It has been well established that the presence of diabetes is accompanied by a chronic inflammatory state promoting various diabetes-associated complications. One potential driver of this enhanced inflammatory state in patients with diabetes is hyperglycemia. Even after blood glucose control is achieved, diabetes-associated complications persist, suggesting the presence of a “hyperglycemic memory.” Innate immune cells, critically involved in various complications associated with diabetes, can build nonspecific, immunological memory (trained immunity) via epigenetic regulation. We examine the potential involvement of hyperglycemia-induced trained immunity in promoting inflammation. Our results show that hyperglycemia induces a trained phenotype in vivo in mice and in vitro in human monocytes, representative by an increased TNF-α secretion after ex vivo stimulation with LPS. These effects were largely mediated by epigenetic changes controlled by the mixed lineage leukemia (MLL) family because treatment with the MLL inhibitor menin-MLL during the process of trained immunity acquisition repressed the proinflammatory phenotype. Collectively, our results identify a novel link between hyperglycemia and inflammation in innate immune cells that might explain the increased proinflammatory state during diabetes potentially contributing to the development of various diabetes-associated complications.

This article is featured in Top Reads, p.665

It has been well established that diabetes is accompanied by a state of chronic low-grade inflammation contributing to the development of various complications, including cardiovascular diseases (CVDs) and an increased risk for infections (1, 2). One of the important factors driving the development of diabetes-associated complications is hyperglycemia (35). Importantly, even transient periods of hyperglycemia increase the risk for diabetes-associated complications. This phenomenon has been termed “hyperglycemic memory” (6) and exemplifies the persistent and harmful effects of hyperglycemia. Despite this observation, the molecular mechanisms linking hyperglycemia to an enhanced inflammatory status ultimately translating into an increased risk for the development of diabetes-associated complications remains unknown.

Macrophages are innate immune cells critically involved in various complications associated with diabetes. Originating primarily from circulating monocytes, macrophages facilitate inflammation by the release of proinflammatory mediators such as TNFα and IL-6 (7). Importantly, proinflammatory cytokine secretion is increased in type 1 diabetes (T1D) subjects (8, 9), indicating a link between the diabetic milieu and monocyte activation. Moreover, hyperglycemia was recently associated with monocyte recruitment and plaque infiltration, thereby promoting the development of atherosclerosis in animal models (10).

Experimental and clinical evidence underscores the ability of innate immune cells to memorize encounters with microbial stimuli and respond in a sensitized and nonspecific manner to subsequent restimulation by a process called trained immunity (11). For example, the fungal-derived cell wall component β-glucan induces a macrophage phenotype bearing the capacity to respond to secondary infection with augmented cytokine secretion. The persistent proinflammatory phenotype is supported not only by profound metabolic changes such as the activation of aerobic glycolysis (12) but also by epigenetic reprogramming associated with distinct histone modifications.

In this current study, we test our hypothesis that hyperglycemia augments a persistent proinflammatory-trained immunity phenotype in diabetes. To test this hypothesis, we employed the experimental model of trained immunity. First, we investigated whether in vivo exposure to hyperglycemia could induce a proinflammatory phenotype in murine bone marrow–derived macrophages (BMDMs). Similarly, we tested whether the preincubation of monocytes in high glucose conditions in vitro could elicit a proinflammatory phenotype in human monocytes/macrophages. Using microarray analyses, we set out to identify transcriptional changes indicative of trained immunity in monocytes isolated from patients with T1D. We aim to support the concept that a proinflammatory phenotype induced by hyperglycemia contributes to the development of various complications in patients with diabetes.

Six-week-old male C57BL/6 mice were obtained from Charles River Laboratories. Mice were housed under standard conditions in conventional cages in a temperature-controlled room with a 12-h light/dark cycle and ad libitum access to food and water. Hyperglycemia was induced by injecting mice on five consecutive days with streptozotocin (STZ) (60 mg/kg, i.p.) or vehicle. After 15 d, STZ-treated mice were excluded that had a nonfasted blood glucose level below 15 mM. After 45 d, 10 mice per group were killed, and bone marrow was isolated and used for culturing of BMDMs. From another 10 mice that either received STZ treatment to induce hyperglycemia or PBS as a control, bone marrow was isolated and used for bone marrow transplantation into C57BL/6 mice. All experiments in this study were carried out in strict accordance with the recommendations in the Dutch law on Animal Experiments and the Federation of European Laboratory Animal Science Associations regulations. The protocol was approved by the Ethics Committee on Animal Experiments of the Radboud University Medical Center.

Bone marrow cells from normoglycemic (n = 6) and hyperglycemic (STZ-treated, n = 4) mice were used for transplantation into recipient mice (n = 12 animals for normo- and hyperglycemic bone marrow). To induce bone marrow aplasia, male C57BL/6 recipient mice (8 wk of age) were irradiated to a dose of 8 Gy at a dosage rate of 3.8 Gy/min (XRAD 320ix; Precision X-Ray). The day thereafter, irradiated recipient C57BL/6 mice received an i.v. injection via the tail vein with 1.2 × 106 bone marrow cells isolated from normoglycemic or hyperglycemic donor mice mixed with 0.3 × 106 freshly isolated splenic cells from Rag1−/− (also C57BL/6J background) male mice. All mice received water containing antibiotics (0.13 mg/kg/d ciprofloxacin, 0.105 mg/kg/d polymyxin B, and 0.15 mg/kg/d amphotericin B) from 1 d before until 4 wk after bone marrow transplantation. After 56 d, the mice were in vivo stimulated with 5 μg/g body weight LPS. After 90 min, the production of TNF-α in blood was measured using Luminex (Bio-Rad Laboratories).

For the murine model of trained immunity (Fig. 1A), bone marrow cells that were in vivo exposed to hyperglycemia or normoglycemia were extracted from the bone marrow with sterile PBS after 45 d. The cells obtained were directly differentiated in DMEM (Thermo Fisher Scientific) containing 1% penicillin/streptomycin (Sigma-Aldrich) and 30% (vol/vol) L929 medium in 48-well plates (per mouse eight wells × 250,000 cells/well). After 6 d, BMDMs were stimulated with 10 ng/ml LPS for 24 h, and supernatants were stored for cytokine analysis. Per group, n = 10 mice were used in this experiment.

For in vitro experiments, buffy coats from the healthy donor were obtained after they provided written informed consent (Sanquin Blood Bank). After PBMCs isolation as stated above, the monocyte fraction was further purified using centrifugation over hyperosmotic Percoll solution (Sigma-Aldrich).

For the in vitro model of trained immunity (Fig. 1A), 1 × 106 monocytes in 100 μl cell suspension were seeded to flat-bottom 96-well plates in Dutch-modified RPMI culture medium (no glucose; Invitrogen) supplemented with 10 μg/ml gentamicin, 20 mM HEPES, and 10 mM pyruvate. Monocyte function following a transient episode of high glucose was tested using a previously described in vitro model of trained immunity (12). Briefly, monocytes were incubated with culture medium containing 10% pooled human serum and normal (5 mM) or high (25 mM) glucose concentration or in combination with 1 μg/ml β-glucan for 24 h. Thereafter, supernatants were taken for lactate analysis, and cells were washed and differentiated for 6 d in culture medium containing 10% serum and 6 mM glucose. Then, cells were stimulated with 10 ng/ml LPS for 24 h, and supernatants were stored for cytokine analysis. For experiments with epigenetic inhibitors, 50 mM WD (Trp-Asp) repeat domain/mixed lineage leukemia (MLL) interaction inhibitor (MM102; Bio-Techne/R&D Systems) and 50 mM menin-MLL inhibitor (MI-2) (Selleck Chemicals) were used during the first 24 h of training. In vitro experiments were performed in individual experiments with three to five donors per experiment. In total, results of n = 26 donors were included in the study.

Cytokine concentrations in the supernatants were determined by ELISA using commercially available kits for TNF-α and IL-6 (R&D Systems) or using available Luminex kits for mouse TNF-α, mouse IL-6, and keratinocyte-derived chemokine (KC) (Merck). Lactate production after 24 h was measured using Lactate Fluorometric Assay Kit (BioVision).

Blood from T1D patients with poor glycemic control and healthy control subjects was taken after written informed consent. Briefly, PBMCs were isolated by density centrifugation over Ficoll-Paque (GE Healthcare). Then, CD14+ monocytes were selected using positive magnetic-bead labeling according to the manufacturer’s instructions (MACS Miltenyi Biotec). Unstimulated RNA was purified from human CD14+ monocytes using TRIzol reagent (Invitrogen), followed by additional round purification with RNeasy Mini Kit columns (QIAGEN). RNA quality was assessed using RNA 6000 nanochips on Agilent 20100 Bioanalyzer (Agilent Technologies). Purified RNA (100 ng) was labeled with WT PLUS Reagent Kit (Affymetrix, Thermo Fisher Scientific) and hybridized to Human Gene 2.1 ST Array Plate (Affymetrix). Hybridization, washing, and scanning were carried out on a GeneTitan platform (Affymetrix), according to the manufacturer’s instructions. Arrays were normalized using the robust multiarray average method (13, 14). Probe sets were defined according to the method of Dai et al. (15). In this method, probes are assigned to Entrez identifiers as a unique gene identifier. The p values were calculated using an intensity-based moderated t-statistic. An overview of the characteristics of the study participants is given in Table I. The microarray dataset has been submitted to the Gene Expression Omnibus under accession number GSE164338.

Data on innate immune training in human monocytes are shown as mean ± SEM or are expressed as fold changes and analyzed using Wilcoxon matched-pairs signed-rank test. Data on in vitro training in murine macrophages are shown as mean ± SEM and analyzed using Mann–Whitney U test. A two-sided p value <0.05 was considered statistically significant. All data were analyzed using Prism version 5.0 (GraphPad Software, La Jolla, CA).

To investigate the effect of hyperglycemia on innate immune cells, we tested murine macrophages and human monocytes. First, to create an in vivo model of hyperglycemia, we injected STZ in C57BL/6 mice (Fig. 1A). Treatment with STZ led to plasma glucose levels of 28 mM, whereas placebo-treated mice had plasma glucose levels of 7.9 mM after 45 d (Fig. 1B). Bone marrow was isolated and used for transplantation. Our results show that upon bone marrow transplantation from hyperglycemic animals, circulating TNF-α levels were enhanced (Fig. 1C, p = 0.06) after LPS treatment compared with mice receiving bone marrow from normoglycemic animals. Similar results were obtained using an in vitro approach (Fig. 1A). In addition to transplantation, bone marrow cells from STZ-treated mice and control were extracted and differentiated toward BMDMs in vitro. Macrophages originating from hyperglycemic (STZ-treated) mice secreted more of the proinflammatory cytokines TNF-α and KC after LPS restimulation at day 6 than BMDMs originating from normoglycemic (control) mice (Fig. 1D–F). Next, an in vitro model of hyperglycemic memory was designed using primary human monocytes (Fig. 1A). Monocytes were maintained in low (5 mM) or high (25 mM) glucose conditions in vitro in the absence or presence of β-glucan, a well-known inducer of trained immunity, for 24 h. Preincubation of monocytes with high glucose led to a significant increase in TNF-α and IL-6 cytokine secretion after restimulation with LPS on day 6 as compared with monocytes that were incubated in normal glucose (Fig. 1G, 1H). Noticeably, trained immunity induced by β-glucan was apparent in monocytes preincubated in high glucose and normal glucose conditions. However, TNF-α secretion (Fig. 1G) but not IL-6 secretion (Fig. 1H) was reinforced by the presence of hyperglycemia during β-glucan training (Fig. 1G, 1H).

FIGURE 1.

Hyperglycemia induces trained immunity in mice in vivo and in humans in vitro. (A) Existence of trained immunity in vivo was tested in bone marrow cells derived from male C57BL/6 mice that either received STZ injections to render them hyperglycemic or a vehicle and were kept for a total study duration of 45 d. Trained immunity in vitro was tested in human monocytes isolated from buffy coats and cultured in the presence of normal (5 mM) or high (25 mM) glucose concentrations or in combination with β-glucan for 24 h. Bone marrow cells and monocytes were differentiated in culture medium for 6 d before the subsequent stimulation with TLR4 ligand LPS for 24 h. (B) Plasma blood levels were measured in unfasted mice at the day of sacrifice after 45 d. (C) Circulating TNF-α levels after 1.5 h of LPS treatment in mice transplanted with bone marrow from normo- or hyperglycemic mice. The secretion of TNF-α (D), IL-6 (E), and KC (F) upon restimulation with LPS or DMEM as a control was measured by ELISA (n = 10 per group). The magnitude of trained immunity for monocytes that were cultured in physiologically normal glucose concentrations or in high glucose concentrations or in combination with β-glucan is shown for (G) TNF-α and (H) IL-6 cytokine secretion, respectively (n = 20–26). Data are presented as mean ± SEM or as fold change; Wilcoxon matched-pairs signed-rank test. *p < 0.05, **p < 0.01, ***p < 0.001.

FIGURE 1.

Hyperglycemia induces trained immunity in mice in vivo and in humans in vitro. (A) Existence of trained immunity in vivo was tested in bone marrow cells derived from male C57BL/6 mice that either received STZ injections to render them hyperglycemic or a vehicle and were kept for a total study duration of 45 d. Trained immunity in vitro was tested in human monocytes isolated from buffy coats and cultured in the presence of normal (5 mM) or high (25 mM) glucose concentrations or in combination with β-glucan for 24 h. Bone marrow cells and monocytes were differentiated in culture medium for 6 d before the subsequent stimulation with TLR4 ligand LPS for 24 h. (B) Plasma blood levels were measured in unfasted mice at the day of sacrifice after 45 d. (C) Circulating TNF-α levels after 1.5 h of LPS treatment in mice transplanted with bone marrow from normo- or hyperglycemic mice. The secretion of TNF-α (D), IL-6 (E), and KC (F) upon restimulation with LPS or DMEM as a control was measured by ELISA (n = 10 per group). The magnitude of trained immunity for monocytes that were cultured in physiologically normal glucose concentrations or in high glucose concentrations or in combination with β-glucan is shown for (G) TNF-α and (H) IL-6 cytokine secretion, respectively (n = 20–26). Data are presented as mean ± SEM or as fold change; Wilcoxon matched-pairs signed-rank test. *p < 0.05, **p < 0.01, ***p < 0.001.

Close modal

We conducted a microarray analysis of CD14+ monocytes isolated from five patients with T1D and five control subjects without diabetes (Table I) to explore differential regulation of intracellular pathways related to trained immunity induced by hyperglycemia (Fig. 2). To further gauge the exclusive impact of glucose, we compared data from our T1D subjects to the gene expression profile of human monocytic THP-1 cells that had been incubated in 5.5 or 25 mM glucose for 72 h (publicly available data: GSE43925). We first set out to study genes related to glucose/energy metabolism known to control trained immunity (16, 17). Although we observed considerable variation in gene expression across donors (Fig. 3), we found a general upregulation of genes involved in glycolysis (Kyoto Encyclopedia of Genes and Genomes M00001) in monocytes from T1D, with GAPDH, GCK, and ENO3 being significantly altered (Fig. 2A). Similarly, high glucose induced a general upregulation of these genes in THP-1 cells (Fig. 2A). In contrast, genes involved in the TCA cycle (HSA 00020) were generally downregulated in monocytes of T1D subjects and THP-1 cells (Fig. 2B), significantly for PC, FH, and SDHD in monocytes of T1D subjects. In addition to energy metabolism, epigenetic pathways are known to govern the trained phenotype (16). When evaluating expression of epigenetic enzymes, we observed changes in the expression of members of the MLL gene family that encode for epigenetic enzymes, which methylate lysine 4 of H3 histone tails (H3K4). In T1D patients, KMT2B (MLL2), KMT2C (MLL3), and KMT2D (MLL4) (Fig. 2C, KMT2D; p = 0.02) were upregulated, and likewise, members of the MLL gene family were increasingly expressed in THP-1 cells incubated in high glucose (Fig. 2C). Other significantly upregulated methyltransferases include SETD1A and SETD1B.

Table I.
Characteristics of T1D subjects and healthy controls
T1DNondiabetic Control Subjects
HbA1c (mmol/mol)Age (y)HbA1c (mmol/mol)Age (y)
 77 22  28 
 93 34  30 
 66 22  29 
 64 41  47 
 68 31  35 
Average 73.6 ± 5.33 30 ± 3.65  33.8 ± 3.51 
T1DNondiabetic Control Subjects
HbA1c (mmol/mol)Age (y)HbA1c (mmol/mol)Age (y)
 77 22  28 
 93 34  30 
 66 22  29 
 64 41  47 
 68 31  35 
Average 73.6 ± 5.33 30 ± 3.65  33.8 ± 3.51 

Circulating CD14+ cells were isolated and used for microarray analysis.

FIGURE 2.

Glycolysis and MLL H3K4 methyltransferase enzymes are upregulated in CD14+ monocytes of T1D subjects and in the THP-1 cell line incubated in high glucose. Microarray analysis was performed on unstimulated CD14+ monocytes of five T1D subjects and five healthy controls, and RNA sequencing was performed on THP-1 cells (publicly available data: GSE43925, n = 3) that were incubated in normal (5.5 mM) or high (25 mM) glucose concentrations for 72 h. Heat maps showing expression of genes associated with (A) glycolysis and (B) the TCA cycle as well as genes encoding (C) H3K4 methyl writers, erasers, and readers.

FIGURE 2.

Glycolysis and MLL H3K4 methyltransferase enzymes are upregulated in CD14+ monocytes of T1D subjects and in the THP-1 cell line incubated in high glucose. Microarray analysis was performed on unstimulated CD14+ monocytes of five T1D subjects and five healthy controls, and RNA sequencing was performed on THP-1 cells (publicly available data: GSE43925, n = 3) that were incubated in normal (5.5 mM) or high (25 mM) glucose concentrations for 72 h. Heat maps showing expression of genes associated with (A) glycolysis and (B) the TCA cycle as well as genes encoding (C) H3K4 methyl writers, erasers, and readers.

Close modal
FIGURE 3.

Lactate production and epigenetic modulators are involved in hyperglycemia-induced trained immunity in human monocytes. Human monocytes were isolated from buffy coat and cultured in the presence of normal (5 mM) or high (25 mM) glucose concentrations or in combination with β-glucan for 24 h. The effect of glucose on 24-h lactate production is shown for cells stimulated with (A) RPMI and (B) β-glucan (n = 18 donors). Correlation graphs comparing lactate production and cytokine secretion are shown for training with (C) RPMI or (D) β-glucan in high glucose concentrations (n = 12–18 donors). (E) The effect of MLL gene inhibition on TNF-α secretion is shown for monocytes cultured in high glucose concentrations alone and (F) in combination with β-glucan during the first 24 h (n = 12 donors). Data are presented as mean ± SEM or as fold change; Wilcoxon matched-pairs signed-rank test. **p < 0.01, ***p < 0.001.

FIGURE 3.

Lactate production and epigenetic modulators are involved in hyperglycemia-induced trained immunity in human monocytes. Human monocytes were isolated from buffy coat and cultured in the presence of normal (5 mM) or high (25 mM) glucose concentrations or in combination with β-glucan for 24 h. The effect of glucose on 24-h lactate production is shown for cells stimulated with (A) RPMI and (B) β-glucan (n = 18 donors). Correlation graphs comparing lactate production and cytokine secretion are shown for training with (C) RPMI or (D) β-glucan in high glucose concentrations (n = 12–18 donors). (E) The effect of MLL gene inhibition on TNF-α secretion is shown for monocytes cultured in high glucose concentrations alone and (F) in combination with β-glucan during the first 24 h (n = 12 donors). Data are presented as mean ± SEM or as fold change; Wilcoxon matched-pairs signed-rank test. **p < 0.01, ***p < 0.001.

Close modal

To validate the biological relevance of our observations, we measured the concentration of lactate, the end product of glycolysis, produced by human monocytes that we exposed to 25 mM glucose for 24 h. Lactate production was not changed in cells solely exposed to a transient episode of high glucose compared with normal glucose concentrations (Fig. 3A). However, we observed that transient exposure to 25 mM of glucose for the 24-h period of training with β-glucan amplified lactate production, the end product of glycolysis, compared with cells that were trained in normal glucose concentrations (Fig. 3B). Interestingly, we observed that after transient periods of high glucose (Fig. 3C) and in combination with β-glucan (Fig. 3D), lactate production correlated with the LPS-induced secretion of TNF-α but not IL-6, indicating that early metabolic changes are relevant for the induction of trained immunity. Finally, we inhibited the MLL gene family or MLL1 gene using MI-2 and MM102, respectively, to test whether trained immunity was conferred via these epigenetic modifiers. No effects of MLL inhibition on cytokine secretion were seen in cells trained by high glucose alone (Fig. 3E). However, inhibition with MI-2 but not with MM102 reduced TNF-α secretion in cells that were exposed to a transient episode of high glucose and β-glucan (Fig. 3F).

Our results demonstrate for the first time, to our knowledge, that high glucose can induce and exacerbate trained immunity both in vivo using an animal model and in vitro using primary human monocytes. This response appears to be mediated by an upregulation in the glycolytic rate of the cells and involves epigenetic changes mediated via histone methyltransferases. Hence, innate immune cells have a metabolic memory toward glucose, leading to an enhanced inflammatory response that may ultimately translate into an increased risk for CVD during diabetes.

A more activated innate immune cell phenotype, as we have shown in this study upon exposure to high glucose, could potentially promote inflammatory-driven complications of diabetes, including CVD (18). Our results demonstrate that a transient period of high glucose, both in vivo in mice and in vitro using human monocytes, increased TNF-α cytokine secretion of innate immune cells in response to subsequent pathogenic stimulation. From a mechanistic point of view, our results demonstrate the involvement of various pathways in governing the trained phenotype. First, our transcriptome analysis revealed an increase in genes involved in glycolysis upon the presence of hyperglycemia. In line with these findings, the magnitude of trained immunity, as represented by increased TNF-α secretion, correlated with the amount of lactate produced during the first 24 h. Glycolysis, with its end product lactate, is a central hallmark of trained monocytes/macrophages (16), and therefore, it can be speculated that in the context of diabetes, changes in the environments of the cells, including high glucose or other disrupted metabolites, impact on immune cell metabolism and ultimately affect innate immune cell function.

Our results show that the hyperglycemic environment most likely induces persistent changes in the bone marrow displayed by an exacerbated proinflammatory response. In this context, Nagareddy et al. (10) were the first to show that diabetes-associated monocytosis in mice promoted the development of atherosclerosis. However, in their study, they did not investigate whether hyperglycemia impacts on the inflammatory phenotype via trained immunity. In this study, we show that a previous period of hyperglycemia also induces a persistent inflammatory bone marrow progenitor phenotype. Together, these results imply that external stimuli such as hyperglycemia are capable of altering hematopoietic stem cell phenotypes, as highlighted in the review by Chavakis et al. (19). This hematopoietic stem cell phenotype gives rise to an inflammatory monocyte population that ultimately might promote CVD. Although trained immunity has been shown to impact the development of atherosclerosis (20), more work is needed to establish the exact contribution of hyperglycemia-induced trained immunity to the development of CVD during diabetes.

One potential molecular mechanism of action translating high glucose into exacerbated immune response may involve epigenetic changes conferred by the MLL family of lysine methyltransferases. Interestingly, in a prediabetic animal model, myeloid-specific deletion of MLL1 led to a decreased H3K4me3 at the NF-κB binding site of inflammatory genes (21) Moreover, MLL gene expression was upregulated in monocytes of patients with type 2 diabetes (T2D) (21), confirming our findings on increased MLL gene expression in patients with T1D indicative of a pronounced role for MLL enzymes in conveying metabolic memory, leading to hyperglycemia-induced inflammatory responses in diabetes. Indeed, trained immunity against oxidized low-density lipoprotein (LDL), fungi, and certain vaccines is associated with the enrichment of H3K4me3 at promoters of genes encoding proinflammatory cytokines (17, 22, 23); however, the epigenetic machinery responsible for these changes has not been clearly defined. Our findings provide the impetus to further characterize the impact of MLL enzymes on the chromatin landscape of trained cells. We conducted experiments in human monocytes obtained from patients with T1D and matched healthy controls (Supplemental Table I) to further investigate the interplay between hyperglycemia and MLL gene expression. We found that there was no difference between the expression of MLL gene 1–5 in CD14+ monocytes isolated from patients with T1D as compared with healthy controls (Supplemental Fig. 1). Further, we could not detect any correlation between HbA1c levels or the duration of T1D and MLL 1–5 expression. It can be speculated that hyperglycemia, rather than upregulating methyltransferases itself, induces an increase in the target genes of MLL1-5. This has been shown for Set7, another important methyltransferase that mediates hyperglycemia-induced changes. The expression of NF-κB subunit p65, the target gene of Set7, was upregulated in normoglycemic mice exposed to a hyperglycemic episode (24). In this way, monocytes acquired a more proinflammatory phenotype. A follow-up experiment could further be carried out to measure H3K4m1/H3K4m3 methylation of promotor regions of inflammatory genes such as NF-κB in hyperglycemic cells with and without MLL inhibition.

In the current study, we showed that hyperglycemia can induce trained immunity in vivo in mice. The first study showing that trained immunity existed in vertebrates was conducted in mice that were lacking functional T and B lymphocytes. Prior exposure to a small dose of a microbial ligand protected mice from a subsequent infection with a lethal dose of Candida albicans. Mechanistically, it was shown that monocytes acquired an unspecific memory, which led to the protection of the “trained” mice in this study (25). Christ et al. (26) recently demonstrated metabolic memory toward Western-type diet (WD) of bone marrow progenitor cells of mice. They exposed animals to a WD, which led to transient hypercholesteremia and systemic inflammation (without causing hyperglycemia), causal for the induction of trained immunity. Animals then were returned to chow diet and eventually were sacrificed. In vitro differentiated and stimulated BMDMs displayed a persistent proinflammatory response as compared with BMDMs from mice that did not receive WD. Interestingly, recent findings have identified oxidized LDL as another metabolic inducer of trained immunity mediated by an upregulation of glycolytic metabolism. When oxidized LDL-treated cells were additionally exposed to 3PO, an inhibitor of the glycolytic enzyme PFKFB3, trained immunity was abolished in monocytes isolated from healthy donors (27). This clearly demonstrates that glycolysis is indispensable for trained immunity. Most likely, glycolytic inhibition would similarly prevent hyperglycemia-induced trained immunity.

Our results demonstrated that BMDMs have a metabolic memory specifically toward hyperglycemia and results from our bone marrow transplantation study suggest long-term changes in bone marrow progenitor cells possibly underlying hyperglycemic memory. In this study, we showed for the first time, to our knowledge, a molecular mechanism of monocyte/macrophage progenitor cell reprogramming by which hyperglycemic memory leads to an increased proinflammatory response that may ultimately help to explain the development of various diabetes-associated complications driven by hyperactivation of innate immune responses including CVD.

Besides hyperglycemia, other factors including insulin, duration of T1D, or fluctuations in HbA1c may have contributed to our observations. In terms of the effect of insulin on monocytes and macrophages, these innate immune cells have been shown to express most molecules relevant for insulin signaling; however, the biological role of the insulin pathway in macrophages remains largely unknown (28). In the context of trained immunity, it has been shown that insulin can activate Akt (28), which is also induced during the acquisition of the trained immunity phenotype (12). With regard to the in vitro model for trained immunity in the current study, the direct addition of insulin to the cells would therefore probably result in an even more pronounced trained immunity phenotype in human monocytes rather than reducing the observed phenotype. However, in vivo, both effects of insulin directly on monocytes/macrophages and indirectly via lowering of plasma glucose levels might lead to differential effects. With regard to the microarray performed in monocytes from patients with T1D, we cannot exclude the effects of insulin or other confounders on increased gene expression of glycolytic genes. However, combining our findings from the microarray with our results from the in vivo/in vitro experiments, it is tempting to speculate that the switch from oxidative phosphorylation to glycolysis is at least partly explained by hyperglycemia-induced trained immunity. This is also supported by the observation that a switch from oxidative phosphorylation to glycolysis in human monocytes occurred during trained immunity induced by β-glucan (12).

Multiple risk factors that predispose patients with diabetes to comorbidities are similar in T1D and T2D. The inciting factor or whether there is one inciting factor for the development of CVD remains unknown. The metabolic syndrome can partly explain the increased risk for CVD (29) in most patients with T2D (30); however, in patients with T1D in which the occurrence of the metabolic syndrome is much less prevalent, the risk for CVD is similarly increased. The linking factor in both forms of diabetes (31) is hyperglycemia. In the current study, we therefore set out to investigate hyperglycemia in T1D patients and in experimental models of hyperglycemia independently of other risk factors that predominately occur in T2D. To this end, our findings do not solely account for T1D but rather indicate which impact hyperglycemia might have on diabetes in general.

In conclusion, the present findings reveal that high glucose can lead to persistent alternations in innate immune cell metabolism and induces epigenetic alterations that lead to a proinflammatory phenotype similar to the concept of trained immunity. Our data suggest that innate immune cells have a hyperglycemic memory that is partly conferred by MLL gene family and may be the underlying mechanism of activation of innate immune response in patients with diabetes, contributing to the development of various complications.

This work was supported by a European Research Council Consolidator Grant (310372) (to M.G.N.), the Netherlands Organization for Scientific Research (a Spinoza Grant to M.G.N. and Veni Grant 91616083 to J.v.D.). R.S. is supported by a senior fellowship from the Dutch Diabetes Research Foundation (2015.82.1824). This work was also supported by the Dutch Heart Foundation (IN-CONTROL).

The online version of this article contains supplemental material.

Abbreviations used in this article:

BMDM

bone marrow–derived macrophage

CVD

cardiovascular disease

KC

keratinocyte-derived chemokine

LDL

low-density lipoprotein

MI-2

menin-MLL inhibitor

MLL

mixed lineage leukemia

STZ

streptozotocin

T1D

type 1 diabetes

T2D

type 2 diabetes

WD

Western-type diet.

1
Lopez-Candales
,
A.
,
P. M.
Hernández Burgos
,
D. F.
Hernandez-Suarez
,
D.
Harris
.
2017
.
Linking chronic inflammation with cardiovascular disease: from normal aging to the metabolic syndrome.
J. Nat. Sci.
3
: e341.
2
Schalkwijk
,
C. G.
,
D. C.
Poland
,
W.
van Dijk
,
A.
Kok
,
J. J.
Emeis
,
A. M.
Dräger
,
A.
Doni
,
V. W.
van Hinsbergh
,
C. D.
Stehouwer
.
1999
.
Plasma concentration of C-reactive protein is increased in type I diabetic patients without clinical macroangiopathy and correlates with markers of endothelial dysfunction: evidence for chronic inflammation.
Diabetologia
42
:
351
357
.
3
Goff
,
D. C.
 Jr.
,
H. C.
Gerstein
,
H. N.
Ginsberg
,
W. C.
Cushman
,
K. L.
Margolis
,
R. P.
Byington
,
J. B.
Buse
,
S.
Genuth
,
J. L.
Probstfield
,
D. G.
Simons-Morton
;
ACCORD Study Group
.
2007
.
Prevention of cardiovascular disease in persons with type 2 diabetes mellitus: current knowledge and rationale for the Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial.
Am. J. Cardiol.
99
(
12A
):
4i
20i
.
4
Nathan
,
D. M.
,
P. A.
Cleary
,
J. Y. C.
Backlund
,
S. M.
Genuth
,
J. M.
Lachin
,
T. J.
Orchard
,
P.
Raskin
,
B.
Zinman
;
Diabetes Control and Complications Trial/Epidemiology of Diabetes Interventions and Complications (DCCT/EDIC) Study Research Group
.
2005
.
Intensive diabetes treatment and cardiovascular disease in patients with type 1 diabetes.
N. Engl. J. Med.
353
:
2643
2653
.
5
Patel
,
M. B.
,
A.
Bhalla
.
2008
.
Images in clinical medicine. Paget’s disease of the mandible.
N. Engl. J. Med.
358
:
625
.
6
Holman
,
R. R.
,
S. K.
Paul
,
M. A.
Bethel
,
D. R.
Matthews
,
H. A.
Neil
.
2008
.
10-year follow-up of intensive glucose control in type 2 diabetes.
N. Engl. J. Med.
359
:
1577
1589
.
7
Moore
,
K. J.
,
F. J.
Sheedy
,
E. A.
Fisher
.
2013
.
Macrophages in atherosclerosis: a dynamic balance.
Nat. Rev. Immunol.
13
:
709
721
.
8
Devaraj
,
S.
,
N.
Glaser
,
S.
Griffen
,
J.
Wang-Polagruto
,
E.
Miguelino
,
I.
Jialal
.
2006
.
Increased monocytic activity and biomarkers of inflammation in patients with type 1 diabetes.
Diabetes
55
:
774
779
.
9
Giulietti
,
A.
,
K.
Stoffels
,
B.
Decallonne
,
L.
Overbergh
,
C.
Mathieu
.
2004
.
Monocytic expression behavior of cytokines in diabetic patients upon inflammatory stimulation.
Ann. N. Y. Acad. Sci.
1037
:
74
78
.
10
Nagareddy
,
P. R.
,
A. J.
Murphy
,
R. A.
Stirzaker
,
Y.
Hu
,
S.
Yu
,
R. G.
Miller
,
B.
Ramkhelawon
,
E.
Distel
,
M.
Westerterp
,
L. S.
Huang
, et al
.
2013
.
Hyperglycemia promotes myelopoiesis and impairs the resolution of atherosclerosis.
Cell Metab.
17
:
695
708
.
11
Netea
,
M. G.
,
J.
Quintin
,
J. W.
van der Meer
.
2011
.
Trained immunity: a memory for innate host defense.
Cell Host Microbe
9
:
355
361
.
12
Cheng
,
S. C.
,
J.
Quintin
,
R. A.
Cramer
,
K. M.
Shepardson
,
S.
Saeed
,
V.
Kumar
,
E. J.
Giamarellos-Bourboulis
,
J. H.
Martens
,
N. A.
Rao
,
A.
Aghajanirefah
, et al
.
2014
.
mTOR- and HIF-1α-mediated aerobic glycolysis as metabolic basis for trained immunity. [Published erratum appears in 2014 Science 346: aaa1503.]
Science
345
: 1250684.
13
Bolstad
,
B. M.
,
R. A.
Irizarry
,
M.
Astrand
,
T. P.
Speed
.
2003
.
A comparison of normalization methods for high density oligonucleotide array data based on variance and bias.
Bioinformatics
19
:
185
193
.
14
Irizarry
,
R. A.
,
B. M.
Bolstad
,
F.
Collin
,
L. M.
Cope
,
B.
Hobbs
,
T. P.
Speed
.
2003
.
Summaries of Affymetrix GeneChip probe level data.
Nucleic Acids Res.
31
: e15.
15
Dai
,
M.
,
P.
Wang
,
A. D.
Boyd
,
G.
Kostov
,
B.
Athey
,
E. G.
Jones
,
W. E.
Bunney
,
R. M.
Myers
,
T. P.
Speed
,
H.
Akil
, et al
.
2005
.
Evolving gene/transcript definitions significantly alter the interpretation of GeneChip data.
Nucleic Acids Res.
33
: e175.
16
Arts
,
R. J.
,
L. A.
Joosten
,
M. G.
Netea
.
2016
.
Immunometabolic circuits in trained immunity.
Semin. Immunol.
28
:
425
430
.
17
Arts
,
R. J.
,
B.
Novakovic
,
R.
Ter Horst
,
A.
Carvalho
,
S.
Bekkering
,
E.
Lachmandas
,
F.
Rodrigues
,
R.
Silvestre
,
S. C.
Cheng
,
S. Y.
Wang
, et al
.
2016
.
Glutaminolysis and fumarate accumulation integrate immunometabolic and epigenetic programs in trained immunity.
Cell Metab.
24
:
807
819
.
18
van Diepen
,
J. A.
,
K.
Thiem
,
R.
Stienstra
,
N. P.
Riksen
,
C. J.
Tack
,
M. G.
Netea
.
2016
.
Diabetes propels the risk for cardiovascular disease: sweet monocytes becoming aggressive?
Cell. Mol. Life Sci.
73
:
4675
4684
.
19
Chavakis
,
T.
,
I.
Mitroulis
,
G.
Hajishengallis
.
2019
.
Hematopoietic progenitor cells as integrative hubs for adaptation to and fine-tuning of inflammation.
Nat. Immunol.
20
:
802
811
.
20
Zhong
,
C.
,
X.
Yang
,
Y.
Feng
,
J.
Yu
.
2020
.
Trained immunity: an underlying driver of inflammatory atherosclerosis.
Front. Immunol.
11
:
284
.
21
Kimball
,
A. S.
,
A.
Joshi
,
W. F.
Carson
IV
,
A. E.
Boniakowski
,
M.
Schaller
,
R.
Allen
,
J.
Bermick
,
F. M.
Davis
,
P. K.
Henke
,
C. F.
Burant
, et al
.
2017
.
The histone methyltransferase MLL1 directs macrophage-mediated inflammation in wound healing and is altered in a murine model of obesity and type 2 diabetes.
Diabetes
66
:
2459
2471
.
22
Arts
,
R. J. W.
,
A.
Carvalho
,
C.
La Rocca
,
C.
Palma
,
F.
Rodrigues
,
R.
Silvestre
,
J.
Kleinnijenhuis
,
E.
Lachmandas
,
L. G.
Gonçalves
,
A.
Belinha
, et al
.
2016
.
Immunometabolic pathways in BCG-induced trained immunity.
Cell Rep.
17
:
2562
2571
.
23
Bekkering
,
S.
,
J.
Quintin
,
L. A.
Joosten
,
J. W.
van der Meer
,
M. G.
Netea
,
N. P.
Riksen
.
2014
.
Oxidized low-density lipoprotein induces long-term proinflammatory cytokine production and foam cell formation via epigenetic reprogramming of monocytes.
Arterioscler. Thromb. Vasc. Biol.
34
:
1731
1738
.
24
El-Osta
,
A.
,
D.
Brasacchio
,
D.
Yao
,
A.
Pocai
,
P. L.
Jones
,
R. G.
Roeder
,
M. E.
Cooper
,
M.
Brownlee
.
2008
.
Transient high glucose causes persistent epigenetic changes and altered gene expression during subsequent normoglycemia.
J. Exp. Med.
205
:
2409
2417
.
25
Quintin
,
J.
,
S.
Saeed
,
J. H. A.
Martens
,
E. J.
Giamarellos-Bourboulis
,
D. C.
Ifrim
,
C.
Logie
,
L.
Jacobs
,
T.
Jansen
,
B. J.
Kullberg
,
C.
Wijmenga
, et al
.
2012
.
Candida albicans infection affords protection against reinfection via functional reprogramming of monocytes.
Cell Host Microbe
12
:
223
232
.
26
Christ
,
A.
,
P.
Günther
,
M. A. R.
Lauterbach
,
P.
Duewell
,
D.
Biswas
,
K.
Pelka
,
C. J.
Scholz
,
M.
Oosting
,
K.
Haendler
,
K.
Baßler
, et al
.
2018
.
Western diet triggers NLRP3-dependent innate immune reprogramming.
Cell
172
:
162
175.e14
.
27
Keating
,
S. T.
,
L.
Groh
,
K.
Thiem
,
S.
Bekkering
,
Y.
Li
,
V.
Matzaraki
,
C. D. C. C.
van der Heijden
,
J. H.
van Puffelen
,
E.
Lachmandas
,
T.
Jansen
, et al
.
2020
.
Rewiring of glucose metabolism defines trained immunity induced by oxidized low-density lipoprotein. [Published erratum appears in 2020 J. Mol. Med. (Berl.) 98: 1051.]
J. Mol. Med. (Berl.)
98
:
819
831
.
28
Liang
,
C. P.
,
S.
Han
,
T.
Senokuchi
,
A. R.
Tall
.
2007
.
The macrophage at the crossroads of insulin resistance and atherosclerosis.
Circ. Res.
100
:
1546
1555
.
29
Reilly
,
M. P.
,
D. J.
Rader
.
2003
.
The metabolic syndrome: more than the sum of its parts?
Circulation
108
:
1546
1551
.
30
Alexander
,
C. M.
,
P. B.
Landsman
,
S. M.
Teutsch
,
S. M.
Haffner
;
Third National Health and Nutrition Examination Survey (NHANES III)
; 
National Cholesterol Education Program (NCEP)
.
2003
.
NCEP-defined metabolic syndrome, diabetes, and prevalence of coronary heart disease among NHANES III participants age 50 years and older.
Diabetes
52
:
1210
1214
.
31
Dorman
,
J. S.
,
R. E.
Laporte
,
L. H.
Kuller
,
K. J.
Cruickshanks
,
T. J.
Orchard
,
D. K.
Wagener
,
D. J.
Becker
,
D. E.
Cavender
,
A. L.
Drash
.
1984
.
The Pittsburgh insulin-dependent diabetes mellitus (IDDM) morbidity and mortality study. Mortality results.
Diabetes
33
:
271
276
.

The authors have no financial conflicts of interest.

Supplementary data