Single-cell transcriptomics reveals markers of regulatory T cell dysfunction in Gestational Diabetes Mellitus

Single-cell transcriptomics reveals markers of regulatory T cell dysfunction in Gestational Diabetes Mellitus | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Single-cell transcriptomics reveals markers of regulatory T cell dysfunction in Gestational Diabetes Mellitus Panicos Shangaris, Nana Mensah, Athina Efthimiou, Nicoleta Mureanu, and 9 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-3773991/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 03 Apr, 2026 Read the published version in Communications Medicine → Version 1 posted You are reading this latest preprint version Abstract Gestational diabetes mellitus (GDM) is a common pregnancy complication associated with hyperglycaemia, chronic inflammation and adverse health outcomes. Regulatory T cells (Tregs) are thought to contribute to GDM due to their role in suppressing inflammation. However, it remains unclear whether specific Treg subsets are impaired in patients with GDM. To investigate transcriptional variation in GDM Tregs, we applied single-cell RNA sequencing to Tregs isolated from the blood of 13 healthy pregnant women and 10 patients with GDM. We identified naive and effector Treg subsets, none of which significantly differ in the proportion of cells captured from GDM and controls. We report a naive Treg subset with reduced expression of AP-1 transcription factor subunits in GDM, including JUN, FOS , and EGR1 , and an effector Treg subset with increased signalling of angiogenesis marker genes. Genes dysregulated in GDM Tregs independently predicted GDM status in pseudobulk and whole blood mRNA from independent cohorts. Remarkably, TXNIP , which regulates glucose levels, emerged as the most reliable standalone predictor in bulk mRNA (minimum AUC 0.7) equivalent to using body mass index (AUC 0.72) in our cohort. This study uncovers a disrupted molecular pathway in Treg cell subsets from GDM patients and proposes a panel of genes with translational potential as early disease biomarkers. Biological sciences/Computational biology and bioinformatics/Gene regulatory networks Biological sciences/Immunology/Immunogenetics Health sciences/Diseases/Endocrine system and metabolic diseases/Diabetes/Gestational diabetes Figures Figure 1 Figure 2 Figure 3 Figure 4 Introduction Gestational diabetes mellitus (GDM) is broadly defined as hyperglycaemia first recognised during pregnancy 1 . GDM is one of the most common complications in pregnancy, affecting 9–26% of pregnancies worldwide, with a rapidly increasing global incidence 2 , 3 . Diagnosis of GDM is linked to clinical risk factors such as obesity, age, ancestry, and a family history of type 2 diabetes. However, no diagnostic threshold has been adopted globally 1 . Furthermore, hyperglycaemia may remain undiagnosed in patients who do not meet specific thresholds 4 . GDM is associated with an increased risk of postpartum Type 2 Diabetes 5 , 6 , an increased risk of cardiovascular disorders 3 and a greater risk of metabolic syndrome in offspring 7 . As such, GDM represents an ongoing global health challenge. Chronic inflammation is a hallmark of GDM, whereby maladaptation of the maternal immune response may contribute to the disease pathophysiology 1 . Regulatory T cells (Tregs) are a subset of CD4 + T cells responsible for maintaining immune homeostasis and inhibiting unwanted immune responses 8 . Tregs are characterised by the intracellular marker FOXP3 which is required for lineage specification 9 , 10 . Although Tregs can be identified by a combination of cell surface markers (CD4+, CD25+, CD127-), they are heterogeneous and can be divided into phenotypic subsets, namely naive and effector cells, based on intra-cellular markers such as CTLA-4, GITR , HLA-DR , and CCR7 9,11,12 . In healthy pregnancies, Tregs prevent the rejection of the fetus by the maternal immune system 13 . Lower Treg percentages have been observed with specific adverse pregnancy outcomes 14 , however, existing literature on the role of Tregs in GDM is conflicting 15 . Several studies have assessed Treg percentages in GDM patients relative to healthy controls with conflicting conclusions 15 – 19 . A meta-analysis of seven publications concludes that Tregs are significantly lower in women with GDM however, the mechanism of this association remains unclear 20 . Functional assays suggest that Tregs in GDM patients may inefficiently regulate immune responses. For example, GDM Tregs are less effective at suppressing IFN-γ and TNF-α production in effector T cells 16 and the activity of CD4 + T cells 18 . Transcriptional networks may therefore be altered in Tregs as a result of GDM. Single-cell RNA sequencing (scRNA-seq) can provide high-resolution maps of transcriptional states in single cells. scRNA-seq has been used to identify T cell subsets based on differential expression of molecular markers, such as CRR7 or TCF7 in naive T cells and CCL5 in effector T cells 21 . In comparisons between GDM and healthy patient cohorts, scRNA-seq has identified altered estrogen signalling in T and NK cells 22 and a greater abundance of lymphocytes with upregulated reactive oxygen species and oxidative phosphorylation pathways 23 . Although evidence for a pro-inflammatory T cell phenotype in GDM patients has mounted in recent years, the role and function of Tregs in GDM pathology remains ambiguous 15 . Blood glucose measurements are used to diagnose GDM, however, earlier detection of at-risk patients could inform timely management and treatment, such as the use of metformin to prevent obstetrical complications 24 . Novel biomarkers may inform predictive models of GDM by capturing disease mechanisms 25 . For example, serum TNF-alpha levels are elevated in GDM patients 17 , and this has been attributed to a failure of Tregs to suppress effector T cell activity 16 . Identifying similarly informative biomarkers from Treg expression profiles may expand our understanding of disease biomarkers and potentially translate to models for earlier detection. This study aimed to investigate the Treg transcriptional landscape in GDM through scRNA-seq. We identified genes differentially expressed within Treg subsets from GDM patients and detected gene sets altered in GDM Tregs. We explore the potential for these cell-type-specific signals to serve as clinical biomarkers by modelling gene expression and GDM status across cohorts. This work provides a compendium of dysregulated genes within FOXP3 + Tregs and expands our understanding of their contribution to the pathophysiology of GDM. Results Single-cell RNA-seq identifies Treg subsets in GDM and healthy controls To investigate Treg transcriptional states in GDM, we sorted peripheral blood mononuclear cells from 10 GDM patients and 13 healthy controls using cell surface markers CD3 + , CD4 + , CD25 + and CD127 − ( Methods, Supplementary Figure S1 ). Single-cell RNA-seq (10X Genomics 5’ VDJ) detected high-quality transcriptional profiles for 9,681 Tregs from GDM patients (mean: 968, SD:468) and 12,537 cells from healthy controls (mean:964, SD:325) ( Methods ). Unsupervised clustering revealed seven Treg subsets (Fig. 1 A). To identify and label each Treg subset, we performed differential expression analysis comparing cells within each cluster to cells from all other clusters ( Methods , Fig. 1 C). We observed high CCR7 expression in two clusters which indicated the presence of naive Tregs. The Naive-1 cluster showed no unique markers, whereas the Naive-2 cluster uniquely overexpressed EGR1, DUSP1, FOS, JUN , and IER2 . JUN and FOS are subunits of the AP-1 transcription factor involved in regulating Treg differentiation and Treg proliferation 26 – 28 , which suggests that the two Naive subsets may present distinct cellular phenotypes. Next, we identified three effector Treg clusters among CCR7 low cells expressing HLA-A , CD74 , and IL32 , respectively. An additional Treg cluster was defined by high MALAT1 expression, which is associated with the cell cycle phase and Treg proliferation 29 , 30 . Lastly, we identified a subset of Tregs expressing Humanin genes MTRNRL28 and MTRNRL12 . Humanin-positive CD4 + T cells have not been described in the context of GDM, however, Humanin expression has been linked to hyperglycaemia in other disease contexts 31 . A complete list of marker genes distinguishing these Treg subsets is summarised in Table 1 . Table 1 Treg cell subtype assignment. Tregs from all patients were assigned to clusters by gene expression profile similarity. Clusters were assigned subtype names based on previous literature and marker genes identified by between-cluster differential expression analysis. N/A: Not applicable. Cluster Canonical Markers Cluster-specific genes (Top 3) Naive-1 CCR7+, CD28-, CD95(FAS)- JUN-, DUSP1-, KLF6- Naive-2 CCR7+, CD28-, CD95(FAS)- JUN+, FOS+, IER2+, EGR1+ Effector-1 CCR7-, CD28+, CD95(FAS)+ HLA-A+, IL32+, GBP5+ Effector-2 CCR7-, CD28+, CD95(FAS)+ ANXA1+, KLRB1+, CRIP1+ Effector-3 CCR7-, CD28+, CD95(FAS)+ CD74+, IL32+, S100A4+ Humanin+ N/A MTRNR2L12+, MTRNR2L8+, AL138963.4(TPT1)+ MALAT1+ N/A MALAT1+, MTRNR2L12+, EEF1B2- To determine whether these marker genes are unique to Treg subsets, we performed scRNA-seq on sorted CD4 + T cells from the same patients, followed by unsupervised clustering ( Methods ). We identified 10 CD4 + T cell subsets including four naive clusters ( CCR7+ ), two memory cell clusters ( CCR6+ ) , an effector cell cluster ( CXCR3+ ), a Humanin + cluster ( MTRNR2L12+ ), a MALAT1 + cluster and one FOXP3 + cluster expected to represent Tregs amongst the sorted CD4 + T cells ( Figure S2A-C, Table S1 ). Surprisingly, the MALAT1 + and Humanin + clusters were common to both sorted CD4 + T cell and sorted Treg datasets, suggesting that these cells are shared among CD4 + T cells in GDM and control patients, regardless of FOXP3 expression ( Figure S2D ). These results characterise the landscape of the T cell subsets present in peripheral blood during pregnancy, from which differential analyses between GDM and control patients can be conducted. The proportion of cells in each cluster may vary between patients as a consequence of disease. Indeed, we observed that Treg subset proportions varied by GDM status (Fig. 2 A) and within individual patients (Fig. 2 B). For example, patient L215-A-AHH03 carried the highest proportion of MALAT1 + cells (0.085) and the lowest proportion of Effector-2 cells (0.035), meanwhile, patient L215-A-AHH05 showed the highest proportion of Effector-2 cells (0.234) and a low relative proportion of Naive-1 cells (0.204). To investigate whether Treg subset proportions are altered by GDM, we performed significance testing using propeller 32 ( Methods ). Differences in Treg subset proportions were not statistically significant (p < 0.05) (Fig. 2 C), however, GDM patients had a notably higher proportion of Humanin + Tregs (ratio = 1.45, p = 0.052) driven by a subset of patients (Fig. 2 B). Interestingly, Memory-1 CD4 + T cell proportions from the sorted CD4 + population were relatively higher in GDM (ratio = 1.39, p = 0.008), suggestive of a reduced immunosuppressive capacity of GDM Tregs. Furthermore, no significant difference was observed in the CD4 + T cell FOXP3 + subset (ratio = 0.987, p = 0.776). These results indicate that while most Treg subsets are consistent between GDM patients and healthy individuals, Humanin + Tregs and Memory-1 CD4 + T cells may vary by disease status. GDM is associated with differential expression in Treg subsets To determine whether Treg subsets have altered transcriptomes in GDM, we performed differential expression analysis using Seurat and MAST ( Methods ). Here, we compared GDM and healthy controls within each Treg subset, excluding the Humanin + and MALAT1 + populations as they were not exclusive to Tregs. We detected an average of 9 and 21 significant differentially expressed genes in effector and naive Treg subsets respectively (Fig. 3 A). JUN, JUND, MT-CO1 and MT-CYB were widely downregulated across GDM patient Tregs. The Naive-2 subset showed significantly reduced expression of all four marker genes JUN, FOS, EGR1 and IER2 , which are components of the AP-1 transcription factor associated with Treg development 28 . To determine the transcriptional pathways altered in each Treg subset, we performed gene set enrichment analysis using the Hallmark Gene Sets from MSigDb 33 . We observed that Effector-2 Tregs were significantly enriched for the Angiogenesis hallmark gene set, driven by TIMP1 , S100A4 , and APP expression (NES = 1.75, FDR = 0.005) (Fig. 3 B). Genes involved in TNF-α (Tumor Necrosis Factor Alpha) signalling by NF-κB (Nuclear Factor Kappa B) were downregulated in the Naive-2 cluster (NES=-1.93, FDR = 0.055). This downregulation was most pronounced in FOSB , EGR1 , KLF2 , NFKBIA , DUSP2, and JUNB , several of which are uniquely expressed in the Naive-2 cluster (Table 1 ). This suggests that the transcriptional programme underlying Naive-2 Treg subsets may be disrupted in GDM. Treg dysregulation identifies candidate whole blood markers of GDM We hypothesised that genes differentially expressed within GDM Treg subsets might serve as molecular markers of GDM in whole blood. To test this hypothesis, we first built a predictive model using clinical features associated with elevated GDM risk, namely patient body mass index (BMI), ethnicity and mean arterial pressure (MAP) 1 . We fit a logistic regression model to these data and established that BMI had the strongest independent association with GDM status (OR: 1.35, 95% CI: 1.01–2.17) (Fig. 4 A-B). Evaluating BMI as a predictor by leave-one-out cross-validation (LOO-CV) returned an AUC of 0.718, representing a baseline for comparison. To evaluate the predictive power of Treg marker gene expression, we compiled a panel of 108 candidate marker genes defined by significant differential expression within and between Treg subsets ( Table S2 ). In all comparisons, we benchmark individual genes using logistic regression with LOO-CV to predict GDM status ( Methods ). First, we sought to confirm that GDM Treg marker genes were predictive of the average cell signals that would be observed in CD4 + cell isolation experiments from blood. We averaged scRNA-seq of Tregs (discovery cohort) and CD4 + T cell populations and fit independent models which determined that the highest AUC was achieved using RPS18 (Treg = 0.79, CD4 = 0.75) and MT-CO3 (Treg = 0.78, CD4 = 0.70) (Fig. 4 C). This was as expected given that these genes were broadly downregulated across Treg clusters (Fig. 3 A). Interestingly, ANXA2 , which identifies the Effector-2 Treg subset, had a higher AUC in CD4 + T cells (Treg = 0.48, CD4 = 0.72), likely due to being a significant marker of the Memory-1 CD4 + T cell population which showed lower percentages in GDM. These data suggest that genes differentially expressed in GDM CD4 + T and Treg subsets may be used to predict disease from bulk mRNA. To investigate whether GDM Treg marker genes are predictive of GDM in bulk RNA-seq datasets, we acquired previously published bulk RNA sequencing data from two independent studies on GDM patients and healthy controls 34 , 35 . Logistic regression models fit to each marker gene, and evaluated with LOO-CV, identified 21/108 markers with AUC ≥ 0.5 in at least one bulk mRNA dataset (Fig. 4 D). We detected several genes whose expression in bulk mRNA produced models with AUC values similar to that of BMI in at least one dataset, including TXNIP, MT-ATP8, H3F3B, RPS18, ANXA2, HLA-DRA and HLA-DRB1 (Fig. 4 E). However, TXNIP , which was downregulated in naive Tregs from scRNA, was the only gene which returned a model with a high AUC in both bulk mRNA datasets (WANG = 1; STIRM = 0.70) despite having a low predictive value in the pseudobulk samples. TXNIP regulates glucose uptake and is an active target for anti-diabetic drugs due to its association with insulin uptake and Metformin activity 36 . This suggests that Treg function may be affected by global TXNIP signalling in GDM patients. Collectively, these findings imply that genes involved in GDM Treg dysregulation may serve as predictive biomarkers for GDM, with performance comparable to known GDM risk factors such as BMI. Discussion Regulatory T cell dysfunction is proposed to underlie the pathology of gestational diabetes mellitus (GDM). We present single-cell transcriptomes of Tregs and CD4 + T cells isolated from PBMCs of women with GDM and healthy pregnant women. Our analysis identifies naive and effector molecular subsets in Tregs with a significant increase in memory CD4 + T cells from GDM patients, suggesting impaired immune suppression. Differential expression analysis revealed a decrease in AP-1 transcription factor subunits in the Naive-2 cluster, and pathway analysis indicated downregulated NF-kB signalling. Additionally, Effector-2 Tregs upregulated genes involved in Angiogenesis. We evaluated the translational potential of scRNA-seq-derived expression markers from CD4 + T cell populations and validated their predictive potential in pseudobulk CD4 + T cells and independent validation cohorts. The percentage of Tregs in GDM patients is observed to be reduced in women with GDM 20 . We identified FOXP3 + cells within sorted CD4 + T cells which we expect to contain Tregs, however, their proportion did not significantly differ between conditions. FOXP3 + cells represent a low proportion of the CD4 + T cell pool. Therefore we may be underpowered to observe the effect at a single cell level. Of note, we observed Humanin + and MALAT1 + cell clusters in both Treg and CD4 + T cell populations. These cells have been previously reported in sorted Tregs from ankylosing spondylitis patients profiled with the 10X scRNA-seq platform 37 . We also observed Humanin + CD4 + T cells, have been reported in the context of Rheumatoid Arthritis 38 , To our knowledge, this is the first report of Humanin-positive cells in GDM. Altered humanin concentrations have been proposed as a biomarker for diabetic conditions, as Humanin is protective under oxidative stress, which can arise from excess ROS generation from diabetic hyperglycaemia 31 , 39 . Further research into the phenotypes of these Treg subsets is required to understand their role in GDM. NF-kB signalling is associated with apoptosis, insulin resistance, systemic inflammation and all major diabetic complications 40 . NFKBIA is upregulated in whole blood from GDM patients 41 . Similarly, the TNF-alpha signalling via NF-kb hallmark pathway was disrupted through RNA-seq of a murine GDM model 42 . We observed a decrease in NFKBIA signalling specific to the Naive-2 GDM Treg subset. Dimerisation between NF-kB and FOXP3 is essential for Treg activation and disruption to this pathway leads to immune dysregulation and altered Treg frequencies 43 . Therefore this suggests a potential mechanism by which GDM Tregs would be less effective at managing inflammatory responses. Additionally, the Effector-2 population was significantly enriched for angiogenesis pathway genes. Although chronic inflammation and angiogenesis are symptoms of GDM promoted by Tregs 44 , 45 , it remains unclear how these genes drive the suggested phenotype. Future work will require isolation and phenotypic characterisation of the Effector-2 and Naive-2 cell populations in order to discern their roles in GDM. Body mass index (BMI) is the most significant risk factor for GDM and maintaining a BMI below 25kg m − 2 is recommended to reduce GDM risk 1 . Additionally, a lack of globally accepted diagnostic thresholds and reliance on BMI may lead to undiagnosed patients, outlining the need for predictive biomarkers for GDM testing 4 . We benchmarked molecular markers against BMI and found them to have similar predictive value. Rapid protocols are readily available for CD4 + T cell isolation from whole blood 46 . Emulating this expression analysis using pseudobulk scRNA-seq expression, we observed that genes broadly altered across Treg subsets were highly predictive of GDM, such as RPS18 , MT-CO1 and MT-CO3 . MT-CO1 and MT-CO3 are subunits of Cytochrome c oxidase II involved in ATP synthesis, and mutations in MT-CO3 are associated with maternally inherited mitochondrial diabetes 47 . Their predictive value may be due to a hyperglycaemic environment in GDM that affects Treg signalling. Expanding our analysis to bulk mRNA from independent cohorts, the Treg marker TXNIP returned the highest AUC. Given its role in diabetes and inflammation, TXNIP represents a promising candidate biomarker for GDM 36 . While our study sheds light on the Treg transcriptional landscape in GDM, the focus on isolated CD4 + cells carries limitations. Several immune subtypes that were not studied are associated with the pathophysiology of GDM, such as Neutrophils, Macrophages and Monocytes 48 . Functional studies in models of GDM may further uncover the role of immune cells in attenuating Treg activity. For example, we observed the under-expression of JUNB , the AP-1 transcription factor, which is demonstrated to regulate intestinal Treg development and its ablation is associated with increased T helper accumulation and inflammation 49 . Although single-cell RNA-seq empowers us to discover cell type-specific expression, this analysis remains limited to protein-coding RNAs. In contrast, several non-coding RNAs such as microRNAs 50 and circular RNAs 51 have been proposed as candidate circulating biomarkers for GDM. Studying the effect of non-coding RNAs on T regulatory cell subtypes may uncover insights into GDM pathology. In conclusion, this study identifies Treg subsets with altered transcriptional programmes in GDM patients and proposes candidate marker genes with translational potential that arise from Treg dysregulation. Chronic inflammation leading to metabolic dysregulation is one coupling factor driving the association between obesity, cardiovascular disease, and diabetes 52 . Future work investigating these molecular markers of Treg dysfunction will improve our understanding of the complex immunometabolic state observed in GDM. Methods PBMC isolation and FACS Venous blood was collected from 23 participating pregnant women at 35–36 weeks of gestation, of whom 10 were diagnosed with GDM and the remainder were healthy controls. Blood was captured in BD Vacutainer® EDTA Tubes with a volume of 10 mL, and the samples were kept at room temperature (18–22°C) while being agitated using an orbital shaker at 90–100 rpm. Peripheral blood mononuclear cells (PBMCs) were isolated as previously described 53 . Briefly, Ficoll-Paque PLUS medium was placed into a SepMate™-50 tube, and the collected blood was transferred to a separate 50 mL Falcon tube where it was diluted 1:1 with PBS-EDTA containing 2% FBS. This diluted sample was added to the SepMate™ tube and then centrifuged at 1200 x g for 20 minutes. After isolation, cells were counted and resuspended in 1 mL of freezing medium containing DMSO. Cryovials containing the cells were kept on wet ice for less than 5 minutes before commencing freezing. The cryovials were then placed in a Corning CoolCell chamber at temperatures below − 70°C. After a minimum of 12 hours, these were transferred to cryo boxes and stored in a -80°C freezer for up to 24 hours before their final transfer to a liquid nitrogen tank. In the thawing process, the cryovial containing the frozen cells was submerged in a 37°C water bath for approximately one minute. Once thawed, 1 mL of pre-warmed medium is promptly added to the cryovial using a transfer pipette. The contents were then transferred to a 15 mL conical Falcon™ tube, which was prefilled with 5 mL of medium also warmed to 37°C. The empty cryovial was then rinsed with an additional 1 mL of medium to ensure complete transfer of cells. Following this, the Falcon™ tube was incubated in the 37°C water bath for 5 minutes. After incubation, the tube was centrifuged at 300×g at room temperature for another 5 minutes. The supernatant was discarded. PBMCs were washed twice in PBS and resuspended. Cells were counted using a hemocytometer and trypan blue to achieve a density of 700–1200 cells per ul. The cells were then stained for CD4 AlexaFLuor 700, CD4 BUV395, CD25 TexasRed and CD127 BV786. DAPI was used to exclude dead cells. The Tregs and CD4 + T cells were sorted in a BD FACS Aria III in 1.5ml Eppendorf using the gating strategy shown in Supplementary Figure S1 . Single-cell RNA sequencing Single-cell mRNA, dual-indexed, sequencing libraries were generated following the Chromium Single Cell 5’ v2 protocol with Feature Barcode technology for Cell Surface Protein and Immune Receptor Mapping (CG000330, 10x Genomics). For each 10x run, patient samples were hash-tagged and pooled, using one antibody-derived tag per patient following the CITE-Seq protocol 54 (Biolegend) and then flow-sorted to generate separate compartments of CD4 + T-cells and FOXP3 + T-cells before proceeding with the 10x assay. Gel beads-in-emulsion (GEMs) were generated on the chromium controller using reagents specific for the 5’ assay. 10x barcoded, full-length cDNA from poly-adenylated mRNA and DNA from protein Feature Barcode were generated and amplified. The amplified cDNA was used to generate gene expression and V(D)J sequencing libraries. To generate the V(D)J sequencing libraries, full-length, 10x barcoded V(D)J segments were enriched from amplified cDNA via further amplification using primers specific to TCR constant regions and converted into sequencing libraries. Cell surface protein Feature Barcode-sequencing libraries were generated from amplified DNA to detect the antibody barcodes used for cell hashing. The dual indexed libraries thus generated for 5’ gene expression, V(D)J and Cell Surface Protein were pooled and sequenced on Illumina NextSeq 2000 and NovaSeq 6000 platform to generate paired-end reads (28-10-10-90) following 10x recommended sequencing depth. Data processing and quality control The cellranger (v6.1.1, 10X Genomics) mkfastq command was used to demultiplex raw sequencing data into FASTQ files and reads were aligned to the GRCh38 and converted to raw feature counts using the cellranger multi command. Count tables were imported into R (v4.1.3) and analysed using Seurat (v4.1.0). For all downstream analyses, gene expression counts for isolated Treg and CD4 + T cells were processed separately. Cells were assigned to hashtag sequences using the Seurat HTODemux function, and only cells identified as Singlets were retained. To detect sample mix-ups, transcript BAM files were demultiplexed with custom scripts and germline SNPs were identified per patient using cellSNP (v0.3.2). Sample mixups were labelled if pairwise SNP allele frequency Pearson correlations exceeded a threshold of 0.8. This procedure correctly identified one known duplicate sample, which was removed from subsequent analysis. To detect outliers cells, cells with UMI counts > 4x the absolute deviation were removed from downstream analysis. Furthermore, cells with transcript counts for fewer than 350 genes were excluded due to low capture. Cells were further excluded if fewer than 20% of known housekeeping genes or over 5% of mitochondrial genes were expressed. Supervised detection of cell types was performed using SingleR (v1.8.1) and the cell dex database (v1.4.0), from which all barcodes assigned the ‘T cell’ label were retained for downstream analysis. Transcript counts were normalised and then integrated using the Seurat sctransform workflow, combining the functions SCTransform, SelectIntegrationFeatures, PrepSCTIntegration, FindIntegrationAnchors and IntegrateData . Cell cycle scores estimated with the function CellCycleScoring were regressed using SCTransform . We captured 22,218 Treg cells (mean: 966, SD:383.5) and 23,737 Tconv cells (mean: 1032, SD: 538.8) per patient following quality control. Across conditions, 43.57% of Tregs and 49.5% of Tconvs were derived from GDM patients. Seurat objects for Treg and CD4 + T cell populations were processed separately in all downstream analyses. T cell subpopulation detection and visualisation Cells were clustered by running Seurat’s RunPCA function with 50 principal components, followed by the functions FindNeighbors and FindClusters (Treg resolution = 0.75, CD4 + resolution = 0.9). T cell receptor (TCR) genes were filtered from variable features in the Integrated dataset to reduce their influence on cell clustering. For visualisation, UMAP and tSNE embeddings were derived from the Seurat RunUMAP (Treg components = 2; Treg neighbors = 50; CD4 + components = 2; CD4 + neighbors = 7) and RunTSNE functions (Treg perplexity = 15; CD4 + perplexity = 35). Clusters detected in CD4 + and Treg Seurat objects were assigned broad groups of Naive, Effector or Memory by computational gating on the expression of phenotypic marker genes previously reported for conventional T cells and Tregs 11 , 55 . Clusters were further distinguished by the top marker genes as detected using the Seurat function FindAllMarkers . Differential abundance The propeller function from the R package speckle (v0.99.7) was used to test for differences in cell type proportions between GDM and control patients, separately for CD4 + T cells and FOXP3 + Tregs. Propeller calculates patient-level cell type proportions from cell annotations, performs a transformation on the counts (logit transform in this analysis), and applies a moderated t- test to find significant differences between cell type proportions 32 . Differential expression Within-cluster differential expression testing was performed using the Seurat FindMarkers with the MAST (v1.25.2) testing framework applied to fit generalised linear models adapted for the bimodality and zero-inflated nature of single-cell gene expression data 56 . Cells from each unique cluster were filtered from Seurat SCT assay objects prior to MAST analysis. Genes with fewer than 10% of cells expressing transcripts in either GDM or control populations, Humanin + clusters and ribosomal proteins were excluded from downstream analysis. Significantly differentially expressed genes were defined as genes where the absolute average Log2 fold-change was more than 0.25 and the adjusted p-value was below 0.05. Gene set enrichment analysis Gene set enrichment analysis was performed by testing all genes analysed in the differential expression analysis for their enrichment in the 50 Human MSigDB Hallmark gene sets (v7.5.1). For each within-cluster comparison, genes were ranked by their average Log2Fold-change and tested against Hallmark gene sets using fgsea (v1.24.0). CD4 + T cells and Treg populations were treated separately. GDM risk factor analysis GDM risk factors were collected from patients at the time of recruitment, including ethnicity, body mass index (BMI), diastolic and systolic blood pressure, and interventions such as aspirin and metformin. Mean arterial pressure (MAP) was calculated using the commonly used formula summing the diastolic pressure and one-third of the pulse pressure, the difference between systolic and diastolic pressure 57 . The R glm function was used to fit logistic regression models to BMI, MAP and ethnicity as predictors of GDM status. Leave-one-out cross-validation was implemented. The area under the receiver operating characteristic curve (AUC) was calculated from the results of each fold to evaluate the model. GDM classification from RNA-seq Pseudobulk expression profiles were generated from CD4 + T cell populations by averaging cell transcript counts from the Seurat RNA assay for each patient using Seurat AverageExpression function. Bulk mRNA profiles were acquired from previously published studies on GDM and control patients 34 , 35 , and GDM status was converted to binary outcome in all datasets prior to downstream analysis. Models were built using marker genes from scRNA-seq differential expression results. Specifically, GDM marker gene sets were defined by combining: differentially expressed genes in sorted Treg clusters, to capture specific signals of Treg malfunction. Logistic regression with leave-one-out cross-validation was applied to mRNA data. A single model was built for each gene as a predictor of GDM status to compare predictor performance across mRNA and CD4 + T cell pseudobulk datasets. Declarations Data availability Single-cell RNA sequencing data has been submitted to Gene Expression Omnibus (accession pending). Bulk RNA sequencing data from published studies is available at GSE154414 and GSE92772. Scripts and analysis pipelines have been deposited at https://github.com/NMNS93/rutepo_gdm_treg. Ethics statement This study was approved by the King’s College Hospital Research Ethics Committee, REC number 02-03-033, dated 01/04/2003. All the experiments conform to the relevant regulatory standards, including proper recruitment and obtaining informed consent. Author contributions N.E.M.* , N.M*. and A.E.* contributed equally to this work. N.E.M.*: Analyzed data, wrote and edited the manuscript. A.E.*: Consent, sample collection, and data analysis. N.M.: Consent, sample collection, and data analysis. S.K.: Data analysis. H.V.: Conducted experiments. A.B.: Data analysis and manuscript editing. A.M.: Data analysis. T.T.: Provided expert opinion. G.L.: Provided expert opinion and edited the manuscript. P.D.: Data analysis and provided expert opinion. K.H.N.: Provided expert opinion and edited the manuscript. C.S.: Provided expert opinion and edited the manuscript. P.S.: Conceptualized the study, conducted experiments, collected samples, and edited the manuscript. References McIntyre, H. D. et al. Gestational diabetes mellitus. Nat Rev Dis Primers 5 , 47 (2019). Jiang, L. et al. A global view of hypertensive disorders and diabetes mellitus during pregnancy. Nat. Rev. Endocrinol. 18 , 760–775 (2022). Sweeting, A., Wong, J., Murphy, H. R. & Ross, G. P. A Clinical Update on Gestational Diabetes Mellitus. Endocr. Rev. 43 , 763–793 (2022). American Diabetes Association. 2. Classification and Diagnosis of Diabetes: Standards of Medical Care in Diabetes-2020. Diabetes Care 43 , S14–S31 (2020). Bellamy, L., Casas, J.-P., Hingorani, A. D. & Williams, D. Type 2 diabetes mellitus after gestational diabetes: a systematic review and meta-analysis. Lancet 373 , 1773–1779 (2009). Noctor, E. & Dunne, F. P. Type 2 diabetes after gestational diabetes: The influence of changing diagnostic criteria. World J. Diabetes 6 , 234–244 (2015). Ornoy, A., Becker, M., Weinstein-Fudim, L. & Ergaz, Z. Diabetes during Pregnancy: A Maternal Disease Complicating the Course of Pregnancy with Long-Term Deleterious Effects on the Offspring. A Clinical Review. Int. J. Mol. Sci. 22 , (2021). Sakaguchi, S., Yamaguchi, T., Nomura, T. & Ono, M. Regulatory T cells and immune tolerance. Cell 133 , 775–787 (2008). Churov, A. V., Mamashov, K. Y. & Novitskaia, A. V. Homeostasis and the functional roles of CD4+ Treg cells in aging. Immunol. Lett. 226 , 83–89 (2020). Hori, S., Nomura, T. & Sakaguchi, S. Control of regulatory T cell development by the transcription factor Foxp3. Science 299 , 1057–1061 (2003). Rosenblum, M. D., Way, S. S. & Abbas, A. K. Regulatory T cell memory. Nat. Rev. Immunol. 16 , 90–101 (2015). Zemmour, D. et al. Single-cell gene expression reveals a landscape of regulatory T cell phenotypes shaped by the TCR. Nat. Immunol. 19 , 291–301 (2018). Krop, J., Heidt, S., Claas, F. H. J. & Eikmans, M. Regulatory T Cells in Pregnancy: It Is Not All About FoxP3. Front. Immunol. 11 , 1182 (2020). Green, S. et al. Regulatory T Cells in Pregnancy Adverse Outcomes: A Systematic Review and Meta-Analysis. Front. Immunol. 12 , 737862 (2021). McElwain, C. J., McCarthy, F. P. & McCarthy, C. M. Gestational Diabetes Mellitus and Maternal Immune Dysregulation: What We Know So Far. Int. J. Mol. Sci. 22 , (2021). Yang, Y. et al. Functional Defects of Regulatory T Cell Through Interleukin 10 Mediated Mechanism in the Induction of Gestational Diabetes Mellitus. DNA Cell Biol. 37 , 278–285 (2018). Lobo, T. F. et al. Impaired Treg and NK cells profile in overweight women with gestational diabetes mellitus. Am. J. Reprod. Immunol. 79 , (2018). Schober, L. et al. The role of regulatory T cell (Treg) subsets in gestational diabetes mellitus. Clin. Exp. Immunol. 177 , 76–85 (2014). Sifnaios, E. et al. Gestational Diabetes and T-cell (Th1/Th2/Th17/Treg) Immune Profile. In Vivo 33 , 31–40 (2019). Arain, H. et al. Regulatory T cells in the peripheral blood of women with gestational diabetes: a systematic review and meta-analysis. Front. Immunol. 14 , (2023). Szabo, P. A. et al. Single-cell transcriptomics of human T cells reveals tissue and activation signatures in health and disease. Nat. Commun. 10 , 4706 (2019). Yang, Y. et al. Transcriptomic Profiling of Human Placenta in Gestational Diabetes Mellitus at the Single-Cell Level. Front. Endocrinol. 12 , 679582 (2021). Jiao, B. et al. Dissecting human placental cells heterogeneity in preeclampsia and gestational diabetes using single-cell sequencing. Mol. Immunol. 161 , 104–118 (2023). Farahvar, S., Walfisch, A. & Sheiner, E. Gestational diabetes risk factors and long-term consequences for both mother and offspring: a literature review. Expert Rev. Endocrinol. Metab. 14 , 63–74 (2019). Rodrigo, N. & Glastras, S. J. The Emerging Role of Biomarkers in the Diagnosis of Gestational Diabetes Mellitus. J. Clin. Med. Res. 7 , (2018). Lee, S.-M., Gao, B. & Fang, D. FoxP3 maintains Treg unresponsiveness by selectively inhibiting the promoter DNA-binding activity of AP-1. Blood 111 , 3599–3606 (2008). Bahrami, S. & Drabløs, F. Gene regulation in the immediate-early response process. Adv. Biol. Regul. 62 , 37–49 (2016). Katagiri, T., Kameda, H., Nakano, H. & Yamazaki, S. Regulation of T cell differentiation by the AP-1 transcription factor JunB. Immunol Med 44 , 197–203 (2021). Dey, S. et al. Downregulation of MALAT1 is a hallmark of tissue and peripheral proliferative T cells in COVID-19. Clin. Exp. Immunol. 212 , 262–275 (2023). Masoumi, F. et al. Malat1 long noncoding RNA regulates inflammation and leukocyte differentiation in experimental autoimmune encephalomyelitis. J. Neuroimmunol. 328 , 50–59 (2019). Boutari, C., Pappas, P. D., Theodoridis, T. D. & Vavilis, D. Humanin and diabetes mellitus: A review of in vitro and in vivo studies. World J. Diabetes 13 , 213–223 (2022). Phipson, B. et al. propeller: testing for differences in cell type proportions in single cell data. Bioinformatics 38 , 4720–4726 (2022). Liberzon, A. et al. The Molecular Signatures Database (MSigDB) hallmark gene set collection. Cell Syst 1 , 417–425 (2015). Stirm, L. et al. Maternal whole blood cell miRNA-340 is elevated in gestational diabetes and inversely regulated by glucose and insulin. Sci. Rep. 8 , 1366 (2018). Wang, J., Wang, K., Liu, W., Cai, Y. & Jin, H. m6A mRNA methylation regulates the development of gestational diabetes mellitus in Han Chinese women. Genomics 113 , 1048–1056 (2021). Masutani, H. Thioredoxin-Interacting Protein in Cancer and Diabetes. Antioxid. Redox Signal. 36 , 1001–1022 (2022). Simone, D. et al. Single cell analysis of spondyloarthritis regulatory T cells identifies distinct synovial gene expression patterns and clonal fates. Commun Biol 4 , 1395 (2021). Argyriou, A. et al. Single cell sequencing reveals expanded cytotoxic CD4+ T cells and two states of peripheral helper T cells in synovial fluid of ACPA+ RA patients. bioRxiv (2021) doi:10.1101/2021.05.28.21255902. Rochette, L., Meloux, A., Zeller, M., Cottin, Y. & Vergely, C. Role of humanin, a mitochondrial-derived peptide, in cardiovascular disorders. Arch. Cardiovasc. Dis. 113 , 564–571 (2020). Indira, M. & Abhilash, P. A. Role of NF-Kappa B (NF-κB) in Diabetes. OT 4 , (2013). Chen, Y.-M. et al. Upregulation of T Cell Receptor Signaling Pathway Components in Gestational Diabetes Mellitus Patients: Joint Analysis of mRNA and circRNA Expression Profiles. Front. Endocrinol. 12 , 774608 (2021). Paolino, M. et al. RANK links thymic regulatory T cells to fetal loss and gestational diabetes in pregnancy. Nature 589 , 442–447 (2021). Cui, Y. et al. A Stk4-Foxp3-NF-κB p65 transcriptional complex promotes Treg cell activation and homeostasis. Sci Immunol 7 , eabl8357 (2022). Joshi, N. P., Madiwale, S. D., Sundrani, D. P. & Joshi, S. R. Fatty acids, inflammation and angiogenesis in women with gestational diabetes mellitus. Biochimie 212 , 31–40 (2023). Lužnik, Z., Anchouche, S., Dana, R. & Yin, J. Regulatory T Cells in Angiogenesis. J. Immunol. 205 , 2557–2565 (2020). Hsu, C.-H., Chen, C., Irimia, D. & Toner, M. Fast sorting of CD4+ T cells from whole blood using glass microbubbles. Technology 3 , 38–44 (2015). Tabebi, M. et al. A novel mutation MT-COIII m.9267G>C and MT-COI m.5913G>A mutation in mitochondrial genes in a Tunisian family with maternally inherited diabetes and deafness (MIDD) associated with severe nephropathy. Biochem. Biophys. Res. Commun. 459 , 353–360 (2015). De Luccia, T. P. B. et al. Unveiling the pathophysiology of gestational diabetes: Studies on local and peripheral immune cells. Scand. J. Immunol. 91 , e12860 (2020). Wheaton, J. D. & Ciofani, M. JunB Controls Intestinal Effector Programs in Regulatory T Cells. Front. Immunol. 11 , 444 (2020). Guarino, E. et al. Circulating MicroRNAs as Biomarkers of Gestational Diabetes Mellitus: Updates and Perspectives. Int. J. Endocrinol. 2018 , 6380463 (2018). Fan, W., Pang, H., Xie, Z., Huang, G. & Zhou, Z. Circular RNAs in diabetes mellitus and its complications. Front. Endocrinol. 13 , 885650 (2022). Hotamisligil, G. S. Inflammation, metaflammation and immunometabolic disorders. Nature 542 , 177–185 (2017). Efthymiou, A. et al. Isolation and freezing of human peripheral blood mononuclear cells from pregnant patients. STAR Protoc 3 , 101204 (2022). Stoeckius, M. et al. Simultaneous epitope and transcriptome measurement in single cells. Nat. Methods 14 , 865–868 (2017). Stroukov, W. et al. OMIP-090: A 20-parameter flow cytometry panel for rapid analysis of cell diversity and homing capacity in human conventional and regulatory T cells. Cytometry A 103 , 362–367 (2023). Finak, G. et al. MAST: a flexible statistical framework for assessing transcriptional changes and characterizing heterogeneity in single-cell RNA sequencing data. Genome Biol. 16 , 278 (2015). Papaioannou, T. G. et al. Mean arterial pressure values calculated using seven different methods and their associations with target organ deterioration in a single-center study of 1878 individuals. Hypertens. Res. 39 , 640–647 (2016). Additional Declarations There is NO Competing Interest. Supplementary Files SupplementaryMaterialsRUTEPOGDM20232.docx Cite Share Download PDF Status: Published Journal Publication published 03 Apr, 2026 Read the published version in Communications Medicine → Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-3773991","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":264835165,"identity":"a29f5dba-c3aa-4ab1-b761-ef807e90cdcb","order_by":0,"name":"Panicos Shangaris","email":"data:image/png;base64,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","orcid":"https://orcid.org/0000-0003-2750-8405","institution":"King's College London","correspondingAuthor":true,"submittingAuthor":false,"prefix":"","firstName":"Panicos","middleName":"","lastName":"Shangaris","suffix":""},{"id":264835166,"identity":"aeed05cd-46f2-4548-84d5-65ce8332f7d4","order_by":1,"name":"Nana Mensah","email":"","orcid":"","institution":"King's College London","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Nana","middleName":"","lastName":"Mensah","suffix":""},{"id":264835167,"identity":"2c66257d-f559-4b4d-b630-36c1f4e77b30","order_by":2,"name":"Athina Efthimiou","email":"","orcid":"","institution":"King's College London","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Athina","middleName":"","lastName":"Efthimiou","suffix":""},{"id":264835168,"identity":"2039fa19-b551-46ff-8f22-a03364a0b6ed","order_by":3,"name":"Nicoleta Mureanu","email":"","orcid":"","institution":"King's College London","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Nicoleta","middleName":"","lastName":"Mureanu","suffix":""},{"id":264835169,"identity":"ab1477e9-8916-4d31-8aa7-c0be7a712656","order_by":4,"name":"Heli Vaikkinen","email":"","orcid":"","institution":"King's College London","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Heli","middleName":"","lastName":"Vaikkinen","suffix":""},{"id":264835170,"identity":"163f878a-f8ed-47e0-bbc0-d027e4eae578","order_by":5,"name":"Shichina Kannambath","email":"","orcid":"","institution":"BRC Genomics Core","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Shichina","middleName":"","lastName":"Kannambath","suffix":""},{"id":264835171,"identity":"1a9b6ccc-6e45-4610-8f8d-c45545ac9a68","order_by":6,"name":"Amanda Bowman","email":"","orcid":"","institution":"King's College London","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Amanda","middleName":"","lastName":"Bowman","suffix":""},{"id":264835172,"identity":"7781a779-943a-4f4e-b68f-5ff1b9ff11ef","order_by":7,"name":"Athul Menon","email":"","orcid":"","institution":"King's College London","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Athul","middleName":"","lastName":"Menon","suffix":""},{"id":264835173,"identity":"ecc69804-d836-4f9c-a528-d0a4d529c309","order_by":8,"name":"Timothy Tree","email":"","orcid":"","institution":"King's College London","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Timothy","middleName":"","lastName":"Tree","suffix":""},{"id":264835174,"identity":"708aacd8-26db-48a7-b23f-221e86415022","order_by":9,"name":"Giovanna Lombardi","email":"","orcid":"https://orcid.org/0000-0003-4496-3215","institution":"King's College London","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Giovanna","middleName":"","lastName":"Lombardi","suffix":""},{"id":264835175,"identity":"f0bd22c2-9119-40b9-aa2f-f1bbd9f71e71","order_by":10,"name":"Pawan Dhami","email":"","orcid":"https://orcid.org/0000-0002-5408-1833","institution":"Guy's and St Thomas' Trust","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Pawan","middleName":"","lastName":"Dhami","suffix":""},{"id":264835176,"identity":"fe02efc0-4f90-4136-ae14-44e9b9155dff","order_by":11,"name":"Kypros Nicolaides","email":"","orcid":"","institution":"King's College Hospital","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Kypros","middleName":"","lastName":"Nicolaides","suffix":""},{"id":264835177,"identity":"ee96d535-795f-4403-ba3c-59b1e1434632","order_by":12,"name":"Cristiano Scottá","email":"","orcid":"","institution":"King's College London","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Cristiano","middleName":"","lastName":"Scottá","suffix":""}],"badges":[],"createdAt":"2023-12-18 23:30:25","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-3773991/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-3773991/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s43856-026-01563-0","type":"published","date":"2026-04-03T04:00:00+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":49434435,"identity":"7a2015e6-fca2-483d-839f-dba3f29d59f5","added_by":"auto","created_at":"2024-01-10 19:31:45","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":1619088,"visible":true,"origin":"","legend":"\u003cp\u003eTreg subsets in GDM patients and healthy controls. A) Treg cell clusters visualised using a t-stochastic neighbour embedding (t-SNE). B) Log-transformed normalized expression values of genes identifying broad and canonical Treg subsets. C) Log-transformed normalized expression values of genes distinguishing Treg clusters.\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-3773991/v1/1b656d879cfe1e41efa70bfe.png"},{"id":49434113,"identity":"d312a69b-d3b2-4c0a-acf4-8f6801bcc325","added_by":"auto","created_at":"2024-01-10 19:23:45","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":278011,"visible":true,"origin":"","legend":"\u003cp\u003eTreg subtypes in GDM patients and healthy controls. A) t-stochastic neighbour embedding (t-SNE) plot of sorted FOXP3+ Treg cell expression. B) Normalised (log) expression of genes marking Treg cell subtypes. C) Marker gene expression across Treg cell clusters.\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-3773991/v1/1a929175ed846d7d7f9e4d4d.png"},{"id":49434116,"identity":"8cc12ac2-918d-4d93-9ca4-695aaf3e648c","added_by":"auto","created_at":"2024-01-10 19:23:45","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":419612,"visible":true,"origin":"","legend":"\u003cp\u003eDifferential expression between GDM and healthy controls within Treg subsets. A) Genes with significant differential expression results within subsets. N.S. = Not significant B) Gene set enrichment plot for all Tregs. ‘**’ : p\u0026lt;=0.01. ‘-’: p\u0026lt;0.1\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-3773991/v1/ff1259b6b74491a32af03b13.png"},{"id":49434114,"identity":"b4d11d90-1bbd-43a1-a85b-51fd9487f512","added_by":"auto","created_at":"2024-01-10 19:23:45","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":529470,"visible":true,"origin":"","legend":"\u003cp\u003ePredicting GDM status from clinical and molecular data. A) Body mass index distribution of GDM and control patients. B) Odds ratio of GDM status calculated by logistic regression using mean arterial pressure, ethnicity and body mass index. C) Area under the receiver operating characteristic curve (AUC) and F1 scores of genes predictive of GDM status in CD4+ T cell and Treg pseudobulk expression profiles. The input gene list is derived from genes differentially expressed in Treg subsets. Scores were calculated using Logistic Regression and leave-one-out cross-validation. Dashed lines connect the performance of individual genes between cohorts. Solid lines connect labels for gene names to points. D) As in C) using bulk mRNA data from Wang et al. (2021) and Stirm et al. (2018). E) ROC-AUC curves for the top-performing marker genes present in all datasets. Treg and CD4 contain pseudobulk data (this study) while Stirms and Wang contain bulk mRNA.\u003c/p\u003e","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-3773991/v1/6af52a2cf24f9db5c33ecdac.png"},{"id":110419074,"identity":"939c96d2-0151-4ad0-a3f0-699dd0d93481","added_by":"auto","created_at":"2026-06-01 07:06:28","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2474783,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-3773991/v1/d5b62f1b-e759-48c8-8c8e-6116d7d4e624.pdf"},{"id":49434117,"identity":"4b13eb8d-36a6-452a-8f88-1f046b3b5258","added_by":"auto","created_at":"2024-01-10 19:23:45","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":1295345,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cbr\u003e\u003c/p\u003e","description":"","filename":"SupplementaryMaterialsRUTEPOGDM20232.docx","url":"https://assets-eu.researchsquare.com/files/rs-3773991/v1/6d3495c885e59422bfb37217.docx"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"Single-cell transcriptomics reveals markers of regulatory T cell dysfunction in Gestational Diabetes Mellitus","fulltext":[{"header":"Introduction","content":"\u003cp\u003eGestational diabetes mellitus (GDM) is broadly defined as hyperglycaemia first recognised during pregnancy \u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e. GDM is one of the most common complications in pregnancy, affecting 9\u0026ndash;26% of pregnancies worldwide, with a rapidly increasing global incidence \u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e,\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e. Diagnosis of GDM is linked to clinical risk factors such as obesity, age, ancestry, and a family history of type 2 diabetes. However, no diagnostic threshold has been adopted globally \u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e. Furthermore, hyperglycaemia may remain undiagnosed in patients who do not meet specific thresholds \u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u003c/sup\u003e. GDM is associated with an increased risk of postpartum Type 2 Diabetes \u003csup\u003e\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e,\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u003c/sup\u003e, an increased risk of cardiovascular disorders \u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e and a greater risk of metabolic syndrome in offspring \u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e. As such, GDM represents an ongoing global health challenge.\u003c/p\u003e \u003cp\u003eChronic inflammation is a hallmark of GDM, whereby maladaptation of the maternal immune response may contribute to the disease pathophysiology \u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e. Regulatory T cells (Tregs) are a subset of CD4\u0026thinsp;+\u0026thinsp;T cells responsible for maintaining immune homeostasis and inhibiting unwanted immune responses \u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e. Tregs are characterised by the intracellular marker \u003cem\u003eFOXP3\u003c/em\u003e which is required for lineage specification \u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e,\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e. Although Tregs can be identified by a combination of cell surface markers (CD4+, CD25+, CD127-), they are heterogeneous and can be divided into phenotypic subsets, namely naive and effector cells, based on intra-cellular markers such as \u003cem\u003eCTLA-4, GITR\u003c/em\u003e, \u003cem\u003eHLA-DR\u003c/em\u003e, and \u003cem\u003eCCR7\u003c/em\u003e \u003csup\u003e9,11,12\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eIn healthy pregnancies, Tregs prevent the rejection of the fetus by the maternal immune system \u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e. Lower Treg percentages have been observed with specific adverse pregnancy outcomes \u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e, however, existing literature on the role of Tregs in GDM is conflicting \u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/sup\u003e. Several studies have assessed Treg percentages in GDM patients relative to healthy controls with conflicting conclusions \u003csup\u003e\u003cspan additionalcitationids=\"CR16 CR17 CR18\" citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u003c/sup\u003e. A meta-analysis of seven publications concludes that Tregs are significantly lower in women with GDM however, the mechanism of this association remains unclear \u003csup\u003e\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e. Functional assays suggest that Tregs in GDM patients may inefficiently regulate immune responses. For example, GDM Tregs are less effective at suppressing IFN-γ and TNF-α production in effector T cells \u003csup\u003e\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e and the activity of CD4\u0026thinsp;+\u0026thinsp;T cells \u003csup\u003e\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e. Transcriptional networks may therefore be altered in Tregs as a result of GDM.\u003c/p\u003e \u003cp\u003eSingle-cell RNA sequencing (scRNA-seq) can provide high-resolution maps of transcriptional states in single cells. scRNA-seq has been used to identify T cell subsets based on differential expression of molecular markers, such as \u003cem\u003eCRR7 or TCF7\u003c/em\u003e in naive T cells and \u003cem\u003eCCL5\u003c/em\u003e in effector T cells \u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e. In comparisons between GDM and healthy patient cohorts, scRNA-seq has identified altered estrogen signalling in T and NK cells \u003csup\u003e\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e and a greater abundance of lymphocytes with upregulated reactive oxygen species and oxidative phosphorylation pathways \u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e. Although evidence for a pro-inflammatory T cell phenotype in GDM patients has mounted in recent years, the role and function of Tregs in GDM pathology remains ambiguous \u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eBlood glucose measurements are used to diagnose GDM, however, earlier detection of at-risk patients could inform timely management and treatment, such as the use of metformin to prevent obstetrical complications \u003csup\u003e\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u003c/sup\u003e. Novel biomarkers may inform predictive models of GDM by capturing disease mechanisms \u003csup\u003e\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e. For example, serum TNF-alpha levels are elevated in GDM patients \u003csup\u003e\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u003c/sup\u003e, and this has been attributed to a failure of Tregs to suppress effector T cell activity \u003csup\u003e\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e. Identifying similarly informative biomarkers from Treg expression profiles may expand our understanding of disease biomarkers and potentially translate to models for earlier detection.\u003c/p\u003e \u003cp\u003eThis study aimed to investigate the Treg transcriptional landscape in GDM through scRNA-seq.\u0026nbsp;We identified genes differentially expressed within Treg subsets from GDM patients and detected gene sets altered in GDM Tregs. We explore the potential for these cell-type-specific signals to serve as clinical biomarkers by modelling gene expression and GDM status across cohorts. This work provides a compendium of dysregulated genes within FOXP3\u0026thinsp;+\u0026thinsp;Tregs and expands our understanding of their contribution to the pathophysiology of GDM.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003eSingle-cell RNA-seq identifies Treg subsets in GDM and healthy controls\u003c/p\u003e \u003cp\u003eTo investigate Treg transcriptional states in GDM, we sorted peripheral blood mononuclear cells from 10 GDM patients and 13 healthy controls using cell surface markers CD3\u003csup\u003e+\u003c/sup\u003e, CD4\u003csup\u003e+\u003c/sup\u003e, CD25\u003csup\u003e+\u003c/sup\u003e and CD127\u003csup\u003e\u0026minus;\u003c/sup\u003e (\u003cb\u003eMethods, Supplementary Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e\u003c/b\u003e). Single-cell RNA-seq (10X Genomics 5\u0026rsquo; VDJ) detected high-quality transcriptional profiles for 9,681 Tregs from GDM patients (mean: 968, SD:468) and 12,537 cells from healthy controls (mean:964, SD:325) (\u003cb\u003eMethods\u003c/b\u003e). Unsupervised clustering revealed seven Treg subsets (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eA).\u003c/p\u003e \u003cp\u003eTo identify and label each Treg subset, we performed differential expression analysis comparing cells within each cluster to cells from all other clusters (\u003cb\u003eMethods\u003c/b\u003e, Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eC). We observed high CCR7 expression in two clusters which indicated the presence of naive Tregs. The Naive-1 cluster showed no unique markers, whereas the Naive-2 cluster uniquely overexpressed \u003cem\u003eEGR1, DUSP1, FOS, JUN\u003c/em\u003e, and \u003cem\u003eIER2\u003c/em\u003e. \u003cem\u003eJUN\u003c/em\u003e and \u003cem\u003eFOS\u003c/em\u003e are subunits of the AP-1 transcription factor involved in regulating Treg differentiation and Treg proliferation \u003csup\u003e\u003cspan additionalcitationids=\"CR27\" citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e, which suggests that the two Naive subsets may present distinct cellular phenotypes. Next, we identified three effector Treg clusters among \u003cem\u003eCCR7\u003c/em\u003e low cells expressing \u003cem\u003eHLA-A\u003c/em\u003e, \u003cem\u003eCD74\u003c/em\u003e, and \u003cem\u003eIL32\u003c/em\u003e, respectively. An additional Treg cluster was defined by high \u003cem\u003eMALAT1\u003c/em\u003e expression, which is associated with the cell cycle phase and Treg proliferation\u003csup\u003e\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e,\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u003c/sup\u003e. Lastly, we identified a subset of Tregs expressing Humanin genes \u003cem\u003eMTRNRL28\u003c/em\u003e and \u003cem\u003eMTRNRL12\u003c/em\u003e. Humanin-positive CD4\u0026thinsp;+\u0026thinsp;T cells have not been described in the context of GDM, however, Humanin expression has been linked to hyperglycaemia in other disease contexts \u003csup\u003e\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e\u003c/sup\u003e. A complete list of marker genes distinguishing these Treg subsets is summarised in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eTreg cell subtype assignment. Tregs from all patients were assigned to clusters by gene expression profile similarity. Clusters were assigned subtype names based on previous literature and marker genes identified by between-cluster differential expression analysis. N/A: Not applicable.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCluster\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCanonical Markers\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCluster-specific genes (Top 3)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNaive-1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cem\u003eCCR7+, CD28-, CD95(FAS)-\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cem\u003eJUN-, DUSP1-, KLF6-\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNaive-2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cem\u003eCCR7+, CD28-, CD95(FAS)-\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cem\u003eJUN+, FOS+, IER2+, EGR1+\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEffector-1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cem\u003eCCR7-, CD28+, CD95(FAS)+\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cem\u003eHLA-A+, IL32+, GBP5+\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEffector-2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cem\u003eCCR7-, CD28+, CD95(FAS)+\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cem\u003eANXA1+, KLRB1+, CRIP1+\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEffector-3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cem\u003eCCR7-, CD28+, CD95(FAS)+\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cem\u003eCD74+, IL32+, S100A4+\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHumanin+\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cem\u003eN/A\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cem\u003eMTRNR2L12+, MTRNR2L8+, AL138963.4(TPT1)+\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMALAT1+\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cem\u003eN/A\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cem\u003eMALAT1+, MTRNR2L12+, EEF1B2-\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eTo determine whether these marker genes are unique to Treg subsets, we performed scRNA-seq on sorted CD4\u0026thinsp;+\u0026thinsp;T cells from the same patients, followed by unsupervised clustering (\u003cb\u003eMethods\u003c/b\u003e). We identified 10 CD4\u0026thinsp;+\u0026thinsp;T cell subsets including four naive clusters (\u003cem\u003eCCR7+\u003c/em\u003e), two memory cell clusters (\u003cem\u003eCCR6+\u003c/em\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e)\u003c/span\u003e, an effector cell cluster (\u003cem\u003eCXCR3+\u003c/em\u003e), a Humanin\u0026thinsp;+\u0026thinsp;cluster (\u003cem\u003eMTRNR2L12+\u003c/em\u003e), a \u003cem\u003eMALAT1\u003c/em\u003e\u0026thinsp;+\u0026thinsp;cluster and one \u003cem\u003eFOXP3\u003c/em\u003e\u0026thinsp;+\u0026thinsp;cluster expected to represent Tregs amongst the sorted CD4\u0026thinsp;+\u0026thinsp;T cells (\u003cb\u003eFigure S2A-C, Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e\u003c/b\u003e). Surprisingly, the \u003cem\u003eMALAT1\u0026thinsp;+\u003c/em\u003e\u0026thinsp;and Humanin\u0026thinsp;\u003cem\u003e+\u003c/em\u003e\u0026thinsp;clusters were common to both sorted CD4\u0026thinsp;+\u0026thinsp;T cell and sorted Treg datasets, suggesting that these cells are shared among CD4\u0026thinsp;+\u0026thinsp;T cells in GDM and control patients, regardless of \u003cem\u003eFOXP3\u003c/em\u003e expression (\u003cb\u003eFigure S2D\u003c/b\u003e). These results characterise the landscape of the T cell subsets present in peripheral blood during pregnancy, from which differential analyses between GDM and control patients can be conducted.\u003c/p\u003e \u003cp\u003eThe proportion of cells in each cluster may vary between patients as a consequence of disease. Indeed, we observed that Treg subset proportions varied by GDM status (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA) and within individual patients (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB). For example, patient L215-A-AHH03 carried the highest proportion of \u003cem\u003eMALAT1\u003c/em\u003e\u0026thinsp;+\u0026thinsp;cells (0.085) and the lowest proportion of Effector-2 cells (0.035), meanwhile, patient L215-A-AHH05 showed the highest proportion of Effector-2 cells (0.234) and a low relative proportion of Naive-1 cells (0.204). To investigate whether Treg subset proportions are altered by GDM, we performed significance testing using propeller\u003csup\u003e\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u003c/sup\u003e (\u003cb\u003eMethods\u003c/b\u003e). Differences in Treg subset proportions were not statistically significant (p\u0026thinsp;\u0026lt;\u0026thinsp;0.05) (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eC), however, GDM patients had a notably higher proportion of Humanin\u0026thinsp;+\u0026thinsp;Tregs (ratio\u0026thinsp;=\u0026thinsp;1.45, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.052) driven by a subset of patients (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB). Interestingly, Memory-1 CD4\u0026thinsp;+\u0026thinsp;T cell proportions from the sorted CD4\u0026thinsp;+\u0026thinsp;population were relatively higher in GDM (ratio\u0026thinsp;=\u0026thinsp;1.39, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.008), suggestive of a reduced immunosuppressive capacity of GDM Tregs. Furthermore, no significant difference was observed in the CD4\u0026thinsp;+\u0026thinsp;T cell \u003cem\u003eFOXP3\u003c/em\u003e\u0026thinsp;+\u0026thinsp;subset (ratio\u0026thinsp;=\u0026thinsp;0.987, \u003cem\u003ep\u0026thinsp;=\u003c/em\u003e\u0026thinsp;0.776). These results indicate that while most Treg subsets are consistent between GDM patients and healthy individuals, Humanin\u0026thinsp;\u003cem\u003e+\u003c/em\u003e\u0026thinsp;Tregs and Memory-1 CD4\u0026thinsp;+\u0026thinsp;T cells may vary by disease status.\u003c/p\u003e\u003cp\u003eGDM is associated with differential expression in Treg subsets\u003c/p\u003e\u003cp\u003eTo determine whether Treg subsets have altered transcriptomes in GDM, we performed differential expression analysis using Seurat and MAST (\u003cb\u003eMethods\u003c/b\u003e). Here, we compared GDM and healthy controls within each Treg subset, excluding the Humanin\u0026thinsp;+\u0026thinsp;and \u003cem\u003eMALAT1\u0026thinsp;+\u003c/em\u003e\u0026thinsp;populations as they were not exclusive to Tregs. We detected an average of 9 and 21 significant differentially expressed genes in effector and naive Treg subsets respectively (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eA). \u003cem\u003eJUN, JUND, MT-CO1 and MT-CYB\u003c/em\u003e were widely downregulated across GDM patient Tregs. The Naive-2 subset showed significantly reduced expression of all four marker genes \u003cem\u003eJUN, FOS, EGR1\u003c/em\u003e and \u003cem\u003eIER2\u003c/em\u003e, which are components of the AP-1 transcription factor associated with Treg development \u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eTo determine the transcriptional pathways altered in each Treg subset, we performed gene set enrichment analysis using the Hallmark Gene Sets from MSigDb\u003csup\u003e\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e\u003c/sup\u003e. We observed that Effector-2 Tregs were significantly enriched for the Angiogenesis hallmark gene set, driven by \u003cem\u003eTIMP1\u003c/em\u003e, \u003cem\u003eS100A4\u003c/em\u003e, and \u003cem\u003eAPP\u003c/em\u003e expression (NES\u0026thinsp;=\u0026thinsp;1.75, FDR\u0026thinsp;=\u0026thinsp;0.005) (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eB). Genes involved in TNF-α (Tumor Necrosis Factor Alpha) signalling by NF-κB (Nuclear Factor Kappa B) were downregulated in the Naive-2 cluster (NES=-1.93, FDR\u0026thinsp;=\u0026thinsp;0.055). This downregulation was most pronounced in \u003cem\u003eFOSB\u003c/em\u003e, \u003cem\u003eEGR1\u003c/em\u003e, \u003cem\u003eKLF2\u003c/em\u003e, \u003cem\u003eNFKBIA\u003c/em\u003e, DUSP2, and \u003cem\u003eJUNB\u003c/em\u003e, several of which are uniquely expressed in the Naive-2 cluster (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). This suggests that the transcriptional programme underlying Naive-2 Treg subsets may be disrupted in GDM.\u003c/p\u003e \u003cp\u003eTreg dysregulation identifies candidate whole blood markers of GDM\u003c/p\u003e\u003cp\u003eWe hypothesised that genes differentially expressed within GDM Treg subsets might serve as molecular markers of GDM in whole blood. To test this hypothesis, we first built a predictive model using clinical features associated with elevated GDM risk, namely patient body mass index (BMI), ethnicity and mean arterial pressure (MAP)\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e. We fit a logistic regression model to these data and established that BMI had the strongest independent association with GDM status (OR: 1.35, 95% CI: 1.01\u0026ndash;2.17) (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eA-B). Evaluating BMI as a predictor by leave-one-out cross-validation (LOO-CV) returned an AUC of 0.718, representing a baseline for comparison.\u003c/p\u003e \u003cp\u003eTo evaluate the predictive power of Treg marker gene expression, we compiled a panel of 108 candidate marker genes defined by significant differential expression within and between Treg subsets (\u003cb\u003eTable S2\u003c/b\u003e). In all comparisons, we benchmark individual genes using logistic regression with LOO-CV to predict GDM status (\u003cb\u003eMethods\u003c/b\u003e). First, we sought to confirm that GDM Treg marker genes were predictive of the average cell signals that would be observed in CD4\u0026thinsp;+\u0026thinsp;cell isolation experiments from blood. We averaged scRNA-seq of Tregs (discovery cohort) and CD4\u0026thinsp;+\u0026thinsp;T cell populations and fit independent models which determined that the highest AUC was achieved using \u003cem\u003eRPS18\u003c/em\u003e (Treg\u0026thinsp;=\u0026thinsp;0.79, CD4\u0026thinsp;=\u0026thinsp;0.75) and \u003cem\u003eMT-CO3\u003c/em\u003e (Treg\u0026thinsp;=\u0026thinsp;0.78, CD4\u0026thinsp;=\u0026thinsp;0.70) (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eC). This was as expected given that these genes were broadly downregulated across Treg clusters (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eA). Interestingly, \u003cem\u003eANXA2\u003c/em\u003e, which identifies the Effector-2 Treg subset, had a higher AUC in CD4\u0026thinsp;+\u0026thinsp;T cells (Treg\u0026thinsp;=\u0026thinsp;0.48, CD4\u0026thinsp;=\u0026thinsp;0.72), likely due to being a significant marker of the Memory-1 CD4\u0026thinsp;+\u0026thinsp;T cell population which showed lower percentages in GDM. These data suggest that genes differentially expressed in GDM CD4\u0026thinsp;+\u0026thinsp;T and Treg subsets may be used to predict disease from bulk mRNA.\u003c/p\u003e \u003cp\u003eTo investigate whether GDM Treg marker genes are predictive of GDM in bulk RNA-seq datasets, we acquired previously published bulk RNA sequencing data from two independent studies on GDM patients and healthy controls \u003csup\u003e\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e,\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u003c/sup\u003e. Logistic regression models fit to each marker gene, and evaluated with LOO-CV, identified 21/108 markers with AUC\u0026thinsp;\u0026ge;\u0026thinsp;0.5 in at least one bulk mRNA dataset (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eD). We detected several genes whose expression in bulk mRNA produced models with AUC values similar to that of BMI in at least one dataset, including \u003cem\u003eTXNIP, MT-ATP8, H3F3B, RPS18, ANXA2, HLA-DRA\u003c/em\u003e and \u003cem\u003eHLA-DRB1\u003c/em\u003e (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eE). However, \u003cem\u003eTXNIP\u003c/em\u003e, which was downregulated in naive Tregs from scRNA, was the only gene which returned a model with a high AUC in both bulk mRNA datasets (WANG\u0026thinsp;=\u0026thinsp;1; STIRM\u0026thinsp;=\u0026thinsp;0.70) despite having a low predictive value in the pseudobulk samples. \u003cem\u003eTXNIP\u003c/em\u003e regulates glucose uptake and is an active target for anti-diabetic drugs due to its association with insulin uptake and Metformin activity\u003csup\u003e\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/sup\u003e. This suggests that Treg function may be affected by global \u003cem\u003eTXNIP\u003c/em\u003e signalling in GDM patients. Collectively, these findings imply that genes involved in GDM Treg dysregulation may serve as predictive biomarkers for GDM, with performance comparable to known GDM risk factors such as BMI.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eRegulatory T cell dysfunction is proposed to underlie the pathology of gestational diabetes mellitus (GDM). We present single-cell transcriptomes of Tregs and CD4\u0026thinsp;+\u0026thinsp;T cells isolated from PBMCs of women with GDM and healthy pregnant women. Our analysis identifies naive and effector molecular subsets in Tregs with a significant increase in memory CD4\u0026thinsp;+\u0026thinsp;T cells from GDM patients, suggesting impaired immune suppression. Differential expression analysis revealed a decrease in AP-1 transcription factor subunits in the Naive-2 cluster, and pathway analysis indicated downregulated NF-kB signalling. Additionally, Effector-2 Tregs upregulated genes involved in Angiogenesis. We evaluated the translational potential of scRNA-seq-derived expression markers from CD4\u0026thinsp;+\u0026thinsp;T cell populations and validated their predictive potential in pseudobulk CD4\u0026thinsp;+\u0026thinsp;T cells and independent validation cohorts.\u003c/p\u003e \u003cp\u003eThe percentage of Tregs in GDM patients is observed to be reduced in women with GDM \u003csup\u003e\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e. We identified FOXP3\u0026thinsp;+\u0026thinsp;cells within sorted CD4\u0026thinsp;+\u0026thinsp;T cells which we expect to contain Tregs, however, their proportion did not significantly differ between conditions. FOXP3\u0026thinsp;+\u0026thinsp;cells represent a low proportion of the CD4\u0026thinsp;+\u0026thinsp;T cell pool. Therefore we may be underpowered to observe the effect at a single cell level. Of note, we observed Humanin\u0026thinsp;+\u0026thinsp;and \u003cem\u003eMALAT1\u003c/em\u003e\u0026thinsp;+\u0026thinsp;cell clusters in both Treg and CD4\u0026thinsp;+\u0026thinsp;T cell populations. These cells have been previously reported in sorted Tregs from ankylosing spondylitis patients profiled with the 10X scRNA-seq platform\u003csup\u003e\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e\u003c/sup\u003e. We also observed Humanin\u0026thinsp;+\u0026thinsp;CD4\u0026thinsp;+\u0026thinsp;T cells, have been reported in the context of Rheumatoid Arthritis \u003csup\u003e\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e\u003c/sup\u003e, To our knowledge, this is the first report of Humanin-positive cells in GDM. Altered humanin concentrations have been proposed as a biomarker for diabetic conditions, as Humanin is protective under oxidative stress, which can arise from excess ROS generation from diabetic hyperglycaemia \u003csup\u003e\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e,\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e\u003c/sup\u003e. Further research into the phenotypes of these Treg subsets is required to understand their role in GDM.\u003c/p\u003e \u003cp\u003eNF-kB signalling is associated with apoptosis, insulin resistance, systemic inflammation and all major diabetic complications \u003csup\u003e\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e\u003c/sup\u003e. \u003cem\u003eNFKBIA\u003c/em\u003e is upregulated in whole blood from GDM patients \u003csup\u003e\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e\u003c/sup\u003e. Similarly, the TNF-alpha signalling via NF-kb hallmark pathway was disrupted through RNA-seq of a murine GDM model \u003csup\u003e\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e\u003c/sup\u003e. We observed a decrease in \u003cem\u003eNFKBIA\u003c/em\u003e signalling specific to the Naive-2 GDM Treg subset. Dimerisation between NF-kB and FOXP3 is essential for Treg activation and disruption to this pathway leads to immune dysregulation and altered Treg frequencies \u003csup\u003e\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e\u003c/sup\u003e. Therefore this suggests a potential mechanism by which GDM Tregs would be less effective at managing inflammatory responses. Additionally, the Effector-2 population was significantly enriched for angiogenesis pathway genes. Although chronic inflammation and angiogenesis are symptoms of GDM promoted by Tregs \u003csup\u003e\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e,\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e\u003c/sup\u003e, it remains unclear how these genes drive the suggested phenotype. Future work will require isolation and phenotypic characterisation of the Effector-2 and Naive-2 cell populations in order to discern their roles in GDM.\u003c/p\u003e \u003cp\u003eBody mass index (BMI) is the most significant risk factor for GDM and maintaining a BMI below 25kg m\u003csup\u003e\u0026minus;\u0026thinsp;2\u003c/sup\u003e is recommended to reduce GDM risk \u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e. Additionally, a lack of globally accepted diagnostic thresholds and reliance on BMI may lead to undiagnosed patients, outlining the need for predictive biomarkers for GDM testing \u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u003c/sup\u003e. We benchmarked molecular markers against BMI and found them to have similar predictive value. Rapid protocols are readily available for CD4\u0026thinsp;+\u0026thinsp;T cell isolation from whole blood \u003csup\u003e\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e\u003c/sup\u003e. Emulating this expression analysis using pseudobulk scRNA-seq expression, we observed that genes broadly altered across Treg subsets were highly predictive of GDM, such as \u003cem\u003eRPS18\u003c/em\u003e, \u003cem\u003eMT-CO1\u003c/em\u003e and \u003cem\u003eMT-CO3\u003c/em\u003e. \u003cem\u003eMT-CO1\u003c/em\u003e and \u003cem\u003eMT-CO3\u003c/em\u003e are subunits of Cytochrome c oxidase II involved in ATP synthesis, and mutations in \u003cem\u003eMT-CO3\u003c/em\u003e are associated with maternally inherited mitochondrial diabetes\u003csup\u003e\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e\u003c/sup\u003e. Their predictive value may be due to a hyperglycaemic environment in GDM that affects Treg signalling. Expanding our analysis to bulk mRNA from independent cohorts, the Treg marker \u003cem\u003eTXNIP\u003c/em\u003e returned the highest AUC. Given its role in diabetes and inflammation, \u003cem\u003eTXNIP\u003c/em\u003e represents a promising candidate biomarker for GDM \u003csup\u003e\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eWhile our study sheds light on the Treg transcriptional landscape in GDM, the focus on isolated CD4\u0026thinsp;+\u0026thinsp;cells carries limitations. Several immune subtypes that were not studied are associated with the pathophysiology of GDM, such as Neutrophils, Macrophages and Monocytes \u003csup\u003e\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e\u003c/sup\u003e. Functional studies in models of GDM may further uncover the role of immune cells in attenuating Treg activity. For example, we observed the under-expression of \u003cem\u003eJUNB\u003c/em\u003e, the AP-1 transcription factor, which is demonstrated to regulate intestinal Treg development and its ablation is associated with increased T helper accumulation and inflammation \u003csup\u003e\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e\u003c/sup\u003e. Although single-cell RNA-seq empowers us to discover cell type-specific expression, this analysis remains limited to protein-coding RNAs. In contrast, several non-coding RNAs such as microRNAs\u003csup\u003e\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e\u003c/sup\u003e and circular RNAs\u003csup\u003e\u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e\u003c/sup\u003e have been proposed as candidate circulating biomarkers for GDM. Studying the effect of non-coding RNAs on T regulatory cell subtypes may uncover insights into GDM pathology.\u003c/p\u003e \u003cp\u003eIn conclusion, this study identifies Treg subsets with altered transcriptional programmes in GDM patients and proposes candidate marker genes with translational potential that arise from Treg dysregulation. Chronic inflammation leading to metabolic dysregulation is one coupling factor driving the association between obesity, cardiovascular disease, and diabetes\u003csup\u003e\u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e\u003c/sup\u003e. Future work investigating these molecular markers of Treg dysfunction will improve our understanding of the complex immunometabolic state observed in GDM.\u003c/p\u003e \n"},{"header":"Methods","content":"\u003ch3\u003ePBMC isolation and FACS\u003c/h3\u003e\n\u003cp\u003eVenous blood was collected from 23 participating pregnant women at 35\u0026ndash;36 weeks of gestation, of whom 10 were diagnosed with GDM and the remainder were healthy controls. Blood was captured in BD Vacutainer\u0026reg; EDTA Tubes with a volume of 10 mL, and the samples were kept at room temperature (18\u0026ndash;22\u0026deg;C) while being agitated using an orbital shaker at 90\u0026ndash;100 rpm. Peripheral blood mononuclear cells (PBMCs) were isolated as previously described \u003csup\u003e\u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e53\u003c/span\u003e\u003c/sup\u003e. Briefly, Ficoll-Paque PLUS medium was placed into a SepMate\u0026trade;-50 tube, and the collected blood was transferred to a separate 50 mL Falcon tube where it was diluted 1:1 with PBS-EDTA containing 2% FBS. This diluted sample was added to the SepMate\u0026trade; tube and then centrifuged at 1200 x g for 20 minutes. After isolation, cells were counted and resuspended in 1 mL of freezing medium containing DMSO. Cryovials containing the cells were kept on wet ice for less than 5 minutes before commencing freezing. The cryovials were then placed in a Corning CoolCell chamber at temperatures below \u0026minus;\u0026thinsp;70\u0026deg;C. After a minimum of 12 hours, these were transferred to cryo boxes and stored in a -80\u0026deg;C freezer for up to 24 hours before their final transfer to a liquid nitrogen tank.\u003c/p\u003e \u003cp\u003eIn the thawing process, the cryovial containing the frozen cells was submerged in a 37\u0026deg;C water bath for approximately one minute. Once thawed, 1 mL of pre-warmed medium is promptly added to the cryovial using a transfer pipette. The contents were then transferred to a 15 mL conical Falcon\u0026trade; tube, which was prefilled with 5 mL of medium also warmed to 37\u0026deg;C. The empty cryovial was then rinsed with an additional 1 mL of medium to ensure complete transfer of cells. Following this, the Falcon\u0026trade; tube was incubated in the 37\u0026deg;C water bath for 5 minutes. After incubation, the tube was centrifuged at 300\u0026times;g at room temperature for another 5 minutes. The supernatant was discarded. PBMCs were washed twice in PBS and resuspended. Cells were counted using a hemocytometer and trypan blue to achieve a density of 700\u0026ndash;1200 cells per ul. The cells were then stained for CD4 AlexaFLuor 700, CD4 BUV395, CD25 TexasRed and CD127 BV786. DAPI was used to exclude dead cells. The Tregs and CD4\u0026thinsp;+\u0026thinsp;T cells were sorted in a BD FACS Aria III in 1.5ml Eppendorf using the gating strategy shown in \u003cb\u003eSupplementary Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e\u003c/b\u003e.\u003c/p\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003eSingle-cell RNA sequencing\u003c/h2\u003e \u003cp\u003eSingle-cell mRNA, dual-indexed, sequencing libraries were generated following the Chromium Single Cell 5\u0026rsquo; v2 protocol with Feature Barcode technology for Cell Surface Protein and Immune Receptor Mapping (CG000330, 10x Genomics). For each 10x run, patient samples were hash-tagged and pooled, using one antibody-derived tag per patient following the CITE-Seq protocol \u003csup\u003e\u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e\u003c/sup\u003e (Biolegend) and then flow-sorted to generate separate compartments of CD4\u0026thinsp;+\u0026thinsp;T-cells and FOXP3\u0026thinsp;+\u0026thinsp;T-cells before proceeding with the 10x assay. Gel beads-in-emulsion (GEMs) were generated on the chromium controller using reagents specific for the 5\u0026rsquo; assay. 10x barcoded, full-length cDNA from poly-adenylated mRNA and DNA from protein Feature Barcode were generated and amplified. The amplified cDNA was used to generate gene expression and V(D)J sequencing libraries. To generate the V(D)J sequencing libraries, full-length, 10x barcoded V(D)J segments were enriched from amplified cDNA via further amplification using primers specific to TCR constant regions and converted into sequencing libraries. Cell surface protein Feature Barcode-sequencing libraries were generated from amplified DNA to detect the antibody barcodes used for cell hashing. The dual indexed libraries thus generated for 5\u0026rsquo; gene expression, V(D)J and Cell Surface Protein were pooled and sequenced on Illumina NextSeq 2000 and NovaSeq 6000 platform to generate paired-end reads (28-10-10-90) following 10x recommended sequencing depth.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003eData processing and quality control\u003c/h2\u003e \u003cp\u003eThe cellranger (v6.1.1, 10X Genomics) \u003cem\u003emkfastq\u003c/em\u003e command was used to demultiplex raw sequencing data into FASTQ files and reads were aligned to the GRCh38 and converted to raw feature counts using the cellranger \u003cem\u003emulti\u003c/em\u003e command. Count tables were imported into R (v4.1.3) and analysed using Seurat (v4.1.0). For all downstream analyses, gene expression counts for isolated Treg and CD4\u0026thinsp;+\u0026thinsp;T cells were processed separately.\u003c/p\u003e \u003cp\u003eCells were assigned to hashtag sequences using the Seurat \u003cem\u003eHTODemux\u003c/em\u003e function, and only cells identified as Singlets were retained. To detect sample mix-ups, transcript BAM files were demultiplexed with custom scripts and germline SNPs were identified per patient using cellSNP (v0.3.2). Sample mixups were labelled if pairwise SNP allele frequency Pearson correlations exceeded a threshold of 0.8. This procedure correctly identified one known duplicate sample, which was removed from subsequent analysis.\u003c/p\u003e \u003cp\u003eTo detect outliers cells, cells with UMI counts\u0026thinsp;\u0026gt;\u0026thinsp;4x the absolute deviation were removed from downstream analysis. Furthermore, cells with transcript counts for fewer than 350 genes were excluded due to low capture. Cells were further excluded if fewer than 20% of known housekeeping genes or over 5% of mitochondrial genes were expressed. Supervised detection of cell types was performed using SingleR (v1.8.1) and the cell dex database (v1.4.0), from which all barcodes assigned the \u0026lsquo;T cell\u0026rsquo; label were retained for downstream analysis.\u003c/p\u003e \u003cp\u003eTranscript counts were normalised and then integrated using the Seurat sctransform workflow, combining the functions \u003cem\u003eSCTransform, SelectIntegrationFeatures, PrepSCTIntegration, FindIntegrationAnchors and IntegrateData\u003c/em\u003e. Cell cycle scores estimated with the function \u003cem\u003eCellCycleScoring\u003c/em\u003e were regressed using \u003cem\u003eSCTransform\u003c/em\u003e. We captured 22,218 Treg cells (mean: 966, SD:383.5) and 23,737 Tconv cells (mean: 1032, SD: 538.8) per patient following quality control. Across conditions, 43.57% of Tregs and 49.5% of Tconvs were derived from GDM patients. Seurat objects for Treg and CD4\u0026thinsp;+\u0026thinsp;T cell populations were processed separately in all downstream analyses.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eT cell subpopulation detection and visualisation\u003c/h3\u003e\n\u003cp\u003eCells were clustered by running Seurat\u0026rsquo;s \u003cem\u003eRunPCA\u003c/em\u003e function with 50 principal components, followed by the functions \u003cem\u003eFindNeighbors\u003c/em\u003e and \u003cem\u003eFindClusters\u003c/em\u003e (Treg resolution\u0026thinsp;=\u0026thinsp;0.75, CD4\u0026thinsp;+\u0026thinsp;resolution\u0026thinsp;=\u0026thinsp;0.9). T cell receptor (TCR) genes were filtered from variable features in the \u003cem\u003eIntegrated\u003c/em\u003e dataset to reduce their influence on cell clustering. For visualisation, UMAP and tSNE embeddings were derived from the Seurat \u003cem\u003eRunUMAP\u003c/em\u003e (Treg components\u0026thinsp;=\u0026thinsp;2; Treg neighbors\u0026thinsp;=\u0026thinsp;50; CD4\u0026thinsp;+\u0026thinsp;components\u0026thinsp;=\u0026thinsp;2; CD4\u0026thinsp;+\u0026thinsp;neighbors\u0026thinsp;=\u0026thinsp;7) and \u003cem\u003eRunTSNE\u003c/em\u003e functions (Treg perplexity\u0026thinsp;=\u0026thinsp;15; CD4\u0026thinsp;+\u0026thinsp;perplexity\u0026thinsp;=\u0026thinsp;35).\u003c/p\u003e \u003cp\u003eClusters detected in CD4\u0026thinsp;+\u0026thinsp;and Treg Seurat objects were assigned broad groups of Naive, Effector or Memory by computational gating on the expression of phenotypic marker genes previously reported for conventional T cells and Tregs \u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e,\u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e55\u003c/span\u003e\u003c/sup\u003e. Clusters were further distinguished by the top marker genes as detected using the Seurat function \u003cem\u003eFindAllMarkers\u003c/em\u003e.\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eDifferential abundance\u003c/h2\u003e \u003cp\u003eThe propeller function from the R package speckle (v0.99.7) was used to test for differences in cell type proportions between GDM and control patients, separately for CD4\u0026thinsp;+\u0026thinsp;T cells and FOXP3\u0026thinsp;+\u0026thinsp;Tregs. Propeller calculates patient-level cell type proportions from cell annotations, performs a transformation on the counts (logit transform in this analysis), and applies a moderated \u003cem\u003et-\u003c/em\u003etest to find significant differences between cell type proportions \u003csup\u003e\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003eDifferential expression\u003c/h2\u003e \u003cp\u003eWithin-cluster differential expression testing was performed using the Seurat \u003cem\u003eFindMarkers\u003c/em\u003e with the MAST (v1.25.2) testing framework applied to fit generalised linear models adapted for the bimodality and zero-inflated nature of single-cell gene expression data \u003csup\u003e\u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e56\u003c/span\u003e\u003c/sup\u003e. Cells from each unique cluster were filtered from Seurat SCT assay objects prior to MAST analysis. Genes with fewer than 10% of cells expressing transcripts in either GDM or control populations, Humanin\u0026thinsp;+\u0026thinsp;clusters and ribosomal proteins were excluded from downstream analysis. Significantly differentially expressed genes were defined as genes where the absolute average Log2 fold-change was more than 0.25 and the adjusted \u003cem\u003ep-value\u003c/em\u003e was below 0.05.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003eGene set enrichment analysis\u003c/h2\u003e \u003cp\u003eGene set enrichment analysis was performed by testing all genes analysed in the differential expression analysis for their enrichment in the 50 Human MSigDB Hallmark gene sets (v7.5.1). For each within-cluster comparison, genes were ranked by their average Log2Fold-change and tested against Hallmark gene sets using fgsea (v1.24.0). CD4\u0026thinsp;+\u0026thinsp;T cells and Treg populations were treated separately.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eGDM risk factor analysis\u003c/h2\u003e \u003cp\u003eGDM risk factors were collected from patients at the time of recruitment, including ethnicity, body mass index (BMI), diastolic and systolic blood pressure, and interventions such as aspirin and metformin. Mean arterial pressure (MAP) was calculated using the commonly used formula summing the diastolic pressure and one-third of the pulse pressure, the difference between systolic and diastolic pressure \u003csup\u003e\u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e57\u003c/span\u003e\u003c/sup\u003e. The R \u003cem\u003eglm\u003c/em\u003e function was used to fit logistic regression models to BMI, MAP and ethnicity as predictors of GDM status. Leave-one-out cross-validation was implemented. The area under the receiver operating characteristic curve (AUC) was calculated from the results of each fold to evaluate the model.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eGDM classification from RNA-seq\u003c/h2\u003e \u003cp\u003ePseudobulk expression profiles were generated from CD4\u0026thinsp;+\u0026thinsp;T cell populations by averaging cell transcript counts from the Seurat RNA assay for each patient using Seurat \u003cem\u003eAverageExpression\u003c/em\u003e function. Bulk mRNA profiles were acquired from previously published studies on GDM and control patients \u003csup\u003e\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e,\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u003c/sup\u003e, and GDM status was converted to binary outcome in all datasets prior to downstream analysis.\u003c/p\u003e \u003cp\u003eModels were built using marker genes from scRNA-seq differential expression results. Specifically, GDM marker gene sets were defined by combining: differentially expressed genes in sorted Treg clusters, to capture specific signals of Treg malfunction.\u003c/p\u003e \u003cp\u003eLogistic regression with leave-one-out cross-validation was applied to mRNA data. A single model was built for each gene as a predictor of GDM status to compare predictor performance across mRNA and CD4\u0026thinsp;+\u0026thinsp;T cell pseudobulk datasets.\u003c/p\u003e \u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eData availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSingle-cell RNA sequencing data has been submitted to Gene Expression Omnibus (accession pending). Bulk RNA sequencing data from published studies is available at GSE154414 and GSE92772. Scripts and analysis pipelines have been deposited at https://github.com/NMNS93/rutepo_gdm_treg.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study was approved by the King\u0026rsquo;s College Hospital Research Ethics Committee, REC number 02-03-033, dated 01/04/2003. All the experiments conform to the relevant regulatory standards, including proper recruitment and obtaining informed consent.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eN.E.M.* , N.M*. and A.E.* contributed equally to this work. N.E.M.*: Analyzed data, wrote and edited the manuscript. A.E.*: Consent, sample collection, and data analysis. N.M.: Consent, sample collection, and data analysis. S.K.: Data analysis. H.V.: Conducted experiments. A.B.: Data analysis and manuscript editing. A.M.: Data analysis. T.T.: Provided expert opinion. G.L.: Provided expert opinion and edited the manuscript. P.D.: Data analysis and provided expert opinion. K.H.N.: Provided expert opinion and edited the manuscript. C.S.: Provided expert opinion and edited the manuscript. P.S.: Conceptualized the study, conducted experiments, collected samples, and edited the manuscript.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eMcIntyre, H. D. \u003cem\u003eet al.\u003c/em\u003e Gestational diabetes mellitus. \u003cem\u003eNat Rev Dis Primers\u003c/em\u003e \u003cstrong\u003e5\u003c/strong\u003e, 47 (2019).\u003c/li\u003e\n\u003cli\u003eJiang, L. \u003cem\u003eet al.\u003c/em\u003e A global view of hypertensive disorders and diabetes mellitus during pregnancy. \u003cem\u003eNat. Rev. Endocrinol.\u003c/em\u003e \u003cstrong\u003e18\u003c/strong\u003e, 760\u0026ndash;775 (2022).\u003c/li\u003e\n\u003cli\u003eSweeting, A., Wong, J., Murphy, H. R. \u0026amp; Ross, G. P. A Clinical Update on Gestational Diabetes Mellitus. \u003cem\u003eEndocr. Rev.\u003c/em\u003e \u003cstrong\u003e43\u003c/strong\u003e, 763\u0026ndash;793 (2022).\u003c/li\u003e\n\u003cli\u003eAmerican Diabetes Association. 2. Classification and Diagnosis of Diabetes: Standards of Medical Care in Diabetes-2020. \u003cem\u003eDiabetes Care\u003c/em\u003e \u003cstrong\u003e43\u003c/strong\u003e, S14\u0026ndash;S31 (2020).\u003c/li\u003e\n\u003cli\u003eBellamy, L., Casas, J.-P., Hingorani, A. D. \u0026amp; Williams, D. Type 2 diabetes mellitus after gestational diabetes: a systematic review and meta-analysis. \u003cem\u003eLancet\u003c/em\u003e \u003cstrong\u003e373\u003c/strong\u003e, 1773\u0026ndash;1779 (2009).\u003c/li\u003e\n\u003cli\u003eNoctor, E. \u0026amp; Dunne, F. P. Type 2 diabetes after gestational diabetes: The influence of changing diagnostic criteria. \u003cem\u003eWorld J. Diabetes\u003c/em\u003e \u003cstrong\u003e6\u003c/strong\u003e, 234\u0026ndash;244 (2015).\u003c/li\u003e\n\u003cli\u003eOrnoy, A., Becker, M., Weinstein-Fudim, L. \u0026amp; Ergaz, Z. Diabetes during Pregnancy: A Maternal Disease Complicating the Course of Pregnancy with Long-Term Deleterious Effects on the Offspring. A Clinical Review. \u003cem\u003eInt. J. Mol. Sci.\u003c/em\u003e \u003cstrong\u003e22\u003c/strong\u003e, (2021).\u003c/li\u003e\n\u003cli\u003eSakaguchi, S., Yamaguchi, T., Nomura, T. \u0026amp; Ono, M. Regulatory T cells and immune tolerance. \u003cem\u003eCell\u003c/em\u003e \u003cstrong\u003e133\u003c/strong\u003e, 775\u0026ndash;787 (2008).\u003c/li\u003e\n\u003cli\u003eChurov, A. V., Mamashov, K. Y. \u0026amp; Novitskaia, A. V. Homeostasis and the functional roles of CD4+ Treg cells in aging. \u003cem\u003eImmunol. Lett.\u003c/em\u003e \u003cstrong\u003e226\u003c/strong\u003e, 83\u0026ndash;89 (2020).\u003c/li\u003e\n\u003cli\u003eHori, S., Nomura, T. \u0026amp; Sakaguchi, S. Control of regulatory T cell development by the transcription factor Foxp3. \u003cem\u003eScience\u003c/em\u003e \u003cstrong\u003e299\u003c/strong\u003e, 1057\u0026ndash;1061 (2003).\u003c/li\u003e\n\u003cli\u003eRosenblum, M. D., Way, S. S. \u0026amp; Abbas, A. K. Regulatory T cell memory. \u003cem\u003eNat. Rev. Immunol.\u003c/em\u003e \u003cstrong\u003e16\u003c/strong\u003e, 90\u0026ndash;101 (2015).\u003c/li\u003e\n\u003cli\u003eZemmour, D. \u003cem\u003eet al.\u003c/em\u003e Single-cell gene expression reveals a landscape of regulatory T cell phenotypes shaped by the TCR. \u003cem\u003eNat. Immunol.\u003c/em\u003e \u003cstrong\u003e19\u003c/strong\u003e, 291\u0026ndash;301 (2018).\u003c/li\u003e\n\u003cli\u003eKrop, J., Heidt, S., Claas, F. H. J. \u0026amp; Eikmans, M. Regulatory T Cells in Pregnancy: It Is Not All About FoxP3. \u003cem\u003eFront. Immunol.\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e, 1182 (2020).\u003c/li\u003e\n\u003cli\u003eGreen, S. \u003cem\u003eet al.\u003c/em\u003e Regulatory T Cells in Pregnancy Adverse Outcomes: A Systematic Review and Meta-Analysis. \u003cem\u003eFront. Immunol.\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, 737862 (2021).\u003c/li\u003e\n\u003cli\u003eMcElwain, C. J., McCarthy, F. P. \u0026amp; McCarthy, C. M. Gestational Diabetes Mellitus and Maternal Immune Dysregulation: What We Know So Far. \u003cem\u003eInt. J. Mol. Sci.\u003c/em\u003e \u003cstrong\u003e22\u003c/strong\u003e, (2021).\u003c/li\u003e\n\u003cli\u003eYang, Y. \u003cem\u003eet al.\u003c/em\u003e Functional Defects of Regulatory T Cell Through Interleukin 10 Mediated Mechanism in the Induction of Gestational Diabetes Mellitus. \u003cem\u003eDNA Cell Biol.\u003c/em\u003e \u003cstrong\u003e37\u003c/strong\u003e, 278\u0026ndash;285 (2018).\u003c/li\u003e\n\u003cli\u003eLobo, T. F. \u003cem\u003eet al.\u003c/em\u003e Impaired Treg and NK cells profile in overweight women with gestational diabetes mellitus. \u003cem\u003eAm. J. Reprod. Immunol.\u003c/em\u003e \u003cstrong\u003e79\u003c/strong\u003e, (2018).\u003c/li\u003e\n\u003cli\u003eSchober, L. \u003cem\u003eet al.\u003c/em\u003e The role of regulatory T cell (Treg) subsets in gestational diabetes mellitus. \u003cem\u003eClin. Exp. Immunol.\u003c/em\u003e \u003cstrong\u003e177\u003c/strong\u003e, 76\u0026ndash;85 (2014).\u003c/li\u003e\n\u003cli\u003eSifnaios, E. \u003cem\u003eet al.\u003c/em\u003e Gestational Diabetes and T-cell (Th1/Th2/Th17/Treg) Immune Profile. \u003cem\u003eIn Vivo\u003c/em\u003e \u003cstrong\u003e33\u003c/strong\u003e, 31\u0026ndash;40 (2019).\u003c/li\u003e\n\u003cli\u003eArain, H. \u003cem\u003eet al.\u003c/em\u003e Regulatory T cells in the peripheral blood of women with gestational diabetes: a systematic review and meta-analysis. \u003cem\u003eFront. Immunol.\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, (2023).\u003c/li\u003e\n\u003cli\u003eSzabo, P. A. \u003cem\u003eet al.\u003c/em\u003e Single-cell transcriptomics of human T cells reveals tissue and activation signatures in health and disease. \u003cem\u003eNat. Commun.\u003c/em\u003e \u003cstrong\u003e10\u003c/strong\u003e, 4706 (2019).\u003c/li\u003e\n\u003cli\u003eYang, Y. \u003cem\u003eet al.\u003c/em\u003e Transcriptomic Profiling of Human Placenta in Gestational Diabetes Mellitus at the Single-Cell Level. \u003cem\u003eFront. Endocrinol. \u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, 679582 (2021).\u003c/li\u003e\n\u003cli\u003eJiao, B. \u003cem\u003eet al.\u003c/em\u003e Dissecting human placental cells heterogeneity in preeclampsia and gestational diabetes using single-cell sequencing. \u003cem\u003eMol. Immunol.\u003c/em\u003e \u003cstrong\u003e161\u003c/strong\u003e, 104\u0026ndash;118 (2023).\u003c/li\u003e\n\u003cli\u003eFarahvar, S., Walfisch, A. \u0026amp; Sheiner, E. Gestational diabetes risk factors and long-term consequences for both mother and offspring: a literature review. \u003cem\u003eExpert Rev. Endocrinol. Metab.\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, 63\u0026ndash;74 (2019).\u003c/li\u003e\n\u003cli\u003eRodrigo, N. \u0026amp; Glastras, S. J. The Emerging Role of Biomarkers in the Diagnosis of Gestational Diabetes Mellitus. \u003cem\u003eJ. Clin. Med. Res.\u003c/em\u003e \u003cstrong\u003e7\u003c/strong\u003e, (2018).\u003c/li\u003e\n\u003cli\u003eLee, S.-M., Gao, B. \u0026amp; Fang, D. FoxP3 maintains Treg unresponsiveness by selectively inhibiting the promoter DNA-binding activity of AP-1. \u003cem\u003eBlood\u003c/em\u003e \u003cstrong\u003e111\u003c/strong\u003e, 3599\u0026ndash;3606 (2008).\u003c/li\u003e\n\u003cli\u003eBahrami, S. \u0026amp; Drabl\u0026oslash;s, F. Gene regulation in the immediate-early response process. \u003cem\u003eAdv. Biol. Regul.\u003c/em\u003e \u003cstrong\u003e62\u003c/strong\u003e, 37\u0026ndash;49 (2016).\u003c/li\u003e\n\u003cli\u003eKatagiri, T., Kameda, H., Nakano, H. \u0026amp; Yamazaki, S. Regulation of T cell differentiation by the AP-1 transcription factor JunB. \u003cem\u003eImmunol Med\u003c/em\u003e \u003cstrong\u003e44\u003c/strong\u003e, 197\u0026ndash;203 (2021).\u003c/li\u003e\n\u003cli\u003eDey, S. \u003cem\u003eet al.\u003c/em\u003e Downregulation of MALAT1 is a hallmark of tissue and peripheral proliferative T cells in COVID-19. \u003cem\u003eClin. Exp. Immunol.\u003c/em\u003e \u003cstrong\u003e212\u003c/strong\u003e, 262\u0026ndash;275 (2023).\u003c/li\u003e\n\u003cli\u003eMasoumi, F. \u003cem\u003eet al.\u003c/em\u003e Malat1 long noncoding RNA regulates inflammation and leukocyte differentiation in experimental autoimmune encephalomyelitis. \u003cem\u003eJ. Neuroimmunol.\u003c/em\u003e \u003cstrong\u003e328\u003c/strong\u003e, 50\u0026ndash;59 (2019).\u003c/li\u003e\n\u003cli\u003eBoutari, C., Pappas, P. D., Theodoridis, T. D. \u0026amp; Vavilis, D. Humanin and diabetes mellitus: A review of in vitro and in vivo studies. \u003cem\u003eWorld J. Diabetes\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, 213\u0026ndash;223 (2022).\u003c/li\u003e\n\u003cli\u003ePhipson, B. \u003cem\u003eet al.\u003c/em\u003e propeller: testing for differences in cell type proportions in single cell data. \u003cem\u003eBioinformatics\u003c/em\u003e \u003cstrong\u003e38\u003c/strong\u003e, 4720\u0026ndash;4726 (2022).\u003c/li\u003e\n\u003cli\u003eLiberzon, A. \u003cem\u003eet al.\u003c/em\u003e The Molecular Signatures Database (MSigDB) hallmark gene set collection. \u003cem\u003eCell Syst\u003c/em\u003e \u003cstrong\u003e1\u003c/strong\u003e, 417\u0026ndash;425 (2015).\u003c/li\u003e\n\u003cli\u003eStirm, L. \u003cem\u003eet al.\u003c/em\u003e Maternal whole blood cell miRNA-340 is elevated in gestational diabetes and inversely regulated by glucose and insulin. \u003cem\u003eSci. Rep.\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e, 1366 (2018).\u003c/li\u003e\n\u003cli\u003eWang, J., Wang, K., Liu, W., Cai, Y. \u0026amp; Jin, H. m6A mRNA methylation regulates the development of gestational diabetes mellitus in Han Chinese women. \u003cem\u003eGenomics\u003c/em\u003e \u003cstrong\u003e113\u003c/strong\u003e, 1048\u0026ndash;1056 (2021).\u003c/li\u003e\n\u003cli\u003eMasutani, H. Thioredoxin-Interacting Protein in Cancer and Diabetes. \u003cem\u003eAntioxid. Redox Signal.\u003c/em\u003e \u003cstrong\u003e36\u003c/strong\u003e, 1001\u0026ndash;1022 (2022).\u003c/li\u003e\n\u003cli\u003eSimone, D. \u003cem\u003eet al.\u003c/em\u003e Single cell analysis of spondyloarthritis regulatory T cells identifies distinct synovial gene expression patterns and clonal fates. \u003cem\u003eCommun Biol\u003c/em\u003e \u003cstrong\u003e4\u003c/strong\u003e, 1395 (2021).\u003c/li\u003e\n\u003cli\u003eArgyriou, A. \u003cem\u003eet al.\u003c/em\u003e Single cell sequencing reveals expanded cytotoxic CD4+ T cells and two states of peripheral helper T cells in synovial fluid of ACPA+ RA patients. \u003cem\u003ebioRxiv\u003c/em\u003e (2021) doi:10.1101/2021.05.28.21255902.\u003c/li\u003e\n\u003cli\u003eRochette, L., Meloux, A., Zeller, M., Cottin, Y. \u0026amp; Vergely, C. Role of humanin, a mitochondrial-derived peptide, in cardiovascular disorders. \u003cem\u003eArch. Cardiovasc. Dis.\u003c/em\u003e \u003cstrong\u003e113\u003c/strong\u003e, 564\u0026ndash;571 (2020).\u003c/li\u003e\n\u003cli\u003eIndira, M. \u0026amp; Abhilash, P. A. Role of NF-Kappa B (NF-\u0026kappa;B) in Diabetes. \u003cem\u003eOT\u003c/em\u003e \u003cstrong\u003e4\u003c/strong\u003e, (2013).\u003c/li\u003e\n\u003cli\u003eChen, Y.-M. \u003cem\u003eet al.\u003c/em\u003e Upregulation of T Cell Receptor Signaling Pathway Components in Gestational Diabetes Mellitus Patients: Joint Analysis of mRNA and circRNA Expression Profiles. \u003cem\u003eFront. Endocrinol. \u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, 774608 (2021).\u003c/li\u003e\n\u003cli\u003ePaolino, M. \u003cem\u003eet al.\u003c/em\u003e RANK links thymic regulatory T cells to fetal loss and gestational diabetes in pregnancy. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e589\u003c/strong\u003e, 442\u0026ndash;447 (2021).\u003c/li\u003e\n\u003cli\u003eCui, Y. \u003cem\u003eet al.\u003c/em\u003e A Stk4-Foxp3-NF-\u0026kappa;B p65 transcriptional complex promotes Treg cell activation and homeostasis. \u003cem\u003eSci Immunol\u003c/em\u003e \u003cstrong\u003e7\u003c/strong\u003e, eabl8357 (2022).\u003c/li\u003e\n\u003cli\u003eJoshi, N. P., Madiwale, S. D., Sundrani, D. P. \u0026amp; Joshi, S. R. Fatty acids, inflammation and angiogenesis in women with gestational diabetes mellitus. \u003cem\u003eBiochimie\u003c/em\u003e \u003cstrong\u003e212\u003c/strong\u003e, 31\u0026ndash;40 (2023).\u003c/li\u003e\n\u003cli\u003eLužnik, Z., Anchouche, S., Dana, R. \u0026amp; Yin, J. Regulatory T Cells in Angiogenesis. \u003cem\u003eJ. Immunol.\u003c/em\u003e \u003cstrong\u003e205\u003c/strong\u003e, 2557\u0026ndash;2565 (2020).\u003c/li\u003e\n\u003cli\u003eHsu, C.-H., Chen, C., Irimia, D. \u0026amp; Toner, M. Fast sorting of CD4+ T cells from whole blood using glass microbubbles. \u003cem\u003eTechnology\u003c/em\u003e \u003cstrong\u003e3\u003c/strong\u003e, 38\u0026ndash;44 (2015).\u003c/li\u003e\n\u003cli\u003eTabebi, M. \u003cem\u003eet al.\u003c/em\u003e A novel mutation MT-COIII m.9267G\u0026gt;C and MT-COI m.5913G\u0026gt;A mutation in mitochondrial genes in a Tunisian family with maternally inherited diabetes and deafness (MIDD) associated with severe nephropathy. \u003cem\u003eBiochem. Biophys. Res. Commun.\u003c/em\u003e \u003cstrong\u003e459\u003c/strong\u003e, 353\u0026ndash;360 (2015).\u003c/li\u003e\n\u003cli\u003eDe Luccia, T. P. B. \u003cem\u003eet al.\u003c/em\u003e Unveiling the pathophysiology of gestational diabetes: Studies on local and peripheral immune cells. \u003cem\u003eScand. J. Immunol.\u003c/em\u003e \u003cstrong\u003e91\u003c/strong\u003e, e12860 (2020).\u003c/li\u003e\n\u003cli\u003eWheaton, J. D. \u0026amp; Ciofani, M. JunB Controls Intestinal Effector Programs in Regulatory T Cells. \u003cem\u003eFront. Immunol.\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e, 444 (2020).\u003c/li\u003e\n\u003cli\u003eGuarino, E. \u003cem\u003eet al.\u003c/em\u003e Circulating MicroRNAs as Biomarkers of Gestational Diabetes Mellitus: Updates and Perspectives. \u003cem\u003eInt. J. Endocrinol.\u003c/em\u003e \u003cstrong\u003e2018\u003c/strong\u003e, 6380463 (2018).\u003c/li\u003e\n\u003cli\u003eFan, W., Pang, H., Xie, Z., Huang, G. \u0026amp; Zhou, Z. Circular RNAs in diabetes mellitus and its complications. \u003cem\u003eFront. Endocrinol. \u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, 885650 (2022).\u003c/li\u003e\n\u003cli\u003eHotamisligil, G. S. Inflammation, metaflammation and immunometabolic disorders. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e542\u003c/strong\u003e, 177\u0026ndash;185 (2017).\u003c/li\u003e\n\u003cli\u003eEfthymiou, A. \u003cem\u003eet al.\u003c/em\u003e Isolation and freezing of human peripheral blood mononuclear cells from pregnant patients. \u003cem\u003eSTAR Protoc\u003c/em\u003e \u003cstrong\u003e3\u003c/strong\u003e, 101204 (2022).\u003c/li\u003e\n\u003cli\u003eStoeckius, M. \u003cem\u003eet al.\u003c/em\u003e Simultaneous epitope and transcriptome measurement in single cells. \u003cem\u003eNat. Methods\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, 865\u0026ndash;868 (2017).\u003c/li\u003e\n\u003cli\u003eStroukov, W. \u003cem\u003eet al.\u003c/em\u003e OMIP-090: A 20-parameter flow cytometry panel for rapid analysis of cell diversity and homing capacity in human conventional and regulatory T cells. \u003cem\u003eCytometry A\u003c/em\u003e \u003cstrong\u003e103\u003c/strong\u003e, 362\u0026ndash;367 (2023).\u003c/li\u003e\n\u003cli\u003eFinak, G. \u003cem\u003eet al.\u003c/em\u003e MAST: a flexible statistical framework for assessing transcriptional changes and characterizing heterogeneity in single-cell RNA sequencing data. \u003cem\u003eGenome Biol.\u003c/em\u003e \u003cstrong\u003e16\u003c/strong\u003e, 278 (2015).\u003c/li\u003e\n\u003cli\u003ePapaioannou, T. G. \u003cem\u003eet al.\u003c/em\u003e Mean arterial pressure values calculated using seven different methods and their associations with target organ deterioration in a single-center study of 1878 individuals. \u003cem\u003eHypertens. Res.\u003c/em\u003e \u003cstrong\u003e39\u003c/strong\u003e, 640\u0026ndash;647 (2016).\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"nature-portfolio","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"Nature Portfolio","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"ejp","reportingPortfolio":"","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-3773991/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-3773991/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eGestational diabetes mellitus (GDM) is a common pregnancy complication associated with hyperglycaemia, chronic inflammation and adverse health outcomes. Regulatory T cells (Tregs) are thought to contribute to GDM due to their role in suppressing inflammation. However, it remains unclear whether specific Treg subsets are impaired in patients with GDM. To investigate transcriptional variation in GDM Tregs, we applied single-cell RNA sequencing to Tregs isolated from the blood of 13 healthy pregnant women and 10 patients with GDM. We identified naive and effector Treg subsets, none of which significantly differ in the proportion of cells captured from GDM and controls. We report a naive Treg subset with reduced expression of AP-1 transcription factor subunits in GDM, including \u003cem\u003eJUN, FOS\u003c/em\u003e, and \u003cem\u003eEGR1\u003c/em\u003e, and an effector Treg subset with increased signalling of angiogenesis marker genes. Genes dysregulated in GDM Tregs independently predicted GDM status in pseudobulk and whole blood mRNA from independent cohorts. Remarkably, \u003cem\u003eTXNIP\u003c/em\u003e, which regulates glucose levels, emerged as the most reliable standalone predictor in bulk mRNA (minimum AUC 0.7) equivalent to using body mass index (AUC 0.72) in our cohort. This study uncovers a disrupted molecular pathway in Treg cell subsets from GDM patients and proposes a panel of genes with translational potential as early disease biomarkers.\u003c/p\u003e","manuscriptTitle":"Single-cell transcriptomics reveals markers of regulatory T cell dysfunction in Gestational Diabetes Mellitus","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-01-10 19:23:40","doi":"10.21203/rs.3.rs-3773991/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"communications-medicine","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"commsmed","sideBox":"Learn more about [Communications Medicine](http://www.nature.com/commsmed)","snPcode":"43856","submissionUrl":"https://mts-commsmed.nature.com/cgi-bin/main.plex","title":"Communications Medicine","twitterHandle":"@commsmedicine","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Communications Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"29ae2d99-12f1-4021-85e3-52df1313d178","owner":[],"postedDate":"January 10th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":27909785,"name":"Biological sciences/Computational biology and bioinformatics/Gene regulatory networks"},{"id":27909786,"name":"Biological sciences/Immunology/Immunogenetics"},{"id":27909787,"name":"Health sciences/Diseases/Endocrine system and metabolic diseases/Diabetes/Gestational diabetes"}],"tags":[],"updatedAt":"2026-06-01T07:06:19+00:00","versionOfRecord":{"articleIdentity":"rs-3773991","link":"https://doi.org/10.1038/s43856-026-01563-0","journal":{"identity":"communications-medicine","isVorOnly":false,"title":"Communications Medicine"},"publishedOn":"2026-04-03 04:00:00","publishedOnDateReadable":"April 3rd, 2026"},"versionCreatedAt":"2024-01-10 19:23:40","video":"","vorDoi":"10.1038/s43856-026-01563-0","vorDoiUrl":"https://doi.org/10.1038/s43856-026-01563-0","workflowStages":[]},"version":"v1","identity":"rs-3773991","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-3773991","identity":"rs-3773991","version":["v1"]},"buildId":"7rjqhiLT3MXkJMwkYKINL","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}