Single-Cell Atlas of Fetal Immune Development Across Lung, Spleen, and Umbilical Cord Blood in Nonhuman Primates
Ilhem Messaoudi, Brianna Doratt, Sheridan Wagner, Katelyn Keen, Uriel Avila, Oleg Varlamov
TL;DR
This study creates a detailed map of fetal immune cells in nonhuman primates across lung, spleen, and umbilical cord blood, revealing tissue-specific immune roles and communication patterns.
Contribution
The paper introduces a high-resolution single-cell atlas of fetal immune development in nonhuman primates across multiple tissues.
Findings
Fetal lung is enriched in myeloid populations and ILC2 cells, showing proinflammatory bias.
Splenic B cells exhibit strong V(D)J recombination and isotype switching signatures.
UCBMC is dominated by T-cells and shows a regulatory immune landscape.
Abstract
The fetal immune system develops within a tightly regulated environment that balances immune tolerance with readiness for postnatal antigen exposure. However, limited access to fetal tissues has constrained our understanding of immune ontogeny across distinct anatomical compartments. Here, we present a high-resolution, multi-tissue single-cell transcriptional atlas of the late-gestation (GD130–135) rhesus macaque (Macaca mulatta) fetal immune system, profiling leukocytes from lung, spleen, and umbilical cord blood mononuclear cell (UCBMC) compartments spanning myeloid, lymphoid, innate lymphoid, and hematopoietic stem cell (HSPC) lineages. The fetal lung was enriched in myeloid populations and ILC2 cells while fetal spleen was comprised primarily of T- and B-cells and UCBMC were dominated by T-cells. Despite reduced overall intercellular communication in lung compared to spleen and…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsSingle-cell and spatial transcriptomics · Immune responses and vaccinations · IL-33, ST2, and ILC Pathways
INTRODUCTION
The development of the human fetal immune system is a tightly orchestrated process that primes the neonate for protective responses against antigenic encounters after birth. Hematopoietic stem cells (HSC) emerge in the yolk sac as early as four weeks of gestation (1). By six weeks, macrophages rapidly develop in the yolk sac and begin to colonize fetal tissues (2, 3), coinciding with the fetal liver becoming the predominant site of hematopoiesis. The liver establishes a self-renewing pool of hematopoietic stem and progenitor cells (HSPC) and serves as a key site for macrophage development between weeks 9 and 13 (1, 4). After gestational week 11, HSC begin colonizing the fetal bone marrow, giving rise to the full hematopoietic lineage (5, 6), which continues to mature throughout gestation as lymphocyte compartments are progressively defined (4, 7).
Accumulating evidence demonstrates that the prenatal environment directly impacts neonatal health and shapes disease susceptibility into adulthood (8). The maternal inflammatory milieu and antigen exposure can mold fetal immunity; for example, maternal infection, even without vertical transmission, can alter neonatal immunity by impairing hematopoiesis (9–12). Human studies investigating immune perturbations in utero are limited by the inaccessibility of fetal tissues throughout pregnancy, often relying instead on postnatal peripheral samples, in vitro approaches, or small animal models (13–16). These strategies cannot fully capture the dynamic trajectory of human fetal immune system development. In contrast, the rhesus macaque (Macaca mulatta), with its phylogenetic proximity to humans, similar placental morphology and developmental trajectory, and gestational timeline (17–19), provides a valuable model for investigating prenatal immune development.
Here, we present a single-cell atlas of the immune transcriptional landscapes of fetal lung, spleen, and umbilical cord blood (UCB) in late-gestational (gestational day (GD) 130) rhesus macaques. UCB was analyzed in place of fetal blood given easier access and greater volume. The spleen was selected as a key secondary lymphoid organ with a microenvironment essential for immune development, and the lung as a non-hematopoietic tissue seeded early in gestation that directly encounters antigens at birth. By leveraging single-cell RNA sequencing (scRNA-seq), we identified diverse immune cell populations across compartments and uncovered putative regulatory circuits underpinning fetal immune development. This atlas provides a foundational framework to investigate molecular mechanisms shaping fetal immunity across gestation.
RESULTS
Immune cell frequencies are tissue-dependent in fetal lung, spleen, and umbilical cord.
Leukocytes isolated from fetal rhesus macaque spleen, lung, and UCB were analyzed via scRNA-seq to generate a fetal immune cell atlas. Using canonical immune markers, we identified 19 distinct leukocyte clusters, with substantial contribution from all three tissue sources (Fig. 1A,B). Within the myeloid compartment, we identified monocytes (CD14, MAMU-DRA, S100A9), macrophages (CD14, MAMU-DRA, MRC1), FLT3 positive plasmacytoid dendritic cells (pDC; CCR7, MAMU-DRA, TCF4, FLT3), and myeloid dendritic cells (mDC; MAMU-DRA, CD1C) (Fig. 1A,C). B-cells expressing MS4A1 (CD20) were subdivided into three groups according to the differential expression of PAX5, EBF1, and CD79A: Bcell_1 (PAX5low, EBF1low, CD79Ahigh), Bcell_2 (PAX5mid, EBF1mid, CD79Amid), and Bcell_3 (PAX5high, EBF1high, CD79Alow) (Fig. 1A,C). Two natural killer (NK) cell clusters were identified based on the expression of KLRB1 and NKG7 and further distinguished by FCGR3 (CD16) expression: CD16 + NK (KLRB1+, NKG7+, FCGR3+) and CD16− NK (KLRB1+, NKG7+, FCGR3−) (Fig. 1A,C). A cluster of natural killer T (NKT) cells was defined by positive ZBTB16 expression, and a cluster of T regulatory (Treg) cells was identified through positive CTLA4 expression (Fig. 1A,C). A cluster of type two innate lymphoid cells (ILC2) was identified by the absence of canonical myeloid and adaptive cell markers and the expression of GATA3 and IL1RL1 (Fig. 1A,C). T-cells were identified by expression of CD3E (Fig. 1A,C). Additional T-cell clusters were delineated into CD4 Naive (CD8A−, CD28mid, CCR7+, IL7R+, LEF1+), CD4 central memory (CM; CD8A−, CD28low, CCR7+, IL7R+, LEF1low), CD4 CAMK4 high (CAMK4; CD8A−, CD28high, CCR7low, IL7Rlow, CAMK4high), CD4 cytotoxic effector memory (EM_CTL; CD8A−, CCR7−, KLRB1+, NKG7+, IL7Rlow) and CD8 (CD8A+, CCR7+, IL7R+) cells (Fig. 1A,C). A proliferating T-cell cluster was identified by high MKI67 expression (Fig. 1A,C). Finally, a small cluster of hematopoietic stem/progenitor cells (HSPC) was distinguished by high CD34 expression, particularly in the spleen, representing extramedullary hematopoiesis (Fig. 1A,C).
Differences in cell cluster abundance were observed between the three tissue compartments. As expected, macrophages were detected only in lung and spleen, while ILC2 cells were found almost exclusively in the lung (Fig. 1B,D). Leukocyte populations in the lung were dominated by myeloid cells (50.3%), followed by T-cells (17.2%) and B-cells (10.8%) (Fig. 1D). The spleen was comprised primarily of T- and B-cell clusters (42.0% and 38.8% respectively), whereas UCBMCs were dominated by T-cells (83.0%) (Fig. 1D).
Increased B-cell maturation within the fetal spleen.
We first performed Gene Ontology (GO) enrichment on the marker genes for each B-cell cluster to delineate subset-specific differences, independent of tissue of origin (Fig. 2A). Marker genes in the Bcell_1 cluster mapped to B-cell receptor (BCR) signaling and immunoglobulin binding while the Bcell_2 cluster enriched for cellular homeostasis and nuclear transport, and the Bcell_3 cluster enriched for response to virus, response to type II interferon, and cellular respiration (Fig. 2A). These enrichments suggest that the Bcell_1 cluster, which was augmented in the fetal lung as indicated by normalization of B cell cluster frequencies (Fig. 2B), exhibited enhanced effector functionality compared to the homeostatic Bcell_2 cluster, while the Bcell_3 cluster enriched in the spleen and UCBMC (Fig. 2B) appeared primed for antiviral responses.
We next evaluated the tissue-specific transcriptional profile of the combined B cell clusters using gene set enrichment analysis (GSEA). Spleen B cell subsets were characterized by increased expression of genes important for B cell isotype switching, differentiation, and antigen processing and presentation compared to B cells in the lung and UCBMC, in line with the role of the spleen as a secondary lymphoid organ in the initiation of antimicrobial responses (Fig. 2C). In contrast, isotype switching and differentiation were significantly downregulated in the UCBMC while differentiation and proliferation were significantly downregulated in the lung (Fig. 2C).
To identify specific differences in transcriptional profiles of B-cells across fetal tissues, we performed DEG analysis comparing gene expression in one cluster from one tissue to the same cell cluster of the other two tissues and retaining only the genes with a positive fold change. Venn diagrams of upregulated genes were generated to identify DEGs unique to each tissue. While most DEGs were tissue-specific, 137 shared upregulated DEGs were detected in both Spleen vs. Lung+UCBMC and UCBMC vs. Lung+Spleen (Fig. 2D). DEG unique to the lung exclusively enriched to processes associated with oxidative phosphorylation (COX7C, ATP5PF) and oxidoreductase activity (GSTP1, ND4L) (Fig. 2E,F). Additional DEGs upregulated in the lung Bcell_1 cluster included the pro-apoptotic gene IL27L2 and the antiviral gene ISG20 (Fig. 2F). DEG unique to spleen enriched to B cell activation (PIK3CD, BLNK, BANK1), positive regulation of cell programmed death (JUN, LTB, BCL2), regulation of cell-cell adhesion (RUNX1, CD47, DOCK8), immunoglobulin recombination (PAX5, IKZF3, IL27RA), and cytokine production (NFKBIA, HIF1A, LTB) (Fig. 2E,F). DEG shared between spleen and UCBMC also enriched to B cell activation (PRKCB, PTPRJ), positive regulation of cell programmed death (FAF1, FOXO1), and stem cell population maintenance (FOXO1) (Fig. 2E,F). DEG unique to UCBMC enriched to regulation of cell-cell adhesion (CD44, LYN) and stem cell population maintenance (BRAF, MED21) (Fig. 2E,F).
Next, we examined the Bcell_3 cluster, which was more abundant in the spleen and UCBMC compared to the lung (Fig. 2B). Only six upregulated DEGs were unique to the lung, four of which were associated with mitochondrial functions (COX1, COX2, ND4, ND4L) (Fig. 2G,I, Sup. Table 2). Compared to UCBMC, Bcell_3 clusters in both lung and spleen exhibited significant upregulation of MEF2C, which limits leukocyte adhesion and migration (Fig. 2I). Although there were no shared upregulated DEGs between spleen and UCBMC, the unique genes in these tissues enriched to similar functions, including histone modification, transcription factor binding, and regulation of cell cycle process (Spleen: BCL11A, HIST1H2AC; UCBMC: KDM7A, JUND, NFKB1) (Fig. 2H,I). Spleen-unique GO terms included B-cell activation (CD19, CD22, CR2), tumor necrosis factor production (TRAF3IP3), and immunoglobulin recombination (BCL11A) (Fig. 2H,I). Other spleen-unique upregulated DEGs play a role in B-cell migration (ITGA4) as well as antigen presentation (MAMU-DBR1) (Fig. 2H,I). UCBMC-unique DEG were important for RNA splicing and receptor internalization (ITCH), B-cell adhesion (PECAM1), and inflammatory response (S100A6 and S100A10) (Fig. 2H,I).
T-cell activation and migration were prominent in the fetal spleen.
This section will focus on the CD4 T-cell compartment as our analysis of T-cell subsets revealed that CD4 T cells exhibited the most pronounced alterations across tissue. Enrichment of CD4_Naive and CD4_CM cluster marker genes indicated that CD4_Naive represents an activated subset, as evidenced by higher expression of genes involved in T-cell activation, antigen receptor signaling, and positive regulation of IL-2 production (CD4, CD28, RHOH, CASP8, TRAC, ICOS, STAT5B) (Fig. 3A, Sup. Table 1). In contrast, CD4_CM markers enriched to GO terms such as cellular respiration (ATP5F1C, COX7C) and TNF-α signaling (RPL6, RPL8, RPL30, RPS13) suggesting heightened metabolic and pro-inflammatory capacity (Fig. 3A, Sup. Table 1). Comparisons of T cell cluster frequencies indicated the CD4_Naive population was most abundant in UCBMC and least abundant in the lung (Fig. 3B). In contrast, the proportion of CD4_CM T-cells was highest in the lung and lowest in the spleen (Fig. 3B). The spleen contained higher proportions of both CD4_EM_CTL and Treg populations compared to lung and UCBMC while the lung had the lowest CD4_CAMK4 percentage (Fig. 3B). We also performed GSEA to identify broad functional programs of the two CCR7 + CD4 clusters (CD4_Naive & CD4_CM) within each tissue. In both the spleen and UCBMC, transcriptional signatures of CD4 CCR7 + cells showed increased T-cell proliferation, anergy, positive selection, and V(D)J recombination (Fig. 3C). Conversely, T-cell proliferation, anergy, positive selection, and V(D)J recombination were decreased in the lung (Fig. 3C). In UCBMC, CD4 T cell transcriptional profile was indicative of regulatory T cell differentiation (Fig. 3C). Overall, these data indicate that CD4 T cells were more activated in the spleen.
Next, we used the same DEG analysis strategy as described previously for the B cells. Within each cluster, we identified DEG upregulated in each tissue relative to the remaining two tissues. DEGs upregulated in the CD4_Naive cluster were largely unique to each tissue, with only six genes shared between Lung vs. Spleen+UCBMC and Spleen vs. Lung+UCBMC (Fig. 3D). DEGs unique to the lung and spleen played a role in T-cell activation (Lung: CCR7, CD3D, ICOS; Spleen: CD2, FOS, JUN) (Fig. 3E,F). CD4_Naive cells in the lung expressed high levels of the antiviral gene IFI16 and numerous T-cell developmental transcription factor genes (IKZF1, BACH2, SOX4, STAT4) (Fig. 3F). DEGs unique to the spleen enriched to positive regulation of immune response and cytokine production (IL16, IL17RA, IL27RA, LTB), key transcription factors (BCL11B, ID3, IRF2, KLF6, JUN), and facilitators of TCR activation (CD40LG, TRAF3IP3) (Fig. 3E,F). DEGs unique to UCBMC enriched to GO terms associated with protein degradation, epigenetics and cell cycle (RELB, EIF2AK3, CASP3 FBXO33, HERC1) (Fig. 3E,F).
DEGs in CD4_CM cell population were also largely tissue-specific (Fig. 3G). DEGs unique to the lung mapped to oxidative phosphorylation (COX7B), T-cell receptor signaling (CD3D, CD7), and antiviral immunity (IFI16) (Fig. 3H,I). DEGs upregulated in the spleen and UCBMC enriched to chromatin binding (Spleen: FOS, IRF2; UCBMC: KDM2A, ARID1A, CREBBP; Spleen/UCBMC: RUNX1) (Fig. 3H,I). Genes uniquely upregulated in the spleen included those important for T-cell activation, signaling, and cytokine response (CD38, FOS, IL27RA, IL6ST, TGFBR2, and TRAC) (Fig. 3I). DEGs unique to UCBMC enriched to regulation of stem cell population maintenance (FOXO1, FOXP1, CTNNB1, MYC, RHOH) (Fig. 3H,I).
NK cells had higher cytotoxic capacity within the fetal lung.
Comparisons of the frequency of the innate, HSPC, and proliferating clusters revealed the lung harbored the largest proportion of CD16 + NK cells and ILC2 cells, while the spleen was home to the largest proportion of CD16− NK cells (Fig. 4A). Finally, the NKT subset was most prominent in the UCBMC (Fig. 4A). Marker genes of both CD16 + and CD16− NK clusters enriched to GO processes associated with oxidative phosphorylation and response to cytokine stimulus (Fig. 4B). Given the role of CD16 + NK cells in ADCC, we noted the enrichment to NF-kB signal transduction, which is crucial for NK cell IFN-gamma production and cytotoxicity (27, 28) (Fig. 4B). Marker genes of CD16− NK cells uniquely enriched to positive regulation of cytokine production, consistent with their cytokine-producing function and reduced ADCC capability compared to CD16 + NK cells (29) (Fig. 4B). GSEA analysis of the combined NK cell clusters showed upregulation of NK cell differentiation and antigen processing/presentation in the spleen and UCBMC (Fig. 4C). On the other hand, leukocyte-mediated cytotoxicity and immunoglobulin-like receptor signaling pathways were upregulated in the lung (Fig. 4C).
For CD16 + NK cells, the lung displayed 165 unique upregulated DEGs while the majority of UCBMC defining DEG were shared with the spleen (Fig. 4D). DEGs in the lung CD16 + NK cluster uniquely enriched to oxidative phosphorylation (COX7C) and NK cell-mediated cytotoxicity (GZMB, KLRB1, NCR3, PRF1) (Fig. 4E,F). Spleen-unique and spleen/UCBMC-shared genes were associated with antigen receptor-mediated signaling and cytokine production (Spleen: IFNG, ITGAM, NKG2D; Spleen/UCBMC: IL2RB, JAK1) as well as nuclear receptor binding (FOXP1, DDX5, NCOR1) and phospholipid binding (RAPGEF2, ITPR2) (Fig. 4E,F). Spleen-unique DEGs were also associated with regulation of inflammation and apoptosis (TNFAIP3, IRF1) (Fig. 4F).
DEG that define the CD16− NK cells were distinct in the spleen and lung while those that defined UCBMC were largely shared with the spleen (Fig. 4G). Lung-unique DEGs enriched to NK cell-mediated cytotoxicity (GZMB, NKG7), antiviral immunity (IFI16, IFI27L2, BST2), and chemotaxis (CCL3, VIM) (Fig. 4H,I). DEG unique to the spleen enriched to leukocyte activation (BCL2, FOXP1, IKZF3), response to type II IFN (IFNG), antigen-receptor-mediated signaling pathway (CD74), and epigenetic regulation (HDAC9, KDM2B, FOS, FOXO3, IKZF3, IRF1, JUN) (Fig. 4H,I). DEG shared between spleen and UCBMC enriched to antigen-receptor-mediated signaling pathway, epigenetic regulation, and intracellular protein transport (Fig. 4H). DEGs specific to UCBMC suggested an anti-inflammatory phenotype as indicated by increased expression of TGFB1 and apoptosis-inducing receptor TNFRSF10A (Fig. 4I).
Lung monocytes are primed for anti-microbial response
Macrophage and pDC clusters were exclusively detected in the lung and spleen with the spleen containing the largest frequency of mDCs, pDCs, and macrophages (Fig. 5A). GSEA analysis of the monocyte population revealed that TLR and complement signaling were uniquely upregulated in lung monocytes (Fig. 5B). On the other hand, antigen processing/presentation and phagocytosis terms were enriched in both spleen and UCBMC (Fig. 5B). Finally, extravasation pathways were upregulated in both lung and spleen monocytes (Fig. 5B). We observed a large overlap between upregulated DEGs defining the spleen and UCBMC monocytes while lung-defining DEG were distinct (Fig. 5C). As expected, DEG from all three tissue types included genes associated with innate immunity (Lung: C5AR1, TLR4, ISG20, NOD2; Spleen: C1QBP, IFNGR1, IRF1; Spleen/UCBMC: IRF8; UCBMC: JAK1) (Fig. 5D,E). DEGs unique to lung monocytes enriched to electron transport chain and cytokine activity (CCL4L1, CXCL3, IL1B, TNFRSF1B) (Fig. 5D,E). DEGs restricted to spleen monocytes enriched to histone modifying activity (HDAC9, KDM2B) and included upregulated genes encoding adhesion molecules (ITGAL, SELL) (Fig. 5D,E). Notable DEGs shared between spleen and UCBMC monocytes were involved in activation (FOS), response to oxidative stress (FOXO3), immune modulation (ILRUN), and immune tolerance (IRAK3) (Fig. 5E). Expression of genes associated with cytokine signaling (IL1RAP, JAK1, TGFB1) was increased in UCBMC monocytes (Fig. 5E). Module scoring of the tissue-specific monocyte populations revealed lung monocytes had highest scores for inflammation while splenic monocytes had the highest score for wound healing, and UCBMC monocytes displayed significantly lower viral/bacterial response (Fig. 5F).
DEG analysis between lung and spleen macrophages (Fig. 5G) indicated that genes upregulated in the lung are involved in proinflammatory response (NLRP3, TLR2, IL1B), response to interferon (IFI30, IL1B), chemotaxis (CCL3, CXCL3), phagosome activity (VAMP8), and wound healing (ANXA1, AREG), (Fig. 5H,I,). Genes upregulated in the spleen mapped to macrophage activation (CD163, CD68, PECAM1), regulation of tumor necrosis factor production (PYCARD), phagocytosis (C1QB, CD163), and pattern recognition receptor signaling (C1QB, FCGR3, IFNGR1) (Fig. 5H,I).
Inferred cell-cell communication is reduced within the lung; while UCBMC displays pronounced HSPC signaling regulation
Next, we used CellChat to identify unique cell-cell communications occurring within each tissue (26). The fetal spleen had the highest number and strongest ligand-receptor interactions, followed by UCBMC, and then the lung (Fig. 6A). Similarly, the incoming and outgoing signal strengths, based on the inferred probabilities of ligand-receptor interactions, were higher for all spleen clusters and most UCBMC clusters compared to the lung (Fig. 6B). Interestingly, HSPCs within UCBMC exhibited nearly twice the incoming interaction strength compared to those in the spleen and lung (Fig. 6B). Additionally, both the CD16 + and CD16− NK cell clusters had increased incoming and outgoing interaction strength in the spleen compared to both lung and UCBMC (Fig. 6B).
Next, we quantified the relative information flow, an aggregate measure of signal number and strength for the selected pathways to identify those that were shared across tissues or unique (Fig. 6C). In the lung, signaling pathways involved in pathogen response (COMPLEMENT), immune cell chemotaxis (PLAU), and proinflammatory response (IL1) were prominent (Fig. 6C). The COMPLEMENT pathway was primarily driven by macrophage signaling within the lung (Fig. 6D). Lung PLAU signaling was predicted to be mediated by macrophages and pDC, while IL1 signaling was driven by all myeloid cells (Fig. 6D).
Signaling pathways that dominated in the spleen were involved in inflammation (TNF, IGF, PTPR, IFN-II), B-cell survival (BAFF, APRIL), chemotaxis (CXCL, CCL), T-cell activation (PD-L1, CD96) and hematopoiesis (FTL3, BAFF, PD-L1) (Fig. 6C). We then evaluated the relative signaling strength of tissue specific cell clusters for each signaling pathway (Fig. 6D). TNF, PTPR, FLT-3, and PD-L1 pathways were primarily driven by signaling from pDC, while the chemotaxis-specific pathways (CXCL and CCL) were predominantly originating from macrophages and CD16− NK cells (Fig. 6D). Finally, BAFF and APRIL pathways were predominantly driven by signaling from the Bcell_2 cluster (Fig. 6D). UCBMC pathways were dominated by Colony stimulating factor (CSF), SELL, CD30 and ANGPT signaling, with higher relative strength and contribution by HSPC (Fig. 6C,D). SELL and ANGPT signaling pathways were absent from the lung, while CD30 had contribution from the lung CD4_CAMK4 cluster which was lessened in the spleen and UCBMC (Fig. 6D). FLT3 signaling which contributes to progenitor cell growth and division was prominent in splenic pDCs and UCBMC mDCs (Fig. 6D).
We evaluated the overall CCL pathway signaling between clusters in each tissue and observed signaling from both splenic NK cell clusters targeting monocytes, macrophages, and mDCs (Fig. 7A). In contrast, fetal lung CCL signaling exclusively targeted NKT cells, whereas UCBMC overall CCL signaling was predominantly directed from NK cells to monocytes (Fig. 7A). In the fetal lung, IL1 signaling was prominent among Bcell_3, monocyte, macrophage, mDC, pDC, and Proliferating clusters (Fig. 7B). In the spleen, IL1 signaling originated from monocyte, macrophage, and mDC clusters, targeting Treg, mDC, pDC, Proliferating, and HSPC clusters (Fig. 7B). Both pathways are important regulators of HSPC function, with MIF preventing accumulation of HSPCs within the bone marrow and TGFB regulating commitment to cell lineage fates. Notably, MIF signaling in UCBMC originated from the lymphoid compartment and targeted HSPCs, while lung and spleen lacked MIF signaling targeting the HSPC cluster (Fig. 7C). In contrast, MIF signaling in the spleen originated from HSPCs to B-cell and myeloid clusters (Fig. 7C). TGFb signaling within the lung and spleen also lacked the targeting to the HSPC cluster which was present in UCBMC, particularly from the lymphoid clusters (Fig. 7D).
DISCUSSION
This study aimed to fill gaps and serve as a reference for fetal immune cell heterogeneity using the rhesus macaque, a well-established animal model of human immunology (17). The study aimed to generate a foundational single-cell dataset from the late-gestational fetal rhesus macaque encompassing three key tissues (blood, spleen, lung) which can be later expanded to include further gestational timepoints and additional tissues. We performed differential gene expression, gene set enrichment, and cell-cell interaction analyses on the integrated dataset to demonstrate its applicability.
The lung is a crucial immunological organ as it is continuously exposed to respiratory pathogens, environmental pollutants, and allergens after birth (32). While maintaining its central role in gas exchange, the lung harbors tissue-resident immune cells that are essential for host defense and must balance immunoregulatory and anti-inflammatory responses to prevent tissue damage. Prominent immune cell populations residing within the lung tissue include alveolar macrophages, interstitial macrophages, innate lymphoid cells (ILCs), and NK cells. Consistent with previous studies (33, 34), we observed a prominent proportion of ILC2 in the fetal lung. These cells produce Th2 cytokines contributing to fetal lung development, the formation of lymphoid tissue, and modulation of the neonatal response to the external environment (35–37). Our transcriptional analysis demonstrated that the fetal lung tissue contained the highest proportion of myeloid cells compared to spleen and UCBMC. Monocytes within the fetal lung were metabolically active, with heightened bacterial responsiveness and patterns associated with molecular recognition signaling. Additionally, our data demonstrated that lung macrophages have enhanced chemotaxis and myeloid cell recruitment capacity. Interestingly, genes upregulated in fetal lung relative to spleen/UCBMC also enriched to immune tolerance, which potentially limits overactivation, and maintain tolerance to the exposure of inhaled antigens at birth.
Fetal lung macrophages showed increased expression of ANXA1, which is involved in positive regulation of macrophage efferocytosis and an anti-inflammatory phenotype (41, 42). ANXA1 signaling has been shown to resolve inflammation and promote the resolution of infections (43–45), highlighting its potential role in limiting neonatal inflammatory responses to novel antigens encountered after delivery. Fetal lung macrophages also upregulated IL1B, a notable proinflammatory cytokine. The production of IL-1b within the fetal lung by myeloid cells has been linked to epithelial development and lung tissue maturation (46). Additionally, CellChat analysis of intercellular communication revealed prominent IL1 signaling between lung cell clusters. While IL1 signaling contributes to antiviral immunity, overproduction can cause detrimental hyperinflammation, highlighting the importance of coupled anti-inflammatory signals such as ANXA1 (48). Collectively, our data indicate that the late-term fetal lung is capable of mounting an initial immune response while remaining highly regulated to maintain self-tolerance and anti-inflammatory capacity prior to delivery.
The spleen is a prominent secondary lymphoid organ, which contains red pulp, that filters the circulating blood via phagocytosis of damaged erythrocytes by macrophages, and white pulp, that initiates defense against pathogens (49). Within white pulp is the periarteriolar lymphoid sheath where splenic T-cells are activated in response to blood-borne antigens and B-cell germinal centers, which elicit T-cell dependent antibody production (49). Consistent with its role as a prominent site of B-cell maturation, the fetal spleen exhibited the highest proportion of B-cells among the tissues studied (50). Critical B-cell signaling pathways BAFF and APRIL which influence B-cell survival and differentiation were solely observed within the spleen further aligning with the anticipated tissue functionality (51, 52). Additionally, our DEG analysis showed an upregulation of immunoglobulin receptor recombination and isotype switching within splenic B-cells. T-cells were the second most prominent population within the fetal spleen. Although the frequency of CD4 populations was lower compared to UCBMC, splenic CD4 T cells were activated. The increased CD4 activation alongside the higher percentage of CD4 effector memory CTL aligns with the spleen’s function as a secondary lymphoid organ (53). The activation observed within the spleen could provide rationale for the higher proportion of splenic Treg cells. Although previous studies have mainly focused on the role of maternal Tregs in maintaining pregnancy, fetal Tregs also contribute to alloantigen tolerance (54). Our CellChat analysis showed the highest predicted signaling strength in the spleen, reflecting its role as a central site for peripheral immune activation and maturation (55). The strength of chemoattractant IL16 signaling to splenic T-cell clusters was notably higher, particularly Tregs, suggesting active recruitment into the spleen (56, 57). Widespread CCL signaling was predicted via CellChat within the spleen between innate, CD4_EM_CTL, Bcell_3, HSPC, and proliferating clusters. Furthermore, CCL/CCR signaling is a key contributor to splenic tolerance and infection responses, as these ligand interactions drive myeloid cell recruitment and macrophage polarization (58). In particular, signaling through CCR1 directs macrophage migration and polarization toward an anti-inflammatory M2 phenotype (58).
Blood mononuclear cells are key mediators of systemic immune responses, as they encompass mobile populations poised for activation and proliferation (60). UCB is often employed in clinical studies as a noninvasive substitute for neonatal blood due to its accessibility (61, 62). Inflammatory perturbations within maternal circulation caused by diseases and disorders such as pre-eclampsia and SARS-CoV-2 infection are reflected in neonatal UCB (63, 64). Fetal UCBMC contain immune cells predominantly in a state of tolerance to prevent immune overactivation prior to delivery (65). We found that UCBMC displayed heightened HSPC signaling strength and upregulation of genes important for stem cell population maintenance particularly among CD4 T cells. This finding aligns with previous studies showing that UCB is a rich source of HSPCs, contributing to fetal immune cell development (66). Notably, fetal spleen and UCB shared numerous upregulated DEGs in B-cell, NK, and monocyte clusters, which were absent in lung cells. Even in the absence of pathogens, UCBMC have shown heightened cytokine responsiveness following viral infection (67). The role of UCBMC and spleen cells in infection control was supported by our observations of increased expression of genes essential for activation and cytokine production.
This study has several limitations. First, animals were not perfused, restricting our ability to distinguish between tissue resident leukocytes and the infiltrating peripheral blood cells. Second, our single-cell profiling was focused on only three fetal tissues. Expansion with publicly available fetal rhesus macaque transcriptional datasets will be needed to incorporate other prominent immunological tissues such as fetal bone marrow, liver, and thymus. Third, the study focused solely on transcriptomic data, constraining the interpretation of cellular functions and interactions. Finally, all fetal samples were derived from macaques at a single gestational age (approximately GD130), which restricts insights into developmental dynamics across gestation. These limitations highlight the need for future studies with larger, more diverse tissue sampling, integrated multi-omics, and functional validation across broader developmental time points. Nevertheless, this initial study provides the first in-depth immune single-cell atlas of the late gestational fetal Macaca mulatta across lung, spleen, and umbilical cord blood. Utilization of this atlas will allow for the cellular characterization of healthy fetal immunity in the Rhesus Macaque model.
METHODS
Animal studies.
Eight healthy female rhesus macaques underwent time-mated breeding at the Oregon National Primate Research Center (ONPRC) resulting in ten fetal macaques (40% Female). All macaques in this study were managed according to the ONPRC animal care program, which is fully accredited by AAALAC International and is based on the laws, regulations, and guidelines set forth by the United States Department of Agriculture (e.g., the Animal Welfare Act and Animal Welfare Regulations, the Guide for the Care and Use of Laboratory Animals, 8th edition [Institute for Laboratory Animal Research]) and the Public Health Service Policy on Humane Care and Use of Laboratory Animals. Animals received ad libitum access to food (Purina 5000 Fiber-balanced Monkey Diet, Purina Mills, Richmond, IN, USA) and fresh water. Animals were fed a diet formulated according to National Research Council recommendations, supplemented with fruits and vegetables, provided through the Behavioral Services Unit’s environmental enrichment program. Animals were maintained in pair housing.
Sample collection and processing.
Fetal samples were obtained via scheduled cesarian section (C-section) between GD130 and GD135 as described to be representative of third trimester human development (20, 21). UCB was collected at C-section in EDTA tubes and UCBMC and plasma were collected following centrifugation over a Ficoll gradient (Lymphoprep; STEMCELL). Final cell counts were obtained, and cells were cryopreserved in 10% DMSO/FBS. Fetal spleen was collected at necropsy and immediately placed on ice in RPMI supplemented with 10% fetal bovine serum (FBS), streptomycin/penicillin, and L-glutamine. Splenic leukocytes were isolated by mechanical disruption. Cells were centrifuged and red blood cells were lysed using 0.84% ammonium chloride pH 7.4, followed by several washes. Splenic leukocytes were cryopreserved in 10% DMSO/FBS. Lung tissue was collected and processed in RPMI-1640 media supplemented with 3% BSA, 1% Penicillin–Streptomycin, 1% L-glutamine, and 10 mM HEPES pH 7.4 (R3 medium). Lung tissue was subjected to enzymatic digestion using 120 mg collagenase II (Gibco), 2.5 mg elastase (Sigma-Aldrich), 40 mg DNase I (Sigma-Aldrich), and 12 mg hyaluronidase (Sigma-Aldrich) in R3 medium and supplemented with 80 μL of 1 M CaCl_2_ for 1 hour at 37°C with gentle rotation. Remaining tissue was mechanically dissociated. Cells were pelleted and subjected to density separation using a discontinuous 60 − 30% Percoll gradient. The gradient was centrifuged at 2500 rpm for 30 minutes. Cells located at the interface between the 30% and 60% Percoll layers were collected and washed in R3 medium. Final cell counts were obtained, and cells were cryopreserved in CryoStor CS10 at a density of less than 20 × 10^6^ cells per vial.
Single-cell RNA sequencing library generation.
Leukocytes were thawed, washed in 2% FBS/DPBS, and incubated with Rhesus Fc Block (Invivogen). Each sample was labeled with a distinct TotalSeq Hashtag Oligo Antibody (HTO, Biolegend) and incubated per manufacturer’s instruction. Pellets were washed twice in 2% FBS/DPBS and pooled by tissue. Cell pools were filtered and counted in duplicate to confirm a viability greater than 80%. Samples with less than 80% viability (spleen and lung) were stained with anti-Rhesus CD45 BV650 (BD Biosciences) and propidium iodine before being sorted for live CD45 + leukocytes using a Sony SH800 Cell Sorter System. Cells were filtered and suspended in 2% FBS/DPBS to a final concentration of 1500–1600 cells/mL. Cell suspensions were then immediately loaded on the 10x Genomics Chromium X Controller Chip G with a target of 30,000 cells. Libraries were prepared using the V3.1 chemistry for gene expression and Single Cell 3 Feature Barcode Library Kit per the manufacturer’s instructions (10x Genomics). Libraries were sequenced on Illumina NovaSeq X with a sequencing target of 30,000 gene expression reads and 5,000 feature barcoding reads per cell.
Single-cell RNA-seq data analysis.
Raw sequencing reads were aligned and HTOs were demultiplexed using Cell Ranger (v6.0.2, 10x Genomic) against the Macaca mulatta reference genome (mmul10) using the multi-option. Downstream processing of aligned reads was performed using Seurat (v5.1.0) (22). Droplets with ambient RNA or potential doublets (< 400 or > 4000 detected genes) and inviable cells (> 20% total mitochondrial gene expression) were excluded during initial QC. Data objects from all libraries were integrated using Harmony (23). Data normalization and variance stabilization were performed on the integrated object using the NormalizeData and ScaleData functions in Seurat (v5.1.0), where a regularized negative binomial regression was corrected for differential effects of mitochondrial and ribosomal gene expression levels. Dimensionality reduction was performed using RunPCA function to obtain the first 30 principal components and clusters visualized using Seurat’s RunUMAP function. Cell types were assigned to individual clusters using FindAllMarkers function (Supplemental Table 1) with a log2 fold change ± 0.4, FDR < 0.05, and canonical scRNA markers for NHP leukocytes.
Differential gene expression (DEG) analysis for each cluster was performed using the FindMarker function in Seurat comparing each tissue to the other two tissues and retaining only the genes with a positive fold change. For example, DEGs in the lung utilized the lung as ident.1 and both spleen and UCBMC as ident.2. Only statistically significant genes maintaining an FDR < 0.05 and a log2 fold change > 0.585 were included in downstream analyses (Supplemental Table 2). Functional enrichment was performed using Metascape (24).
Gene set enrichment analysis (GSEA) was performed using the irGSEA (25) package (v2). Briefly, the irGSEA.score function was used to calculate the enrichment score for each cell for genes sets in the MsigBD gene ontology (GO) GO0002376 Immune processes. The irGSEA.integrate function was then used to perform a Wilcox test of the enrichment score matrix comparing the tissues compartments. The heatmaps were visualized using the irGSEA.heatmap functions.
The R CellChat (26) package was employed to infer probable intercellular communication networks. A CellChat object was generated from a Seurat v5 object with the createCellChat function. The CellChatDB.human database was utilized. Data was preprocessed with the function identifyOverExpressedGenes and identifyOverExpressedInteractions and communication probabilities between clusters were determined with computeCommunProb (truncatedMean, trim = 0.1, interaction.range = 250, contact.range = 100). Communications were filtered (filterCommunication) to a minimum number of 10 cell and signaling pathway probabilities were calculated (computeCommunProbPathway).
Supplementary Material
Supplementary Files
This is a list of supplementary files associated with this preprint. Click to download.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Popescu DM, Botting RA, Stephenson E, Green K, Webb S, Jardine L (2019) Decoding human fetal liver haematopoiesis. Nature 574(7778):365–37131597962 10.1038/s 41586-019-1652-y PMC 6861135 · doi ↗ · pubmed ↗
- 2Ginhoux F, Guilliams M (2016) Tissue-Resident Macrophage Ontogeny and Homeostasis. Immunity 44(3):439–44926982352 10.1016/j.immuni.2016.02.024 · doi ↗ · pubmed ↗
- 3Hoeffel G, Ginhoux F (2018) Fetal monocytes and the origins of tissue-resident macrophages. Cell Immunol 330:5–1529475558 10.1016/j.cellimm.2018.01.001 · doi ↗ · pubmed ↗
- 4Feyaerts D, Urbschat C, Gaudilliere B, Stelzer IA (2022) Establishment of tissue-resident immune populations in the fetus. Semin Immunopathol 44(6):747–76635508672 10.1007/s 00281-022-00931-x PMC 9067556 · doi ↗ · pubmed ↗
- 5Charbord P, Tavian M, Humeau L, Peault B (1996) Early ontogeny of the human marrow from long bones: an immunohistochemical study of hematopoiesis and its microenvironment. Blood 87(10):4109–41198639768 · pubmed ↗
- 6Jardine L, Webb S, Goh I, Quiroga Londono M, Reynolds G, Mather M (2021) Blood and immune development in human fetal bone marrow and Down syndrome. Nature 598(7880):327–33134588693 10.1038/s 41586-021-03929-x PMC 7612688 · doi ↗ · pubmed ↗
- 7Park JE, Jardine L, Gottgens B, Teichmann SA, Haniffa M (2020) Prenatal development of human immunity. Science 368(6491):600–60332381715 10.1126/science.aaz 9330 PMC 7612900 · doi ↗ · pubmed ↗
- 8Jazwiec PA, Sloboda DM (2019) Nutritional adversity, sex and reproduction: 30 years of DO Ha D and what have we learned? J Endocrinol 242(1):T 51–T 6831013473 10.1530/JOE-19-0048 · doi ↗ · pubmed ↗