How does fetal immunity differ across lung, spleen, and cord blood in primates?

Three compartments, one late-gestation snapshot

The study asks whether the fetal immune system is uniform across the body or organized differently by tissue before birth. To answer this, the authors built a single-cell transcriptional atlas of leukocytes from three sites in the rhesus macaque fetus: lung, spleen, and umbilical cord blood mononuclear cells (UCBMC). Cord blood was used in place of fetal blood because it is easier to access, yields more volume, and is commonly used as a surrogate for neonatal peripheral blood.

The timing matters for interpreting every result. Fetuses were delivered by scheduled cesarean section between gestational day (GD) 130 and GD 135. Full-term rhesus gestation is about 165 days, so this is roughly 80% of term, corresponding to the third trimester and an estimated human age of about 32 weeks. The atlas therefore describes a late-gestation snapshot, not the whole trajectory of immune ontogeny.

The sample base was ten time-mated female macaques producing thirteen fetuses (46% female). Leukocytes were profiled with 10x single-cell RNA sequencing, and the transcriptional findings were then checked with spectral flow cytometry and functional assays measuring immune mediator secretion. That combination is what lets the paper move from cell frequencies and gene signatures to statements about what the cells actually do.

The fetal lung leans myeloid and proinflammatory

In the sequencing data, lung leukocytes were dominated by myeloid cells at 50.3%, followed by T cells at 17.2% and B cells at 10.8%. ILC2 cells, identified by GATA3 and IL1RL1 expression and the absence of canonical myeloid and adaptive markers, were found almost exclusively in the lung. Macrophages were detected only in lung and spleen.

Lung monocytes carried upregulated Toll-like receptor and complement signaling, and lung-unique differentially expressed genes enriched for electron transport chain and cytokine activity, including CCL4L1, CXCL3, IL1B, and TNFRSF1B. Module scoring put lung monocytes highest for inflammation. The authors also note that genes upregulated in fetal lung relative to spleen and UCBMC enriched for immune tolerance, which they interpret as a possible brake on overactivation to inhaled antigens at birth.

The functional assays support a resting proinflammatory bias. Without stimulation, fetal lung cells secreted broadly higher levels of immune mediators than spleen and UCBMC cells. After stimulation, however, lung cells showed a smaller response, while spleen and UCBMC showed greater inducible responses. Resting secretion and stimulated response are different measurements, and the paper's point depends on keeping them separate.

The spleen hosts B cell maturation and activated CD4 T cells

The spleen was composed primarily of T and B cell clusters, at 42.0% and 38.8% respectively. Its B cells showed enhanced signatures of V(D)J recombination and isotype switching, and its CD4 T cells showed increased activation and a higher Treg frequency. A small HSPC cluster marked by high CD34 expression was found particularly in the spleen, consistent with extramedullary hematopoiesis.

One discrepancy is worth flagging because it affects how the composition numbers should be read. Sequencing showed the lung had the highest myeloid proportion, but flow cytometry indicated a higher macrophage portion in the spleen. The authors offer several explanations: the sequencing was performed on sorted live CD45+ cells while the flow panel did not include CD45, single-cell resolution may classify cell states differently than flow, and fragile red pulp and marginal zone macrophages may be lost during preparation and quality control. The two methods therefore do not give interchangeable cell-frequency estimates.

Cord blood T cells carry a regulatory landscape

UCBMC was dominated by T cells at 83.0%. Its transcriptional profile pointed toward regulation rather than inflammation: UCBMC-specific differentially expressed genes suggested an anti-inflammatory phenotype, marked by increased TGFB1 and the apoptosis-inducing receptor TNFRSF10A. UCBMC monocytes also showed increased expression of cytokine-signaling genes IL1RAP, JAK1, and TGFB1, and had significantly lower viral and bacterial response module scores.

The functional data add a practical constraint. Viral stimulation was performed only on lung and spleen cells because limited UCBMC numbers precluded it. Any comparison of antiviral responses therefore cannot include cord blood, and the cord blood picture rests on bacterial and PMA/Ionomycin stimulation plus its resting transcriptional state.

NK cells split by tissue and by CD16 status

Two NK clusters were defined by KLRB1 and NKG7 expression and then separated by FCGR3 (CD16): CD16+ and CD16− NK cells. Their tissue-specific gene programs differed. Lung CD16+ NK cells uniquely enriched for oxidative phosphorylation (COX7C) and NK cell-mediated cytotoxicity (GZMB, KLRB1, NCR3, PRF1). Lung CD16− NK cells uniquely enriched for cytotoxicity (GZMB, NKG7), antiviral immunity (IFI16, IFI27L2, BST2), and chemotaxis (CCL3, VIM).

Spleen-unique NK genes pointed in a different direction, toward leukocyte activation, type II interferon response, antigen-receptor signaling, and epigenetic regulation. UCBMC-defining NK genes largely overlapped with the spleen, and UCBMC-specific genes suggested an anti-inflammatory phenotype through TGFB1 and TNFRSF10A. The practical implication is that an NK cell's tissue of origin, and its CD16 status, change which functional program its transcriptome implies.

Lung networks are smaller but inflammatory

Cell-cell communication analysis found reduced overall intercellular communication in the lung compared with spleen and UCBMC, yet the lung networks were marked by a proinflammatory bias. Prominent IL1 signaling appeared between lung cell clusters. The authors connect this to IL-1β production by fetal lung myeloid cells, which has been linked to epithelial development and lung maturation, while noting that overproduction can cause hyperinflammation and that coupled anti-inflammatory signals such as ANXA1 may restrain it.

These are predicted communication networks inferred from expression data, not measured signaling events. They are best read as hypotheses about which cell subsets could talk to which, and the paper's own framing treats them as a framework for future mechanistic work rather than as demonstrated causal interactions.

What the atlas supports and what it does not

The strongest claim the data support is comparative: at late gestation, the three compartments are specialized rather than interchangeable. Lung is myeloid- and ILC2-enriched with high resting mediator secretion but a blunted stimulated response; spleen carries B cell maturation and activated CD4 T cells; UCBMC is T cell-dominated with a regulatory profile. Flow cytometry and functional assays corroborate these tissue-specific differences, which is why the findings are not purely transcriptional.

Several boundaries limit transfer. The study covers a narrow late-gestation window, so it cannot describe how these compartments changed earlier or how they shift after birth. It uses rhesus macaque, a model chosen for phylogenetic proximity, similar placental morphology, and comparable gestational timing, but the results are not human measurements. Cord blood substitutes for fetal blood. Viral stimulation was not performed on UCBMC. And where sequencing and flow disagree on macrophage frequency, the discrepancy is unresolved.

Sex effects appeared only in a limited form: significant sex-by-concentration interactions were identified for IFNγ and CXCL2 in uncorrected stimulated mediator concentrations, with male fetal spleen showing reduced IFNγ secretion after bacterial and viral stimulation, and no significant sex interactions were detected for corrected concentrations. The paper's most useful takeaway is therefore a comparison rule: when evaluating fetal immune readiness, specify the compartment and the measurement condition, because resting secretion, stimulated response, and transcriptional signature can point in different directions for the same tissue.

Sources used in this answer