Reversal of New-Onset Type 1 Diabetes With an Agonistic TLR4/MD-2 Monoclonal Antibody
Type 1 diabetes (T1D) is currently an incurable disease, characterized by a silent prodromal phase followed by an acute clinical phase, reflecting progressive autoimmune destruction of insulin-producing pancreatic β-cells. Autoreactive T cells play a major role in β-cell destruction, but innate immune cell cytokines and costimulatory molecules critically affect T-cell functional status. We show that an agonistic monoclonal antibody to TLR4/MD-2 (TLR4-Ab) reverses new-onset diabetes in a high percentage of NOD mice. TLR4-Ab induces antigen-presenting cell (APC) tolerance in vitro and in vivo, resulting in an altered cytokine profile, decreased costimulatory molecule expression, and decreased T-cell proliferation in APC:T-cell assays. TLR4-Ab treatment increases T-regulatory cell (Treg) numbers in both the periphery and the pancreatic islet, predominantly expanding the Helios+Nrp-1+Foxp3+ Treg subset. TLR4-Ab treatment in the absence of B cells in NOD.scid mice prevents subsequent T cell–mediated disease, further suggesting a major role for APC tolerization in disease protection. Specific stimulation of the innate immune system through TLR4/MD-2, therefore, can restore tolerance in the aberrant adaptive immune system and reverse new-onset T1D, suggesting a novel immunological approach to treatment of T1D in humans.
Type 1 diabetes (T1D) is a complex disease caused by genetic and environmental factors that result in an autoimmune response to pancreatic β-cells (1–3). Innate and adaptive immune responses interact to produce soluble or contact-dependent factors that destroy islet β-cells (3–7). The incidence of T1D has risen rapidly worldwide since the mid-20th century, increasing by 3–4% per annum in Europe and the U.S. (8–11). One explanation for the rising incidence of T1D is the hygiene hypothesis, which suggests that insufficient microbiological stimulation of the innate immune system skews the adaptive immune system toward autoimmunity (12,13).
In support of this hypothesis, both NOD mice and patients with T1D demonstrate abnormalities in innate immunity. Serreze et al. (14) showed that NOD macrophages are developmentally defective, whereas Tisch and colleagues (15) showed an intrinsic signaling defect in NOD dendritic cells (DCs). Macrophages and DCs are among the earliest invading cells in NOD insulitis, produce proinflammatory tumor necrosis factor-α (TNF-α) (16), and are needed to maintain disease pathogenesis (17). Other studies have found similar enhanced innate immune responses in patients with T1D (18,19). Dysregulated innate immune signaling critically affects antigen-presenting cell (APC) cytokine production and costimulatory capacity, producing downstream alterations in T-cell immunity and thus altering the balance between regulatory and effector adaptive immune mechanisms (20–23). In the setting of overactive innate immune responses, a therapy that could retune innate immune responses might have beneficial downstream effects on adaptive immunity and thus potentially restore tolerance.
Toll-like receptors (TLRs) are innate pattern recognition receptors mediating the host defense against pathogens (24). The TLR4 receptor responds to lipopolysaccharide (LPS), an integral component of gram-negative bacteria, by forming a complex with MD-2 (25). TLR4/MD-2 signaling is mediated through two main adaptor proteins, MyD88 and Trif, that direct production of inflammatory modulators such as TNF-α, interferon-β (IFN-β), and IL-10 (24). TLR4 often is considered a proinflammatory molecule, and abnormalities of TLR4 signaling have been shown in both mouse and human T1D (26–28). Patients at genetic risk for the development of T1D also have dysregulated TLR-induced IL-1β and IL-6 responses compared with healthy individuals (29). However, studies on the effect of TLR4 knockout on T1D development have been inconsistent, showing increased, decreased, and no effect on T1D incidence (30–32). Experiments in hepatic immunology demonstrated that TLR4-expressing APCs can undergo endotoxin tolerance with repeated TLR4 stimulation (33,34). Endotoxin-tolerized immune cells have an altered immune response that has profound downstream effects on adaptive immunity (35). We hypothesized that the TLR4 abnormalities in T1D could be corrected by therapeutic targeting of TLR4-expressing APCs and that this could alter the course of new-onset T1D. To induce APC tolerance, we used an agonistic antibody to mouse TLR4/MD-2 (UT18 [hereafter referred to as TLR4-Ab]) compared with a control nonagonistic antibody to TLR4/MD-2 (UT15 [hereafter referred to as Ctrl-Ab]). We previously showed that the agonistic antibody induced tolerance to LPS and prevented allergic asthma (36–38); however, we never previously used these antibodies to reverse clinically established disease. TLR4-Ab was enormously successful in the setting of new-onset clinically evident T1D, permanently reversing disease in >70% of mice.
Research Design and Methods
Mice, Reagents, Flow Cytometry
NOD, NOD.BDC2.5, and NOD.scid mice were bred and maintained under specific pathogen-free conditions in accordance with institutional animal care guidelines at the University of Cincinnati College of Medicine Laboratory Animal Medical Services. Immunomagnetic beads were purchased from Miltenyi. Anti-CD3/CD28 beads and recombinant mouse IL-2 were purchased from Invitrogen (Carlsbad, CA). Carboxyfluorescein succinimidyl ester (CFSE) (Invitrogen) was used to assess cell division using the manufacturer’s protocol. Ultrapure LPS was purchased from InvivoGen. The production/characterization of TLR4/MD-2 monoclonal antibodies was previously described (36).
Splenocytes or bead-purified CD4 cells were incubated with 2.4G2 Fc-block and stained with the indicated antibodies (purchased from BD Biosciences or BioLegend). For intracellular staining, the cells were prepared according to the fixation/permeabilization kit/protocol (eBioscience). FACS data were collected using a FACSCalibur or FACSCanto system (BD Biosciences) and analyzed by FlowJo software (Tree Star).
Prevention of T1D
Female NOD mice (27–44 days old) were treated once a week for 3 weeks with an intraperitoneal injection of either Ctrl-Ab (5 μg) or TLR4-Ab (5 μg) and monitored for diabetes incidence (along with an untreated control group) until end-stage T1D (blood glucose [BG] >500 mg/dL) or after 200 days. TLR4-Ab dosing was determined by our prior experience (37). Female NOD mice were monitored starting at 12 weeks of age for clinical signs of diabetes (polyuria). Diabetes onset was assessed using urine glucose paper testing (Tes-Tape; Nasco) and confirmed and quantified with a standard one-step blood glucose meter.
Pancreatic insulitis was blindly scored as previously described (39) (0, no visible infiltration; 1, peri-insulitis with a cell depth less than five; 2, peri-insulitis with a cell depth greater than five; 3, insulitis with <50% islet infiltration; 4, insulitis with >50% islet infiltration; 5, islet scar). For quantification of immunohistochemistry, we digitally recreated the histology slide using Photomerge (Adobe Photoshop) and used ImageJ software to quantify insulin and glucagon-positive staining areas. Pixel area was measured using the ImageJ Measure function, and each islet was measured for total pixelated area. Individual Foxp3+ cells were counted per section.
Macrophage Isolation and Stimulation
For macrophage cytokine production experiments, 5 mL 3% thioglycollate (BD) was injected intraperitoneally into NOD mice, and peritoneal macrophages were harvested after 72 h. Fifty thousand macrophages per well were plated with serial dilutions of TLR4-Ab, Ctrl-Ab, and LPS. Supernatant was removed after 24 h and tested for TNF-α by ELISA (eBioscience). IFN-β was quantified using an interferon-stimulated response element reporter cell line.
T-Cell and APC:T-Cell Proliferation Assays
Splenic CD4+ T cells were purified using immunomagnetic beads and stained with CFSE (0.5 μmol/L), and 100,000 CD4+ T cells/well were cultured as follows: unstimulated; 20,000 anti-CD3/CD28 beads/well; and LPS, Ctrl-Ab, or TLR4-Ab (2 μg each). After 72 h, the cells were harvested, blocked with 2.4G2, stained with CD4-APC, and the CFSE dilution assessed by FACS.
For APC:T-cell assays, CD11c+ cells were purified with immunomagnetic beads, and 25,000 cells/well were cultured with 10 ng TLR4-Ab or Ctrl-Ab (or untreated) for 1 h. All wells were then washed and cocultured with either 1 μg anti-CD3 or 5 mmol/L BDC2.5 mimic peptide for an additional 1 h. The CD11c+ cells were subsequently restimulated with 100 ng TLR4-Ab, and 200,000 purified CFSE-labeled CD4+ cells were added. The CD4+ T-cell CFSE dilution was measured after 3 days.
Reversal of T1D
Prediabetic female NOD mice were randomly assigned to either Ctrl-Ab or TLR4-Ab treatment groups. Mice were assessed for diabetes as described previously herein. After onset of polyuria, at a BG between 200 and 250 mg/dL, mice were treated twice, 1 week apart, with either Ctrl-Ab (5 μg) i.p. or TLR4-Ab (5 μg) i.p. Mice were excluded from the study if 1) they never developed T1D, 2) they developed end-stage T1D before the second dose of antibody, or 3) their initial BG was not in the prespecified range. After the second treatment, mice were retreated if BG again exceeded 200 mg/dL. A control group of untreated NOD mice was used to quantify typical disease course after BG 200 mg/dL. A second reversal study was performed as aforementioned, but the initial BG was specified to fall between 300 and 400 mg/dL before treatment with TLR4-Ab.
In Vivo T1D Transfer Studies
NOD.scid mice (4–6 weeks) were treated once a week for 3 weeks with TLR4-Ab (5 μg) i.p. (n = 13) or were untreated (n = 10). Two hours after the final treatment, prediabetic NOD CD4+ and CD8+ T cells were transferred to all groups. Mice were sacrificed at onset of end-stage T1D (BG >500 mg/dL) or after 75 days. Islet pathology was assessed by histology as aforementioned, and splenic cell populations were assessed using flow cytometry as indicated in Fig. 6.
All statistical analyses were performed using an unpaired t test, Mann-Whitney test, log-rank (Mantel-Cox) test, or Fisher exact test using GraphPad Prism version 5.0 for Windows (GraphPad Software, San Diego, CA).
TLR4-Ab Prevents T1D and Targets APCs
We first assessed whether TLR4-Ab could prevent T1D in NOD mice. We treated young (27–44 days old) prediabetic female NOD mice with TLR4-Ab or Ctrl-Ab and monitored for T1D up to 200 days. TLR4-Ab significantly protected NOD mice from T1D, whereas Ctrl-Ab and untreated mice were statistically indistinguishable and had normal T1D incidence (Fig. 1A). TLR4-Ab–treated mice had decreased overall insulitis and significantly fewer highly infiltrated islets (Fig. 1B and C).
To determine TLR4-Ab cellular targets, we measured the kinetic and dynamic changes in cellular subsets after treatment with TLR4-Ab. TLR4-Ab in vivo resulted in significant increases of only CD11b+ cell subsets at 24 h (Fig. 1D). CD4+ cells treated in vitro with LPS or TLR4-Ab did not proliferate (Fig. 1F) or upregulate CD69 (data not shown), despite a prior suggestion that LPS could stimulate CD4+ cells (40). Moreover, in vitro LPS or TLR4-Ab treatment had no effect on the percentages of Foxp3+ regulatory T cells (Tregs) (Fig. 1G), consistent with the majority of prior reports showing no direct effect of LPS on Tregs (41). Despite this, at 14 days after TLR4-Ab treatment, there was a significant increase in Foxp3+ Tregs (data not shown), which persisted to at least day 28 post-TLR4-Ab treatment (Fig. 1E). These experiments suggest that initial targeting of innate immune cells by TLR4-Ab results in subsequent increases in Tregs, consistent with previous reports of the downstream effects of LPS (41).
Reversal of T1D Using an Agonistic TLR4/MD-2 Monoclonal Antibody
Because of the current lack of widespread screening to detect individuals at risk for T1D, most patients present clinically with elevated BG levels, polyuria, and weight loss. A clinical cure requires reversal of new-onset disease, so we assessed whether TLR4-Ab could reverse new-onset T1D. At the onset of polyuria, the earliest clinical sign of T1D in the current colony, the average BG was 208.3 ± 16.1 mg/dL, and T1D-induced weight loss had already begun (Fig. 2E). Mice were treated at this point with either Ctrl-Ab or TLR4-Ab or were left untreated. Ctrl-Ab treated and untreated mice rapidly (17.6 ± 2.9 and 16.7 ± 3.2 days, respectively) progressed to end-stage T1D (Fig. 2A–C and Table 1). In striking contrast, TLR4-Ab reversed new-onset T1D in a majority of mice (15 of 21) (Fig. 2A and D). Most TLR4-Ab–treated mice (19 of 21) had a clinical response (Fig. 2A and D and Table 1), defined by either permanent reversal of T1D (n = 15) or a significant delay to end-stage T1D (n = 4). In contrast, both the Ctrl-Ab and the untreated groups showed no clinical response (Fig. 2B and C and Table 1). TLR4-Ab–treated mice fell into two groups: those receiving only the initial two treatments of TLR4-Ab (11 of 21) and those that had to be retreated (10 of 21) (Fig. 2D, red marks indicate first retreatment). Remarkably, 9 of 11 mice receiving only the initial treatments had permanent reversal of T1D (Fig. 2D1). In the retreated group, 6 of 10 mice had persistent reversal of T1D, whereas 4 of 10 eventually progressed to end-stage T1D (Fig. 2D2). Overall, 71% of TLR4-Ab–treated mice had permanent reversal of T1D and 90% a clinical response, whereas Ctrl-Ab–treated mice had neither disease reversal nor clinical response (Fig. 2A–D). As shown in Table 2, TLR4-Ab treatment was clearly effective at reversing T1D at a BG level as high as 400 mg/dL; four mice had clinical responses to TLR4-Ab despite a BG level as high as 463 mg/dL (Table 2). New-onset T1D reversal with TLR4-Ab caused significantly decreased BG levels (from 204.7 to 159.7 mg/dL) (Table 1) and recovery from weight loss (Fig. 2E).
Our initial treatment protocol prespecified a BG range of 200–250 mg/dL. Although all untreated or Ctrl-Ab–treated mice defined in this cohort rapidly progressed to end-stage T1D (Fig. 2B and C), clinical T1D in NOD mice is often defined as >250 mg/dL. To test TLR4-Ab effectiveness in more-advanced T1D, we aged a second cohort but allowed the BG levels to rise to between 300 and 400 mg/dL (324.8 ± 10.1 mg/dL) before TLR4-Ab treatment. TLR4-Ab was highly and significantly effective in this cohort (Fig. 2F), confirming that TLR4-Ab treatment alone can reverse new-onset T1D.
TLR4-Ab Decreases Insulitis Severity and Preserves Insulin-Staining Islets in Diabetic Mice
We examined islet histology in five groups: 1) prediabetic mice, 2) untreated new-onset diabetic mice at the initial treatment point (when BG approximated 200 mg/dL [designated as BG200 mice]), 3) untreated end-stage diabetic mice, 4) Ctrl-Ab–treated mice, and 5) TLR4-Ab–treated mice. The overall insulitis score was significantly decreased in TLR4-Ab–treated compared with Ctrl-Ab–treated mice (Fig. 3A). The distribution of severely infiltrated islets was similar in Ctrl-Ab–treated mice and end-stage untreated diabetic mice (Fig. 3A). Of great interest, however, TLR4-Ab–treated mice had significantly increased stage 0 and significantly decreased stage 4 islets compared with both Ctrl-Ab–treated and BG200 mice (P < 0.0001 and P = 0.003 by χ2 test, respectively).
Total islet insulin staining in the TLR4-Ab–treated group was comparable to that in the nondiabetic control group (Fig. 3B and C), whereas the end-stage diabetic and Ctrl-Ab groups had minimal insulin staining (Fig. 3B and C). TLR4-Ab–treated mice had similar numbers of insulin-positive islets as the BG200 group, indicating preservation of insulin-positive islets by TLR4-Ab treatment. The amount of insulin per islet, however, was significantly increased in the TLR4-Ab–treated group compared with the other groups (Fig. 3E). Overall, the increased islet insulin-positive area in TLR4-Ab–treated mice correlates with the decreased numbers of severely infiltrated islets (Fig. 3A) and provides further evidence of a therapeutic effect of TLR4-Ab at the level of pancreatic islet β-cell insulin production.
TLR4-Ab Decreases APC Costimulatory Marker Expression, Expands Helios+Nrp-1+Foxp3+ Cells, and Increases IL-2, IL-4, and IL-10 In Vivo
We tested the effect of the TLR4-Ab reversal treatment on APC function in vivo. Two treatments with TLR4-Ab, 1 week apart, significantly downregulated CD80, CD86, and CD40 on CD11c+ cells in vivo compared with Ctrl-Ab–treated or untreated cells (Fig. 4A) and significantly increased serum IL-10 24 h after the second TLR-Ab treatment (Fig. 4B). We assayed serum cytokine levels at the end points for the five groups of mice described in Fig. 3A. TLR4-Ab–treated mice had significantly increased serum IL-2 and IL-4 compared with all other groups (Fig. 4C). Other cytokines were also increased in the TLR4-Ab group at the end points (although not significantly), including IL-33 (P = 0.06 vs. end-stage diabetic) (Fig. 4C); IL-21 (P = 0.10 vs. end-stage diabetic), and IL-10 (P = 0.32 vs. Ctrl-Ab) (Mann-Whitney test [data not shown]).
To test whether the peripheral Treg increase (Fig. 1) translated to the target tissue, we performed Foxp3 islet immunohistochemistry and demonstrated a significant increase in Foxp3+ cells in the TLR4-Ab–treated islets compared with Ctrl-Ab or nondiabetic islets (Fig. 4D). Thymic origin Tregs (tTregs) or peripheral origin Tregs (pTregs) can be distinguished by Helios and neuropilin-1 (Nrp-1) expression (42). TLR4-Ab treatment increased total Treg numbers and the Foxp3:CD4 ratio in vivo (Fig. 4E). Furthermore, the Treg increase was largely a result of a significant increase in the number of Helios+Nrp-1+Foxp3+ Tregs (Fig. 4E).
TLR4-Ab Treatment Decreases APC-Mediated T-Cell Proliferation
To further assess the effects of TLR4-Ab on APC function, NOD CD11c+ cells were tested for the capacity to stimulate T cells after a tolerogenic TLR4-Ab regimen (Fig. 5A and B). CD11c+ cells were treated twice with either TLR4-Ab or Ctrl-Ab and then cocultured with CD4+ T cells. Tolerizing treatment of CD11c+ cells with TLR4-Ab significantly decreased anti-CD3–mediated CD4+ T-cell expansion (Fig. 5A). We tested the effects of TLR4-Ab tolerization on APC-mediated antigen-specific T-cell proliferation using BDC2.5 transgenic T cells. TLR4-Ab treatment reduced APC-mediated BDC2.5 T-cell proliferation compared with Ctrl-Ab treatment or peptide alone (Fig. 5B).
A single dose of TLR4-Ab increased the number of CD11b+ APCs (Fig. 1D), raising an issue of the safety profile of the initial TLR4-Ab treatment. However, none of the 42 treated mice shown in Figs. 1 and 2 had any adverse effects after the initial TLR4-Ab treatment. To begin to understand this finding, we performed a side-by-side comparison of LPS and TLR4-Ab macrophage stimulation. The maximum TLR4-related (TNF-α, IFN-β) cytokine production mediated by TLR4-Ab was significantly below that of LPS, and the cytokine-producing effect of TLR4-Ab plateaued, whereas LPS-stimulated production continued to increase (Fig. 5C). This blunted cytokine response to TLR4-Ab plus increased production of immunomodulatory cytokines such as IL-10 may explain why a single administration of TLR4-Ab at the doses used here did not cause any apparent adverse clinical effects.
Targeted TLR4-Ab Treatment of Innate APCs Alone, in the Absence of B or T Cells, Is Sufficient to Prevent T1D in a Transfer Model
To directly demonstrate that targeting of innate immune cells with TLR4-Ab could mediate T-cell tolerization and protect against T1D, we treated 6-week-old female NOD.scid mice (lacking T or B cells) with TLR4-Ab, reconstituted them with prediabetic NOD CD4+ and CD8+ T cells, and assessed diabetes incidence compared with untreated NOD.scid recipients. TLR4-Ab pretreatment of NOD.scid significantly protected against the transfer of T1D by prediabetic CD4+ and CD8+ cells (Fig. 6A), contrasting sharply with rapid disease transfer in the untreated NOD.scid recipients. This decrease in disease incidence correlated with decreased islet pathology (Fig. 6B). Consistent with increased Tregs in TLR4-Ab–treated NOD mice (Fig. 4D and E), pretreatment of NOD.scid mice with TLR4-Ab increased the numbers and percentages of Tregs compared with the untreated group (Fig. 6C). These results clearly demonstrate that TLR4-Ab treatment causes a downstream increase in Treg responses, likely mediated through stimulation of innate immune cells, and that B cells are not necessary for TLR4-Ab–mediated protection (although these data do not exclude any role for B cells).
By targeting TLR4/MD-2 with an agonistic monoclonal antibody, we both prevented T1D and reversed new-onset T1D. Prevention of diabetes using a TLR4 agonist was not surprising in the NOD mouse model. Many agents, including LPS, can prevent diabetes but have subsequently shown no effect in treating or reversing the acute disease (i.e., LPS has no effect on prevention beyond 10 weeks and has no efficacy in new-onset clinical T1D ). We speculate that prevention studies may not predict efficacy in reversal of clinical T1D because they primarily affect checkpoint 1 (the onset of insulitis starting at ∼3 weeks and the subsequent nonprogression of T1D until ∼20 weeks) rather than reverse the immunological and inflammatory chaos (rapid and devastating onset of clinical T1D) initiated by checkpoint 2 (6). In striking contrast, TLR4-Ab reversed clinical T1D in a high percentage of mice, including mice with BG levels in the 400 mg/dL range, without any other disease modulators, such as islet transplantation or exogenous insulin. These results offer a new interpretation to the literature showing increased TLR4 activation in NOD and human new-onset T1D (26–29) and suggest that stimulation of TLR4, with resultant induction of APC tolerance, is the proper approach rather than global blockade of TLR4. This is a major difference, with implications for both therapy and safety. Global TLR4 blockade may introduce global immunosuppression. In contrast, stimulation through TLR4 initially releases some inflammatory cytokines, and subsequent treatment enhances a tolerogenic APC phenotype, as shown here. Is TLR4-Ab treatment immunosuppressive? Most current treatments of autoimmunity (e.g., methotrexate, cyclophosphamide, anti-TNF blockade) involve substantial immunosuppression. In contrast, TLR4-Ab treatment initially stimulates the immune system, subsequently enhances Treg responses, and then decreases (but does not eliminate) APC-mediated T-effector responses. Ultimately, we need to test the degree of immunosuppression mediated by TLR4-Ab by using coinfection at the time of TLR4-Ab and testing the ability of the altered immune system to fight infection.
A second safety-related question raised by this approach concerns the initial TLR4-Ab treatment. A single dose of TLR4-Ab has many downstream effects (Fig. 1), including release of inflammatory cytokines (38). Is this dangerous? None of 42 mice treated in this study showed adverse effects from the initial dose of TLR4-Ab. The blunted TLR4-Ab cytokine response compared with LPS (Fig. 5C) may explain this finding. TLR4-Ab appears less effective than LPS at causing maximal inflammatory cytokine release; in addition, immunomodulatory cytokines, such as IL-10, are increased (Fig. 4B). These results are consistent with our previous studies of TLR4-Ab compared with LPS and suggest that this antibody has a broader therapeutic window and margin of safety than LPS (38). However, many more studies are needed before this approach is tested in humans. Other TLR4 agonists have Food and Drug Administration approval (44,45); ultimately, the proper agent to stimulate TLR4 in humans may need an even broader therapeutic window.
Why TLR4-Ab, but not LPS, can reverse T1D is unclear. The mechanisms should be pursued in future studies; however, one obvious factor explaining TLR4-Ab therapeutic efficacy is its increased half-life. The half-life of LPS in vivo is <8 min (46), whereas IgG has a half-life of at least 7 days. Prolonging the antibody in circulation may lead to persistent signaling that contributes to the APC tolerization phenotype. We have also shown that TLR4-Ab induces less cytokine production than LPS, resulting in a lower plateau level of cytokine production (Fig. 5C). More-prolonged signaling (due to persistence in circulation) combined with altered APC effects could make TLR4-Ab more therapeutically effective.
One aspect of TLR4-Ab disease reversal unexplored here is its effects on islet β-cells. We show preservation of insulin staining and insulin-positive islet numbers in TLR4-Ab–treated mice; the increase in insulin staining per islet in these mice raises the question of whether the islets were hypertrophied or hyperplastic. Further study is needed to show whether TLR4-Ab treatment induces islet hyperplasia or hypertrophy, including the effects of TLR4-Ab on Reg gene family members.
We have shown by various approaches that TLR4-Ab does not directly stimulate Tregs but leads to downstream increases in Treg numbers in both the periphery and the islet. The present data suggest that a direct effect of TLR4-Ab on APCs leads to expansion of Tregs by these APCs, which is consistent with the most widely accepted and supported view of how LPS increases Treg numbers (41). We show that TLR4-Ab treatment primarily increases the Helios+Nrp-1+Foxp3+ Treg subset, which (although the field of tTregs vs. pTregs is controversial and evolving) has generally been designated as tTregs (42). tTregs play a critical role in regulating autoimmunity and may be functionally superior in this role compared with induced or pTreg subsets (42). The increase seen in tTregs after TLR4-Ab treatment thus supports the idea that TLR4-Ab treatment can reverse disease at least in part by boosting Treg-mediated protection. Given that Treg expansion takes several weeks in vivo, this pathway is likely to be quite complicated. However, future studies can directly test the role of Tregs in TLR4-Ab–mediated protection by depleting them after antibody treatment.
We found that several cytokines were increased in vivo after TLR4-Ab treatment, including IL-10 (significantly increased 24 h after treatment and increased, but not significantly, at the end point), IL-2, and IL-4 (at the end point after TLR4-Ab treatment). In addition, IL-33 was increased (although not significantly) and expressed at high levels at the TLR4-Ab end point. IL-4 and IL-10 have previously been implicated in T1D protection (47). IL-2 has recently emerged as a promising therapy for T1D, is undergoing trials in patients with T1D, and may act selectively on Tregs (48,49). Very little has been published on IL-33 in T1D, although it is protective in one model of T1D (50). Many more studies, including detailed kinetic studies, are needed to thoroughly define the cytokine profile in TLR4-Ab–treated mice over time before undertaking antibody blocking studies to determine the mechanistic role of any particular cytokine in TLR4-Ab therapy.
These results collectively show that stimulation through TLR4 can alter both the innate and the adaptive immune response and restore immune tolerance in new-onset T1D. The results, therefore, provide a strong foundation for exploring TLR4-mediated pathways as a novel means to reverse new-onset, clinically apparent T1D. The mechanisms of this strong clinical effect should be pursued in depth because the availability of Food and Drug Administration–approved agonistic anti-TLR4 agents (44,45) will make it possible to test this approach in new-onset human T1D.
Funding. K.J.B. was supported by the UC2019 Immunology Graduate Student Research Award. H.T. was supported by Japan Society for the Promotion of Science KAKENHI grant number 26460058. W.M.R. was supported by a U.S. Department of Veterans Affairs merit award grant BX000827-01A1 and by a University of Cincinnati Department of Medicine Distinguished Scientific Research Award.
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. K.J.B., H.T., K.K., and Y.W. performed experiments, analyzed results, and wrote the manuscript. S.O., J.D.K., and D.P.A. provided reagents and experimental ideas and edited the manuscript. W.M.R. supervised all experiments, analyzed data, and edited the manuscript. W.M.R. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
- Received December 8, 2014.
- Accepted June 23, 2015.
- © 2015 by the American Diabetes Association. Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered.