CD38-activated macrophages drive age-related placental senescence by depleting NAD⁺ in decidual stromal cells

Abstract

As delayed childbearing becomes increasingly common, elucidating the mechanisms of age-related placental senescence is critical for the development of effective therapeutic strategies. Here, we integrate metabolomics, single-cell RNA sequencing, and spatial transcriptomics to uncover conserved features of aged placentas in humans, mice, and pigs, namely, increased macrophage CD38 expression and NAD⁺ deficiency. Through pharmacological and genetic approaches, we demonstrate that macrophage CD38 depletes NAD⁺ in decidual stromal cells, thereby promoting placental senescence. Mechanistically, early-onset inflammation and senescence-associated secretory phenotype activity drive CD38 expression in macrophages via the IRF5 pathway. Importantly, treatment with NAD⁺ precursors or CD38 inhibitors attenuates age-related placental senescence, rescues intrauterine growth restriction, and improves long-term metabolic outcomes in offspring. These findings reveal a critical role for macrophage-driven metabolic dysregulation in reproductive aging and establish CD38 as a potential therapeutic target for age-associated pregnancy complications.

The authors show that CD38-activated macrophages deplete NAD+ in decidual stromal cells, driving placental senescence and fetal growth restriction during aging. NAD+ restoration in animal models improves pregnancy outcomes and offspring health.

Introduction

With the progression of modern society, delayed childbearing has become increasingly prevalent and is closely linked to a heightened risk of adverse maternal and neonatal outcomes^1–3. The placenta, a transient yet highly heterogeneous organ, forms the critical interface between mother and fetus^4,5. Early onset placental senescence associated with advanced maternal age is well-established as a major contributor to a spectrum of complications, including intrauterine growth restriction (IUGR), preterm birth and preeclampsia^6,7. Despite advances in assisted reproductive technologies, effective strategies to prevent age-related placental senescence and mitigate associated complications remain limited. A deeper understanding of the molecular mechanisms underlying placental aging is therefore crucial for developing targeted interventions to protect both maternal and fetal health.

Cellular senescence is defined by irreversible cell-cycle arrest in aged or damaged cells, curtailing their proliferative and differentiation capacities^8,9. The placenta displays a marked increase in senescence markers as gestation progresses, suggesting an accelerated aging trajectory distinct from other maternal tissues^10,11. Indeed, elevated levels of senescence markers are observed in healthy placental membranes as gestational age advances¹². This premature senescence initiates the senescence-associated secretory phenotype (SASP), a complex and potent pro-inflammatory program¹⁰. SASPs drive immune cell recruitment and amplify systemic inflammation¹³. However, despite the placenta’s complex metabolic landscape and cellular diversity, the spatial organization of SASP activation and the molecular triggers of placental senescence remain poorly defined. These knowledge gaps hinder our ability to link cellular senescence with adverse pregnancy outcomes.

Nicotinamide adenine dinucleotide (NAD⁺), a central regulator of cellular metabolism, maintains homeostasis both as a redox cofactor and as a substrate for NAD⁺-consuming enzymes^14,15. Disruptions in NAD⁺ metabolism have been increasingly implicated in the acceleration of cellular senescence across species¹⁶. NAD⁺ homeostasis is finely balanced by its biosynthetic enzymes (e.g., NMNAT1–3 and NAMPT) and its consumers (e.g., SIRT1–6, PARP1/2, and CD38)^14–17. NAD⁺ levels decrease with chronological aging in worms, flies, mice, and humans, representing a conserved feature of aging across species¹⁸. However, despite the unique metabolic alterations observed in female reproductive organs¹⁹, most studies have remained focused on major metabolic tissues such as the liver, adipose tissue, and skeletal muscle^20–23(Supplementary Fig. 1), predominantly using male animals as experimental models. Physiological accumulation of senescent cells contributes to placental maturation and the adaptations required for labor onset¹¹. Premature or excessive placental senescence, in particular, is closely tied to mitochondrial dysfunction and SASP activation²⁴, processes that are highly sensitive to NAD⁺ availability. Nonetheless, the spatial and temporal regulation of NAD⁺ metabolism in the placenta, and its interactions with broader metabolic networks, remain largely unexplored.

Given the placenta’s high energy demands and rapid aging⁴, it provides an ideal model to investigate NAD⁺ dynamics in a tissue-specific context. Although NAD⁺ precursor supplementation has shown promise in restoring NAD⁺ levels, their pharmacokinetics and bioavailability vary considerably across tissues and cell types^25,26. The specific effects of NAD⁺ precursors on placental aging and related pregnancy outcomes, however, remain poorly characterized.

To address these questions, we systematically investigated the immunometabolic basis of age-related placental senescence across humans, mice, and pigs. We identified a conserved decline in placental NAD⁺ levels with advancing maternal age, which may contribute to impaired fetal growth. Single-cell and spatial transcriptomic analyses revealed a significant accumulation of CD38⁺ macrophages within the decidual basalis (Db) layer of aged murine placentas. Through a series of in vitro and in vivo experiments (Supplementary Fig. 2), we demonstrate that CD38⁺ macrophages promote placental senescence by inducing NAD⁺ depletion in decidual stromal cells (DSCs). Collectively, our findings establish dysregulated NAD⁺ metabolism, driven by macrophage CD38 activity, as a key mechanism in placental aging and highlight CD38 as a potential therapeutic target to improve pregnancy outcomes in older mothers.

Results

Age-related reduction in placental NAD⁺ is associated with fetal growth delay

To explore the metabolic alterations underlying age-related placental senescence, we conducted untargeted metabolomic profiling of placental samples collected from young mothers (aged 20–25 years, n = 13) and aged mothers (aged 35–44 years, n = 13) (Supplementary Table 1). Among the 869 metabolites analyzed, 60 (6.90%) displayed significant differences between the two groups (Fig. 1a–c). These metabolites were predominantly enriched in nutrient metabolism pathways, including cysteine, methionine, pyruvate, and nicotinate/nicotinamide (NAM) metabolism (Fig. 1d). Notably, aged placentas exhibited substantial reductions in three key metabolites within the NAD⁺ biosynthetic pathway: NAD⁺, NADH, and NADP⁺ (Fig. 1b–c).

Fig. 1. Age-related reduction in placental NAD⁺ is associated with fetal growth deficiencies.

a Heatmap showing differentially expressed metabolites in placentas from young (20–25 years, n = 13) and aged (35–44 years, n = 13) mothers. The analytical method for differential abundance metabolites was described in the Methods section. b Differential metabolites related to NAD⁺ metabolism (n = 13). The P value was calculated using the limma package. The center line denotes the median, while the box contains the 25 to 75th percentiles and the whiskers mark 1.5x interquartile range. c Volcano plot of metabolite changes in aged human placentas (fold change > 2, FDR < 0.05). Red and blue dots indicate elevated and reduced metabolites, respectively. d Pathways with significantly altered metabolites in aged placentas. e–g Correlations between birth weight (x-axis) and placental NAD⁺ (e), NADH (f), and NADP⁺ (g) levels (y-axis), determined by Pearson’s correlation. h mRNA expression of NAD⁺-consuming enzymes (n = 13). i Summary of changes in NAD⁺ levels and the expression of key NAD⁺-synthesizing and -consuming enzymes. Red, purple, and grey denote significant increase, decrease, or no change, respectively. j,k Levels of NAD⁺, NADH, and NADP⁺ in placentas from young and aged mice (n = 6) (j) and pigs (n = 14) (k). Data are presented as mean ± s.e.m. Unpaired two-tailed Student’s t-test was used in (b,h,j,k). *P  <  0.05, and **P  <  0.01. Each symbol represents one biological replicate.

Advanced maternal age is well known to correlate with reduced fetal birth weight^6,7 (Supplementary Fig. 3a,b). In line with this, we observed a positive correlation between placental levels of NAD⁺, NADH, and NADP⁺ with fetal birth weight (Fig. 1e–g), suggesting that disrupted NAD⁺ homeostasis in aged placentas may contribute to impaired fetal growth.

We next examined the expression of four NAD⁺-synthesizing and eight NAD⁺-consuming enzymes to assess their roles in maintaining NAD⁺ homeostasis. Although the expression levels of NAD⁺-synthesizing enzymes remained unchanged, we observed a significant increase in the expression of four NAD⁺-consuming enzymes (Cd38, Parp2, Sirt3, and Sirt5) in aged placentas compared to those from young mothers (Fig. 1h, i and Supplementary Fig. 3c, d). These findings imply that age-related NAD⁺ depletion may be driven by upregulated catabolic enzyme activity, thereby contributing to fetal growth restriction.

To determine whether this phenomenon is evolutionarily conserved and amenable to mechanistic study in animal models, we assessed placental NAD⁺ levels in mice (aged 3 vs. 8 months) and pigs (aged 11 vs. 59 months). Consistent with human data, aged dams in both species displayed adverse pregnancy outcomes, including fewer live fetuses, higher pregnancy loss rates, reduced fetal weights, and significant placental NAD⁺ depletion (Fig. 1j,k and Supplementary Fig. 3e–l). These findings suggest that placental NAD⁺ decline and associated fetal growth impairment constitute a conserved hallmark of maternal aging across species, despite their distinct placental structures and classifications.

Age-related NAD⁺ decline is localized to the decidual basalis and junctional zones

To delineate the spatial pattern of NAD⁺ depletion associated with maternal aging, we quantified NAD⁺ and NADH levels in anatomically distinct regions of the mouse placenta. These included the outer maternal Db and adjacent junctional zone (Jz), analyzed as a unit, and the inner fetal labyrinth zone (Lz). Compared to placentas from young mothers, those from aged mothers exhibited significant reductions in NAD⁺ and NADH levels, with the most pronounced decreases localized to the Db-Jz compartment (Fig. 2a and Supplementary Fig. 4a). This regional depletion aligned with stronger alterations in NAD⁺-regulatory enzyme activity in the Db-Jz compared to the Lz of aged placentas (Supplementary Fig. 4b, c).

Fig. 2. Placental NAD⁺ decline in the Jz and Db correlates with alterations in cell populations.

a Levels of NAD⁺ and NADH in the Db-Jz of placentas from young and aged mice at embryonic day 18.5 (E18.5) (n = 6). b Schematic of the scRNA-seq workflow. Each group had three biological replicates, each consisting of Db and Jz tissues from two male and two female fetuses from different litters. c UMAP plot of 52,362 cells with major cell types annotated by marker gene expression. d Fold changes in cell-type abundance with age, calculated using a Poisson generalized linear mixed model. e Bee-swarm plot of log₂ fold changes in cell-type abundance from Milo analysis. Significant neighborhoods (10% Spatial FDR) are colored. f Sankey diagram showing shifts in cellular ratio between placentas from young and aged mothers. g,h Ridge plots comparing SASP and inflammatory response signature scores in DSCs from young and aged placentas (Wilcoxon P < 2.2×10⁻¹⁶). i Heatmap of differentially expressed genes in DSCs with pathway annotations. Differential expressed gene testing and GO functional enrichment analyses are described in the Methods section. j,k Circulating levels of estrogen (E₂) (j) and progesterone (P₄) (k) in young and aged mothers (n = 8 per group). l Representative 3D reconstructions of feto-placental arteries at E18.5. m,n Ridge plots comparing NAD⁺ metabolism (m) and NADH complex I assembly (n) activity scores in DSCs (Wilcoxon P < 2.2×10⁻¹⁶). o Differential intercellular interaction strength between DSCs and placental cell types in aged and young maternal placenta. p Heatmap of interaction strength between DSCs and other cell types in young and aged mothers. q Chord diagram of interaction strength between DSCs and other placental cell types in young and aged mothers. Data are shown as mean ± s.e.m. Unpaired two-tailed Student’s t-test used in (a,j,k). *P  <  0.05. Cell-type abbreviations: DSCs, decidual stromal cells; GCs, glycogen cells; S-TGCs, sinusoid trophoblast giant cells; SpTs, spongiotrophoblasts; ECs, endothelial cells; MSCs, mesenchymal stromal cells; EryCs, erythroid cells; GMPs, granulocyte-macrophage progenitors; DCs, dendritic cells.

Moreover, histological analysis revealed an expansion of the Db-Jz compartment in placentas from aged mothers (Supplementary Fig. 4d). This morphological change was accompanied by enhanced senescence, as indicated by increased levels of p16 (mRNA) and P21 (mRNA and protein) in the placenta, along with altered expression of senescence-associated β-galactosidase (SA-β-Gal), Ki67, γ-H2AX and Lamin B1, as determined by histological approaches, relative to placentas from young mothers (Supplementary Fig. 4e–o). The cell-cycle arrest machinery is widely recognized as an essential prerequisite for the induction of senescence²⁷. To assess cell proliferation activity, we administered a single dose of 5-bromo-2′-deoxyuridine (BrdU) to the mice. In line with the elevated senescence marker levels, we found that cell proliferation activity was reduced in placentas from aged mothers (Supplementary Fig. 4p, q). These data suggest that the outer placental layers Db and adjacent Jz, which comprise the maternal-fetal interface, are particularly susceptible to age-related metabolic disruption. Their heightened sensitivity to NAD⁺ depletion and senescence underscore their central role in mediating the physiological impact of maternal aging on placental function.

Age-related alterations in placental cell populations

To characterize maternal age-associated changes in placental cellular composition, we performed single-cell RNA sequencing (scRNA-seq) on the combined Db and Jz compartments from placentas of young and aged mouse mothers (Fig. 2b). Analysis of 52,362 high-quality cells using Uniform Manifold Approximation and Projection (UMAP) revealed 15 major cell types, defined by canonical marker gene expression (Fig. 2c and Supplementary Fig. 5a, b). Using a Poisson linear mixed model²⁸, we detected a significant age-related reduction in parenchymal cell types, particularly DSCs, accompanied by an expansion of immune populations, most notably macrophages (Fig. 2d). These trends were further validated by differential cell abundance analysis using the Milo algorithm²⁹ (Fig. 2e).

DSCs are essential for sustaining pregnancy by promoting maternal vascular remodeling, modulating immune tolerance, and supplying nutrients to the fetus³⁰. We observed a striking reduction in DSC proportion with age, from 44.8% in young placentas to 23.9% in aged placentas (Fig. 2f and Supplementary Table 2). Concurrently, DSCs from aged mothers exhibited significantly higher activity scores for SASP and inflammation-related gene signatures (Fig. 2g, h), suggesting compromised functionality and a proinflammatory microenvironment. This inference was supported by pseudobulk analysis showing enriched expression of immune-related genes in DSCs from aged placentas compared to those from young mothers (Fig. 2i). These age-related cellular and transcriptomic alterations correlated with physiological disruptions: aged mothers displayed perturbed steroid hormone profiles (reduced circulating estrogen and progesterone) and impaired feto-placental vascular architecture (Fig. 2j–l).

In alignment with NAD⁺ depletion observed in the Db-Jz compartment, aged DSCs also showed significantly reduced gene activity scores for NAD⁺ metabolism and NADH complex I assembly compared to young counterparts (Fig. 2m, n). Given the established capacity of immune cells to modulate NAD⁺ metabolism in neighboring parenchymal cells by regulating precursor availability³¹, we next examined potential age-related changes in DSC–immune cell interactions. We found a notable increase in macrophage abundance with age, rising from 7.7% of total cells in young placentas to 17.1% in aged placentas (Fig. 2f). In line with our hypothesis, cell-cell interaction analysis revealed significantly enhanced communication between DSCs and macrophages in aged placentas (Fig. 2o–q), suggesting that macrophages may actively modulate the metabolic state of aging DSCs.

Age-related increase in spatial proximity and interaction between DSCs and macrophages

To spatially resolve the interactions between DSCs and macrophages in the aging placenta, we employed the Visium HD platform (10x Genomics), generating high-resolution (2 µm) spatial transcriptomic maps of placentas from young and aged mice. The outer Db, adjacent Jz, and inner Lz were delineated based on marker gene expression, with Prl8a9 (a marker of glycogen trophoblasts) used to define the Jz (Fig. 3a). Cell types in each spot were inferred via reference-based deconvolution using scRNA-seq data and the robust cell type decomposition (RCTD) algorithm from the SpaceXR package³². As macrophage-specific genes showed sparse expression without forming distinct spot clusters (Supplementary Fig. 6a, b), we defined “macrophage spots” as those with a macrophage signature score exceeding the 90th percentile within each tissue section (Fig. 3b; see Methods). These high-scoring macrophage spots were primarily localized within the Db region (Fig. 3b).

Fig. 3. Age-related increase in spatial proximity and interactions between DSC and macrophage spots.

a Spatial transcriptomic architecture of the placenta analyzed via Visium HD. Compartment boundaries were defined by Prl8a9 expression to delineate the Jz. b Spatial distribution of cell types within the Db-Jz of placentas from young and aged mothers. c Heatmap of neighborhood enrichment of spatial cell types in young and aged placentas. d Spatial co-occurrence scores between DSC spots and other cell-type spots across radial distances. e Heatmap of ligand-receptor interaction strength between macrophage spots and other spatially defined cell-type spots. f Classification of DSC spots by proximity to macrophage spots: macrophage-proximal (immediately adjacent), macrophage-intermediate (within five spots), and macrophage-distal (more than five spots away). g Spatial maps showing distributions of DSC spot classes and macrophage spots in placentas of young (top) and aged (bottom) mothers. h,i Activity scores of SASP (h) and inflammatory response (i) gene sets in DSC spots stratified by spatial proximity to macrophage spots in young and aged placentas.

Neighborhood enrichment analysis using Squidpy³³ revealed significant spatial co-enrichment of macrophage and DSC spots in both young and aged placentas (Fig. 3c). Spatial co-occurrence analysis further confirmed that DSC spots were in significantly closer proximity to macrophage spots in aged placentas than in young counterparts (Fig. 3d), indicating a tighter spatial association. The spatial transcriptomic maps generated by the Xenium platform (10× Genomics) recapitulated these findings (Supplementary Fig. 6c–j). This pattern was further supported by ligand-receptor interaction analysis, which demonstrated enhanced communication between DSC and macrophage spots in aged placentas (Fig. 3e). Notably, DSC spots adjacent to macrophage spots exhibited significantly elevated SASP and inflammatory response scores, along with reduced activity of genes associated with the NAD⁺ metabolic process, compared to those located at greater distances. These effects were particularly pronounced in aged placentas, and further spatial analysis across macrophage-proximal, macrophage-intermediate, and macrophage-distal DSC spots corroborated the heightened inflammatory and senescence-associated signaling in aged mothers (Fig. 3f–i and Supplementary Fig. 6k–m). These results suggest that increased spatial proximity and interaction between DSCs and macrophages may exacerbate inflammatory signaling and metabolic dysfunction during placental aging.

Altogether, the pronounced co-localization and enhanced communication between DSCs and macrophages in aged placentas support the existence of a macrophage–DSC regulatory axis that influences local NAD⁺ metabolism. We propose that in aged placentas, macrophages are actively recruited by DSCs and contribute to a local microenvironment that promotes NAD⁺ depletion and senescence.

CD38⁺ macrophages mediate NAD⁺ decline in the placenta during aging

To investigate the influence of macrophages on NAD⁺ metabolism in DSCs, we analyzed the expression of 11 NAD⁺-related metabolic enzymes in macrophages utilizing our scRNA-seq dataset. Among these, CD38, a key NAD⁺-consuming enzyme, exhibited the most pronounced age-related upregulation in its expression and was predominantly enriched in macrophages (Fig. 4a–c and Supplementary Fig. 7a). This finding was further corroborated by flow cytometry and immunofluorescence staining, which confirmed an increased density of CD38 in macrophages from placentas of aged mothers (Fig. 4d–g). Given the extracellular enzymatic activity of CD38, which enables it to deplete NAD⁺ precursors from neighboring cells³¹, these results suggest that macrophage-derived CD38 may be a major contributor to NAD⁺ depletion in DSCs during maternal aging.

Fig. 4. Early-onset SASP and inflammation drive CD38-mediated NAD⁺ depletion.

a Dot plot showing expression of NAD⁺ metabolism genes in macrophages from young and aged placentas. Dot size indicates cell proportion; color intensity reflects normalized expression. b UMAP plots showing Cd38 expression across Db-Jz macrophages in young and aged placentas. c Violin plots comparing Cd38 expression in macrophages. d,e Representative FACS plots quantifying CD38⁺F4/80⁺ macrophages in Db-Jz from young (n = 10) and aged (n = 12) mothers. f,g Representative immunofluorescence of CD68 and CD38 protein in placentas from young and aged mothers (f) and quantification of CD38⁺ and CD68⁺ cells from immunofluorescence images (g) (n =   6; each symbol represents one biological replicate, ten fields of view were selected for each biological replicate). h Schematic of inflammatory (LPS) or senescence (Doxo) induction in pregnant mice (E13.5–E15.5); tissues collected at E16.5. i,j Western blot analysis of CD38 levels in Lz, Db-Jz compartments (n = 4). k, RT-qPCR quantification of Cd38 mRNA in Db-Jz (Young, Young+LPS, Aged+Doxo, n = 8; Young+Doxo, Aged, Aged+LPS, n = 6). l, Schematic of macrophage depletion strategy using clodronate (CL) vs. control liposomes (PL) in LPS-treated pregnant mice. m, Western blot of CD38 in Db-Jz following macrophage depletion (n = 4). n, NAD⁺ levels in Db-Jz measured by mass spectrometry after CL vs. PL treatment (n = 6). o, Schematic of CD38 inhibition (78c) following LPS or Doxo induction. p, NAD⁺ levels in Db-Jz after 78c treatment (n = 6). q, NAD⁺ levels in Db-Jz of aged Cd38^Lyz2 vs. wild-type mice at E18.5 (n = 6). Each point represents an individual biological replicate. Data are presented as mean ± s.e.m. Statistical significance determined by unpaired two-tailed Student’s t-tests, whereas panel k was analyzed using two-way ANOVA followed by Tukey’s multiple comparisons test. *P  <  0.05, **P  <  0.01, and ***P  <  0.001.

To investigate the role of aged-related factors in the elevation of CD38 in the placenta, we administered either doxorubicin (Doxo), a DNA damage-inducing chemotherapeutic, or lipopolysaccharide (LPS), a proinflammatory stimulus, to pregnant mice (Fig. 4h). In line with our earlier observations that the Db-Jz compartment is particularly susceptible to aging, both treatments induced greater CD38 protein increases in the Db-Jz than in the Lz (Fig. 4i,j), recapitulating the spatial expression pattern seen in naturally aged placentas. Moreover, expression of Cd38, the macrophage marker F4/80, and inflammatory cytokines (Tnfα, Il1β, Il6) was significantly elevated in aged placentas following Doxo or LPS treatment compared to younger counterparts (Fig. 4k; Supplementary Fig. 7b-e), suggesting that age-related senescence and inflammation are key drivers of placental CD38 upregulation, particularly in the context of advanced maternal age.

To directly assess the role of macrophages in CD38-mediated NAD⁺ depletion, we selectively depleted macrophages via intraperitoneal injection (i.p.) of clodronate liposomes (CL), using PBS liposomes (PL) as a control (Fig. 4l). Depletion of Macrophages significantly ameliorated LPS-induced impairments in pregnancy outcomes (Supplementary Fig. 7f-i). Additionally, this intervention markedly reduced the expression of F4/80 and proinflammatory cytokines (Supplementary Fig. 7j-n), thereby confirming the effective depletion of macrophages. Notably, CD38 protein expression in the Db-Jz was markedly reduced in CL-treated mice (Fig. 4m), and LPS-induced NAD⁺ decline was significantly rescued (Fig. 4n). These results underscore the central role of macrophage-derived CD38 in mediating placental NAD⁺ loss under inflammatory or senescent stress.

To further establish the functional relevance of CD38 activity, we treated pregnant mice with 78c, a potent CD38 inhibitor, following Doxo or LPS challenge (Fig. 4o). While both stressors significantly lowered placental NAD⁺ levels and intensified inflammatory responses, 78c treatment effectively reversed these effects (Fig. 4p; Supplementary Fig. 7o-r). Importantly, in macrophage-specific Cd38 knockout mice (Cd38^Lyz2), placental NAD⁺ levels in the Db-Jz were significantly higher than those in age-matched wild-type (WT) controls (Fig. 4q), providing genetic evidence that CD38 in macrophages is essential for the observed NAD⁺ decline.

Together, these data identify macrophage-derived CD38 as a key effector of NAD⁺ depletion in aging placentas, acting downstream of inflammation and senescence signals. Targeting CD38 may thus offer a promising strategy to mitigate placental aging and improve pregnancy outcomes in older mothers.

Macrophage-derived CD38 drives age-associated fetal growth restriction

To further define the role of macrophage-derived CD38 in maternal age-related fetal growth restriction, we assessed reproductive outcomes in aged Cd38^Lyz2 mice. Consistent with our earlier findings of placental metabolic dysfunction and impaired pregnancy outcomes during aging (Fig. 2and Supplementary Fig. 3), deletion of CD38 in macrophages markedly ameliorated these deficits. Specifically, aged Cd38^Lyz2 mice exhibited increased litter sizes, improved fetal weights, and restored circulating estrogen and progesterone levels (Fig. 5a-c; Supplementary Fig. 8a,b). In contrast, aged WT mice displayed pronounced Lz atrophy and elevated placental senescence, as evidenced by SA-β-Gal staining (Fig. 5d-f; Supplementary Fig. 8c-l), whereas these phenotypes were substantially rescued in age-matched Cd38^Lyz2 mice (Fig. 5d-f; Supplementary Fig. 8c-l). These results support a central regulatory role for macrophage-expressed CD38 in placental remodeling and gestational adaptation during aging.

Fig. 5. Macrophage-derived CD38 regulates placental NAD⁺ and fetal development during aging.

a Fetal weights in young WT (n = 54), aged WT (n = 45), and age-matched Cd38^Lyz2 (n = 45) mice at E18.5. b Fetal-to-placental weight ratio (Young-Cd38^flfl, n = 52; Aged-Cd38^flfl, Aged-Cd38^Lyz2, n = 45). c Proportion of growth-restricted fetuses (below 10th percentile of young WT controls). d Lz-to-Db-Jz ratios quantified from H&E-stained sections (Young-Cd38^flfl, n = 7; Aged-Cd38^flfl, n = 8; Aged-Cd38^Lyz2, n = 6). e,f Representative images of placental SA-β-Gal staining (e) and quantification of SA-β-Gal-positive cells from Cd38^Lyz2 mice and age-matched Cd38^fl/fl mice (f) (n = 10 mice per group). g Gross placental images and 3D vascular reconstructions from young WT, aged WT, and aged Cd38^Lyz2 mice. h,i Spatial visualization of gene set scores related to hematopoiesis (h) and CD31 immunofluorescence staining of vascular formation genes (i) in the placental sections from young-WT, aged-WT, and age-matched Cd38^Lyz2 mice at E18.5. Representative images from five biologically independent mice per group with similar results. j,k Ridge plots comparing SASP (j) and inflammatory response (k) scores in DSC spots adjacent to macrophage across groups. (Wilcoxon rank-sum test, P < 2.2 × 10^-16). l Heatmap of neighborhood enrichment among spatial cell types in maternal placenta with CD38 deletion. m, Spatial co-occurrence scores between DSC spots and other cell types. n Differential intercellular interaction strength between macrophage and other cell types across conditions. Each dot represents a biological replicate. Data are shown as mean ± s.e.m. Statistical analysis was performed using unpaired two-tailed Student’s ttests for (a,b,d,f), and the chi-square test for (c). *P  <  0.05, **P  <  0.01, and ***P  <  0.001.

Placental vascular density is critical for maternal-fetal nutrient exchange. In aged WT mothers, large white plaques were evident on the fetal side of the placenta, accompanied by a marked reduction in vascularization (Fig. 5g). These defects were absent in Cd38^Lyz2 mice (Fig. 5g), suggesting that CD38 deletion in macrophages restores placental angiogenesis. Using spatial transcriptomic profiling (Visium HD), we observed age-associated suppression of hematopoietic gene programs and endothelial markers in the Lz of WT mice (Fig. 5h). In contrast, Cd38^Lyz2 mice retained gene expression profiles resembling those of young placentas, which correlated with preserved fetal vessel arborization (Fig. 5h). Immunofluorescent staining of CD31, an endothelial marker, further validated the restoration of vascular integrity in Cd38^Lyz2 placentas (Fig. 5i). These findings indicate that CD38 expression in macrophages impairs angiogenic capacity and transcriptional plasticity, thereby compromising fetal nutrient supply.

Consistent with our scRNA-seq results, spatial transcriptomics showed that DSC spots from aged WT placentas displayed significantly elevated activity scores for SASP and inflammatory response gene sets (Supplementary Fig. 8m–o), particularly in proximity to macrophage spots (Fig. 5j, k). In contrast, CD38 deletion in macrophages reduced both SASP and inflammatory activity scores, especially in macrophage-adjacent DSCs (Fig. 5j, k and Supplementary Fig. 8m–o). These data suggest that early-onset senescence in DSCs may recruit macrophages via CD38-dependent mechanisms, establishing a local inflammatory microenvironment.

Spatial neighborhood enrichment and co-occurrence analyses revealed reduced clustering between macrophage and DSC spots in Cd38^Lyz2 mice compared to aged WT controls (Fig. 5l, m). This was further corroborated by decreased spatial interaction strength between macrophages and DSC spots in Cd38^Lyz2 mice relative to aged WT controls (Fig. 5n). Those results were further validated by Xenium In Situ platform (10x Genomics) (Supplementary Fig. 8p–t), supporting the notion that CD38 is required for macrophage recruitment by senescent DSCs.

Together, these findings demonstrate that CD38-expressing macrophages act as key drivers of placental inflammation, DSC senescence, and impaired vascular remodeling. By disrupting nutrient flow at the maternal-fetal interface, this axis contributes to fetal growth restriction. Importantly, selective deletion of CD38 in macrophages effectively breaks this pathological loop, highlighting its therapeutic potential in age-related pregnancy complications.

Age-related senescence and inflammation drive CD38 upregulation in pro-inflammatory M1 macrophages

To explore the mechanisms underlying age-dependent CD38 elevation in macrophages, we analyzed its expression during M1/M2 polarization. Bone marrow-derived macrophages (BMDMs) were induced toward pro-inflammatory (M1) or anti-inflammatory (M2) phenotypes. CD38 expression, at both the mRNA and protein levels, was robustly upregulated in M1 macrophages but not in M2 (Fig. 6a and Supplementary Fig. 9a, b). This M1-specific induction aligned with single-cell transcriptomic data from aged placentas, where CD38⁺ cells were enriched in inflammatory gene modules (Fig. 6b, c). The overlap of CD38 expression with canonical M1 polarization signatures suggests a feedforward loop: inflammation drives M1 polarization, which in turn amplifies CD38-mediated NAD⁺ depletion.

Fig. 6. Inflammation promotes CD38 accumulation in M1 macrophages via IRF5 activation.

a Immunofluorescent staining of CD38 (green) in M1 macrophages. b UMAP plot of CD38⁺ macrophages from the Db-Jz of placentas in young and aged mothers. c GO-BP enrichment of genes upregulated in CD38⁺ macrophages. d Western blot analysis of CD38 in macrophages treated with conditioned media from ESCs of young or aged mice (n = 4). e Schematic of the co-culture system using NMN-treated DSCs and M1 macrophages. f,g Representative images of placental SA-β-Gal staining (f) and quantification of SA-β-Gal-positive cells from DSCs (g) (n = 4 wells per group). h Western blot for CD38 in macrophages (n = 4). i DSCs were cultured in the presence or absence of macrophages and treated with 200 μM NMN or vehicle for 20 hours. DSCs were harvested at 20 hours after NMN administration, and the NAD⁺ levels were detected (n = 6). j DSCs were culture with 200 μM NMN in the presence or absence of pre-mixed human recombinant CD38 (rCD38) (1 μg/mL), and NAD⁺ levels in DSCs were quantified (n = 4). k Macrophages were incubated with or without isatuximab (5 μg/mL) for 3 h, followed by treatment with 200 μM NMN. After 6 h, conditioned media were collected and used to treat DSCs. DSCs were harvested at 8 h post-treatment, and the NAD⁺ levels were quantified (n = 6). l Western blot analysis of CD38 and IRF5 during M1 polarization (n = 3). m CD38 protein levels following IRF5 inhibition with YE6144 (n = 3). n EMSA validation of IRF5–DNA binding. o Luciferase assay assessing IRF5-dependent transcriptional activation of Cd38 (n = 5). p Chromatin immunoprecipitation (ChIP) assay showing that IRF5 binds to the Cd38 promoter in placentas from aged mothers (n = 4). q Schematic of in vivo YE6144 treatment following LPS challenge from E13.5 to E16.5. r Western blot analysis of placental CD38 expression (n = 4). s NAD⁺ levels in the Db-Jz, measured by mass spectrometry (n = 6). Each dot represents one biological replicate. Data are presented as mean ± s.e.m. Statistical significance in d,g,h,j,k,l,m,r,s was determined using unpaired two-tailed Student’s ttests, whereas (i,o,p) were analyzed using two-way ANOVA followed by Tukey’s multiple comparisons test. *P  <  0.05, **P  <  0.01, and ***P  <  0.001.

To assess how the SASP regulates macrophage-derived CD38, we treated BMDMs with conditioned media (CM) from DSCs of young or aged mice. CM from aged DSCs induced a 3.2-fold increase in CD38 protein levels in macrophage within 24 h (Fig. 6d), indicating that SASP factors act as potent inducers of CD38. Indeed, SASP factors levels were higher in DSCs differentiated from endometrial stromal cells (ESCs) of aged females (Supplementary Fig. 9c), and these factors potently induced Cd38 expression in macrophage (Supplementary Fig. 9d). We further established a DSC-macrophage Transwell co-culture system to mimic stromal-immune crosstalk (Supplementary Fig. 9e). Supplementation with antibodies against SASP factors to the CM significantly reduced Cd38 expression in macrophages (Supplementary Fig. 9f), further validating this regulatory role of SASP factors in Cd38 expression. Supplementation with nicotinamide mononucleotide (NMN), a NAD⁺ precursor, attenuated DSCs senescence and significantly reduced CD38 expression in macrophages (Fig. 6e–h). These findings highlight the critical role of stromal NAD⁺ availability in regulating paracrine-induced CD38 expression.

To determine the role of macrophage-derived CD38 activity in the availability of NAD⁺ precursors to DSCs, we performed transwell co-culture experiments with macrophages in the upper chamber and DSCs in the lower chamber. Under basal conditions, macrophages exerted no regulatory effect on NAD⁺ levels in DSCs. In contrast, the NMN-mediated enhancement of NAD⁺ in DSCs was abrogated by macrophages (Fig. 6i), indicating that macrophages can modulate the uptake of NAD⁺ precursors by adjacent cells.

We then investigated the role of CD38 protein in regulating cellular NAD⁺ and its precursors. We treated DSCs with NAD⁺ precursors in the presence or absence of recombinant CD38 protein (rCD38) to mimic the pure ecto-enzymatic activity of CD38. Consistent with the results of macrophage co-culture experiments, rCD38 prevented the NMN-induced increase in NAD⁺ levels in DSCs (Fig. 6j). We further investigated whether the ectoenzymatic or intracellular activity of CD38 regulates NAD⁺ levels in DSCs. Macrophage were pre-treated with isatuximab, a cell-impermeable antibody that specifically inhibits the ectoenzymatic activity of CD38, followed by incubation with NAD⁺ precursors. CM was collected from these macrophages and used to treat DSCs. The addition of isatuximab reversed the inhibition of NMN-induced NAD⁺ enhancement in DSCs (Fig. 6k). Collectively, these results indicate that the ecto-enzymatic activity of CD38 controls the availability of NMN to DSCs.

Together, these results suggest a feedforward mechanism in which stromal NAD⁺ depletion enhances SASP signaling, further promoting CD38 upregulation in macrophages. This pro-inflammatory loop may be exacerbated by age- and pregnancy-associated declines in intestinal barrier integrity, which are linked to elevated systemic inflammation, hallmarks of senescent microenvironments^34,35. To evaluate the transcriptional control of Cd38 by inflammatory cues, we stimulated BMDMs with various pathogen-associated molecular patterns (PAMPs), including LPS, poly(I:C), FSL-1, and ODN826. All stimuli robustly induced Cd38 expression, with LPS eliciting the strongest response (Supplementary Fig. 9g), supporting a central role for pathogen-sensing pathways in regulating CD38 during aging.

Transcription factor IRF5 mediates age-associated CD38 upregulation in M1 macrophages

To dissect the molecular mechanisms by which pathogen-sensing pathways regulate Cd38 expression, we investigated the involvement of Toll-like receptor (TLR) signaling, a convergent pathway activated by both PAMPs and SASP factors³⁶. During M1 polarization of macrophages, we observed robust upregulation of IRF5, a canonical downstream effector of TLR signaling (Fig. 6l), prompting the hypothesis that IRF5 mediates Cd38 transcription in response to inflammatory cues. Cd38-positive macrophages are predominantly enriched in placentas of aged mother (Supplementary Fig. 9h, i). We therefore performed SCENIC analysis³⁷ to further evaluate the transcriptional activity of IRF5 in Cd38⁺ relative to Cd38⁻ macrophages (Supplementary Fig. 9j). The regulon specificity score (RSS) value ranking of IRF5 in Cd38⁺ macrophages indicates that IRF5 is more specifically activated relative to other regulons in this population, which is consistent with its elevated regulon activity (Supplementary Fig. 9k).

To test this, we co-treated macrophages with LPS and the IRF5-specific inhibitor YE6144. Pharmacological blockade of IRF5 abolished LPS-induced CD38 expression (Fig. 6m), indicating that IRF5 activity is required for CD38 upregulation. In silico analysis using the JASPAR and hTFtarget databases identified a predicted IRF5-binding motif (5’-CACACATGGC-3’) within the Cd38 promoter (Supplementary Fig. 9l). DNA–protein docking simulations further supported a high binding affinity between IRF5 and this promoter region (Supplementary Fig. 9m). To experimentally validate IRF5 binding to the Cd38 promoter, we performed electrophoretic mobility shift assays (EMSA). Given that the bacterial lysate proteins expressing IRF5 contained non-specific proteins, we focused on the super-shift bands. A specific super-shift band was observed only in the presence of IRF5 protein, the WT probe, and an IRF5 antibody (Fig. 6n). This band disappeared when either a mutant (MT) probe was used or a competitive unlabeled probe was added, confirming specific IRF5–DNA interaction (Fig. 6n). Luciferase reporter assays provided functional evidence for IRF5-mediated transcriptional activation. Overexpression of IRF5 significantly increased Cd38 promoter-driven luciferase activity, which was partially diminished when the IRF5-binding motif was mutated (Fig. 6o). To determine whether IRF5 directly binds to the Cd38 promoter in vivo, we performed chromatin immunoprecipitation-quantitative PCR (ChIP–qPCR) analysis on placental tissues derived from young and aged mothers. Consistent with our in vitro findings (Fig. 6n,o), IRF5 exhibited direct binding to the Cd38 promoter, and this interaction was markedly enhanced in placentas from aged mothers (Fig. 6p). These results demonstrate that IRF5 directly binds to the Cd38 promoter and promotes its transcription in vivo.

To assess the in vivo relevance of IRF5 in regulating placental CD38 expression, pregnant mice were administered YE6144 following LPS injection (Fig. 6q). As anticipated, LPS induced elevated placental CD38 levels and NAD⁺ depletion; however, IRF5 inhibition significantly attenuated both effects (Fig. 6r, s). Collectively, these findings identify IRF5 as a key transcriptional regulator linking inflammation to CD38 accumulation in M1 macrophages during maternal aging.

Restoration of placental NAD⁺ homeostasis mitigates maternal age-related pregnancy complications

To assess the therapeutic potential of restoring NAD⁺ homeostasis in aged pregnancies, we administered either nicotinamide riboside (NR), a NAD⁺ precursor, or 78c, a CD38 inhibitor, daily from E8.5 to E17.5 (Fig. 7a). Both treatments significantly elevated placental NAD⁺ levels in aged mice (Fig. 7b) and ameliorated key hallmarks of adverse pregnancy outcomes. These included improved fetal weights and reversal of IUGR phenotypes (Fig. 7c–e and Supplementary Fig. 10a). To assess whether the repletion of NAD⁺ during late gestation, subsequent to the establishment of IUGR, yields beneficial effects, we administered daily doses of NR or 78c from E13.5 to E17.5 (Supplementary Fig. 10b). Consistent with our observations from mid-gestational treatment (Fig. 7a–e), both interventions resulted in a significant elevation of placental NAD⁺ levels in aged mice and ameliorated key indicators of adverse pregnancy outcomes (Supplementary Fig. 10c–g). These findings suggest that the restoration of NAD⁺ even during late gestation can effectively alleviate pregnancy-related impairments in aged mice.

Fig. 7. NAD⁺ precursor supplementation or CD38 inhibition restores placental NAD⁺ levels and improves fetal outcomes in aged mice.

a Experimental schematic. Aged pregnant mice received daily injections of NR (400 mg/kg) or 78c (20 mg/kg) from E8.5 to E17.5; tissues collected at E18.5. b Mass spectrometry quantification of NAD⁺ in the Db-Jz (n = 6). c Representative images of fetuses, placentas, and embryo resorption. d Fetal weights (Young, n = 148; Aged, n = 84; Aged+NR, n = 104; Aged+78c, n = 55). e, Incidence of fetal growth restriction (<10th percentile of young control group). f Fetal-to-placental weight ratio (Young, n = 137; Aged, n = 83; Aged+NR, n = 104; Aged+78c, n = 55). g,h Representative images of placental SA-β-Gal staining (g) and quantification of SA-β-Gal-positive cells from NR- or 78c- treated mice (h) (Young, Aged, n = 11; Aged+NR, n = 14; Aged+78c, n = 13). i, Summary of aging-associated differentially expressed genes (DEGs) reversed by NR and 78c. j,k PCA (j) and Euclidean distance clustering (k) of RNA-seq data. The center line denotes themedian, while the box contains the 25–75th percentiles and the whiskersmark 1.5x interquartile range. l In silico deconvolution of cell types based on scRNA-seq reference. m Cell-type proportions heatmap from bulk RNA-seq deconvolution. n Heatmap of gene sets involved in NADH metabolism, SASP, and inflammation. Data are presented as mean ± s.e.m. Statistical significance in (b,d,f,h,k) was determined using unpaired two-tailed Student’s ttests, whereas panel e was analyzed using the chi-square test. *P  <  0.05, **P  <  0.01, and ***P  <  0.001.

Systemic endocrine profiling revealed that NR and 78c corrected age-associated hypoestrogenemia, hypoprogesteronemia, and fetal hypoglycemia (Supplementary Fig. 10h–j), suggesting restored placental steroidogenesis and metabolic homeostasis. In addition, neither NR nor 78c induced detectable maternal toxicity, as evidenced by stable maternal body weight and normal circulating biomarkers of hepatic, renal, and cardiac injury (Supplementary Fig. 10k–n). Morphometric analysis demonstrated that NR and 78c reversed placental hypertrophy and improved fetal-to-placental weight ratios (Fig. 7f and Supplementary Fig. 10o). Notably, Lz hypoplasia and cellular rarefaction, hallmarks of aged placentas, were also ameliorated by treatment (Supplementary Fig. 10p, q). Furthermore, SA-β-Gal staining revealed that placental senescence was significantly reduced in both treatment groups (Fig. 7g, h), which was corroborated by concordant changes in additional senescence markers (Supplementary Fig. 11a–i), supporting the rejuvenating effect of NAD⁺ restoration on placental aging.

To uncover the molecular mechanisms underlying these therapeutic benefits, we performed bulk RNA-seq on placentas from young, aged, and aged+NR/78c-treated mice. Aged placentas exhibited extensive transcriptomic alterations, with 258 genes upregulated and 602 downregulated compared to young controls (Fig. 7i). Remarkably, NR reversed 92.3% of upregulated genes and 88.9% of downregulated genes, while 78c reversed 55.4% and 35.9%, respectively (Fig. 7i), demonstrating potent transcriptomic rescue effects. Principal component analysis (PCA) and Euclidean distance clustering confirmed that NR- and 78c-treated aged placentas were transcriptionally more similar to young controls than to untreated aged samples (Fig. 7j, k). Deconvolution of bulk RNA-seq using scRNA-seq-derived cell type markers revealed elevated macrophage infiltration in aged placentas, which was attenuated by both NR and 78c treatments (Fig. 7l, m). Those findings were further validated by flow cytometry and immunofluorescence staining (Supplementary Fig. 11j–m). Furthermore, expression profiles of genes related to NADH metabolism, SASP, and inflammatory signaling were significantly restored in treated placentas (Fig. 7n), indicating broad functional rejuvenation.

Together, these results establish placental NAD⁺ depletion as a central driver of age-related reproductive decline. Therapeutic interventions that restore NAD⁺ levels, either by precursor supplementation or by inhibiting CD38, can effectively suppress placental senescence, normalize immune cell infiltration, and improve fetal development in aged pregnancies.

Gestational NAD⁺ restoration reverses sex-specific metabolic maladaptation in offspring of aged mothers

To determine whether restoring maternal NAD⁺ levels during gestation confers long-term metabolic benefits to offspring, we subjected weaned pups born to aged dams treated with either NR or 78c to a high sucrose-high fat diet (HSHFD) for 12 weeks (Fig. 8a and Supplementary Table 3). Progeny of aged dams displayed increased body weight in both sexes (Fig. 8b–e), consistent with clinical observations from human cohorts showing higher BMI among offspring of older mothers^38,39. Remarkably, prenatal administration of NR or 78c substantially attenuated this weight gain in offspring from aged pregnancies (Fig. 8b–e).

Fig. 8. Maternal NAD⁺ repletion improves offspring insulin sensitivity and reduces hepatic lipid accumulation following gestational NR or 78c treatment.

a Experimental design for offspring follow-up. Weaned pups were fed a high sucrose-high fat diet (HSHFD) for 12 weeks. b–e Time course (b,d) and final body weights (c,e) at 15 weeks in male (Young, Aged, Aged+NR, n = 12; Aged+78c, n = 11) and female (Young, Aged+78c, n = 9; Aged, Aged+NR, n = 10) offspring. Each point represents a distinct litter. f–k Intraperitoneal glucose tolerance tests, glucose AUC, and HOMA-IR scores for male (Young, Aged+78c, n = 10; Aged, Aged+NR, n = 11) (f–h) and female (i–k) offspring (Young, Aged+78c, n = 9; Aged, Aged+NR, n = 10). l–o, Circulating ALT and AST levels in male (l–m; n = 10) and female (n–o; Young, Aged+78c, n = 9; Aged, Aged+NR, n = 10) offspring on HSHFD at 15 weeks. p,q Oil Red O staining of liver tissue and quantification of lipid droplet area in male and female offspring (n = 6 per group per sex). r–u Liver fatty acid profiles and total fatty acid quantification in male (r–s) and female (t–u) offspring (n = 7 per group per sex). Data are presented as mean ± s.e.m. Statistical significance was determined using unpaired two-tailed Student’s t-tests, except for (b,d,f,i), which were analyzed using two-way repeated-measures ANOVA followed by Dunnett’s multiple comparisons test. Black asterisks indicate comparisons between the Young and Aged groups, red asterisks indicate comparisons between the Aged and Aged+NR groups, and blue asterisks indicate comparisons between the Aged and Aged+78c groups *P  <  0.05, **P  <  0.01, and ***P  <  0.001.

In male offspring, maternal NAD⁺ restoration led to reductions in liver mass and in the weight of two distinct white adipose tissue depots, gonadal (gWAT) and inguinal (iWAT), while female offspring showed no such improvements, indicating a sex-specific effect on adipose tissue remodeling and lipid metabolism (Supplementary Fig. 12a–h). Both male and female offspring of aged mothers exhibited impaired glucose tolerance; however, maternal treatment with NR or 78c markedly improved insulin sensitivity in both sexes (Fig. 8f–k and Supplementary Fig. 12i, j). Additionally, maternal NAD⁺ repletion selectively corrected hyperlipidemia, characterized by elevated circulating cholesterol and triglycerides in male offspring (Supplementary Fig. 12k–n), further emphasizing the sexually dimorphic nature of these metabolic outcomes.

Hepatic dysfunction, a prominent feature in male offspring of aged mothers, was evident by elevated serum alanine aminotransferase (ALT) and aspartate aminotransferase (AST) levels (Fig. 8l–o). These abnormalities were fully reversed by maternal treatment with either NR or 78c (Fig. 8l–o). Histological examination and lipidomic profiling of the liver revealed extensive steatosis in the offspring of aged dams, with notable accumulation of lipid droplets and elevated levels of diabetogenic fatty acids (C16:0, C18:1n9c, C18:2n6c; Fig. 8p–u). Prenatal NAD⁺ restoration significantly suppressed hepatic lipid accumulation and normalized fatty acid composition in male offspring (Fig. 8p–u). These data reveal a partial dissociation between sex-specific adipose tissue expansion and systemic insulin sensitivity, suggesting that targeting hepatic metabolism may be sufficient to reverse key features of metabolic dysfunction in offspring born to aged mothers.

In summary, these findings demonstrate that gestational NAD⁺ repletion during advanced maternal age not only prevents intergenerational transmission of metabolic disorders but also reverses sex-specific maladaptive responses to obesogenic stress in offspring. This highlights placental NAD⁺ metabolism as a central regulator of long-term offspring metabolic health across generations.

Discussion

A detailed investigation of immune cell dynamics in the context of placental senescence offers substantial promise for the identification of therapeutic targets aimed at mitigating age-related placental dysfunction. In this study, by integrating scRNA-seq with spatial transcriptomics, we identified an enrichment of CD38⁺ macrophages and a concurrent depletion of NAD⁺ as defining molecular features of placental aging in older mothers. Both pharmacological and genetic manipulations confirmed that macrophage-expressed CD38 is a key driver of placental NAD⁺ decline, thereby accelerating senescence (Supplementary Fig. 13). These findings establish CD38 as a central molecular node linking NAD⁺ metabolism to immune activity, mediating macrophage-stromal crosstalk within the placental microenvironment.

While physiological placental senescence is inescapable and necessary for normal fetal development, aberrant or premature senescence is closely associated with adverse pregnancy outcomes^10,11. Despite its clinical significance, the mechanisms regulating placental senescence remain poorly defined. Recent studies have highlighted the role of NAD⁺ metabolic reprogramming in cellular senescence^14–17. Here, we show that NAD⁺ deficiency is a conserved hallmark of placental aging across species, including humans, mice, and pigs. This decline is paralleled by a robust upregulation of CD38 expression in placental macrophages. CD38 has previously been characterized as a key modulator of tissue NAD⁺ homeostasis^{19,31,40–42} and is broadly expressed across immune cell lineages, including lymphocytes, monocytes, macrophages, dendritic cells, granulocytes, NK cells, and endothelial cells^31,43–46.

Our functional analyses, encompassing both gain- and loss-of-function approaches, provide compelling evidence that macrophage CD38 plays a critical role in regulating NAD⁺ levels within the placenta. Importantly, CD38 expression is enriched in macrophages of the decidua basalis, a region central to hormone biosynthesis and vascular remodeling³⁰. Consistent with this spatial distribution, macrophage-specific CD38 deletion in aged dams restored placental vascular density and normalized syncytiotrophoblast–derived estrogen and progesterone production, underscoring the pivotal role of CD38 in preserving placental endocrine and vascular function during aging. Furthermore, using placental macrophage–specific Cd38 deletion models will provide a more definitive approach to delineate the causal role of placental macrophage CD38 in age-related placental senescence.

The maternal-fetal interface requires tightly regulated immunological tolerance to maintain a successful pregnancy, a balance predominantly maintained by leukocytes, including macrophages, NK cells, and T cells^47–49. Our data reveal increased infiltration of maternal immune cells at the placental interface in aged mothers. Immune cells can influence neighboring stromal cell metabolism through mechanisms such as cytokine secretion and modulation of nutrient uptake⁵⁰. In line with this, we observed significant alterations in intercellular signaling networks within aged placentas, particularly between macrophages and DSCs. We show that inflammatory cues and senescent cell-derived factors cooperatively induce CD38 expression in macrophages, further accelerating NAD⁺ depletion. Notably, CD38 is abundantly localized to both the cell membrane and cytoplasm of M1-polarized macrophages^51,52. Future studies should define the subcellular localization of CD38 in placental macrophages and assess its mechanistic contribution to local NAD⁺ metabolism. Furthermore, although current technical limitations constrain the reliable in situ detection of NAD⁺ and NADH, future advances in spatial single-cell metabolomics are expected to enable more precise mapping of NAD⁺ metabolism across distinct placental cell types.

The intrauterine environment exerts a pivotal influence on fetal development and long-term metabolic health⁵³. Our findings show that gestational NAD⁺ repletion in the context of advanced maternal age not only mitigates intergenerational transmission of metabolic disorders but also attenuates sex-specific maladaptive responses to dietary obesogenic stress in offspring. Notably, as NAD⁺ acts as a substrate for SIRT1- and PARP1-mediated epigenetic regulation^54,55, restoring NAD⁺ availability may reprogram the epigenetic landscapes of the placenta to optimize nutrient supply and fetal growth. Consistent with this notion, we observed sex-specific differences in adipose tissue expansion and hepatic function, with males showing greater susceptibility to metabolic dysfunction. These results suggest that NAD⁺ repletion during pregnancy mitigates offspring metabolic risk may through sex-dependent epigenetic remodeling of the placenta. While restoring NAD⁺ homeostasis in the context of advanced maternal age via NR supplementation or CD38 inhibition (78c) confers long-term benefits to offspring, the potential impact of macrophage depletion as an alternative strategy for gestational NAD⁺ repletion on postnatal development remains to be further explored.

In this study, we establish a conceptual framework for the development of NAD⁺-targeted strategies to alleviate placental aging in humans. The conserved decline in placental NAD⁺ observed in pigs further implies broader relevance to reproductive performance and offspring vitality in livestock. While mouse models have proven invaluable for elucidating fundamental mechanisms, species-specific differences in placental architecture, immune composition, and metabolic regulation may restrict direct clinical translation^56,57. Additionally, the genetic and environmental uniformity of laboratory mice fails to fully reflect human diversity. Thus, the efficacy and safety of potential NAD⁺-based interventions for age-related pregnancy complications must be validated through long-term, large-scale clinical trials.

In conclusion, we identify NAD⁺ deficiency and elevated CD38 expression as core signatures of placental senescence in aged mothers. Our findings demonstrate that macrophage-derived CD38 promotes NAD⁺ depletion in the decidua, thereby driving placental aging. These insights offer a mechanistic foundation for therapeutic interventions aimed at restoring placental function, preventing fetal growth restriction, and improving the long-term metabolic health of offspring from aged pregnancies.

Methods

Human sample collection

Placental tissues were collected exclusively from biologically female participants (pregnant women); therefore, gender was not considered as a variable in this study. Term placental tissues were collected from two maternal age cohorts: young mothers (aged 20–25 years, gestational age 37.14–40.43 weeks, n = 13) and aged mothers (aged 35–44 years, gestational age 37.71–40.71 weeks, n = 13). Participants with major complications, including spontaneous abortion, ectopic pregnancy, pre-existing hypertension, gestational diabetes mellitus, premature rupture of membranes, renal disease, preterm labor, or pregnancies conceived via in vitro fertilization and embryo transfer were excluded. Clinical characteristics are detailed in Supplementary Table 1. Placental tissues were immediately immersed in ice-cold Dulbecco’s Modified Eagle Medium (DMEM) supplemented with DMEM/F12 or high snap-frozen in liquid nitrogen and stored at −80 °C. The human sample collection was approved by the Human Research Ethics Committee of the Second Affiliated Hospital of Nanchang University (No. 202409005), and written informed consent was obtained from all participants prior to sample collection.

Mice

Female wild-type C57BL/6 J mice were obtained from Beijing Vital River Experimental Animals Center. Mice aged 12 weeks and 32 weeks were classified as young and aged groups, respectively. CD38-floxed (CD38^fl/fl) and Lyz2-Cre mice were purchased from Cyagen Biosciences. For timed pregnancies, females were mated with C57BL/6 J males, and vaginal plug detection (between 8:00–9:00 AM) was designated as embryonic day 0.5 (E0.5). Mice were housed under specific-pathogen-free conditions at an ambient temperature of 21 °C–26 °C and relative humidity of 50–60%, under a 12-hour light/dark cycle. Mice were fed with a normal chow diet (ND) with 10% calories from fat (D12450J) or an HSHF diet, which consisted of 35% carbohydrate, 20% protein, and 45% fat (D12451), purchased from Research Diets (New Brunswick, NJ, USA). Composition of the diets are listed in Supplementary Table 3. Mice were euthanized by carbon dioxide inhalation followed by cervical dislocation. Blood samples were collected by cardiac puncture, and serum was isolated and stored at −20 °C for biochemical analyses. Tissues were rapidly excised, snap-frozen on dry ice, and stored at −80 °C until further analysis. All mouse experiments were approved by the Animal Care and Use Committee of Sichuan Agricultural University (No. 20220092) and conducted in accordance with the Guide for the Care and Use of Laboratory Animals (China).

Pigs

Pregnant sows from a commercial farm were housed under standardized management conditions in environmentally controlled facilities maintained at 18 °C–22 °C, with free access to water and experimental diets formulated according to the National Research Council recommendations (2012). Placentas were collected from 11-month-old (n = 30) and 59-month-old (n = 30) sows. Immediately following parturition, freshly delivered placentas were visually inspected, and only those without congestion, discoloration, or calcification were used. Tissue samples were excised within 10 cm of the umbilical cord insertion, avoiding large vessels. Samples were flash-frozen in liquid nitrogen and stored at −80 °C. All pig experiments were approved by the Animal Care and Use Committee of Sichuan Agricultural University (No. 20240154) and complied with national animal welfare guidelines.

Cell isolation and culture

Macrophages: Bone marrow cells were isolated from the femurs and tibias of 6–8-week-old female mice and cultured in DMEM containing 10% fetal bovine serum (FBS) and 20 ng/mL macrophage colony-stimulating factor (M-CSF) at 37 °C and 5% CO₂. The medium was changed on day 2 and refreshed every 48 h. For polarization, cells were stimulated for 24 h with LPS (100 ng/mL) and IFN-γ (20 ng/mL) for M1 differentiation or IL-4 (20 ng/mL) for M2 polarization. M0 macrophages were maintained in DMEM + 10% FBS without additional stimulation.

Endometrial Stromal Cells: Uteri from young (8–12 weeks) and aged (10–12 months) C57BL/6 females were excised post-euthanasia. After trimming fat and removing cervical regions, uterine horns were longitudinally incised and minced into ~2 × 2 mm fragments. Tissue was digested in serum-free, phenol red-free DMEM/F12 containing 2 mg/mL collagenase type II under sequential conditions (4 °C for 1 h, room temperature for 1 h, and 37 °C for 10 min, with agitation every 10 min). After settling for 3 min to remove epithelial-enriched supernatant, digestion was halted by adding phenol red-free DMEM/F12 + 10% charcoal-stripped FBS. The suspension was filtered (70 μm), centrifuged (250 × g, 4 min), and the pellet resuspended in complete medium (phenol red-free DMEM/F12 + 10% charcoal-stripped FBS + 100 U/mL penicillin + 0.1 mg/mL streptomycin). ESCs were plated in a 37 °C, 5% CO₂ incubator; non-adherent cells were removed after 2 hours. The medium was changed every 48 h.

ESC Decidualization: ESCs were cultured to ~80% confluence and treated for 5 days with decidualization induction medium (10 nM 17β-estradiol + 1 μM progesterone in complete medium), with media changed every 48 hours.

Co-culture experiment: DSCs were cultured in upper chambers and treated with NMN (200 μM). DSCs were then co-cultured with BMDMs polarized to the M1 phenotype for 18 hours. Subsequently, SA-β-Gal staining was performed on DSCs, and CD38 protein expression was measured on macrophages.

Drug treatment

Inflammation or SASP induction: For LPS or Doxo experiments, 3–4-month-old pregnant mice received daily i.p. injections of LPS (20 μg/kg), Doxo (2.5 mg/kg), or PBS (vehicle) starting at E13.5, and were euthanized at E16.5.

Macrophage Depletion: After LPS injection on E13.5, pregnant mice received daily i.p. injections of 200 μL clodronate liposomes (CL) to deplete macrophages. PBS liposomes (PL) were used as a negative control. Mice were sacrificed on E16.5.

NAD⁺ Restoration: For NAD⁺ precursor supplementation, aged (8-month-old) pregnant mice were administered NR (400 mg/kg/day) or the CD38 inhibitor 78c (20 mg/kg/day) via i.p. injection from E8.5 to E17.5; additionally, a late-gestation treatment regimen was performed with daily administration from E13.5 to E17.5.

Offspring Studies: Dams delivered naturally, and litters were standardized to six pups (3 males, 3 females when possible) by postnatal day 3. Pups were weaned at day 21 and randomized to a HSHF diet. At 14 weeks, one male and one female pup per litter underwent intraperitoneal glucose tolerance testing. All offspring were euthanized at 15 weeks, and one animal per sex per litter was used for endpoint analyses to avoid pseudoreplication and maximize statistical power.

Dissection of placental compartments (Db-Jz and Lz)

After euthanasia, pregnant mice were dissected, and the uteri removed. Placentas were isolated by excising the yolk sac and umbilical cord, then placed in ice-cold PBS. Using a scalpel, placentas were sectioned transversely into ~1 mm slices. Under a dissection microscope, the Lz was separated from the decidua basalis-junctional zone (Db-Jz) using fine Dumont tweezers.

Senescence-associated β-galactosidase staining

Cultured cells were washed once with PBS and fixed in SA-β-Gal fixative for 15 min at room temperature. After three PBS washes, cells were incubated with SA-β-Gal staining solution at 37 °C for 16–20 h. Plates were sealed with Parafilm to minimize evaporation. Following incubation, cells were rinsed with PBS and visualized under a bright-field microscope.

For placental tissue, frozen sections were air-dried at 37 °C for 20–30 min, then fixed in SA-β-Gal fixative for 15 minutes at room temperature. Sections were washed three times with PBS and stained with SA-β-Gal solution overnight at 37 °C. Post-staining, slides were counterstained with eosin for 1–2 min, rinsed under running water for 1 minute, differentiated in 1% acid alcohol for 10–20 s, and rinsed again. Slides were then dehydrated through graded ethanol, cleared in xylene, and mounted with coverslips. After drying, samples were examined microscopically.

Histology and immunostaining

Tissues were fixed in 4% paraformaldehyde for 24 h at room temperature, followed by dehydration, paraffin embedding, and sectioning at a thickness of 4 μm for H&E staining. Adipocyte size (μm²) was quantified across 20 sections per mouse using the Adiposoft ImageJ plugin. For Oil Red O staining, snap-frozen liver samples embedded in OCT were sectioned at a thickness of 8 μm.

For immunohistochemistry, paraffin sections were deparaffinized, rehydrated, and subjected to antigen retrieval in citrate buffer (pH 6.0) using an autoclave for 15 min. Sections were then blocked with 1% BSA in PBS containing 0.05% Tween-20 and incubated overnight at 4 °C with anti-Lamin B1 antibody. Detection was performed using standard HRP–DAB procedures according to the manufacturer’s instructions.

For immunofluorescence staining, sections were permeabilized with 0.05% Triton X-100 in PBS, blocked with 1% BSA in PBS containing 0.05% Tween-20, and incubated overnight at 4 °C with the following primary antibodies: rabbit anti-CD38 (SCT, 68336), anti-F4/80 (Abcam, ab6640), anti-CD68 (Abcam, ab283654), anti-P21 (Abcam, ab188224), anti-Ki67 (Abcam, ab15580), and anti-γ-H2AX (Abcam, ab81299). Appropriate fluorescent secondary antibodies were applied, and nuclei were counterstained with DAPI. Images shown are representative of at least six biological replicates.

For BrdU labeling, pregnant mice received an intraperitoneal injection of BrdU (10 mg/ml, 200 μl per mouse) at E17.5, and placentas were harvested at E18.5.

Placental vascular imaging

Mice were anesthetized with sodium pentobarbital (50 mg/kg) and perfused transcardially with cold 1× PBS, followed by 4% paraformaldehyde (PFA). Placentas were fixed in 4% PFA overnight at 4 °C with gentle agitation, then washed twice with PBS (~2 h each). Tissue clearing was performed using a Nuohai Enhanced Tissue Clearing Kit (NH-CR-230701, Nuohai Life Science, Shanghai). 3D fluorescence imaging was conducted with a Nuohai LS18 Tiling Light Sheet Microscope using laser lines (405, 488, 561, 637 nm), a 4-tile light sheet, and a 1×/0.25 NA Olympus MVPLAPO objective. Imaging was performed at 4× magnification with a resolution of ~3.3 × 3.3 × 7 μm³. Images were processed using LS18 ImageCombine software and rendered in Amira (Thermo Fisher Scientific).

Metabolite profiling

Approximately 50 mg of tissue was extracted in 500 μL of ice-cold methanol/water (4:1, v/v) supplemented with 0.224 mM phenylhydrazine. Samples were mechanically disrupted using a bead homogenizer and centrifuged at 250 × g for 30 min at 4 °C. The resulting supernatants were incubated at −20 °C for 1 h to derivatize α-keto acids, followed by centrifugation at 16,000 × g for 10 min at 4 °C. The remaining pellets were subjected to a second extraction using the same solvent, and the combined supernatants were evaporated to dryness in a SpeedVac concentrator. Dried extracts were reconstituted in 5% acetonitrile before LC–MS analysis on an Agilent 1290 II UPLC system coupled to a Sciex 5600 + QTOF mass spectrometer. Chromatographic separation was performed using either an ACQUITY HSS T3 column (3 × 100 mm, 1.8 μm) or an ACQUITY BEH Amide column (2.1 × 100 mm, 1.7 μm; Waters), depending on metabolite class. Quantification was achieved using isotopically labeled internal standards (Cambridge Isotope Laboratories). Peak intensities were normalized to their corresponding internal standards whenever available; metabolites lacking appropriate isotopic analogues were normalized using the internal standard with the lowest coefficient of variation. Differential metabolite abundance was evaluated using a OPLS-DA model, and significance was defined as FDR < 0.05, VIP > 1.5, and |r| > 0.5.

Serum estrogen and progesterone quantification

Serum levels of estradiol (E₂) and progesterone (P₄) were measured using commercial ELISA kits (KGE014, R&D Systems; FRE-2700, LDN) per manufacturer’s instructions.

Flow cytometry

Placental Db-Jz compartments were dissociated using collagenase V/DNase I in a gentleMACS™ Dissociator (30 min). Homogenates were filtered through a 70 μm strainer, centrifuged at 300 × g for 10 mins at 4 °C, and subjected to leukocyte enrichment via density gradient centrifugation. Mononuclear cells were stained with LIVE/DEAD^® fixable blue dye, incubated for 30 min on ice in the dark, then labeled with antibodies: CD45-BV510, F4/80-PE, CD11b-BV450 and CD38-BB700. For intracellular staining, cells were treated with Transcription Factor Buffer Set (BD Biosciences) and stained with anti-CD206 Alexa Fluor^® 647. Analysis was performed on a BD Verse Cell Analyzer using FlowJo vX.0.7.

Dual-Luciferase reporter assays

HEK 293 T cells were seeded in 24-well plates and transfected with wild-type or mutant CD38 luciferase reporter plasmids (0.5 µg), IRF5 expression plasmid (0.5 µg), and Renilla luciferase control (pRL-CMV, 0.01 µg). At 48 hours post-transfection, luciferase activities were measured using the Dual-Luciferase Reporter Assay System (Promega), with firefly luciferase normalized to Renilla luciferase.

EMSA

A biotin-labeled oligonucleotide probe (5’-AAAGAATGTATTTTTAAAAAAGAAAAAGAAAACACAGGAAA-3’) was incubated with bacterial lysates expressing IRF5 at room temperature for 30 minutes. Complexes were resolved on a non-denaturing 6% polyacrylamide gel in 0.5× TBE buffer at 60 V and 4 °C for 1 hour. DNA was transferred to a Biodyne^® B nylon membrane (Pall Corporation) and detected using a chemiluminescent EMSA kit, visualized via ChemiDoc XRS system (Bio-Rad).

Chromatin immunoprecipitation (ChIP)-PCR assay

ChIP was performed using the SimpleChIP® Enzymatic Chromatin IP Kit (Cat. No. 9003, CST). Briefly, tissues were cross-linked with 1% formaldehyde, and the crosslinking reactions were terminated by the addition of glycine. The tissues were washed with ice-cold PBS, and nuclei were extracted following lysis. Micrococcal nuclease was added to the nuclei suspension to digest genomic DNA. The nuclei were recovered and resuspended in IP Dilution Buffer, followed by sonication to disrupt the nuclear membrane. Chromatin extracts containing DNA fragments with an average size of 500 bp were immunoprecipitated overnight at 4 °C using an anti-IRF5 antibody (Cat. No. 11312-1-AP, Proteintech), or IgG control (included in the kit). ChIP-grade Protein A/G Magnetic beads were added, follower by incubation at 4 °C. RNAse A and Proteinase K were used to digest RNA and protein, respectively. The purified DNA was used for qPCR, and the primers designed to amplify selected regions of the Cd38 promoter are 5’- CTCAGACACTGAGCCACCAA -3’ (forward) and 5’- CAGCTCCGTATGTCAGGAGG -3’ (reverse).

RNA extraction and qRT-PCR

Total RNA from placental tissues (whole, Lz, or Db-Jz) was extracted using RNAiso Plus (TaKaRa). RNA concentration was measured using a NanoDrop 2000. Reverse transcription was performed with TransScript^® All-in-One First-Strand cDNA Synthesis SuperMix with genomic DNA removal (TransGen Biotech). Real-time PCR was conducted on a QuantStudio™ 5 (Thermo Fisher) using PerfectStart^® Green qPCR SuperMix (+ Universal Passive Reference Dye, TransGen Biotech). Expression was normalized to β-ACTIN, and relative expression levels were calculated using the ΔΔCT method. Primer sequences are listed in Supplementary Table 4.

Protein extraction and Western blotting

Db-Jz placental tissues were homogenized in RIPA buffer (Beyotime) containing protease and phosphatase inhibitors, sonicated, and centrifuged. Protein concentrations were determined by BCA assay. Equal protein amounts were denatured at 98 °C for 8 min, resolved on 12% SDS-PAGE gels, and transferred to nitrocellulose membranes at 300 mA for 1 h. Membranes were blocked with 5% BSA for 1 h, incubated overnight at 4 °C with CD38 primary antibody (Abcam, ab216343), and probed with HRP-conjugated secondary antibody (1:2000, Cell Signaling Technology). Signal was detected using enhanced chemiluminescence (ECL) and imaged using a Bio-Rad Image Lab system.

Single-cell RNA sequencing analysis

Library preparation

Db-Jz compartments from mouse placentas were dissected into sterile, RNase-free culture dishes containing calcium- and magnesium-free 1× PBS on ice. Tissues were minced into ~0.5 mm² fragments, washed with PBS, and non-target components such as blood and adipose tissue were removed. Tissues were enzymatically dissociated in a solution containing 0.35% collagenase IV and 5.2 mg/mL papain at 37 °C with shaking (500× g) for 20 min. Digestion was halted by adding 10% FBS in PBS, followed by gentle trituration (5–10 times) with a Pasteur pipette. The suspension was filtered sequentially through stacked 70 and 30 μm strainers, then centrifuged at 300 × g for 5 min at 4 °C. The pellet was resuspended in 100 μL PBS with 0.04% BSA, and 1 mL of 1× red blood cell lysis buffer (MACS 130-094-183) was added. After 2–10 min of incubation at room temperature or on ice, samples were centrifuged again at 300 × g for 5 min.

Dead cells were removed using 100 μL of Dead Cell Removal MicroBeads (MACS 130-090-101) and a Miltenyi^® Dead Cell Removal Kit, followed by two rounds of washing with PBS (0.04% BSA) and centrifugation at 300 × g for 3 min at 4 °C. The final pellet was resuspended in 50 μL PBS (0.04% BSA). Cell viability (>85%) was assessed by trypan blue exclusion, and concentrations were adjusted to 700–1200 cells/μL using a Countess II Automated Cell Counter.

Single-cell suspensions were processed using the 10x Chromium system to capture ~10,000 cells per sample, following the manufacturer’s protocol for a 10x Genomics Chromium Single Cell 3’ kit (v3). cDNA amplification and library preparation were performed according to standard instructions. Libraries were sequenced on an Illumina NovaSeq 6000 platform (paired-end, 150 bp) by LC-Bio Technology Co., Ltd. (Hangzhou, China) with a minimum depth of 20,000 reads per cell.

Preprocessing, clustering, and cell type annotation

Cluster differential gene expression was performed using Wilcoxon rank-sum tests via Seurat’s FindAllMarkers. The p-value adjustment method was explicitly set to Benjamini–Hochberg (FDR), and adjusted p-values (p_val_adj) are reported. The applied thresholds were: avg_logFC > 0.25, p_val_adj <0.05, and min.pct > 0.10.

Raw sequencing data were aligned to the GRCm39 reference genome using Cell Ranger (v8.0.1) to generate count matrices. A total of 72,476 cells across six samples were profiled. Ambient RNA contamination was removed using SoupX (v1.6.2)⁵⁸, and doublets were filtered out with scDblFinder (v1.16.0)⁵⁹.

Cells were retained if they met the following quality thresholds: (i) >250 expressed genes; (ii) >1,250 UMIs; (iii) mitochondrial gene fraction below the median plus five median absolute deviations (MADs); and (iv) ribosomal gene fraction below the median plus five MADs. Genes expressed in fewer than 20 cells were also excluded. After filtering, 53,542 high-quality cells remained.

Data integration across the six libraries was performed using Seurat (v5.1.0)⁶⁰. Counts were normalized and scaled via the SCTransform function, regressing out mitochondrial and ribosomal gene content as well as cell cycle effects. Batch effects were corrected using canonical correlation analysis (CCA) based on the top 2000 highly variable genes (HVGs), followed by PCA, neighbors, clustering, and UMAP on the integrated representation.

Dimensionality reduction was performed using principal component analysis (PCA), and the top 30 PCs were used to construct a shared nearest neighbor (SNN) graph via the FindNeighbors function. Clustering was performed using the FindClusters function (resolution = 1.0), and visualization was achieved via UMAP (RunUMAP function). Cell clusters were annotated using known lineage-specific markers from previous placenta studies. Differential gene expression for each cluster was calculated using the Wilcoxon rank-sum test with thresholds of avg_logFC > 0.25, adjusted P < 0.05, and min.pct > 0.10.

Immune cells were sub-clustered at resolution 1.0, macrophages at 0.8, and trophoblasts at 0.2 using integrated analysis methods. Labyrinthine trophoblasts were excluded, resulting in a final dataset of 52,362 cells for downstream analysis. Pseudobulk expression profiles were generated using AggregateExpression, and differential expression was assessed using DESeq2⁶¹ with adjusted P-values corrected via Independent Hypothesis Weighting (IHW) and log₂ fold changes stabilized using the lfcShrink function.

Analysis of cell type composition and abundance

Differential abundance analysis was conducted using miloR (v1.10.0)²⁹, a k-nearest neighbor graph-based framework. To ensure balanced comparisons between young and aged groups, samples were down-sampled to equal cell numbers. A SingleCellExperiment object was created from the count matrix. Neighborhood graphs were constructed using buildGraph (k = 30, d = 50) and makeNhoods (prop = 0.2, k = 30, d = 50), yielding 5,468 representative neighborhoods.

Differential abundance was tested using testNhoods, with significant neighborhoods (spatial FDR < 0.1) visualized via plotNhoodGraphDA. Changes in cell type composition were quantified by computing the local true sign rate (LTSR)²⁸, which estimates the probability that the direction of log₂ fold change is correct given its mean and variance. A threshold of LTSR > 0.9 was used to define significantly altered cell types.

Signature score definition

To evaluate transcriptional changes associated with placental aging, we computed a SASP score using a predefined SASP gene set¹⁹. Gene set activity scores were calculated with AUCell (v1.24.0). Additional scores for inflammatory response, NADH metabolism, NADH complex I assembly, and NAD⁺ metabolism were derived using curated gene sets from the Gene Ontology and KEGG databases.

Cell–cell interaction analysis

Cell–cell communication was inferred using CellChat (v2.1.2)⁶² on both young and aged scRNA-seq datasets following the standard workflow. The gene expression matrix and annotated cell types were input, and communication probabilities between cell types were computed using the computeCommunProb function with a trim parameter set to 0.1. Separate CellChat objects were created for young and aged conditions, then merged to enable comparison of intercellular communication networks between age groups.

SCENIC analysis

Gene regulatory networks and regulons were elucidated using the pySCENIC (v 0.12.1) pipeline, which includes tools such as ‘grn’, ‘ctx’,and ‘aucell’ (https://doi.org/10.1038/s41596-020-0336-2). The cell × gene matrix was derived from macrophages within the entire dataset, which had undergone quality-controlled and annotation. Briefly, raw count data and a list of known TFs in mice were used as inputs for network inference (GRNBoost2), followed by the generation of candidate regulon (cisTarget). The activity score of candidate regulons in individual cells was quantified using AUCell. Regulons in macrophages across the two conditions were ranked by importance, key regulons were highlighted.

Visium HD spatial transcriptomics analysis

Library preparation

A 10-μm section from E18.5 fresh-frozen placenta was mounted onto a histology slide, air-dried at 42 °C for 3 h, dehydrated overnight, baked at 60 °C for 30 min, and deparaffinized according to the Visium HD protocol (CG000685). Mouse-specific reagents (Visium HD Mouse Transcriptome, 6.5 mm, PN-1000676) were used for H&E imaging, probe hybridization, RNA digestion, and probe amplification. Hybridized analytes were transferred onto the Visium HD slide via the CytAssist system (v2.0). Libraries were constructed and sequenced on an Illumina NovaSeq X Plus using the recommended configuration.

Preprocessing and annotation of spatial transcriptomics data

To characterize macrophage spatial organization, neighborhood enrichment analysis was performed using Squidpy (v1.6.1). After data normalization and scaling, we computed neighborhood enrichment score using sq.gr.spatial_neighbors with default parameters. Briefly, this enrichment score reflects the spatial proximity of cell types: if spots belonging to two different cell types are frequently adjacent, the score is high (defined as enriched); if they are rarely adjacent, the score is low (defined as depleted). This score is based on a permutation-based test, with the number of permutations set to the default values of the n_perms argument. We then visualized cell type co-occurrence in spatial dimensions. The co-occurrence score is defined as p(exp|cond)/p(exp), where p(exp|cond) denotes the conditional probability of observing cell type “exp” given the presence of cell type “cond”, and p(exp) represents the probability of observing cell type “exp” within the radius of interest. The score was computed across increasing radius sizes around each observation using sq.pl.co_occurrence with default parameters.

Spaceranger (v3.1.2) was used to generate spatial gene expression matrices at 2-μm and 8-μm bin resolutions; 8 × 8 μm binning was selected to balance transcript coverage and resolution. The dataset included two analysis areas: area 1 (young and aged placentas) and area 2 (Cd38^Lyz2 and aged placentas). The aged placenta from area 2 was excluded due to low transcript counts per bin.

Using Loupe Browser (v8.0.0), placental regions were segmented into compartments based on Prl8a9, a marker of the Jz, and only the Db-Jz compartments were retained. Scaled gene expression matrices were generated using Seurat (v5.1.0) after cell normalization and regression. For cell type deconvolution, each spatial sample was paired with a matched scRNA-seq reference: young samples with young scRNA-seq, aged with aged, and Cd38^Lyz2 samples with combined scRNA-seq datasets. Deconvolution was performed using robust cell type decomposition (RCTD) from the SpaceXR package (v2.2.1)³².

Due to consistent annotation failure for immune cells, MSCs, and erythroid cells in RCTD, spatial deconvolution was limited to GCs, DSCs, SpTs, S-TGCs, and ECs. Macrophage identification was based on signature scores calculated using Seurat’s AddModuleScore function. Bins with scores above the 90th percentile in each section were classified as macrophage-positive. Signature genes were selected as the top 10 macrophage markers (ranked by log₂ fold change) from our scRNA-seq data, following methods described by Guo et al⁶³.

Spatial classification of DSC spots based on macrophage proximity

To spatially classify DSCs relative to macrophages, bin coordinates were retrieved from the tissue_positions.parquet file of each spatial transcriptomics dataset. Bin barcodes were used to integrate cell type labels and spatial positions. The spatial coordinates of DSC and macrophage bins were extracted. DSC bins within a 3 window macrophage bins, which were designated as macrophage-proximal DSCs. Euclidean distances between DSC and macrophage bins were computed using the rdist function (fields v15.2). DSC bins >5 bins away from any macrophage bin were designated macrophage-distal. Intermediate bins not meeting either criterion were classified as macrophage-intermediate DSCs.

Neighborhood enrichment, centrality, and co-occurrence analyses

To characterize macrophage spatial organization, neighborhood enrichment analysis was performed using Squidpy (v1.6.1). After data normalization and scaling, spatial neighbor graphs were constructed via gr.spatial_neighbors, and neighborhood enrichment was assessed with gr.nhood_enrichment. Co-occurrence probabilities between cell types were calculated using gr.co-occurrence, all using default parameters.

Cell–cell interaction analysis in spatial transcriptomics

Cell–cell communication in the Visium HD dataset was analyzed with CellChat (v2.1.2). Gene expression matrices, spatial bin coordinates, and cell annotations were imported. Communication probabilities were computed using computeCommunProb with the following parameters: type = truncatedMean, trim = 0.1, interaction.range = 250, scale.distance = 0.05, distance.use = TRUE, contact.dependent = TRUE, and contact.range = 21. Only interactions detected in ≥30 bins were retained.

Signature score definition

Feature scoring followed the same methodology used in the scRNA-seq analysis. Hematopoiesis signatures were based on the top 10 EryC-specific genes (highest log₂ fold change) identified in our scRNA-seq dataset.

Xenium spatial transcriptomics analysis

Library preparation

Tissue processing and library preparation were performed according to the manufacturer’s standard protocols for fresh-frozen samples (10× Genomics). Briefly, tissue sections were prepared, fixed, and permeabilized. For targeted transcript detection in mouse placentas, the Xenium Prime 5 K Mouse Pan-Tissue Panel was employed. Gene-specific padlock probes were hybridized in situ, followed by ligation, rolling‑circle amplification, and optional nuclear staining. Samples were imaged using the Xenium Analyzer (v3.4.1.0). Raw image processing and transcript decoding were performed using the integrated Xenium Onboard Analysis software (v3.3.0.1) to generate spatially resolved gene expression matrices.

Processing and annotation pipeline for Xenium data

Using Xenium Explorer (v4.1.1), the Db-Jz compartment of the placenta was selectively isolated based on Prl8a6 expression. Cells within the Db-Jz region were extracted, and those with an nCount_Xenium value of <10 were excluded prior to downstream analyses. Normalized and scaled gene expression matrices were generated using Seurat (v5.1.0) following cell normalization and regression. Cell type annotations for spatial transcriptomic data were transferred from their paired scRNA-seq datasets via the FindTransferAnchors and TransferData functions provided in the Seurat toolkit. Cells assigned a prediction confidence score below 0.7 were labeled as “Unassigned”.

Neighborhood enrichment and co-occurrence analyses for DSCs and macrophages in the Xenium dataset

To characterize the spatial organization of macrophages and DSCs in the Xenium data, neighborhood enrichment analysis was conducted using Squidpy (v1.6.1). Following data normalization and scaling, spatial neighbor graphs were constructed using gr.spatial_neighbors, and neighborhood enrichment was evaluated via gr.nhood_enrichment. Co-occurrence probabilities between the two cell types were calculated using gr.co_occurrence and the resulting scores were normalized.

Cell–cell interaction analysis

Ligand–receptor (LR) pairs were sourced from the CellChat database. These interactions were further refined by retaining only those LR pairs for which the corresponding ligand and receptor genes were included in the Xenium Prime 5 K Mouse Pan-Tissue Panel. For LR complexes containing multiple subunits, only interactions with detectable panel coverage of the required ligand and receptor components were retained for subsequent analysis., with an emphasis on those involved in “Secreted signaling”, “ECM-receptor” interactions and “cell-cell contact”. Macrophage–DSC cell pairs with a centroid-to-centroid distance of ≤ 500μm were identified using scipy.spatial.cKDTree and defined as spatially proximal cell pairs. Subsequent LR scoring was performed following the method described by Qiao et al⁶⁴. Distal cell pairs, serving as a control group, were randomly selected sender-receiver pairs beyond the contact radius. For each LR pair, expression levels were extracted from the Xenium spatial dataset for both proximal and distal cell pairs. Given that distal cell pairs are less likely to engage in direct LR interactions due to the lack of physical proximity, we performed a one-sided t-test to determine the significance of LR pairs with higher expression levels in spatially proximal cell pairs compared to distal pairs. Following the statistical test, false discovery rate (FDR) correction was applied to adjust the P values. The LR score was calculated using the formula: LR score = log₂(ligand × receptor + 1). The resulting LR pairs were then filtered based on criteria of adjusted P value < 0.05 and LR score > 1 and log₂FC > 0.5.

Bulk RNA-seq analysis

Library preparation

Total RNA was isolated with TRIzol reagent (Invitrogen) according to the manufacturer’s instructions. RNA quantity and purity were determined using a NanoDrop ND-1000 spectrophotometer, and RNA integrity was evaluated on an Agilent 2100 Bioanalyzer. Only samples with an RNA integrity number (RIN) greater than 7.0 were used for library construction, and RNA quality was further verified by denaturing agarose gel electrophoresis. For each sample, 1 μg of total RNA was subjected to two successive rounds of poly(A)^ + RNA enrichment using Dynabeads Oligo(dT)_25 (Thermo Fisher). Purified mRNA was chemically fragmented with the Magnesium RNA Fragmentation Module (NEB) at 94 °C for 5–7 min before cDNA synthesis. First-strand cDNA was generated using SuperScript™ II Reverse Transcriptase (Invitrogen), followed by second-strand synthesis with E. coli DNA polymerase I, RNase H, and dUTP (NEB and Thermo Fisher). The resulting double-stranded cDNA fragments were end-repaired and adenylated at their 3′ termini to facilitate adapter ligation. Indexed sequencing adapters, including both single- and dual-index formats, were ligated to the A-tailed fragments, after which libraries were purified and size-selected using AMPure XP beads. Uracil-containing second strands were subsequently removed with heat-labile Uracil-DNA Glycosylase (NEB) to preserve strand specificity.

Libraries were amplified by PCR using the following cycling program: 95 °C for 3 min; eight cycles of 98 °C for 15 s, 60 °C for 15 s, and 72 °C for 30 s; followed by a final extension at 72 °C for 5 min. The resulting libraries had an average insert size of approximately 300 ± 50 bp and were sequenced in paired-end 150-bp mode (PE150) on an Illumina NovaSeq 6000 platform (LC-Bio Technology Co., Ltd.) according to the manufacturer’s standard protocol.

Quality control, alignment, and differential gene expression analysis

Quality control of raw sequencing data was performed using trim-galore (v-0.6.10)⁶⁵ with parameters: -q 25 -j 8 --phred33 --length 74 -e 0.1 --stringency 4 -paired. Transcript quantification was carried out using kallisto (v0.48.0) against the GRCm39 reference genome (GCF_000001635.27) and Ensembl v110 annotations⁶⁶. Transcript-level abundance estimates were aggregated to gene-level TPMs using the tximport R package (v1.30.0)⁶⁷. Unsupervised clustering of all 24 high-quality FASTQ files across four experimental conditions was used to assess potential batch effects or outliers.

Protein-coding genes with TPM > 0.1 in at least 50% of replicates per condition were retained. Differential gene expression analysis was performed using DESeq2 (v1.42.1)⁶¹ with the Wald test. Adjusted p-values (Padj) were calculated using the independent hypothesis weighting (IHW) method, and log₂ fold changes were shrunk using the apeglm algorithm. Genes were defined as differentially expressed if Padj <0.05 and the absolute value of log₂-transformed fold changes (| log₂ FC |) ≥ 0.5849 (equivalently, fold change ≥ 1.5 or fold change ≤ 0.667, where 0.667 denotes the reciprocal of 1.5).

Principal component analysis and Euclidean distance calculation

Principal component analysis (PCA) was conducted using FactoMineR (v2.11)⁶⁸ and factoextra (v-1.0.7)⁶⁹, visualizing sample distributions in PC1 vs. PC2 space. Pairwise Euclidean distances between groups were calculated with the dist function in R, and statistical significance was assessed using the Wilcoxon rank-sum test.

Deconvolution

Bulk RNA-seq deconvolution was carried out using CIBERSORTx⁷⁰, employing cell-type-specific expression profiles from our scRNA-seq dataset as the reference signature matrix.

Gene Ontology (GO) enrichment analysis

Gene Ontology enrichment of differentially expressed genes was performed using Metascape (https://pubmed.ncbi.nlm.nih.gov/30944313/)⁷¹.

Statistical analysis

The number of biological replicates is specified in the corresponding figure legends. Statistical analyses were conducted using GraphPad Prism 9.5.0. Unless otherwise indicated, data are presented as mean ± SEM. Comparisons between two groups were performed using unpaired two-tailed Student’s t-tests. Two-way ANOVA or two-way repeated-measures ANOVA followed by the appropriate multiple comparisons test was used where applicable, as indicated in the corresponding figure legends. Categorical data were analyzed using the chi-square test.Pearson’s correlation coefficients (R values) were considered significant at P < 0.05, and 95% confidence intervals are displayed as shaded areas along the regression line. Unless otherwise indicated, statistical significance is denoted as *P < 0.05, **P < 0.01, ***P < 0.001.

Correspondence and requests for materials

All data necessary to support the conclusions of the study are either included in the paper or provided in the Supplementary Materials. The datasets generated and/or analyzed during the current study are available upon request from the corresponding author.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

L.H., M.L., Y.Z., T.L., and D.W. designed the research. L.H., Xh.Y., D.G., Xy.J., L.M., W.L., Lu.L., Y.W., Y.Y., C.Z., Li.L., Xm.J., X.H., C.J., G.L., X.Y., and B.F. performed the research. Xh.Y., D.G., Xt.Y., J. L., Z.S., L.C., S.X., Y.L., Z.F., and L.J. analyzed the data. L.H., Xh.Y., M.L., and D.W. wrote the paper. All authors reviewed and edited the manuscript.

Peer review

Peer review information

Nature Communications thanks Fredrick Joseph, Augusto Schneider, Shixuan Wang and other anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Data availability

All high-throughput sequencing datasets are deposited at the Genome Sequence Archive (GSA) (https://ngdc.cncb.ac.cn/gsa/) of the China National Center for Bioinformation National Genomics Data Center (CNCB-NGDC) under accession code CRA023378 (https://ngdc.cncb.ac.cn/gsa/search?searchTerm=CRA023378). The Visium HD high-resolution spatial transcriptomics image data has been uploaded to the Open Archive for Miscellaneous Data (OMIX) (https://ngdc.cncb.ac.cn/omix/) of the China National Center for Bioinformation National Genomics Data Center (CNCB-NGDC) under accession code OMIX010607 (https://ngdc.cncb.ac.cn/omix/release/OMIX010607). The Xenium image data has been uploaded to the Open Archive for Miscellaneous Data (OMIX) (https://ngdc.cncb.ac.cn/omix/) of the China National Center for Bioinformation National Genomics Data Center (CNCB-NGDC) under accession code OMIX014447 (https://ngdc.cncb.ac.cn/omix/release/OMIX014447). The metabolome data has been uploaded to the Open Archive for Miscellaneous Data (OMIX) (https://ngdc.cncb.ac.cn/omix/) of the China National Center for Bioinformation National Genomics Data Center (CNCB-NGDC) under accession code OMIX018414 (https://ngdc.cncb.ac.cn/omix/release/OMIX018414). All data supporting the findings described in this manuscript are available in the article and the Supplementary Information and from the corresponding author upon request, and that Source data are provided with this paper, in accordance with the journal’s requirements. Source data are provided with this paper.

Code availability

The code used for scRNA-seq and Visium HD dataset analysis is accessible at https://github.com/dfgao/Mouse_placenta_scRNA-HD.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-77217-5.