MX1+ effector T cells hyperactivation at the maternal-fetal interface in unexplained recurrent pregnancy loss.

This study was conducted in accordance with the Declaration of Helsinki. All participants provided written informed consent under protocols approved by the Medical Ethics Committee of the Third Affiliated Hospital of Guangzhou Medical University (No. 2,025,022). In this study, URPL was defined as the occurrence of two or more consecutive pregnancy losses before 20 weeks of gestation with no identifiable etiology, consistent with the criteria established by the European Society of Human Reproduction and Embryology [ 9 ]. The controls group included women aged 20 to 40 years old with one prior live birth, no prior unknown reason miscarriages, undergoing elective termination of a uncomplicated singleton pregnancy at 5 to 9 weeks. The URPL group included patients aged 20 to 40 years old with two or more previous unexplained pregnancy losses at 5 to 9 weeks. All the participants with known cause of miscarriage were excluded, including parental and embryonic chromosomal abnormalities, uterine malformation, endocrine disorders (such as thyroid dysfunction, diabetes or polycystic ovary syndrome), autoimmune diseases, infection or chronic endometritis as confirmed by histopathological examination within three months, tumor and endometriosis and smoking or alcohol consumption.[ 9 ] Decidual tissues were collected from two cohorts: (1) healthy controls, 5 for scRNA-seq, 12 for immunohistochemistry (IHC), multiplex immunohistochemistry (mIHC) and 11 for FCM; and (2) URPL patients, 6 for scRNA-seq, 12 for IHC and mIHC and 11 for FCM. Peripheral blood sample were collected from healthy control ( n  = 6) and URPL patients ( n  = 6). The demographic characteristics of participants used for scRNA-seq, IHC, mIHC and FCM were displayed in Supplementary Table S1 and S3 - 5 . The specificity of the MX1 antibody used in IHC was validated by IgG negative control staining (Supplementary Figure 3 ). The decidual samples were collected by ultrasound-guided curettage immediately after the diagnosis of missed abortion. The human peripheral blood and decidua tissues were freshly collected at the first trimester. Fresh human peripheral blood was collected in K 2 EDTA tubes. An equal volume of Phosphate-Buffered Saline (PBS) was mixed with blood and carefully layered over Ficoll Hypaque solution (MP Biomedicals). After centrifugation for 30 min at 400 rcf at room temperature, the mononuclear cell layer was washed two times with PBS for 10 min at 300rcf. Cell sedimentation was treated with erythrocyte lysis buffer (Invitrogen, 00–4300) for 10 minutes, followed by termination with PBS. Cells were again washed twice with PBS and finally resuspended in PBS for subsequent FCM. Decidual tissues were macroscopically identified, rinsed in phosphate-buffered saline to remove blood clots, and dissociated into small fragments using collagenase IV (0.5 mg/mL, Sigma-Aldrich C5138) and DNase I (0.1 mg/mL, Sigma-Aldrich DN25) for 30 minutes at 37 °C [ 16 ]. Cell suspensions were filtered through 40 µm strainers, treated with red blood cell lysis buffer (Invitrogen 00–4300) for 10 minutes, and resuspended in PBS. Cell viability was assessed, and only samples with viability exceeding 90% were subjected to subsequent scRNA-seq and FCM. Single-cell 3′ cDNA libraries were prepared using the 10× Genomics Chromium platform. Libraries were sequenced on an Illumina HiSeq X Ten (150 bp paired-end). Raw FASTQ files were processed via Cell Ranger v4.0.0, aligned to GRCh38 using STAR. This matrix was then imported into the Seurat (v5.1.0) package in R (v4.4.1) to apply for quality control and downstream analysis. All functions were run with default parameters unless specified otherwise. Low-quality cells with fewer than 500 genes/cell and more than 20% total mitochondrial genes were exclude gene expression. Finally, a filtered gene-barcode matrix of all samples was integrated with to remove batch effects across different samples using the SCT (SCTransform) method. Data normalization and variance stabilization were performed on the integrated object using the NormalizeData and ScaleData functions in Seurat where a regularized negative binomial regression corrected for differential effects of mitochondrial gene expression levels. PrepSCTIntegration and FindIntegrationAnchors were applied to identify the anchoring vectors across datasets and integration steps performed with IntegrateData (normalization method set to SCT). Principal component analysis (PCA) was conducted on the top 2000 variable genes to obtain the first 30 principal components. Uniform Manifold Approximation and Projection (UMAP) was used to visualize clusters with resolution 0.3 to identify distinct groups of cells. Cell types were assigned by the expression of well-characterized marker genes and the top differentially expressed genes in each cluster. The FindAllMarkers function was used to identify the signature genes of each cell cluster. The Wilcoxon Rank Sum test was used, and the p -values were corrected using the false discovery rate (FDR) method. Genes were considered significantly differentially expressed if they met the criteria of an adjusted p -value  1. We characterized the identities of cell types based on published annotation. [ 17 , 18 ] Major cell types were first determined and T cells were further sub-clustered. In addition, the percentages of different cell types or cell subtypes were calculated accordingly. To assess the functional differences between MX1 + T cells and other T cell subsets, as well as between URPL patients and healthy controls, we performed GO analysis using the clusterProfiler package (v4.12.0). This analysis was conducted to identify enriched biological processes (BP), cellular components (CC), and molecular functions (MF) associated with differentially expressed genes (DEGs) between the groups. First, we identified DEGs using the Seurat package, with a threshold of adjusted p -value  1. These DEGs were then mapped to GO terms using the org.Hs.eg.db (v3.19.1) annotation database. Enrichment analysis was performed using the Fisher’s exact test, and the significance of enrichment was determined by adjusting p -values. Only GO terms with an adjusted p -value < 0.05 were considered statistically significant. The results were visualized using dot plots and bar charts to highlight the top enriched GO terms for each comparison. ScRNAtoolVis (v0.1.0) was used to analyze the differential gene expression of stromal cell subpopulations between normal and abnormal patients. The gene list for inhibitory genes, co-accessory genes and MHC genes was obtained from the Molecular Signatures Database (MSigDB), which included BTLA, VTCN1, ENTPD1, PDCD1, CTLA4, BATF, LAG3, TIGIT, HAVCR2, NCR3LG1, LGALS9, SIPRA, FGL1, 2B4, TGFB1, VSIR, PVR, FASLG, TYROBP, ICOSLG, OX40, CD28, CD40, CD47, CD70, CD80, CD86, CD96, CD99, CD137, CD160, CD200, CD274, CD276, SDC4, ITGB1, SIRPG, CSF1R, VISTA, GITR, LAIR1, TDO2, IDO1, INFSF4, TNFSF8, TNFSF9, TNFRSF14, TNFRSF18, HLA-A, HLA-B, HLA-C, HLA-DM, HLA-E, HLA-F and HLA-DR.[ 19 ] To analyze the expression patterns of specific gene sets in our dataset, we utilized the AddModuleScore in the R package Seurat to calculate module scores for T cells according to predefined gene lists. [ 20 , 21 ] These scores were computed to evaluate the relative expression levels of genes within each gene set. The different state of T cell gene sets were analyzed: activation, proliferation, cytotoxicity, IFN response, HLA Type II, inhibitory, senescence, and exhaustion. The activation status of cells was assessed using a gene set that included CD40LG, CD69, HLA-DPA1, HLA-DPB1, HLA-DQB1, HLA-DRB5, ICOS, TNFRSF4, ID3, TNFSF14, TNFAIP6, TNFAIP8, TNFSF13B, MX1, MX2, NFAT5, NFATC3, NFATC2, LCP2, LAT, ZAP70 and LCK. Proliferation was evaluated based on the expression of genes such as MKI67, STMN1, CDKI, TOP2A and PCNA. Cytotoxic potential was determined using a gene set that included GZMB, GZMH, GZMK, GZMA, TIA1, PRF1, LAMP1, GNLY, FASLG, CD69, IFNG, NFKB1 and TNF. For the assessment of interferon-related pathways, we computed a module score using the IFN Score gene set, which included ISG15, IFI44, IFI27, CXCL10, URPLD2, IFIT1, IFI44L, CCL8, XAF1, GBP1, IRF7 and CEACAM1. To evaluate the expression of HLA II molecules, we generated a module score based on the HLA II gene set, comprising HLA-DRA, HLA-DQA1, HLA-DPA1, HLA-DRB1, HLA-DPB1, HLA-DRB5, HLA-DQB1, HLA-DMA and HLA-DMB. For the analysis of inhibitory pathways, we utilized a gene set consisting of BATF3, BATF, BATF2, CSNK2B, CSNK2A3, CSNK2A2, CSNK2A1, CTLA4, PDCD1, LAG3, HAVCR2, BTLA and TIGIT. Senescence-related processes were assessed using genes such as CD247, KLRG1, B3GAT1, TP53, MAPK14, MAPK8, MAPK1 and CDKN2A. Finally, exhaustion-related pathways were evaluated using a gene set that included PDCD1, HAVCR2, LAG3, TOX, TIGIT, CTLA4, BTLA and DDX4. The module scores derived from these gene sets were used to assess the relative activity of each biological process or functional pathway within the dataset. To assess the metabolic differences in T cell subsets, we employed the ScMetabolism from the databases of KEGG to associate metabolic activity scores with specific T cell subpopulations. [ 22 ] By leveraging methods of AUCell, VISION and ssGSEA, the relative expression of metabolic pathways and their correlation with T cell functions were evaluated. Trajectory analysis was performed for the subset of T cells using scTour (version 1.0.0) in python (version 3.12.4). First, we preprocessed the single-cell data, including normalization and scaling, to ensure the data quality. Then, a neural network model within the scTour framework was defined, which consisted of multiple layers to capture the complex patterns in the data. The model was trained with the aim of classifying cell types or developmental stages based on gene expression profiles. During the training process, a cross-validation approach was used to evaluate the model’s performance and avoid overfitting. Once the model was trained, it was used to predict the developmental trajectories of T cell subsets. This involved mapping the cells onto a low-dimensional space that represented the inferred developmental paths, which was visualized using trajectory plots to illustrate the progression of T cell differentiation and maturation. CellChat (version 2.1.2) package was used to further analyse and compare the intercellular communication differences between normal and URPL decidua immune microenvironment. After classifying the T cell subpopulations, we reintegrated them with other cell types and then inferred intercellular communications among T cell and other cell subsets. The ranknet function was applied with a significance threshold of 0.05. Only common cell types could be included in group comparisons. The information flow bar charts highlight significant signaling pathways between groups, which can be further examined through chord plots of interaction probability to identify the receiver and sender cells. The contribution of each L-R pair to a pathway is also displayed in circle plots of interaction probability. For histological analysis, decidual tissues were fixed with 10% buffered formalin, dehydrated in ethanol and embedded with paraffin and sectioned at a thickness of 5 um. Multiplex immunofluorescence was performed using the tyramine signal amplification (TSA) Fluorescence System Kits. Primary antibodies against CD3 (ab16669, 1:200; Abcam), PD-L1 (BP0101, 1:200; BioXCell), Ki67 (14–98-82, 1:100, Thermo Fisher Scientific), M×1 (13750–1-AP, 1:200; Proteintech), CLEC9A (222185, 1:200; Zenbio) and ICOSL (14922–1-AP, 1:500; Proteintech) were incubated sequentially, followed by horseradish peroxidase-coupled secondary antibody incubation and TSA (Supplementary Table S6 ). Tyramide-mediated signal amplification was performed to covalently link the Opal fluorophores. Then, the antibodies were stripped to initiate the next immunostaining cycle. The sequence of Panel 1 was MX1/Opal 690, CD3/Opal 620, Ki67/Opal 570, and PD-L1/Opal 520. The Panel 2 sequence was MX1/Opal 690, CD3/Opal 520, CLEC9A/Opal 620 and ICOSL/Opal 570. All slides were stained for nuclei with DAPI (D9542, 1:500, Sigma) after labeling with all antigens described above. Multispectral imaging was performed on the Panoramic tissue single-cell identification and image analysis system (Akoya PhenoImager Fusion, Akoya) and Laser Scanning Confocal Microscope (LSM 980 With Airyscan 2, ZEISS). Slides were imaged at × 20 and × 40 magnification and the fluorophore signals were captured by a multispectral camera module. Autofluorescence, obtained from an unstained slide, was removed from the composite images and pseudo-colored images were exported as TIF files. Cells were incubated with surface antibodies in FACS buffer (PBS containing 2.5% fetal bovine serum) for 20 min at 4 °C in the dark. All antibodies were used at concentrations recommended by the manufacturer. Stained cells were acquired on an Attune NxT (Life Technologies). Data were analyzed with FlowJo software (version 10.8.1, BD Biosciences). Primary antibodies used for the intercalating antibodies were M×1 (13750–1-AP, Proteintech) and CLEC9A (222185, Zenbio), corresponding to the fluorescent secondary antibodies BV421 (565014, BD Biosciences) and APC (ab150115, Abcam), respectively. Direct antibodies used included CD3-FITC (300452, Biolegend), ICOSL-PE (309403, Biolegend), CD25-PE-Cy7 (557741, BD Biosciences), CD69–BV605 (562989, BD Biosciences), CD45-APC-Cy7 (557833, BD Biosciences)(Supplementary Table S7 ). Statistical analysis was performed with the R, SPSS (v27, IBM, Armonk,NY) and GraphPad Prism (9.0.0, SanDiego, CA) software. G*power (v3.1.9.7) was used to determine the appropriate sample size and prove power justification. Gaussian distribution of the data was tested using the Shapiro–Wilk test, and Levene’s test was utilized to assess the homogeneity of variances. Student’s t-test or the Wilcoxon rank-sum test were used as appropriate. Data were presented as mean±SD. p values < 0.05 was considered to be significant.

First-trimester decidual tissues (5–9 weeks gestation) were obtained from 5 healthy controls and 6 URPL patients (Fig. 1 A). The two cohorts were well-matched for key clinical parameters, including maternal age, gestational age and body mass index (BMI) (Supplementary Table S1 ). After rigorous quality filtering, a total of 66,078 high-quality single-cell transcriptomes were obtained for downstream analysis (Supplementary Figure S1 A, Supplementary Table S2 ). Unsupervised clustering identified 21 distinct cellular populations of early decidual immune microenvironment (Fig. 1 B), including a T cell cluster (identified by CD3G), five decidual natural killer (dNK) subsets (identified by NCAM1), three macrophage populations (identified by CD14), one B cell cluster (identified by CD79A), one dendritic cell cluster (identified by CLEC9A), three decidual stromal cell subsets (dS1, dS2, dS3, identified by HAND2 and CITED2), three endothelium cell subsets (Endo1, Endo2, Endo3, identified by PECAM1), one perivascular subsets (PV, identified by EPCAM), one epithelium cluster (Epi, identified by EPCAM) and trophoblastic (EVT, identified by HLA-G) (Fig. 1 C). Fig. 1 Increased T cell infiltration at the maternal-fetal interface in women with URPL. ( A ) schematic overview of sample collection and analytical workflow. ( B ) UMAP visualization of immune cell clusters in combined, normal and URPL groups. ( C ) dot plot displaying expression levels of canonical cell type-specific markers across clusters. ( D ) proportional distribution of cell populations between normal and URPL groups. ( E ) differentially expressed genes in stromal cell subsets (dS1, dS2, dS3) between normal and URPL groups. ( F - G ) immunohistochemical staining and quantification of CD3+ and PD-L1 in decidual tissues ( n  = 12 for controls, n  = 12 for URPL patients). ( H ) Representative flow cytometry plots showing the expression and flow cytometry analysis of the percentage of positive cells for CD3 in decidual and peripheral blood immune cells (normal decidual, n  = 11; URPL decidual, n  = 11, normal peripheral blood, n  = 6; URPL peripheral blood, n  = 6). Color coding: blue represents normal decidual; red represents URPL decidual; green represents normal peripheral blood; orange represents URPL peripheral blood. Data represent mean ± SD; *** p  < 0.001, * p   0.05 (unpaired t-test or Wilcoxon rank-sum test); scale bar, 50 µm in TSA and 20 µm in IHC Increased T cell infiltration at the maternal-fetal interface in women with URPL. ( A ) schematic overview of sample collection and analytical workflow. ( B ) UMAP visualization of immune cell clusters in combined, normal and URPL groups. ( C ) dot plot displaying expression levels of canonical cell type-specific markers across clusters. ( D ) proportional distribution of cell populations between normal and URPL groups. ( E ) differentially expressed genes in stromal cell subsets (dS1, dS2, dS3) between normal and URPL groups. ( F - G ) immunohistochemical staining and quantification of CD3+ and PD-L1 in decidual tissues ( n  = 12 for controls, n  = 12 for URPL patients). ( H ) Representative flow cytometry plots showing the expression and flow cytometry analysis of the percentage of positive cells for CD3 in decidual and peripheral blood immune cells (normal decidual, n  = 11; URPL decidual, n  = 11, normal peripheral blood, n  = 6; URPL peripheral blood, n  = 6). Color coding: blue represents normal decidual; red represents URPL decidual; green represents normal peripheral blood; orange represents URPL peripheral blood. Data represent mean ± SD; *** p  < 0.001, * p   0.05 (unpaired t-test or Wilcoxon rank-sum test); scale bar, 50 µm in TSA and 20 µm in IHC Proportion analysis revealed significant compositional shifts of decidual immune microenvironment in URPL (Fig. 1 D). Notably, stromal remodeling was evidenced by decidual stromal cell redistribution, with dS1 (16.81% vs. 10.22%) and dS2 (16.45% vs. 10.02%) decreased, dS3 increased from 0.03% to 1.31 between the two groups. In URPL patients, total T cell infiltration was elevated from 4.25% to 8.43%, indicating heightened proliferation activity. Specifically, NK cells decreased, with dNK1 dropping from 8.73% to 6.83% and dNK2 from 6.07% to 5.08% suggesting functional subset specialization. Macrophage polarization shifted toward dMc1 (8.87% vs. 12.19%) and away from dMc2 (4.08% vs. 2.93%) while no significant changes were detected in DCs (0.65% vs. 0.50%). Given the markedly reduction of stromal cells among URPL patients, differential gene expression analysis was performed, which indicated downregulated expression of immune checkpoint related genes in patient-derived stromal cells (Fig. 1 E). IHC and mIHC were performed on decidual tissues from 12 individuals per group. No significant differences in key clinical characteristics, such as maternal age, gestational age or BMI, were observed between the URPL and control groups (Supplementary Table S3 ). Which confirmed significantly higher infiltration of CD3+T cell and lower immune checkpoint PD-L1 in URPL decidual tissues (Fig. 1 F), with CD3+T cells increased from 8.82 ± 1.81% to 21.86 ± 6.99% ( n  = 12 , p  < 0.001) and PD-L1 decreased from 5.28 ± 1.88% to 2.82 ± 0.45% ( n  = 6, p  < 0.05) (Fig. 1 G). Flow cytometry analysis was performed on peripheral blood mononuclear cells (PBMC) and decidual tissues (PBMC, n  = 6 per group; decidua, n  = 11 per group) from healthy women and URPL patients. The clinical characteristics of the participants, including maternal age, gestational age and BMI, were comparable between the study groups (Supplementary Table S4 and S5 ). The gating strategy was detailed in Supplementary Figure S2 . The percentage of CD3+T cells in peripheral blood did not differ significantly between URPL patients (83.62 ± 3.11%) and healthy controls (78.10 ± 3.04%) ( n  = 6, p  > 0.05). However, a significant increase in CD3+ T cells expression was observed in decidual tissues from URPL patients (37.35 ± 2.12%) compared to healthy controls (30.54 ± 1.86%, p  < 0.05; Fig. 1 H). The data indicated the significant augment of decidual T cell in URPL, while the decline of PD-L1 in decidual represent the suppression of immune tolerance, suggesting their potential role in URPL pathogenesis. To identify different function of T cells between healthy women and women with URPL, we used clinical signature score to compare the feature of total T cells. Comparative gene set scoring revealed significant up-regulation of proliferative, cytotoxic and activation pathways in URPL-derived T cells compared to healthy controls, while senescence-associated pathways remained unchanged (Fig. 2 A). Fig. 2 Hyperactivated T cell signatures at the maternal-fetal interface in URPL. ( A ) module scores for T cell functional pathways (activation, proliferation, cytotoxicity, senescence) in normal vs. URPL groups. ( B ) gene Ontology enrichment analysis of T cells in URPL compared to controls. ( C ) metabolic pathway enrichment profiles of T cells. ( D ) differential gene scores for xenobiotic metabolism, interferon response, apoptosis and hypoxia-related pathways. ( E - F ) Representative flow cytometry plots showing the expression and flow cytometry analysis of the percentage of positive cells for CD25, CD69 in decidual and peripheral blood T cells (normal decidual, n  = 11; URPL decidual, n  = 11, normal peripheral blood, n  = 6; URPL peripheral blood, n  = 6). ( G - H ) Immunohistochemical (IHC), multiplex immunofluorescence (TSA) and quantification of Ki67+CD3+ proliferating T cells ( n  = 12 for controls, n  = 12 for URPL patients). Color coding: blue represents normal decidual; red represents URPL decidual; green represents normal peripheral blood; orange represents URPL peripheral blood. Data represent mean ± SD; *** p  < 0.001,** p   0.05 (unpaired t-test or Wilcoxon rank-sum test); scale bar, 50 µm in TSA and 20 µm in IHC; TCA cycle, tricarboxylic acid cycle Hyperactivated T cell signatures at the maternal-fetal interface in URPL. ( A ) module scores for T cell functional pathways (activation, proliferation, cytotoxicity, senescence) in normal vs. URPL groups. ( B ) gene Ontology enrichment analysis of T cells in URPL compared to controls. ( C ) metabolic pathway enrichment profiles of T cells. ( D ) differential gene scores for xenobiotic metabolism, interferon response, apoptosis and hypoxia-related pathways. ( E - F ) Representative flow cytometry plots showing the expression and flow cytometry analysis of the percentage of positive cells for CD25, CD69 in decidual and peripheral blood T cells (normal decidual, n  = 11; URPL decidual, n  = 11, normal peripheral blood, n  = 6; URPL peripheral blood, n  = 6). ( G - H ) Immunohistochemical (IHC), multiplex immunofluorescence (TSA) and quantification of Ki67+CD3+ proliferating T cells ( n  = 12 for controls, n  = 12 for URPL patients). Color coding: blue represents normal decidual; red represents URPL decidual; green represents normal peripheral blood; orange represents URPL peripheral blood. Data represent mean ± SD; *** p  < 0.001,** p   0.05 (unpaired t-test or Wilcoxon rank-sum test); scale bar, 50 µm in TSA and 20 µm in IHC; TCA cycle, tricarboxylic acid cycle Gene Ontology (GO) enrichment analysis demonstrated dominant enrichment of immune response activation signaling pathway in URPL decidual, including immune response-regulating cell surface receptor signaling pathway, T cell differentiation and antigen receptor-mediated signaling pathway (Fig. 2 B). Metabolic profiling via scMetabolism uncovered distinct immunometabolic reprogramming: URPL T cells exhibited enhanced metabolic pathways associated with carbohydrate remodeling and redox regulation, including starch and sucrose metabolism, pentose phosphate pathway, pentose and glucuronate interconversions and nucleotide sugar metabolism. In contrast, healthy controls exhibited marked enrichment in pathways linked to classical energy production pathway such as citrate acid (TCA) cycle (Fig. 2 C). The irGSEA-based differential gene set scoring (AUCell) identified coordinated upregulated pathways in URPL patients, including interferon-alpha-response, xenobiotic-metabolism, apoptosis and hypoxia (Fig. 2 D). Considering the CD25 and CD69 as T cell activation markers, we next characterized peripheral blood and decidual tissues by FCM. Peripheral blood T cells exhibited no significant difference, with CD25 expression at 5.52 ± 1.80% in controls and 10.30 ± 2.62% in URPL patients, and CD69 was 0.98 ± 0.07% in controls and 1.35 ± 0.29 in URPL patients ( p  > 0.05). However, decidual T cells from URPL patients exhibited significantly increased activation, with CD25 expression at 6.63 ± 1.14% in controls and 23.00 ± 1.38% in URPL patients. For CD69, the expression was 14.37 ± 3.88% in healthy controls and 44.46 ± 6.66% in URPL patients ( p  < 0.001, Fig. 2 E). IHC and mIHC confirmed significantly higher protein expression of proliferation marker Ki67 from 2.83 ± 3.57% to 22.5 ± 9.93% in URPL decidual tissues (p  < 0.01) (Fig. 2 F). Taken together, our analyses demonstrate the activation of tissue-resident T cells among women with URPL. Unsupervised UMAP clustering delineated 13 transcriptionally distinct decidual T cell subsets, including effector memory T cells (MX1+Tem, three clusters of CD8+Tem), tissue-resident memory (CD8+Tm), progenitor exhausted (CD8+Tpex, CD4+Tpex), terminally exhausted (Tex), naive T, helper CD4+Th, γδT, MAIT, and IGFBP3+ T cells (Fig. 3 A), which comprehensive mapping highlighted the functional diversity of T cell populations at the maternal-fetal interface. Fig. 3 Enhanced interferon response in MX1+ Tem cells at the maternal-fetal interface among URPL patients. ( A ) UMAP clustering of T cell subsets. ( B ) dot plot showing expression of subset-specific markers. ( C ) pseudotemporal trajectory analysis (scTour) of T cell differentiation. ( D ) Representative flow cytometry plots showing the expression and flow cytometry analysis of the percentage of positive cells for MX1 in decidual and peripheral blood T cells (normal decidual, n  = 11; URPL decidual, n  = 11, normal peripheral blood, n  = 6; URPL peripheral blood, n  = 6). ( E ) comparative gene scores (activation, proliferation, interferon response, HLA scoring, inhibitory, exhausion) in MX1+ T cells between groups. ( F ) GO enrichment for MX1+ T cells. ( G ) cross-subset similarity analysis of effector programs (Spearman’s rank correlation coefficient). ( H ) Heatmap comparing cytotoxic, interferon-responsive, inhibitory and proliferation gene expression across CD8+ T cell subsets. Color coding: blue represents normal decidual; red represents URPL decidual; green represents normal peripheral blood; orange represents URPL peripheral blood. Data represent mean ± SD; *** p  < 0.001,* p  < 0.05 (unpaired t-test) Enhanced interferon response in MX1+ Tem cells at the maternal-fetal interface among URPL patients. ( A ) UMAP clustering of T cell subsets. ( B ) dot plot showing expression of subset-specific markers. ( C ) pseudotemporal trajectory analysis (scTour) of T cell differentiation. ( D ) Representative flow cytometry plots showing the expression and flow cytometry analysis of the percentage of positive cells for MX1 in decidual and peripheral blood T cells (normal decidual, n  = 11; URPL decidual, n  = 11, normal peripheral blood, n  = 6; URPL peripheral blood, n  = 6). ( E ) comparative gene scores (activation, proliferation, interferon response, HLA scoring, inhibitory, exhausion) in MX1+ T cells between groups. ( F ) GO enrichment for MX1+ T cells. ( G ) cross-subset similarity analysis of effector programs (Spearman’s rank correlation coefficient). ( H ) Heatmap comparing cytotoxic, interferon-responsive, inhibitory and proliferation gene expression across CD8+ T cell subsets. Color coding: blue represents normal decidual; red represents URPL decidual; green represents normal peripheral blood; orange represents URPL peripheral blood. Data represent mean ± SD; *** p  < 0.001,* p  < 0.05 (unpaired t-test) Effector T cells demonstrated robust expression of cytotoxic mediators (GZMA, GZMH, GZMK), with hierarchical functional specialization as following. CD8+Tem1 exhibited a broad cytotoxic profile and CD8+Tem2 displayed chemokine signaling specialization (CCL4), suggesting roles in immune cell recruitment. CD8+Tem3 marked terminal differentiation (PRF1/perforin, GZMB), indicative of potent target cell elimination. Tissue-resident CD8+Tm cells expressed niche-adaptation markers (SH3BGRL3, HOPX, CD52), consistent with endometrial retention. Strikingly, the MX1+Tem subset uniquely co-expressed interferon-stimulated genes (MX1, MX2, IFITM1, ISG15, ISG20, IFIT2) and the co-stimulatory receptor ICOS, suggesting dual roles in IFN-response and adaptive immune regulation (Fig. 3 B). Progenitor exhausted populations exhibited divergent transcriptional regulation. CD8+Tpex expressed TOB1, a brake on effector differentiation. CD4+Tpex upregulated stress-responsive factors (JUNB, NR4A1). Terminally exhausted Tex cells were defined by LAG3, a master regulator of exhaustion. Naive T cell populations exhibited distinctive transcriptional signatures of CCR7, SELL and IL7R, reflecting their antigen-inexperienced state. CD4+Th cells showed T helper cell identity gene CD40LG, while γδT and MAIT cells were distinguished by TRDC and EGR1/SLC16A3 respectively. The IGFBP3+ T subset relatively high expression expressed IGFBP3 (Fig. 3 B). Pseudotemporal trajectory analysis using scTour positioned naive T cells at the origin, effector subsets (including MX1+ T cells) at intermediate states, and exhausted populations at terminal differentiation, revealing a progressive loss of functionality (Fig. 3 C). We next explored whether these developmental trajectories differed between normal pregnancies and URPL cases. Pseudotemporal analysis revealed that T cell differentiation in healthy individuals followed a relatively straightforward progression, whereas URPL specimens exhibited a more complex trajectory with divergent branching (Supplementary Figure S1 B). FCM revealed no significant difference in MX1+Tem cells in peripheral blood between URPL patients (2.18 ± 0.28%) and controls (1.95 ± 0.50%, n  = 6, p  > 0.05). However, decidual MX1+Tem cells were significantly increased in URPL patients (36.05 ± 4.79%) compared to controls (11.29 ± 2.09%, p  < 0.05, Fig. 3 D). In total, these analyses demonstrated augmented MX1+Tem cells in URPL decidua. Subsequently, comparative gene scores of MX1+ T cells across cohorts revealed upregulation of proliferation and interferon-response signatures (Fig. 3 E). These cells also showed enhanced human leukocyte antigen (HLA) associated antigen presentation capacity (HLA-scores), concurrent with elevated inhibitory receptor expression, suggesting a hyperresponsive state. GO analysis of MX1+Tem cell further confirmed preferential enrichment for defense responses to symbionts and type II interferon signaling compared to other T cell subsets (Fig. 3 F), aligning with their activation transcriptional identity. Cross-subset similarity analysis revealed shared effector programs between MX1+Tem and CD8+Tem clusters (Fig. 3 G). To dissect functional divergence across T cell subsets, we analyzed effector molecules, interferon-responsive factors, immune checkpoint markers and proliferative drivers. Strikingly, MX1 + T cells uniquely amplified interferon-stimulated genes (ISGs, including MX1, MX2, IFITM1 and ISG15) while retaining moderate expression of cytotoxic (GZMH) and proliferative (PCNA) markers (Fig. 3 H). This dual transcriptional signature positions MX1+Tem cells as specialized sentinels that integrate innate interferon-driven antigen presentation with adaptive cytotoxic functions at the maternal-fetal interface. Overall, these analyses demonstrated the activation characteristics of MX1+Tem cells among women with URPL. The single-cell atlas reveals pathological redistribution of T cell subsets in URPL decidua. Unsupervised clustering identified profound compositional shifts in decidual T cells between URPL and normal decidual (Fig. 4 A). Notably, MX1+Tem cells demonstrated more than 2 fold expansion of in URPL (1.54% vs. 3.17%), accompanied by a dramatic increase in CD8+Tpex (0.15% vs. 7.06%) and exclusive detection of terminally exhausted Tex cells (4.27% in URPL, absent in controls) (Fig. 4 B). Fig. 4 Expansion of hyperactivated MX1+ Tem cells at the maternal-fetal interface in URPL. ( A ) UMAP visualization of T cell subsets between normal and URPL groups. ( B ) proportional changes in T cell subpopulations. ( C ) mapping showing MX1 and ICOS expression in T cells. ( D ) metabolic pathway analysis of oxidative phosphorylation in T cell subsets. ( E ) metabolic pathway analysis of carbohydrate metabolism. ( F - G ) Immunohistochemical (IHC), multiplex immunofluorescence (TSA) and quantification of MX1+Ki67+CD3+T cells ( n  = 12 for controls, n  = 12 for URPL patients). ( H ) Representative flow cytometry plots showing the expression and flow cytometry analysis of the percentage of positive cells for CD25, CD69 in decidual and peripheral blood MX1+Tem cells. (normal decidual, n  = 11; URPL decidual, n  = 11, normal peripheral blood, n  = 6; URPL peripheral blood, n  = 6). Color coding: blue represents normal decidual; red represents URPL decidual; green represents normal peripheral blood; orange represents URPL peripheral blood. Data represent mean ± SD; *** p  < 0.001,** p  < 0.01, * p  < 0.05 (unpaired t-test or Wilcoxon rank-sum test); scale bar, 20 µm in TSA and IHC; scale bar, 50 µm in TSA and 20 µm in IHC Expansion of hyperactivated MX1+ Tem cells at the maternal-fetal interface in URPL. ( A ) UMAP visualization of T cell subsets between normal and URPL groups. ( B ) proportional changes in T cell subpopulations. ( C ) mapping showing MX1 and ICOS expression in T cells. ( D ) metabolic pathway analysis of oxidative phosphorylation in T cell subsets. ( E ) metabolic pathway analysis of carbohydrate metabolism. ( F - G ) Immunohistochemical (IHC), multiplex immunofluorescence (TSA) and quantification of MX1+Ki67+CD3+T cells ( n  = 12 for controls, n  = 12 for URPL patients). ( H ) Representative flow cytometry plots showing the expression and flow cytometry analysis of the percentage of positive cells for CD25, CD69 in decidual and peripheral blood MX1+Tem cells. (normal decidual, n  = 11; URPL decidual, n  = 11, normal peripheral blood, n  = 6; URPL peripheral blood, n  = 6). Color coding: blue represents normal decidual; red represents URPL decidual; green represents normal peripheral blood; orange represents URPL peripheral blood. Data represent mean ± SD; *** p  < 0.001,** p  < 0.01, * p  < 0.05 (unpaired t-test or Wilcoxon rank-sum test); scale bar, 20 µm in TSA and IHC; scale bar, 50 µm in TSA and 20 µm in IHC Dual up-regulation of the interferon-responsive gene M×1 and co-stimulatory factor ICOS in URPL-derived MX1+Tem cells further indicates synergistic interferon-α signaling and activated adaptive immune engagement (Fig. 4 C). This hybrid phenotype likely drives pathological crosstalk at the maternal-fetal interface. Metabolic profiling via scMetabolism revealed distinct immunometabolic reprogramming: TEM cells exhibited high oxidative phosphorylation, whereas Tex cells were relatively quiescent (Fig. 4 D). In normal controls, MX1+ T cells predominantly utilized TCA cycle, whereas URPL cases exhibited marked upregulation of lipid metabolism pathways (ether lipid pathway and fatty acid elongation) and glutathione metabolism, indicating a potential compensatory mechanism in response to oxidative stress and cellular injury (Fig. 4 E). Notably, immunofluorescence and TSA quantification confirmed a significant increase in MX1+ CD3+T cells (15.08 ± 8.17% to 28.61 ± 11.68%, P ＜0.05) and proliferating MX1+Ki67+CD3+T cells (0.46 ± 1.14% to 3.63 ± 1.90%, P ＜0.01) within URPL decidual tissues ( P ＜0.01) (Fig. 4 F–G). To characterize MX1+Tem cells activation, we analyzed CD25 and CD69 expression by FCM. Specifically, no significant difference was observed peripheral blood. CD25 expression was 24.43 ± 5.70% in URPL patients and 17.42 ± 1.83% in controls, while CD69 expression was 9.48 ± 0.97% in patients and 12.38 ± 1.98% in controls ( p  > 0.05). In contrast, decidual T cells from URPL patients exhibited significantly elevated CD25 as 23.14 ± 2.07% and 14.95 ± 2.02% in controls ( p  < 0.05), with CD69 expressed as 48.34 ± 6.29% and 18.23 ± 3.92% in controls ( p  < 0.001, Fig. 4 H). Collectively, these analyses revealed the activation of decidual MX1+Tem cells among women with URPL. Systematic Cellchat analysis identified 117 unique intercellular communication axes versus 116 pairs in healthy controls, indicating rewiring of the maternal-fetal immune microenvironment. Focusing on myeloid-T cell interactions, two key pathological signaling hubs emerged. Firstly, ICOS signaling was exclusively detected in URPL cases, with DCs-MX1+Tem interactions showing the highest signaling strength (Fig. 5 A). What’s more, enhanced MHC-II signaling between DCs and MX1+T cells suggested dysregulated antigen presentation and failed tolerance induction (Fig. 5 B, Supplementary Figure S1 C). Fig. 5 Enhanced dendritic cell (DC)-mediated activation signals drive MX1+ T cell hyperactivation in URPL. ( A ) CellChat analysis of ICOS signaling between DCs and T cells in URPL women (line thickness indicates the strength of interaction). ( B ) MHC-II signaling network in normal (top) vs. URPL (bottom) groups. ( C ) flow cytometry analysis of the percentage of positive cells for ICOSL+CLEC9A in decidual and peripheral blood DC cells (normal decidual, n  = 11; URPL decidual, n  = 11, normal peripheral blood, n  = 6; URPL peripheral blood, n  = 6). ( D - F ) Immunohistochemical (IHC), multiplex immunofluorescence (TSA) and quantification of CLEC9A, MX1, CD3 and ICOSL in decidual tissues ( n  = 12 for controls, n  = 12 for URPL patients). Color coding: blue represents normal decidual; red represents URPL decidual; green represents normal peripheral blood; orange represents URPL peripheral blood. Data represent mean ± SD; *** p  < 0.001,* p   0.05 (unpaired t-test); scale bar, 50 µm in TSA and 20 µm in IHC Enhanced dendritic cell (DC)-mediated activation signals drive MX1+ T cell hyperactivation in URPL. ( A ) CellChat analysis of ICOS signaling between DCs and T cells in URPL women (line thickness indicates the strength of interaction). ( B ) MHC-II signaling network in normal (top) vs. URPL (bottom) groups. ( C ) flow cytometry analysis of the percentage of positive cells for ICOSL+CLEC9A in decidual and peripheral blood DC cells (normal decidual, n  = 11; URPL decidual, n  = 11, normal peripheral blood, n  = 6; URPL peripheral blood, n  = 6). ( D - F ) Immunohistochemical (IHC), multiplex immunofluorescence (TSA) and quantification of CLEC9A, MX1, CD3 and ICOSL in decidual tissues ( n  = 12 for controls, n  = 12 for URPL patients). Color coding: blue represents normal decidual; red represents URPL decidual; green represents normal peripheral blood; orange represents URPL peripheral blood. Data represent mean ± SD; *** p  < 0.001,* p   0.05 (unpaired t-test); scale bar, 50 µm in TSA and 20 µm in IHC FCM revealed significantly increased ICOSL expression on decidual DC in URPL patients compared to healthy controls (14.09 ± 4.83% vs 0.68 ± 0.41%, p   0.05), indicating decidual-specific dysregulation. To investigate the spatial distribution of DCs and MX1+Tem cells, we employed mIHC ( n  = 12). Spatial profiling through mIHC showed distinct cellular interaction patterns. URPL decidua uniquely contained ICOSL+DCs-MX1+Tem clusters, with significantly elevated frequencies compared to controls (7.51 ± 1.14% vs 0.38 ± 0.27%, p  < 0.001, Fig. 5 D–F). Concurrently, we observed enhanced physical co-localization between ICOSL+DCs and T cells in miscarriage specimens (8.75 ± 1.24% vs 1.35 ± 0.59%, p  < 0.001). The co-localization indices for MX1+Tem cells and DCs (5.43 ± 1.39% vs 20.02 ± 3.74%, p= 0.001) and elevated frequencies of T and DCs complexes (14.78 ± 1.89% vs 28.34 ± 4.43%, p  < 0.05) also observed in URPL decidua (Fig. 5 D–F, Supplementary Figure S1 D). The infiltration of ICOSL+ DCs was also augmented in URPL decidua (28.64 ± 4.65% vs 12.25 ± 4.62%, p  < 0.05, Supplementary Figure S1 D). Consistently, the irGSEA scoring with AUCell analysis also revealed enhanced interferon-α/γ response signatures in URPL decidual (Supplementary Figure S1 E), aligning with the hyperactivated interferon-driven microenvironment of MX1+Tem cells among women with URPL. Collectively, these multimodal datasets revealed an enhanced ICOS-ICOSL interaction network between DCs and MX1+Tem cells within the pathological decidual niche in URPL (Fig. 6 ). Fig. 6 Aberrant dendritic cell-derived ICOSL signaling promotes proliferative expansion and interferon-driven hyperactivation of MX1+ Tem cells Aberrant dendritic cell-derived ICOSL signaling promotes proliferative expansion and interferon-driven hyperactivation of MX1+ Tem cells

In summary, our study indicated the infiltration of an activated interferon-responsive MX1+Tem population associated with DCs in the decidual among patients with URPL. These findings position MX1+ Tem cells and ICOSL+ DCs as potential biomarkers for patient stratification, which may guide targeted therapies to prevent URPL.

Our findings revealed profound T cell dysregulation at the maternal-fetal interface in URPL, characterized by expanded T cell infiltration and hyperactivation of the MX1+Tem subset. These cells, defined by co-expression of interferon-stimulated genes such as MX1, activated genes and proliferative genes, represent a unique immunological hallmark of URPL. Crucially, our data indicated a potential pivotal role for DCs-mediated ICOSL signaling in this process. The sustained T cell activation, potentially fueled by this axis, mirrors mechanisms in allograft rejection and provides a plausible explanation for the breakdown of fetal tolerance in URPL. The elevated T cell infiltration and activation observed in URPL decidual correlated with immune microenvironment dysregulation. Under physiological conditions, T cells account for 5–15% of decidual immune cells, maintaining fetal tolerance through regulatory dominance. [ 23 ] However, T cell proportion was found surged to 20.16% among URPL women in our study, suggesting antigen-driven clonal proliferation, which was further supported by enrichment of proliferation and activation biomarkers in T cells. Our findings align with the recent single-cell study in pregnancy loss, which revealed a prominent CD11c+CD8+ T cells, contributing to trophoblast injury through the secretion of GZMB and IFN-γ [ 24 ]. In the other hand, metabolic reprogramming toward other glycolytic pathways may mirror the bioenergetic demands of activated effector T cells, which relied on more energy for rapid effector function and membrane biosynthesis. [ 25 ] This metabolic shift is critically aligned with recent frameworks in cancer research, where altered cellular metabolism is shown to shape immune dysfunction and govern therapeutic responses. [ 26 , 27 ] What’s more, GO analysis further corroborated this state, linking T cell hyperactivation to chronic long-term antigen stimulation, which likely disrupts fetal tolerance and decidual immunosuppression, resembling lymphocyte infiltration patterns in cancer models.[ 28 ] Our study further demonstrated reduced PD-L1 expression specifically in decidual stromal cells of URPL patients. This reduction coincided with pronounced activation of decidual T cells, particularly the MX1+Tem subset, which suggested a potential link between impaired stromal PD-L1-mediated immunosuppression and T cell hyperactivation. Crucially, PD-L1 sustains fetal-maternal tolerance by engaging PD-1 on activated T cells, delivering inhibitory signals that constrain excessive T-cell responses and protect the semi-allogeneic fetus. [ 29 ] The observed PD-L1 reduction in URPL may represent a pathological failure of this immunosuppressive checkpoint, weakening physiological restraint on T cells and permitting unrestrained hyperactivation. This scenario is consistent with mechanisms of allograft rejection, where insufficient immunosuppression leads to graft failure.[ 30 ] Among patients with URPL, a distinct T-cell subset characterized by high expression of ISGs, including MX1, has been identified and termed MX1+Tem cells. M×1 is a dynamin-like GTPase and a member of the ISGs family. Robustly induced by IFN-α/β, it thus serves as a sensitive indicator of localized IFN pathway activation. [ 31 ] We hypothesize that in URPL, aberrant recognition of fetal-derived antigens triggers a similar IFN-I response at the maternal-fetal interface. This persistent IFN-I signaling, akin to that seen in chronic viral infections or the tumor microenvironment, may drive the hyperactivation and altered function of these MX1+Tem cells.[ 32 , 33 ] Mechanistically, IFN-I activation initiates a feedforward loop. It robustly induces the expression of ISGs, which in turn stabilizes STAT1 and thereby amplifies a pro-inflammatory transcriptional program in T cells [ 34 ]. This ISG15/STAT1 axis likely operates in parallel to MX1-associated pathways to promote a cytotoxic T-cell phenotype. Concurrently, sustained IFN-β feedback further augments local inflammation and T-cell recruitment, potentially through the activation of antigen-presenting cells such as DCs and macrophages [ 35 ]. Furthermore, M×1 upregulation may synergize with this inflammatory milieu to induce metabolic stress in T cells, characterized by ROS accumulation and mitochondrial fragmentation, which could ultimately contribute to their dysfunction [ 36 ]. Collectively, these interconnected mechanisms provide a plausible framework for how MX1+Tem cells acquire and exert enhanced cytotoxicity in the context of URPL. Two lines of evidence support the activated state of interferon-responsive MX1+Tem cells. Firstly, the MX1+Tem cells exhibited simultaneous upregulation of cytotoxic effectors such as GZMB and proliferative markers such as MKI6 7 . Secondly, the metabolic profile of MX1+Tem cells, marked by attenuated oxidative phosphorylation and enhanced lipid metabolism, paralleled the state of activated T cells in chronic infections and tumor microenvironment. [ 29 ] The identification of these specific T cell subsets mirrors recent advances where single-cell dissection of immune heterogeneity has successfully identified immunomodulatory biomarkers that predict clinical responses in complex immune environments. [ 37 ] Strikingly, pseudotemporal trajectory analysis positioned MX1+Tem cells at an intermediate state between effector and exhausted subsets, reflecting a balance between activation and metabolic stress. This transitional state aligns with stem-like exhausted T cells in melanoma, where TCF1 retain self-renewal activation capacity and inhibition.[ 38 ] Our study observed altered interactions between decidual DCs and MX1+Tem cells in URPL, suggesting that DCs may contribute to MX1+Tem cell hyperactivation. DCs typically activate T cells via three primary mechanisms, including antigen presentation through MHC-II complexes, co-stimulatory signals and cytokine secretion. [ 39 ] In our study, DCs exhibited upregulated ICOSL interaction and MHC-II expression to MX1+Tem cells, suggesting a potentially enhanced antigen-presenting capacity. Notably, ICOSL upregulation on decidual DCs appears associated with IFN-I hyperactivity (evidenced by MX1+Tem signatures), positioning ICOSL as a critical downstream effector of IFN-I signaling. Furthermore, persistent ICOSL signaling which potentially induced by decidual IFN-I and acting on the ICOS receptor on MX1+Tem cells, may contribute to disrupting fetal-maternal tolerance by directly sustaining hyperactivation and metabolic stress in these IFN-responsive cells. This axis recapitulates allograft rejection mechanisms, which is associated with pregnancy failure in URPL. Targeting this axis may offer therapeutic benefits, analogous to PD-1 blockade in cancer immunotherapy.[ 40 ] While this study leverages single-cell transcriptomics to delineate T cell heterogeneity and landscape in URPL decidual, the inter-individual heterogeneity of scRNA-seq analysis among clinical patients may influence the transcriptional profiles, the modest cohort size and single-center design limit generalizability. Longitudinal studies linking MX1+Tem dynamics to clinical outcomes are needed. Further validation through functional co-culture assays including blocking assays and animal models assessing interaction between MX1+Tem cells and ICOSL+DCs are needed. TCR sequencing to track clonal expansion are also needed in the future. Nevertheless, our findings provide critical insights into the DC-MX1+Tem axis and highlight ICOSL as a potential therapeutic target. These data contribute to the burgeoning Human Cell Atlas and pave the way for developing immunotherapies to prevent URPL.

T lymphocytes exhibit remarkable functional plasticity, transitioning through differentiation states ranging from naive to helper, regulation, effector, effector memory (Tem), and exhaustion subsets. [ 1 ] This adaptive plasticity enables T cells to balance immune surveillance and tolerance, a critical mechanism at the maternal-fetal interface where fetal antigens evoke allogeneic-like immune challenges. [ 2 ] Recent evidences have unmasked the complexity within this niche. For instance, CCR8+ decidual regulatory T cells have been identified as a uniquely suppressive subset essential for maintaining maternal-fetal tolerance during early pregnancy. [ 3 ] Furthermore, classical T cell activation involves antigen recognition by dendritic cells (DCs) through major histocompatibility complex (MHC) peptide complexes, complemented by co-stimulatory signals such as CD28–B7 interactions. [ 4 ] Specifically, emerging costimulatory family members such as BTN2A2 have been shown to regulate maternal-fetal tolerance by modulating T cell receptor signaling.[ 5 ] Beyond costimulatory signals, T cell activation states are shaped by complex microenvironmental cues, including cytokine gradients such as interferon and metabolic reprogramming. [ 6 ] Notably, metabolic shifts toward aerobic glycolysis and lipid biosynthesis drive T cell activation in tumor microenvironments, often marked by blunted cytotoxicity. [ 7 ] These dynamics are particularly critical at the maternal-fetal interface, where fetal trophoblasts evade immune rejection through mechanisms analogous to tumor immune tolerance. [ 8 ] While T cell activation has been well-characterized in oncology, the role in pregnancy complications, particularly unexplained recurrent pregnancy loss (URPL), remains enigmatic. URPL, defined as two or more unexplained consecutive pregnancy loss before 20 weeks of gestation, is increasingly linked to dysregulation of the decidual immune microenvironment. [ 9 ] Emerging single-cell studies have unmasked the decidual immune landscape as a tightly regulated niche populated by T cells, DCs, natural killer (NK) cells and macrophages, suggesting a tightly regulated crosstalk network. [ 10 ] Although NK cell-mediated immune tolerance has been extensively studied in URPL, the role of T cell subsets remains poorly understood. [ 11 ] In chronic inflammatory conditions, such as tumors, persistent antigen exposure drives T cell activation and dysfunction. [ 12 ] During pregnancy, fetal trophoblasts which are analogous to semi-allografts, orchestrate immune tolerance through mechanisms resembling tumor immune evasion, including regulatory T cell induction. [ 13 , 14 ] Recent findings regarding T cell exhaustion-like states and aberrant protein modifications, such as Tim-3 palmitoylation, suggest that specific T cell subsets may play a more dominant role in URPL than previously recognized. [ 15 ] However, the drivers of T cell dysfunction in URPL, particularly their interplay with DCs, remained unexplored. In this study, we aimed to indentify and characterize the specific T-cell subset responsible for the breakdown of maternal-fetal tolerance in URPL and to delineate its regulatory crosstalk with DCs, which was a critical step toward targeted immunotherapy for URPL.

Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1 Supplementary Material 2 Supplementary Material 2 Supplementary Material 3 Supplementary Material 3 Supplementary Material 4 Supplementary Material 4