Diabetes mellitus (DM) significantly increases the risk for cancer and cancer progression. Hyperglycemia is the defining characteristic of DM and tightly correlates with a poor prognosis in patients with cancer. The hexosamine biosynthetic pathway (HBP) is emerging as a pivotal cascade linking high glucose, tumor progression, and impaired immune function. Here we show that enhanced glucose flow through the HBP drives cancer progression and immune evasion by increasing O-GlcNAcylation in tumor-associated macrophages (TAM). Increased O-GlcNAc skewed macrophage polarization to a M2-like phenotype supporting tumor progression. Finally, we found an upregulation of M2 markers on TAMs in DM2 patients with colorectal cancer compared with nondiabetic normoglycemic patients. Our results provide evidence for a new and targetable mechanism of cancer immune evasion in patients with hyperglycemia, advocating for strict control of hyperglycemia in patients with cancer.

Diabetes mellitus (DM) is one of the most common metabolic disorders with a high morbidity and mortality leading to uncontrolled hyperglycemia (1). Prevalence of DM is increasing at an alarming rate and on a global scale (2). Epidemiologic evidence shows that patients with DM display a significantly higher risk of developing multiple types of cancers (3–6) and also demonstrated that cancer recurrence, metastasis, or fatal outcome occurs with much higher incidence in patients with cancer with hyperglycemia (7–9). Although an association of a poor prognosis with hyperglycemia in patients with cancer has been shown, mechanistic studies often focus on metabolic changes in tumor cells (10). Studies on the influence of hyperglycemia on the tumor immune microenvironment are often lacking, and further investigations are needed.

Glucose metabolism, and in particular the hexosamine biosynthetic pathway (HBP), emerged as an important cascade linking cancer progression and hyperglycemia (10–13). The HBP provides UDP-N-acetyl-D-glucosamine (UDP-GlcNAc) and its derivatives for glycosylation using glucose, glutamine, Acetyl-CoA, and UTP as substrates (14). UDP-GlcNAc is highly responsive to cell nutrient fluctuations as its synthesis depends on products of the several metabolic pathways and is therefore considered to be a metabolic sensor (14). Our recent work showed that increasing extracellular glucose concentrations provoke aberrant glycosylation, increasing cell proliferation and invasion in vivo due to increased flux through the HBP (14). Hyperglycemia increased levels of intracellular O-GlcNAcylation in tumor cells (12), a posttranslational modification of intracellular proteins catalyzed by O-GlcNAc transferase (15) and counteracted by O-GlcNAcase (OGA; ref. 16). Increase of O-GlcNAc levels induced by high glucose (HG) can affect the expression of genes and deregulation of biochemical pathways to respond to environmental changes (10).

Metabolic changes can strongly influence immune cell function (17–21). Although glucose is needed for proper effector function as T-cell activation and M1 macrophage polarization (21, 22), it is largely unknown how a systemic increase of glucose in circulation of diabetic patients with cancer instructs the local tumor immune microenvironment. In order to improve our understanding of the influence of hyperglycemia, we used a mouse model of hyperglycemia to study the changes in the tumor immune microenvironment. We characterized alterations of cell infiltration, activation, and polarization and provide a molecular mechanism for the effect of hyperglycemia on tumor growth.

Study approval

All experiments involving mice were approved by the local ethical committees in Basel, Switzerland (approval number 2747), and in Rio de Janeiro, Brazil (approval number IBCCF214-09/16), and were performed in accordance with the regulations. All human tumor samples were anonymized and obtained from the Marcílio Dias Hospital following informed written consent (approval 39063014.8.0000.5256, Brazil).

Cell culture

Mouse colon adenocarcinoma MC38 cells were donated by Dr. M. Smyth (QIMR Biomedical Research Institute, Brisbane, Queensland, Australia) and mouse melanoma B16D5 cells by Dr. L. Weiner (Georgetown University, Washington, DC). Tumor cell lines were cultured in DMEM (25 mmol/L Glucose, Sigma-Aldrich), supplemented with 2 mmol/L l-glutamine, 1 mmol/L sodium pyruvate, 1% MEM nonessential amino acids, 500 U/mL streptomycin/penicillin, and 10% heat-inactivated FBS (Sigma-Aldrich) at 37°C and 5% CO2. All cells were regularly tested for Mycoplasma by PCR.

Mice

Male C57BL/6 and NOD scid gamma (NSG) mice were obtained from Janvier Labs (23), bred in-house at the Department of Biomedicine, University Hospital Basel, Switzerland, and were used at 10 to 12 weeks old. Hyperglycemia was induced with a single i.p. injection of 150 mg/kg streptozotocin (STZ, Sigma-Aldrich) diluted with 0.1 mmol/L sodium citrate buffer as vehicle (pH 4.3). Euglycemic control mice (EuG) were treated with vehicle only. Seven days after STZ injection, 5 × 105 syngeneic tumor cells in 100 μL PBS were injected s.c. into the flank of mice. Tumor growth and health scores were measured 3 times per week until a maximum size of 1,500 mm3 (advanced tumor stage) or 300 to 400 mm3 (early tumor stage). Perpendicular tumor diameters were measured using a caliper, and tumor volume was calculated according to the following formula: tumor volume (mm3) = (d2*D)/2, where d and D are the shortest and longest diameters (in mm) of the tumor, respectively. Mice that developed ulcerated tumors were sacrificed and excluded from further analysis.

Blood glucose measurement

Blood sampling was performed from the tail vein 7 days after STZ injection and on the day of tumor resection. Glucose levels were measured using a validated glucose meter (FreeStyle lite, Abbott). STZ-treated mice were considered to be hyperglycemic if the blood glucose levels exceeded 250 mg/dL at the end of the experiment.

Intratumoral glucose measurement

Tumors were collected, weighed, cut in small pieces, and mechanically dissociated using a micropipette. Each tumor was resuspended in 1 mL PBS, filtered, glucose concentrations measured by the glucose colorimetric detection Kit (Invitrogen), and normalized to tumor mass.

Treatments

Depletion of intratumoral macrophages was achieved by i.p. anti-CSF1R antibody treatment (CD115, clone AFS98, BioXCell), with a total of 7 injections, 1 mg before tumor inoculation, followed by 6 injections of 400 μg each every 3 to 4 days over 3 weeks.

HBP inhibition in vivo was achieved by treatment of tumor-bearing mice with 3 doses (10 mg/kg) of DON (6-Diazo-5-oxo-L-norleucine, Sigma-Aldrich) every 3 days, beginning when tumors reached 300 to 400 mm3. In addition, mice were treated with a single i.p. injection of DON (10 mg/kg) or the OGA-inhibitor Thiamet G (TMG, 20 mg/kg, Sigma-Aldrich), and animals sacrificed 2 days after the treatment.

Tumor digestion

Tumors were surgically collected, cut in small pieces, and digested in a solution of collagenase IV (Worthington), DNase I (Sigma-Aldrich), Hyaluronidase (Sigma-Aldrich), and accutase (PAA Laboratories) for 1 hour at 37°C. After incubation, cells were filtered, washed, and separated by density centrifugation with Histopaque-1119 (Sigma-Aldrich), and cell suspensions frozen at -80°C for future analyses. All samples were stained using LIVE/DEAD Fixable Blue Dye (Invitrogen) and various panels of antibodies to identify different population of immune cells. All samples were acquired with the Fortessa LSR II Flow Cytometer (BD Biosciences) and analyzed in the FlowJo software (BD Biosciences).

Bone marrow–derived macrophages and treatments

Bone marrow–derived macrophages (BMDM) were differentiated in petri dishes with M-CSF (20 ng/mL, Peprotech) or 20% L929 supernatant for 6 days in HG (25 mmol/L) or low glucose (LG, 5 mmol/L) DMEM. Cells were plated at 5 × 105/well in 12-well plates and treated for 24 hours with 10 μmol/L of the inhibitors DON, TMG, and OSMI-1 (Sigma-Aldrich). BMDMs were polarized to M1 with IFNγ 50 ng/mL (Peprotech) and lipopolysaccharide (LPS) 20 ng/mL (Sigma-Aldrich) or to M2 with IL4 20 ng/mL (Peprotech) for 48 hours. BMDMs were washed in FACS buffer, followed by incubation with anti-CD16/CD32 blocking antibodies (eBioscience) for Fc-receptor blocking. Polarization was confirmed by staining against CD206, CD86, and MHC-II. Cells were also collected for metabolite extraction and immunoblotting.

BMDM coculture assay

BMDMs were differentiated in petri dishes with M-CSF (20 ng/mL, Peprotech) for 6 days in HG (25 mmol/L) or LG (5 mmol/L) DMEM. MC38 cells were cocultured with macrophages in the ultralow attachment 96-well U-bottom plates (Corning). The coculture was carried out using serum-free LG or HG DMEM at different time points (2, 4, or 24 hours), and in each well was added 1 × 105 MC38 cells and 5 × 104 macrophages. After the incubation time, polarization was accessed by staining against CD206, CD86, and MHC-II.

Metabolite extraction and nucleotide sugar analysis

Polar metabolites were extracted from 5 × 106 macrophages with chloroform, methanol, and water (1:1:1). Each sample was subjected to chromatographic separation on a Hypercarb PGC column (Thermo Fisher) running in HPLC Prominence (Shimadzu). Nucleotide sugars were eluted using a discontinuous linear gradient of mobile phases A (0.2% formic acid and 0.75% ammonium hydroxide in water) and B (95% acetonitrile with 0.1% formic acid and 0.07% ammonium hydroxide). The detection was achieved using a diode array detector (SPM-M20A Prominence, Shimadzu) set to 250 to 400 nm, coupled to a Q-TOF mass spectrometer (maXis Impact, Bruker Daltonics). The retention times of nucleotide sugars were previously established using standards. The chromatograms and mass spectra were analyzed with DataAnalysis 4.2 (Bruker Daltonics). UDP-GlcNAc concentration was quantified by measuring the area of its correspondent UV peak at 262 nm normalized to the total area of the chromatogram from 5 to 30 minutes, using a calibration curve. For the other nucleotide sugars, the peaks of their respective extracted ion chromatograms were evaluated similarly.

Immunoblotting

BMDMs were washed with PBS and homogenized in lysis buffer [150 mmol/L NaCl, 30 mmol/L Tris-HCl, pH 7.6, 1 mmol/L EDTA, 1 mmol/L EGTA, 0.1% SDS, 1 mmol/L phenylmethylsulfonyl fluoride, and 1 μmol/L O-(2-acetamido-2-deoxy-d-glucopyranosylidene)amino-N-phenylcarbamate (PUGNAc)]. Cell lysates were sonicated and centrifuged. Supernatant was collected, protein concentration was determined, and modified Laemmli buffer was added. Samples were separated on SDS-polyacrylamide gels and were subsequently electroblotted to nitrocellulose membrane (Bio-Rad). The membranes were blocked in Tris-buffered saline with 0.1% (v/v) Tween 20 with either 3% (w/v) BSA or 3% (w/v) nonfat dry milk. The blocked membranes were then incubated overnight at 4°C with primary antibodies against O-GlcNAc (CTD 110.6 or RL-2, Santa Cruz Biotechnology), arginase-1 (clone H52, Santa Cruz Biotechnology), NOS2 (clone N20, Santa Cruz Biotechnology), and β-actin (clone 13E5, Cell Signaling Technology). The blots were then washed, incubated with the appropriate secondary antibody, developed using ECL (GE Healthcare), and exposed to Image Quant LAS 4000 (GE Healthcare). ImageJ software was used for densitometric analysis of immunoblots, and measurements were normalized against β-actin.

Histology and immunofluorescence

Paraffin sections of patients with colorectal cancer with DM and control patients were obtained from the Marine Hospital Marcílio Dias, Rio de Janeiro, Brazil. All samples were from surgical resections and were evaluated by a board-certified pathologist. Sections were deparaffinized automatically, and antigen retrieval was performed in citrate buffer. Tissues were permeabilized with 0.1% Triton X-100 and blocked with 1% BSA. Primary antibodies were incubated overnight in 1% BSA solution. The antibodies used were anti-CD68 (clone Y1-82A, BD Biosciences), anti-CD206 (polyclonal, Abcam, Cat.: 64693), and anti-CD163 (polyclonal, Abcam, Cat.: 87099). After incubation, slides were washed and incubated with the respective secondary antibody and then washed. Slides were further stained with DAPI (Thermo Fisher) and mounted with ProLong diamond antifade mountant. Pictures were taken in a Nikon Eclipse Ti-S microscope and analyzed using ImageJ.

Statistical analysis

Graphs and statistical tests were done with Prism 8 (GraphPad). Statistical significance between two groups was tested using paired Student t tests. Comparison between multiple time points was analyzed by one-way ANOVAs with multiple comparisons and Tukey posttest. Differences in survival curves were analyzed by the Gehan–Breslow–Wilcoxon test.

Hyperglycemia increases tumor growth and reduces survival in two different syngeneic mouse models

In order to study the influence of systemic hyperglycemia on tumor progression, we used a mouse model of acute hyperglycemia by applying the cytotoxic drug STZ, destroying pancreatic beta cells (24). We tested the growth of MC38 and highly metastatic mouse melanoma B16D5 cells subcutaneously transplanted into C57BL/6 mice treated with STZ (Fig. 1A). Survival was significantly shorter, and tumor growth increased in hyperglycemic mice bearing subcutaneous B16D5 tumors (Fig. 1B and C). Similarly, hyperglycemic mice injected with MC38 tumor cells showed a significantly shorter survival than euglycemic mice (Fig. 1D and E). In addition, hyperglycemia was observed in the peripheral blood as well as within the extracellular space of the tumor microenvironment, demonstrated by increased glucose concentrations in the interstitial tumor fluid (Fig. 1F). Tumor growth was not due to STZ treatment because STZ reduced MC38 viability in vitro in a concentration-dependent manner (Supplementary Fig. S1A). These experiments demonstrate that STZ treatment, which induces hyperglycemia, increases tumor growth and thereby reduces survival in two different syngeneic mouse models.

Sialylation is not a key mechanism in hyperglycemia-induced tumor growth

Recent studies have shown that hyperglycemia can lead to an increase of sialylation in tumor microenvironment (12). More importantly, hypersialylation of tumors can mediate immune evasion by engaging the sialoglycan-Siglec immune checkpoint (25, 26). In our model system, hyperglycemia induced an increase in sialylation demonstrated by increased Siglec-E-Fc binding when compared with cells grown in normoglycemia (Supplementary Fig. S2A). We therefore tested the role of sialylation in the hyperglycemia model by using GlcNAc-epimerase (GNE)–deficient MC38 tumor cells, which display a strong reduction in cell surface sialylation (25). As previously published (25), GNE-deficient MC38 cells grew slower compared with wild-type cells. Yet, hyperglycemia still increased tumor growth, conserving the key difference in tumor growth between hyperglycemic and euglycemic mice (Supplementary Fig. S2B). This finding was confirmed in mice deficient for Siglec-E, which showed similar tumor growth as wild-type animals, but were smaller in the hyperglycemic compared with euglycemic mice (Supplementary Fig. S2D). We concluded that engagement of the sialoglycan-Siglec pathway is not the main mechanism of enhanced tumor growth in hyperglycemic mice.

Hyperglycemia skews intratumoral macrophage polarization

To delineate the cellular immune mechanisms involved in hyperglycemia-induced tumor progression, we characterized the immune cell infiltrates in size-matched tumors by flow cytometry (see gating strategy, Supplementary Fig. S3). We found a significant reduction of CD8+ tumor-infiltrating lymphocytes (TIL; Fig. 2A), as well as of conventional CD11c+MHC-II+ dendritic cells (Supplementary Fig. S3A) in tumors from hyperglycemic mice. We found a reduction of CD4+CD25+FOXP3+ regulatory T cells (Fig. 2B), whereas no significant differences in the frequencies of CD4+ TILs (Fig. 2C) and B cells (Fig. 2D) were observed in the infiltrates of hyperglycemic tumors. In addition, we found that the total percentages of CD3+ cells were the same among both groups (Supplementary Fig. S4B). The presence of Ly6G+ cells was decreased in tumors from hyperglycemic mice (Supplementary Fig. S4C), whereas no difference was seen in the Ly6C+ cells (Supplementary Fig. S4D). The numbers of CD11b+F4/80+Gr1 tumor-associated macrophages (TAM) were similar in both conditions (Supplementary Fig. S4E). However, a significant increase in CD206 expression was found in hyperglycemic MC38 (Fig. 2E; Supplementary Fig. S4F) and B16D5 tumors (Supplementary Fig. S4I). This result was accompanied by a decrease of the M1 marker MHC-II in MC38 (Fig. 2F; Supplementary Fig. S4G), but not in B16D5 tumors (Supplementary Fig. S4J), whereas MHC-II+CD206+ cells were not altered in both cases (Supplementary Fig. S4H and S4K).

Taken together, the analysis of inflammatory infiltrates suggests that hyperglycemia affects the immune response against the tumor.

Macrophage polarization is not modulated by tumor cells

The interaction between the cancer cells, its metabolites, and other soluble factors are known to influence TAM function (21). Previously, we have shown that HG concentration induces aberrant glycosylation in colon cancer cell. By modulating the activity of the HBP rate-limiting enzyme, glutamine-fructose-6-phosphate amidotransferase (GFAT), we abrogate hyperglycemia effects on tumor cells (12). To address the influence of tumor cell surface glycosylation in macrophage polarization, shGFAT-MC38 and the control MC38 cells, transduced with a scramble GFAT, were injected in hyperglycemic mice treated with STZ or in the euglycemic group, and cell infiltrate was analyzed. As shown in Supplementary Fig. S5A and S5B, GFAT downregulation does not affect the expression of TAM polarization markers MHCII and CD206. These results are supported by experiments using GNE-deficient MC38 tumors (Supplementary Fig. S2C) and by in vitro coculture experiments (Fig. 2G and H; Supplementary Fig S5C–S5J). Indeed, the increase of CD206 (Fig. 2I; Supplementary Fig. S5C and S5D) and decrease of MHCII (Supplementary Fig. S5E–S5G) were observed coculturing MC38 and BMDMs in HG medium at different time points (2, 4, and 24 hours). Furthermore, we observed the decrease of the M1 marker CD86 in the coculture using HG medium comparing with LG (Supplementary Fig. S5H–S5J). Coculture experiment and data showing that hyperglycemia also accumulate M2 macrophages in the peritoneum after pristine injection in vivo, an inflammatory murine model of systemic Lupus erythematosus, support that macrophage polarization is affected by HG concentrations rather than by tumor cells or its metabolites. Analysis of pristane-induced macrophages in the peritoneal lavage showed a decrease of MHC-II (Supplementary Fig. S5K) and increase of CD206 expression (Supplementary Fig. S5L) in hyperglycemia. In addition, the proportions of Ly6C+ (Supplementary Fig. S5M) and Ly6G+ cells (Supplementary Fig. S5N) were not affected by hyperglycemia.

TAMs mediate immune evasion in hyperglycemic mice

Next, we aimed to determine the functional contribution of the immune system and specific immune cell types to the observed effect of hyperglycemia-promoted tumor growth. Therefore, we induced hyperglycemia in NSG mice that are devoid of a functional immune system. Tumor growth of s.c. injected MC38 cells (Fig. 3A) was faster than in immunocompetent mice (Fig. 3B), whereas no differences were observed between hyperglycemic and euglycemic mice (Fig. 3A). The analysis of the TAMs showed no significantly difference in the expression of CD206 (Supplementary Fig. S5O) and MHCII (Supplementary Fig. S5P) comparing hyperglycemic and euglycemic mice. This finding indicates that growth difference observed in our mouse model was mainly mediated by changes in the immune response. Further characterization by gene expression analysis showed a decreased presence of inflammation-associated genes, using immunocompetent C57bl/6 mice. Genes involved in immune cell recruitment and activation such as Il1b, Cxcl1, Ccl5, S100a9, and S100a8 were significantly downregulated in hyperglycemic tumors compared with euglycemic tumors (Supplementary Fig. S6A). In contrast, genes involved in tumor-promoting inflammation as the M2 macrophage markers Cd163 and Cd276, an immune checkpoint member, were increased supporting an important function of macrophages in this model (Supplementary Fig. S6A). We therefore tested the contribution of macrophages to the hyperglycemia-induced tumor growth by treating mice with anti-CSF1R antibodies. Hyperglycemic mice treated with anti-CSF1R showed similar tumor growth as treated euglycemic mice (Fig. 3B). The treatment of mice with anti-CSF1R showed a significant reduction of intratumoral macrophages (Fig. 3CE). These results suggest a key role for intratumoral macrophages in mediating the effect of hyperglycemia on tumor growth and suggest an interplay between innate and adaptive responses in the observed effect.

Flux through the HBP induces M2 polarization

Hyperglycemia can increase the rate of flux through the HBP, resulting in increased levels of UDP-GlcNAc in cells in HG environments (12). Previous work has shown that UDP-GlcNAc is a characteristic of M2 polarization (27) and acts as a substrate for O-GlcNAcylation (14). In addition, increased O-GlcNAcylation has been demonstrated to influence macrophage function (28–30). We therefore investigated whether hyperglycemia could change macrophage polarization through increased O-GlcNAcylation. First, we tested whether polarization of in vitro cultured BMDMs would be affected by the glucose concentration of the cell culture medium (Supplementary Fig. S6B). HG medium augmented UDP-GlcNAc levels (Fig. 4A) and O-GlcNAcylation in BMDMs, with the highest levels being reached under M2-polarizing conditions (Fig. 4B; Supplementary Fig S6C). Other activated monosaccharides were differentially affected by HG, with CMP-Neu5Ac levels decreasing under HG conditions (Supplementary Fig. S6D–S6F). HG changed the macrophage phenotype measured as a decrease of the M1 markers CD86 (Fig. 4C, F, and I), MHC-II (Fig. 4D, G, and J), and NOS2 (Supplementary Fig. S6G), whereas M2 markers such as CD206 (Fig. 4E, H, and K) and Arginase1 (Supplementary Fig. S6G) were increased.

To determine whether O-GlcNAcylation would induce an M2 phenotype in BMDMs, we inhibited the enzyme O-GlcNAc-transferase (15) using the inhibitor OSMI-1 (Fig. 4CE). Treatment decreased overall cell O-GlcNAcylation (Supplementary Fig. S6H) and reversed the effects induced by HG, leading to an increase in CD86 and MHC-II expression (Fig. 4C and D) and a decrease in CD206 (Fig. 4E). Applying the glutamine antagonist DON, which inhibits the rate-limiting enzyme of HBP GFAT (Fig. 4FH) and leads to a decrease in O-GlcNAcylation (Supplementary Fig. S6I), showed similar effects.

Conversely, increasing O-GlcNAcylation by inhibiting OGA with TMG (Supplementary Fig. S6J) augmented the expression of CD206 (Fig. 4K) and decreased CD86 in both HG and LG cultured macrophages (Fig. 4I). These experiments suggest that HG concentrations enhance M2 polarization in murine BMDMs by increasing the flux of glucose through the HBP, resulting in an increase of O-GlcNAcylation.

To further investigate the influence of O-GlcNAcylation on hyperglycemia-promoted tumor growth and macrophage fate in vivo, we treated hyperglycemic and euglycemic mice with the inhibitor of the HBP, DON, thereby inhibiting O-GlcNAcylation. Treatment impaired tumor growth in both hyperglycemic and euglycemic mice (Fig. 5A) and resulted in a similar tumor growth in both groups, abrogating the effect of hyperglycemia (Fig. 5A). We observed a clear reduction of polarization to a protumorigenic phenotype of macrophages (Fig. 5B; Supplementary Fig. S7A) and an increase in M1 macrophage markers (Supplementary Fig. S7B and S7C) in the tumor microenvironment of hyperglycemic mice. Confirming this finding, opposite effects were seen when mice were treated with the inhibitor of the enzyme that cleaves O-GlcNAc (OGA), TMG, increasing O-GlcNAcylation in mice. We found an increase in protumorigenic M2-like macrophages (Fig. 5B), whereas no differences were observed in the CD206+ and MHC-II+ double-positive population (Supplementary Fig. S7D). Together, these results show that the inhibition of O-GlcNAcylation reverts the effect of hyperglycemia on tumor growth and intratumoral macrophage polarization in vivo, reprograming macrophages toward an antitumor phenotype.

Diabetes alters the tumor immune microenvironment in colorectal cancer

Hyperglycemia correlates with a worse prognosis in colorectal cancer (31), and our data from preclinical models provided evidence that hyperglycemia could induce intratumoral macrophage polarization to a protumorigenic M2-like phenotype. Analysis of The Cancer Genome Atlas gene expression data showed that the expression of M2 macrophage markers correlated with a reduced overall survival for all types of cancers (Supplementary Fig. S7E) and that the increased expression of OGT was similarly associated with poor survival (Supplementary Fig. S7F). To investigate if hyperglycemia was associated with a protumorigenic M2-like polarization of macrophages in patients with colorectal cancer, we analyzed samples from patients with colorectal cancer with type 2 DM for changes in macrophage infiltration and polarization (Fig. 5C). No significant differences in the frequency of CD68+ intratumoral macrophages were observed between nondiabetic and diabetic patients (Fig. 5D). However, IHC analysis of CD68+ macrophages for the M2 markers CD206 (Fig. 5E; Supplementary Fig. S8A) and CD163 (Fig. 5F; Supplementary Fig. S8B) showed a higher frequency of M2-like macrophages in diabetic patients. This analysis suggests that patients with colorectal cancer with hyperglycemia have a higher frequency of M2 macrophages, although larger studies are needed to further corroborate a direct association between DM2 and M2 macrophage polarization.

Here, we show that systemic and intratumoral hyperglycemia significantly alters the tumor immune microenvironment and increases the number of alternatively M2-like polarized tumor-promoting macrophages. We demonstrate that these macrophages enhance tumor growth by mediating immune evasion and secondary inhibition of the adaptive immune response. Intratumoral macrophages are known to support cancer progression by creating an inhibitory microenvironment and to strongly inhibit antitumor immunity (32–34). M2-like polarized macrophages secrete mediators such as IL10 and TGFβ leading to the dampening of adaptive antitumor immunity (32–34). M2-like polarized TAMs are also defective of phagocytosing tumor cells (35–37). Due to the heavy involvement in immune evasion and cancer progression, TAMs have become an interesting target for cancer immunotherapy. For example, blockade of CD47 on tumor cells or its receptor SIRPα on macrophages was successfully tested in preclinical models, and antibodies against CD47 are currently in clinical development (35–37). Targeting the recruitment and polarization of macrophages has also been tested by using blocking agents against CSF1 or CSF1R (38, 39). In our model, treatment with anti-CSF1R led to a significant reduction in TAMs and M2-like polarization. This reduction of protumorigenic macrophages reversed the tumor-promoting effect of hyperglycemia, which clearly implicates intratumor macrophages in the observed effect.

Polarization of TAMs is not binary but rather a continuum (32–34), with LPS and IFNγ inducing an antitumoral M1 phenotype, whereas cytokines including IL4 and IL13 lead to an alternative, anti-inflammatory M2-like phenotype. Previous evidence showed that M2 polarization in murine macrophages is accompanied by an increase in UDP-GlcNAc (27). We demonstrate that hyperglycemia increases M2 macrophage polarization through an increase in HBP flux and enhanced UDP-GlcNAc production. Inhibition of the HBP with DON showed a reduction in M2 polarization and also abolished the effect of hyperglycemia in vivo, although the results of systemic DON treatment need to be interpreted cautiously due to its pleiotropic effects on many different cell types including direct effects on tumor cells. In further support of an important role of O-GlcNAcylation in hyperglycemia-driven M2 polarization, inhibition of the O-GlcNAc transferase by OSMI-1 inhibited the M2 polarization in hyperglycemic condition. O-GlcNAcylation has been associated in regulation of inflammatory responses and other immune reactions (21, 40).

Our data support a role for O-GlcNAcylation in regulating macrophage polarization and differentiation, with increased O-GlcNAcylation directing macrophages toward M2-like phenotype. We show for the first time that intratumoral hyperglycemia leads to mediate cancer immune evasion by O-GlcNAc–mediated macrophage polarization. In agreement, recent work has shown that LPS-activated macrophages (M1 phenotype) display reduced HBP activity and protein O-GlcNAcylation (30). Previous studies have reported how O-GlcNAc cycling can influence macrophage function. Depletion of OGT in the human macrophage cell line THP-1 induces M1 polarization, whereas M2 genes were downregulated (41). In addition, pharmacologic inhibition of OGA in an experimental model of ischemic stroke showed an increased expression of M2 markers and decreased expression of M1 markers in microglia cells (42). OGT-mediated O-GlcNAcylation of STAT3 in macrophages inhibits STAT3 phosphorylation and IL10 production (43). Several proteins which regulate macrophage function, i.e., Akt/mTOR, HIFα, as well as the NF-κB and NFAT families (34, 44–46), were shown to be O-GlcNAcylated in other models. Together, these studies point to OGT as an important mediator of M2 polarization in macrophages.

A limitation of our study is the use of STZ to induce hyperglycemia, since the model induces hypoinsulinemia, where the acute type of hyperglycemia resembles more the situation in patients with type 1 DM. Future experiments will make use of other DM models, including diet-induced mouse models that mimic type 2 DM. We have also not identified exact target proteins of O-GlcNAcylation in our experiments, but have focused on overall levels of O-GlcNAcylation. Further studies are needed to determine the distinct influence of enhanced O-GlcNAcylation on different signaling pathways.

Our results support a metabolic reprogramming of the tumor immune microenvironment by hyperglycemia resulting in an immune-inhibitory, tumor-promoting condition mediated by increased M2-like macrophage polarization. Accordingly, diabetic patients with colorectal cancer had an increased number of M2-like polarized macrophages compared with patients with colorectal cancer without diabetes. Although our study advocates for a strict control of glycemia in diabetic patients with cancer, further functional analyses and clinical examinations are needed to determine the target range of blood glucose in patients with cancer. One could also speculate that diabetic patients with cancer could benefit from an additional macrophage-targeting agent such as CSF-1R blockade. Furthermore, the description of O-GlcNAcylation mediating hyperglycemia-induced tumor progression provides new targets for cancer immunotherapy, including enzymes involved in the HBP and O-GlcNAcylation. We demonstrated that OGT inhibition reprograms TAMs to an antitumor profile. By directly targeting macrophages, the specific delivery of OGT inhibitors could be achieved, greatly reducing the toxicity of the compound and influencing the tumor microenvironment only, reducing potential side effects.

No potential conflicts of interest were disclosed.

N. Rodrigues Mantuano: Conceptualization, formal analysis, funding acquisition, investigation, methodology, writing–original draft, writing–review and editing. M.A. Stanczak: Formal analysis, investigation, methodology. I.A. Oliveira: Investigation. N. Kirchhammer: Investigation. A.A. Filardy: Formal analysis, investigation. G. Monaco: Data curation, software, investigation. R.C. Santos: Resources, investigation. A.C. Fonseca: Resources. M. Fontes: Resources. C.S. Bastos Jr: Resources. W.B. Dias: Conceptualization, data curation. A. Zippelius: Conceptualization, resources, supervision, funding acquisition. A.R. Todeschini: Conceptualization, resources, supervision, funding acquisition, writing–original draft, writing–review and editing. H. Läubli: Conceptualization, resources, supervision, funding acquisition, investigation, writing–original draft, writing–review and editing.

This work was supported by funding from the Goldschmidt-Jacobson Foundation (to H. Läubli), Promedica Foundation (to M.A. Stanczak and A. Zippelius), Krebsliga Beider Basel (KLBB, to H. Läubli), Schoenemakers Foundation (to H. Läubli), Swiss National Science Foundation (SNSF grant #310030-184720, to H. Läubli), Swiss Government Excellence Scholarships for Foreign Scholars and Artists (FCS, to N. Rodrigues Mantuano), Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES, to N. Rodrigues Mantuano and R.C. Santos), Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq, to A.R. Todeschini, I.A. Oliveira, N. Rodrigues Mantuano, and W.B. Dias), Fundação de Amparo à Pesquisa do Estado do Rio de Janeiro (FAPERJ, to A.R. Todeschini), and Fundação do Câncer (to W.B. Dias). The authors thank the Centro de Espectrometria de Massas de Biomoléculas (CEMBIO) and plataforma de Imuno-análise (PIA; UFRJ, Rio de Janeiro, Brazil). They also thank all the patients that allowed the authors to use their material and made this work possible.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Ngwa
VM
,
Edwards
DN
,
Philip
M
,
Chen
J
. 
Microenvironmental metabolism regulates antitumor immunity
.
Cancer Res
2019
;
79
:
4003
8
.
2.
Wild
S
,
Roglic
G
,
Green
A
,
Sicree
R
,
King
H
. 
Global prevalence of diabetes: estimates for the year 2000 and projections for 2030
.
Diabetes Care
2004
;
27
:
1047
53
.
3.
Hardefeldt
PJ
,
Edirimanne
S
,
Eslick
GD
. 
Diabetes increases the risk of breast cancer: a meta-analysis
.
Endocr Relat Cancer
2012
;
19
:
793
803
.
4.
Bansal
D
,
Bhansali
A
,
Kapil
G
,
Undela
K
,
Tiwari
P
. 
Type 2 diabetes and risk of prostate cancer: a meta-analysis of observational studies
.
Prostate Cancer Prostatic Dis
2013
;
16
:
151
8
, S1.
5.
Lee
JY
,
Jeon
I
,
Lee
JM
,
Yoon
JM
,
Park
SM
. 
Diabetes mellitus as an independent risk factor for lung cancer: a meta-analysis of observational studies
.
Eur J Cancer
2013
;
49
:
2411
23
.
6.
de Kort
S
,
Simons
CC
,
van den Brandt
PA
,
Goldbohm
RA
,
Arts
IC
,
de Bruine
AP
, et al
Diabetes mellitus type 2 and subsite-specific colorectal cancer risk in men and women: results from the Netherlands Cohort Study on diet and cancer
.
Eur J Gastroenterol Hepatol
2016
;
28
:
896
903
.
7.
Barone
BB
,
Yeh
HC
,
Snyder
CF
,
Peairs
KS
,
Stein
KB
,
Derr
RL
, et al
Long-term all-cause mortality in cancer patients with preexisting diabetes mellitus: a systematic review and meta-analysis
.
JAMA
2008
;
300
:
2754
64
.
8.
Dehal
AN
,
Newton
CC
,
Jacobs
EJ
,
Patel
AV
,
Gapstur
SM
,
Campbell
PT
. 
Impact of diabetes mellitus and insulin use on survival after colorectal cancer diagnosis: the Cancer Prevention Study-II Nutrition Cohort
.
J Clin Oncol
2012
;
30
:
53
9
.
9.
Peairs
KS
,
Barone
BB
,
Snyder
CF
,
Yeh
HC
,
Stein
KB
,
Derr
RL
, et al
Diabetes mellitus and breast cancer outcomes: a systematic review and meta-analysis
.
J Clin Oncol
2011
;
29
:
40
6
.
10.
Vasconcelos-Dos-Santos
A
,
de Queiroz
RM
,
da Costa Rodrigues
B
,
Todeschini
AR
,
Dias
WB
. 
Hyperglycemia and aberrant O-GlcNAcylation: contributions to tumor progression
.
J Bioenerg Biomembr
2018
;
50
:
175
87
.
11.
Alisson-Silva
F
,
Freire-de-Lima
L
,
Donadio
JL
,
Lucena
MC
,
Penha
L
,
Sa-Diniz
JN
, et al
Increase of O-glycosylated oncofetal fibronectin in high glucose-induced epithelial-mesenchymal transition of cultured human epithelial cells
.
PLoS One
2013
;
8
:
e60471
.
12.
Vasconcelos-Dos-Santos
A
,
Loponte
HF
,
Mantuano
NR
,
Oliveira
IA
,
de Paula
IF
,
Teixeira
LK
, et al
Hyperglycemia exacerbates colon cancer malignancy through hexosamine biosynthetic pathway
.
Oncogenesis
2017
;
6
:
e306
.
13.
Phoomak
C
,
Vaeteewoottacharn
K
,
Silsirivanit
A
,
Saengboonmee
C
,
Seubwai
W
,
Sawanyawisuth
K
, et al
High glucose levels boost the aggressiveness of highly metastatic cholangiocarcinoma cells via O-GlcNAcylation
.
Sci Rep
2017
;
7
:
43842
.
14.
Hart
GW
. 
Nutrient regulation of signaling and transcription
.
J Biol Chem
2019
;
294
:
2211
31
.
15.
Aigner
S
,
Sthoeger
ZM
,
Fogel
M
,
Weber
E
,
Zarn
J
,
Ruppert
M
, et al
CD24, a mucin-type glycoprotein, is a ligand for P-selectin on human tumor cells
.
Blood
1997
;
89
:
3385
95
.
16.
Dias
WB
,
Hart
GW
. 
O-GlcNAc modification in diabetes and Alzheimer's disease
.
Mol Biosyst
2007
;
3
:
766
72
.
17.
Jung
J
,
Zeng
H
,
Horng
T
. 
Metabolism as a guiding force for immunity
.
Nat Cell Biol
2019
;
21
:
85
93
.
18.
DiMeglio
LA
,
Evans-Molina
C
,
Oram
RA
. 
Type 1 diabetes
.
Lancet
2018
;
391
:
2449
62
.
19.
Buck
MD
,
Sowell
RT
,
Kaech
SM
,
Pearce
EL
. 
Metabolic instruction of immunity
.
Cell
2017
;
169
:
570
86
.
20.
Bantug
GR
,
Fischer
M
,
Grahlert
J
,
Balmer
ML
,
Unterstab
G
,
Develioglu
L
, et al
Mitochondria-endoplasmic reticulum contact sites function as immunometabolic hubs that orchestrate the rapid recall response of memory CD8(+) T cells
.
Immunity
2018
;
48
:
542
55e6
.
21.
Mantuano
NR
,
Nunes
MC
,
Silva
FA
,
Dias
WB
,
Todeschini
AR
. 
Emerging role of glycosylation in the polarization of tumor-associated macrophages
.
Pharmacol Res
2019
;
146
:
104285
.
22.
Buck
MD
,
O'Sullivan
D
,
Pearce
EL
. 
T cell metabolism drives immunity
.
J Exp Med
2015
;
212
:
1345
60
.
23.
Mozzi
A
,
Forcella
M
,
Riva
A
,
Difrancesco
C
,
Molinari
F
,
Martin
V
, et al
NEU3 activity enhances EGFR activation without affecting EGFR expression and acts on its sialylation levels
.
Glycobiology
2015
;
25
:
855
68
.
24.
Szkudelski
T
. 
The mechanism of alloxan and streptozotocin action in B cells of the rat pancreas
.
Physiol Res
2001
;
50
:
537
46
.
25.
Stanczak
MA
,
Siddiqui
SS
,
Trefny
MP
,
Thommen
DS
,
Boligan
KF
,
von Gunten
S
, et al
Self-associated molecular patterns mediate cancer immune evasion by engaging Siglecs on T cells
.
J Clin Invest
2018
;
128
:
4912
23
.
26.
Adams
OJ
,
Stanczak
MA
,
von Gunten
S
,
Laubli
H
. 
Targeting sialic acid-Siglec interactions to reverse immune suppression in cancer
.
Glycobiology
2018
;
28
:
640
7
.
27.
Jha
AK
,
Huang
SC
,
Sergushichev
A
,
Lampropoulou
V
,
Ivanova
Y
,
Loginicheva
E
, et al
Network integration of parallel metabolic and transcriptional data reveals metabolic modules that regulate macrophage polarization
.
Immunity
2015
;
42
:
419
30
.
28.
Moriwaki
K
,
Asahi
M.
Augmented TME O-GlcNAcylation promotes tumor proliferation through the inhibition of p38 MAPK
.
Mol Cancer Res
2017
;
15
:
1287
98
.
29.
O'Neill
LA
,
Pearce
EJ
. 
Immunometabolism governs dendritic cell and macrophage function
.
J Exp Med
2016
;
213
:
15
23
.
30.
Li
X
,
Gong
W
,
Wang
H
,
Li
T
,
Attri
KS
,
Lewis
RE
, et al
O-GlcNAc transferase suppresses inflammation and necroptosis by targeting receptor-interacting serine/threonine-protein kinase 3
.
Immunity
2019
;
50
:
576
90e6
.
31.
Mills
KT
,
Bellows
CF
,
Hoffman
AE
,
Kelly
TN
,
Gagliardi
G
. 
Diabetes mellitus and colorectal cancer prognosis: a meta-analysis
.
Dis Colon Rectum
2013
;
56
:
1304
19
.
32.
Mantovani
A
. 
Wandering pathways in the regulation of innate immunity and inflammation
.
J Autoimmun
2017
;
85
:
1
5
.
33.
Cassetta
L
,
Kitamura
T
. 
Macrophage targeting: opening new possibilities for cancer immunotherapy
.
Immunology
2018
;
155
:
285
93
.
34.
Yang
M
,
McKay
D
,
Pollard
JW
,
Lewis
CE
. 
Diverse functions of macrophages in different tumor microenvironments
.
Cancer Res
2018
;
78
:
5492
503
.
35.
Gholamin
S
,
Mitra
SS
,
Feroze
AH
,
Liu
J
,
Kahn
SA
,
Zhang
M
, et al
Disrupting the CD47-SIRPalpha anti-phagocytic axis by a humanized anti-CD47 antibody is an efficacious treatment for malignant pediatric brain tumors
.
Sci Transl Med
2017
;
9
:
eaaf2968
.
36.
Weiskopf
K
. 
Cancer immunotherapy targeting the CD47/SIRPalpha axis
.
Eur J Cancer
2017
;
76
:
100
9
.
37.
Willingham
SB
,
Volkmer
JP
,
Gentles
AJ
,
Sahoo
D
,
Dalerba
P
,
Mitra
SS
, et al
The CD47-signal regulatory protein alpha (SIRPa) interaction is a therapeutic target for human solid tumors
.
Proc Natl Acad Sci U S A
2012
;
109
:
6662
7
.
38.
Neubert
NJ
,
Schmittnaegel
M
,
Bordry
N
,
Nassiri
S
,
Wald
N
,
Martignier
C
, et al
T cell-induced CSF1 promotes melanoma resistance to PD1 blockade
.
Sci Transl Med
2018
;
10
:
eaan3311
.
39.
Pyonteck
SM
,
Akkari
L
,
Schuhmacher
AJ
,
Bowman
RL
,
Sevenich
L
,
Quail
DF
, et al
CSF-1R inhibition alters macrophage polarization and blocks glioma progression
.
Nat Med
2013
;
19
:
1264
72
.
40.
de Jesus
T
,
Shukla
S
,
Ramakrishnan
P
. 
Too sweet to resist: control of immune cell function by O-GlcNAcylation
.
Cell Immunol
2018
;
333
:
85
92
.
41.
Surdziel
E
,
Clay
I
,
Nigsch
F
,
Thiemeyer
A
,
Allard
C
,
Hoffman
G
, et al
Multidimensional pooled shRNA screens in human THP-1 cells identify candidate modulators of macrophage polarization
.
PLoS One
2017
;
12
:
e0183679
.
42.
He
Y
,
Ma
X
,
Li
D
,
Hao
J
. 
Thiamet G mediates neuroprotection in experimental stroke by modulating microglia/macrophage polarization and inhibiting NF-kappaB p65 signaling
.
J Cereb Blood Flow Metab
2017
;
37
:
2938
51
.
43.
Li
X
,
Zhang
Z
,
Li
L
,
Gong
W
,
Lazenby
AJ
,
Swanson
BJ
, et al
Myeloid-derived cullin 3 promotes STAT3 phosphorylation by inhibiting OGT expression and protects against intestinal inflammation
.
J Exp Med
2017
;
214
:
1093
109
.
44.
Facundo
HT
,
Brainard
RE
,
Watson
LJ
,
Ngoh
GA
,
Hamid
T
,
Prabhu
SD
, et al
O-GlcNAc signaling is essential for NFAT-mediated transcriptional reprogramming during cardiomyocyte hypertrophy
.
Am J Physiol Heart Circ Physiol
2012
;
302
:
H2122
30
.
45.
Ferrer
CM
,
Lynch
TP
,
Sodi
VL
,
Falcone
JN
,
Schwab
LP
,
Peacock
DL
, et al
O-GlcNAcylation regulates cancer metabolism and survival stress signaling via regulation of the HIF-1 pathway
.
Mol Cell
2014
;
54
:
820
31
.
46.
Very
N
,
Steenackers
A
,
Dubuquoy
C
,
Vermuse
J
,
Dubuquoy
L
,
Lefebvre
T
, et al
Cross regulation between mTOR signaling and O-GlcNAcylation
.
J Bioenerg Biomembr
2018
;
50
:
213
22
.

Supplementary data