Elevated CD4+ T Cell Glucose Metabolism in HIV+ Women with Diabetes Mellitus

Abstract

Objective

Immune dysfunction and chronic inflammation are characteristic of HIV infection and diabetes mellitus (DM), with CD4+ T cell metabolism implicated in the pathogenesis of each disease. However, there is limited information on CD4+ T cell metabolism in HIV+ persons with DM. We examined CD4+ T cell glucose metabolism in HIV+ women with and without DM.

Design

A case-control study was used to compare CD4+ T cell glucose metabolism in women with HIV with or without DM.

Methods

Non-diabetic (HIV+DM−, N = 20) or type 2 diabetic HIV+ women with (HIV+DM+, N = 16) or without (HIV+DMTx+, N = 18) anti-diabetic treatment were identified from the WIHS and matched for age, race/ethnicity, smoking status and CD4 count. CD4+ T cell immunometabolism was examined by flow cytometry, microfluidic qRT-PCR of metabolic genes, and Seahorse extracellular flux analysis of stimulated CD4+ T cells.

Results

HIV+DM+ displayed a significantly elevated proportion of CD4+ T cells expressing the immunometabolic marker GLUT1 compared to HIV+DMTx+ and HIV+DM− (p=0.04 and p=0.01, respectively). Relative expression of genes encoding key enzymes for glucose metabolism pathways were elevated in CD4+ T cells of HIV+DM+ compared to HIV+DMTx+ and HIV+DM−. TCR-activated CD4+ T cells from HIV+DM+ showed elevated glycolysis and oxidative phosphorylation compared to HIV+DM−.

Conclusions

CD4+ T cells from HIV+DM+ have elevated glucose metabolism. Treatment of DM among women with HIV may partially correct CD4+ T cell metabolic dysfunction.

Introduction

People living with HIV (PLWH) have an increased prevalence of diabetes mellitus (DM) and are at increased risk of developing DM compared to the general population [1,2]. Immune dysfunction is a defining feature of HIV infection, with characteristic CD4+ T cell activation persisting even during suppressive antiretroviral therapy (ART) [3]. Similar to HIV, DM is an inflammatory disease associated with chronic immune activation, including activation of CD4+ T cells [4,5]. Although chronic immune activation is well-described individually for DM and HIV, it remains largely unexplored with respect to DM in the context of chronic HIV infection.

CD4+ T cell function is regulated by the metabolic program of the cell which has therefore been investigated as a target for anti-inflammatory therapies [6,7]. Activated T cells increase glucose transporter 1 (GLUT1) expression and their reliance on glycolytic metabolism to produce lactic acid in a process termed aerobic glycolysis [8–10]. Increased influx of glucose into the cell increases usage of the pentose phosphate pathway for production of nucleotides that are essential for activated T cell [11], and there is a repurposing of the oxidative phosphorylation (OXPHOS) machinery to further achieve efficient function [12]. These changes in cellular metabolism dictate immune functions critical for disease outcomes [10,13]. Compared to CD4+ T cells from persons without HIV, CD4+ T cells from PLWH have increased GLUT1 expression, glucose influx, lactic acid production and cellular activation [14].

CD4+ T cell metabolism has not been investigated in PLWH with co-morbid DM. As both diseases are associated with CD4+ T cell activation, dysregulated CD4+ T cell metabolism observed during chronic HIV infection could be exacerbated by co-morbid DM. With this background, we evaluated metabolic processes of CD4+ T cells from HIV+ women with DM compared to those without DM.

Methods

Study design

Women in the Women’s Interagency HIV Study (WIHS) were selected for a case-control study. The WIHS, prior to being combined with the Multicenter AIDS Cohort Study, was a longitudinal cohort that actively followed 1647 women with HIV and 716 women without HIV who were assessed semi-annually by way of an interview, physical examination, gynecological assessment and frailty assessment [15]. Informed consent was received for all participants, and the study was approved by the institutional review board at each study site. Cases in our study (HIV+DM+) were defined as HIV+ with DM and no report of anti-diabetic medication use to eliminate the confounding variable of anti-diabetic medication which can be anti-inflammatory [16]. DM was defined by the presence of one of the following criteria confirmed with a second of the remaining criteria: self-report of diabetic medication, ≥6.5% haemoglobin A1C (HbA1c), ≥126 mg/dL fasting plasma glucose (FPG) or self-report of DM diagnosis [2]. This definition provides accurate diagnosis and prevents underestimation of the DM and HIV association [2]. All eligible cases (n = 20) with at least one study visit between April 2012 and August 2015, had DM, no history of DM treatment, and at least 3 vials of PBMCs stored in the specimen repository from a single study visit were included in the study. Two control groups were initially developed: 1) HIV+ women with DM and reported use of anti-diabetic medication (HIV+DMTx+) and 2) HIV+ women with no history of DM (HIV+DM−). Each HIV+DM+ case was individually matched to a corresponding HIV+DMTx+ control and HIV+DM− control by age, race/ethnicity, and smoking status. These triplets were selected to have stored PBMCs within +/− 1 year of each other. Samples with cell viability <70% were excluded, reducing the sample number to 54 (n = 16, HIV+DM+; n = 18, HIV+DMTx+; n = 20, HIV+DM−). A set of HIV− samples were selected as additional controls and were matched by age, race/ethnicity and smoking status to each HIV+ sample, resulting in an additional three groups: 1) HIV− women with DM and no report of anti-diabetic medication (n=11; HIV−DM+), 2) HIV− women with DM and reported use of anti-diabetic medication (n = 13; HIV− DMTx+) and 3) HIV− women with no history of DM (n = 10; HIV−DM−). All available specimens in each group (both HIV+ and HIV−) were included to maximize power while maintaining similarity in age, race and smoking status. Vials of peripheral blood mononuclear cells (PBMC) containing 2–22 million cells per vial were obtained from the WIHS specimen repository for each participant, which were collected between April 2012 and August 2015 to minimize extended cryopreservation storage time. There was an average viability of 76.6% (trypan blue exclusion) for all samples after the thawing process.

Assessment of T cell proportions, activation and glucose transporter 1 expression

PBMC from cases and controls were assessed for T cell activation and GLUT1 expression by flow cytometry with an LSR Fortessa II (BD Biosciences). Cryopreserved PBMC were rapidly thawed in a 37°C water bath and washed using RPMI-1640 media (Corning Cellgro Mediatech Inc) containing 10% fetal bovine serum (Gemini Bio-products), 200 IU penicillin (Sigma Aldrich), 20 μg/mL streptomycin (Sigma Aldrich) and 2 mmol/L L-glutamine (Sigma Aldrich). Following washing, cells were stained with Live/Dead fixable AQUA stain (Life Technologies Corporation, L34957) followed by staining with the antibody conjugated fluorophores CD3-AF700 (BD Biosciences, 557943), CD4-PECF594 (BD Biosciences, 562281), CD8-APC-H7 (BD Biosciences, 560179), CD45RA-APC (BD Biosciences, 550855), CCR7-PECy7 (BD Biosciences, 557648), HLA-DR-FITC (BD Biosciences, 556643), CD38-BV786 (BD Biosciences, 563964) and GLUT1-PE (R&D Systems, FAB1418P). Antibody concentrations were selected according to manufacturer’s recommendations. Samples were run daily with matched samples from each group. After staining, samples were fixed and stored at 4°C until they were analyzed on the flow cytometer. SPHERO™ rainbow calibration particles (Spherotech Inc; Lake Forest, Illinois, USA; RCP-30-5A) were used each day to check the sensitivity and linearity of the flow cytometer. Data was analyzed using FlowJo Software (TreeStar Inc; Ashland, Oregon, USA).

The CD4+ T cell population was identified according to the gating strategy shown in Fig. S1. Briefly, scatter properties were used to determine the initial T cell populations which were then gated to select live cells and singlets. CD3, CD4 and CD8 expression was used to identify the CD4+ T cell population. CD45RA and CCR7 expression was used to further classify CD4+ T cell subpopulations. CD38 and HLA-DR co-expression was used to assess T cell activation, and GLUT1 expression was used as metabolic marker.

Evaluation of relative glucose metabolism gene expression

The relative expression of genes encoding the enzymes in the glycolysis pathway, pentose phosphate pathway, tri-carboxylic acid (TCA) cycle, glucose metabolic regulation and glutamine metabolism were determined using real-time PCR (RT-PCR). Samples were assessed for 96 genes using the microfluidic chip of the Fluidigm HD Biomark System which has a limit of detection of one copy per chamber (Ct ≈ 27) (Table S1). Of these 96 genes assessed, 12 assays did not detect a signal because of assay failure or absence of target, and two were used to identify a reference gene that is most consistently expressed. Thus, 82 genes were analyzed that encode for the most relevant enzymes and isozymes in the glucose metabolism pathways indicated (Table S2). An analysis of these 82 genes was also conducted for samples from HIV− participants (Table S3). The reference sequence (RefSeq) mRNA number was used by the manufacturer for primer and probe design and optimization of the reactions. CD4+ T cell cDNA from each sample was used for analysis. Thirty (30) samples of the 54 HIV+ samples (10 from each group) were selected to allow for multiple assays for samples and control reactions. Each sample was thawed and CD4+ T cells were purified by 2 rounds of negative selection using the CD4+ T cell isolation kit (MACS Miltenyi, 130-096-133) according to the manufacturer’s instructions. Five (5) samples had CD4+ T cell purity <85% as assessed by flow cytometry and were not included in further assays. RNA extraction was completed using the Qiagen RNeasy Mini Kit according to manufacturer’s instructions and assessed for quality by the absorbance at 260/280 ratio of which 2 samples had a ratio <1.90 and were excluded from further assays. Purified RNA was converted to cDNA using the QuantiNova Reverse Transcription Kit (Qiagen) according to manufacturer’s instructions. RNA was stored at −80°C and cDNA was stored at −20°C. Two (2) samples failed (>50% total failed reactions) in the RT-PCR analysis leaving a total of 21 samples run in quadruplicate for data analysis.

Results obtained from the Fluidigm HD Biomark system were visualized using the Fluidigm RT-PCR Analysis Program. The relative fold change in gene expression was calculated using the 2^−ΔΔCt method with RPL30 as the reference gene because it is not metabolism related and is stably expressed [17] and the control group without DM formed the reference samples.

Real-time assessment of glucose metabolism using Seahorse extracellular flux analysis

Glucose metabolism of stimulated CD4+ T cells from HIV+ samples was measured using the Seahorse XF96 analyzer according to manufacturer’s instructions (insufficient samples were available to assess CD4+ T cell from HIV– women). The purified CD4+ T cells were incubated overnight with 5×10⁵ anti-CD3/CD28 Dynabeads per 1×10⁶ cells (Gibco, 11131D) to stimulate the cells. Titration experiments were conducted to determine the concentration of cells to be added to each well. Two hundred thousand (2×10⁵) stimulated CD4+ T cells were adhered to Cell-Tak (Corning, 354240) coated XFe96 96-well culture plates (Agilent Technologies, 102416–100) in glucose-free XF base media (Agilent Technologies, 10253–100) supplemented with 2 mM glutamine (Sigma Aldrich, G7513). Cells were assessed for extracellular acidification rate (ECAR, mph/minute) and oxygen consumption rate (OCR, pmol/minute) at baseline and after perturbations with sequential additions of 10 mM glucose (Sigma Aldrich, G7021), 1 μM oligomycin (Agilent Technologies Inc., 103325–100), 1.5 μM carbonyl cyanide-p-trifluoromethoxyphenylhydrazone (FCCP) (Agilent Technologies Inc., 103325–100) and 50 mM 2-deoxyglucose (2-DG) (Sigma Aldrich, D8375-10MG) (all concentrations determined by titration experiments and compounds dissolved in glucose-free XF base media supplemented with 2 mM glutamine). The addition of glucose indicates ECAR associated with glycolytic metabolism, the addition of oligomycin inhibits complex V of the electron transport chain and induces maximum glycolytic metabolism and the addition of 2-DG inhibits glycolytic metabolism by binding to hexokinase [18]. While, FCCP is a potent uncoupler of mitochondrial oxidative phosphorylation and interferes with the proton gradient leading to maximum rates of oxidative phosphorylation [18]. Following acquisition of data on the XF96 analyzer, normalization of the results was done by counting the number of viable cells in each well. Wells containing Cell-Tak immobilized CD4+ T cells were trypsinized and counted using trypan blue exclusion. Data was analyzed using Wave software (Agilent Technologies Inc.). Samples were analyzed in triplicate.

Statistical analysis

The Mann-Whitney U test was used for comparison of unpaired data and correlations were assessed by the Spearman rank test. Statistical analysis of relative gene expression was completed using one-way ANOVA with Tukey post-hoc test for multiple comparisons. As groups were already matched for important confounding variables, the χ² test was used for comparison of categorical data. Analyses were conducted using IBM SPSS Statistics for Windows version 20 (IBM Corp., Armonk NY, USA) and GraphPad Prism version 8.3.1 (GraphPad Software, San Diego California USA). p-values<0.05 were considered significant.

Results

Demographic and clinical characteristics

Demographic and clinical characteristics for HIV+ women are summarized in Table 1 & 2, respectively. The majority of participants had a HIV-1 RNA <200 copies/mL, which were similar between groups. Forty-nine (49) participants (91%) reported ART use, with 96% taking a nucleoside/nucleotide reverse transcriptase inhibitor, 31% taking a non-nucleoside reverse transcriptase inhibitor, 63% taking a protease inhibitor, and 24% taking an integrase inhibitor. There was no association between ART use and DM among the participants (Table S4). Compared to controls without DM, HIV+DM+ or HIV+DMTx+ women had significantly greater median fasting glucose (84.5 mg/dL vs. 126 mg/dL, and 84.5 mg/dl vs. 117 mg/dl, respectively) and hemoglobin A1c (HbA1c) (5.5% vs. 6.4%, and 5.5% vs. 7.9%, respectively). HIV+DMTx+ women had significantly greater HbA1c than HIV+DM+ women (7.9% vs. 6.4%) and had significantly greater median serum insulin than HIV+DM– women (21.5 uIU/mL vs. 8.1 uIU/mL). HIV+DM+ women had a shorter duration of DM (median 8.14 years vs. 11.94 years) and may have diet controlled HbA1c which does not require medical intervention, providing a possible explanation for higher HbA1c in HIV+DMTx+. No differences were observed between the groups for levels of total cholesterol, LDL cholesterol, HDL cholesterol and triglycerides. The majority of HIV+DMTx+ women were treated with metformin (56%) of which 40% were treated with metformin alone, 40% were dual therapy treatment with metformin and another antidiabetic other than insulin, and 20% treated with metformin and insulin (Table S5). Thirty-three percent (33%) of all HIV+ participants were using statins at the time of sample collection, with HIV+DMTx+ women more likely to be using statins (χ² = 6.08, p < 0.05).

Table 1.

Demographic characteristics of HIV+ study participants

Variable	Groups			p-value	p-value	p-value
	HIV+DM+ N=16	HIV+DMTx+ N=18	HIV+DM− N=20
	A	B	C	(AvB)	(AvC)	(BvC)
Mean age (yrs) ± st. dev.	54.13 ± 8.65	52.58 ± 7.91	50.14 ± 8.48	0.59 a	0.17 a	0.37 a
Smoking history (%)				0.29	0.84	0.10
Never smoked	50	28	60
Current smoker	25	22	20
Former smoker	25	50	20
African American (%)	56	61	65	1.00	0.73	1.00
Overweight/obese (%)	79	94	65	0.30	0.47	0.05
CD4 ≥500 cells/mm3 (%)	69	89	75	0.21	0.72	0.41
Median CD4/CD8 Ratio (IQR)	0.90 (0.82 – 1.11)	0.83 (0.64 – 1.06)	0.94 (0.56 – 1.35)	0.47 b	0.77 b	0.38 b
Viral load <200 HIV RNA copies/mL (%)	75	83	90	0.68	0.13	0.65

^aGroups compared using students t-test.

^bGroups compared using Mann-Whitney U test.

Categorical variables compared using the Fisher’s exact test. HIV+DM+: HIV+ women with untreated DM. HIV+DMTx+: HIV+ women with treated DM. HIV+DM−: HIV+ women without DM.

Table 2.

Clinical characteristics of HIV+ study participants

Variable	Groups			p-value	p-value	p-value
	HIV+DM+ N=12	HIV+DMTx+ N=16	HIV+DM− N=19
	A	B	C	(AvB)a	(AvC) a	(BvC) a
Median fasting glucose mg/dL (IQR)	126.0 (113.0 – 139.5)	117.0 (104.0 – 183.0)	84.5 (75.8 – 91.5)	0.69	<0.001	<0.001
Median HbA1c % (IQR)	6.4 (5.5 – 7.1)	7.9 (6.5 – 9.1)	5.5 (5.2 – 5.7)	<0.01	<0.01	<0.001
Median serum insulin uIU/mL (IQR)	18.4 (8.8 – 26.3)	21.5 (17.8 – 39.0)	8.1 (2.8 – 13.5)	0.21	0.05	<0.001
DM duration, years (IQR)	8.14 (2.76 – 11.40)	11.94 (8.29 – 14.13)	-	0.02	-	-
Median cholesterol mg/dL (IQR)	190.0 (139.8 – 208.0)	183.5 (158.3 – 226.8)	182.5 (156.8 – 210.5)	0.37	0.89	0.31
Median LDL cholesterol mg/dL (IQR)	93.5 (68.0 – 137.0)	102.0 (87.0 – 135.0)	102.0 (81.8 – 123.0)	0.79	0.99	0.43
Median HDL cholesterol mg/dL (IQR)	39.0 (35.0 – 60.3)	47.5 (43.0 – 62.5)	52.5 (44.5 – 62.8)	0.13	0.51	0.61
Median triglycerides mg/dL (IQR)	120.0 (99.0 – 220.3)	135.5 (105.8 – 184.8)	106.5 (64.8 – 215.5)	0.70	0.74	0.18

^aGroups compared using Mann-Whitney U test.

HIV+DM+: HIV+ women with untreated DM. HIV+DMTx+: HIV+ women with treated DM. HIV+DM−: HIV+ women without DM.

Demographic and clinical characteristics for HIV– women are summarized in Tables S6 and S7, respectively. Compared to HIV−DM− women, HIV−DM+ and HIV−DMTx+ women had higher fasting glucose (77 mg/dL vs 103 mg/dL, and 77 mg/dL vs 111 mg/dL, respectively) and HbA1c (5.6% vs 6.2%, and 5.6% vs 7.2%, respectively). HIV−DMTx+ women had higher serum insulin (14.4 uIU/mL vs. 6.5 uIU/mL) than HIV−DM− women. There was no difference in the duration of DM between HIV−DM+ and HIV−DMTx+ (median 6.96 years vs. 4.69 years). No differences were observed in levels of total cholesterol, LDL cholesterol and HDL cholesterol. However, there was a higher level of triglycerides in HIV−DMTx+ women compared to HIV−DM+ women (126 mg/dL vs. 98 mg/dL). The majority (77%) of HIV−DMTx+ women were treated with metformin of which 80% were treated with metformin alone and 20% treated with metformin and another anti-diabetic other than insulin. Twenty-seven percent (27%) of HIV− participants were using statins, with no association to DM status (χ² = 4.42, p = 0.11).

HIV+ women with diabetes mellitus have elevated proportion of glucose transporter 1 expressing CD4+ T cells

We assessed the expression of GLUT1 on CD4+ T cells from HIV+ women with or without DM to determine if there are differences in metabolic programming between these groups, as GLUT1 is the major glucose transporter of T cells and the relative expression of GLUT1 is associated with function [10,13]. The percentage of total CD4+ T cells and the proportions of each CD4+ T cell subpopulation (naïve [N], effector memory [EM], central memory [CM], and terminally differentiated EM [TEMRA]) were similar between all groups (Fig. 1A & B). The percentage of GLUT1+CD4+ T cells was significantly greater in HIV+DM+ women compared to HIV+DM– (0.51% vs. 0.23%, p=0.01) and HIV+DMTx+ women (0.51% vs. 0.23%, p=0.04) (Fig. 1C). The elevated percentage of GLUT1+CD4+ T cells between HIV+DM+ women and controls was observed for EM (0.54% vs. 0.34%, p=0.04) and CM (0.89% vs. 0.40%, p = 0.009) CD4+ T cell subpopulations (Fig. 1D). Compared to HIV+DMTx+ women, HIV+DM+ women had a significantly higher percentage of GLUT1+CD4+ CM (0.89% vs. 0.39%, p=0.02) T cells (Fig. 1D). Comparing HIV−DM+ to HIV−DM− or HIV−DMTx+, the proportion of total CD4+ T cells and CD4+ T cell subpopulations did not differ, except for CD4+ EM T cells that were lower compared to HIV−DMTx+ (Fig S2A & B). No differences between HIV−DM+ and HIV−DM− or HIV−DMTx+ were observed for the proportion of total GLUT1+CD4+ T cells and GLUT1+CD4+ T cells subpopulations (Fig. S2C & D).

Figure 1.

The proportions of total CD4+ T cells and CD4+ T cell subpopulations expressing GLUT1. GLUT1 expressing T cells were identified by light scatter, AQUA Live/Dead and CD3 and CD4 expression. CD4+ T cell subpopulations were then identified by CD45RA and CCR7 expression. (A) Number of CD4+ T cells as a proportion of cells. (B) Proportions of naïve (CD45RA+CCR7+), effector memory (CD45RA-CCR7-), central memory (CD45RA-CCR7+) and terminally differentiated (TemRA) (CD45RA+CCR7-) live CD4+ T cells. (C) Proportion of GLUT1-expressing CD4+ T cells and (D) proportion of GLUT1-expressing CD4+ T cell subpopulations. Open squares represent women with viral loads >200 copies/mL and not on ART at the time of sampling. Closed squares represent women on ART with viral load >200 copies/mL at the time of sampling. Blue circles represent women not on ART with viral load <200 copies/mL at the time of sampling. Groups were compared using the Mann Whitney U test. A p-value <0.05 was considered significant. HIV+DM+ women, n = 13; HIV+DMTx+ women, n = 15; HIV+DM− women, n = 17.

To determine if CD4+ T cell GLUT1 expression is associated with CD4+ T cell activation, we examined the percentage of CD4+ T cells expressing CD38 and HLA-DR. The proportion of GLUT1+CD4+ T cells was significantly associated with CD38+HLA-DR+CD4+ T cells (Fig. S3A). In contrast to GLUT1+ T cells, assessment of CD38+HLA-DR+CD4+ T cells levels between groups was not significantly different by DM status or treatment (Fig. S3B). These results highlight the relationship between increased CD4+ T cell metabolic activity and activation and suggest that differences in metabolic programming are more readily apparent than conventional measures of activation between closely matched, virally suppressed HIV+ women with and without DM.

HIV+ women with diabetes mellitus have increased relative expression of genes encoding enzymes of glucose metabolism pathways.

As we observed differences in the expression of GLUT1, suggesting an increased reliance on glucose, we next assessed metabolic pathways in CD4+ T cells by gene expression analysis. Of the total genes assessed, almost all genes showed elevated relative expression and 5 genes showed significant difference between the groups (Fig. 2A). These genes represent enzymes within the glycolysis pathway (hexokinase–3 [HK3]), the TCA cycle (pyruvate dehydrogenase E1 subunit alpha-1 [PDHA1], isocitrate dehydrogenase-1 [IDH1], malate dehydrogenase 1B [MDH1B]) and the regulatory gene pyruvate dehydrogenase kinase 2 (PDK2) (Fig. 2B&C). Compared to HIV+DM− women, HIV+DM+ women had a relative gene expression increase of 4.9-fold for HK3, 3.0-fold for MDH1B, 3.5-fold for IDH1, 2.5-fold for PDK2 and 2.2-fold for PDHA1 (Fig. 2C). PDK2, MDH1B and IDH1 relative gene expression were also significantly increased for HIV+DM+ women compared to HIV+DMTx+ women (Fig. 2C). The relative gene expression of HK3 and IDH1 was positively correlated with the percentage of GLUT1+CD4+ T cells (ρ=0.58, p<0.01 and ρ=0.47, p=0.04 respectively). Among HIV− women with or without DM there were no differences in the relative expression of any gene assessed, including HK3, MDH1B, IDH1, PDK2 and PDHA1 (Fig. S4).

Figure 2.

CD4+ T cell expression of glucose metabolism genes. Relative gene expression of 82 genes encoding enzymes and isozymes in glucose metabolism pathways assessed by Fluidigm HD Biomark RT-PCR of RNA extracted from CD4+ T cells purified by magnetic bead negative selection. (A) Heatmaps depicting the relative expression of metabolic pathway gene sets as listed in Table S2. (B) CD4+ T cell glucose metabolism pathways highlighting enzymes encoded by genes with increased relative expression in HIV+ women with DM compared to controls. (C) Gene encoding enzymes with elevated relative expression in HIV+ women with DM compared to controls. Fold-change in gene expression calculated using 2^−ΔΔCt with RPL30 as reference gene and HIV+DM− as reference samples. Genes with 95% C.I. >1 or 95% C.I. <1 were assessed for differences between groups. Fold change was compared between groups using one-way ANOVA with Tukey post-hoc test for multiple comparison. A p < 0.05 was considered significant. HIV+DM+ women, n = 5; HIV+DMTx+ women, n = 10; HIV+DM− women, n = 6. Created with BioRender.com

Elevated metabolic activity in TCR-stimulated CD4+ T cells from HIV+ women with diabetes mellitus

To functionally validate the increased relative expression of glycolysis and OXPHOS genes in CD4+ T cell and increased proportions of GLUT1 expressing CD4+ T cells we utilized Seahorse extracellular flux as a proxy to measure CD4+ T cell metabolic activity. The acidification of the extracellular milieu and the depletion of oxygen over time provide an indication of the rates of glycolysis and OXPHOS, respectively. CD3/CD28-stimulated CD4+ T cells from HIV+DM+ and HIV+DMTx+ women showed increased ECAR and OCR compared to HIV+DM− women, pointing to increased glycolysis and OXPHOS respectively (Fig. 3A&B). This data agrees with the elevated expression of specific genes encoding rate-limiting enzymes in both the glycolytic and OXPHOS pathways.

Figure 3.

Glycolytic metabolism and oxidative phosphorylation of stimulated CD4+ T cells. CD4+ T cells were purified by magnetic bead negative selection and stimulated using anti-CD3/CD28 Dynabeads. (A) Extracellular acidification rate (ECAR) and (B) oxygen consumption rate (OCR) were evaluated. All measurements were normalized to cell count. HIV+DM+ women, n = 4; HIV+DMTx+ women, n = 4; HIV+DM− women, n = 4. Each sample had 3 technical replicates.

Discussion

In this study we demonstrate altered CD4+ T cell metabolism in HIV+ women with DM as evidenced by elevated GLUT1+CD4+ T cells, increased relative expression of key metabolic enzymes, and increased glycolytic and OXPHOS metabolic activity. Despite higher HbA1c and longer duration of DM, HIV+DMTx+ showed lower levels for many of these CD4+ T cell metabolic parameters as compared to HIV+DM+ women. Our data suggests that anti-diabetic medications may have a beneficial effect in reducing metabolic activation in HIV+ women with DM.

Differentiating PLWH with DM based on diabetic treatment status allows for a clearer understanding of the immunometabolic characteristics of CD4+ T cells because anti-diabetic medications such as metformin could potentially mask associations specific to HIV and DM. The majority of HIV+DMTx+ women in the current study were treated with metformin, which has been reported to be anti-inflammatory and to have direct effects on the mitochondria and OXPHOS machinery of cells [19–21]. Metformin directly inhibits complex 1 of the electron transport chain resulting in a decrease in OXPHOS which is important in the effector T cell immune response [19] and also suppresses the secretion of inflammatory chemokines from activated monocytes [21]. We show here that diabetic treatment is associated with significantly lower proportions of GLUT1+CD4+ T cells and a reduced expression of the TCA cycle genes pdk2, idh1 and mdh1B. This data suggests that diabetic treatment of HIV+ women with DM may partially correct CD4+ T cell immunometabolic dysfunction. Metformin treatment in PLWH has previously been shown to have beneficial immune effects and favorable metabolic alterations have been observed in metformin-treated PLWH without DM [22–24]. These data together with our findings suggest that anti-diabetics including metformin may be a beneficial treatment not only in PLWH with DM but also PLWH without DM.

Increased glycolysis is a hallmark of activated CD4+ T cells [14,25]. We identified that GLUT1 expression on CD4+ T cells of HIV+ women is associated with activation of these cells. This is in keeping with the fact that activated CD4+ T cells reprogram their metabolic machinery to increase the use of glucose in aerobic glycolysis to achieve function [25]. While we show elevated GLUT1+CD4+ T cells in HIV+DM+ women compared to controls and that the proportion of GLUT1+CD4+ T cells is positively correlated with the proportion of CD38+HLADR+CD4+ T cells, we did not observe any differences in activation as measured by CD38/HLA-DR co-expression. These data suggest that the CD4+ T cell metabolic programming is a more sensitive indicator of dysfunction than the commonly utilized activation markers CD38/HLA-DR. Along these lines, ART-treated PLWH have dramatically fewer CD38+HLA-DR+CD4+ T cells than untreated PLWH, whereas the difference in GLUT1+CD4+ T cells between these two groups is much less pronounced [14].

Gene expression analysis identified that CD4+ T cells from HIV+DM+ women had elevated HK3 expression compared to HIV+DM– women. Hexokinase is responsible for the first, irreversible, rate-limiting step of glycolysis which phosphorylates glucose to glucose-6-phosphate [26] and HK3 overproduction is associated with increased ATP production, preservation of mitochondrial membrane potential and increased mitochondrial biogenesis [27]. In addition to elevated HK3 gene expression and increased percentage of GLUT1+CD4+ T cells in HIV+ women with DM, we also observed a heightened ECAR for these stimulated cells, suggesting a functional increase in glycolysis. These data suggest that CD4+ T cells from HIV+ women with DM are poised for increased glycolytic activity even before stimulation, consistent with previous observations in human CD4+ T cells from persons without HIV [28].

Several key genes relating to the TCA cycle including genes encoding rate limiting enzymes (PDHA1) [29], enzymes determining substrate fate (IDH1) [30] and enzymes involved in redox balance (MDH1B) [31,32] were increased in HIV+ women with DM. Consistent with the elevated expression of these key TCA cycle genes, stimulated CD4+ T cells from HIV+ women with DM showed elevated OCR compared to controls. The increased expression of key TCA cycle genes and OCR are likely utilized to maintain high energy production and/or to facilitate other functions such as anabolic metabolism and cellular redox balance [30–32]. As these processes have been associated with activation and cellular survival and growth, their interrogation in the context of HIV persistence in persons with DM would be informative.

We observed an elevated expression of both pdha1 (a subunit of PDC) and pdk2, genes encoding enzymes with opposing effects for the fate of pyruvate, in HIV+DM+ women. While these findings seem incongruous, a possible explanation is the identification of non-canonical functions of metabolic enzymes. For example, the glycolytic enzyme GAPDH can affect IFN-γ production by CD4+ T cells through its interaction with IFN-γ mRNA [33]. In the case of the elevated expression of both pdha1 and pdk2 in CD4+ T cells, future work is needed to assess whether non-canonical functions of these metabolic enzymes are implicated in the setting of HIV and DM. Namely, previous studies have shown that translocation of PDC from the mitochondria to the nucleus generates acetyl-CoA, responsible for histone acetylation of genes implicated in cell growth and proliferation [34,35], which would be functionally relevant in activated CD4+ T cells.

Our study was limited to assessing the metabolic profile in circulating CD4+ T cells, but future studies would also need to examine tissue resident CD4+ T cells. These cells could reveal even more profound dysfunction as they are more likely to be in direct contact with stimulating factors. For example, adipose tissue has been identified as a key site in understanding the association between DM and HIV as this site could be an HIV reservoir and a contributor to inflammation leading to insulin resistance in DM [36,37]. Additionally, our findings of increased GLUT1+CD4+ T cells, elevated relative expression of metabolic genes including HK3 and IDH1 and heightened metabolic activity coincide with the metabolic characteristics of CD4+ T cells susceptible to HIV infection [38,39]. This points to the need for experiments to assess the contribution of residual HIV replication to metabolic dysfunction in CD4+ T cells from HIV+ women with DM.

In summary, CD4+ T cells from HIV+ women with DM show increased glucose metabolism, with evidence of increased glycolysis and OXPHOS. Our finding that anti-diabetic treatment of HIV+ women with DM may partially correct CD4+ T cell glucose metabolism identifies a potential future role for immunometabolism drug targeting in PLWH with DM that may also be of benefit to PLWH without DM.