Epigenetic Regulation of TLR4 in Diabetic Macrophages Modulates Immunometabolism and Wound Repair

The failure of wound healing in type 2 diabetes (T2D) represents the most common cause of amputation in the United States and has an associated 5-y mortality rate of 50% (1, 2). Chronic dysregulated inflammation is a hallmark of diabetic wounds and prevents tissue repair. Although our understanding of the pathophysiology of wound healing remains incomplete, it is clear that macrophage plasticity, allowing the transition of macrophages from an inflammatory to a reparative phenotype, is critical for normal wound healing. The molecular mechanisms that program and sustain these macrophage phenotypes in wounds have not been completely identified.

Wound repair is a complex process that occurs in overlapping stages of coagulation, inflammation, proliferation, and remodeling (2, 3). During the inflammatory phase, macrophage plasticity is essential for the repair and remodeling of wounds. Specifically, infiltrating blood monocyte–derived macrophages (monocyte/macrophages) are critical for the initial inflammatory phase of wound healing (4–6). Blood monocytes originate from macrophage–dendritic cell (DC) precursors in the bone marrow and ultimately differentiate into macrophages and DCs in the tissues (7, 8). Although there is some literature on the role of resident macrophages (F4/80⁺) and DCs in wound healing, there is growing evidence that infiltrating monocyte/macrophages provide the mandatory drive for acute inflammation, recruiting additional leukocytes, and promoting tissue/pathogen destruction (4, 6, 9, 10). After this early inflammatory phase, macrophages undergo a phenotypic switch producing TGF-β, IL-10, and other mediators important in the transition from the inflammatory to the proliferative phase of wound healing. The predominance of these phenotypically distinct macrophages at specific times during healing facilitates the development of a tailored macrophage-dependent response. Prior studies in diabetic murine models have demonstrated that the proinflammatory–to–anti-inflammatory macrophage phenotype switch is impaired, resulting in a persistent hyperinflammatory macrophage phenotype (11–14). Thus, the examination of the molecular mechanisms underlying macrophage plasticity in wounds is necessary to address the pathology seen in diabetes.

Epigenetic regulation of gene expression plays a major role in the phenotype and function of immune cells in both normal and pathologic conditions by controlling downstream protein expression patterns (13–16). We and others have shown that histone methylation regulates immune-mediator expression in in vitro and in vivo macrophages (17, 18). Histone methylation is important in maintaining active or suppressed gene expression depending on the specific methylation site. Methylation of lysine 4 (K4) on histone 3 (H3) keeps the chromatin in a confirmation such that the promoter for specific genes is available for transcription, and thus genes are actively transcribed (17). Histone H3K4 can be methylated by several different members of the SET domain-containing family of methyltransferase. In particular, the histone methyltransferase, mixed-lineage leukemia 1 (MLL1), promotes expression of inflammatory genes in a NF-κB–dependent manner (19–21). We have recently identified that MLL1 regulates macrophage cytokine expression; however, the role of MLL1 in regulating upstream signaling pathways in macrophages and inflammatory immune cells remains poorly defined (14).

One upstream receptor-signaling pathway that is instrumental in the regulation of innate immunity, specifically macrophages, is the TLR4 signaling pathway. TLRs are a family of evolutionarily conserved receptors, which have a key role in host defense by regulating both innate and adaptive immune responses (22). TLR2 recognizes the peptidoglycan and lipopeptide in the cell walls of Gram-positive bacteria, whereas TLR4 recognizes LPS, which is an integral component of the outer membranes of Gram-negative bacteria. Importantly, Gram-negative bacteria are common organisms found in diabetic wounds (23). Further, the TLR4 receptor is also activated by other ligands, such as saturated fatty acids, which are abundantly presented in diabetic patients (24). Recent studies suggest that TLR4 plays an important role in sterile inflammation, tissue repair, and response to a variety of injuries (15, 25). Similarly within diabetes, TLR4 expression and signaling are significantly increased in diabetic patients and db/db mice (26). Yet the mechanism of this hyperresponsiveness to TLR4 stimulation in diabetic inflammatory cells remains undefined. Further, despite the importance of TLR4 in the regulation of cytokines, there remains a paucity of data on the role of TLR4 in cutaneous wound healing. The limited literature focuses primarily on early wound healing in keratinocytes (27), and thus the in vivo role of TLR4 in myeloid cells during the course of healing remains unknown.

Given the importance of TLR4 on immune cell function, particularly macrophage function, we investigated the role of MLL1 in regulating TLR4 expression in diabetic wound macrophages. In this study, we show TLR4 expression is significantly elevated in macrophages in a murine model of diet-induced insulin resistance resulting in altered inflammation and metabolism. Further, we demonstrate by using a myeloid-specific MLL1 knockout murine model (Mll1^f/fLyz2^Cre+) that this increase in Tlr4 expression is driven by the epigenetic enzyme MLL1 and its corresponding H3K4 trimethylation (H3K4me3) on the Tlr4 promoter in diabetic bone marrow and wound myeloid cells. Last, genetic depletion (Tlr4^−/− + diet-induced obesity [DIO]) or small molecule pharmacological inhibition (TAK-242) of TLR4 as well as myeloid-specific TLR4 deficiency (Tlr4^f/fLyz2^Cre+) decreased the inflammatory macrophage response and improved diabetic wound healing. Taken together, our findings suggest that MLL1 regulates myeloid-specific Tlr4 expression and renders diabetic macrophages hyperinflammatory in response to the TLR4 pathway. Importantly, we demonstrate that TLR4 signaling plays an integral role in prolonged myeloid cell–mediated inflammation during aberrant diabetic wound repair. This work identifies practical therapeutic targets for abrogating dysregulated inflammation in diabetic wounds.

Mice were maintained in the University of Michigan pathogen-free animal facility, and all protocols were approved by and in accordance with the guidelines established by the Institutional Animal Care and Use Committee (University Committee on Use and Care of Animals). Male C57BL/6 (Tlr4^+/+) and Tlr4^−/− mice purchased from The Jackson Laboratory (Bar Harbor, ME) were maintained on a normal chow diet (13.5% kcal fat; LabDiet) or high-fat diet (60% kcal fat; Research Diets) for 12 wk to generate the DIO model of glucose intolerance/insulin resistance. Mice with the Mll1 or Tlr4 gene deleted in myeloid cells were generated by mating Mll1^f/f (16) or Tlr4^f/f (kind gift from Timothy Billiar, University of Pittsburgh) mice with LysM-Cre mice (The Jackson Laboratory). Of note, only male mice were used for all studies as female mice fail to develop glucose intolerance/insulin resistance following high-fat diet administration. Animals underwent all procedures at 20–24 wk of age. Body weights were determined prior to experimentation.

All experiments using human samples were approved by the Institutional Review Board at the University of Michigan and were conducted in accordance with the principles in the Declaration of Helsinki. Briefly, wounds were isolated from male age-matched patients with or without T2D who were undergoing amputation for medical reasons. For description of patient cohort, see Supplemental Table I. Comorbid conditions were not statistically different between the groups. Wounds were obtained from the lateral edge of wound specimens using an 8-mm punch biopsy tool.

Before wounding, mice were anesthetized, hair was removed with Veet (Reckitt Benckiser), and skin was cleaned with sterile water. Full-thickness back wounds were created by 4-mm punch biopsy, as previously described (13). Initial wound surface area was recorded, and digital photographs were obtained daily using an Olympus digital camera. Photographs contained an internal scale to allow for standard measurement calibration. Wound area was quantified using ImageJ software (National Institutes of Health, Bethesda, MD) and was expressed as the percentage of original wound size over time.

TAK-242 (13871; Cayman Chemical) was dissolved in DMSO according to manufacture instructions and stored at −20°C until use. For in vivo studies, the dissolved TAK-242 was prepared daily and injected i.p. (3 mg/kg body weight). TAK-242 injections started 2 d prior to creation of s.c. wound as described above and continued daily throughout wound healing experiment.

Following sacrifice, wounds were collected from the backs of the mice post mortem following CO₂ asphyxiation using a 6-mm wound biopsy. Sharp scissors were used to excise the full-thickness dermis with a 1- to 2-mm margin around the wound, ensuring collection of granulation tissue, and wounds were placed in RPMI 1640. Wounds were then carefully minced with sharp scissors and digested by incubating in a 50 mg/ml Liberase Thermolysin Medium (Roche) and 20 U/ml DNaseI (Sigma-Aldrich) solution. Wound cell suspensions were then gently plunged and filtered through a 100-μm filter to yield a single-cell suspension. Cells were then either sorted via MACS for RNA studies or cultured ex vivo for application of GolgiStop and subsequent staining for intracellular flow cytometry (28).

Wounds were digested as described above. Single-cell suspensions were incubated with FITC-labeled anti-CD3, anti-CD19, and anti-Ly6G (BioLegend) followed by anti-FITC microbeads (Miltenyi Biotec). Flow through was then incubated with anti-CD11b microbeads (Miltenyi Biotec) to isolate the nonneutrophil, nonlymphocyte, CD11b⁺ cells. Cells were saved in TRIzol (Invitrogen) for quantitative RT-PCR analyses. For human monocyte isolation, peripheral blood was collected and subjected to RBC lysis and Ficoll separation (GE Healthcare). Cell suspensions were then treated with anti-human CD14 microbeads. Magnetic separation yielded 95% purity by flow cytometry.

Whole wounds were excised from humans using a 6- to 8-mm punch biopsy. Wound sections were fixed in 10% formalin overnight before embedding in paraffin. Five-micromolar sections were stained with Masson trichrome for evaluation of reepithelialization, granulation, and collagen deposition. For immunohistochemistry, paraffin-embedded tissue sections were heated at 60°C for 30 min, deparaffinized, and rehydrated. Slides were placed in Ag retrieval buffer (pH 9.0) and heated at 95°C for 20 min in a hot water bath. After cooling, slides were treated with 3% H₂O₂ (5 min) and blocked using 10% goat serum (30 min). Overnight incubation (4°C) was then performed using first Ab at a working concentration. Slides were then washed and treated with secondary Ab, peroxidase (30 min), and diaminobenzidine substrate. Abs used were human anti-TLR4 (catalog no. AF1478, 2.5 μg/ml; Thermo Fisher Scientific). Images were quantified via ImageScope software and Image J at ×20 magnification. For immunofluorescence, tissue sections were heated at 65°C to remove parafilm and clarified in a clearing agent. Following rehydration in an ethanol gradient, Ag retrieval was performed using a citrate buffer (pH 6.0). After blocking with serum at room temperature for 30 min, the slides were incubated with the primary Abs, rabbit anti-human TLR4 (1:50, ab13556; Abcam), and mouse anti-human CD163 (1:200, NCL-L-CD163; Leica) at 4°C overnight. The sections were washed in PBS containing 0.05% Tween 20 and incubated with the secondary Abs, TRITC-conjugated donkey anti-rabbit IgG (1:50, 711-025-152; Immunoresearch) and AF488-conjugated donkey anti-mouse IgG (1:100, 715-545-151; Immunoresearch), at room temperature for 30 min. After a second wash, the slides were mounted with DAPI mounting medium and coverslipped for imaging under a fluorescence microscope (Axio Observer; Zeiss).

Chromatin immunoprecipitation (ChIP) assay was performed as described previously (16). Briefly, cells fixed in paraformaldehyde were lysed and sonicated to generate 100- to 300-bp fragments. To immunoprecipitate, samples were incubated in anti-H3K4trimethyl Ab (Abcam) or isotype control (rabbit polyclonal IgG) (Millipore) in parallel samples overnight followed by addition of Protein A Sepharose beads (Thermo Fisher Scientific). Bound DNA was eluted and purified using phenol:chloroform:isoamyl alcohol extraction and ethanol precipitation. Primers were designed using the Ensembl genome browser to search the TLR4, IL-1β, and TNF-α promoter and then National Center for Biotechnology Information Primer-BLAST was used to design primers that flank this site. TLR4 forward primer was 5′-CCAAGCCCAGAGGTCAGATG-3′, and TLR4 reverse primer was 5′-CCGTCGCAGGAGGGAAGTTA-3′. IL-1β forward primer was 5′-ACCTTTGTTCCGCACATC-3′, and IL-1β reverse primer was 5′-GGGATTATTTCCCCCTGG-3′. TNF-α forward primer was 5′-TCCTGATTGGCCCCAGATTG-3′, and TNF-α reverse primer was 5′-TAGTGGCCCTACACCTCTGT-3′.

Single-cell suspensions were collected and washed two times with cold PBS and filtered into a 96-well plate for surface staining. Cells were initially stained with Pacific Orange LIVE/DEAD fixable viability dye (Thermo Fisher Scientific) and then washed two times with cold PBS. Cells were then resuspended in Flow Buffer (PBS, FBS, NaN3, and HEPES buffer) and Fc receptors were blocked with anti-CD16/32 (BioLegend) prior to surface staining. mAbs for surface staining included the following: anti-CD3 (catalog no. 100304, 1:400 dilution; BioLegend), anti-CD19 (catalog no. 115504, 1:400 dilution; BioLegend), anti-Ter-119 (catalog no. 116204, 1:400 dilution; BioLegend), anti-NK1.1 (catalog no. 108704, 1:400 dilution), anti-Ly6G (catalog no. 127604, 1:400 dilution; BioLegend), anti-CD11b (catalog no. 101230, 1:400 dilution; BioLegend), anti-TLR4 (catalog no. 145406, 1:400 dilution; BioLegend), anti-Ter119 (catalog no.116204, 1:200 dilution; BioLegend), anti-GR.1 (catalog no. 108404, 1:200 dilution; BioLegend), anti-B220 (catalog no. 103204, 1:200 dilution; BioLegend), anti-cKit (catalog no. 105812, 1:200 dilution; BioLegend), anti-Sca (catalog no. 56-5981-82, 1:200 dilution; eBiosciences), anti-FcgRIII (catalog no. 101308, 1:200 dilution; BioLegend), anti-CD105 (catalog no. 120410, 1:1000 dilution; BioLegend), and anti-CD150 (catalog no. 115922, 1:200 dilution; BioLegend). Following surface staining, cells were washed twice, and biotinylated Abs were labeled with streptavidin allophycocyanin-Cy7 or streptavidin Pacific Orange. Next, cells were either washed and acquired for surface-only flow cytometry or were fixed with 2% formaldehyde and then washed/permeabilized with BD perm/wash buffer (BD Biosciences) for intracellular flow cytometry. After permeabilization, intracellular stains included the following: anti-H3K4me3 (abcam), anti-IL1β (mature IL-1β; BD Biosciences), and anti-TNF-α (BioLegend). For intracellular histone methylation, a secondary FITC anti-rabbit IgG (R&D Systems) was used. After washing, samples were then acquired on a 3-Laser Novocyte Flow Cytometer (Acea Biosciences). Data were analyzed using FlowJo software version 10.0 (Treestar). To verify gating and purity, all populations were routinely back-gated.

Bone marrow cells were collected by flushing mouse femurs and tibias with RPMI 1640. Bone marrow–derived macrophages (BMDMs) were cultured as previously detailed (16). On day 6, the cells were replated, and after resting for 24 h, they were incubated with or without LPS (100 ng/ml; Sigma-Aldrich [L2880] purified by phenol extraction <3% impurities) for 2–6 h, after which cells were processed for metabolite analysis or placed in TRIzol (Invitrogen) for RNA analysis. For human monocyte-derived macrophages, CD14⁺ monocytes were cultured in complete media supplemented with 50 ng/ml M-CSF (R&D Systems) for 1 wk. Adherent cells were washed and harvested with trypsin/EDTA (Lonza).

Following stimulation with media or LPS + IFN-γ to provide maximal generation “M1” macrophage phenotype, BMDM plates were rinsed and metabolism quenched, and metabolites were extracted and analyzed using a procedure described previously (29–33). Briefly, cell plates were rapidly rinsed with water and quenched with liquid nitrogen. Metabolites were extracted with 8:1:1 methanol/chloroform/water and assayed by high-performance liquid chromatography (HPLC) with quadrupole time-of-flight mass spectrometry (MS). Chromatographic separations were performed with an Agilent Technologies (Santa Clara, CA) 1200 HPLC system equipped with a Phenomenex (Torrance, CA) Luna NH2 HPLC column (1.0-mm inner bore × 150 mm long and packed with 3-μm particles) and a 1.0 × 4 mm guard column. Mobile phase A was 100% acetonitrile, and mobile phase B (MPB) was 100% 5 mM ammonium acetate adjusted to pH 9.9 with ammonium hydroxide. The gradient started at 20% MPB and ramped till 100% MPB over 20 min, was held at 100% MPB for 5 min, and then returned to 20% MPB for an additional 7 min. Detection was performed using a 6520 Agilent Technologies quadrupole time-of-flight system equipped with a dual electrospray ionization source operated in negative-ion mode for polar metabolites. Raw results are present in Supplemental Table II. For heat map analysis, metabolite levels are expressed as a fold change over wild-type control BMDM stimulated for 2 h with media.

Total RNA extraction was performed using TRIzol (Invitrogen) according to the manufacturer’s instructions. RNA was then reverse transcribed to cDNA using iScript (BioRad). PCR was performed with 2× TaqMan PCR mix using the 7500 Real-Time PCR System. Primers for Il1b (Mm00434228_m10), Tnfa (Mm00443258_m1), Tlr4 (Mm00445273_m1), Mll1 (Mm01179235_m1), and human Tlr4 (Hs00152939_m1) were purchased (Applied Biosystems). 18S was used as the internal control. Data were then analyzed relative to 18s rRNA (2Δ cycle threshold) and expressed as a fold change in comparison with control group. All samples were assayed in triplicate. The threshold cycle values were used to plot a standard curve. Data were compiled in Microsoft Excel and presented using Prism software (GraphPad).

GraphPad Prism software (RRID:SCR_002798) version 6.0 was used to analyze the data. All the data were assessed for normality and equal variance using Shapiro-Wilk test and Levene test, respectively. Unpaired two-tailed Student t test was used to determine statistical difference between two groups for normally distributed continuous variables. For comparison of multiple groups, one-way ANOVA test followed by Newman–Keuls post hoc test was used. For data with small sample size or nonnormally distributed data, nonparametric Mann-Whitney U test or Kruskal-Wallis test were used for analysis. All data are representative of at least two independent experiments as detailed in the figure legends. A p value ≤0.05 was significant.

Chronic inflammation is a hallmark of impaired diabetic wound healing. BMDMs from DIO mice exhibited increased proinflammatory inflammatory cytokine production as compared with controls, shown by increased Il1b, Tnfa, Il12, and Il23 expression (Fig. 1A, Supplemental Fig. 1A–E). This suggests that diabetic BMDMs are primed toward a proinflammatory phenotype. Recent evidence suggests that intermediary metabolism alters gene expression and thus phenotype in macrophages, providing a molecular link between metabolism and innate immunity (34–36). Given the hyperinflammatory phenotype in diabetic BMDMs, we sought to examine the variation in intermediary metabolism from DIO and control BMDMs. As such, BMDMs were stimulated and analyzed using targeted liquid chromatography–MS for ∼65 metabolites representing central carbon metabolism, including acyl-CoAs, acylcarnitines, and amino acids as well as metabolites in glycolysis, the pentose phosphate pathway, and the TCA cycle. Several glycolytic metabolites were upregulated in the DIO BMDMs, including the glucose 6-phosphate and fructose-1,6-bisphosphate (Fig. 1B). Likewise, many TCA cycle and nucleotide intermediates were increased in DIO BMDMs. Specifically, itaconate and malonyl-CoA were the two most highly upregulated TCA cycle intermediates in DIO BMDMs following stimulation. Prior metabolomic analysis by ourselves and others has demonstrated that dysregulation in itaconate production results in metabolic remodeling in macrophages toward an inflammatory state (32, 34, 37, 38). Within nucleotide metabolism, multiple nucleotide intermediates were aberrantly expressed in DIO BMDMs compared with wild-type BMDMs. Interestingly, NAD⁺, which has been recently linked to innate immune cell dysfunction and inflammation (39), was found to be significantly elevated in DIO BMDMs. For amino acids, there was less discrepancy between wild-type and DIO BMDMs compared with other metabolic pathways; however, aspartate, taurine, and alanine were three of the most highly upregulated amino acids in DIO BMDMs in both unstimulated and LPS-stimulated cells. Lastly, within this analysis, one of the most highly upregulated metabolites was S-adenosylmethionine (SAM). Additional analysis of SAM levels in normal and DIO BMDMs demonstrated that SAM was significantly upregulated in DIO BMDMs at both baseline and following stimulation (Fig. 1C).

It has recently been established that SAM is instrumental to changes in the epigenetic methylation status of histones and nucleic acids via the actions of methyltransferase enzymes (40–43). SAM serves as the universal methyl donor for these enzymes, resulting in the transfer of its methyl group to yield S-adenosylhomocysteine (SAH) and a methylated substrate. Genes that encode these enzymes are frequently altered in pathological states, such as T2D (14, 44, 45), leading to alterations in methylation and providing a link between the metabolism that regulates SAM and SAH and the epigenetic status of cells. Taken together, these data identify that DIO BMDM display increased responsiveness to inflammatory stimuli and altered metabolic pathways.

Evidence suggests that epigenetic regulation via histone methylation of gene expression plays a key role in influencing inflammatory phenotypes (14, 44, 45). Given the markedly elevated SAM levels in diabetic BMDMs, we examined if prolonged diet-induced insulin resistance results in alterations in histone methylation. As such, we examined several histone methylation marks associated with gene activation by flow cytometry. We found that H3K4me3 was increased in diabetic bone marrow progenitor cells (lineage⁻[CD3⁻CD19⁻Ly6G⁻], c-kit⁺/sca⁻, CD41^low, FcRγIII^low, CD105⁻, H3K4me3⁺) in comparison with controls (Fig. 2A, 2B). Histone methylation, particularly H3K4me3, influences immune cell phenotypes through the regulation of downstream inflammatory mediator expression in monocyte-derived macrophages (13–16). TLR4 is a major receptor that initiates a downstream signaling cascade that promotes inflammation, mostly through MyD88-dependent pathway and NF-κB expression. To evaluate if altered histone methylation impacted Tlr4 expression, BMDMs from DIO and control mice were isolated and analyzed for H3K4me3 on the promoter of the Tlr4 gene. We found that DIO BMDMs demonstrated increased H3K4me3 on NF-κB binding sites of the Tlr4 promoter (Fig. 2C). The H3K4me3 methylation mark maintains the chromatin in a conformation so specific genes are effectively activated. This was demonstrated with DIO BMDMs displaying upregulation of Tlr4 expression (Fig. 2D).

Previous investigations have demonstrated that the methyltransferase, MLL1, has site specificity for H3K4 (12, 30). Because H3K4me3 was increased on the Tlr4 promoter and we have previously identified MLL1 to influence macrophage phenotype (14), we examined MLL1 expression in normal diet and DIO BMDMs. MLL1 was significantly increased in DIO BMDMs compared with controls corresponding to the increased TLR4 receptor on diabetic macrophages (Fig. 2E). To evaluate the ability of MLL1 to regulate Tlr4 expression, we generated mice deficient in Mll1 in cells of the myeloid lineage with lysosomes by using the Cre-lox system. Myeloid-specific depletion of Mll1 was confirmed in vivo by examining sorted splenic monocytes from Mll1^f/fLyz2^Cre+ mice and littermate controls (Mll1^f/fLyz2^Cre-) (14). These mice were then placed on a high-fat diet to induce insulin resistance (DIO Mll1^f/fLyz2^Cre+ or DIO Mll1^f/fLyz2^Cre-). Following confirmation of hyperglycemia (data not shown), BMDMs were isolated from DIO Mll1^f/fLyz2^Cre+ mice and littermate controls to determine whether Mll1 alters Tlr4 expression. Mll1-deficient myeloid cells demonstrated a significant decrease in H3K4me3 on the Tlr4 promoter as well as a corresponding decrease in Tlr4 expression in the DIO Mll1^f/fLyz2^Cre+ compared with littermate controls but no significant decrease in expression of the majority of other TlrR (Fig. 2F, 2G, Supplemental Fig. 1F–H). Taken together, these results suggest that DIO bone marrow myeloid cells exhibit increased H3K4me3 at the Tlr4 promoter that primes cells toward increased Tlr4 receptor expression.

Proper wound healing requires the establishment of a regulated inflammatory response mediated by macrophages, and persistent macrophage inflammation results in poorly healing diabetic wounds (8, 27, 28). The mechanism(s) responsible for the persistent macrophage-inflammatory phenotype in diabetic wound repair are incompletely understood. Given the increased Tlr4 expression exhibited in DIO BMDMs, we next sought to determine if aberrant Tlr4 expression was also present in diabetic wounds. To evaluate the role of TLR4 in vivo in wounds, peripheral monocytes and wound macrophages (CD11b⁺[CD3⁻CD19⁻Ly6G⁻]) were examined for TLR4 receptors by flow cytometry. In comparison with normal diet controls, DIO peripheral blood monocytes and wound macrophages display significantly increased levels of TLR4 receptor (Fig. 3A). As a translational corollary, human wound tissue was also examined from patients with nonhealing wounds and T2D, and we found markedly increased TLR4 transcript and TLR4 on histologic assessment in wounds from T2D patients compared with controls (Fig. 3B, 3C).

To evaluate if the increased TLR4 in wound myeloid cells is due to epigenetic regulation of the Tlr4 gene in vivo consistent with that seen in vitro, we examined several histone methylation marks associated with gene activation. We sorted macrophages from DIO and control mice on day 3 postwounding and found that H3K4me3 was significantly increased on the Tlr4 promoter in DIO wound myeloid cells in comparison with controls, resulting in a marked upregulation of Tlr4 mRNA expression (Fig. 3D, 3E). Because the methyltransferase, MLL1, specifically methylates H3K4 (12, 30), we examined the expression of Mll1 in wound macrophages and found it significantly increased at day 3 following tissue injury, which corresponds to the increased TLR4 levels (Fig. 3F). Lastly, to determine whether Mll1 alters Tlr4 expression, wound macrophages were isolated on day 5 from DIO Mll1^f/fLyz2^Cre+ mice and littermate controls. Mll1-deficient (Mll1^f/fLyz2^Cre+) wound macrophages demonstrated a significant decrease in H3K4me3 on the Tlr4 promoter and a corresponding decrease in Tlr4 expression compared with littermate controls (Fig. 3G, 3H). Additionally, the macrophage-specific depletion of Mll1 resulted in a significant reduction of H3K4me3 on the promoters of inflammatory cytokines with a reciprocal increase in anti-inflammatory gene expression (Supplemental Fig. 2A–D). These data suggest that TLR4 is significantly upregulated in human and murine diabetic wound tissue. Further, MLL1-derived H3K4me3 methylation increases Tlr4 gene expression in murine wound myeloid cells, and that may control, at least in part, the increased inflammatory response of diabetic myeloid cells seen during tissue repair.

Given the increased TLR4 expression in diabetic bone marrow and wound macrophages, we examined if TLR4 expression was detrimental for cutaneous wound repair in diabetes. Mice with a TLR4 deficiency (Tlr4^−/−) were placed on a high-fat diet for 12 wk to induce insulin resistance and diet-induced obesity. We then wounded these mice and monitored the course of wound healing daily. DIO mice with TLR4 deficiency (DIO Tlr4^−/−) demonstrated improved healing throughout the entire wound course compared with controls (Fig. 4A). Because DIO Tlr4^−/− showed improved wound healing, we examined if this was due to changes in wound macrophage phenotype. This is important as it is well established that regulated inflammation is necessary for tissue repair (33, 34). Wound macrophages (CD11b⁺[CD3⁻CD19⁻Ly6G⁻]) were thereby isolated on day 3 from normal diet and DIO mice with or without a TLR4 deficiency. Consistent with previous reports (9, 13), diabetic wound macrophages had increased expression of proinflammatory cytokines. However, this diabetic proinflammatory phenotype was negated in DIO Tlr4^−/− mice, indicated by a reduction in inflammatory cytokines and increased anti-inflammatory cytokine expression (Fig. 4B, Supplemental Fig. 2E–G). Genetic depletion of Tlr4 did not affect histone methylation on the Tlr4 promoter (Supplemental Fig. 2H). To further confirm the detrimental role of the TLR4 overexpression in diabetic wound macrophages, we performed pharmacological inhibition of the TLR4 pathway in DIO mice and examined wound healing. Diabetic mice were treated with daily injections of either a TLR4 inhibitor, TAK-242 (3 mg/kg), or PBS control starting 2 d prior to wound creation and continued daily throughout the wound healing course. TAK-242 administration markedly improved diabetic wound healing as well as decreased wound macrophage inflammation (Fig. 4C, 4D). Lastly, to confirm the importance of myeloid-specific TLR4 in cutaneous wound healing, we generated mice deficient in Tlr4 in cells of the myeloid lineage with lysosomes (monocytes, macrophages, granulocytes) by using the Cre-lox system (Tlr4^f/fLyz2^Cre+). These Tlr4^f/fLyz2^Cre+ mice and littermate controls (Tlr4^f/fLyz2^Cre-) were then placed on a high-fat diet to induce insulin resistance/glucose intolerance. Following confirmation of glucose intolerance (data not shown), wounds were created in DIO mice with Tlr4^f/fLyz2^Cre+ (DIO Tlr4^f/fLyz2^Cre+) mice and controls (DIO Tlr4^f/fLyz2^Cre-), and wound closure was analyzed daily. Wound closure was markedly improved in DIO Tlr4^f/fLyz2^Cre+ mice compared with controls (Fig. 4E). These findings suggest that upregulation of the TLR4 signaling pathway in diabetic wound myeloid cells is detrimental to wound closure, and local inhibition of the pathway may improve diabetic tissue repair (Fig. 5).

It is well established that macrophages drive increased inflammation in obesity and T2D, contributing to the chronic inflammation seen in diabetic wounds; however, the etiology of this increased inflammatory state is unclear (13, 28, 46–48). In this study, we identify that TLR4 expression is significantly elevated in macrophages in human diabetic patients and a murine model of diet-induced insulin resistance, resulting in altered inflammation and metabolism. This increased TLR4 receptor expression is in part due to increased expression of the histone methyltransferase MLL1 in DIO macrophages and its resulting H3K4me3 on the Tlr4 promoter. Myeloid-specific deletion of MLL1 (Mll1^f/fLyz2^Cre+) resulted in significant reduction in H3K4me3 on the Tlr4 promoter and decreased Tlr4 expression. Diabetic wound healing was improved with either genetic depletion (Tlr4^−/− + DIO) or pharmacological inhibition (TAK-242) of TLR4 as well as myeloid-specific TLR4 deficiency (Tlr4^f/fLyz2^Cre+), resulting in a reduction in macrophage-mediated inflammation. Taken together, our findings suggest that MLL1 regulates Tlr4 expression in diabetic myeloid cells and that TLR4 signaling plays an integral role in prolonged macrophage-mediated inflammation during wound repair (Fig. 5). Further, these findings define a potential therapeutic target to correct impairments in the inflammatory program in diabetic wound macrophages that contribute to dysregulated inflammation.

The role of TLR4 in wound healing has previously been investigated in multiple disease states. Within the context of thermal burn injury, TLR4 signaling provides an important role in leukocyte adhesion and cytokine release (42). Further, upregulation of the TLR4 is known to be detrimental following renal, cardiac, or cerebral ischemia reperfusion (49–51). Within the context of nonpathologic wound healing, previous work demonstrates that wounds in Tlr2^−/−, Tlr4^−/−, and double-knockout Tlr2^−/−/Tlr4^−/− mice exhibit attenuated healing and decreased global wound Tgfβ and Ccl5 expression relative to wild-type animals (52). We recently expanded upon this understanding by demonstrating that in nonpathologic wound healing, TLR4-MyD88 signaling is as important for regulated tissue repair as it is instrumental in the initial inflammatory response following tissue injury (15). In contrast to the beneficial effects of TLR4 signaling in normal tissue repair, within diabetic wound healing, prolonged signaling through the TLR4 pathway likely has a detrimental impact (43). Within murine investigations, knockdown of TLR4 in a streptozotocin model of type I diabetes resulted in decreased circulating chemokines (53). Further investigations using a diet-induced obesity model demonstrated that TLR4 deficiency reduces adipose tissue inflammation concomitant with a shift in adipose tissue macrophage polarization toward an alternatively activated state (54). The importance of TLR4 regulation and signaling likely has relevance to other secondary complications of diabetes as recent studies have shown that overexpression of TLR4 in the human diabetic kidney correlates with CD68⁺ macrophage cell infiltration, suggesting a possible role for TLR4 in mediating monocyte/macrophage recruitment and tubulointerstitial inflammation in diabetic nephropathy (55).

Although these prior studies have provided insight on the role of TLR4 in diabetes and wound healing, they have been limited by several factors. First, in the study by Dasu et al. (56), a murine model of type 1 diabetes, the streptozotocin model is used; however, this model does not recapitulate the clinical characteristics observed in T2D. In fact, animals actually lose weight during the streptozotocin-induced model, which is contrast to the obesity-induced T2D model, and hence, this could impact wound healing physiology (57). Additionally, the streptozotocin-induced type 1 diabetes model has previously been demonstrated to alter immune function separately from the induction of hyperglycemia (58). Thereby any conclusions on the role of innate immune cell biology in diabetic wound healing may be clouded by the chemical effect of streptozotocin. Separately, multiple studies investigating the relation between TLR4 deficiency and diabetes analyzed either adipose tissue macrophages (54) or peritoneal macrophages (53) to draw their conclusions. Macrophage phenotypes can vary depending on their environmental niche, and thereby drawing correlations about diabetic wound macrophages from other macrophage populations may be inadequate (59). Last, prior studies that have investigated the impact of TLR4 deficiency on diabetes and wound healing (54, 58) have mostly been observational, whereas wound healing is observed in a whole-body TLR4-deficient murine model and not an in vivo cell-specific depletion model. Given that TLR4 can be expressed in multiple cell types, including inflammatory cells, keratinocytes, and fibroblasts, it is important to identify which cell type is most impacted by TLR4 deficiency and, hence, may be the most influential in diabetic wound healing (27). Because many of these earlier studies failed to identify regulators of TLR4 expression in diabetic macrophages, clinical translation will depend on identifying the mechanisms behind these alterations in TLR4 expression in diabetic wound macrophage. In our study, we use a myeloid-specific TLR4-deficient murine strain to demonstrate in vivo improvement in diabetic wound healing and identify the mechanisms that regulate TLR4 expression in diabetic wound macrophages. We demonstrate that epigenetic upregulation of myeloid-specific TLR4 signaling drives macrophages toward increased inflammatory cytokines and altered metabolism.

The current study supports the theory that the diabetic milieu alters immune cell phenotypes through epigenetics. Accumulating evidence suggests that epigenetic regulation of gene expression influences immune cell phenotypes in both disease states as well as during the normal response to injury (8, 26, 31). Further, a link between metabolic derangements and the epigenetic status of cellular pathways has recently been demonstrated. Specifically, Mentch et al. (42) show that modulation of methionine metabolism regulates SAM and SAH levels to drive specific histone methylation at H3K4, thereby affecting gene expression. Hence, we demonstrate that diabetic macrophages display increased SAM levels and MLL-mediated H3K4me3 of the TLR4 promoter, resulting in upregulated TLR4 expression in diabetic wound macrophages. The dependence of TLR4 on histone methylation via MLL1 was further supported when analyzing mice with myeloid-specific MLL1 depletion (Mll1^f/fLyz2^Cre+) in which wound macrophages from these mice showed significantly decreased Tlr4 expression. The dynamic epigenetic regulation of TLR4 is important as previous studies have shown that immune cell phenotypes are continuing to evolve during the course of wound healing and aberrances in this process can lead to delayed tissue repair (40). Additional studies demonstrate that epigenetic regulation of other TLRs occur in diabetic wound healing, such as TLR2 in which altered CpG promoter methylation correlated with diabetic foot ulcer severity (60). However, to date, no studies have shown a role of epigenetic modification of the TLR4 promoter as a mechanism for regulating macrophages in diabetes and diabetic wound repair.

Although this study produces insight into the mechanism(s) behind diabetic myeloid-mediated inflammation in cutaneous wound healing, some limitations must be addressed. Myeloid cells play an important role in tissue repair following injury; however, there is evidence that TLR4 is also expressed in keratinocytes, fibroblasts, and B cells (61, 62). Additionally, although H3K4me3 suggests a potential mechanism for increased TLR4 expression in diabetic wound macrophages, we acknowledge that other epigenetic modifications may play a role in macrophage Tlr4 expression and downstream NF-κB mediate cytokine production. Indeed, other epigenetic enzymes have been shown to play a role in aberrant myeloid cell function in pathological states (13, 14). Thus, further studies assessing the role of other specific epigenetic enzymes in the regulation of TLR4 signaling and other pathways in macrophage-mediated inflammation would be useful.

In summary, we have established that alterations in MLL1-mediated H3K4me3 on the Tlr4 promoter in macrophages are instrumental in the dysregulated inflammation and impaired wound healing in diabetes. Further, we have shown that elevated TLR4 levels in diabetic myeloid cells alter macrophage metabolism through hyperresponsiveness to TLR4 ligands. These findings suggest that MLL1 plays a significant role in dictating wound macrophage phenotype; furthermore, it may have significant relevance to macrophage-mediated inflammation in other secondary complications of diabetes (63–65). Pharmacological inhibition of the TLR4 pathway may be a reasonable therapeutic strategy for regulating the inflammatory response in diabetic repair.

We thank Robin Kunkel for assistance with the graphical illustrations.

The authors have no financial conflicts of interest.