Unique maternal immune and functional microbial profiles during prenatal stress

Abstract

Maternal stress during pregnancy is widespread and is associated with poor offspring outcomes, including long-term mental health issues. Prenatal stress-induced fetal neuroinflammation is thought to underlie aberrant neurodevelopment and to derive from a disruption in intrauterine immune homeostasis, though the exact origins are incompletely defined. We aimed to identify divergent immune and microbial metagenome profiles of stressed gestating mice that may trigger detrimental inflammatory signaling at the maternal-fetal interface. In response to stress, maternal glucocorticoid circuit activation corresponded with indicators of systemic immunosuppression. At the maternal-fetal interface, density of placental mononuclear leukocytes decreased with stress, yet maternal whole blood leukocyte analysis indicated monocytosis and classical M1 phenotypic shifts. Genome-resolved microbial metagenomic analyses revealed reductions in genes, microbial strains, and metabolic pathways in stressed dams that are primarily associated with pro-inflammatory function. In particular, disrupted Parasutterella excrementihominis appears to be integral to inflammatory and metabolic dysregulation during prenatal stress. Overall, these perturbations in maternal immunological and microbial regulation during pregnancy may displace immune equilibrium at the maternal-fetal interface. Notably, the absence of and reduction in overt maternal inflammation during stress indicates that the signaling patterns driving fetal outcomes in this context are more nuanced and complex than originally anticipated.

Figure 1

Restraint stress restricts gestational weight gain. The study design is depicted in (a). While body weight gain (b) positively correlated with litter size regardless of treatment (NS: R² = 0.27, p = 0.01; S: R² = 0.34, p = 0.001; linear regression slope comparison, p = 0.71), linear regression y-intercepts differed across treatment groups (p = 0.04), reflecting a reduced weight gain from GD0 to GD16 with stress. This is further reflected when weight gain is normalized to litter size, revealing a (c) decreased weight gain per pup in the stressed group. Data are mean ± SEM. NS = non-stressed; n = 22, S = stressed; n = 27. *p < 0.05.

Figure 2

Restraint stress impacts systemic glucocorticoid and immune circuits. Stressing pregnant dams (a) increased serum corticosterone and (b) upregulated expression of CRH in the hypothalamus and glucocorticoid receptor (NR3C1) in the amygdala (p = 0.099). Immunosuppression was indicated by reduced (c) spleen weights and relative concentration of splenic IL-1β, but not TNFα (p = 0.37). A general increase in CCR2 expression is demonstrated in (d) representative scatter plots of whole blood CD11b⁺CD45⁺Ly6G^-SSC^low mononuclear cells (Mo/M-MDSCs). When quantified, CCR2 expression increased within the mononuclear cell population as a whole, evidenced by (e) percent CCR2⁺ cells and CCR2 median fluorescence intensity (MFI). However, populations of mononuclear (f) Ly6C^-CCR2^- alternative M2 monocytes (p = 0.12), Ly6C^intCCR2^- transitional monocytes (p = 0.14), and Ly6C^hi CCR2⁺ classical M1 monocytes (p = 0.59) did not differ, nor were there overt differences in populations of circulating (g) CD11b⁺CD45⁺Ly6G⁺ polymorphonuclear cells (neutrophils or PMN-MDSCs; p = 0.21). Circulating plasma chemokine CCL2 concentrations (h) tended to decrease with stress (p = 0.056). Data are mean ± SEM, with each dot representing a single dam. NS = non-stressed, S = stressed; Mo/M-MDSCs = monocyte/monocytic myeloid-derived suppressor cells. CORT assay: n = 16/group; Spleen mass: n = 21–26/group; Spleen protein: n = 6–11/group; Gene expression: n = 7–10/group; Flow cytometry: n = 6/group. *p < 0.05, ^#p < 0.10.

Figure 3

GD17 uterine CD11b⁺CD45⁺ leukocytes appear to be stress resistant. Flow cytometric analysis of CD11b⁺CD45⁺ cells revealed that stress did not impact uterine (a) polymorphonuclear cell percentages (p = 0.74; neutrophils or PMN-MDSCs), or (b) percent of SSC^low mononuclear cells within the three phenotypes (Ly6C^- alternative M2, p = 0.77; Ly6C^int transitional, p = 0.30; or Ly6C^hi classical M1, p = 0.62). (c) CCR2 MFI within uterine Mo/Mφ/M-MDSCs isolated from restrained dams did not differ from controls (p = 0.13). Data are mean ± SEM, with each dot representing a single dam. NS = non-stressed, n = 6; S = stressed, n = 9; PMN-MDSCs = polymorphonuclear myeloid-derived suppressor cells; Mo/Mφ/M-MDSCs = monocyte/macrophage/monocytic myeloid-derived suppressor cells; MFI = median fluorescence intensity.

Figure 4

Placental CD11b⁺CD45⁺SSC^low leukocyte density decreases with prenatal stress at GD17, but cell proportions remain stable. Flow cytometric analysis of CD11b⁺CD45⁺ cells revealed (a) no change in polymorphonuclear Ly6G⁺ cell numbers (p = 0.45) but (b) a dramatic decrease in total SSC^low mononuclear cells (p < 0.0001) due to prenatal stress that was apparent across all three Ly6C subtypes (Ly6C^- alternative M2, p = 0.10; Ly6C^int transitional, p = 0.04; and Ly6C^hi classical M1, p = 0.04). However, percent of placental CD11b⁺CD45⁺ cells classified as (c) Ly6G⁺ polymorphonuclear cells (neutrophils or PMN-MDSCs; p = 0.28) and (d) SSC^low mononuclear cells did not significantly differ across treatment (Ly6C^- alternative M2, p = 0.77; Ly6C^int transitional, p = 0.50; and Ly6C^hi classical M1, p = 0.71). Likewise, (e) CCR2 median fluorescence intensity did not shift among the Mo/Mφ/M-MDSC population (p = 0.73). Data are mean ± SEM, with each dot representing a pooled sample (containing a minimum of 2 placentas) from 1 litter. NS = non-stressed, n = 6, S = stressed, n = 9; Mo/Mφ/M-MDSCs = monocytes/macrophages/monocytic myeloid-derived suppressor cells; MFI = median fluorescence intensity. ***p < 0.001, *p < 0.05, ^#p ≤ 0.10.

Figure 5

GD17 colonic microbiota metagenome analyses reveal restricted bacterial abundances and metabolic pathways due to psychological stress. (a) Differentially abundant metagenome-assembled genomes (MAGs) in non-stressed and stressed samples (p < 0.05, LDA > 2.0), as identified by LEfSe, using genome copies per million reads. Taxonomy is GTDB taxonomy with the lowest possible classification listed. Some MAGs have higher level taxonomy listed for easier identification. (b) Parasutterella excrementihominis genome copies per million reads (mean ± SEM; Kruskal Wallis and Wilcoxon Signed Rank tests, p = 0.006). (c) KEGG pathway completion for select KEGG pathways in whole sample assemblies that displayed presence/absence differences between sample groups using KEGGDecoder. Samples are grouped by non-stress and stress, with the KEGG pathway completions for the consensus Parasutterella excrementihominis MAG on the far right. NS = non-stressed, S = stressed; n = 6/group; **p < 0.01.