Prenatal immune stress blunts microglia reactivity, impairing neurocircuitry

Abstract

Recent studies suggested that microglia, the primary brain immune cells, can affect circuit connectivity and neuronal function 1,2. Microglia infiltrate the neuroepithelium early in embryonic development and are maintained in the brain throughout adulthood 3,4. Several maternal environmental factors-such as an aberrant microbiome, immune activation and poor nutrition-can influence prenatal brain development 5,6. Nevertheless, it is unknown how changes in the prenatal environment instruct the developmental trajectory of infiltrating microglia, which in turn affect brain development and function. Here we show that, after maternal immune activation (MIA) in mice, microglia from the offspring have a long-lived decrease in immune reactivity (blunting) across the developmental trajectory. The blunted immune response was accompanied by changes in chromatin accessibility and reduced transcription factor occupancy of the open chromatin. Single-cell RNA-sequencing analysis revealed that MIA does not induce a distinct subpopulation but, rather, decreases the contribution to inflammatory microglia states. Prenatal replacement of microglia from MIA offspring with physiological infiltration of naive microglia ameliorated the immune blunting and restored a decrease in presynaptic vesicle release probability onto dopamine receptor type-two medium spiny neurons, indicating that aberrantly formed microglia due to an adverse prenatal environment affect the long-term microglia reactivity and proper striatal circuit development. Microglia, the primary brain immune cells, have many functions across the lifespan. Their primary function is rapid immune activation to protect the brain, but dysfunction of microglia immune activation can result in poor outcomes for the surrounding neurons and glia 7,8. As microglia are long-lived cells with early brain entry, aberrant micro-glia arriving in early development may have a key role in modulating these cell-cell interactions across the trajectory of health and disease. Several studies showed that prenatal stress has an effect on brain development and function in later life 9,10. Nevertheless, the role and mechanism of microglia in the long-term changes elicited by prenatal stress are unclear. Blunted microglia reactivity in adult MIA offspring To address this knowledge gap, we tested whether a prenatal immune stressor (MIA) influenced microglia functions, including immune activation. To generate MIA offspring, we delivered an immune activator (polyinosinic:polycytidylic acid (PIC)) to pregnant dams at embryonic day 9.5 (E9.5) to target the first wave of microglia infiltrating the neuroepithelium 3. To study how MIA influences the offspring immune response, we injected either a proinflammatory stimulus (lipopolysaccharide (LPS)) or saline (SAL) to adult offspring in vivo, isolated microglia and analysed the gene expression profiles (Extended Data Fig. 1). First, we compared gene expression between microglia from MIA offspring (MIA microglia) and those from control offspring (CON microglia). We observed only seven differentially expressed genes (DEGs) between MIA and CON microglia after SAL treatment (Fig. 1a and Extended Data Fig. 2). By contrast, we identified 401 DEGs between MIA and CON microglia after LPS treatment (Fig. 1a and Extended Data Fig. 2). Interestingly, the majority of the DEGs (76%) were downregulated in MIA compared with CON microglia (Fig. 1a). Furthermore, the DEGs showed significant negative enrichment for many immune response pathways (Fig. 1b). Together, adult immune reactivity is markedly reduced in MIA microglia, although the baseline (SAL) gene expression is similar between MIA and CON microglia. To address the decreased MIA microglia immune reactivity directly, we compared the microglia gene expression between LPS and SAL treatment (LPS response genes). In CON microglia, 5,624 genes were differentially expressed between LPS and SAL treatment, whereas, in MIA microglia, fewer genes (4,284) were differentially expressed between LPS and SAL treatment (Fig. 1c and Extended Data Fig. 2).