We previously showed that treating NOD mice with an agonistic monoclonal anti-TLR4/MD2 antibody (TLR4-Ab) reversed acute type 1 diabetes (T1D). Here, we show that TLR4-Ab reverses T1D by induction of myeloid-derived suppressor cells (MDSCs). Unbiased gene expression analysis after TLR4-Ab treatment demonstrated upregulation of genes associated with CD11b+Ly6G+ myeloid cells and downregulation of T-cell genes. Further RNA sequencing of purified, TLR4-Ab–treated CD11b+ cells showed significant upregulation of genes associated with bone marrow–derived CD11b+ cells and innate immune system genes. TLR4-Ab significantly increased percentages and numbers of CD11b+ cells. TLR4-Ab–induced CD11b+ cells, derived ex vivo from TLR4-Ab–treated mice, suppress T cells, and TLR4-Ab–conditioned bone marrow cells suppress acute T1D when transferred into acutely diabetic mice. Thus, the TLR4-Ab–induced CD11b+ cells, by the currently accepted definition, are MDSCs able to reverse T1D. To understand the TLR4-Ab mechanism, we compared TLR4-Ab with TLR4 agonist lipopolysaccharide (LPS), which cannot reverse T1D. TLR4-Ab remains sequestered at least 48 times longer than LPS within early endosomes, alters TLR4 signaling, and downregulates inflammatory genes and proteins, including nuclear factor-κB. TLR4-Ab in the endosome, therefore, induces a sustained, attenuated inflammatory response, providing an ideal “second signal” for the activation/maturation of MDSCs that can reverse acute T1D.
Introduction
Type 1 diabetes (T1D) is a complex autoimmune disease triggered by genetic and environmental factors that ultimately results in the destruction of pancreatic islet β-cells. The progression of T1D is mediated by cellular responses of the innate and adaptive immune system. T cells are primarily responsible for destruction of the β-cells; however, soluble and cellular mediators of the innate immune system have a crucial role in promoting T-cell effector function. In T1D, tissue-resident macrophages and dendritic cells are among the first to respond to β-cell insult and are essential for the retention of diabetogenic T cells during peri-insulitis (1–5). These antigen-presenting cells (APCs) produce proinflammatory cytokines, which drive lymphocytic infiltration of the islet and further promote priming and differentiation of destructive effector T cells (6–9). Despite the critical role of APCs in T1D pathogenesis, our ability to reverse acute T1D with agonistic anti–toll-like receptor 4 (TLR4)/MD2 antibody (TLR4-Ab) was surprising, and the mechanism was unclear (10). Given the high expression of pathogen recognition receptors, such as TLR4 on APCs, our initial hypothesis was that TLR4-Ab induced “endotoxin tolerance.” However, the presence of downregulated, but still elevated amounts of inflammatory cytokines did not explain reversal of acute autoimmunity (11). The studies presented here provide further data on the mechanism of TLR4-Ab–mediated reversal of T1D.
The TLR4 signaling axis plays a role in the recruitment of myeloid-derived suppressor cells (MDSCs) (12–15), and nuclear factor-κB (NF-κB) activation plays a role in MDSC expansion (16). MDSCs are morphologically and functionally heterogeneous but have a well-defined role in the suppression of natural killer and effector T cells. Although they are most commonly studied in the context of cancer, MDSCs accumulate in many other disease models of chronic inflammation and autoimmunity, including T1D (17–19). MDSCs are unique in that they are not classically differentiated by growth factors; rather, they are recruited in at least two steps: 1) mobilization from the bone marrow as immature myeloid cells and 2) activation by chronic immune stimulation at a site of infection, inflammation, or tumor microenvironment (20). Here, we show induction of MDSCs, with specific gene expression profiles by RNA sequencing (RNA-seq), that suppress T cells in TLR4-Ab–treated NOD mice and reverse acute T1D. Furthermore, we establish how TLR4-Ab differentially activates APCs, compared with lipopolysaccharide (LPS), by sequestering in the early endosome. These results provide novel insight into the molecular mechanisms that underlie the mild agonist phenotype and reversal of acute T1D.
Research Design and Methods
Mice and Treatment With TLR4 Ab
NOD mice were bred and maintained and all procedures involving mice conducted in accordance with institutional animal care and use committee guidelines at the University of Cincinnati Laboratory Animal Medical Services. The production and characterization of TLR4/MD2 (control [Ctrl]-Ab/TLR4-Ab) monoclonal antibodies was previously described (21). Prediabetic female NOD mice were randomly assigned to either Ctrl-Ab or TLR4-Ab treatment groups. Diabetes was confirmed by blood glucose [BG] meter. After onset of T1D (BG 200–250 mg/dL), mice were treated twice, 1 week apart with either Ctrl-Ab (5 μg) or TLR4-Ab (5 μg) injected intraperitoneally.
Flow Cytometry
Intra-islet hematopoietic cells (IIHCs) and splenocytes were isolated from PBS- and TLR4-Ab–treated mice as previously described (22). These cells were treated with Fc block (BD Biosciences), stained with the indicated antibodies and propidium iodine, run on an LSR II flow cytometer (BD Biosciences), and analyzed using FlowJo software.
RNA-Seq Analysis and PCR Array
RNA-seq was performed on splenocytes and on CD11b+ FACS-sorted cells from TLR4-Ab– versus Ctrl-Ab–treated diabetic mice as previously described (23). RNA-seq analysis was performed using Bowtie (24), TopHat2 (25), and the Cufflinks2 pipeline (26), as previously described. Differentially expressed gene signatures were identified using Audic-Claverie test (P < 0.05) and Student t test (false discovery rate <0.05), followed by a twofold change requirement. Gene ontology/biological network analysis was carried out using ToppGene (27) and ToppCluster (28) and analyzed in Cytoscape. For PCR arrays, cDNA was amplified on a StepOne Real-Time PCR System using the RT2 Profiler PCR Array Mouse Innate and Adaptive Immune Responses Kit (QIAGEN) and RT2 SYBR Green qPCR Mastermix (QIAGEN). Cycle threshold values were uploaded to the QIAGEN GeneGlobe Analysis Center website for analysis.
Bone Marrow Transfer Experiments
NOD bone marrow cells were cultured in the presence of either TLR4-Ab (250 ng/mL) or Ctrl-Ab (250 ng/mL) for 7 days. Cells were trypsinized and harvested for flow analysis and transfer. Two million harvested cells were transferred into diabetic (BG >200 mg/dL) NOD recipients. Five separate experiments were done. The end point was BG >500 mg/dL.
CD11b-Mediated T-Cell Suppression Assay
NOD mice were treated twice with 10 μg of TLR4-Ab or Ctrl-Ab. One week after the second treatment, splenic CD11b cells were purified using magnetic cell separation (MACS) beads. The cells were FACS sorted for CD11b+Gr1+ MDSCs. CD4 cells were MACS purified from untreated NOD spleen and stained with CellTrace Violet (Invitrogen). One hundred thousand CD4 T cells were cultured with 50,000 CD11b+Gr1+ MDSCs and CD3/28 activation beads (Dynabeads). Control cells without MDSCs were either left unstimulated or stimulated with CD3/28 beads. On day 3, the cells were stained with CD4 PerCP (BD Biosciences), followed by viability dye eFluor 780 (Thermo Fisher Scientific) and analyzed by FACSCanto and FlowJo.
TLR4 Trafficking and Colocalization Experiments
Bone marrow was harvested from C57BL/6 mice and cultured in complete DMEM (DMEM + 10% FBS + 1% penicillin/streptomycin) with 40 ng/mL macrophage colony-stimulating factor (CSF) for 5 days to generate bone marrow–derived macrophages (BMDMs). BMDMs were plated in 24-well plates at 2 × 106 cells/mL in complete DMEM with Fc block (1:5,000; BioLegend), then pulsed with TLR4-Ab (UT18 100 ng), Ctrl-Ab (UT15 100 ng), or LPS (100 ng) for 5 min. BMDMs were stained in FIX & PERM (Invitrogen) with antibodies against CD11b (BioLegend), EEA1 (Novus Biologicals), and UT18/UT15-specific secondary antibodies (BioLegend). Cells stimulated with LPS were fixed and stained with TLR4-Ab to detect TLR4/MD2 localization. Internalization and colocalization was evaluated using ImageStream. TLR4/EEA1 colocalization was quantified using the bright detail similarity wizard. The internalization wizard generates an internalization erode score; scores >1.0 indicate internalization of TLR4. IDEAS quantification data from two experiments (>5,000 events, gated on CD11b+ cells) was imported to GraphPad Prism software, and P values were calculated using two-way ANOVA.
Western Blots
Two million RAW 264.7 cells (TIB-71; ATCC) in complete DMEM were stimulated with UT18 (100 ng), UT15 (100 ng), or LPS (100 ng) for various times, pelleted, washed, resuspended in 2× Laemmli sample buffer (Bio-Rad), and sonicated. Proteins were separated by electrophoresis in 4–20% PAGE and transferred to nitrocellulose membranes, blocked with 5% BSA in Tris-buffered saline with Tween, and probed with antibodies against phosphorylated extracellular signal–regulated kinase (P-ERK), phosphorylated p38 mitogen-activated protein kinase (P-p38 MAPK), phosphorylated c-Jun N-terminal kinase (P-JNK), inhibitor of κBα (IκBα), and GAPDH (Cell Signaling Technology). After incubation with a horseradish peroxidase–conjugated secondary antibody (Cell Signaling Technology), blots were incubated with electrochemiluminescence substrate (Thermo Fisher Scientific) and developed with film. Densitometry analysis of protein bands was quantified using ImageJ software as previously described (29).
Data Analysis
All statistical analyses were performed using GraphPad Prism 6 for Windows (GraphPad Software). Significance testing was done using either the unpaired t or Mann-Whitney tests for sample comparisons and the log-rank test for survival curve analysis.
Data and Resource Availability
Resources generated and/or analyzed during the current study are available from the corresponding authors upon reasonable request.
Results
TLR4-Ab Reverses Acute T1D in NOD Mice
We previously demonstrated that TLR4-Ab treatment of NOD mice with acute T1D results in reversal of T1D disease (in mice with initial BG up to 400 mg/dL), with significantly decreased BG levels, recovery from weight loss, and a significantly improved islet infiltration score (10). Successful treatment was accompanied by significantly increased insulin area per insulin-positive islet cell, reflecting larger insulin-positive islets in treated mice compared with untreated or newly diabetic mice (10). We have subsequently studied a large cohort of NOD diabetic mice treated twice, 1 week apart, with TLR4-Ab versus Ctrl-Ab and have focused our studies on the immunological consequences of treatment 1 week after the second antibody injection. TLR4-Ab successfully treated acute T1D in this group, confirming our previously published results (Fig. 1). In the TLR4-Ab–treated group, only 15% (7 of 46) of the mice progressed to end-stage T1D within 2 weeks of T1D onset compared with 53% (21 of 40 mice) that progressed in the Ctrl-Ab–treated group. Only 3 mice in the TLR4-Ab group progressed to end-stage T1D (BG >500 mg/dL) before the second dose compared with 11 mice in the Ctrl-Ab group. The final BG in the TLR4-Ab group was significantly less than in the Ctrl-Ab group (Fig. 1). These results in a large cohort of treated mice support our prior findings of significant reversal of acute T1D with two treatments of an agonist TLR4-Ab.
TLR4-Ab Treatment In Vivo Increases CD11b+ MDSCs in the Spleen and Islets of NOD Mice
Given that the majority of TLR4-Ab–treated mice achieved lower BG levels and reduced islet infiltration, we sought to immunophenotype the cells in the spleen and islets of TLR4-Ab–treated mice that may contribute to disease reversal. Splenocytes harvested from NOD mice 24 h after the second TLR4-Ab treatment showed increased numbers and percentages of CD11b+ cells compared with splenocytes from PBS- or Ctrl-Ab–treated mice (Fig. 2A and B). Within this CD11b+ population, increases in both monocytic and granulocytic MDSC subsets were observed in TLR4-Ab–treated mice in addition to increased macrophage and granulocyte (polymorphonuclear neutrophil) populations (Fig. 2B). Splenocytes and IIHCs harvested 7 days after the final treatment still had significantly increased percentages of CD11b+ cells, and splenocytes had significantly increased numbers compared with their PBS counterparts (IIHC CD11b numbers were clearly increased but did not reach significance) (Fig. 2C and D), suggesting that TLR4-Ab treatment results in the retention of these cells rather than transient recruitment and infiltration.
RNA-Seq Data From TLR4-Ab–Treated Mice Show Upregulation of Genes Associated With CD11b+Ly6G+ Cells and Downregulation of Genes Associated With CD4+ T Cells
Given the longevity of the CD11b+ cell population in both the spleen and islet, we sought to characterize the transcriptional profile of these cells. RNA-seq of splenocytes was performed to build cluster enrichment networks based on genes associated with biological processes and pathways that were significantly upregulated or downregulated in TLR4-Ab–treated mice compared with those treated with Ctrl-Ab. Splenocytes harvested 1 week after the second TLR4-Ab treatment had an upregulation of genes associated with CD11b+Ly6G+ cells (granulocytic MDSCs) compared with their PBS and Ctrl-Ab counterparts (Fig. 3 and Supplementary Table 1). Additionally, T-cell–associated genes were downregulated in TLR4-Ab–treated mice, supporting the hypothesis that CD11b+ cells mediate suppressive effects on T cells in vivo.
CD11b+ Cells Sorted From TLR4-Ab–Treated Mice Display Upregulation of Genes Associated With a Bone Marrow–Derived Myeloid Cell Transcriptional Profile
To determine the transcription profile of TLR4-Ab–treated CD11b+ cells, CD11b+ splenocytes were FACS sorted from diabetic mice treated with either TLR4-Ab or Ctrl-Ab, as in Fig. 3, and analyzed by RNA-seq. A heat map was generated using 261 significantly upregulated genes in the TLR4-Ab–treated group (Fig. 4A), these genes were used to build a cluster enrichment network of significantly associated gene ontology (GO) categories (Fig. 4B). The significantly upregulated genes were significantly associated with myeloid, CD11b+, bone marrow–derived cells in coexpression atlas databases (Fig. 4B, light green GO categories). Furthermore, the first neighbors of the indicated GO coexpression atlas categories showed that a large proportion of the upregulated genes are linked to bone marrow–derived, CD11b+ gene signatures from the coexpression databases (Supplementary Fig. 1). These results provide strong evidence that the TLR4-Ab–induced increased CD11b+ population is derived from the bone marrow. Notably, the RNA-seq data show a significant association with both granulocytic/neutrophil and macrophage pathways and overall, show a significant association with activation of the innate immune system. Several novel gene families (e.g., kinesins, flotillins, formyl peptide receptors) are upregulated and will be addressed in the discussion.
CD11b+Gr1+ Cells Sorted From TLR4-Ab–Treated Mice Suppress T-Cell Proliferation Ex Vivo
Having met the criteria for the first step of MDSC generation (expansion and mobilization of immature myeloid cells), we next assayed whether TLR4-Ab could produce functionally active MDSCs and suppress T-cell activation. CD11b+Gr1+ cells were therefore sorted from NOD mice 1 week after treatment with either TLR4-Ab or Ctrl-Ab and cocultured with donor-matched CD4+ T cells. T cells cultured with CD11b+Gr1+ cells from TLR4-Ab–treated mice demonstrated reduced proliferation compared with T cells cultured with Ctrl-Ab–treated CD11b+Gr1+ cells (Fig. 4C and D). As MDSCs share phenotypic markers with monocytes and neutrophils, they are predominantly defined by their functional ability to suppress T-cell activation through metabolic paralysis (30–32). Our data demonstrate that highly purified CD11b+Gr1+ cells from TLR4-Ab mice suppress T-cell proliferation, further supporting our hypothesis that TLR4-Ab induces MDSCs.
CD11b+ Cells Generated From TLR4-Ab–Treated Bone Marrow Ameliorate Progression of Acute T1D
Next, we tested the effect of TLR4-Ab–treated bone marrow–derived cells on acute T1D. All hematopoietic cell subsets except megakaryocyte-erythroid progenitors have been reported to express functional TLR4/MD2, with the highest expression observed on hematopoietic stem cells and granulocyte-monocyte progenitors (12). One theory of MDSC generation suggests a two-step model wherein immature myeloid cells are first expanded in the bone marrow and introduced into the circulation and second, stimulated by chronic inflammation to become functionally suppressive MDSCs (16). We therefore tested whether TLR4-Ab treatment of bone marrow was sufficient to generate a population of suppressor cells in vitro. Bone marrow grown in the presence of TLR4-Ab yielded an enrichment of CD11b+ cells that express Ly6C and/or Ly6G, compared with Ctrl-Ab–treated bone marrow (Fig. 5A), showing that TLR4-Ab alone was sufficient to drive differentiation of CD11b+Gr1+ cells from bone marrow. TLR4-Ab–differentiated bone marrow cells transferred into acutely diabetic NOD recipient mice significantly reduced progression to end-stage T1D and final BG levels compared with diabetic recipients that received Ctrl-Ab–treated bone marrow cells (Fig. 5B and C). Thus, bone marrow–derived, TLR4-Ab–induced MDSCs can protect mice from progression of acute T1D.
TLR4-Ab Upregulates Expression of Inhibitory IκBα and Demonstrates an Intermediate Activation of MAPKs Compared With Proinflammatory LPS
LPS, the prototypical TLR4 agonist, drives differentiation of hematopoietic stem cells and granulocyte-monocyte progenitors into monocytes and macrophages with an inflammatory phenotype (12), but the phenotype of cells generated from TLR4-Ab–conditioned bone marrow is markedly different. LPS, in contrast to TLR4-Ab, cannot reverse acute T1D (33). Given that TLR4-Ab can induce suppressor MDSCs, we next sought to understand how TLR4-Ab activates target cells differently at the molecular level compared with LPS. We quantified phosphorylation of key signaling adaptors in the NF-κB pathway in TLR4-Ab–treated RAW 264.7 cells. We observed an overall downregulation of P-JNK/stress-activated protein kinase, P-p38 MAPK, and P-ERK, as well as upregulation of IκBα in cells stimulated with TLR4-Ab compared with LPS (Fig. 6). Sustained expression of IκBα is particularly interesting since it blocks nuclear translocation of p50/p65 (NF-κB) and the activation of genes that produce proinflammatory cytokines. These data suggest that TLR4 MyD88-dependent signaling is activated to a significantly lesser degree than with LPS. Furthermore, PCR array data from splenocytes harvested from NOD mice 1 week after TLR4-Ab treatment shows broad downregulation of multiple inflammatory genes compared with mice treated with LPS (Fig. 7). Interestingly, even 1 week after the last TLR4-Ab treatment, we still observe a significantly increased ratio of IκBα:NF-κB compared with splenocytes from LPS-treated mice (Fig. 7). These results demonstrate a prolonged, suppressed inflammatory phenotype in TLR4-Ab cells compared with LPS, which may explain the inability of LPS to reverse acute disease. To further understand this critical difference in two TLR4 agonists, we next studied the effect of both treatments on TLR4 endosomal cycling.
TLR4-Ab Remains Colocalized With EEA1 and Sequesters TLR4/MD2 Inside the Endosome for at Least 24 h
Given the decreased NF-κB activation observed after TLR4-Ab treatment compared with LPS, we tested whether TLR4-Ab was inducing internalization of the TLR4/MD2 receptor complex. Receptor localization impacts the receptor’s accessibility to signaling mediators. In TLR4 signaling, while MyD88 signaling can be initiated at the cell surface and internally from the endosome, TRIF/TRAM signaling requires internalization of TLR4 to the endosome (34). For these experiments, we used imaging flow cytometry to perform internalization and colocalization analyses. In cells pulsed with TLR4-Ab, TLR4 colocalizes with early endosomal marker EEA1 within minutes (Fig. 8). These data correspond with a previous report that UT12, another TLR4/MD2-Ab agonist, induces internalization of TLR4 (35). Ctrl-Ab did not induce internalization of TLR4 in any experiments (Supplementary Fig. 2). In contrast, cells pulsed with LPS also demonstrated internalization of TLR4, and as expected, LPS was quickly processed and TLR4 recycled back to the cell surface within 30–60 min after the stimulation (Fig. 8). Surprisingly, TLR4-Ab remains colocalized with EEA1 for a prolonged period of time, retaining TLR4 in the early endosome for at least 24 h (Fig. 8). The data in Figs. 6–8 support the hypothesis that TLR4-Ab initiates a prolonged, chronic proinflammatory response by sequestering TLR4/MD2 within the endosome for a much longer time frame than the natural TLR4 agonist LPS. The effect of this sequestration is to alter downstream signaling events for an extended period of time; protein expression of key signaling molecules was altered by 120 min after treatment (Fig. 6) while gene expression (Fig. 7) was altered for at least 1 week after treatment compared with LPS. The prolonged, but decreased inflammatory response compared with LPS is ideally suited to provide the “second signal” for MDSC activation of bone marrow–mobilized immature myeloid cells.
Discussion
Our data provide novel insight into the mechanisms underlying TLR4-Ab reversal of acute T1D in NOD mice. An increase in the peripheral MDSC population (most likely derived from bone marrow precursors), evidence of T-cell immunosuppression, and recapitulation of disease protection through adoptive transfer are consistent with published standards for MDSC identification and functional characterization (36). Protection from T1D by MDSCs has been previously shown but not reversal of acute T1D as we show here (17,18). Patients with T1D show an increase in the peripheral population of MDSCs, but they are not maximally suppressive (19,37). Whitfield et al. (19) described weak T-cell suppressive ability of this population that is lost with the addition of granulocyte-macrophage CSF and interleukin-1β in culture. These results indicate that cells of an MDSC phenotype are induced in response to T1D but that full licensing of their functional suppressive capacity has not been achieved. Unlike in cancer and infectious diseases, which induce a strong MDSC response, autoimmune diseases induce a weak MDSC response (38). The evidence would therefore suggest that in T1D, an MDSC response is suboptimal and that amplification of MDSC number and suppressive ability has therapeutic potential.
MDSCs develop in response to chronic inflammatory conditions and require two signals to become functionally immunosuppressive. A primary myelopoietic signal (e.g., granulocyte-macrophage CSF, granulocyte CSF, CSF-1, interleukin-6, Notch ligands, adenosine) expands and mobilizes the MDSC precursor population from the bone marrow, while chronic inflammatory mediators provide a second signal critical for MDSC maturation (16). Clinical conditions in which strongly immunosuppressive MDSCs are generated include postoperative stress (39), burns (40), and stroke (41), and these conditions leave patients susceptible to infections and sepsis. A common thread among these diverse pathologies is the presence of overt tissue damage and cellular stress that can provide a second signal necessary to activate and maintain the MDSC population. In autoimmunity, however, disease progression is gradual. In T1D the destruction of β-cells is regulated through highly controlled apoptotic processes, and reduction of β-cell numbers occurs over a period of years (42). Our hypothesis is that despite significant insulitis and chronic inflammation within the islet, sufficient to provide the first, myelopoietic signal and increase peripheral populations of MDSCs, the islets lack sufficient damage-associated signals to activate their suppressive potential and long-term survival. This scenario is consistent with the findings of Whitfield et al. (19).
Our molecular data establish TLR4-Ab as a weak/moderate TLR4 agonist with a unique ability to induce sustained activation and sequestration of the receptor. The cellular response is strong enough to result in cytokine production that is sustained for at least 7 days (10), altered gene expression in immune cells for at least 7 days (Fig. 7), and TLR4 sequestered in the early endosome for at least 24 h (Fig. 8). TLR4 is a unique innate immune receptor that has both inflammatory and tolerogenic functions; it can unleash a lethal cytokine storm during sepsis or can induce tolerance to protect against autoimmunity. The accessory molecules CD14 and MD2 partly determine which response predominates; they aid in the recognition of LPS as well as influence two signaling pathway options, the classical MyD88/NF-κB pathway or the alternative TRIF/IRF3 pathway. CD14 binds to LPS/TLR4 and brings them into lipid rafts where TIRAP/MyD88 signal via NF-κB and initiate inflammation. MD2 binds LPS and dimerizes TLR4 to initiate endocytosis. In the endosome, the alternative TRIF/TRAM signaling pathway can activate the IRF3 transcription factor to initiate a type 1 interferon response (34,43). In the NOD T1D mouse model, deletion of MyD88 is protective, but only if TRIF is functional and the mice have a microbial community in the gut. If the mice are germ free or if TRIF is knocked out, the disease course is unchanged, implying that activation of the alternative pathway is critical for protection (44–46). There is evidence that our TLR4-Ab binds to a region close to the LPS activation site (21), and we show here internalization and sequestration of the TLR4-Ab complex within the early endosomes. Our data show not only trafficking of TLR4-Ab to the early endosome but also prolonged endosome sequestration compared with LPS, suggesting enhancement of the alternative pathway. Taken together, our results suggest that TLR4-Ab is mimicking a damage-associated molecular pattern–mediated tissue response by sequestration in the endosome that provides prolonged second signals necessary for MDSC activation/generation.
We propose that cell surface binding of TLR4-Ab increases mobilization of bone marrow–derived myeloid cells (“signal one” in MDSC generation) and that prolonged sequestration of TLR4-Ab in myeloid cells (both systemically and in resident myeloid cells in the islet) provides the second signal for these cells to become differentiated/activated MDSCs. Furthermore, we have observed that acutely diabetic mice are most effectively treated when they receive at least two doses of TLR4-Ab. The initial TLR4-Ab dose may increase mobilization of immature MDSCs, while subsequent doses prolong endosomal sequestration and enhance generation and maintenance of mature, suppressive MDSCs, which promote diabetogenic T-cell suppression. TLR4-Ab would therefore be critical for maintaining this population of MDSCs rather than simply initiating a transient recruitment of suppressor cells. TLR4-Ab is attractive as a potential therapeutic because it reverses acute T1D, even in advanced disease (10). Part of the therapeutic efficacy is likely the long half-life of the antibody that can perpetuate the second signal long enough to generate a substantial population of MDSCs. The RNA-seq studies of purified CD11b+ cells treated with TLR4-Ab provide strong unbiased evidence that these cells are bone marrow derived. In addition, genes upregulated by TLR4-Ab involve both myeloid/granulocytic- and monocyte-associated gene subsets and pathways and are associated with genes upregulated in myeloid processes, such as myeloid leukemia and sepsis. Of note, the gene families associated with the upregulated genes include several novel families, such as kinesins, flotillins, and formyl peptide receptors. Understanding the role of such genes may provide novel mechanisms of MDSC action, which we will pursue in future studies.
A limitation of our study is that we did not study mechanisms of TLR4-Ab in NOD APCs compared with nondiabetic strains (e.g., B6) or congenic NOD mice protected from T1D. NOD macrophages are known to differ from other strains in important ways that could affect TLR4 receptor processing and retention in the endosome; some of these mechanisms might contribute to the MDSC formation and T1D reversals seen here (47–49). These issues will need to be addressed in future studies.
Overall, these studies have greatly enhanced our understanding of the role of TLR4-Ab in T1D reversal and identified a novel role for MDSCs in reinstating immune tolerance during acute T1D. Remedying the aberrant adaptive response by initiating profound changes to the underlying innate immune response is a promising approach for treating acute T1D and possibly other autoimmune diseases.
K.C.S.L. and K.K. contributed equally.
This article contains supplementary material online at https://doi.org/10.2337/figshare.17185826.
Article Information
Funding. This work was supported by National Institutes of Health grant 1R21-AI-120084-01A1 (W.M.R.).
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. K.C.S.L., K.K., Y.W., K.J.B., D.A., C.P., H.T., L.S.H., and B.J.A. performed experiments, analyzed data, created figures, and wrote up the results. A.B.H. and W.M.R. planned the experiments, analyzed data, and wrote 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.