CD3⁺ T-cell: CD14⁺ monocyte complexes are dynamic and increased with HIV and glucose intolerance

Abstract

Persistent systemic inflammation is associated with an elevated risk of cardiometabolic diseases. However, the characteristics of the innate and adaptive immune systems in individuals who develop these conditions remain poorly defined. Doublets, or cell-cell complexes, are routinely eliminated from flow cytometric and other immune phenotyping analyses, which limits our understanding of their relationship to disease states. Using well-characterized clinical cohorts, including participants with controlled human immunodeficiency virus (HIV) as a model for chronic inflammation and increased immune cell interactions, we show that circulating CD14⁺ monocytes complexed to CD3⁺ T cells are dynamic, biologically relevant, and increased in individuals with diabetes after adjusting for confounding factors. The complexes form functional immune synapses with increased expression of proinflammatory cytokines and greater glucose utilization. Furthermore, in persons with HIV, the CD3⁺ T cell: CD14⁺ monocyte complexes had more HIV copies compared to matched CD14⁺ monocytes or CD4⁺ T cells alone. Our results demonstrate that circulating CD3⁺ T-cell: CD14⁺ monocyte pairs represent dynamic cellular interactions that may contribute to inflammation and cardiometabolic disease pathogenesis. CD3⁺ T-cell: CD14⁺ monocyte complexes may originate or be maintained, in part, by chronic viral infections. These findings provide a foundation for future studies investigating mechanisms linking T cell-monocyte cell-cell complexes to developing immune-mediated diseases, including HIV and diabetes.

Open in new tabDownload slide

Introduction

Chronic inflammation is linked to metabolic diseases like diabetes and atherosclerosis, which can worsen in people living with human immunodeficiency virus (PLWH) due to elevated cytokine and chemokine levels despite antiretroviral therapy (ART).^1–8 Traditional analysis of immune cells using flow cytometry often misses interacting cell populations, dismissing cell aggregates as artifacts of sample processing.⁹ However, recent evidence suggests the presence of significant immunologically relevant cell-cell complexes in various disease states, including tuberculosis, and after vaccination with immunogenic vaccines such as the yellow fever vaccine.^10–12 Importantly, these cell-cell complexes are not artifactual from cryopreservation of peripheral blood mononuclear cells (PBMCs), with a strong correlation between these complexes from fresh and cryopreserved samples.¹⁰ However, none of these studies have shown an association between cell-cell complexes and metabolic disease. Our study investigates these complexes in the context of controlled HIV, using it as a model to understand chronic immune activation's effects on diseases like diabetes and cardiovascular disease.

Methods

Study participants

All PLWH included in this study were previously recruited for the HIV, Adipose Tissue Immunology, and Metabolism cohort at the Vanderbilt Comprehensive Care clinic. The cohort has individuals without diabetes (hemoglobin A1c [HbA1c] <5.7% and fasting blood glucose [FBG] <100 mg/dl), with prediabetes (HbA1c 5.7–6.4% and/or FBG 100–125 mg/dl), and diabetes (HbA1c ≥6.5%, FBG ≥ 126 mg/dl and/or on medications to treat diabetes). All PLWH were on ART with sustained viral suppression for at least 12 months before the study, with a CD4⁺ T cell count > 350 cells/ml (Table S1). The cohort excluded individuals with inflammatory illnesses, substance abuse, greater than 11 alcoholic drinks per week, and active hepatitis B/C. The study is registered at ClinicalTrials.gov (NCT04451980).² The second cohort has ten adults without HIV who were enrolled in ongoing research to understand the role of immune cells in aging and cardiovascular disease (CVD) (Table S2). All studies were approved by the Vanderbilt University of Medicine Institutional Review Boards. Participants provided written informed consent. The studies were conducted using the United States Department of Health and Human Services guidelines.

Sex as a biological variable

Our study examined males and females and similar parameters and measured and reported for both sexes.

Mass cytometry

Mass cytometry was conducted on cryopreserved PBMCs using a validated 37-marker antibody panel (Table S3). PBMCs were stained for live/dead cells with Cisplatin, surface markers with a master mix, and fixed in paraformaldehyde (PFA). Post-fixation, cells were stored in methanol at −20°C, later stained with intracellular markers for 20 minutes at room temperature, followed by the addition of 2ul (250 nM) of DNA intercalator (Ir) in phosphate-buffered saline (PBS) with 1.6% PFA. Just before running and analyzing the samples on the mass cytometer, we washed the cells in PBS, followed by Millipore water. For analysis, we resuspended 500,000 cells/ml of Millipore water. The ¹/₁₀ volume of equilibration beads were added to the cells, which we then filtered and analyzed on the Helios. FCS files from the Helios cytometer were bead-normalized using the premessa R package's normalizer GUI method.¹³ FCS files were analyzed in Flowjo to clean the data of debris (DNA-), Fluidigm beads (175++165++), and dead cells (cisplatin⁺). Data gating was performed using Flowjo (Fig. S1), followed by downstream analysis in R programming language (version 4.2.1) and the flowCore package.¹⁴ We downsampled all samples and processed them through the following workflow: Subset parameters were transformed using the function asinh(x/5). A nearest neighbor search produced a weighted adjacency matrix with several nearest neighbors set to the dimension of subset + 1.¹⁵ The Leiden community detection algorithm was used to cluster the adjacency matrix.¹⁶ Uniform Manifold Approximation and Projection (UMAP) was performed for subset visualization using the uwot R package.¹⁷

Tracking responders expanding populations (T-REX) and marker enrichment modeling (MEM) of enriched features

The T-REX algorithm was performed as published.¹⁸ Briefly,we classified cells of interest measured using mass cytometry, including CD45⁺ cells, CD3⁺ CD4⁺ T cells, CD3⁺ CD8⁺ T cells, CD3^- CD19^- HLA-DR⁺ monocytes, and CD3^- CD56⁺ CD16⁺ NK cells. UMAP analyses were performed for concatenated non-diabetic participants (group 1) and prediabetic/diabetic participants (group 2). This was followed by K nearest neighbor (KNN) analyses to search for the nearest neighbors for each cell. The difference in the percentage change per cell between group 1 and group 2 was calculated based on the abundance of these cells in each group in the KNN region. ≤5% and ≥95% changes in cell percentages were considered significant, which we clustered using Density-Based Spatial Clustering of Applications with Noise (DBSCAN). The phenotype of the clusters that were significantly different between the groups was determined using the MEM package.

Flow cytometry

PBMCs were stained with fluorescently tagged antibodies as previously published.¹⁹ Briefly, thawed and washed PBMCs were stained with Aqua (Live/Dead marker) for 10 minutes at room temperature, followed by the addition of a mastermix of fluorescently tagged antibodies (Table S4, BD Aria). Stained cells were analyzed using a 4-laser BD FACS ARIA III with a cell sorter. The two-dimensional gates used to sort CD3⁺ T cell: CD14⁺ monocyte complexes are shown (Fig. S2A, Table S4). The stained PBMCs were resuspended in PBS with RNA later to stabilize the RNA (ThermoFisher #AM7022). Single-cell indexed sorting was done using a 100 µM nozzle to sort CD3⁺ T cell: CD14⁺ monocyte complexes as a single entity into each 96 well plate, as previously published.²⁰ For bulk sequencing, we performed a 4-way sort through a 100 µM nozzle into four 1.5 ml Eppendorf tubes (CD4⁺ T cells, CD8⁺ T cells, CD14⁺ monocytes and CD3⁺ T cells: CD14⁺ monocyte complexes). A separate panel was used to identify cell-cell complexes that include dendritic cells (Table S3). PBMC preparation and staining were done as above. The gating strategy used to identify T cell-DC complexes is shown in Fig. S2. We did not gate out doublets, as per traditional flow cytometry gating.

Single-cell ENergetIc metabolism by profiling translation inhibition (SCENITH) assay

PBMCs were prepared for SCENITH as published.²¹ We added 10 µl of inhibitors (oligomycin [1.5 µM], 2DG [100 mM], 2DG + Oligomycin) to the cells. Media only was included as a control. All samples were then incubated at 37 °C for 30 minutes. We then added puromycin (10 µM) to each condition (PBMCs with inhibitors) and incubated the PBMCs at 37 °C for 45 minutes. After this, we washed the PBMCs in PBS, stained them with surface antibodies against CX3CR1 and CCR7, and stained them at 37 °C for 15 minutes. The cells were then incubated with the master mix containing the other surface markers (Cytek antibodies, Table S2) for 20 minutes at room temperature. The cells were fixed with 4% PFA for 15 minutes at room temperature. We added 0.1% triton permeabilization solution and incubated the cells for 15 minutes. Anti-puromycin in the permeabilization buffer was added to the cells for 15 minutes at room temperature. The cells were then washed and resuspended in PBS for analysis with Cytek Aurora.

Droplet digital polymerase chain reaction (PCR)

We sorted CD3⁺ CD4⁺ T cells, CD14⁺ monocytes, and CD3⁺ CD14⁺ T cell-monocyte complexes into separate Eppendorf tubes with PBS. Cells were pelleted and resuspended in lysis buffer [Tritonx100 (0.1%), Tris HCL (10 mM), and Proteinase K (400ug/ml)] at 55 °C for 10 hours. Additional proteinase K was added during the heat inactivation stage at 95 °C for 5 minutes. For HIV DNA quantitation, we used LTR primers (forward primer -LTR 5’-AGC ACT CAA GGC AAG CTT TA-3’, and reverse primer -LTR 5’-TGT ACT GGG TCT CTC TGG TTA G-3’, and probe 5’-FAM-GCA GTG GGT TCC CTA GTT AGC CAG AGA G-3IABkFQ-3’).²² HIV transcripts were quantified as copies/million cells. 19 µl of the ddPCR SuperMix (LTR primers & RPP30 housekeeping gene primers and probes), and 6ul of cell lysates were mixed and aliquoted per well (96-well twin tec plate) and droplets generated with an AutoDG. Droplets were read using a plate reader, and the positive droplet threshold was manually set using the negative droplet control (media only).

Time-lapse imaging

CD3⁺ CD14⁺ T cell-monocyte complexes were sorted as above and resuspended in RPMI with 10% fetal bovine serum (FBS). The cells were then plated on poly-L-Lysine pre-coated coverslips at a density of 15,000 to 40,000 complexes per 100 µl media. The cells on the coverslip were placed in a 24-well plate, and time-lapse imaging was captured using an EVOS M5000 imaging system. Image J Version 1.53t 24 August 2022 was used for image analysis.

Single-cell T-cell receptor (TCR) sequencing

Single-cell TCR sequencing involved sorting CD3⁺ CD14⁺ T cell-monocyte complexes, storing them at -80°C, and using uniquely tagged primers for reverse transcription.²⁰ cDNA amplification was performed with KAPA HiFi HotStart ReadyMix (Roche, Basel, Switzerland).^23–25 TCR gene expression was quantified via UMIs and nested PCRs targeting TCRαβ genes. After pooling and purifying the products, indexed sequencing libraries were created using Truseq adapters and quantified with the Jetseq qPCR Library Quantification Kit (Meridian Biosciences Inc., Ohio, USA). Samples were sequenced on an Illumina MiSeq with paired-end reads, quality-filtered, and demultiplexed. Reads were assigned to TCRA and TCRB loci and TCR clonotypes using MIXCR software,²⁶ with data visualization by the Visual Genomics Analysis Studio (VGAS).²⁷

Transmission electron microscopy (TEM)

CD3⁺ CD14⁺ T cell-monocyte complexes, CD3⁺ T cells, and CD14⁺ monocytes were sorted as above. For day 3 samples, we added RPMI media supplemented with human IL-2 [10 ng/ml]. The cells were plated on a poly-L-lysine coated coverslip for 1 to 2 hours for doublet imaging. When the cells were bound to the coverslip, the media was aspirated, and then the cells were fixed with 2.5% glutaraldehyde solution in 0.1 M sodium cacodylate buffer.²⁸ After secondary fixation, samples were washed for 5 minutes with 0.1 M sodium cacodylate buffer (7.3 pH). Followed by two five-minute washes with diH₂O. While keeping all solutions and plates at room temperature, the samples were then incubated with 2.5% uranyl acetate, diluted with H₂O, at 4 °C overnight. The samples were dehydrated using an ethanol gradient series. After dehydration, the ethanol was replaced with Eponate 12™ mixed in 100% ethanol in a 1:1 solution, then incubated at room temperature for 30 mins. This was repeated three times for 1 hour using 100% Eponate 12™. The plates were finally placed in new media and cured in an oven at 70 °C overnight. The plates were cracked upon hardening, and the cells were separated by submerging the plate in liquid nitrogen. An 80 nm thickness jeweler’s saw was used to cut the block to fit in a Leica UC6 ultramicrotome sample holder. The section was placed on formvar-coated copper grids counterstained in 2% uranyl acetate for 2 mins. Then, the grids were counterstained by Reynold’s lead citrate for two minutes. TEM acquired images on either a JEOL JEM-1230, operating at 120 kV, or a JEOL 1400, operating at 80 kV.²⁹

Single-cell RNA (scRNA) sequencing

PBMCs were thawed, washed with PBS, and incubated with Fc receptor-blocking solution. Surface antibody staining was performed (CD3 clone UCHT1 #300479, CD4 clone SK3 #344651, CD8a clone SK1 #344753, CD14 clone 63D3 #367137, CD16 clone 3G8 #302065, CD69 clone FN50 #310951), followed by encapsulation and barcoding using the Chromium Single Cell 5’ assay. Library preparation, cDNA amplification, and sequencing were done, aligning reads to the human genome. Cell identification and downstream analyses, including feature selection, PCA, and UMAP, were executed using Seurat V4.³⁰ Cells with abnormal gene or mitochondrial counts were filtered out. DoubletFinder identified doublets for exclusion. Differential gene expression analysis was conducted on sorted cell populations using WebGestalt (WEB-based Gene SeT Analysis Toolkit).³¹

Statistical analysis

This cross-sectional study establishes whether T cell-monocyte complexes are associated with metabolic disease variables and outcomes in PLWH. We reported summary statistics of clinical demographic characteristics using medians and interquartile ranges. The Wilcoxon test was used to examine differences for continuous variables, and Pearson χ² test was used for categorical variables. We selected partial Spearman correlation for analysis because it is less sensitive to outliers. We used a nonparametric test, partial Spearman’s correlation analysis, to test the relationship between T cell-monocyte complexes and clinical variables, including hemoglobin A1C, fasting blood glucose, high-density lipoprotein (HDL), low-density lipoprotein (LDL), triglycerides, coronary arterial calcium, and fat volume (pericardial, subcutaneous, and visceral). We adjusted for possible confounders that could influence the relationship between the immune complexes and the outcomes. These included age, sex, and body mass index (BMI). We selected partial Spearman correlation because it is less sensitive to outliers. Similarly, we used partial Spearman’s correlation analysis to test the relationship between T cell-monocyte complexes and plasma cytokines. We adjusted for possible HIV-related confounders that could influence the immune complexes' relationship with the plasma cytokines. These included CD4: CD8 ratio, hemoglobin A1C, and duration of years on ART.

Other statistical analyses comparing two continuous variables were performed using the Mann-Whitney U and Kruskal-Wallis tests, where more than 2 variables were compared. Statistical analysis in this study was performed in Graph Pad Prism version 9.5.0 and R version 4.2.1. Details of transcriptomic analysis above under single-cell sequencing.

Results

Characteristics of PLWH

The HIV cohort comprised 38 individuals on ART with long-term suppression of plasma viremia: 14 without diabetes and 24 with prediabetes or diabetes (Table S1). Details of the cohort and clinical visit procedures were previously published.¹⁹ In downstream analysis, PLWH with prediabetes and diabetes were combined into a single metabolic disease group. The characteristics of the groups were largely similar, except for parameters linked to glucose intolerance that we have highlighted and adjusted for in the downstream analysis. These include body mass index (P < 0.05), waist and hip circumference (P < 0.05, P = 0.01 respectively), and fasting blood sugar (P < 0.001). Glucose-tolerant individuals were younger by about ten years of age (P = 0.1). No notable differences were observed in HIV-related laboratory values (CD4 at ART start, CD4 at T cell enrollment, current ART, duration on ART, and hepatitis C antibody status. Cell-associated DNA and RNA were higher in PLWH without diabetes but not statistically significant (P = 0.1). Similarly, visceral fat volume was higher with glucose intolerance but insignificant (P = 0.1). Lastly, 33% of PLWH with diabetes had coronary arterial calcium (CAC), while none of the participants without HIV had CAC (P = 0.02). The differences between PLWH with and without glucose intolerance included known risk factors associated with metabolic disease, including age, BMI, hip/waist circumference, fasting blood glucose, and CAC prevalence.

Circulating cells of the innate and adaptive immune system differ by metabolic health

We used mass cytometry to examine immune cells in cryopreserved PBMCs of all PLWH using markers and gates depicted in the workflow (Fig. S1A–C). We identified six primary clusters, including CD4⁺ T cells, senescent/cytotoxic CD4⁺ T cells,¹⁹ CD8⁺ T cells, monocytes, B cells, and NK cells (Fig. 1A). A comparison of all cell clusters (abundance/size differences) between PLWH with diabetes/prediabetes and those without revealed several clusters in participants with glucose intolerance that were fewer in PLWH without diabetes depicted with the red dotted circles, all P < 0.05 (Fig. 1B). Other cell types that were more abundant in PLWH without diabetes included classical monocytes, CD14⁺ CD16^+/- Monocytes, and CD4⁺ T regulatory cells(Table S5).

The heatmap in Fig. 1C represents the median relative expression of immune markers on clusters and the median fold difference in cluster sizes. Magenta clusters were more abundant in prediabetic/diabetic PLWH, whereas CD4⁺ T regulatory cell clusters marked with a blue oval were more abundant in non-diabetic PLWH. CGC⁺ CD4⁺ T cells, a population we have previously reported as associated with metabolic and cardiovascular disease conditions in controlled HIV, were also increased with diabetes.^{2,  19,20} As previously published, we calculated the association constant between T cells/B cells and monocytes for the cell-cell complex clusters.¹⁰ To calculate the constant, we divided the proportion of the cell-cell complex by the proportion of T cells multiplied by the proportion of monocytes as published (Fig. 1D). We also individually analyzed CD4 (Fig. S2A) and CD8 (Fig. S3S) T cell clusters. Similarly, we can identify T cell-monocyte clusters among CD4 and CD8 T cells using unbiased approaches. The markers expressed on the largest CD4⁺ T cell: CD14⁺ monocyte cluster included FOXP3⁺⁵ CTLA4⁺³ CRTH2⁺² CD27⁺² CD57⁺² CD14⁺² (Fig. S2B–D). A similar analysis with CD8⁺ T cell markers expressed on the largest CD4+ T cell: CD14⁺ monocyte cluster included FOXP3⁺⁴ CTLA4⁺³ CD57⁺³ CD28⁺² CD27⁺² CD14⁺² (Fig. S3B–D). Note that CD28 was significantly lower in CD4 T cell clusters.

In a separate analysis, we included markers that allowed us to look for dendritic cell/T cell complexes (Fig. S4A). We observed fewer DCs complexed with T cells than monocyte/T cell complexes in PLWH and PWoH with diabetes (Fig. S4B). CD209 expression (Z axis) is shown on gated CD3⁺ T cell CD1c⁺ complexes, displayed in ring shape with larger diameters (Fig. S4C). Other markers of T cells (CD3, CD4, CD8), monocytes (CD14, CD16), B cells (CD19), activation and antigen presentation (HLA-DR), and additional phenotypic markers (CD206, CD209, CXCR4, CD207, CD169) are also shown for different cell clusters in the histogram. The geometric mean expression of each marker is displayed as a heatmap (Fig. S4D). We also detected the cell-cell complexes marked by arrows in a 3-dimensional UMAP representation of the cells from this analysis (Fig. S5A, B).

T cell-monocyte complexes are increased with glucose intolerance

T-REX workflow, an unbiased machine learning approach, was used to visualize distinct cell populations based on diabetes status and marker enrichment modeling.^{18,  32} The UMAP displays clusters that differ between non-diabetic (blue) and prediabetic/diabetic (red) PLWH (Fig. 2A). Complex clusters 1, 5, 7, and 8 remained significantly higher in PLWH with pre-diabetes/diabetes, and clusters 9 and 10 were higher in PLWH without diabetes (Fig. 2B). The expression of CD14⁺ in clusters outside of the monocyte population was confirmed, and other markers like FOXP3 and CTLA4 defined the clusters with cell-cell complexes (Fig. 2C). Due to their larger proportion, we focused on CD3⁺ T cell: CD14⁺ monocyte complexes. Additional studies using flow cytometry validated these complexes (Fig. 2D). Notably, CD3 and CD14 markers were sufficient to help define the T cell-monocyte complexes (Fig. 2Di–iv).

We sorted CD14⁺ monocytes, CD3⁺CD4⁺ T cells, and CD3⁺ T cell: CD14⁺ monocyte complexes and used light microscopy to image the cells. Compared to CD14⁺ monocytes alone, the complexes had a higher proportion of cells with presumed immunological synapses due to the proximity of cells and flattening at the connecting points (Fig. 2E). In longitudinal analysis, the proportion of CD3⁺ CD14⁺ complexes increased significantly from the first to second visit in non-diabetic PLWH (Fig. 2F). The higher increase in PLWH without diabetes over time compared to PLWH with diabetes might point to a modifying effect of therapies used to treat diabetes. Finally, we analyzed CD3⁺ T cell: CD14⁺ monocyte complexes among the 6 PLWH compared with 6 PWoH individuals with diabetes (Fig. S6A–C). We observed that metabolic syndrome, regardless of HIV status, was associated with the formation of these cell-cell complexes (Fig. S6D). The lack of significance between PLWH without diabetes and PWoH with diabetes is likely due to a small number of participants in the latter group. Further studies are needed to link complex formation and diabetes in PWoH.

We also examined singlet monocyte subsets and their connection to metabolic status (Fig. S7A). Most classical monocyte clusters were higher in PLWH without diabetes (Fig. S7B). Our analysis further revealed that PLWH with metabolic syndrome had more non-classical monocytes in their blood and some sub-clusters of classical monocytes expressing CX3CR1 than PLWH without diabetes (Fig. S7C). We also detected CD3⁺ T cell CD16⁺ monocyte complexes, separate from the NK cell cluster. We show these cells in three people with HIV and three without HIV (Fig. S8A–C). A comparison of other markers expressed by the complexes suggests that there might be CD56⁺ NK cells among the T cells complexed with intermediate monocytes. As expected, mean CD16 expression was highest on non-classical monocytes complexed with T cells. Both intermediate and non-classical monocytes had a higher mean expression of CXCR4 than T cells complexed with classical monocytes, which may play a role in HIV infection. While CD209 expression was similar among all monocyte complexes, the mean expression of CD207 was higher in intermediate and non-classical monocytes (Fig. S8C).

T cell-monocyte complexes in PLWH are positively associated with fasting blood glucose and hemoglobin A1C

Based on the observed differences in CD3⁺ T cell: CD14⁺ monocyte complexes by diabetes among PLWH, we posited that the cell-cell complex formation is promoted with glucose intolerance and influenced by factors associated with metabolic disease. To this end, we used partial Spearman rank correlation analysis to assess whether circulating cell-cell complexes identified by mass cytometry (Fig. 1) were associated with fasting blood glucose and hemoglobin A1C. We adjusted for age, sex, and BMI, 2 of which were different between PLWH with and without glucose intolerance (Table S1). CD8⁺ T cell-CD14⁺ monocyte complexes were associated with fasting blood glucose and hemoglobin. CD4⁺ T cell-CD14⁺ monocyte complexes were positively associated with hemoglobin A1C and triglycerides (Fig. 3A).

Systemic inflammation is associated with an increased risk of metabolic disease.³³ PLWH on antiretroviral therapy have elevated levels of plasma cytokines at baseline compared to persons without HIV. Among the 38 participants with HIV in this study, there were no differences in select plasma cytokines by metabolic group (Table S6). T cell-monocyte complexes were negatively correlated with circulating CD4⁺ T regulatory cells (Fig. 3A) and plasma IL-10 (Fig. S9). The negative correlation between T cell-monocyte complexes and circulating CD4⁺ T regulatory cells and IL-10 was modulated by blood glucose levels as determined by hemoglobin A1C as an interaction term (Fig. 3B, C). This indicates a diminished impact of IL-10 and CD4⁺ T regulatory cells on cell-cell complex formation as glucose increases. Overall, this suggests that there may be a greater tendency for complex formation with metabolic disease, yet the specific roles of the varied immune cells within these complexes remain undetermined.

T cell-monocyte pairs form stable and dynamic complexes with HIV

Time-lapse imaging (∼5 hours) revealed dynamic T cell-monocyte interactions (Fig. 4A, B, Videos S1 and S2), with some forming stable complexes (Fig. 4C, D) and others transient (Fig. 4E). Control experiments with CD3⁺ singlet T cells and CD14⁺ singlet monocytes showed no complex formation (Video S3). We sorted CD3⁺ T cell: CD14⁺ monocyte complexes and used TEM to view the interactions. T cells (∼7 to 12µm with large nuclei) and monocytes (15 to 18 µm) were identified by morphology (Fig. 4F-i, ii).^34,35 We also detected 100 nm viral-like particles in these complexes (Fig. 4G). Although our donors were on ART and with no detectable virus, their T cells and monocytes may still contain HIV viral reservoirs, as previously published.³⁶

CD4⁺ T cell-CD14⁺ monocyte complexes are more activated with higher proportions of TH17 cells than singlet CD4⁺ T cells

To better characterize the CD4⁺ T cells complexed with CD14⁺ monocytes, we first analyzed the memory subsets using CCR7 and CD45RO (Fig. 5A). CD14⁺ monocytes were predominantly complexed with CD4⁺ TCM and TEM cells (Fig. 5B). Several markers were used to define activated CD4⁺ T cells (CD137/OX40 and HLADR/CD38) (Fig. 5C). A significantly higher proportion of activated cells in prediabetic/diabetic PLWH than in non-diabetic CD4⁺ T cells (Fig. 5D, E, left panels). Focusing on cell-cell complexes, we observed that CD4⁺ T cell-CD14⁺ monocyte complexes had a higher proportion of activated cells (Fig. 5D, right panel). Irrespective of metabolic status, all cells within cell-cell complexes were HLADR⁺ CD38⁺ (Fig. 5E, right panel). Circulating activated T cells and cell-cell complexes were correlated with fasting blood glucose (Fig. 5F). We compared the activation profile between the CD3⁺ T-cell: CD14⁺ monocyte complexes and singlet T cells and found a higher proportion of CD137⁺OX40⁺ T cells in the cell-cell complexes (Fig. 5G). Using chemokine receptor markers, we defined CD4⁺ T helper subsets within the CD4⁺ T cells and CD4⁺ T cell-CD14⁺ monocyte complexes (Fig. S1C). CD3⁺ T cell:CD14⁺ monocyte complexes had significantly higher proportions of TH17 cells than CD4⁺ T cells (Fig. 5H). CD4⁺ T cell-CD14⁺ monocyte complexes from prediabetic/diabetic PLWH had a substantially higher proportion of TH2, TH17, and TH1 cells than non-diabetic PLWH (Fig. 5I). In summary, the T-cell monocyte complexes consist of activated immune cells enriched for TH17 memory subsets.

T cell-monocyte complexes in PLWH show higher HIV copies and gene expression related to activation and adhesion than singlet T cells and monocytes

We quantified HIV DNA in CD4⁺ T cells, CD14⁺ monocytes, and CD3⁺ T cell: CD14⁺ monocyte complexes from 6 PLWH on ART (selected based on higher proportions of complexes). Representative images show blue droplets (HIV LTR copies) and green droplets (RNAse P copies) (Fig. 6A–C). A higher count of HIV DNA copies per million cells was observed in CD3⁺ T cell: CD14⁺ monocyte complexes compared to paired single CD4⁺ T cells and CD14⁺ monocytes (Fig. 6D and E). Cells within these complexes express HIV-binding lectin and HIV entry receptors, which may facilitate the transfer of HIV to CD4⁺ T cells within the complexes (Fig. S10A and B).

To determine if the T cells in the complexes are clonally expanded, 4 PLWH with a more significant proportion of cell-cell complexes and paired αβ TCR chains were sequenced from sorted complexes (Fig. 6F). We used TCRmatch (http://tools.iedb.org/tcrmatch/) to predict the antigen specificity of the clonal TCRs among those identified. The majority of the clonal TCRs (with identical CDR3 > 2) were predicted to bind viral antigens, including HIV, in PLWH (Table S7). Herpes virus TCRs were predicted in both PLWH and PWoH.

PBMCs from PLWH with a large proportion of T cell-monocyte complexes as determined by Cytof and flow cytometry were processed for single-cell transcriptomic analysis. In addition, validation was based on CD3⁺ and CD14⁺ CITE-seq markers on the same cells, as expected. T cell-monocyte complexes had significantly more reads per cell than all singlet clusters (Fig. S11). For this sample, classical monocytes complexed with an almost equal representation of T cells with clonal and non-clonal TCRs. Non-classical monocytes, on the other hand, were mostly paired with clonal TCRs (Fig. 6G). Artificial complexes (a group of singlet monocytes and singlet CD3⁺ T cells combined for analysis) were compared to paired T cell-classical monocyte complexes from ten PWoH who consented to study immune cells in CVD (Clinical demographics in Table S2, Fig. 6H). Cell-cell complexes were compared from the same participant for each artificial complex group. Compared to artificial complexes, T cell-classical monocyte complexes expressed higher levels of GNLY with an overrepresentation of the MHCII antigen presentation and TCR signaling pathways, consistent with an activated inflammatory response (Fig. 6I and J, Table S8). Similarly, we compared T cell-nonclassical monocyte complexes to artificial complexes (Fig. S12A–C). GNLY and GZMH were also highly expressed. The comparisons of the T cell-monocyte complexes with paired artificial complexes in older PWoH individuals, showing high inflammation and antigen presentation, suggest that HIV does not entirely drive these complexes.

Maintenance of CD3⁺ T cell:CD14⁺ monocytes by oxidative phosphorylation

Changes in metabolism can be informative of the functional profile of immune cells.³⁷ While immune cells can rely on glucose and mitochondria for energy production, activated cells mostly rely on glycolysis.³⁸ We measured the energy dependencies of PBMCs from 15 PLWH (5 non-diabetic and 10 pre-diabetic and diabetic) ex vivo using SCENITH.²¹ Based on the puromycin uptake, CD3⁺ T cell:CD14⁺ monocyte complexes and CD14⁺ monocytes were more metabolically active than T cells, given their uptake of puromycin at baseline (Fig. 7A). The CD3⁺ T cell:CD14⁺ monocyte complexes had a higher mitochondrial dependence than the singlet CD14⁺ monocytes (Fig. 7B–D). Next, we quantified the proportion of persistent CD3⁺ T cells:CD14⁺ monocyte complexes after the incubation of PBMCs with metabolic pathway inhibitors (2DG, oligomycin). While the proportion of CD3⁺ T cell:CD14⁺ monocyte complexes was unchanged after inhibition of glycolysis with 2DG, there was a decrease with inhibition of oxidative phosphorylation (Fig. 7E). In summary, CD3⁺ T cell:CD14⁺ monocyte complexes use glycolysis and oxidative phosphorylation, though mitochondrial ATP synthesis may play a more significant role in maintaining cell-cell complexes.

Discussion

Our findings highlight the biological relevance of CD3⁺ T cell: CD14⁺ monocyte complexes in PLWH, particularly in glucose intolerance. Many of these complexes remain intact over a 4-hour time-lapse ex-vivo, suggesting they can form stable complexes. Furthermore, in one of the Videos, we capture a T cell complex with an antigen-presenting cell dividing (Video S2), suggesting that the complexes are biologically functional. Stable immune complexes with prolonged synapses may contribute to systemic inflammation greater than the contribution of single cells.³⁹ Importantly, the cell-cell complexes have increased expression of activation markers and inflammatory gene transcripts. Additionally, there is an inverse association between the cell-cell complexes, plasma IL-10, and CD4⁺ T regulatory cells.

IL-10 is an anti-inflammatory cytokine expressed by several cell types, including macrophages and regulatory T cells.⁴⁰ Several studies, including the Leiden 85-plus study, have shown that immune cells from individuals with metabolic syndrome expressed lower levels of IL-10 upon stimulation.⁴¹ HIV infection is associated with increased expression of IL-10, which in turn can suppress T cell responses.⁴² Over time and with ART, IL-10 expression decreases and may be associated with metabolic disease.^43,44 In this study, the relationship between the T cell-monocyte complexes and IL-10 and the interaction between hemoglobin A1C and IL-10 reinforces the notion that complexes are increased in the setting of inflammation.⁴⁵

In PLWH, replicating and integrated HIV can be detected in circulating monocytes.^46–49 Although some studies suggest that CD14^lo CD16⁺ non-classical monocytes are more prone to HIV infection,⁵⁰ we found that HIV is detected in CD4⁺ T cell-CD14⁺ monocyte complexes. Within tissues, macrophages infected with HIV can transmit HIV to CD4⁺ T cells, suggesting these cellular interactions may contribute to T cell loss and the establishment of HIV reservoirs.^51–53

A recent study on immunometabolism of CD4⁺ T cells in the context of HIV showed that the infectivity of the CD4⁺ T cells was more dependent on the metabolic activity of the T cells and less on the activation status.^54,55 They also showed fewer cells with latent HIV infection when they partially inhibited glycolysis using 2DG, suggesting that the steps required for HIV to establish latency are glucose dependent. In our study, CD3⁺ T cell:CD14⁺ monocyte complexes were metabolically active with greater dependence on glucose than oxidative phosphorylation. Forming these cell-cell complexes is an energy-demanding process, and inhibiting oxidative phosphorylation may have been sufficient to affect some of these immunological synapses.⁵⁶

Based on our observation, these complexes are metabolically regulated. The association between cell-cell complex abundance and A1C but not fasting blood glucose suggests that prolonged exposure to high glucose over a protracted time might drive complex formation (Fig. S13). Because monocytes circulate in the blood for a shorter period, we could also speculate that the complex formation is primarily driven by T cells, which can circulate for months.⁵⁷ If monocytes drive complex formation, then trained immunity could explain the relationship with A1C. Another important observation is the lack of a significant relationship between the complexes and visceral fat volume. This may point to factors present in circulating blood as drivers of complex formation in PBMCs. Future studies should investigate the formation of these complexes ex vivo and consider non-infectious factors that contribute to the formation of these inflammatory complexes.

In summary, CD3⁺ T cell:CD14⁺ monocyte complexes are increased in PLWH with glucose intolerance and may contribute to inflammation and viral persistence. The complex interplay between inflammatory and metabolic disorders makes these cells particularly interesting. Future studies investigating these cells in vivo and characterizing HIV within the cell-cell complexes will provide insight into their role in metabolic disease and complications that may arise from this.

Limitations of the study

This is a cross-sectional study with non-diabetic PWH and pre-diabetic/diabetic PWH (n = 14 and n = 24, respectively) and differs in variables associated with glucose intolerance, including age, BMI, and waist circumference. While we report an association between the cell-cell complexes and glucose intolerance, we cannot show causality in this study. Our study's cross-sectional design limits our ability to establish causality between CD3⁺ T cell:CD14⁺ monocyte complexes and metabolic disease progression. While the inflammatory cell-cell complexes are increased in persons with HIV with increased glucose tolerance and appear to carry HIV, we are currently unable to make conclusions as to whether these cell-cell complexes drive the pathogenesis of the metabolic disease or are a consequence of metabolic disease.

Acknowledgments

The graphical abstract was created using BioRender.

Author contributions

Conceptualization, C.N.W.; Methodology, C.N.W., L.M.O., J.S., J. O., X.Z., C.M.W., Q.S., S.P., J.S., L.M., M.M., S.A.M., R.G., H.B., A. H., E.W., K.N., T.A., Y.R.S., S.A.K., T.T., C.L.G., D.G.H., E.J.P., J.R.K., A.K.; Statistics, C.N.W., J.S., Q.S., J.S.; Formal Analysis, C.N.W., L.M.O., J.S., Q.S., J.S., J.O., C.M.W., R.G., A.C., S.P.; Investigation, C.N.W., L.M.O., C.M.W., J.R.K, A.K.; Resources, J.R.K., A.K., S.A.K.; Data Curation, C.N.W., S.B., Q.S., J.S., J. R. K.; Writing—Original Draft, C.N.W.; Writing—Review and Editing original draft., C.N.W., L.M.O., J.S., J. O., X.Z., M.M., S.A.M., H.B., A. H., E.W., S.A.K., T.T., D.G.H., E.J.P., A.K., J.R.K.; Visualization, C.N.W., S.P., J.W., S.B., Q.S., J.S., H.B., A.H.; Supervision, C.N.W, J.R.K., Project Administration, C.N.W., J.R.K.; Funding Acquisition, C.N.W., J.R.K.

Supplementary material

Supplementary material is available at The Journal of Immunology online.

Funding

Doris Duke CSDA 2021193 (CNW), K23 HL156759 (CNW), Burroughs Wellcome Fund 1021480 (CNW), 1021868.01 (AHJ) and Burroughs Wellcome Fund/ PDEP #1022376 (HKB), R01 DK112262 (JRK), R01HL144941 (AK), R03HL155041 (AK), the Tennessee Center for AIDS Research grant P30 AI110527 (SAM), KL2TR002245 (MM and EMW), K08AR080808 (EMW), The Myositis Association Pilot Award Grant (EMW), the Vanderbilt Flow Cytometry Shared Resource is supported by the Vanderbilt Ingram Cancer Center (P30 CA068485) and the Vanderbilt Digestive Disease Research Center (DK058404), National Center for Advancing Translational Sciences (NCATS) Clinical Translational Science Award (CTSA) Program, Award Number 5UL1TR002243-03, The UNCF/ Bristol-Myers Squibb (UNCF/BMS)- E.E. Just Postgraduate Fellowship in Life sciences Fellowship, NIH Small Research Pilot Subaward to 5R25HL106365-12 from the National Institutes of Health PRIDE Program, DK020593, Vanderbilt Diabetes and Research Training Center for DRTC Alzheimer’s Disease Pilot & Feasibility Program. CZI Science Diversity Leadership grant number 2022- 253529 from the Chan Zuckerberg Initiative DAF, an advised fund of Silicon Valley Community Foundation (AHJ).

Conflicts of interest

The authors have no competing interests.

Data availability

Gene expression data from this study have been deposited in the National Institutes of Health (NIH) Gene Expression Omnibus (GEO) accession numbers GSE229707 and GSE230276. Requests for further details on protocols and data included in this study are available upon request from the lead contact, [email protected].

References