cells (n = 7). White bar indicates the scale for the immunofluorescent images. (E) Fold change in annexin V/PI–positive apoptotic INS1 cells assayed by flow cytometry after treatment without (C) or with cytokines (CTK) and vehicle (V) or OPG (1000 ng/ml; O) for 16 hours (n = 4 to 5); basal cell death in C-V was 6.88 ± 1.57%. Representative Western blot analysis and quantification of the ratio with tubulin of (F and G) p-NF-κB and (H and I) p-STAT1, in mouse islets treated without or with CTK, and Veh or OPG (100 ng/ml) for 24 hours (n = 6 to 7). *P < 0.05, **P < 0.01, ***P < 0.001 versus C-V; ^#P < 0.05 versus CTK-V; ^$P < 0.05 versus all other groups. Secreted insulin (nanograms per milliliter)/total insulin content at 2.2 and 22.2 mM glucose from mouse islets (J) treated with Veh or OPG (100 ng/ml) for 45 min (n = 5); or (K) treated with Veh, CTK, CTK + IgG (100 ng/ml), or CTK + OPG (100 ng/ml) for 24 hours (n = 3); *P < 0.05, ***P < 0.001 versus 2.2 mM of the same treatment group; ^#P < 0.05 versus 22.2 mM of CTK and CTK-Ig groups. Experiments were done in duplicate or triplicate; 5 to 10 fields and 1545 ± 183 β cells per sample were analyzed for (A) to (D). Individual symbols in the graphs represent independent experiments from individual mice, averaging duplicate (B, D, E, J, and K) or single (G and I) samples. All data represent means ± SEM. All statistical analysis was by analysis of variance (ANOVA) with Tukey’s post hoc analysis.

Fig. 2.. OPG enhances survival, and OPG and DMB reduce activation of proinflammatory signaling pathways induced by cytokines in primary human β cells.

Human islet cells treated without (Ctrl; C) or with cytokines (CTK), and vehicle (Veh; V) or human OPG (OPG, O) at 25 to 100 ng/ml (O-25, O-50, O-100) (A) were stained for insulin (red), TUNEL (green), and DAPI (blue), as represented in the confocal images; and (B) quantified for percent TUNEL-positive β cells after 24 hours (n = 4 in duplicate; 5 to 10 fields and 1145 ± 252 β cells per sample were analyzed). Representative Western blot analysis and quantification of the ratio with glyceraldehyde-3-phosphate dehydrogenase (GAPDH) or tubulin of (C and D) p-NF-κB and (E and F) p-STAT1, in human islets treated without (Ctrl) or with CTK, and vehicle (V), IgG (100 ng/ml; Ig), DMB (100 ng/ml; D), or OPG (100 ng/ml; O) for 24 hours (n = 4 to 7). Human islet cells treated without (Ctrl) or with CTK, and vehicle (V) or OPG (O) (G) were stained for p-NF-κB (red), insulin (green), and DAPI (blue), as represented in the individual and merged confocal images; (H) quantified for β cell–specific MFI of p-NF-κB after 90 min, (n = 3); or (I) were stained for p-STAT1 (green), insulin (red), and DAPI (blue), as represented in the individual and merged confocal images and (J) quantified for β cell–specific MFI of p-STAT1 after 15 min (n = 5), *P < 0.05, **P < 0.01, ***P < 0.001 versus Ctrl or C-V; ^#P < 0.05, ^##P < 0.01, ^###P < 0.001 versus CTK-V and CTK-Ig. White bar indicates the scale for the immunofluorescent images. Individual symbols in the graphs represent independent experiments on different human islet preps, averaging duplicate (B) or single (D, F, H, and J) samples. All data represent means ± SEM. All statistical analysis was by ANOVA with Tukey’s post hoc analysis.

Fig. 3.. Cytokine-induced rodent and human β cell death require RANKL/RANK.

Adv-LacZ– or Adv-Cre–transduced mouse islet cells after 48 hours (A) were stained for insulin (green), Cre-recombinase (red), and DAPI (blue), as represented in the individual and merged images (white bar indicates the scale for the images); or (B) the ratio of Rank/actin mRNA was analyzed by real-time qPCR (n = 3). *P < 0.05 versus LacZ. (C) Percent TUNEL-positive β cells in mouse islet cells from WT or Rank-floxed (RANKFl) mice transduced with Adv-Cre for 48 hours and treated without (Ctrl) or with cytokines (CTK) for an additional 24 hours (n = 5 to 6). **P < 0.01 versus WT-Ctrl; ^#P < 0.05 versus WT-CTK. Percent of TUNEL-positive β cells in human islet cells treated without (Ctrl) or with CTK for 24 hours and (D) with Veh (V), 100 ng/ml of hOPG (O), or hRANKL (R) alone, or in combination at a 1:1, 1:5, or 5:1 ratio of OPG/RANKL (O + R), respectively; ^$P < 0.05, ^$$P < 0.01 versus all other groups without symbols, on duplicate samples of n = 9; or (E) with Veh or DMB (0.1 to 100 ng/ml), with Ctrl-Veh represented as 100% (Ctrl-Veh value 1.54 ± 0.42%; n = 4 to 8); ***P < 0.001 versus Ctrl-Veh; ^#P < 0.05, ^##P < 0.01 versus CTK-Veh. Experiments were done in duplicate with 5 to 10 fields and 1646 ± 124 β cells per sample analyzed for mouse islets (C), and 5 to 10 fields and 1144 ± 167 β cells per sample analyzed for human islets (D and E). Individual symbols in the graphs represent independent experiments on individual mouse or human islet preps, averaging duplicate samples for all experiments (A to E). All data represent means ± SEM. All statistical analysis was by ANOVA with Tukey’s post hoc analysis.

Fig. 4.. RANK/TRAF6 interaction is required for cytokine-induced β cell death and for NF-κB nuclear translocation.

Percent cleaved caspase-3 (CC3)–positive INS1 cells treated without (ctrl) or with CTK for 16 hours and (A) with veh or TRAF6 chemical inhibitor (Chem-inh; 5 μM) (n = 3) or (B) with Veh, control-peptide (ctrl-pep), or TRAF6 peptide inhibitor (Pep-inh; 30 μM) (n = 3). (C) Percent TUNEL-positive β cells from mouse islet cells treated without (ctrl) or with CTK in the presence of ctrl-pep or Pep-inh for 24 hours (n = 3). INS1 cells treated without (ctrl) or with CTK in the presence of Veh, ctrl-pep, or Pep-inh for 30 min (D) immunostained for NF-κB (green) and DAPI (blue) and (E) quantified for percent NF-κB–positive nuclei (n = 4). **P < 0.01 versus ctrl in the same treatment group; ^#P < 0.05, ^##P < 0.01 versus CTK/Veh and CTK/ctrl-pep. Experiments were done in duplicate; with 8 fields and 7678 ± 1208 INS1 cells per sample (A and B), 3 fields and 2898 ± 270 INS1 cells per sample (E), and 5 to 10 fields and 889 ± 96 β cells per sample (C), analyzed. Individual symbols in the graphs represent independent experiments in INS1 cells or individual mouse islet preps, averaging duplicate samples for all experiments (A to E). All data represent means ± SEM. All statistical analysis was by ANOVA with Tukey’s post hoc analysis.

Fig. 5.. Effects of RANKL, OPG, and DMB on human immune cells.

Flow cytometry analysis of (A) percent RANK⁺ cells in human PBMCs from healthy blood donors. (B) Percent intracellular RANKL-positive CD4 and CD8 T lymphocytes before (Basal) and after (Stim) activation of human PBMCs with plate-bound anti-CD3 (0.5 μg/ml) (n = 3). IFN-γ, TNF-α, and IL-17 cytokine production (C) in CD4 (left) and CD8 (right) T cells in PBMCs stimulated with plate-bound anti-CD3 (0.5 μg/ml) or (D) in CD45RA⁺ naïve CD4 T cells stimulated in the presence of CD14⁺ monocytes (MN) with plate-bound anti-CD3 (0.5 μg/ml) and soluble anti-CD28 (1.0 μg/ml) (n = 3 to 6). Cytokines were assessed in the absence (untreated) or presence of 500 ng/ml each of OPG (blue bar) or DMB (pink bar) for 72 hours. Intracellular cytokine production is shown on the y axis as the ratio (fold) of % cytokine⁺ cells treated with OPG or DMB versus untreated group. Dotted line (at 1.0) represents no difference in intracellular cytokine production between untreated and treated. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001 versus untreated. (E) Levels (picograms per milliliter) of IL-1β and IL-6 in supernatants of CD14⁺ MNs incubated with RANKL (100 ng/ml) for 24 hours in the presence or absence of OPG (n = 6). ****P < 0.0001 versus basal; ^##P < 0.05 versus RANKL. Individual symbols in the graphs represent independent experiments on different human immune cell donors, averaging triplicate samples for all experiments (A to E). All data represent means ± SEM. Statistical analysis was by t test (C and D) and by ANOVA with Tukey’s post hoc analysis (E).

Fig. 6.. OPG reduces hyperglycemia and insulitis and reverses recent-onset diabetes in NOD/Ltj female mice.

Twelve-week-old NOD/Ltj female mice treated every other day with Veh (olive green, circle) or OPG at 1.0 μg/g (blue, triangle) body weight for 16 weeks (n = 5 mice per group) were assessed for (A) weekly blood glucose; (B) percent diabetes incidence (defined as blood glucose >250 mg/dl), no significance between the two groups by log-rank test. Eleven-week-old NOD/Ltj female mice treated daily for 5 weeks with Veh (olive green, circle) or OPG (blue, triangle) at 1.0 μg/g body weight were assessed for (C) weekly blood glucose; (D) insulitis by H&E staining of pancreatic sections with representative images showing leukocyte infiltration; and (E) quantification of percent degree insulitis scored manually on the H&E-stained pancreatic sections using a published formula (93), *P < 0.05 for >75% insulitis OPG versus Veh. (F) Schematic representation of experimental design for NOD/Ltj female mice with recent-onset diabetes (defined as blood glucose >250 mg/dl for three consecutive days), treated with Veh or OPG at 0.3 μg/g or 1.0 μg/g (daily) or 10.0 μg/g (2×/week), for 30 days, with assessment of body weight and blood glucose every 5 days. NOD/Ltj female mice with recent-onset diabetes treated with Veh (V, olive green, n = 8), OPG at 0.3 μg/g (O-0.3, pink, n = 3), 1.0 μg/g (O-1.0, blue, n = 7), or 10.0 μg/g (O-10, brown, n = 3) for 30 days were assessed for (G) blood glucose; (H) AUC for blood glucose and (I) percent diabetes incidence; *P < 0.05 versus Veh; ^$P < 0.05, ^$$P < 0.01 versus all other groups; by mixed model analysis (A and G), by ANOVA with Tukey’s post hoc analysis (H), or by log-rank test (I). Individual symbols in the graphs represent individual mice, and average values represent the average of individual mice in that group. All data represent means ± SEM.

Fig. 7.. OPG reverses recent-onset T1D, improves glucose tolerance, and increases plasma insulin in NOD/Ltj female mice.

(A) Schematic representation of experimental design for NOD/Ltj female mice with recent-onset diabetes (defined as blood glucose >250 mg/dl for three consecutive days), treated daily with Veh, IgG (1.0 μg/g), or OPG (1.0 μg/g), for 60 days. Body weight and blood glucose were measured every 5 days, IPGTT with plasma insulin was assessed at week 4, and blood and pancreas were harvested at day 60. NOD/Ltj female control mice that do not develop diabetes (Ctrl, purple, circle, n = 11), or mice with recent-onset diabetes treated with Veh (olive green, circle, n = 16), IgG (orange, square, n = 7), or OPG (blue, triangle, n = 10) for 60 days, were assessed for (B) blood glucose; (C) AUC for blood glucose; (D) glucose clearance during an IPGTT (n = 4 to 11 per group); (E) AUC for IPGTT. ^#P < 0.05, ^###P < 0.001, ^####P < 0.0001 versus Veh and IgG; ^$P < 0.05, ^$$$P < 0.001, ^$$$$P < 0.0001 versus all other groups; by mixed model analysis (B and D) or by ANOVA with Tukey’s post hoc analysis (C and E); (F) blood glucose and (G) plasma insulin at 0- and 15-min time points during the IPGTT; *P < 0.05, **P < 0.01, ***P < 0.001 versus Ctrl at the same time point; ^#P < 0.05, ^##P < 0.01 versus Veh and IgG at the same time point; ^^P < 0.01 versus Veh at the same time point; and ^&&&P < 0.001 versus 0 min of the same treatment; (H) plasma insulin at day 60; (I) percent diabetes incidence; ***P < 0.001 versus Ctrl; ^#P < 0.05, ^###P < 0.001 versus Veh and IgG; by ANOVA with Tukey’s post hoc analysis (F, G, and H) or by log-rank test (I). Individual symbols in the graphs represent individual mice, and average values represent the average of individual mice in that group. All data represent means ± SEM.

Fig. 8.. OPG treatment significantly increases β cell mass and proliferation in recent-onset T1D mice while maintaining normal β cell survival.

Pancreata from NOD/Ltj female mice with recent-onset diabetes defined as blood glucose >250 mg/dl for three consecutive days were harvested immediately (new onset, n = 5) or after 60 days of daily treatment with vehicle (Veh, n = 10), 1.0 μg/g IgG (n = 6), or 1.0 μg/g OPG (n = 10), or from control ND NOD/Ltj female mice of similar age (Ctrl, n = 7), and (A) stained for insulin as shown in the representative images; (B) assessed for β cell mass, ***P < 0.001 versus Ctrl; ^###P < 0.001 versus Veh and IgG, by ANOVA with Tukey’s post hoc analysis; (C) stained for insulin (green), pHH3 (red), and DAPI (blue); (D) quantified for percent pHH3-positive β cells (n = 5 to 11), ^P < 0.05 versus new onset by ANOVA with Tukey’s post hoc analysis; (E) stained for insulin (red), TUNEL (green), and DAPI (blue); and (F) quantified for percent TUNEL-positive β cells (n = 5 to 9). The bar indicates the magnification scale for the images. Individual symbols in the graphs represent individual mice. All data represent means ± SEM.

Fig. 9.. Serum from patients with T1D induces β cell cytotoxicity and dysfunction in human islets, but not α cell cytotoxicity; OPG and DMB protect against this cytotoxicity and improve function.

Human islet cells (n = 7 preps) cultured for 24 hours in islet media (ctrl), treated with cytokines (CTK), or in media in which the FCS was substituted with serum (10% v/v) from T1D or control ND donors (n = 9 each), were (A) stained for TUNEL (green), DAPI (blue), and insulin (red) (top) or glucagon (far red) (bottom), as shown in the representative confocal images (bar indicates the scale for the images); and (B) quantified for percent TUNEL-positive β cells (n = 9) or α cells (n = 4). Percent TUNEL-positive β cells in human islet cells (n = 3 to 5 preps) cultured for 24 hours in ctrl media or in medium containing serum from either T1D or ND (n = 6 to 9 each) subjects and treated with (C) either Veh or OPG (100 ng/ml) or (D) either Veh, IgG (100 ng/ml), or DMB (100 ng/ml); **P < 0.01, ***P < 0.001 versus Ctrl or ND of the same treatment group; ^#P < 0.05, ^###P < 0.001 versus Veh-T1D and IgG-T1D. (E) Secreted insulin (nanograms per milliliter)/total insulin content from human islets (n = 2 preps) at 2.2 and 22.2 mM glucose for 45 min, after culturing for 24 hours in regular media (ctrl) or in media containing ND or T1D serum (n = 4 each) in the presence of Veh, 100 ng/ml of IgG, OPG, or DMB, represented as fold over ctrl at 2.2 mM (insulin secretion/total insulin content at 2.2 mM in ctrl islets is 0.096 ± 0.03 ng/liter); *P < 0.05, **P < 0.01, ***P < 0.001 versus 2.2 mM of the same treatment group; ^^P < 0.01 versus 22.2 mM of Veh and IgG of ND serum. Five to ten fields and 1093 ± 199 β cells per sample in duplicate (A to D) were analyzed. Individual symbols in the graphs represent individual human serum samples, averaging duplicate samples for all experiments (A to E). All data represent means ± SEM. All statistical analysis was by ANOVA with Tukey’s post hoc analysis.