Human PBMC’s were isolated using Lymphocyte separation medium (LSM 1077, PAA, Somerset, UK). PBMC’s were then derived by harvesting the cell interface. All procedures performed on human participants were in accordance with the ethical standards of the Helsinki Declaration and ethically approved by HRA and Health and Care Research Wales (HCRW) with IRAS project ID 236524, REC reference 18/LO/1814. Informed consent was obtained from all individual participants involved in the study.

To isolate CD4⁺ T cells, the RosetteSep™ Human CD4⁺ T Cell Enrichment Cocktail (STEMCELL Technologies UK Ltd, Cambridge, UK) was used as per the manufacturer’s instruction. Tregs were characterised as CD4⁺CD25⁺CD127^low and were enriched prior to cell sorting through positive selection with CD25 MicroBeads II (Miltenyi Biotech, Surrey, UK) following manufacturer’s instructions. Two distinct fractions were obtained: a CD25⁺ positive fraction enriched for Tregs and a CD25^- negative fraction enriched for conventional T cells (Teffs). Teffs were classified based on the expression of CD4⁺CD25^-CD127^high. Samples were then stained with mouse anti-human antibodies for CD4 (BD Pharmingen), CD127 (Biolegend) and CD25 (BioLegend) prior to cell-sorting into a highly pure population of CD4⁺CD25⁺CD127^low Tregs.

Isolated Tregs were expanded in vitro using the same protocol as our clinical-grade preparation, which was utilised in our previous clinical trials and publications (14, 16–18, 24, 27). Briefly, cells were cultured in X-Vivo (Lonza, UK) supplemented with 5% of Human Serum AB Male (BioWest, France) and 100 nM of rapamycin (LC-Laboratories, USA). Cells were then activated with anti-CD3/CD28 beads (ratio bead:cell of 1:1; Invitrogen, UK). IL-2 (1,000 IU/mL; Proleukin, Novartis, UK) was added at day 4 post activation and replenished every 2 days. Cells were re-stimulated every 10–12 days and used after 24 days from the first activation. Expanded cells were frozen and used when needed.

CD14⁺ monocytes were isolated from total PBMCs using CD14 MicroBeads (Miltenyi Biotech, Surrey, UK) following manufacturer’s instructions. Macrophages were generated from isolated CD14⁺ cells by culturing in RPMI 1640 (Gibco) supplemented with 10% FCS, 2mM L-glutamine (Gibco) and 1% penicillin/streptomycin (Sigma). M1-like macrophages (M_LPS) were generated in the presence of 10ng/mL GM-CSF (R&D systems) and M2-like macrophages (M_IL4) in the presence of 25ng/mL M-CSF (R&D Systems) using a 5-day culture period (see Table 1 ). Maturation was subsequently achieved by culturing M_LPS macrophages with 100ng/mL LPS (Sigma) and 20ng/mL IFNγ (R&D systems) and M_IL4 macrophages with 20ng/mL IL-4 (R&D systems) for 48h.

All the experiments were performed by co-culturing macrophages with autologous Tregs at a ratio of 1:1 in serum free RPMI for 24h. Where indicated macrophages were incubated for 1h with either the connexin mimetic peptide GAP27 (Cambridge Biosciences, Cat Number HY-P0139) at 300 µM or PKA inhibitor H89 at 5 µM before starting the coculture.

To assess the phenotypic effect of oxLDL on both M_IL4 and M_LPS macrophages, the cells were cultured for 24h either alone or in the presence of 10μg/mL of oxLDL (Invitrogen, cat. Number L34357) in either the presence or absence of Tregs.

To assess uptake of both LDL and oxLDL following the 24h co-culture period, native LDL (Dil-LDL; Invitrogen, cat number L3482) or oxLDL (Dil-oxLDL; Invitrogen, cat number L34358) complexed with a fluorescent lipophilic cationic indocarbocyanine dye (1,1’-Dioctadecyl-3,3,3’,3’-Tetramethylindocarbocyanine Perchlorate fluorescent dye; Ex 554/Em 571) were added at a concentration of 10µg/mL for 6h. After this incubation, the cells were detached using StemPro^®Accutase^®. Uptake was analysed using flow cytometry.

To assess whether the observed effects of the co-culture experiments were contact-dependent, Transwell inserts were utilised to spatially separate cell types. Macrophages were seeded into 24-well plates whilst Tregs were separated from the macrophage monolayer using 6.5mm Transwell with 0.4µm Pore Polyester Membrane Insert, Sterile (Corning).

To determine whether Tregs affected the reverse cholesterol transport, a cholesterol efflux assay was performed using the cholesterol efflux assay Kit from Sigma-Aldrich (Cat number MAK192) to the manufacture’s specifications. Both supernatant and cell lysate samples were read using a SpectraMax i3® (Molecular Devices) fluorescent plate reader (Excitation 485nm Emission 523nm). The rate of cholesterol efflux was calculated using C = [Fm/(Fm +Fc)] × 100% where: Fm = fluorescent intensity of supernatant Fc = fluorescent intensity of cell lysate. To test whether the observed effects of the co-culture experiments were contact-dependent, a Transwell insert was utilised to spatially separate cell types as previously described (28).

To determine whether any effects of T cells on oxLDL uptake by macrophages was specific to this molecule, a phagocytosis assay was performed. Macrophages were cultured for 24h in either in the presence or absence of T cells at a 1:1 ratio. Following this, cells were cooled to 4°C, 100 particles per macrophage of Alexa Fluor 488 conjugate Zymosan A (S. cerevisiae) (BioParticles, cat number Z23373) were added to the cells for 90 mins at 37°C. Cells were then washed five times with ice-cold PBS before being detached with StemPro^®Accutase^®. Zymosan uptake was analysed using flow cytometry.

Cytosolic cAMP concentrations were measured using the Complete cAMP ELISA Kit (Cat. No. ADI-900-163A, ENZO Life Sciences). Briefly, adherent macrophages and T cells from co-culture experiments were separated and washed three times with ice-cold PBS, then lysed in 0.1 M HCl containing 0.1% Triton X-100 (10⁶ cells/ml) for 10 minutes at room temperature. Lysates were centrifuged at ≥600 × g to pellet cellular debris, and the supernatants were collected for analysis. cAMP levels were normalized to total protein content according to the manufacturer’s instructions.

RNA-sequencing was performed at Novogene. RNA samples were subjected to polyA selection, library preparation and sequencing. Fastq files were trimmed and aligned using kallisto (29), and differential gene expression using DESeq2 (30). The RNA sequencing data has been deposited in GEO database (GSE265832) and can be accessed by token opebwewuzfidxmn. Differentially expressed genes were those that had a p-adjusted value (p-adj) of less than 0.05 (Supplemental Supplementary Table S1 ). Human genes related with cholesterol metabolism and macrophage differentiation were obtained from gene ontology data on AmiGO (release data 2024-01-17) (Supplemental Supplementary Table S2 ) (31–33). Enrichment (Supplemental Supplementary Table S3 ) was calculated employing a Fisher’s two-sided test considering 19,890 human protein coding genes as background (34). Heatmaps and volcano plot were done using R. All packages are available in CRAN.

Total RNA was extracted from cells deriving from coculture by using AllPrep DNA/RNA/Protein Mini Kit (QIAGEN, Cat Number 80004) according to the manufacturer’s instructions. RNA was reverse-transcribed to single-stranded cDNA using High-Capacity cDNA Reverse Transcription Kit (ThermoFisher, Cat Number 4368814). Quantitative RT-PCRs were performed using primers from Applied Biosystems: ATP-binding cassette A1 (ABCA1; Assay ID: Hs01059118_m1), ATP-binding cassette G1 (ABCG1; Assay ID: Hs00245154_m1), paraoxonase 1 (PON1; Assay ID: Hs00166557_m1) and ubiquitin C (UBC; Assay ID: Hs05002522_g1). Samples analysed by quantitative RT-PCR were assessed in triplicates on an Applied Biosystems cycler (ViiA7 Real-time PCR system) using the TaqMan™ Universal Master Mix II, with UNG (ThermoFisher Scientific, Catalog number: 4440038) according to manufacturer’s instructions. To quantify the data, the comparative threshold cycle method was used. Relative quantity was defined as 2^-ΔΔCt (35). UBC was used as reference gene.

Cells from co-culture experiments were washed 3 times in ice-cold PBS and lysed in Pierce RIPA buffer (ThermoFisher Scientific; Cat Number 89900) supplemented with Protease and Phosphatase Inhibitor Cocktail (ThermoFisher Scientific; Cat Number 78440). Protein concentration was determined by BCA assay (Pierce™, cat Number 23225). Equal amounts of total protein were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), transferred to polyvinylidene difluoride (PVDF) membrane and probed overnight at 4C with the respective antibodies: monoclonal anti-ABCA1 antibody (CellSignaling; Cat Number 96292), polyclonal anti-PON1 antibody (Proteintech; Cat Number 18155-1-AP), monoclonal anti-GAPDH antibody (CellSignaling; Cat Number 5174), recombinant monoclonal anti-ABCA1 (phosho-S2054) antibody (Abcam; Cat Number ab125064). Immunocomplexes were detected using enhanced chemiluminescence (GE Healthcare). All data were analysed with Image Lab Software (BIO-RAD, Hercules, CA, USA), and GAPDH was used as an internal control.

Statistical analysis was carried out using GraphPad Prism 9 (GraphPad Software Inc., USA). All measures of variance were expressed as mean ± standard deviation (SD). Datasets were compared using a t-test, one- or two-way ANOVA as indicated. Data were considered statistically significant with p<0.01, p<0.001 or p<0.0001 and represented on the figures as indicated.

Freshly isolated Tregs, Teff and Treg_exp have been phenotypically evaluated by flow cytometry using antibodies listed in Supplemental Supplementary Table S4 and following previously published procedures (16, 24). After detaching, monocytes were incubated with human TruStain FcX™ (Fc receptor blocking solution, Biolegend, USA) for 10 minutes and then stained with Fixable Viability Stain 780 (BD Biosciences, USA) and extracellular antibodies as listed in Supplemental Supplementary Table S4 for 30 minutes at 4 °C. Samples were acquired on LSR-Fortessa™ flow cytometer and files analysed using FlowJo™ 10.8.1 (BD Life Sciences, USA). Cytokine secretion was measured employing LEGENDplex™ (BioLegend) assay as per the manufacturer’s instructions using cellular supernatant. Samples were run using a BD FACSCanto™ flow cytometer. Data analysis was performed using LEGENDplex™ Data Analysis Software V8.0.

Regulatory T cells (Tregs) are known to play pivotal roles throughout the progression of atherosclerosis. Beyond suppressing inflammation and limiting immune-mediated damage to the arterial wall, Tregs also contribute to plaque stability and tissue repair processes (1, 36). Previous studies, including our own, have shown that Tregs can influence monocyte-to-macrophage differentiation, promoting a more tolerogenic phenotype (24, 26).

Given the impact of lipid accumulation on macrophage phenotype and function, we hypothesised that clinical-grade ex vivo expanded Tregs (Treg_exp) could modulate macrophage responses and mitigate the detrimental effects of intracellular cholesterol buildup.

To generate Treg_exp, CD4⁺CD25⁺CD127^low T cells were isolated from the peripheral blood of healthy donors and expanded using anti-CD3/CD28 stimulation (1:1 bead-to-cell ratio), high-dose IL-2 (1,000 IU/mL), and rapamycin (100 nM), following protocols from our previous clinical trials (16–18). After two weeks, Treg_exp were characterised by flow cytometry, confirming high expression of CD25, FOXP3, CTLA-4, CD39, and TIM-3, along with low levels of CD127 and PD-1, hallmarks of a highly suppressive Treg phenotype (16, 37, 38) ( Figures 1A, B ).

To assess their suppressive function, Treg_exp were co-cultured with CFSE-labelled conventional T cells (Teff) activated with anti-CD3/CD28 beads. Flow cytometric analysis of CFSE dilution demonstrated potent suppression of Teff proliferation, consistent with our previous findings (14, 16, 24).

We next examined whether Treg_exp retained their suppressive capacity in an atherogenic environment. Increasing concentrations of oxLDL (1, 5, and 10 µg/mL) did not impair their ability to suppress Teff proliferation ( Figure 1C ). Supporting this, Treg_exp did not uptake Dil-labelled oxLDL ( Figure 1D ), likely due to the absence or very low expression of key oxLDL receptors such as CD36, CD204, and LOX-1 ( Figure 1E ).

To explore the impact of Treg_exp on macrophage function under atherogenic conditions, we first generated macrophages using established differentiation protocols (24). Monocytes were cultured with either LPS, IFNγ, and GM-CSF or IL-4 and M-CSF (yielding M_LPS and M_IL4 cells respectively; see Table 1 ; Supplementary Figure S1 ). These phenotypes differed in their surface marker expression ( Figure 2A ; Supplementary Figure S2 ). Notably, M_IL4 cells showed higher levels of oxLDL receptors (CD36, CD204, LOX-1) and demonstrated greater oxLDL uptake compared to M_LPS ( Figure 1D ) or native LDL ( Figure 2B ).

Treg_exp were re-activated with anti-CD3/CD28 beads and cultured with IL-2 (500 IU/mL), which was necessary to sustain Treg activity but did not alter the M_IL4 phenotype ( Supplementary Figure S3 ). For the coculture experiments, we used a 1:1 Treg-to-macrophage ratio. Although this ratio is supraphysiological and does not reflect in vivo conditions, it was chosen in line with previous mechanistic coculture studies, where comparable proportions were required to observe measurable effects of Tregs on macrophage phenotype and function. Under these conditions, Treg_exp modestly increased CD14 expression on M_IL4 in the absence of oxLDL, while CD16, CD80, CD86, CD163, CD206 and scavenger receptor expression remained largely unchanged ( Figures 2C, D ).

In the presence of oxLDL, M_IL4 cells underwent phenotypic changes, characterized by increased expression of CD16 and CD80, and decreased levels of CD163 and CD36. Notably, M_IL4 cells pre-treated with oxLDL were able to expand a population of T_eff cells that produced significantly more IFNγ compared to stimulation with untreated M_IL4 cells ( Figure 2C ; Supplementary Figure S2D ). Co-culture with Treg_exp attenuated these changes, restoring CD16 and CD80 expression toward baseline and counteracting the reduction of CD36 ( Figures 2C, D ).

To confirm the specificity of this effect, we replaced Treg_exp with T_eff in parallel co-cultures. Unlike Treg_exp, T_eff promoted a shift toward a pro-inflammatory M_LPS-like phenotype, characterised by reduced CD14 and increased CD80 and CD86 expression ( Supplementary Figure S4A ), without altering scavenger receptor levels.

Additional analyses supported a broader influence of Tregs on maintaining M2-like features. Furthermore, analysis of culture supernatants revealed the presence of IL-10 in Treg–macrophage cocultures, both with and without oxLDL, consistent with the induction of a more tolerogenic environment ( Figure 2E ). These observations are in agreement with our previous work which showed that Tregs can drive macrophages toward an IL-10–producing, alternatively activated phenotype (24).

Collectively, these findings indicate that Tregexp mitigate oxLDL-induced pro-inflammatory changes in macrophages and help maintain features of the M_IL4 phenotype, supporting the concept that Tregs can modulate macrophage responses under conditions of lipid stress in vitro.

To further explore the effect of Treg_exp on the M_IL4 macrophages during the co-culture, their transcriptome profile was investigated.

M_IL4 were cultured alone or in the presence of Treg_exp (ratio 1:1) for 24 hours. Then, medium supernatant (containing Treg_exp in the coculture condition) was removed and M_IL4 cells were detached using Accutase™ and further purified by removing any T cell contaminants by fluorescence-activated cell sorting. Total RNA was then isolated from purified M_IL4 cells and sequenced using Illumina technology.

The analysis of the transcripts of M_IL4 cultured either alone or in the presence of Treg_exp revealed 2,395 differentially expressed genes (DEG; p-adj <0.05, which provide insights into how Treg_exp influenced the M_IL4 cellular transcriptome ( Figure 3A ; Supplementary Table S1 ).

To interpret the biological significance of the observed gene expression changes in lipid-related or mediated processes, we sought for pathways present in our DEG list that related to cholesterol metabolism, localisation, and macrophage differentiation. We restricted the gene ontology analysis to the biological processes ( Supplementary Table S2 ) where 10 or more of our DEGs were present ( Figure 3B ) and performed an enrichment analysis of those pathways over the expected frequency in the human transcriptome. The analysis revealed that the presence of Treg_exp was affecting the expression of genes in M_IL4 cells involved in the biological processes associated not only with the macrophage differentiation but also with the localisation and transport of cholesterol ( Figure 3B , Supplementary Figure S5 and Supplementary Table S3 ). The Gene Ontology pathways associated to “Regulation of macrophage differentiation”, “Regulation of lipid localization”, “Regulation of macrophage foam cell differentiation” shown in Figure 3C along with “Negative regulation of transport activity” and “Cholesterol import” ( Supplementary Figure S6 ) were all significantly enriched with genes affected by the presence of Tregs ( Figure 3C ).

These findings prompted us to investigate the capacity of Treg_exp to regulate the increase of oxLDL within M_IL4. We first analysed the capacity of M_IL4 to accumulate oxLDL in the presence of Treg_exp by adding Dil-oxLDL to the co-culture. As shown in Figure 4A , M_IL4 cultured with Treg_exp showed a significantly lower oxLDL accumulation than in the absence of Treg_exp (approximately 40% reduction, p = 0.03). Reducing the number of Treg_exp decreased the inhibition in oxLDL uptake ( Supplementary Figure S2C ). This was an effect specifically associated with oxLDL, as no differences were observed in the uptake of fluorescent zymosan particles in the same culture conditions ( Figure 4A ).

It has been previously published that under homeostatic conditions, macrophages remove the excess of cellular cholesterol via reverse cholesterol transport (RCT) (39). We thus investigated whether the reduced accumulation of oxLDL observed in the presence of Treg_exp ( Figure 4A ) could be due to increase cholesterol efflux. After 24h of co-culture with Tregs_exp, the cholesterol efflux in M_IL4 was significantly increased ( Figure 4B ). This effect was Treg-specific since M_IL4 co-cultured with Teff did not show the same outcome ( Figure 4B ).

To investigate the molecular mechanisms used by Treg_exp to reduce cholesterol accumulation, we tested whether Tregs required close contact with the target cell or produced soluble factors to affect macrophages. To address this question Treg_exp and M_IL4 were spatially separated by a porous membrane insert. The results in Figure 4C showed that when cell-cell contact between the two cells was prevented the increased cholesterol efflux was abolished ( Figure 4C ). Similarly, the previously observed decrease of oxLDL accumulation in M_IL4 due to the co-culture with Treg_exp was abolished by the same spatial separation ( Figure 4D ).

Altogether these findings demonstrate that Treg_exp can decrease the accumulation of oxLDL in M_IL4 by favouring its efflux in a cell contact manner.

To understand which molecules were involved in the cholesterol efflux enhanced by Treg_exp, we investigated the RCT system and focused on the membrane adenosine triphosphate (ATP)-binding cassette (ABC) transporters. M_IL4 were co-cultured with Treg_exp, or T_effs as a control. After 4h, macrophages were separated from the other cells, lysed and mRNA extracted. Then, the gene expression of ABCA1 and the other gene involved in the RCT system such as ABCG1 was quantified by qRT-PCR. Results confirmed the RNA sequencing data of the increase of ABCA1 gene expression in M_IL4 co-cultured with Treg_exp when compared to M_IL4 alone (2.5-fold increase) and this difference was significantly higher than the increase observed in the same cells co-cultured with T_effs (1.4-fold increase) ( Figure 5A ).

The production and transfer of cAMP into the target cell has been described as one of the mechanisms used by Tregs to suppress T cell proliferation (1, 12). As cAMP can stimulate ABCA1 gene expression and enhance cholesterol efflux in human fibroblasts and THP-1 or RAW264.7 macrophages (40, 41), we investigated whether Treg_exp could use this mechanism to influence the same molecular pathway in M_IL4. We analysed cAMP in Treg_exp alone and in M_IL4 co-cultured or not with Treg_exp. The results in Figure 5B show elevated levels of cAMP in Treg_exp, and a noteworthy rise in intracellular cAMP in M_IL4 when co-cultured with Treg_exp compared to when these cells are cultured independently. In contrast, co-culturing M_IL4 with T_effs did not show any increase in cAMP ( Supplementary Figure S4B ). To determine if cAMP transfer by the Treg_exp to macrophages depends on gap junctions as described in the suppression of conventional CD4⁺ T cell proliferation (42), we performed the same co-culture experiment described above in the presence of the mimetic peptide GAP27 to block gap junctions. The results showed that the preincubation of M_IL4 with GAP27 drastically inhibited the increase in ABCA1 mRNA transcript induced by Treg_exp ( Figure 5C ). Likewise, preincubating M_IL4 with the protein kinase A (PKA) inhibitor H89 led to a significant decrease in ABCA1 transcript levels ( Figure 5D ). Together, these findings showed that ABCA1 mRNA expression induced by cAMP transfer from Treg_exp to M_IL4 is controlled by the activation of PKA pathway and depends on gap junctions.

To confirm these results at the protein level, we analysed the cell lysates of macrophages co-cultured with either Treg_exp or T_effs by western blot. The analysis of ABCA1 expression in M_IL4 after 24h with Treg_exp showed a significant increase of this protein ( Figure 5E ). In contrast, the presence of T_eff did not produce any increase of ABCA1 ( Figure 5E ). ABCA1 activity is controlled not only at transcriptional level, but also through post-translational modifications. One of these modifications involves the phosphorylation of Ser-2054 of ABCA1 which protects the protein from its rapid degradation (43). The analysis of ABCA1 in Figure 5E shows that the presence of Treg_exp in the coculture induced the increase of ABCA1 phosphorylated in Ser-2054 in M_IL4. These findings support that Treg_exp regulate cholesterol efflux by increasing both the expression and protein stability of ABCA1.

Human PON1 is a high-density lipoprotein (HDL)-associated lipolactonase which contributes to the antioxidant function of HDL (44). In animal models, Pon1 has been reported to reduce macrophage oxidative stress, prevent LDL oxidation, inhibit cholesterol synthesis, and enhance cholesterol efflux, thereby conferring atheroprotective effects (45, 46). Furthermore, human studies have shown that PON1 gene expression and serum activity inversely correlate with cardiovascular disease risk (47, 48). To investigate whether Treg_exp influence PON1 expression, we analysed M_IL4 macrophages co-cultured with Tregs or Teff cells. After 4 hours, PON1 mRNA levels were significantly higher in M_IL4 co-cultured with Treg_exp compared to M_IL4 alone (1.74-fold increase), whereas no induction was observed in the presence of Teff ( Figure 6A ). Consistently, western blot analysis after 24 hours confirmed an increase in PON1 protein levels in Treg-treated M_IL4 ( Figure 6B ). We next explored whether the induction of PON1 followed the same cAMP-dependent mechanism identified for ABCA1 regulation. When M_IL4 were co-cultured with Treg_exp in the presence of the gap junction blocker GAP27 or the PKA inhibitor H89, PON1 mRNA induction was markedly reduced ( Figure 6C ). These results indicate that Treg_exp promote PON1 expression in macrophages through cAMP transfer and activation of the PKA pathway.

Given the diverse functions of PON1 in modulating oxidative stress and cholesterol metabolism, these observations highlight an additional mechanism by which Treg_exp can influence macrophage biology in vitro. While further in vivo validation is required, the ability of Treg_exp to induce PON1 suggests a potential contribution to creating a more protective macrophage phenotype under lipid stress conditions.