Multimodal profiling of term human decidua demonstrates immune adaptations with pregravid obesity

SUMMARY

Leukocyte diversity of the first-trimester maternal-fetal interface has been extensively described; however, the immunological landscape of the term decidua remains poorly understood. We therefore profiled human leukocytes from term decidua collected via scheduled cesarean delivery. Relative to the first trimester, our analyses show a shift from NK cells and macrophages to T cells and enhanced immune activation. Although circulating and decidual T cells are phenotypically distinct, they demonstrate significant clonotype sharing. We also report significant diversity within decidual macrophages, the frequency of which positively correlates with pregravid maternal body mass index. Interestingly, the ability of decidual macrophages to respond to bacterial ligands is reduced with pregravid obesity, suggestive of skewing toward immunoregulation as a possible mechanism to safeguard the fetus against excessive maternal inflammation. These findings are a resource for future studies investigating pathological conditions that compromise fetal health and reproductive success.

In brief

Sureshchandra et al. describe the changes in the decidual immune landscape in the third-trimester placenta compared with early pregnancy and the establishment of immunoregulatory mechanisms in decidual tissues as a mechanism of fetal protection from obesity-induced maternal inflammation.

Graphical abstract

INTRODUCTION

The placenta is a complex organ composed of maternal-derived decidua and fetal-derived chorionic villi.¹ Functionally, it facilitates gas, nutrient, and waste exchanges between the fetus and the mother. The maternal compartment, called the decidua, harbors a unique immune landscape, which facilitates trophoblast invasion, maternal tolerance, debris clearance, and antimicrobial defense.² Studying longitudinal changes in frequencies, phenotype, and function of decidual immune cells with healthy gestation is of great interest, given the role of decidua in the pathophysiology of several obstetric complications such as preterm labor,³ preeclampsia,^4–6 and chorioamnionitis.⁷ First-trimester decidua (10–12 weeks) is composed mostly of decidual Natural Killer cells (dNKs, ~70% of immune cells), decidual macrophages (20%–25%), and T cells (3%–10%).^8,9 The dNKs are the earliest immune cells detected in the decidua, differentiating from blood NK cells, and have been demonstrated to play a critical role in regulating trophoblast invasion and decidualization.¹⁰ Decidual macrophages and T cells play important roles in maintaining immune tolerance,¹¹ mediating antimicrobial responses,² promoting spiral artery remodeling,¹² and clearing apoptotic bodies from the placental vasculature.¹³ While the diversity and characteristics of decidual immune cells during the first trimester have been well described, those at term remain poorly defined. Current models predict that over the course of healthy gestation, dNK and macrophage numbers dwindle while those of T cells remain intact in the decidua basalis.¹⁴ However, there is still no consensus about the phenotypic and functional diversity of decidual T cells and macrophages at term.

Maternal pre-pregnancy (pregravid) obesity is associated with an increased risk for several obstetric complications, including gestational diabetes and gestational hypertension,^15,16 pre-eclampsia,¹⁷ placental abruption,¹⁸ preterm delivery,^19,20 and need for cesarean delivery.^21,22 These predispositions are mediated by altered placental structure,^23,24 increased levels of oxidative stress,²⁵ and elevated expression of proinflammatory factors.²⁶ Studies in animal models suggest that changes in the decidual inflammatory environment can be attributed to dysregulated cell turnover and immune cell rewiring^27,28; however, these changes have been poorly defined in humans. For example, first-trimester uterine NK cells in pregnant women with obesity exhibitan imbalance ofinhibitoryand activatingreceptors resulting in aberrant activation.²⁹ Immunohistochemistry and flow cytometry-based studies have reported increased CD68+ macrophage frequencies in chorionic villi with maternal obesity.³⁰ In line with these observations, higher transcript levels of proinflammatory genes IL6, CCL2, IL8, and TLR4 were detected in a baboon model of high-fat diet-induced obesity,^31,32 suggesting potential immune cell adaptations with obesity at the maternal-fetal interface. However, a clear consensus on these observations is lacking, perhaps due to differences in choices of placental anatomic sections analyzed, immune markers used, experimental readouts measured, and confounding variables in human studies.²³

A detailed understanding of immune cell adaptations in the decidua that support late pregnancy is currently lacking. More importantly, how maternal obesity reprograms the immune milieu of decidua at term is yet to be fully assessed. To address these gaps in our knowledge, we carried out a multi-pronged study using single-cell RNA sequencing (scRNA-seq), flow cytometry, Luminex, and functional assays to profile the diverse subsets of macrophages, T cells, and NK cells in term decidua and compared them with the single-cell atlas of early (6–12 weeks of gestation) maternal-fetal interface reported by Vento-Tomo et al.⁹ The use of decidua isolated from a cohort of pregnant mothers undergoing scheduled cesarean delivery in the absence of any obstetric complications allowed us to study the impact of pregravid obesity on immune cell adaptations at term without the confounders of labor.

Our data show that relative to the first trimester, decidual immune cells exhibit signs of heightened activation status at term. Furthermore, the relative abundance of T cells increases at term with a concomitant reduction in frequencies of macrophages and NK cells. We also report a shift from the dNK1 subset dominant at 12 weeks, to the dNK3 subset at term. At term, compared with their blood counterparts, decidual T cells exhibited enhanced regulatory and Th17 responses, whereas Th1 responses were attenuated, in line with immune tolerance. Finally, we observed two major macrophage subsets that can be distinguished based on HLA-DR and CD11c expression that may represent a tissue-resident and a monocyte-derived subset. Pregravid obesity led to a reduction of T cell numbers but increased accumulation of macrophages and functional rewiring of HLA-DR^highCD11c^high macrophages. Furthermore, decidual macrophages from mothers with obesity generated dampened responses to bacterial ligands, suggesting skewing toward regulatory phenotype. Collectively, our study provides an atlas of the immune landscape of term decidua and adaptations associated with pregravid obesity, shedding light on immune mechanisms limiting fetal exposure to maternal inflammation.

RESULTS

The immune cell landscape of decidua at term

While the diversity of human immune cells at the maternal-fetal interface during the first trimester has recently been described,⁹ the phenotypic and functional diversity of decidual immune cells at term is less understood. Here, we characterized CD45+ decidual leukocytes and matched peripheral blood mononuclear cells (PBMCs) obtained from pregnant women undergoing scheduled cesarean delivery using a combination of single-cell sequencing, flow cytometry, and functional assays (Figure 1A). Live CD45+ decidual leukocytes and PBMCs collected at term from four subjects were sorted and profiled using scRNA-seq (Figure 1A). Major decidual immune cell subsets were identified following dimensionality reduction and removal of potential blood contaminants following the integration of immune cells from blood and decidua (Figures S1A and S1B). The expression of structural genes defining tissue-resident macrophages (VIM) and T cells (ANXA1, EZR, TUBA1A) was higher in decidual leukocytes compared with PBMCs (Figure S1C), confirming their tissue-resident identity.

Figure 1. Defining the immune landscape of term decidua.

(A) Experimental design: CD45+ leukocytes were isolated from maternal blood and decidua from term cesarean deliveries (n = 4). Single-cell suspensions of decidual leukocytes and matched PBMCs were subjected to gene expression profiling using 10× 3′ single-cell gene expression protocol (scRNA-seq).

Additionally, sorted decidual and peripheral T cells were surface stained with DNA oligo-tagged antibodies and hashing antibodies and subjected to 5′ gene expression and TCR on the 10x platform (n = 5). Finally, the frequency, phenotype, and function of various immune cells were assayed using flow cytometry.

(B) Uniform Manifold Approximation and Projection (UMAP) of 12,698 CD45+ decidual leukocytes from four donors.

(C) Violin plots showing log-transformed normalized expression levels of major cell-type determining markers identified using Seurat.

(D) Stacked bar graph comparing frequencies of major immune cell types in term decidua (n = 62) using multiparameter flow cytometry. Error bars represent the mean and standard error of the mean (SEM).

(E) Changes in frequencies of major immune cell subsets in the decidua based on scRNA-seq obtained from first-trimester (T1, weeks 6–12) decidua⁹ integrated with week 37 (T3) data generated in this study.

(F) Feature plots comparing expression of transcription factors, surface markers, and cytokines associated with immune activation at T3 relative to T1.

Overall, the decidual immune landscape was dominated by T cells, NK cells, and macrophages (Figure 1B). Decidual CD4 and CD8 T cells were identified based on the relative expression of IL7R, CCR7, and XCL1, while regulatory T cells (Tregs) were identified based on the high expression of CTLA4 and FOXP3 (Figures 1B and 1C, Table S1). dNKs were identified based on the expression of NKG7, NCAM1, and CD160; macrophages based on the expression of CD14 and CD68; and dendritic cells (DC) based on the expression of FCER1A (Figure 1C and Table S1). Decidual immune cell distribution was validated using flow cytometry in a larger cohort of samples (n = 62) (Figures S1D and 1D).

We next integrated our data with those obtained from 6- to 12-week-old (first-trimester) decidua to understand longitudinal immune changes (Figures S1E and S1F). Compared with the first trimester, the relative frequencies of dNK cells and macrophages decreased while those of T cells, notably Tregs, increased at term (Figures 1E, S1E, and S1F). Moreover, both T and NK cells expressed high levels of NFKB and IFNΓ (Figure 1F), while decidual macrophages expressed higher levels of NFKB1 and NLRP3, CD83, and proinflammatory cytokines TNF and IL1B (Figure 1F), suggesting a state of heightened immune activation at term.

To eliminate the possibility of contaminating fetal cells in decidua, we integrated decidual leukocytes with matched chorionic villi and cord blood cells (n = 4) (Figures S2A–S2D). This analysis showed a lack of shared clusters with cord blood or contaminating Hofbauer cells (cluster 16) (Figure S2E), fetal cells found exclusively in the chorionic villous, in the decidua. Furthermore, monocytes were restricted to the cord blood mononuclear compartment (Figure S2E). This analysis confirms the clear separation of the decidua from the chorionic villous during manual dissection of the placenta at delivery and the lack of contaminating fetal cells in the decidua single-cell analysis.

Gestational shifts in dNK cells

We next investigated shifts in the dNK subsets with pregnancy. Within the dNK cell population, the frequency of dNK1 (expressing B4GALNT1 and CYP26A1), which is the largest cluster found at T1,⁹ was reduced at T3 (Figure S3A). Similarly, dNK2, a subset of cytotoxic dNKs expressing high levels of granulysin (GNLY) and granzymes (GZMB and GZMH) (Figures S3A and S3B) also decreased in proportion between T1 and T3. On the other hand, the frequency of dNK3 (expressing inhibitory receptors CD160 and KLRB1) increased at T3 relative to T1 (Figures S3A and S3B).

We then used flow cytometry to confirm and validate these shifts in dNK subsets. The dNK subsets were identified using CD103, CD49a, CD39, and ITGB2, as previously described,⁹ as well as activation markers NKG2A, NKG2C, KIR2DL4 (CD158d), KIR2DL2/3 (CD158b), KIR2DL1 (CD158a), and CD160 (Figure S3C). The frequencies of dNK2 and dNK3 were significantly higher than dNK1 at T3 (Figure S3D). Expression of activating (NKG2A) and inhibitory (NKG2C) receptors, as well as killer immunoglobulin receptor KIR2DL4, an activating receptor for HLA-G expressed on extravillous trophoblasts, were highest on dNK2 followed by dNK3 and lowest on dNK1 (Figures S3E and S3F). Expression of HLA-C receptors also varied among dNK subsets—a higher proportion of dNK3 cells expressed KIR2DL1, whereas KIR2DL2/3 expression was highest in dNK2 (Figure S3G). The inhibitory receptor CD160 was expressed at a lower level on dNK3 cells (Figure S3H). Finally, all dNK subsets were capable of degranulation as indicated by increased levels of CD107a in response to PMA/ionomycin stimulation (Figure S3I).

Phenotypic and clonal diversity of decidual T cells at term

In contrast to circulating T cells, decidual T cells were predominantly of effector memory phenotype (Figures 2A and S4A). To compare the phenotypic and clonal diversity of decidual T cells relative to blood T cells, we performed 5′ gene expression and T cell repertoire analyses at the single-cell level coupled with cell surface protein analysis using TotalSeq-C technology (Figure 1A, n = 4/group). Dimensionality reduction, clustering, and integration of blood and decidual T cells showed 32 clusters (Figure S4B). Naive T cell clusters were identified based on high expression of IL7R, SELL, and CCR7 (Figure S4C) and excluded from further analysis. The remaining 13 non-naive T cell clusters (red font) were reorganized into 17 clusters of memory T cells (Figures 2B and 2C, and Table S2). Interestingly, there was very little overlap between the decidual and blood compartments (Figure S4D), occurring only within proliferating, activated CD8, and CM CD4 T cell clusters (Figures 2B and S4D). Overall, decidual T cells expressed higher levels of CD69 and PD-1 (both at protein and gene expression levels) but lower levels of SELL (encoding CD62L) compared with blood memory T cells (Figures 2D, 2E, and S4E). Single-cell analyses also showed an increased abundance of regulatory T cells (expressing high levels of FOXP3 and CTLA4 and surface CD25) (Figures 2C and S4F), which was confirmed by flow cytometry (Figure 2F).

Figure 2. Phenotypic and clonal diversity of decidual T cells at term.

(A) Frequencies of naive and memory CD4 and CD8 T cell subsets in blood (n = 55) and decidua (n = 6) at T3 using flow cytometry. Error bars represent the mean and standard error of the mean (SEM).

(B) UMAP of matched decidual and blood memory T cells at term (n = 4), annotated by surface- and gene expression markers.

(C) Stacked violin plots of key gene markers of memory T cell subsets within clusters using normalized transcript counts. Red font indicates clusters shared between blood and decidual leukocytes.

(D) Frequencies of CD4 (top) and CD8 T cells (bottom) in blood (n = 3) and decidua (n = 9) expressing CD69 and CD103. Error bars represent the mean and standard error of the mean (SEM).

(E and F) Frequencies of (E) PD-1+ CD4 and CD8 T cells (n = 3/group) and (F) Tregs within CD4 (n = 3 for blood and n = 9 for decidua). Error bars represent the mean and standard error of the mean (SEM).

(G) Th1 cytokine responses of blood (n = 3) vs. decidual (n = 9) CD4 (left) and CD8 (right) T cells in response to 4 h PMA/ionomycin stimulation.

(H) Treg and Th17 cytokine responses of blood (n = 3) vs. decidual (n = 3) CD4 T cells following 4 h PMA/ionomycin stimulation.

(I) T cell clone sizes in blood and decidua at term.

(J) Chao diversity of T cell clones from blood and decidua within each donor.

(K) Clonal tracking of the 10 most abundant blood T cell clones in the decidua (top panel) and blood (bottom panel) in each donor. ****p < 0.0001, ***p < 0.001, **p < 0.01, and *p < 0.05.

Gene expression profiles suggest heightened activation of decidual T cells relative to those in circulation as indicated by increased expression of NFKB1, IFNΓ, and TNF (Figure S4G). However, Th1 cytokine production (tumor necrosis factor [TNF] α, macrophage inflammation protein [MIP]-1β, and interferon [IFN]γ) by decidual T cells following PMA/ionomycin stimulation was attenuated compared with circulating T cells (Figure 2G). Single-cell analyses also identified a Th17 (expressing RORA) cluster in the decidua (Figures 2B and 2C), which aligned with significantly higher IL-17 production by decidual compared with circulating T cells (Figure 2H). Finally, enhanced transforming growth factor (TGF)-β production following PMA/ionomycin stimulation correlated with higher frequencies of regulatory T cells (Figure 2H).

We next compared T cell receptor (TCR) repertoire diversity between blood and decidual memory T cells based on paired TCR alpha and beta genes captured in the single-cell assay (Figure 2B). Analysis of clonal sizes shows reduced TCR repertoire diversity in the decidua as indicated by the absence of large T cell clones (Figure 2I) and reduced Chao diversity (Figure 2J). Additionally, in all four donors, T cell clones that were predominant in the blood were detected in the decidua, albeit at lower proportions (Figure 2K, top panel). On the other hand, predominant decidual T cell clones were only detected in the blood of three of the four donors (Figure 2K, bottom panel), with proportions higher than in the blood in only one of the four donors.

Two distinct macrophage subsets exist in term decidua

Macrophages made up roughly 20% of leukocyte frequencies at term (compared with 28% in the first trimester) (Figure 1E). Within the decidua, we identified two major subsets of macrophages based on surface expression of HLA-DR and CD11c (Figures 3A and 3B), as well as size/granularity (Figure S5A). Both subsets had distinct transcriptional profiles compared with matched peripheral monocytes (Figure 3A). The subset expressing lower levels of HLA-DRA (HLA-DR^low) expressed higher levels of alarmins S100A8, S100A9, S100A10, IL1B, CXCL8, and markers associated with in vitro IFNγ conditioned macrophages (TREM1) (Figures 3A and 3B). On the other hand, the HLA-DR^high subset expressed higher levels of tetraspanins (CD9 and CD81) and canonical markers associated with in vitro IL-4 conditioned macrophages (TREM2, MSR1, APOE) (Figures 3A and 3B). Both subsets of macrophages were also detected in first-trimester decidua (Figure S1E), with a modest increase in HLA-DR^low subset (22% at T1 vs. 34% at T3) and modest decrease in HLA- DR^high subset (79% at T1 vs. 67% at T3) with gestational age. However, both subsets of macrophages were highly activated at term, with roughly 50%–75% of HLA-DR^high macrophages expressing activation marker CD83, cytokine IL1B, and chemokine CXCL8 (Figure S5B). The presence of these two decidual macrophage subsets at term was confirmed by flow cytometry (Figure 3C) and imaging flow cytometry (Figure 3D) with the HLA-DR^high subset expressing higher levels of CD11c compared with the HLA-DR^low subset.

Figure 3. Two distinct subsets of macrophages exist in term decidua.

(A) Heatmap of top marker genes for the decidual HLA-DR^low and HLA-DR^high subsets, classical and non-classical blood monocytes. Each column is a bin of a fixed number of cells and each row represents a gene. A subset of genes within each cluster is annotated. High and low expression are represented as red and blue, respectively.

(B) Violin plots comparing expression of gene markers within decidual macrophage subsets.

(C) Surface expression of HLA-DR and CD11c measured by flow cytometry.

(D) Brightfield and fluorescence profiles of macrophage subsets captured using imaging flow cytometry.

(E) Median Fluorescence Intensities (MFI) of macrophage surface markers on decidual macrophage subsets (n = 7/group).

(F) Frequencies of TREM2+ FOLR2+ (n = 5/group) and CD163+ (n = 7/group) within decidual macrophages.

(G) Phospho p65 signal in blood monocytes and decidual macrophage subsets (n = 3/group) by flow cytometry. Error bars represent the mean and standard error of the mean (SEM).

(H) Comparison of baseline secreted cytokine profiles (median) of sorted decidual macrophage subsets (n = 4/group) and blood monocytes (n = 2). * and # represent significant differences relative to blood monocytes between HLA-DR^low and HLA-DR^high macrophages, respectively.

(I) Responses to overnight lipopolysaccharide (LPS) stimulation of total decidual leukocytes (n = 8/group). Error bars represent the mean and standard error of the mean (SEM).

(J) MFI of intracellular cellROX (n = 7/group).

(K) Percentage of E. coli pHrodo+ (n = 8/group) in decidual macrophages.

(L) Oxygen consumption rate and extracellular acidification rates of decidual macrophages (n = 6/group). ****p < 0.0001, **p < 0.01, *p < 0.05, and #p<0.1.

To further characterize these decidual macrophage subsets at term, we measured protein expression of several markers associated with in vitro polarized macrophages (Figure S5C). Interestingly, surface expression of TREM-1 was lower on the HLA-DR^low subset despite the higher transcript levels. Moreover, surface expression of CCR2 was higher in the HLA-DR^low subset (Figure 3E), suggesting that this population may likely be infiltrating monocyte-derived macrophages. On the other hand, HLA-DR^high macrophages expressed higher levels of surface markers CD86, CD11c, CD16, CD64, and TLR4 (Figures 3E, S5C, and S5D) that are traditionally associated with IFNγ conditioned macrophages. Interestingly, this subset was also enriched for regulatory CD163+ and TREM2+FOLR2+ macrophages (Figures 3E, 3F, and S5E), suggesting that this population may be a tissue-resident.

Functional diversity of macrophage subsets in term decidua

We next investigated if the transcriptional heterogeneity of decidual macrophages correlated with functional diversity. Flow analyses showed that the HLA-DR^high macrophage subset exhibited a more heightened state of activation, expressing higher levels of phosphorylated p65 (Figure 3G). This heightened activation state was further evident when we assessed their baseline cytokine/chemokine secretion profiles. Sorted blood monocytes, and both macrophage subsets from the same donors were cultured in RPMI for 16 h, and their secretomes were analyzed using Luminex. Decidual macrophages secreted higher IL-12p70, IL-23, IL-10, CXCL9, S100B, granulocyte-macrophage colony-stimulating factor (GM-CSF), and CXCL11 compared with blood monocytes (Figure 3H). However, secreted levels of both TNFα and IL-10 were significantly higher in HLA-DR^high relative to HLA-DR^low macrophages (Figure 3H), suggesting differential activation states between the two subsets.

Moreover, a higher percentage of HLA-DR^high macrophages responded to lipopolysaccharide (LPS) stimulation as evidenced by the increased frequency of IL-6+TNFα+ producing cells (Figure 3I), expressed higher levels of reactive oxygen species (ROS) (Figures 3J and S6A), and were more phagocytic (Figure 3K). Since these functions are intricately linked to the metabolic state of the cell, we next compared several aspects of cellular metabolism. Our analysis shows that HLA-DR^high macrophages are more lipid-laden than HLA-DR^low macrophages (Figure S6B). Moreover, seahorse energy phenotype assay indicated that at baseline, the HLA-DR^high macrophages had a higher oxygen consumption rate (OCR) and lower extracellular acidification rate (ECAR) (Figure 3L) correlating with higher mitochondrial membrane potential (Figure S6C). Finally, HLA-DR^high macrophages exhibited greater uptake of glucose analog 2-NBDG and increased expression of glucose transporter GLUT1 (Figure S6D). Collectively these observations suggest HLA-DR^high macrophages are more metabolically active than HLA-DR^low cells.

Identification of heterogeneous macrophage subsets within HLA-DR^high subset

Given the elevated expression of both inflammatory (TREM1) and regulatory markers (TREM2, CD163, FOLR2) within HLA-DR^high macrophages (Figures 3E and 3F), we next investigated the heterogeneity of this subset. High-resolution clustering and visualization of Uniform Manifold Approximation and Projection (UMAP) showed three distinct macrophage clusters within HLA-DR^high cells (Figure 4A) that were transcriptionally organized along a pseudo temporal trajectory originating from HLA-DR^low to cluster 1, cluster 2, and terminating in cluster 3 of HLA-DR^high macrophages (Figures 4A and 4B). The genes defining the origin of this trajectory were primarily proinflammatory LYZ, NLRP3, TREM1, and S100 alarmins, which were highly expressed in cluster 1 (Figures 4C and 4D). Cells in cluster 2 expressed high levels of HLA-DRA, scavenger receptors (CD36, MARCO, MSR1), lipid signaling molecules (FABP5, CD9, CD63), and regulatory molecules (TREM2 and IL1RN) (Figures 4C and 4D). Finally, cells in cluster 3 express high levels of complement proteins (C1QA, C1QB), folate receptor FOLR2, and negative regulators of inflammation (ATF3, KLF2, KLF4) (Figures 4C and 4D). We then compared the transcriptomes of these clusters with recently published data of macrophages stimulated with TPP (TNF, Pam3CSK4, and prostaglandins PGE2), a model for chronic inflammation.³³ In concordance with the trajectory analysis and the shift from proinflammatory to regulatory gene expression patterns, cluster 1 of HLA-DR^high and HLA-DR^low macrophages had the highest scores while cluster 3 had the lowest score (Figure 4E).

Figure 4. Diverse subsets within HLA-DR^high macrophages.

(A) UMAP highlighting diverse subsets of macrophages within HLA-DR^high macrophages when analyzed under a higher resolution. Arrows point in the direction of pseudo temporal trajectory predicted by Monocle.

(B) Trajectory of cells ordered by predicted pseudo time colored by pseudotime (top) or by the cluster of origin (bottom).

(C) Heatmap of top 100 genes that define the trajectory originating from HLA-DR^low macrophages, through clusters 1, 2, and 3 of HLA-DR^high macrophages. Low and high gene expression is indicated by blue and red, respectively.

(D) Violin plots comparing the expression of the top eight markers from each of the HLA-DR^high macrophages across all three clusters.

(E) Violin plot comparing aggregate score for modules of genes highly expressed in macrophages stimulated with TPP (TNF+Pam3CSK4+PGE2), an in vitro model for chronic inflammation.

(F) Gating strategy for flow cytometry-based identification of HLA-DR^high macrophage clusters based on markers discovered from differential gene analyses.

(G) Three-way functional enrichment of top genes expressed by each HLA-DR^high subset performed in Metascape. The size of the bubble denotes the number of genes mapping to each gene ontology (GO) term, and the intensity of color denotes the statistical strength of prediction.

Canonical markers that can differentiate the three HLA-DR^high subclusters were validated using intracellular staining of S100A8/A9, and surface expression of FOLR2 and CD9 by flow cytometry (Figure 4F). Functional enrichment of the top 100 marker genes from each sub-cluster showed that cluster 1 has a dominant signature associated with LPS response, while cluster 2 had features associated with wound healing and antigen presentation, and finally cluster 3 was enriched in regulatory pathways (Figure 4G).

Pregravid obesity is associated with reduced frequency of decidual T cells and functional rewiring of HLA-DR^high decidual macrophages

The prevalence of pregravid obesity is increasing³⁴ and is implicated in elevated risk for perinatal complications, which may have a long-term impact on maternal and offspring health.^{15,22,35–41} Therefore, we next investigated the impact of maternal pregravid obesity on the immunological landscape of the decidua. Frequency of major decidual immune subsets from lean pregnant women (BMI <25) was compared with those with obesity (BMI >30). Obesity had no impact on dNK cell frequencies (Figure 5A). In contrast, the proportions of CD4 and CD8 T cells within the CD45 compartment were significantly reduced (Figure 5A). Moreover, decidual CD4 and CD8 T cell frequencies negatively correlated with maternal pregravid BMI (Figure S7A). However, no differences in the relative abundance of naive/memory T cell subsets (Figure S7B) or tissue-resident subsets (Figure S7C) were observed with pregravid obesity. Furthermore, T cell responses to PMA/ionomycin stimulation were comparable between the two groups (Figure S7D). Regulatory T cell frequencies, on the other hand, increased with pregravid obesity (Figure 5A).

Figure 5. Decidual immune cell composition with maternal obesity.

(A and B) Dot plots comparing (A) lymphocyte and (B) macrophage proportions within CD45+ decidual cells in lean pregnant women (n = 6 for Tregs, 32 for rest) and those with obesity (n = 6 for Tregs, 31 for rest).

(C) Gating strategy and dot plots comparing frequencies of S100A9+ cells within HLA-DR^high macrophages (n = 5/group). Error bars represent the mean and standard error of the mean (SEM).

(D) Stacked bar graph comparing frequencies of decidual leukocytes within total cells sequenced in each group (within CD45+ cells).

(E) Stacked bar comparing HLA-DR^high macrophage cluster within total CD45+ cells.

(F and G) Gene Ontology (GO) terms of genes that are up- (F) and downregulated (G) with pregravid obesity within HLA-DR^high macrophages predicted by Metascape. Only differentially expressed genes with a Log₂ fold change cutoff of 0.4 were analyzed.

(H) Violin plots compared normalized transcripts of select genes that are significantly up- (left) or downregulated (right) with obesity.

(I) Dot plots comparing frequencies of TREM1+ cells or cytokine expressing cells within HLA-DR^high macrophages (n = 5/group). Error bars represent the mean and standard error of the mean (SEM).

(J) Violin plots comparing secreted factors at baseline by FACS sorted HLA-DR^high macrophages (n = 3/group).

(K) Experimental design for measuring ex vivo macrophage responses to E. coli or bacterial TLRs.

(L) Bar graphs comparing cytokine-producing cells TNFα+IL-6+ (left) and TNFα+IL-6+IL1β+ (right) following stimulation with bacterial TLRs and E. coli (n = 6/group). Error bars represent the mean and standard error of the mean (SEM). ****p < 0.0001, ***p < 0.001, **p < 0.01, *p < 0.05.

Reduction in decidual T cells with pregravid obesity was accompanied by an expansion of decidual macrophages (Figure S7E), driven by the HLA-DR^high subset (Figure 5B). Moreover, the frequency of HLA-DR^high but not HLA-DR^low subsets positively correlated with maternal pregravid BMI (Figure S7F). Flow analyses of HLA-DR^high macrophages showed an increase in S100A9+ macrophages (Figure 5C), a marker of subclusters 1 and 2 (Figure 4F), and recently infiltrated macrophages.⁴²

Next, we leveraged scRNA-seq to further interrogate the impact of pregravid obesity on the transcriptional landscape of decidual leukocytes (Figure S8A). In line with the flow cytometry data, this analysis showed a contraction of CD4 and CD8 T cells and expansion of decidual macrophages (Figure 5D), driven by increased frequencies of all three HLA-DR^high subsets (Figure 5E). Differential gene expression analyses of HLA-DR^high macrophages between lean and obese groups showed an up-regulation of pathways associated with response to lipoproteins and cholesterol transport (FABP5, CD36, APOE, APOC1, CD9), response to stress (CEBPA, CEBPB, HSPA5, HSPA8), and antigen processing and presentation (HLA-DRA) (Figures 5F and 5H left column, Figure S8B left column, and Table S3). Proinflammatory signatures, on the other hand, were downregulated with pregravid obesity (Figure 5G and Table S3). This included cytokine signaling pathways (CCL4, CXCL8, TNF, NFKB1, REL), interferon-stimulated genes (ISG15, IFITM3), metabolic genes (COX6A1, COX6B1), and genes important for T cell activation (CD55, CD300E) (Figures 5G and 5H right column, and Figure S8B right column).

To validate the transcriptional changes described above, we assessed the protein expression of inflammatory mediators and markers using flow cytometry. While frequencies of TREM1+, MIP-1β+, and CXCL8+ decidual macrophages were reduced with pregravid obesity (Figure 5I), the frequency of IL-1RA+ macrophages increased (Figure 5I). Additionally, spontaneous production of several proinflammatory cytokines (TNFα, IL-1β, IL-12p70) was attenuated within sorted HLA-DR^high macrophages (Figures 5J and S8C) in the absence of changes in baseline levels of activation-associated signaling molecules (Figure S8D). To define the biological significance of the downregulation of genes important for antimicrobial function with pregravid obesity (Figure 5G), we measured the frequency of cytokine-producing decidual macrophages following stimulation with a cocktail of bacterial TLR ligands (activating TLRs 1/2/4/6) or E. coli using flow cytometry (Figures 5K and S8E). Pregravid obesity was associated with reduced frequency of TNFα+IL-6+IL-1β+ macrophages following both stimulation conditions within both HLA-DR^high (Figure 5L) and HLA-DR^low subsets (Figure S8F).

DISCUSSION

Roughly 30%–40% of decidual cells are immune cells that play a critical role in the maintenance of fetomaternal homeostasis.^2,43 The phenotype and frequency of immune cells in the decidua dynamically adapt to fetal needs,^44,45 either by proliferating locally or via recruitment from circulation.⁴⁶ These time-sensitive changes support fetal growth and tolerance, protection from infections, and ultimately promote parturition.^43,47 Recent advances in single-cell technologies have facilitated the unbiased profiling of rare populations of immune cells residing in the early decidua.^9,48 While the transcriptional profiles of placental trophoblast populations from chorionic villi in homeostasis^49,50 and in preterm labor have been recently investigated,⁵¹ an atlas of immune cell diversity of human decidua at term is still lacking. The availability of such a high-resolution map of decidual immune cells collected in the absence of pregnancy complications would pave the way for future research on the role of decidual T cells, NK cells, and macrophages in the pathophysiology of adverse outcomes such as preeclampsia, spontaneous abortion, and preterm labor. We, therefore, profiled the immune compartment of term decidua isolated from mothers undergoing scheduled cesarean delivery to generate a detailed map of transcriptomic, phenotypic, and functional diversity of term decidual T cells, NK cells, and macrophages. Finally, given the strong association between maternal pre-pregnancy BMI and pregnancy complications,^3–7 we investigated the impact of pregravid obesity on the functional rewiring of decidual immune cells.

The first-trimester decidua is populated by NK cells (dNKs),^52,53 which account for about 70% of decidual leukocytes that are recruited from blood by factors such as IL-15 and progesterone.^54,55 Early dNKs facilitate embryonic development and promote decidual tolerance to the embryo. Furthermore, animal studies have shown the critical role of dNK-derived factors (vascular endothelial growth factor C [VEGF-C], TGF-β, IFNγ) in uterine vascular modifications and extravillous trophoblast (EVT) differentiation.^56,57 scRNA-seq and mass cytometry analyses of first-trimester decidua^9,10 showed three subpopulations of dNKs. Of the three populations, dNK1s (Eomes+) are the most abundant, expressing granzymes, KIR inhibitory receptors, and LILRB1 suggesting their preferential interactions with EVTs. This subset of dNKs produce growth factors such as pleiotrophin and osteoglycin, promoting fetal growth in both mice and humans,⁵⁸ suggesting their critical role in early pregnancy. On the other hand, mass cytometry analyses suggest that dNK3s are predominantly Tbet+ and mounted the strongest polyfunctional (CD107a, MIP-1α, MIP-1β, XCL1, IFNγ) response to ex vivo PMA stimulation.¹⁰ Interestingly, these populations have also been described in the endometrium (uterine NKs [uNKs]) of non-pregnant women.⁵⁹ Our analyses show a sharp drop in total dNK proportions at term, with a substantial shift toward dNK2 and dNK3 cells, as recently described in NK cells within the uterine mucosa (uNKs).⁵⁹ This reduction is in line with the reduced need for vascular remodeling and a shift toward fetal growth and enhanced immune activation associated with labor in the third trimester.⁶⁰ It is known that dNKs at term are transcriptionally distinct compared with their first-trimester counterparts and exhibit a lower capacity to respond to cytomegalovirus (CMV).⁶¹ Furthermore, dNKs become less granular over the course of a healthy pregnancy.⁶² However, their transcriptional diversity and activation status in response to labor, parturition, and adverse pregnancy outcomes associated with late-stage pregnancies is still unclear.

Reduction and redistribution of dNK cell subsets at term was associated with an expansion of T cells, potentially indicating a shift from tissue remodeling to host defense at the maternal-fetal interface. These observations are in line with recent studies that have used imaging mass cytometry to analyze immune cells in decidua in situ.⁶³ T cells are recruited into the decidua from the blood by trophoblast-derived CXCL16.⁶⁴ Tissue-resident memory (TRM, CD69+) CD8 T cells are present in term decidua, and a subset of these cells expressed CD103.^65–67 TRM T cells have been reported to express high levels of PD-1 but low levels of effector molecules such as GZMB and perforin compared with their blood counterparts, in line with their ability to trigger trophoblast antigen-specific tolerance.^68,69 We also observed increased PD-1 expression on decidual CD4 T cells; however, these TRMs were predominantly CD103–. Aberrant Th1 responses in placental CD4 T cells have been associated with pathologies such as recurrent spontaneous abortions.⁷⁰ Our data indicate reduced ability of decidual CD8 T cells to produce IFNγ, MIP-1β, and TNFα compared with circulating CD8 T cells. While decidual and circulating CD4 T cell IFNγ responses were comparable, production of MIP-1β and TNFα were lower in decidual CD4 T cells compared with their blood counterparts. Th17 responses were significantly higher in the decidua. Both Th1 and Th17 responses at the maternal-fetal interface have been shown to increase over the course of pregnancy and play a role in the initiation of labor.⁷¹ Finally, CD4 T cells generated higher TGF-β responses in line with the increased frequency of Tregs, which play an important role in maintaining fetal tolerance and preventing miscarriages.⁷²

This study compared the clonality of matched decidual and blood T cells. In all four donors, we observed that the decidual T cells repertoire was more restricted. However, in all four donors, the dominant decidual T cell clones were also detected in the blood, suggesting their recruitment from circulation. The process by which these specific clones are recruited and retained in the decidua, as well as their antigen specificity, remains poorly understood. Studies have shown that decidual CD4 and CD8 T cells can recognize viral antigens during Zika and dengue infections⁷³ and that frequency of Epstein-Barr virus and CMV-specific CD8 T cells is greater in the decidua compared with blood.⁷⁴ These findings suggest that decidual T cells can mediate antimicrobial responses.

Compared with decidual lymphocytes, macrophage frequencies modestly decreased at term relative to first-trimester decidua. A large macrophage population with the near absence of DCs suggests their potential role in antigen presentation at the maternal-fetal interface. Additionally, these cells play a critical role in regulating T cell responses, tissue remodeling, and clearance of apoptotic cells. Two distinct subsets of macrophages have been defined in decidua during early gestation—characterized by differential surface expression of CD11c and CCR2.⁷⁵ Gene expression profiling of these subsets suggested that the CD11c^low subset plays a role in tissue homeostasis while CD11c^high is important for antigen presentation.⁷⁶ However, neither subset conformed with the in vitro derived IFNγ or IL-4 conditioned macrophages, suggesting an underappreciation of their plasticity at the maternal-fetal interface. Using unbiased scRNA-seq of matched term decidual macrophages and blood monocytes, we report two major macrophage subsets characterized by co-expression of HLA-DR and CD11c as described for first-trimester decidua.⁷⁶

Trajectory analysis, flow cytometry, and profiling of secreted cytokines at baseline showed that the HLA-DR^low macrophages are an inflammatory subset that are transcriptionally similar to blood monocytes. On the other hand, despite high expression level of regulatory molecules (TREM2, CD163, CD206), HLA-DR^high macrophages exhibited an enhanced capacity for TLR4-mediated responses and phagocytosis. Furthermore, HLA-DR^high macrophages also accumulated higher levels of lipids and are metabolically primed for oxidative phosphorylation, supporting their identity as bona fide tissue-resident macrophages, as previously described in other compartments.⁷⁷ Within the decidua, these macrophages have been shown to be more likely to be located within the vicinity of T cells at term.⁶³ Further characterization of the heterogeneity within this subset showed three subpopulations predicted to play a role in acute inflammatory responses and killing (S100A8+ CD9−), response to wounding (S100A8+ CD9+), and apoptosis and fat-cell differentiation (S100A8− FOLR2+). Therefore, decidual macrophages span the spectrum from inflammatory blood-derived HLA-DR^low cells to HLA-DR^high FOLR2+ regulatory cells involved in tissue remodeling and clearance of cellular debris.

A consistent theme across both decidual lymphocyte and macrophage populations was a heightened state of immune activation at term compared with the first trimester. This is evident from increased levels of inflammatory cytokine transcripts (IL1B, IFNΓ, TNF) across dNK, T cell, and macrophage subsets and higher levels of spontaneous secretion of cytokines and chemokines by sorted macrophage subsets compared with blood monocytes. We posit that this state of activation is important for parturition⁷⁸ and is in line with recent studies highlighting progressive peripheral immune activation associated with gestation.^79,80 Understanding these pregnancy-associated shifts in immune states is therefore critical in addressing the immunological basis of pregnancy complications.

Our analyses suggest a dramatic drop in CD4 and CD8 T cell numbers and concomitant increases in macrophage frequencies with pregravid obesity driven in a BMI-dependent manner by increased frequency of the HLA-DR^high macrophage subset. Despite the overall loss of T cells, we report no preferential loss of any major T cell subset or functional loss in cytokine responses. Regulatory T cell frequencies, however, increased with obesity. In contrast, macrophage responses were significantly attenuated—with the downregulation of genes involved in pathogen sensing and cytokine signaling. Functionally, this translated to reduced levels of secreted inflammatory cytokines at baseline (TNFα, IL-1β) but also poor responses to ex vivo E. coli stimulation. We have previously demonstrated that pregravid obesity alters activation in peripheral monocytes at term, resulting in dampened ex vivo cytokine responses to TLR4 stimulation.²² These observations suggest continuous recruitment of circulation monocytes to the decidua.^22,81–83

These observations align with previous reports of reduced CD163-macrophages in decidua parietalis of mothers with obesity.⁸⁴ Taken together, these findings highlight an establishment of a tolerance-like program in term decidual macrophages with maternal obesity, mirroring the phenotype observed in peripheral monocytes reported in our previous study.²² Additionally, we posit that the accumulation of regulatory macrophages and Tregs with obesity is mediated by changes in the decidual microenvironment (cytokine/chemokine changes, stromal cell crosstalk),⁸⁵ to protect the developing fetus from the undesirable effects of maternal inflammation. Suboptimal activation and attenuated responses can have untoward consequences for both host defense and initiation of labor. These changes could also contribute to obesity-associated pathologies during the late stages of pregnancy.

Limitations of the study

Our study is limited by restricting analyses to decidua collected following cesarean deliveries and the lack of investigating dietary contributions to maternal obesity-associated decidual adaptations. Moreover, while pregravid obesity has been shown to attenuate cord blood monocyte and T cell responses, its influence on fetal macrophage populations within the chorionic villi is still unknown. Future studies need to focus on identifying key immunological changes occurring in the placenta stratified by mode of delivery. Animal models of maternal obesity could be leveraged to fully elucidate longitudinal changes in vascularization, tissue remodeling, and immune activation in the decidua.

STAR★METHODS

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Dr. Ilhem Messaoudi (ilhem.messaoudi@uky.edu).

Materials availability

This study did not generate new unique reagents.

Data and code availability

• The datasets supporting the conclusions of this article are available on NCBI’s Sequence Read Archive and project ID’s are listed in the key resources table.

• This paper does not report original code.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
CD3-FITC	BD Biosciences	Cat#556611; RRID:AB_396484 Clone-SP34
CD20-BV510	BioLegend	Cat#302340; RRID:AB_2561941 Clone-2H7
CD56-BV711	BioLegend	Cat#318336; RRID:AB_2562417 Clone-HCD56
CD16-PB	BioLegend	Cat#302021; RRID:AB_492978 Clone-3G8
CD14-AF700	BioLegend	Cat#301822; RRID:AB_493747 Clone-M5E2
HLA-DR-APC-Cy7	BioLegend	Cat#307618; RRID:AB_493586 Clone-L243
CD11c-APC	BioLegend	Cat#301614; RRID:AB_493023 Clone-3.9
CD123-PCP-Cy5.5	BioLegend	Cat#306016; RRID:AB_2264693 Clone-6H6
CD86-BV605	BioLegend	Cat#305430; RRID:AB_2563824 Clone-IT2.2
CD4-FITC	BioLegend	Cat#300538; RRID:AB_2562052 Clone-RPA-T4
CD4-PE	BioLegend	Cat#317410; RRID:AB_571955 Clone-OKT4
CD8b-ECD	Beckman Coulter	Cat#6607123; RRID:AB_1575983 Clone-2ST8.5H7
CD45RA-PerCP-Cy5.5	Thermo Fisher Scientific	Cat#45-0458-42; RRID:AB_10718536 Clone-HI100
CCR7-PE-Cy7	BioLegend	Cat#353226; RRID:AB_11126145 Clone-GO43H7
CD69-PE	BioLegend	Cat#310906; RRID:AB_314841 Clone-FN50
CD103-BV605	BioLegend	Cat#350218; RRID:AB_2564283 Clone-BER Act1
CTLA4-PeDazzle594	BioLegend	Cat#369616; RRID:AB_2632878 Clone-BNI3
CD25-APC-Cy7	BioLegend	Cat#302614; RRID:AB_314284 Clone-BC96
FOXP3-A488	BioLegend	Cat#320106; RRID:AB_439752 Clone-206D
CD49a-PE-Cy7	BioLegend	Cat#328312; RRID:AB_2566272 Clone-TS2/7
CD39-BV510	BioLegend	Cat#328219; RRID:AB_2563265 Clone-A1
ITGB2 (CD11a/CD18)-PCP-Cy5.5	BioLegend	Cat#302120; RRID:AB_2565587 Clone-TS1/18
KIR2DL4 (CD158d)-APC	BioLegend	Cat#347008; RRID:AB_2130691 Clone-mAb33
KIR2DL1 (CD158a)-APC-Cy7	Novus Biologicals	Cat#NB100-63267APCCY7; RRID:AB_2941073 Clone-NKVFS1
KIR2DL2/3 (CD158b)-PE	BD Biosciences	Cat#559785; RRID:AB_397326 Clone-CH-L
CD160-AF700	Novus Biologicals	Cat#NBP2-90074AF700; RRID:AB_2941072 Clone-275
NKG2A-PE-Dazzle594	BioLegend	Cat#375121; RRID:AB_2888869 Clone-S19004C
NKG2C-PE	BioLegend	Cat#375003; RRID:AB_2888871 Clone-S19005E
CD9-PE-Cy7	BioLegend	Cat#312116; RRID:AB_2728256 Clone-HI9a
CD68-APC	BioLegend	Cat#333810; RRID:AB_2275735 Clone-Y1/82A
TLR4-BV711	BD Biosciences	Cat#564404; RRID:AB_2738794 Clone-tf901
CD163-APC/Fire750	BioLegend	Cat#333634; RRID:AB_2734333 Clone-GHI/61
CD206-BV510	BioLegend	Cat#321138; RRID:AB_2721530 Clone-15-2
CD209-FITC	BioLegend	Cat#330104; RRID:AB_1134048 Clone-9e9a8
TREM1-AF405	R&D Systems	Cat#FAB12781V; RRID:AB_2941074 Clone-888111
TREM2-PerCP	R&D Systems	Cat#FAB17291C; RRID:AB_2941075 Clone-237920
CD64-BV711	BioLegend	Cat#305042; RRID:AB_2800778 Clone-10.1
CD64-APC	BioLegend	Cat#305014; RRID:AB_1595428 Clone-10.1
GLUT1-AF488	R&D Systems	Cat#FAB1418G; RRID:AB_2941076 Clone-202915
FOLR2-PE	BioLegend	Cat#391704; RRID:AB_2721336 Clone-94b/folr2
CCR2-BV605	BioLegend	Cat#357213; RRID:AB_2562702 Clone-K036C2
CD19-PE	BioLegend	Cat#302208; RRID:AB_314238 Clone-HIB19
S100A9-APC	Invitrogen	Cat#MA5-28129; RRID:AB_2745112 Clone-10.1
CD107a-FITC	BD Biosciences	Cat#555800; RRID:AB_396134 Clone-H4A3
PD-1-BV510	BioLegend	Cat#329932; RRID:AB_2562256 Clone-Eh12.2h7
CD45-FITC	BioLegend	Cat#304038; RRID:AB_2562050 Clone-HI30
TNFα-APC	BioLegend	Cat#502912; RRID:AB_315264 Clone-Mab11
IL-6-FITC	BD Biosciences	Cat#554696; RRID:AB_395514 Clone-MQ2-6A3
IL-1β-PB	BioLegend	Cat#511710; RRID:AB_2124350 Clone-H1b-98
IFNγ-APC	Invitrogen	Cat#17-7319-82; RRID:AB_469506 Clone-4S.B3
IFNγ-PE-Cy7	eBioscience	Cat#502527; RRID:AB_1626154 Clone-4S.B3
IL-2-AF700	BioLegend	Cat#500320; RRID:AB_528929 Clone-MQ-17H12
IL-4-APC	Tonbo Biosciences	Cat#50-210-2786; RRID:AB_2941077 Clone-MP425D2
IL-17-FITC	Invitrogen	Cat#11-717942; RRID:AB_10805390 Clone-Ebio64dec17
TGF-β-PerCP-Cy5.5	BioLegend	Cat#141410; RRID:AB_2561592 Clone-TW7-16B4
MIP-1β-PE	BD Biosciences	Cat#550078; RRID:AB_393549 Clone-D21-1351
GZMB-AF700	BD Biosciences	Cat#560213; RRID:AB_1645453 Clone-GB11
NF-κB p50-AF488	Luminex	Cat#4700-1674; RRID:AB_2941078
7-Amino-Actinomycin D (7-AAD)	Luminex	Cat#400-0290; RRID:AB_2941079
NF-κB p65 (pS529)-AF647	BD Biosciences	Cat#558422; RRID:AB_647136 Clone-K10-895.12.50
IκBa-PE	eBioscience	Cat#12-903642; RRID:AB_2572683 Clone-MFRDTRK
Phospho-p38 MAPK-APC	eBiosceince	Cat#17-9078-42; RRID:AB_2573290 Clone-4NIT4KK
CD3-PE	BD Biosciences	Cat#556612; RRID:AB_396485 Clone-SP34
TotalSeq-C0072-CD4	BioLegend	Cat#300567; RRID:AB_2800725 Clone-RPA-T4
TotalSeq-C0046-CD8	BioLegend	Cat#344753; RRID:AB_2800922 Clone-SK1
TotalSeq-C0148-CCR7	BioLegend	Cat#353251; RRID:AB_2800943 Clone-G043H7
TotalSeq-C0063-CD45RA	BioLegend	Cat#304163; RRID:AB_2800764 Clone-HI100
TotalSeq-C0146-CD69	BioLegend	Cat#310951; RRID:AB_2800810 Clone-FN50
TotalSeq-C0145-CD103	BioLegend	Cat#350233; RRID:AB_2800933 Clone-Ber-ACT8
TotalSeq-C0088-PD-1	BioLegend	Cat#329963; RRID:AB_2800862 Clone-EH12.2H7
TotalSeq-C0085-CD25	BioLegend	Cat#302649; RRID:AB_2800745 Clone-BC96
TotalSeq-C0251	BioLegend	Cat#394661; RRID:AB_2801031 Clone-LNH-94, 2M2
TotalSeq-C0254	BioLegend	Cat#394667; RRID:AB_2801034 Clone-LNH-94, 2M2
TotalSeq-C0256	BioLegend	Cat#394671; RRID:AB_2820042 Clone-LNH-94, 2M2
TotalSeq-C0260	BioLegend	Cat#394679; RRID:AB_2820046 Clone-LNH-94, 2M2
Biological Samples
Fetal Bovine Serum, USDA Certified, Heat Inactivated	Omega Scientific	Cat#FB-02
FetalPlex™ Animal Serum Complex	GeminiBioProducts	Cat#100-602
Chemicals, Peptides, and Recombinant Proteins
RPMI 1640	VWR	Cat#45000-396
Hanks Balanced Salt Solution (HBSS)	VWR	Cat#21-023-CV
Penicillin Streptomycin	GeminiBio	Cat#400109
L-Glutamine	GeminiBio	Cat#400106
N-2-hydroxyethylpiperazine-N′2-ethanesulfonic acid HEPES	ThermoFisher	Cat#15630080
Collagenase	Sigma	Cat# C9722 Source: Clostridium histolyticum
Ammonium chloride - NH4Cl (RBC Lysis Buffer)	Sigma	Cat#A9434
Sodium hydrogen carbonate - CHNaO3 (RBC Lysis Buffer)	Fisher	Cat#AAA170050E
EDTA (RBC Lysis Buffer)	Invitrogen	Cat#AM9260G
Percoll Density Gradient	Neta Scientific	Cat# 17-0891-01
Ficoll-Paque	GE Healthcare	Cat# 17144003
Dimethyl sulfoxide (DMSO)	Sigma	Cat# D2650-100ML
FOXP3 Fix/Perm Buffer Set	BioLegend	Cat#421403
Ghost Dye 510	TONBO Biosciences	Cat#13-0870-T 100
Ghost Dye 540	TONBO Biosciences	Cat#13-0879
SYTOX Blue Dead Cell Stain	ThermoFisher Scientific	Cat#S34857
Human TruStain FcX	BioLegend	Cat#422302
Human True-Stain Monocyte Blocker	BioLegend	Cat#426103
Lipopolysaccharide (LPS) TLR4 ligand E. coli 055:B5	Invivogen	Cat#tlrl-b5lps
Pam3CSK4 TLR1/2 agonist	Invivogen	Cat#tlrl-pm2s-1
FSL-1 TLR2/6 agonist	Sigma	Cat#SML1420
E. coli	Migula. Castellani and Chalmers ATCC 11775	Cat#1132342
Phorbol myristate acetate-NF-κB Activator (PMA)	Invivogen	Cat#tlrl-pma
Ionomycin	Invivogen	Cat#inh-ion
Brefeldin A (BFA)	BioLegend	Ca#420601
Fixation buffer	BioLegend	Cat#420801
Amnis NF-κB translocation kit	Luminex Corporation	Cat #ACS10000
Permeabilization wash buffer	BioLegend	Cat#421002
Bovine Serum Albumin (BSA), Fraction V— Molecular Biology Grade	GeminiBio	Cat#700-106P
Cytofix fixation buffer	BD Biosciences	Cat#554655
BD Perm/Wash buffer	BD Biosciences	Cat#554723
Critical Commercial Assays
BODIPY 493/503	ThermoFisher Scientific	Cat#D3922
pHrodo Deep Red E. coli BioParticle Conjugates	ThermoFisher Scientific	Cat#P35360
CellROX Deep Red Flow Cytometry Assay Kit	ThermoFisher Scientific	Cat#C10491
2-NBDG	ThermoFisher Scientific	Cat#N13195
MitoTracker Red CMXRos	ThermoFisher Scientific	Cat#M7512
Seahorse XF Cell Energy Phenotype kit	Agilent Technologies	Cat#103592-100
Human Premixed 29-plex Magnetic Luminex Assay	R&D Systems	Cat#Custom-LxSA-H-29 Lot#C0003514
TotalSeq B Hashtag Oligos 1-10	BioLegend	384631-392649
Chromium Single Cell 3’ Reagent Kits v3	10X Genomics	PN-1000075
3′ Feature Barcode Kit	10X Genomics	PN-1000262
Chromium Single Cell 5’ Reagent Kits V2	10X Genomics	PN-1000263
5′ Feature Barcode Kit	10X Genomics	PN-1000256
Deposited Data
3′ Single Cell GEX data - Decidua	This paper	NCBI Sequence Read Archive: PRJNA817521
5′ Single Cell TCR data - Decidua	This paper	NCBI Sequence Read Archive: PRJNA817521
5′ Single Cell TCR data - Maternal PBMC	This paper	NCBI Sequence Read Archive: PRJNA817521
3′ Single Cell GEX data -Villous	This paper	NCBI Sequence Read Archive:PRJNA946160
3′ Single Cell GEX data - UCBMC	Sureshchandra et al. 202166 10.3389/fimmu.2021.617592	NCBI Sequence Read Archive:PRJNA690128
3′ Single Cell GEX data - Maternal PBMC	This paper and Sureshchandra etal. 202166 10.1016/j.isci.2021.102690	NCBI Sequence Read Archive:PRJNA887020
Software and Algorithms
Prism	GraphPad	Version#8
Immunarch	R package	Version#0.6.7
Metascape	www.metascape.org	N/A
Attune NxT Flow Cytometer Software	ThermoFisher Scientific	Version#2.5
FlowJo	TreeStar	Version#10.5
Ideas Analysis Software	Luminex Corporation	Version#6.1
Seahorse Wave	Agilent Technologies	Version#2.6.1
xPONENT® Software for Luminex Instruments	Luminex Corporation	Version#4.3
Cell Ranger Single-Cell Software Suite	10X Genomics	Version 6.0.2
Seurat	R package	Version 3.1.5
Monocle	R package	Version 2.8.0

EXPERIMENTAL MODEL AND SUBJECT PARTICIPANT DETAILS

This study was approved by the Institutional Ethics Review Board of Oregon Health & Science University and the University of California, Irvine. A total of 102 non-smoking women (47 lean, 12 overweight, and 43 obese) who had an uncomplicated pregnancy were enrolled for this study. Cohort characteristics are outlined in Table 1. Written informed consent was obtained from all participants. Due to the strong positive correlation between pre-pregnancy BMI and total body fat in our previous studies,⁸⁶ we used pre-pregnancy BMI as a surrogate for maternal pregravid obesity and stratified participants into lean or obese based on pre-pregnancy BMI (Table 1). Participants Only decidua from patients who met study inclusion criteria undergoing scheduled cesarean deliveries for a variety of medical and elective indications, including repeat cesareans, breech presentation, maternal request, and medical contra-indication to vaginal birth. There were no indications of complications for any of the study participants. The decision to use decidua obtained via cesarean section was made to avoid confounders associated with delivery (length of labor, methods of pain control, etc.). Exclusion criteria include active maternal infection, documented fetal congenital anomalies, substance use disorder, chronic illness requiring regular medication use, preeclampsia, gestational diabetes, chorioamnionitis, and significant medical conditions (active cancers, cardiac, renal, hepatic, or pulmonary diseases).

Table 1.

Cohort characteristics

	Lean	Overweight	Obese
Enrollment	47	12	43
BMI (kg/m2)a	21.9 (1.7)	27.4 (1.6)	38 (7.9)
Maternal age (years)	33.7 (4.6)	34.7 (4.71)	32.4 (4.5)
Gestational age (weeks)	38.8 (0.7)	39.02 (0.7)	38.3 (0.9)
%Female	55.8	33.3	33.3

Avg (SEM).

^ap ≤ 0.0001.

METHOD DETAILS

Sample collection and processing

Placental biopsies were taken from all four quadrants of the placenta and near the umbilical cord. Decidua basalis was separated from chorionic villi immediately following delivery by experienced personnel from the dedicated maternal-fetal medicine research team at Oregon Health & Science University, Portland OR, who undergo specialized placenta collection training. Separated tissue was immersed in RPMI supplemented with 10% FBS and antibiotics then shipped overnight for processing by the Messaoudi lab. Tissues were washed in HBSS to remove contaminating blood and any visible blood vessels macroscopically separated and then washed for 10 min in HBSS before processing. Tissues were minced into approximately 0.2–0.3 mm³ cubes and enzymatically digested in 0.5 mg/mL collagenase V (Sigma, St. Louis, MO, C-9722) solution in 50 mL R3 media (RPMI 1640 with 3% FBS, 1% Penicillin-Streptomycin, 1% L-glutamine, and 1M HEPES) at 37°C for 1 h. The disaggregated cell suspension was passed through cell strainers to eliminate visible chunks of fat/tissue. Cells were pelleted from the filtrate, passed through a 70-μm cell sieve, centrifuged, and resuspended in R3 media. Red blood cells were lysed using RBC lysis buffer (155 mM NH₄Cl, 12 mM NaHCO₃, 0.1 mM EDTA in double-distilled water) and resuspended in 5 mL R3. The cell suspension was then layered on a discontinuous 60% and 40% percoll gradient and centrifuged for 60 min with the brakes off. Immune cells at the interface of 40% and 60% gradients were collected, washed in HBSS, counted, and cryopreserved for future analysis.

Blood processing

Peripheral blood collected before cesarean section was layered on a Ficoll-Paque (GE HealthCare Technologies Inc, Chicago, IL) gradient and centrifuged at 2000 rpm for 30 min without brakes. Peripheral Blood Mononuclear cells (PBMC) were resuspended in FetalPlex (Gemini BioProducts, West Sacramento, CA) with 10% DMSO (Sigma, St. Louis, MO) and stored in liquid nitrogen for future analysis. Plasma samples were stored at −80°C until they were used.

Immunophenotyping

For broad phenotyping of immune cells within term decidua, 1–2 X 10⁶ fresh decidual leukocytes were washed with PBS and stained using the following cocktail of antibodies: CD45, CD20, CD4 (BioLegend, San Diego, CA), CD8b (Beckman Coulter, Brea, CA), CD14, HLA-DR, CD56, CD16, CD11c, CD123 (BioLegend, San Diego, CA), for 30 min in the dark at 4°C.

For T cell phenotyping, 0.5 X 10⁶ freshly thawed decidual leukocytes were stained using the following panel of surface antibodies: CD8 (Beckman Coulter, Brea, CA), CD4, CD45RA, CCR7, CD69, CD103, CTLA4, CD25, and PD-1 (BioLegend, San Diego, CA) for 30 min at 4°C. Cells were then fixed and permeabilized for 30 min followed by intra-nuclear staining for FOXP3 for an additional 4 h using FOXP3 Fix/Perm Buffer set (BioLegend, San Diego, CA).

For identification of NK cell subsets, 0.5 X 10⁶ freshly thawed decidual leukocytes were stained using the following panel of antibodies: CD56, CD16, CD103, CD49a, CD39, ITGB2, KIR2DL4, NKG2A, NKG2C, CD9 (BioLegend, San Diego, CA), KIR2DL1, CD160 (Novus Biologicals, Englewood, CO), KIR2DL2/3 (BD Biosciences, San Jose, CA).

For a more comprehensive analysis of macrophage phenotypes, 0.5 X 10⁶ thawed decidual cells were stained using the following panel of antibodies CD14, HLA-DR, CD11c, CD16, CD68, CD86, CD163, CD206, CD209 (BioLegend, San Diego, CA), and TLR4 (BD Biosciences, San Jose, CA). A second aliquot of cells was surface stained with CD14, HLA-DR, CD64, CD9, FOLR2, CCR2 (BioLegend, San Diego, CA), TREM1, TREM2, and GLUT1 (R&D Systems, Minneapolis, MN), followed by fixation/permeabilization before staining intracellularly with TREM2 (R&D Systems, Minneapolis, MN) and S100A9 (ThermoFisher Scientific, Waltham, MA) for 4 h. Unstained and Fluorescence minus one (FMO) controls were included for all major macrophage markers.

All surface staining was done in the presence of Human Fc block and Monocyte Blocker (BioLegend, San Diego, CA). Following staining, all samples were washed twice in FACS buffer and acquired with the Attune NxT Flow Cytometer (ThermoFisher Scientific, Waltham, MA), immediately after the addition of SYTOX Live Dead Cell Stain (ThermoFisher Scientific, Waltham MA) for unfixed samples. Data were analyzed using FlowJo 10.5 (Beckton Dickinson, Ashland, OR).

Ex vivo stimulation and intracellular cytokine staining

To measure cytokine responses of decidua macrophages, 10⁶ thawed cells were stimulated for 6h at 37°C in RPMI supplemented with 10% FBS in the presence or absence of bacterial TLR cocktail containing 1 μg/mL LPS (TLR4 ligand, E.coli 055:B5; Invivogen, San Diego, CA), 2 μg/mL Pam3CSK4 (TLR1/2 agonist, Invivogen), and 1 μg/mL FSL-1 (TLR2/6 agonist, Sigma, St. Louis, MO), or 6 X 10⁵ cfu/mL E. coli (Escherichia coli (Migula) Castellani and Chalmers ATCC 11775).

For NK and T cell responses, 10⁶ thawed cells were stimulated for 4h at 37°C in RPMI supplemented with 10% FBS in the presence or absence of 500 ng PMA and 1 μg Ionomycin (Sigma, St. Louis, MO). For assessment of NK cell degranulation, CD107a (BD Biosciences, San Jose, CA) was added during stimulation.

For all samples, Brefeldin A (Sigma, St. Louis, MO) was added after 1-h incubation and cell were cultured for an additional 5 h before surface and intracellular cytokine staining.

Cells were stained with the following surface antibodies: [1] CD14 and HLA-DR for macrophages; [2] CD4 and CD8 for T-cells; and [3] CD56, CD16, CD103, CD49a, CD39, and ITGB2 (BioLegend, San Diego, CA) for NK cells for 30 min in the dark at 4°C. Samples were fixed and permeabilized using fixation and permeabilization wash buffer (BioLegend, San Diego, CA) at 4°C for 20 min and stained intracellularly overnight for TNFα and IL-6, and IL-1β for macrophages, IFNγ, TNFα, IL-2, IL-4, IL-17, TGF-β and MIP-1β for T-cells and IFNγ, GZMB, and MIP-1β for NK cells. Cells were then washed and acquired using the Attune NxT Flow Cytometer (ThermoFisher Scientific, Waltham MA) and analyzed using FlowJo 10.5 (BD, Ashland OR).

Imaging flow cytometry

For visualization of macrophage subsets and nuclear translocation of NF-κB p50, decidual leukocytes were thawed, then 500,000 cells were washed and stained for dead cell exclusion (Ghost Violet 510, 1:1000 dilution) for 30 min at 4°C. Cells were then surface stained (CD14, CD11c, HLA-DR) and fixed with 1X fixation buffer (Luminex Corporation, Austin, TX) for 10 min at room temperature (RT) before the addition of p50-AF488 (1:20 dilution) in 1X Assay Buffer for 30 min at RT in the dark. At the end of the incubation, cells were washed twice in 1X Assay buffer and resuspended in 50 μL 0.25X Fixation Buffer in polypropylene Eppendorf tubes. After the addition of nuclear dye (7-AAD, 1:50), samples were run on Amnis ImageStream XMark II Imaging flow cytometer (Luminex Corporation, Austin, TX) and data were analyzed on Ideas Analysis Software (Luminex Corporation, Austin, TX).

Phospho flow

Decidual leukocytes were thawed, then washed with FACS buffer and surface stained for CD14 and HLA-DR in FACS tubes. Pellets were washed in FACS buffer and resuspended in 100 μL prewarmed PBS (Ca+ Mg+ free). Cells were fixed immediately by the addition of equal volumes of prewarmed Cytofix Buffer (BD Biosciences) and thorough mixing and incubating at 37C for 10 min. Cells were then centrifuged at 600g for 8 min. Supernatants were removed leaving no more than 50 μL residual volume. Cells were then permeabilized by the addition of 1 mL 1X BD PermWash Buffer I, mixed well, and incubated at RT for 30 min. Pellets were then spun, aspirated, and stained intranuclearly with antibodies against NF-κB p65 (pS529) AF647 (Clone K10–895.12.50, Cell Signaling Technology) or IκBa PE (Clone MFRDTRK, eBioscience, San Diego, CA) and Phospho-p38 MAPK-APC (Clone 4NIT4KK, eBioscience) for 1h at RT in the dark. Samples were washed twice in Permwash Buffer I, resuspended in FACS buffer and acquired on the Attune NxT flow cytometer.

BODIPY staining

Decidual leukocytes were thawed, then 500,000 cells surface stained (CD14, HLA-DR) for 20 min at 4°C, washed twice, and resuspended in 500 μL warm 1X PBS containing 1 μg/mL BODIPY 493/503 (ThermoFisher Scientific, Waltham, MA). Cells were incubated at 37°C for 10 min and acquired on the Attune NxT flow cytometer.

Phagocytosis assay

Decidual leukocytes were thawed, then 500,000 cells were incubated for 2 h at 37°C in media containing pH-sensitive pHrodo E. coli BioParticles conjugates (ThermoFisher Scientific, Waltham, MA) (1 mg/mL concentration). Pellets were washed twice, surface stained (CD14, HLA-DR), and resuspended in ice-cold FACS buffer. Samples were analyzed using flow cytometry with no pHrodo controls.

Cytosolic ROS assay

Decidual leukocytes were thawed, then 500,000 cells were incubated with 2.5 μM CellROX Deep Red (Life Technology, Carlsbad, CA) at 37°C for 30 min. For negative control, cells were incubated in serum-free media containing 200 μM antioxidant N-acetylcysteine (NAC) for 1.5 h. Both negative and positive controls were incubated with tert-butyl hydroperoxide (TBHP) for 30 min to induce oxidative stress. All samples were then surface stained (CD14, HLA-DR) and analyzed using flow cytometry.

Glucose uptake assay

Decidual leukocytes were thawed, then 500,000 cells were resuspended in glucose-free media and incubated with 60 mM NBDG (ThermoFisher, Waltham MA), a fluorescent glucose analog for 30 min at 37°C. Samples were washed in 1X PBS, surface stained (CD14, HLA-DR), and median fluorescence of NBDG recorded using flow cytometry.

MitoTracker analysis

Decidual leukocytes were thawed, then 500,000 cells were resuspended in 100μL of 1X MitoTracker Red CMXRos (ThermoFisher, Waltham MA) (1:1000 of stock) in 96 well plates, incubated at 37°C for 30 min, washed, surface stained (CD14, HLA-DR), and median fluorescence of MitoTracker recorded using flow cytometry.

Seahorse assay

Oxygen Consumption Rate (OCR) and Extracellular Acidification Rate (ECAR) were measured using Seahorse XF Cell Energy Phenotype kit on Seahorse XFp Flux Analyzer (Agilent Technologies, Santa Clara, CA) following the manufacturer’s instructions. Briefly, decidual leukocytes were thawed, then macrophages were sorted (pooled n = 3/group) ~200,000 cells were seeded on Cell-Tak (Corning, Corning, NY) coated 8-well culture plates in phenol-free RPMI media containing 2 mM L-glutamine, 10 mM L-glucose, and 1mM sodium pyruvate. Seeded plates were placed in 37°C incubator without CO₂ then run on the XFp with extended basal measurements, followed by injection of a stressor cocktail of 1 μg/mL LPS for 1 h in a 37°C incubator without CO₂. Plates were run on the XFp for 8 cycles of basal measurements, followed by acute injection of L-glucose (100 mM), oligomycin (ATP synthase inhibitor) carbonilcyanide p-triflouromethoxyphenylhydrazone (FCCP; a mitochondrial uncoupling agent).

Cytokine production

Decidual leukocytes and matched PBMCs were thawed. Live decidua macrophage subsets and peripheral blood monocytes (CD14+ HLA-DR^high and HLA-DR^low, and SYTOX Blue negative) were sorted using BD FACS Aria Fusion. 50,000-sorted cells were placed in a 96-well plate overnight to measure cytokine production. Supernatants were collected and stored at −80°C until further analysis using Luminex (Luminex Corporation, Austin, TX). Briefly, immune mediators in cell supernatants were measured using Human Custom 29-plex Multi-Analyte Kit (R&D Systems, Minneapolis, MN) measuring the following analytes: CCL2 (MCP-1), CCL4 (MIP-1β), CXCL9 (MIG), CXCL11 (I-TAC), GM-CSF, IFNγ, IL-1RA, IL-4, IL-7, IL-12p70, IL-18, PD-L1, S100B, VEGF, IL-15, CCL3, CCL11, CXCL10, CXCL13, IFNβ, IL-1β, IL-2, IL-6, CXCL8 (IL-8), IL-17A, IL-23, PDGF-BB, TNFα, and IL-10. Samples were processed per manufacturer’s instructions, recorded, and analyzed on the Magpix xPONENT system version 4.2 (Luminex Corporation, Austin, TX).

FACS and 3′ single cell RNA library preparation

Freshly thawed immune cells from decidua, villi, maternal, and cord blood were stained with CD45-FITC at 4°C in 1% FBS in DPBS without calcium and magnesium. Cells were washed twice and sorted on BD FACS Aria Fusion into RPMI (supplemented with 30% FBS) following the addition of SYTOX Blue stain for dead cell exclusion. Cells were then counted in triplicates on a TC20 Automated Cell Counter (BioRad, Hercules, CA), washed, and resuspended in PBS with 0.04% BSA in a final concentration of 1200 cells/μL. Single-cell suspensions were then immediately loaded on the 10x Genomics Chromium Controller with a loading target of 17,600 cells. Libraries were generated using the V3 chemistry per the manufacturer’s instructions (10x Genomics, Pleasanton CA). Libraries were sequenced on Illumina NovaSeq with a sequencing target of 50,000 reads per cell.

3′ single cell RNA-Seq data analysis

Raw reads were aligned and quantified using the Cell Ranger Single-Cell Software Suite (version 3.0.1, 10x Genomics) against the GRCh38 human reference genome internally running the STAR aligner. Downstream processing of aligned reads was performed using Seurat (version 3.1.5). Droplets with ambient RNA (cells with fewer than 400 detected genes), potential doublets (cells with more than 4000 detected genes and dying cells (cells with more than 20% total mitochondrial gene expression) were excluded during initial QC. Data objects from the lean and obese groups were integrated using Seurat.⁸⁷ Data normalization and variance stabilization were performed using SCTransform function⁸⁷ using a regularized negative binomial regression, correcting for differential effects of mitochondrial and ribosomal gene expression levels and cell cycle.

We integrated scRNA-Seq dataset with those from matched maternal PBMC using Seurat’s IntegrateData function to remove potential contaminating blood cells. Decidual cells clustering with PBMC were removed from downstream analyses. To rule out any infiltrating fetal Hofbauer cells or monocytes, data were integrated with matched CD45+ cells isolated from chorionic villi and mononuclear cells from umbilical cord blood.

Dimensionality reduction was performed using RunPCA function to obtain the first 30 principal components followed by clustering using the FindClusters function in Seurat. Clusters were visualized using the UMAP algorithm as implemented by Seurat’s runUMAP function. Cell types were assigned to individual clusters using FindMarkers function with a fold change cutoff of at least 0.4 (FDR ≤ 0.05) and using a known catalog of well-characterized scRNA markers for human PBMC,⁸⁸ tissue-resident lymphoid cells,⁹ and leukocytes in first-trimester human placentas.⁹

To study the evolution of the immune landscape of decidua with gestational age, we used the recently described atlas of first trimester (weeks 6–12) decidual CD45+ cells.⁹ Raw data were downloaded for 6 individuals (D6–10, D12) and analyzed as described above resulting in 24,849 cells. Cells were downsampled (6,116) to match the number of cells from term decidua of this study and integrated using Seurat’s IntegrateData function and markers compared using FindMarkers function.

Temporal trajectory was predicted using Monocle (version 2.8.0). Briefly, t-SNE was used for cell clustering. Monocle’s differentialGeneTest function was used to identify differential genes. Differential genes with a q-value less than 1e-10 were used to plot macrophages in a pseudotime scale. Metascape was used to identify Gene ontology (GO) terms.⁸⁹

5′ GEX and scTCR library preparation

Matched PBMC and decidual leukocytes were thawed, washed, filtered, and stained with Ghost Violet 540 live-dead stain (Tonbo Biosciences, San Diego, CA) for 30 min in the dark at 4°C. Given the parallel assessment of both TCR and gene expression, we used TotalSeq-C, an antibody-based multiplexing technology as it is the only reagent compatible with 5′ gene expression analysis. Samples were washed thoroughly with a cell staining buffer (1X PBS with 0.5% BSA), and Fc blocked for 10 min (Human TruStain FcX, BioLegend, San Diego, CA), and incubated with a cocktail containing CD3 (SP34, BD Pharmingen), 0.5 μg each of oligo tagged CD4 (TotalSeq-C0072, BioLegend, San Diego, CA), CD8 (TotalSeq-C0046, BioLegend, San Diego, CA), CCR7 (TotalSeq-C0148, BioLegend, San Diego, CA), CD45RA (TotalSeq-C0063, BioLegend, San Diego, CA), CD69 (TotalSeq-C0146, BioLegend, San Diego, CA), CD103 (TotalSeq-C0145, BioLegend, San Diego, CA), PD-1 (TotalSeq-C0088, BioLegend, San Diego, CA), CD25 (TotalSeqTM-C0085, BioLegend, San Diego, CA), and a unique hashing antibody (TotalSeq-C0251, C0254, C0256, or C0260, BioLegend, San Diego, CA) for an additional 30 min at 4°C. Samples were washed four times with 1X PBS (serum and azide-free), filtered using Flowmi 1000 μL pipette strainers (SP Bel-Art, Wayne, NJ), and resuspended in 300 μL FACS buffer.

CD3+ T-cells were sorted on the BD FACS Aria Fusion into RPMI (supplemented with 30% FBS). Sorted cells were counted in triplicates on a TC20 Automated Cell Counter (BioRad, Hercules, CA), washed, and resuspended in PBS with 0.04% BSA in a final concentration of 1500 cells/μL. Single-cell suspensions were then immediately loaded on the 10x Genomics Chromium Controller with a loading target of 20,000 cells. Libraries were generated using the 5′ V2 chemistry (for gene expression) and Single Cell 5ʹ Feature Barcode Library Kit per manufacturer’s instructions (10x Genomics, Pleasanton CA). Libraries were sequenced on Illumina NovaSeq 6000 with a sequencing target of 30,000 gene expression reads and 10,000 feature barcode reads per cell.

5′ single cell RNA-Seq data analysis

For 5′ gene expression, alignments were performed using the feature and vdj option in Cell Ranger. Following alignment, hashing (HTO) and cell surface features (Antibody Capture) from feature barcoding alignments were manually updated in cellranger generated features file. Doublets were then removed in Seurat using the HTODemux function, which assigned sample identity to every cell in the matrix. Droplets with ambient RNA (cells with fewer than 400 detected genes) and dying cells (cells with more than 20% total mitochondrial gene expression) were excluded during initial QC. 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 and ribosomal gene expression levels. Dimensionality reduction was performed using the RunPCA function to obtain the first 30 principal components and clusters visualized using Seurat’s RunUMAP function. Cell types were assigned to individual clusters using FindMarkers function with a log2 fold change cutoff of at least 0.4 (FDR ≤ 0.05) and using a known catalog of well-characterized scRNA markers for human PBMC and decidual leukocytes. 5′ feature barcoding reads were normalized using centered logratio (CLR) transformation. Differential markers between clusters were then detected using FindMarkers function. A combination of gene expression and protein markers was used to define T cell subsets. CCR7 staining did not exhibit a significant positive peak, likely due to low signal in freshly thawed cells and hence was excluded from all downstream analyses. Only non-naïve T cell clusters (clusters 1, 4, 5, 7, 9, 10, 13, 5, 21, 23, 24, 26, 27, Figure S3C) based on relative gene expression of SELL, IL7R, and CCR7) and protein expression of CD45RA were retained for downstream analyses.

scTCR analyses

TCR reads were aligned to VDJ-GRCh38 ensembl reference using Cell Ranger 6.0 (10x Genomics) generating sequences and annotations such as gene usage, clonotype frequency, and cell-specific barcode information. Only cells with one productive alpha and one productive beta chain were retained for downstream analyses. CDR3 sequences were required to have a length between 5 and 27 amino acids, start with a C, and not contain a stop codon. Clonal assignments from cellranger were used to perform all downstream analyses using the R package immunarch. Data were first parsed through repLoad function in immunarch, and clonality was examined using repExplore function. Diversity estimates (Chao Diversity) were calculated using repDiversity function.

QUANTIFICATION AND STATISTICAL ANALYSIS

All statistical analyses were conducted in Prism 8 (GraphPad). All definitive outliers in two-way and four-way comparisons were identified using ROUT analysis (Q = 0.1%) after testing for normality using Shapiro-Wilk test (alpha = 0.05). If data were normally distributed across all groups, differences were tested using ordinary one-way ANOVA with unmatched samples. Multiple comparisons were corrected using Holm-Sidak test adjusting the family-wise significance and confidence level at 0.05. If the Gaussian assumption was not satisfied, differences were tested using the Kruskall-Wallis test (alpha = 0.05) followed by Dunn’s multiple hypothesis correction tests. Differences in normally distributed two groups were tested using an unpaired t test with Welch’s correction (assuming different standard deviations). Two group comparisons that failed normality tests were tested for differences using Mann-Whitney test. For subsets of cells within a sample, differences were tested using paired t-tests. p-values are indicated as ^# - p < 0.1, * - p < 0.05, ** - p < 0.01; *** - p < 0.001; **** - p < 0.0001, unless otherwise indicated.

Highlights.

• Decidual immune changes include T cell expansion and macrophage/NK cell contraction

• Term decidual T cells are skewed to Treg/Th17 and are clonally restricted

• Two subsets of macrophages with distinct characteristics exist in term decidua

• Maternal obesity alters decidual immunity and dampens antimicrobial responses