Maternal SARS-CoV-2 impacts fetal placental macrophage programs and placenta-derived microglial models of neurodevelopment

38 The SARS-CoV-2 virus activates maternal and placental immune responses, 39 which in the setting of other infections occurring during pregnancy are known to impact 40 fetal brain development. The effects of maternal immune activation on 41 neurodevelopment are mediated at least in part by fetal brain microglia. However, 42 microglia are inaccessible for direct analysis, and there are no validated non-invasive 43 surrogate models to evaluate in utero microglial priming and function. We have 44 previously demonstrated shared transcriptional programs between microglia and 45 Hofbauer cells (HBCs, or fetal placental macrophages) in mouse models. Here, we 46 assessed the impact of maternal SARS-CoV-2 on HBCs isolated from term placentas 47 using single-cell RNA-sequencing. We demonstrated that HBC subpopulations exhibit 48 distinct cellular programs, with specific subpopulations differentially impacted by SARS-49 CoV-2. Assessment of differentially expressed genes implied impaired phagocytosis, a 50 key function of both HBCs and microglia, in some subclusters. Leveraging previously 51 validated models of microglial synaptic pruning, we showed that HBCs isolated from 52 placentas of SARS-CoV-2 positive pregnancies can be transdifferentiated into 53 microglia-like cells, with altered morphology and impaired synaptic pruning behavior 54 compared to HBC models from negative controls. These findings suggest that HBCs 55 isolated at birth can be used to create personalized cellular models of offspring 56 microglial programming. 57 58

Hofbauer cells; microglia; single-cell RNA sequencing; fetal brain; placenta; 59 neurodevelopment; neuroimmune; SARS-CoV-2, COVID-19 60 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 3

61 Multiple population-based studies have suggested that maternal infection during 62 pregnancy may have a transgenerational impact on offspring neurodevelopment. Initial 63 work found that the incidence of schizophrenia was increased after influenza pandemics 64 in Finland (1), Denmark (2), and the UK (3). Subsequent registry studies directly 65 examining the association of maternal infection requiring hospitalization during 66 pregnancy with diagnoses of autism and other neurodevelopmental disorders in 67 offspring also found risk to be increased (4, 5). Using electronic health records, we 68 identified an increased risk of delayed acquisition of speech and motor milestones, 69 beyond that attributable to prematurity, in a US cohort of offspring whose mothers had 70 SARS-CoV-2 during pregnancy. (6, 7). Similarly, authors of a prospective cohort study 71 of 127 children in Brazil found an increased risk of neurodevelopmental delay with in 72 utero SARS-CoV-2 exposure during pregnancy (8), and a recent meta-analysis of 73 smaller studies identified additional evidence of neurodevelopmental sequelae – 74 including reductions in fine motor and problem-solving skills – in infants with in utero 75 SARS-CoV-2 exposure compared to unexposed and pre-pandemic cohorts (9). If these 76 early signals foreshadow an increased risk of neurodevelopmental disorders in 77 childhood and adulthood, the public health implications could be profound, given the 78 significant number of pregnancies exposed to SARS-CoV-2 infection. 79 Despite the convergence of studies suggesting that maternal viral infection may 80 increase offspring risk for neurodevelopmental disorders, the precise biological 81 mechanisms leading to offspring neurodevelopmental vulnerability are not known. 82 Direct placental and fetal infection with SARS-CoV-2 virus is uncommon based on 83 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 4 current evidence (10–13), and thus vertical transmission is unlikely to be a major cause 84 of neurodevelopmental sequelae. Animal models of maternal immune activation (MIA), 85 in which offspring of pregnant dams treated with an immune stimulus recapitulate the 86 behavioral hallmarks of human neurodevelopmental disorders, have been used for 87 decades to investigate candidate in utero mechanisms of neurodevelopmental 88 programming (14–17). Embryonic microglia have emerged as central mediators of 89 offspring neuropathology in the setting of MIA (14). However, microglia from surviving 90 offspring are inaccessible for direct analysis in humans, necessitating alternative 91 models for evaluating the impact of SARS-CoV-2 on the fetal brain. 92 Prior work from our group has identified remarkable similarities in the 93 transcriptional programs and reactivity of fetal placental macrophages, or Hofbauer cells 94 (HBCs), and fetal brain microglia isolated from mouse embryos (18, 19). These two cell 95 types share an embryonic origin in the fetal yolk sac (20, 21), and both carry the imprint 96 of the in utero environment, with fetal yolk sac-derived macrophages serving as the 97 progenitors for the lifelong pool of microglia (22, 23). Here, we investigate the impact of 98 SARS-CoV-2 exposure on the transcriptional profiles of HBC subpopulations to gain 99 insight into fetal resident tissue macrophage programming. Our results demonstrate 100 that HBCs are a heterogeneous cell type, with eight subpopulations exhibiting distinct 101 cellular programs, and that maternal SARS-CoV-2 infection is associated with varying 102 impact on function in these subpopulations. Assessment of differentially expressed 103 genes implies impaired phagocytosis in specific subclusters, a key function of both 104 HBCs and microglia; we confirm these effects using a previously validated assay of 105 microglial synaptic pruning via synaptosome phagocytosis. In aggregate, we 106 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 5 demonstrate the application of HBC-based cellular models to gain non-invasive insight 107 into the impact of in utero exposures on fetal brain development. 108 109

110 Hofbauer cells are a heterogeneous population with subclusters demonstrating 111 both M1- and M2-like transcriptional signatures 112 Placental chorionic villous tissues were collected from N=12 birthing individuals: 113 N=4 from individuals who had a positive SARS-CoV-2 nasopharyngeal PCR test during 114 pregnancy, and N=8 from birthing individuals with a negative PCR at delivery and no 115 history of a positive SARS-CoV-2 test during pregnancy. In SARS-CoV-2 positive 116 maternal cases, infections occurred remote from delivery (median 19.5 weeks). No 117 participants had received a COVID-19 vaccine prior to or during pregnancy and no 118 placental samples were infected with SARS-CoV-2 at delivery (defined as having 119 detectable SARS-CoV-2 viral load in a validated assay sensitive to 40 copies/mL) (24, 120 25). Thus, SARS-CoV-2 positive cases were defined by maternal infection in 121 pregnancy, not placental or fetal infection. Additional participant characteristics are 122 provided in Table 1. 123 To assess the HBC transcriptome, we first used a previously-described protocol 124 to obtain primarily HBCs from placental villi; in this protocol, a Percoll-based gradient 125 and negative bead-based selection steps are used to isolate putative HBCs from other 126 cell types present in the chorionic villi (including trophoblasts, fibroblasts) (26). Single-127 cell RNA sequencing was then performed on all cell suspensions (10x Genomics). After 128 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 6 quality control filtering to remove putative doublets and cells with less than 300 129 identified genes, we obtained a dataset comprised of a total of 70,817 cells. We then 130 performed sample integration and graph-based clustering to identify broad cell types 131 (Figure S1A). Based on analyses of marker gene expression (Figure S1B), we found 132 that the majority of cells in our dataset had marker gene expression consistent with 133 monocytes/macrophages, and that other cell types were represented to a lesser extent, 134 including some fibroblasts, vascular endothelial cells (VECs), extravillous trophoblasts 135 (EVTs), leukocytes (NK cells, CD8+ T cells, B cells), neutrophils and red blood cells 136 (Figure S1C). From this dataset, we excluded all cell types that were not identifiable as 137 macrophages/monocytes. After additional quality control filtering for nUMIs, gene 138 counts, and percent mitochondrial reads (see Methods), this resulted in a dataset 139 containing 31,719 high-quality placental macrophages/monocytes. All subsequent 140 analyses were performed with this final dataset. 141 After re-processing selected cells for quality control as described, we identified 142 10 total subclusters of macrophages/monocytes (Figure 1A), with representation of 143 each subcluster across donors from both SARS-CoV-2 positive cases and controls 144 (Figure S2A). To distinguish HBCs, which are placental macrophages of fetal origin, 145 from macrophages or monocytes of maternal origin, we used cells isolated from male 146 placentas (N=10). Male fetal origin was confirmed by high expression of DDX3Y and 147 low expression of XIST in 8 subclusters; these were labeled HBC 0-7 (Figure 1B). The 148 macrophage cluster with high expression of XIST (consistent with maternal origin) was 149 annotated as placenta-associated maternal macrophages and monocytes (PAMMs, 150 Figure 1B) (27). A small cluster of monocytes – identified as such by high expression of 151 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 7 monocyte marker genes S100A8, S100A9, and TIMP1 – demonstrated equal 152 expression levels of both DDX3Y and XIST, suggesting that both fetal and maternal 153 cells were present in this monocyte cluster. To further support HBC cluster annotation, 154 we next compared the overall gene expression profiles of each cluster to a previously 155 published single-cell dataset derived from human first-trimester placenta and decidua 156 (28). In this analysis, all putative HBC subclusters showed highest correlation with HBC 157 expression profiles from this dataset, whereas the monocyte and PAMM clusters had 158 higher correlation with decidual macrophages than HBCs (Figure 1C). 159 To delineate differences in the identity and functions of HBC subclusters, we next 160 assessed top marker genes defining each subcluster, shown as an average heatmap 161 (Figure 1D). Marker genes for HBC clusters 3, 4, 5, and 7 suggested involvement in 162 classic M1 macrophage/pro-inflammatory activities. HBC cluster 3 demonstrated high 163 expression of chemokine (C-X-C motif) ligand genes (CXCLs) as well as pro-164 inflammatory marker genes IL1B, IL1A, TNF, and NFKB1. HBC cluster 4 demonstrated 165 high expression of multiple CC chemokine ligand genes (CCLs) in a profile similar to 166 that observed in HBCs responding to lipopolysaccharide stimulation in vitro (29). HBC 167 cluster 5 was characterized by high expression of genes encoding major 168 histocompatibility complex (MHC) class II molecules (human leukocyte antigen (HLA)-169 DRA/B1 and -DP) and Fc-gamma binding protein (FCGBP), suggesting a role in antigen 170 presentation to CD4+ T cells. MHC class II molecules are critical to antigen-specific 171 responses, and upregulation of HLA complexes and antigen presentation pathways has 172 been observed in proteomic analyses of HBCs stimulated with the viral dsRNA mimic 173 poly(I:C) (29). HBC cluster 7 demonstrated marker genes from the interferon-induced 174 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 8 protein with tetratricopeptide repeats (IFIT) family (IFIT2, IFIT3), CXCL10, ISG15, and 175 MX1, associated with a pro-inflammatory type 1 interferon antiviral response. To gain 176 further insight into the biological processes reflected in each HBC subcluster, we 177 performed Gene Ontology (GO) enrichment analyses of cluster marker genes (Figure 178 1E). As would be expected given marker gene expression noted previously, pathways 179 involved in M1-like immune/inflammatory responses were enriched in HBC 3, 4, 5, and 180 7, including “response to interleukin-1,” “response to interferon-gamma,” “response to 181 tumor necrosis factor,” and “positive regulation of cytokine production.” 182 Gene signatures of HBC clusters 0, 1, and 2 reflected engagement in specific 183 stress response processes, particularly response to inflammation and/or tissue damage, 184 to ultimately support placental function. HBC 0 and HBC 1 were characterized by genes 185 encoding heat shock proteins and other proteins involved in endoplasmic reticulum 186 stress and the unfolded protein response, such as HSPA6, HSPA1B, DNAJB1, 187 HSP90B1, HSPA5 and BAG3. The unfolded protein response represents a homeostatic 188 response to restore balance when endoplasmic reticulum stress is sensed and to 189 modulate and/or resolve inflammation (30). Additionally, HBC 0’s high expression of 190 PDK4 and KLF2 may suggest involvement in attenuating oxidative stress responses 191 and reducing pro-inflammatory cytokine production (31, 32), and HBC 1’s high 192 expression of FABP5 and HMOX1 suggests engagement in anti-inflammatory 193 responses against heme-induced toxicity and induction towards M2 polarization (33, 194 34). GO enrichment analysis of these clusters similarly demonstrated enrichment of 195 pathways such as “response to unfolded protein”, “response to heat”, “response to 196 endoplasmic reticulum stress”, and pathways related to cellular stress response and 197 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 9 apoptosis (e.g. “extrinsic apoptotic signaling pathway”, “response to oxidative stress” 198 and “ERK1 and ERK2 cascade”). 199 GO enrichment analysis also suggested both HBC 0 and HBC 1 were engaged in 200 homeostatic functions including “receptor-mediated endocytosis”, “regulation of 201 angiogenesis” and “vascular development”, and nutrient-sensing functions such as 202 “response to nutrient levels” and “response to starvation.” HBC 2 was characterized by 203 high expression of the genes encoding secreted phosphoprotein 1 (SPP1) and 204 Nicotinamide N -methyltransferase (NNMT), both associated with M2 (anti-205 inflammatory) macrophage polarization in the context of tumor-associated macrophages 206 (35, 36); SPP1, also known as osteopontin, is secreted by HBCs and plays an important 207 role in endothelial biology and angiogenesis (37). GO analysis of HBC cluster 2 also 208 demonstrated enrichment in “receptor-mediated endocytosis” (involved in intracellular 209 transport of macromolecules), as well as “cellular lipid catabolic process,” and 210 processes associated with stromal tissue development. 211 HBC 6 was characterized by expression of genes involved in regulation of actin 212 polymerization and cytoskeleton organization (TMSB4X and AIF1, which encodes the 213 canonical microglial marker Iba1 (38)) and several ribosomal proteins including RPS18, 214 RPS23, and RPS4Y1. GO enrichment analysis of this cluster demonstrates highly 215 specific enrichment of protein processing pathways (e.g. “protein targeting” pathways, 216 “cytoplasmic translation,” “translational initiation”) and pathways related to RNA 217 catabolism, oxidative phosphorylation, and ATP metabolism. Marker genes of the 218 maternal PAMM subcluster included APOE, APOC1, VIM, LGALS1, and GPNMB 219 among others, an expression pattern consistent with previously reported maternal 220 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 10 placental macrophage transcriptional profiles (39, 40). High expression of LGALS1/3 221 and GPNMB by the PAMM cluster suggests a role in inflammation regulation (41–43), 222 which was echoed by GO analyses identifying enrichment in immune response and 223 immunomodulatory pathways (e.g. “antigen processing and presentation, “positive 224 regulation of cytokine production”, and “regulation of innate immune response”) . In 225 addition, GO enrichment analysis demonstrated PAMM were engaged in lipid metabolic 226 processes and receptor-mediated endocytosis, consistent with their known involvement 227 in lipid engulfment, and tissue repair/scar formation (37, 39). 228 229 Maternal SARS-CoV-2 infection drives cluster-specific differences in immune 230 signaling and metabolic pathways 231 Once the baseline functions of HBC and PAMM subclusters had been 232 established, we then sought to characterize the impact of maternal SARS-CoV-2 233 infection on the transcriptomic profile of HBC subclusters. To do so, we identified 234 differentially expressed genes (DEG) by maternal SARS-CoV-2 status within each 235 cluster. DEG were defined using log fold-change threshold of 0.2 and adjusted p-value 236 of 0.05 (see Methods for full details). We first verified that each subcluster included 237 representation from both SARS-CoV-2+ cases and controls (Figure 2A, top panel). The 238 proportion of cells from cases versus controls was consistent across all subclusters, 239 except for HBC 0, which demonstrated a relatively high contribution of control donor 240 cells (Figure S2A). Of the 8 HBC clusters, a majority (5) were significantly impacted by 241 maternal SARS-CoV-2 infection: HBC 0, 1, 2, 3, and 5 (Figure 2A, bottom panel). In 242 contrast, HBC clusters 4, 6, and 7 had very few DEG in the setting of maternal SARS-243 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 11 CoV-2 exposure, with three, zero, and two DEG respectively. Of the 5 highly impacted 244 clusters, HBC 1 and HBC 5 had the highest number of DEG by maternal SARS-CoV-2, 245 with 723 and 566 DEG, respectively. PAMMs were impacted by SARS-CoV-2 to a 246 lesser extent, with 67 DEG identified. Both up- and down-regulated DEG were 247 identified across all impacted clusters. 248 GO pathway enrichment analysis of DEG indicated that in the context of maternal 249 SARS-CoV-2 infection, specific pathways involved in immune responses were enriched 250 in all impacted HBC clusters (HBC 0, 1, 2, 3 and 5), including response to 251 lipopolysaccharide, response to interferon-gamma, cellular response to tumor necrosis 252 factor, and regulation of T cell activation (Figure 2B). Additionally, all subclusters were 253 enriched for pathways related to cellular movement, such as cell chemotaxis, leukocyte 254 cell-cell adhesion, and myeloid leukocyte migration; heat shock-related pathways 255 (unfolded protein response); and phagocytosis pathways. 256 GO enrichment analysis also indicated some biological processes that were only 257 dysregulated in specific clusters in the setting of maternal SARS-CoV-2 infection (Figure 258 2B). For example, regulation of vascular development was only dysregulated in HBC 0 259 and the PAMM cluster, regulation of lipid metabolism and transport were dysregulated 260 in all clusters except HBC 5, more cellular energy utilization pathways (e.g. ATP 261 metabolism, electron transport chain/oxidative phosphorylation and cellular respiration) 262 and cellular stress/apoptosis pathways were impacted in HBC clusters relative to the 263 PAMM cluster, and protein processing and actin cytoskeleton organization pathways 264 were only dysregulated in HBC clusters but not PAMMs. Taken together, these 265 functional analyses suggest that in the context of maternal SARS-CoV-2 infection, HBC 266 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 12 subclusters and PAMMs are differentially impacted, with key dysregulated biological 267 processes including innate immune and pro-inflammatory signaling, cell chemotaxis and 268 migration, cellular ATP and lipid metabolism, cellular phagocytosis, and the unfolded 269 protein response. 270 To better understand the impact of SARS-CoV-2 on HBC functions, we next used 271 Ingenuity Pathway Analysis (IPA), which predicts strength and directionality (i.e. 272 activation or suppression) of enriched canonical pathways by subcluster. In this 273 analysis, pathways with absolute Z-score value greater than 1 (consistent with IPA 274 being able to “call” a direction of dysregulation of the pathway) and Benjamini 275 Hochburg-adjusted p<0.05 were included and displayed by subcluster (Figure 2C and 276 Figure S2B-F). Z-scores ³ 1 indicate upregulated signaling in the pathway and ≤ 1 277 indicate downregulated pathway signaling (44). Both HBC 1 and HBC 2 subclusters 278 exhibited a primarily anti-inflammatory response to SARS-CoV-2, with activation of 279 PPAR signaling and oxidative phosphorylation, a metabolic profile associated with an 280 anti-inflammatory/pro-resolution phase macrophage signature (45, 46). In HBC 1 and 281 2, suppression of IL-6, IL-1, and IL-17 pathways, and activation of LXR/RXR signaling 282 pathways in SARS-CoV-2+ cases suggests involvement in resolution of inflammation, 283 as LXR/RXR pathway activation in macrophages is associated with inhibition of 284 inflammatory gene expression and promotion of lipid metabolism (47). Also consistent 285 with an anti-inflammatory role, HBC 2 showed strong suppression of the Coronavirus 286 Pathogenesis Pathway and activation of Oxytocin Signaling Pathway, the latter of which 287 is involved in attenuating oxidative and cellular inflammatory responses in macrophages 288 (48). 289 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 13 Conversely, HBC 0 and 3 demonstrated primarily activated pro-inflammatory 290 immune responses in SARS-CoV-2+ cases, with increases in LPS/IL-1 mediated 291 inhibition of RXR (HBC 3), Interferon induction (HBC 3), Neuroinflammation signaling 292 (HBC 0 and 3), T-cell signaling (HBC 0 and 3), and Production of nitric oxide and 293 reactive oxygen species (HBC 0). Metabolic processes were suppressed in both 294 clusters, including Oxytocin signaling pathway (HBC 0), Sirtuin signaling (HBC 3), and 295 MSP-RON signaling (HBC 0 and 3) (48–50). In the context of maternal SARS-CoV-2 296 infection, subcluster HBC 5 presented a mixed picture of both pro- and anti-297 inflammatory signaling, with upregulation of interferon, EIF 2, neuroinflammation and T 298 cell related signaling pathways, balanced by upregulation of anti-inflammatory pathways 299 such as PPAR signaling and downregulation of pro-inflammatory signaling pathways 300 such as Coronavirus Pathogenesis pathway, FAK and TNF-mediated signaling 301 pathways. 302 Compared to HBC subclusters, PAMMs were less impacted overall by maternal 303 SARS-CoV-2 at a transcriptomic level, with 67 DEG identified. In the setting of maternal 304 SARS-CoV-2 infection, PAMMs showed activation of pathways involved in immune 305 responses including Production of Nitric Oxide and Reactive Oxygen Species, B-cell 306 signaling pathways, Interferon induction, and activation of the pattern recognition 307 receptor TREM-1 signaling. Similar pro-inflammatory patterns were observed for 308 monocytes, including activation of antiviral response pathways and Th1 signaling 309 pathways, and suppression of MSP-RON signaling (Figure S2C). Taken together, 310 these analyses point to transcriptional shifts in some but not all subclusters in response 311 to SARS-CoV-2, with a greater response by HBCs compared to PAMMs, driven by a 312 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 14 combination of immune activation/pro-inflammatory signature in subclusters HBC 0 and 313 HBC 3 and an anti-inflammatory tissue repair signature in clusters HBC 1 and HBC 2. 314 315 Maternal SARS-CoV-2 infection impacts HBC transcriptional programs associated 316 with phagocytosis, neuroinflammation, and neurological disorders 317 Tissue-resident macrophages promote resolution of inflammation through 318 phagocytosis of apoptotic cells, invading pathogens, or cellular debris (51, 52). 319 Phagocytosis is also a key function of microglia in early brain development (53–55). IPA 320 functional analysis of SARS-CoV-2-specific HBC signatures demonstrated that 321 macrophage phagocytosis (Figure 3A) and neurological disease-related pathways 322 (Figure 3B) were key functions and pathways implicated by the cluster-specific gene 323 expression signatures. Figure 3 summarizes the impact of maternal SARS-CoV-2 324 infection on placental macrophage phagocytosis (Figure 3A, 3C), illustrating the 325 potential for altered HBC gene programs to provide insight into both fetal brain 326 microglial function and the impact of maternal SARS-CoV-2 infection on 327 neurodevelopment (Figure 3B, 3D). These analyses predicted SARS-CoV-2-associated 328 suppression of phagocyte chemotaxis and cell movement pathways (e.g. reduced 329 “activation of phagocytes”, “recruitment of phagocytes”, “cell movement of phagocytes”, 330 “adhesion of phagocytes”) in HBC 1, 2 and 5, consistent with the suppression of 331 synaptosome phagocytosis (a proxy for synaptic pruning) observed in subsequent 332 experiments using in vitro Hofbauer cell induced microglial assays, detailed below. In 333 contrast to the consistent suppression of phagocytosis in HBC clusters 1, 2 and 5, HBC 334 clusters 0 and 3 and the PAMM cluster demonstrated activation of phagocytosis-related 335 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 15 pathways including “Phagocytosis” (HBC 0), “Immune response of phagocytes” (HBC 336 0), “Phagocytosis by macrophages” (HBC 3) and “Cellular infiltration by phagocytes” 337 (HBC 3). A representative phagocytosis pathway from IPA and expression of its 338 constituent genes by cluster is depicted as a heatmap in Figure 3C. Cluster-specific 339 alterations in phagocyte movement in the setting of maternal SARS-CoV-2 infection 340 were primarily driven by expression differences in CXCL2, NFKB1A, NFKB1Z, IL1B, 341 CXCL8, CD36, and ICAM1 by cluster (Figure 3C). Concordant with patterns observed in 342 canonical pathways analyses, HBC 1 and 2 (and to a lesser extent HBC 5), which 343 showed primarily immunomodulatory signatures, also show suppressed phagocytosis 344 and phagocytic movement pathways, versus proinflammatory clusters HBC 0 and 3, 345 which demonstrate activation of phagocytosis. 346 In addition to phagocytosis, pathways relevant to neurologic disease and 347 microglial functions emerged as key dysregulated pathways in the setting of maternal 348 SARS-CoV-2 infection. We therefore assessed whether DEG of HBC subclusters map 349 to neuroinflammatory/neurodevelopmental pathways and functions in IPA, and plotted 350 pathway activation Z-scores by subcluster (Figure 3B). The transcriptional signature of 351 HBC 1 – in which Fc-gamma receptor-mediated phagocytosis (Figure 2C) and other 352 previously-described phagocytosis pathways (Figure 3A) are suppressed in SARS-CoV-353 2+ cases – is also associated with increased neuroinflammation, including positive 354 activation Z-scores for “Inflammation of central nervous system”, “Myelitis”, and 355 “Encephalitis.” Cluster-specific expression of the genes in the “Inflammation of central 356 nervous system pathway” is depicted in Figure 3D, with upregulation of signaling in this 357 pathway driven by increased expression of ANXA1, FN1, CCL3, SLC1A3, NLRP3, and 358 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 16 MERTK, among others. Interestingly, HBC 3 – in which “Inflammation of Central 359 Nervous System” is also predicted to be activated after maternal SARS-CoV-2 infection 360 and whose transcriptional signature is consistent with activation in phagocytosis 361 pathways, also implicates increased “Apoptosis of neurons” and “Neuronal cell death” 362 (Figure 3B). In a developmental context, this pattern may represent a functional rather 363 than pathologic gene signature in response to SARS-CoV-2, as microglia (resident brain 364 macrophages) play a key role in neuronal cell turnover, regulation of neural progenitors, 365 and synaptic rewiring in early neurodevelopment, all via phagocytosis (56). Thus, 366 increased phagocytosis by tissue-resident macrophages might be an adaptive response 367 to SARS-CoV-2-associated inflammation, while reduced phagocytosis could be a 368 pathologic or maladaptive response to maternal immune activation (e.g., reduced 369 microglial phagocytosis and reduced synaptic pruning associated with maternal immune 370 activation is thought to be a key aspect of the pathogenesis of autism (57–59)). Taken 371 together, these data support the notion that HBC transcriptional signatures provide 372 insight into protective versus pathologic microglial programming in the setting of an 373 immune challenge such as SARS-CoV-2. 374 Prior work from our group in a mouse model has shown that HBCs and fetal 375 brain microglia share transcriptional profiles and responses to maternal obesity, an 376 immune-activating exposure (19). To further probe the potential connection between 377 transcriptional signatures of HBC subclusters and brain microglia in humans, we next 378 compared marker genes from HBC subclusters with gene modules from published 379 human single cell atlases of macrophages derived from adult and embryonic brain 380 (Figure 3E) (60, 61). Nearly all HBC subclusters scored highly for gene signatures 381 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 17 found in microglia, yolk sac macrophages, or CNS border associated macrophages, 382 compared with monocytes and PAMMs. HBC 2 and 1 exhibited greatest similarity to 383 yolk sac macrophages whereas HBC 3 and 4 were most like microglia isolated from 384 adult brain samples. In contrast, monocytes and PAMMs were most similar to 385 circulating monocytes, which is concordant with their shared myeloid lineage (37). This 386 analysis supports the concept that HBCs isolated from full term human placenta share 387 transcriptional signatures with yolk sac macrophages and fetal brain microglia, and thus 388 may offer insights into global reprogramming of fetal macrophage populations, including 389 those of the fetal brain, in the setting of maternal immune-activating exposures. 390 391 HBCs isolated from placentas of SARS-CoV-2 positive pregnancies can be 392 transdifferentiated to microglia-like cells (HBC-iMGs) 393 To gain insight into the functional consequences of SARS-CoV-2 exposure on 394 HBC populations, we used a previously-validated model in which HBCs isolated from 395 SARS-CoV-2 positive cases (N=4) and a subset of SARS-CoV-2 negative controls 396 (N=4) were cultured in media containing IL-34 and GM-CSF to obtain transdifferentiated 397 microglia-like cells (HBC-iMGs), as we have previously described (see Methods) (62–398 64). Following culture, we assessed the expression of multiple markers associated with 399 microglial identity, including IBA1, TMEM119, PU.1, and P2RY12 (65, 66), and 400 identified expression of all markers in the majority of HBC-iMGs from both SARS-CoV-2 401 positive cases and negative controls (Figure 4A, Figure S3A-B). 402 403 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 18 SARS-CoV-2-exposed HBC-iMGs demonstrate increased amoeboid morphology 404 and impaired phagocytic behavior compared to HBC-iMGs from uninfected 405 control placentas 406 To assess differences in cell phenotype by SARS-CoV-2 exposure, we first 407 evaluated cellular morphology of IBA1-positive HBC-iMGs by quantitative assessment 408 of two morphological characteristics: eccentricity (amoeboid vs bipolar shape) and 409 solidity (amoeboid/bipolar vs ramified shape) (Figure 4B-C). In this analysis, HBC-iMGs 410 from SARS-CoV-2 negative controls demonstrated a more ramified morphology than 411 those from positive cases, indicated by their lower solidity and high eccentricity (Figure 412 4B-C). More ramified microglial morphology is generally typical of resting-state, tissue-413 surveilling microglia in vivo (67, 68). In contrast, a greater proportion of HBC-iMGs 414 generated from SARS-CoV-2 positive cases demonstrated higher solidity and smaller 415 cell size (Figure 4B-C, Figure S3C), consistent with a more amoeboid appearance. 416 While this morphology is classically attributed to an immune-activated state (69), it is 417 also typical of microglial patterns observed in fetal states (67, 70). 418 Transcriptional analyses of HBC clusters pointed to a cluster-specific impact of 419 maternal SARS-CoV-2 on phagocytosis pathways. HBC clusters with the greatest 420 similarity to embryonic/yolk sac microglia (e.g. HBC 1, 2) also exhibited cellular 421 programs suggestive of impaired phagocytosis. Using a previously-validated model of 422 synaptic pruning (62–64), a key physiologic function of microglia in early brain 423 development, we next tested the functional capability of HBC-iMGs to engage in 424 phagocytosis. In this assay, HBC-iMGs were co-cultured for 3 hours with pHrodo Red-425 labeled neuronal synaptosomes derived from human induced pluripotent stem cells 426 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 19 prior to fixation and imaging. This pH-sensitive label fluoresces following intracellular 427 engulfment (see Methods). Synaptosome engulfment by IBA1-positive cells was then 428 measured by quantifying fluorescence using confocal microscopy images with 429 CellProfiler software applied for segmentation and thresholding (Figure 5A). Compared 430 to SARS-CoV-2 negative cases, HBC-iMGs from positive cases demonstrated 431 significant impairment in synaptosome phagocytosis, reflected by a reduced phagocytic 432 index (Figure 5B). Phagocytic index was reduced across all SARS-CoV-2 positive 433 samples, and was driven by reduced phagocytic uptake per cell, regardless of the 434 proportion of cells engaged in phagocytosis in any given sample. (Figure S3D-E). In 435 conjunction with the morphologic changes, and consistent with the transcriptomic 436 signatures observed in a subset of HBC, these functional phenotypes support a 437 dysregulated activation state following SARS-CoV-2 infection. 438 439

440 Data from observational cohorts suggests an increased neurodevelopmental risk 441 of offspring exposed in utero to maternal SARS-CoV-2 infection (6–8) but the underlying 442 mechanism for offspring brain vulnerability remains unknown. Studies have consistently 443 demonstrated that maternal SARS-CoV-2 infection drives alterations in immune cell 444 populations and pro-inflammatory responses at the maternal-fetal interface (71–78) that 445 have the capacity to impact the fetal brain (79). Even in the absence of direct viral 446 transmission to the fetus, profiling of umbilical cord blood immune cell populations and 447 the serum proteome demonstrates that maternal SARS-CoV-2 infection can shape fetal 448 and neonatal immunity (40, 80–82). Prior bulk and single-cell transcriptomic analyses 449 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 20 have also revealed significant reprogramming at the maternal-fetal interface following 450 SARS-CoV-2 infection during pregnancy (40, 71, 72, 77), yet granular information on 451 fetal placental cell populations has been relatively limited by their lower representation 452 in these studies. 453 Here we report single-cell RNA-seq data that provide new insights into the 454 heterogeneous functions that fetal placental macrophages, or Hofbauer cells, and 455 maternal resident placental macrophages and monocytes or PAMMs, perform at 456 baseline, and how these programs are altered in the setting of maternal SARS-CoV-2 457 infection. We found that maternal SARS-CoV-2 infection in pregnancy, even distant 458 from delivery and in the absence of placental infection, was associated with significant 459 alterations in the transcriptional programs of Hofbauer cells. These programs were more 460 significantly impacted than those of maternal placental macrophages, as indicated by 461 number of DEG. Effects of maternal SARS-CoV-2 infection were subcluster-specific, 462 with phagocytosis being a key dysregulated function, and affected Hofbauer cell 463 clusters exhibited signatures consistent with neuroinflammation and neurologic disease. 464 We directly tested this predicted dysregulation using validated in vitro models of HBC-465 based induced microglia (HBC-iMGs) (62–64), confirming that SARS-CoV-2 infection 466 altered HBC-iMG morphology and function. SARS-CoV-2 exposed HBC-iMGs were 467 more ameboid in shape and exhibited reduced synaptosome phagocytosis, an assay 468 that serves as a proxy for synaptic pruning. Notably, reduced synaptic pruning by 469 microglia has been identified as a key mechanism in the pathogenesis of autism (57, 470 83), a neurodevelopmental disease associated with maternal immune activation and 471 viral infection in pregnancy (59, 84, 85). Considering the shared fetal origin between 472 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 21 Hofbauer cells and brain microglia (20, 21), this work indicates Hofbauer cells’ potential 473 to serve as a more accessible cell type at birth that could provide information about fetal 474 brain immune programming, which in turn could alter neurodevelopmental trajectories 475 after in utero exposure to maternal SARS-CoV-2 infection. 476 Our study is unique in its in-depth, focused interrogation of fetal immune cell 477 populations of the placenta in the context of a remote maternal viral infection. Through 478 sex-chromosome-specific gene expression mapping, we were able to reliably assign 479 fetal cell identity to 8 subtypes of HBCs, engaged in a myriad of functions at baseline. 480 Similar to the work of Thomas et al. in first trimester placenta, we identified subclusters 481 with transcriptional programs associated with angiogenesis and tissue remodeling, as 482 well as clusters enriched for immune defense functions (39). Concordant with prior 483 single-cell RNA sequencing studies of the placenta in the context of SARS-CoV-2 484 infection (40, 86), we identified that even in the absence of direct placental infection or 485 active COVID-19 disease at the time of delivery, maternal exposure to SARS-CoV-2 486 remote from delivery had a profound impact on the transcriptional programs of the fetal 487 macrophage population, and to a lesser extent maternal PAMMs. 488 We defined a broad range of responses to SARS-CoV-2 across HBC 489 subclusters, including some clusters with relatively few DEG and others with significant 490 transcriptional shifts. Impacted HBC subclusters demonstrated transcriptional programs 491 evoking the changes observed in neuroinflammation, and the same subclusters 492 exhibited alterations in phagocytosis and in chemotaxis and cellular movement. To 493 investigate these results further we created induced microglial cellular models from the 494 same samples (HBC-iMG). Phenotypic and functional analyses of HBC-iMGs from 495 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 22 SARS-CoV-2 positive samples demonstrated a shift toward more amoeboid morphology 496 and significant impairments in synaptosome phagocytosis. Reduced phagocytic 497 efficiency appeared to result from reduced capacity for synaptosome uptake within the 498 cell, rather differences in the proportion of cells engaged in phagocytosis. 499 We demonstrated for the first time that HBCs can be used to create microglia-like 500 cell models, applying this approach to gain insight into fetal brain immune programming 501 in the context of maternal SARS-CoV-2 infection. As yolk-sac derived macrophages 502 that colonize the fetal brain early in development (20), microglia play a fundamental 503 role in neurogenesis by promoting neural precursor cell proliferation, axonal outgrowth, 504 and synaptic wiring throughout development (53–55). A key function of microglia in 505 normal neurodevelopment also includes selective phagocytosis of excess neuronal 506 precursors and synapses to edit and refine the architecture of neuronal communication 507 (55, 87). Evidence from animal models of maternal immune activation (MIA) suggests 508 that microglia are keenly responsive to maternal innate immune signaling, and MIA-509 induced disruption of normal microglial function can recapitulate social deficits and other 510 behaviors correlative of those observed in neurodevelopmental disorders such as 511 autism spectrum disorder and schizophrenia (57, 58, 88, 89). 512 A primary strength of our study is inclusion of rigorously phenotyped individuals 513 without a history of prior SARS-CoV-2 infection or vaccination and of 514 contemporaneously enrolled control subjects. We thus were able to examine the 515 impact of maternal SARS-CoV-2 on an immunologically naïve cohort in the absence of 516 prior immunity to SARS-CoV-2, with a consequence being that we could not assess the 517 impact of prior vaccination. Neither the impact of COVID-19 severity nor fetal sex could 518 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 23 be assessed in this study due to the study design (primarily focused on symptomatic 519 infection and male samples as a proof of principle study) and small sample size. Sex 520 differences will be particularly important to assess in future work, given the importance 521 of fetal sex on offspring neurodevelopmental vulnerability and fetoplacental 522 programming (90, 91). Taken together, our results suggest the ability of HBC-iMGs to 523 serve as personalized cellular models of microglial programming in the setting of 524 maternal exposures, including SARS-CoV-2 and potentially other environmental 525 exposures that might impact neurodevelopment. They demonstrate potential 526 mechanisms by which these exposures may contribute to adverse neurodevelopmental 527 outcomes. 528 529 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 24

530 Study design and participant enrollment 531 In this study, 12 pregnant individuals with full-term, singleton pregnancies delivering at 532 Massachusetts General Hospital (March 2021 - August 2021) were included. 533 Participants were classified as SARS-CoV-2 positive (N=4) if they had symptomatic 534 COVID-19 infection during pregnancy, confirmed by positive SARS-CoV-2 535 nasopharyngeal PCR test. Participants were classified as SARS-CoV-2 negative (n=8) 536 if they did not have a positive SARS-CoV-2 nasopharyngeal PCR or COVID-19 537 symptoms during pregnancy and had a negative SARS-CoV-2 nasopharyngeal PCR at 538 delivery upon universal COVID-19 screening on Labor and Delivery. Pregnant 539 individuals were eligible for inclusion if they were 18 years or older and were delivering 540 during the COVID-19 pandemic. For this study, individuals with prior COVID-19 541 vaccination were excluded. A study questionnaire and review of the electronic health 542 record was used to determine key demographic variables such as maternal age, 543 gestational age at delivery, gestational age at positive COVID-19 test, COVID-19 544 disease severity at diagnosis, any prior diagnoses of COVID-19 or history of COVID-19 545 vaccination, and infant sex and birthweight. 546 547 Placenta collection and processing 548 Placentas were obtained within 20 minutes after delivery and submerged in Cytowash 549 media (Dulbecco’s Modified Eagle Medium (DMEM) containing 2.5% FBS, 1% 550 Penicillin-Streptomycin, 0.1% Gentamicin) and stored at 4°C until cell isolation. 551 Isolation of Hofbauer cells was performed using previously described protocols (26); 552 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 25 reagents are listed in Supplemental Table S1 and isolation workflow and study 553 procedures are depicted in Supplemental Figure S4. Briefly, placental chorionic villi 554 were separated from fetal membranes and decidua, washed in DPBS wash, and 555 mechanically homogenized. Placental tissue was then serially digested in Collagenase 556 Digestion Buffer, Trypsin Digestion Buffer, and Collagenase Digestion Buffer 2. 557 Undigested tissue was removed by passage through sterile gauze and 100uM filter. The 558 cell suspension was centrifuged at 257g for 8 min at 4°C, washed, spun again, and 559 resuspended in media. The cells were then suspended in 4mL of 20% Percoll and 5mL 560 of 35% Percoll was underlayed through a #1 glass Pasteur pipette (92). After 561 centrifugation for 30 minutes at 4°C without brake at 1000g, cells were isolated from the 562 interphase layer, washed in media, and spun at 257g for 8 minutes at 4°C. Cell pellets 563 were immunopurified by negative selection by incubation with anti-EGFR (to remove 564 syncytiotrophoblasts) and anti-CD10 (to remove fibroblasts) conjugated to magnetic 565 Dynabeads, prepared as previously described (76), for 20 minutes at 4°C. Tubes were 566 placed on a DynaMagTM magnet for 5 minutes to magnetically bind cytotrophoblasts 567 (anti-EGFR) and fibroblasts (anti-CD10) – allowing media containing unbound placental 568 macrophages to pass through into collection tubes. Cells were cryopreserved in 90% 569 FBS and 10% dimethyl sulfoxide (DMSO) at 1-10 million cells/vial and stored at -80°C 570 for downstream analyses. SARS-CoV-2 viral loads were assessed in all placental 571 tissues using qPCR as previously described, with 40 copies/mL as limit of detection (24, 572 25). 573 574 Single-cell RNA-sequencing (scRNA-Seq) data analysis 575 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 26 576 Sequencing. Cryopreserved HBCs were thawed at 37°C and diluted with RPMI 1640 577 including 10% FBS and 1% Pen/Strep. The cell suspension was centrifuged at 300g for 578 5 min at room temperature, with the brake off. The supernatant was aspirated and the 579 cell pellet was resuspended in media. Dead cells were removed using OptiPrepTM 580 Density Gradient Medium (Sigma) and cell count and viability of cells were calculated 581 using LunaFX7 automated counter. Cells were immediately loaded onto 10x Genomics 582 platform with a loading target of approximately 10,000 viable cells/sample. Libraries 583 were sequenced on an Illumina NextSeq 2000 P3 flowcell machine with a sequencing 584 target of 25,000 reads per cell. 585 586 Initial cluster identification. Raw reads were aligned to reference genome GRCh38 and 587 quantified using Cell Ranger (version 6.0.1, 10x Genomics) and after initial cellranger 588 filtering an average of 6,295 cell/sample and 20,459 reads/cell were present. Putative 589 doublet cells were removed using predictions generated from DoubletFinder (v2.0.3) as 590 were cells containing less than 300 identified genes, which resulted in an object 591 containing 70,817 cells. All samples were integrated to remove batch effects from 592 individuals using the Seurat Single Cell Transform workflow (Seurat version 4.3.0) with 593 the top 2,000 variable features. Cells were clustered using the Louvain algorithm on the 594 shared nearest neighbor graph and visualized by UMAP using the first 30 principal 595 components. Several clustering resolutions were used to scan through the data to 596 identify a resolution (0.3) that allowed us to identify top-level cell types based on marker 597 genes. Marker genes were identified using the Wilcoxon rank sum test with the following 598 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 27 parameters: only.pos = TRUE, min.pct = 0.2, logfc.threshold = 0.5. Additionally, 599 expression of well-known cell-type markers was assessed to refine top-level identities: 600 hofbauer, fibroblasts, NK cells/CD8 T cells, VECs (vascular endothelial cells), EVTs 601 (extravillous trophoblasts), RBCs (red blood cells), B cells, and neutrophils 602 (Supplemental Figure S1). For subsequent analyses, we created a subset of the data 603 including only cells identified as “Hofbauer,” which included macrophage and monocyte 604 populations. 605 606 Subcluster analysis. The data were re-integrated and processed similarly as described 607 above to identify macrophage/monocyte subclusters. Only high-quality cells were 608 retained (mitoRatio 1000 and < 9681, or 3 standard deviations 609 above the mean UMI count). The number of subclusters was optimized by iteratively 610 clustering across several cluster resolutions, and identifying the resolution that provided 611 non-redundant clusters (resolution = 0.3) as determined by marker gene identification 612 with Seurat’s Wilconcon rank-sum test (only.pos = TRUE, min.pct = 0.3, and 613 logfc.threshold = 0.5). Subclusters were then assigned as HBCs (0-7), PAMMs or 614 Monocytes based on marker genes. To delineate fetal from maternal origin of 615 subclusters, we evaluated sex-specific markers using only cells from placentas with a 616 male fetus (N=10). This allowed for maternal vs fetal cell differentiation, as fetal cells 617 would be expected to have increased expression of the male-specific Y-linked gene 618 DEAD-Box Helicase 3 Y-Linked (DDX3Y) and maternal cells would exhibit high 619 expression of the female-specific gene X-inactive specific transcript (XIST). The 620 macrophage cluster with high expression of XIST consistent with maternal origin was 621 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 28 annotated as maternal placenta-associated macrophages and monocytes (PAMMs). To 622 further support cell cluster annotation, we also compared the overall gene expression 623 profiles of each cluster to a previously published single-cell dataset derived from human 624 first-trimester placenta and decidua (28). 625 626 Differential gene expression by SARS-CoV-2 status. For differential gene expression 627 analysis between cells from SARS-CoV-2 positive cases and negative controls, we 628 used the Seurat FindMarkers() within each cluster with the following parameters: 629 test.use = “MAST”, min.pct =0.3, logfc.threshold = 0.2, latent.vars = “donor”. Genes with 630 Benjamini-Hochberg adjusted p-value < 0.05 were considered significant. 631 632 Functional enrichment analyses. Gene Ontology (GO) Biological Process enrichment 633 analysis was performed on both cluster marker genes and SARS-CoV-2 differentially 634 expressed genes in each cluster using R package clusterProfler (v. 3.18.1) (93) and 635 underlying database AnnotationDb org.Hs.eg.db (v3.12.0). GO terms were considered 636 significant with adjusted p-value < 0.05. IPA Canonical Pathway and Diseases and 637 Functions analysis were performed with IPA (Content Version: 90348151) with 638 pathways considered significant with adjusted p-value <0.05. 639 640 Derivation of Hofbauer cells transdifferentiated toward microglia-like cells (HBC-641 iMGs) by direct cytokine reprogramming 642 HBC-iMGs were derived from HBCs using previously described methods (62–64), with 643 modifications as noted. Briefly, thawed HBCs were plated on Geltrex-coated 24-well 644 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 29 plates (1 × 106 cells in 0.5 mL per well) or 96-well plates (2 x 105 cells in 0.1mL per well) 645 depending on cell availability. After cells were incubated at 37°C for 24 h, the media was 646 completely replaced with RPMI 1640 including GlutaMAX, 1% penicillin–streptomycin, 647 100 ng/mL of human recombinant IL-34 (Peprotech), and 10 ng/mL of GM-CSF 648 (Peprotech). At day 6 of transdifferentiation, the cultures were assayed and 649 subsequently fixed with 4% PFA to perform endpoint analysis using 650 immunocytochemistry. 651 652 HBC-iMG Immunocytochemistry 653 HBC-iMGs were washed twice with PBS and blocked for 1 h with 5% FBS and 0.3% 654 Triton-X (Sigma Aldrich) in PBS at room temperature. Next, they were washed three 655 times with 1% FBS in PBS and incubated with primary antibodies in 5% FBS and 0.1% 656 Triton-X overnight at 4°C (Anti-IBA1, 1:500; Abcam #ab5076; Anti-TMEM119, 1:500; 657 Abcam #ab18537; Anti-CX3CR1, 1:100, Abcam #ab8021; Anti-PU.1, 1:1000, Abcam 658 #ab183327, and Anti-P2RY12, 1:100, Alomone Labs). Cells were then washed three 659 times with 1% FBS in PBS and incubated in secondary antibodies (Invitrogen Alexa 660 Fluor, 1:500) and Hoechst 33342 (1:5000) in 5% FBS and 0.1% Triton-X in PBS for 661 45 min at 4°C. Cells were washed two final times and imaged using the IN Cell Analyzer 662 6000 (Cytiva). Marker characterization was analyzed using CellProfiler (94). Cells were 663 segmented using one of the four microglia markers used and percent marker positive 664 cells calculated by dividing the number of marker positive cells by the number of 665 identified nuclei, per image. A total of 12 20x images per sample were analyzed. 666 667 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 30 Synaptosome derivation and phagocytosis assay 668 Synaptosome generation from neural progenitor cell cultures. Induced pluripotent stem 669 cells were reprogrammed from fibroblasts and used to derive expandable neural 670 progenitor cells, and large-scale differentiated neural cultures, as previously described 671 (62–64). After media removal, neural cultures were collected by scraping in 10ml per 672 T1000 flask 1× gradient buffer (ice-cold 0.32 M sucrose, 600 mg/L Tris, 1 mM 673 NaH3CO3, 1 mM EDTA, pH 7.4 with added HALT protease inhibitor—Thermo-Fisher 674 Scientific # 78442) and homogenized using a Dounce Tissue Grinder (Wheaton 675 #357544 15ml) with the ‘tight’ plunger. Homogenate was collected and centrifuged at 676 700g g for 10 min at 4°C to remove large debris. The gradient buffer was removed by 677 aspiration and saved on ice then the pellet was resuspended in 10 mL of 1× gradient 678 buffer and homogenization was repeated as above. The final homogenates were 679 combined and centrifuged at 15,000g for 15 min at 4°C. This second pellet was 680 resuspended in 12ml 1× gradient buffer and slowly added on top of a pre-formed 681 sucrose gradient in Ultracentrifuge Tubes (Beckman Coulter Ultra-Clear #344058) 682 containing 12ml each 1.2 M (bottom) and 0.85 M (middle) sucrose layers. The gradients 683 were centrifuged using Ultracentrifuge swinging bucket rotor #SW32TI at 26,500 RPM 684 (~80,000g) for 2 h at 4°C with the brake off. The synaptosome band (in between 0.85 685 and 1.2 M sucrose layers) was removed using a 5-mL syringe and 19Gx1 ½″ needle, 686 diluted with 5-fold 1x gradient buffer then centrifuged at 20,000g for 20 min at 4°C. The 687 final pellet was resuspended in an appropriate volume of 1× gradient buffer containing 688 1 mg/mL bovine serum albumin (BSA) with HALT protease and phosphatase inhibitors 689 added, aliquoted and slowly frozen at −80°C. Protein concentration was measured by 690 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 31 BCA and enrichment of pre-synaptic (synapsin, SNAP-25) and post-synaptic (PSD-95) 691 markers was determined by western blot analysis. 692 693 Phagocytosis Assay. Synaptosomes were thawed and labeled with pHrodo Red SE 694 (Thermo-Fisher Scientific #P36600) at 1:2 (mg dye: mg synaptosome) and incubated at 695 room temperature for 1 hour. Labeled synaptosomes were sonicated for 1 hour before 696 adding to HBC-iMGs at 15mg total protein per well in 24-well plates, or 3mg per well in 697 96-well format. HBC-iMGs with synaptosomes were incubated at 37°C for three hours 698 and then fixed with 4% PFA for 15 minutes at room temperature. Immunocytochemistry 699 was performed to quantify phagocytosis, with images analyzed in CellProfiler (v4.2.4). 700 HBC-iMGs were segmented as described below using IBA1 staining and phagocytic 701 index was calculated by dividing the signal area of pHrodo Red by the number of 702 segmented cells, per image. 703 704 Image analysis 705 CellProfiler (v4.2.4) was used to identify cellular and subcellular structures in the 706 confocal images (94). The module CorrectIlluminationCalculate and 707 CorrectIlluminationApply were used in all channels to correct uneven illumination and 708 uneven background. Nuclei and cell bodies were each identified using 709 IdentifyPrimaryObjects. Specifically, pixel diameter ranges and the automatic 710 thresholding method Otsu were applied. The module RelateObjects was used to drop 711 structures incorrectly identified as nuclei by ensuring they were only accepted when 712 they have a surrounding microglia-like cell. IdentifySecondaryObjects was used to more 713 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 32 accurately outline cells around these nuclei and avoid debris. The module 714 IdentifyTertiaryObjects identified cytoplasm by subtracting the area of the nucleus from 715 the cell. IdentifySecondaryObjects was used with Otsu thresholding to identify and 716 measure synaptosomes. Background red signal was eliminated by increasing the lower 717 bounds on the automatic threshold, using reference images from Cytochalasin 718 treatment as a positive control of diminished phagocytosis. MaskObjects was used with 719 the cell and synaptosome objects to omit red signal from outside the cell. Overlays of 720 the outlines of all generated image structures were created for quality check purposes 721 using the OverlayOutlines module. Cell area, count, and signal intensity were created 722 with MeasureObjectSizeShape, MeasureObjectIntensity, ExportToSpreadSheet, and 723 ExportToDatabase. RStudio 2 (1.4) was used to organize the metadata exported from 724 CellProfiler. Phagocytic index was calculated as area of Synaptosomes divided by cell 725 count per image. As a confirmation, the integrated intensity of Synaptosome signal 726 divided by cell area was also checked to make sure both measures corresponded. 727 Images containing >80 cells were omitted due to procedure inaccuracy with dense 728 fields. Outliers were excluded by calculating a phagocytic index threshold of 3 SD above 729 the mean. Morphology data was produced using cell level metadata from CellProfiler 730 followed by a cleaning process matching the field level dataset cleaning. Cells with an 731 area or synaptosome area of greater than the mean plus 3 SD were omitted. In total, 12 732 20x images per sample were analyzed. 733 734 Statistics 735 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 33 Group differences were assessed using Mann-Whitney U tests. P values less than 0.05 736 were considered statistically significant. Dark lines represent median and dotted lines 737 interquartile range, unless otherwise specified. Statistical analyses were performed in 738 GraphPad Prism (version 9.3). 739 740 Study approval 741 This study was approved by the Mass General Brigham Institutional Review Board 742 (Protocol #2020P003538). All participants provided written informed consent prior to 743 participation. 744 745 Data availability 746 Sequencing data will be made available for download on GEO upon acceptance. R 747 code supporting the conclusions of this manuscript is made available here: 748 https://github.com/rbatorsky/covid-placenta-edlow. 749 750 Author Contributions 751 L.L.S. and R.A.B. contributed equally, and as co-first authors. A.G.E. conceived the 752 study and, together with R.H.P., designed the experiments. Acquisition of data: L.L.S., 753 R.A.B., R.M.D., L.T.M., S.M.B., S.D.S., J.Z.L., S.B., J.E.H., B.A.G., R.H.P., A.G.E. 754 Analysis and interpretation of data: L.L.S., R.A.B., R.M.D., L.T.M., O.K., S.D.S., A.M.C., 755 B.A.G., R.H.P., A.G.E. Drafting of the manuscript: L.L.S., R.A.B., R.M.D., L.T.M., A.G.E. 756 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 34 Revising the manuscript critically for important intellectual content: L.L.S., R.A.B., 757 R.M.D., L.T.M., S.M.B., O.K., J.E.H., S.D.S., A.M.C., J.Z.L., B.A.G., R.H.P., A.G.E. All 758 authors have given final approval for submission. 759 760

NIH/NICHD: 1R01HD100022-01, 3R01HD100022-02S2, and 761 1U19AI167899-01 to A.G.E; 1K12HD103096 to L.L.S.; NIH/NIMH: 1RF1MH132336-01 762 to A.G.E. and R.H.P.; NIH: 5T32HG010464 to R.M.D.; B.A.G. was supported in part by 763 the Geisel School of Medicine at Dartmouth’s Center for Quantitative Biology by 764 NIH/NIGMS: P20GM130454. J.Z.L. was supported by a grant from the Massachusetts 765 Consortium for Pathogen Readiness (MassCPR). 766 767 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 35

768 1. Mednick SA, et al. Adult schizophrenia following prenatal exposure to an influenza 769 epidemic. Arch Gen Psychiatry. 1988;45(2):189–192. 770 2. Barr CE, Mednick SA, Munk-Jorgensen P. Exposure to influenza epidemics during 771 gestation and adult schizophrenia. A 40-year study. Arch Gen Psychiatry. 772 1990;47(9):869–874. 773 3. Cooper SJ. Schizophrenia after prenatal exposure to 1957 A2 influenza epidemic. Br 774 J Psychiatry. 1992;161:394–396. 775 4. Al-Haddad BJS, et al. Long-term Risk of Neuropsychiatric Disease After Exposure to 776 Infection In Utero. JAMA Psychiatry. 2019;76(6):594–602. 777 5. Atladóttir HO, et al. Maternal infection requiring hospitalization during pregnancy and 778 autism spectrum disorders. J Autism Dev Disord. 2010;40(12):1423–1430. 779 6. Edlow AG, et al. Sex-Specific Neurodevelopmental Outcomes Among Offspring of 780 Mothers With SARS-CoV-2 Infection During Pregnancy. JAMA Netw Open. 781 2023;6(3):e234415. 782 7. Edlow AG, et al. Neurodevelopmental Outcomes at 1 Year in Infants of Mothers Who 783 Tested Positive for SARS-CoV-2 During Pregnancy. JAMA Netw Open. 784 2022;5(6):e2215787. 785 8. Santos CAD, et al. Developmental impairment in children exposed during pregnancy 786 to maternal SARS-COV2: A Brazilian cohort study. Int J Infect Dis. [published online 787 ahead of print: December 5, 2023]; https://doi.org/10.1016/j.ijid.2023.12.001 788 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 36 9. Pinheiro GSMA, et al. Effects of intrauterine exposure to SARS-CoV-2 on infants’ 789 development: a rapid review and meta-analysis. Eur J Pediatr. 2023;182(5):2041–2055. 790 10. Kotlyar AM, et al. Vertical transmission of coronavirus disease 2019: a systematic 791 review and meta-analysis. Am J Obstet Gynecol. 2020;0(0). 792 https://doi.org/10.1016/j.ajog.2020.07.049 793 11. Woodworth KR, et al. Birth and Infant Outcomes Following Laboratory-Confirmed 794 SARS-CoV-2 Infection in Pregnancy - SET-NET, 16 Jurisdictions, March 29-October 14, 795 2020. MMWR Morb Mortal Wkly Rep. 2020;69(44):1635–1640. 796 12. Flaherman VJ, et al. Infant Outcomes Following Maternal Infection with SARS-CoV-797 2: First Report from the PRIORITY Study. Clin Infect Dis. [published online ahead of 798 print: 2020]; https://doi.org/10.1093/cid/ciaa1411 799 13. Dumitriu D, et al. Outcomes of Neonates Born to Mothers With Severe Acute 800 Respiratory Syndrome Coronavirus 2 Infection at a Large Medical Center in New York 801 City. JAMA Pediatr. [published online ahead of print: October 12, 2020]; 802 https://doi.org/10.1001/jamapediatrics.2020.4298 803 14. Otero AM, Antonson AM. At the crux of maternal immune activation: Viruses, 804 microglia, microbes, and IL-17A. Immunol Rev. 2022;311(1):205–223. 805 15. Simões LR, et al. Maternal immune activation induced by lipopolysaccharide 806 triggers immune response in pregnant mother and fetus, and induces behavioral 807 impairment in adult rats. J Psychiatr Res. 2018;100:71–83. 808 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 37 16. Baines KJ, et al. Maternal Immune Activation Alters Fetal Brain Development and 809 Enhances Proliferation of Neural Precursor Cells in Rats. Front Immunol. 2020;11:1145. 810 17. Careaga M, Murai T, Bauman MD. Maternal Immune Activation and Autism 811 Spectrum Disorder: From Rodents to Nonhuman and Human Primates. Biol Psychiatry. 812 2017;81(5):391–401. 813 18. Edlow AG, et al. Placental Macrophages: A Window Into Fetal Microglial Function in 814 Maternal Obesity. Int J Dev Neurosci. [published online ahead of print: November 20, 815 2018]; https://doi.org/10.1016/j.ijdevneu.2018.11.004 816 19. Batorsky R, et al. Hofbauer cells and fetal brain microglia share transcriptional 817 profiles and responses to maternal diet-induced obesity. bioRxiv. 818 2023;2023.12.16.571680. 819 20. Ginhoux F, et al. Fate Mapping Analysis Reveals That Adult Microglia Derive from 820 Primitive Macrophages. Science. 2010;330(6005):841–845. 821 21. Gomez Perdiguero E, et al. Tissue-resident macrophages originate from yolk-sac-822 derived erythro-myeloid progenitors. Nature. 2015;518(7540):547–551. 823 22. Ginhoux F, Prinz M. Origin of microglia: current concepts and past controversies. 824 Cold Spring Harb Perspect Biol. 2015;7(8):a020537. 825 23. Ginhoux F, et al. Origin and differentiation of microglia. Front Cell Neurosci. 826 2013;7:45. 827 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 38 24. Fajnzylber J, et al. SARS-CoV-2 viral load is associated with increased disease 828 severity and mortality. Nat Commun. 2020;11(1):5493. 829 25. Edlow AG, et al. Assessment of Maternal and Neonatal SARS-CoV-2 Viral Load, 830 Transplacental Antibody Transfer, and Placental Pathology in Pregnancies During the 831 COVID-19 Pandemic. JAMA Netw Open. 2020;3(12):e2030455. 832 26. Tang Z, et al. Isolation of Hofbauer Cells from Human Term Placentas with High 833 Yield and Purity: ISOLATION OF PLACENTAL HOFBAUER CELLS. Am J Reprod 834 Immunol. 10/2011;66(4):336–348. 835 27. Megli C, Coyne CB. Gatekeepers of the fetus: Characterization of placental 836 macrophages. J Exp Med. 2021;218(1). https://doi.org/10.1084/jem.20202071 837 28. Suryawanshi H, et al. A single-cell survey of the human first-trimester placenta and 838 decidua. Sci Adv. 2018;4(10):eaau4788. 839 29. Pantazi P, et al. Placental macrophage responses to viral and bacterial ligands and 840 the influence of fetal sex. iScience. 2022;25(12):105653. 841 30. Heck TG, et al. Suppressed anti-inflammatory heat shock response in high-risk 842 COVID-19 patients: lessons from basic research (inclusive bats), light on conceivable 843 therapies. Clin Sci . 2020;134(15):1991–2017. 844 31. Gao X, et al. PDK4 Decrease Neuronal Apoptosis via Inhibiting ROS-ASK1/P38 845 Pathway in Early Brain Injury After Subarachnoid Hemorrhage. Antioxid Redox Signal. 846 2022;36(7–9):505–524. 847 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 39 32. Das H, et al. Kruppel-like factor 2 (KLF2) regulates proinflammatory activation of 848 monocytes. Proc Natl Acad Sci U S A. 2006;103(17):6653–6658. 849 33. Vijayan V, Wagener FADTG, Immenschuh S. The macrophage heme-heme 850 oxygenase-1 system and its role in inflammation. Biochem Pharmacol. 2018;153:159–851 167. 852 34. Hou Y, et al. FABP5 controls macrophage alternative activation and allergic asthma 853 by selectively programming long-chain unsaturated fatty acid metabolism. Cell Rep. 854 2022;41(7):111668. 855 35. Dong B, et al. Macrophage-Related SPP1 as a Potential Biomarker for Early Lymph 856 Node Metastasis in Lung Adenocarcinoma. Front Cell Dev Biol. 2021;9:739358. 857 36. Li Y, et al. Nicotinamide N -methyltransferase promotes M2 macrophage 858 polarization by IL6 and MDSC conversion by GM-CSF in gallbladder carcinoma. 859 Hepatology. 2023;78(5):1352–1367. 860 37. Thomas JR, et al. The Ontogeny and Function of Placental Macrophages. Front 861 Immunol. 2021;12:771054. 862 38. Sasaki Y, et al. Iba1 is an actin-cross-linking protein in macrophages/microglia. 863 Biochem Biophys Res Commun. 2001;286(2):292–297. 864 39. Thomas JR, et al. Phenotypic and functional characterization of first-trimester 865 human placental macrophages, Hofbauer cells. J Exp Med. 2021;218(1). 866 https://doi.org/10.1084/jem.20200891 867 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 40 40. Doratt BM, et al. Mild/asymptomatic COVID-19 in unvaccinated pregnant mothers 868 impairs neonatal immune responses. JCI Insight. 2023;8(19). 869 https://doi.org/10.1172/jci.insight.172658 870 41. Ripoll VM, et al. Gpnmb is induced in macrophages by IFN-gamma and 871 lipopolysaccharide and acts as a feedback regulator of proinflammatory responses. J 872 Immunol. 2007;178(10):6557–6566. 873 42. Zhang H, et al. GPNMB plays an active role in the M1/M2 balance. Tissue Cell. 874 2022;74:101683. 875 43. Yaseen H, et al. Galectin-1 Facilitates Macrophage Reprogramming and Resolution 876 of Inflammation Through IFN-β. Front Pharmacol. 2020;11:901. 877 44. Tugendreich SM. Understanding biological mechanisms in transcriptomics or 878 proteomics datasets with Ingenuity Pathway Analysis (IPA) and Analysis 879 Matchhttps://www.dkfz.de/genomics-880 proteomics/fileadmin/Ingenuity/IPA_AnalysisMatch_White_paper.pdf. 881 45. Abdalla HB, et al. Activation of PPAR-γ induces macrophage polarization and 882 reduces neutrophil migration mediated by heme oxygenase 1. Int Immunopharmacol. 883 2020;84:106565. 884 46. Viola A, et al. The Metabolic Signature of Macrophage Responses. Front Immunol. 885 2019;10:1462. 886 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 41 47. Hong C, et al. Constitutive activation of LXR in macrophages regulates metabolic 887 and inflammatory gene expression: identification of ARL7 as a direct target. J Lipid Res. 888 2011;52(3):531–539. 889 48. Szeto A, et al. Regulation of the macrophage oxytocin receptor in response to 890 inflammation. Am J Physiol Endocrinol Metab. 2017;312(3):E183–E189. 891 49. Chen YQ, Fisher JH, Wang MH. Activation of the RON receptor tyrosine kinase 892 inhibits inducible nitric oxide synthase (iNOS) expression by murine peritoneal exudate 893 macrophages: phosphatidylinositol-3 kinase is required for RON-mediated inhibition of 894 iNOS expression. J Immunol. 1998;161(9):4950–4959. 895 50. Yoshizaki T, et al. SIRT1 inhibits inflammatory pathways in macrophages and 896 modulates insulin sensitivity. Am J Physiol Endocrinol Metab. 2010;298(3):E419-28. 897 51. A-Gonzalez N, et al. Phagocytosis imprints heterogeneity in tissue-resident 898 macrophages. J Exp Med. 2017;214(5):1281–1296. 899 52. Mosser C-A, et al. Microglia in CNS development: Shaping the brain for the future. 900 Prog Neurobiol. 2017;149–150:1–20. 901 53. Frost JL, Schafer DP. Microglia: Architects of the Developing Nervous System. 902 Trends Cell Biol. 2016;26(8):587–597. 903 54. Squarzoni P, et al. Microglia modulate wiring of the embryonic forebrain. Cell Rep. 904 2014;8(5):1271–1279. 905 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 42 55. Cunningham CL, Martínez-Cerdeño V, Noctor SC. Microglia regulate the number of 906 neural precursor cells in the developing cerebral cortex. J Neurosci. 2013;33(10):4216–907 4233. 908 56. Hattori Y. The behavior and functions of embryonic microglia. Anat Sci Int. 909 2022;97(1):1–14. 910 57. Koyama R, Ikegaya Y. Microglia in the pathogenesis of autism spectrum disorders. 911 Neurosci Res. 2015;100:1–5. 912 58. Zhan Y, et al. Deficient neuron-microglia signaling results in impaired functional 913 brain connectivity and social behavior. Nat Neurosci. 2014;17(3):400–406. 914 59. Knuesel I, et al. Maternal immune activation and abnormal brain development 915 across CNS disorders. Nat Rev Neurol. 2014;10(11):643–660. 916 60. Bian Z, et al. Deciphering human macrophage development at single-cell resolution. 917 Nature. 2020;582(7813):571–576. 918 61. Askenase MH, et al. Longitudinal transcriptomics define the stages of myeloid 919 activation in the living human brain after intracerebral hemorrhage. Sci Immunol. 920 2021;6(56). https://doi.org/10.1126/sciimmunol.abd6279 921 62. Sellgren CM, et al. Patient-specific models of microglia-mediated engulfment of 922 synapses and neural progenitors. Mol Psychiatry. 2017;22(2):170–177. 923 63. Sellgren CM, et al. Increased synapse elimination by microglia in schizophrenia 924 patient-derived models of synaptic pruning. Nat Neurosci. 2019;22(3):374–385. 925 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 43 64. Sheridan SD, et al. Umbilical cord blood-derived microglia-like cells to model 926 COVID-19 exposure. Transl Psychiatry. 2021;11(1):179. 927 65. Satoh J-I, et al. TMEM119 marks a subset of microglia in the human brain. 928 Neuropathology. 2016;36(1):39–49. 929 66. Stratoulias V, et al. Microglial subtypes: diversity within the microglial community. 930 EMBO J. 2019;38(17):e101997. 931 67. Torres-Platas SG, et al. Morphometric characterization of microglial phenotypes in 932 human cerebral cortex. J Neuroinflammation. 2014;11:12. 933 68. Tremblay M-È, et al. The role of microglia in the healthy brain. J Neurosci. 934 2011;31(45):16064–16069. 935 69. Leyh J, et al. Classification of Microglial Morphological Phenotypes Using Machine 936 Learning. Front Cell Neurosci. 2021;15:701673. 937 70. Menassa DA, et al. The spatiotemporal dynamics of microglia across the human 938 lifespan. Dev Cell. 2022;57(17):2127-2139.e6. 939 71. Sureshchandra S, et al. Single-cell RNA sequencing reveals immunological rewiring 940 at the maternal-fetal interface following asymptomatic/mild SARS-CoV-2 infection. Cell 941 Rep. 2022;39(11):110938. 942 72. Lu-Culligan A, et al. Maternal respiratory SARS-CoV-2 infection in pregnancy is 943 associated with a robust inflammatory response at the maternal-fetal interface. Med (N 944 Y). 2021;2(5):591-610.e10. 945 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 44 73. Argueta LB, et al. Inflammatory responses in the placenta upon SARS-CoV-2 946 infection late in pregnancy. iScience. 2022;25(5):104223. 947 74. Bordt EA, et al. Maternal SARS-CoV-2 infection elicits sexually dimorphic placental 948 immune responses. Sci Transl Med. 2021;13(617):eabi7428. 949 75. Gao L, et al. Single-cell analysis reveals transcriptomic and epigenomic impacts on 950 the maternal-fetal interface following SARS-CoV-2 infection. Nat Cell Biol. 951 2023;25(7):1047–1060. 952 76. Doratt BM, et al. Mild/Asymptomatic Maternal SARS-CoV-2 Infection Leads to 953 Immune Paralysis in Fetal Circulation and Immune Dysregulation in Fetal-Placental 954 Tissues. bioRxiv. [published online ahead of print: May 11, 2023]; 955 https://doi.org/10.1101/2023.05.10.540233 956 77. Garcia-Flores V, et al. Maternal-fetal immune responses in pregnant women 957 infected with SARS-CoV-2. Nat Commun. 2022;13(1):320. 958 78. Gomez-Lopez N, et al. Pregnancy-specific responses to COVID-19 revealed by 959 high-throughput proteomics of human plasma. Commun Med. 2023;3(1):48. 960 79. Shook LL, et al. COVID-19 in pregnancy: implications for fetal brain development. 961 Trends Mol Med. 2022;28(4):319–330. 962 80. Carbonnel M, et al. Plasticity of natural killer cells in pregnant patients infected with 963 SARS-CoV-2 and their neonates during childbirth. Front Immunol. 2022;13:893450. 964 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 45 81. Foo S-S, et al. The systemic inflammatory landscape of COVID-19 in pregnancy: 965 Extensive serum proteomic profiling of mother-infant dyads with in utero SARS-CoV-2. 966 Cell Rep Med. 2021;2(11):100453. 967 82. Matute JD, et al. Single-cell immunophenotyping of the fetal immune response to 968 maternal SARS-CoV-2 infection in late gestation. Pediatr Res. [published online ahead 969 of print: November 8, 2021]; https://doi.org/10.1038/s41390-021-01793-z 970 83. Tang G, et al. Loss of mTOR-dependent macroautophagy causes autistic-like 971 synaptic pruning deficits. Neuron. 2014;83(5):1131–1143. 972 84. Carter M, et al. Maternal Immune Activation and Interleukin 17A in the Pathogenesis 973 of Autistic Spectrum Disorder and Why It Matters in the COVID-19 Era. Front 974 Psychiatry. 2022;13:823096. 975 85. Malkova NV, et al. Maternal immune activation yields offspring displaying mouse 976 versions of the three core symptoms of autism. Brain Behav Immun. 2012;26(4):607–977 616. 978 86. Gomez-Lopez N, et al. Distinct Cellular Immune Responses to SARS-CoV-2 in 979 Pregnant Women. J Immunol. 2022;208(8):1857–1872. 980 87. Paolicelli RC, et al. Synaptic pruning by microglia is necessary for normal brain 981 development. Science. 2011;333(6048):1456–1458. 982 88. Kim HJ, et al. Deficient autophagy in microglia impairs synaptic pruning and causes 983 social behavioral defects. Mol Psychiatry. 2017;22(11):1576–1584. 984 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 46 89. Lenz KM, Nelson LH. Microglia and Beyond: Innate Immune Cells As Regulators of 985 Brain Development and Behavioral Function. Front Immunol. 2018;9:698. 986 90. Baines KJ, West RC. Sex differences in innate and adaptive immunity impact fetal, 987 placental, and maternal health. Biol Reprod. 2023;109(3):256–270. 988 91. Weinhard L, et al. Sexual dimorphism of microglia and synapses during mouse 989 postnatal development: Sexual Dimorphism in Microglia and Synapses. Dev Neurobiol. 990 2018;78(6):618–626. 991 92. Bordt EA, et al. Isolation of Microglia from Mouse or Human Tissue. STAR Protoc. 992 2020;1(1). https://doi.org/10.1016/j.xpro.2020.100035 993 93. Wu T, et al. clusterProfiler 4.0: A universal enrichment tool for interpreting omics 994 data. Innovation (Camb). 2021;2(3):100141. 995 94. Stirling DR, et al. CellProfiler 4: improvements in speed, utility and usability. BMC 996 Bioinformatics. 2021;22(1):433. 997 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 47 COVID-19 infection in pregnancy COVID-19 severity1 GA at SARS-CoV-2 infection (weeks) Infant Sex Maternal Age* (years) GA at Delivery (weeks) Infant birthweight (grams) No N/A N/A M >40 39 2850 No N/A N/A F <21 40.4 3270 No N/A N/A F 36-40 39 2785 No N/A N/A M 31-35 37.1 3360 No N/A N/A M 26-30 40.3 3670 No N/A N/A M 26-30 38.9 3350 No N/A N/A M 26-30 38.9 3450 No N/A N/A M 36-40 39.3 3200 Yes Severe 24 M 31-35 39 3395 Yes Mild 28 M 26-30 40.1 3125 Yes Mild 11 M 31-35 39.9 3370 Yes Mild 16 M 26-30 40.3 3985 Table 1. Clinical information of study participants. GA: Gestational age. M: male. F: female. N/A: not 998 applicable. No participants had received a COVID -19 vaccine prior to delivery. 1COVID-19 severity was 999 defined by NIH criteria. *Maternal age is provided as a range to preserve participant anonymity. 1000 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 48 1001 Figure 1. Transcriptomic profiles of fetal and maternal macrophages and monocytes isolated from term 1002 placentas with and without SARS -CoV-2 infection during pregnancy. (A) Uniform Manifold Approximation and 1003 Projection (UMAP) visualization of 31,719 high -quality placental macrophage and monocyte cells enriched from 1004 placentas of pregnancies with (N=4) and without (N=8) SARS-CoV-2 infection shows 10 clusters. HBC: Hofbauer cell; 1005 PAMM: placenta-associated maternal monocyte/macrophage. (B) Cluster-specific expression of DDX3Y, expressed 1006 only in fetal cells, and XIST, expressed only in maternal cells, in placentas from individuals carrying a male fetus (N=10). 1007 (C) Correlation of cluster-average gene expression with annotated cell types identified by Suryawanshi et al., Sci Adv, 1008 2018. Each heatmap shows Spearman correlation coefficients. Highest correlation coefficient per cluster is indicated 1009 by black dots. HBC clusters were most highly correlated with Suryawanshi HBC clusters, PAMM cluster most correlated 1010 with decidual macrophages. (D) Heatmap displaying expression (log 2 fold change) of the top 5 marker genes per 1011 cluster. (E) Gene Ontology (GO) Biological Process enrichment analysis for cluster marker genes. GO terms displayed 1012 were curated from the top significant GO terms in each cluster, selecting the processes most relevant to macrophage 1013 function, and reducing redundancy. Gene Count gives the number of genes in the query set that are annotated by the 1014 relevant GO category. GO terms with an adjusted p-value < 0.05 were considered significantly enriched. 1015 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 49 1016 Figure 2. Impact of maternal SARS-CoV-2 infection on Hofbauer cell subclusters. DEG: differentially expressed 1017 genes. (A) Barplot demonstrating proportion of cells per cluster from SARS-CoV-2 positive cases (red) and negative 1018 controls (gray), top panel. Number of DEG upregulated (dark blue) and downregulated (light blue) by SARS-CoV-2 1019 exposure per cluster, bottom panel. (B) Gene Ontology (GO) Biological Process enrichment analysis for DEG. Gene 1020 Count gives the number of genes in the query set that are annotated by the relevant GO category. GO terms with an 1021 adjusted p-value < 0.05 were considered significantly enriched. (C) Ingenuity Pathway Analysis (IPA) of DEG for HBC 1022 clusters 0 (left) and 1 (right). Canonical pathways with absolute Z-score ≥1 and adjusted p-value < 0.05 are shown. 1023 IPA analysis for remaining HBC clusters depicted in Supplement. 1024 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 50 1025 Figure 3. Impact of maternal SARS-CoV-2 on HBC gene programs associated with phagocytosis and neurologic 1026 disease. (A-B) Ingenuity Pathway Analysis (IPA) phagocytosis diseases and functions pathways (A), and neurologic 1027 diseases or functions (B), enriched for ≥ 3 DEGs, with absolute Z-score ≥1 and adjusted P-value < 0.05. Activation Z-1028 score represented by color and number of DEGs by circle size, with red color indicating pathway activation and blue 1029 color indicating suppression. (C-D). Heatmap of gene expression in “Cellular Infiltration by Phagocytes” IPA Pathway 1030 (C) and “Inflammation of Central Nervous System” ( D) by cluster. Color represents gene expression level (log 2 fold 1031 change), *adjusted P-value < 0.05. (E). Module score by subcluster in comparison to cluster-specific gene expression 1032 of single-cell datasets from human brain: Microglia and Border Associated Macrophages (Askenase et al., Sci Immunol, 1033 2021) and Yolk Sac Macrophages and Monocytes (Bian et al., Nature, 2020). Color indicates module score. 1034 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 51 1035 Figure 4. Phenotypic characterization of HBC -iMGs by direct cytokine reprogramming . HBC-iMGs: Hofbauer 1036 cells transdifferentiated toward microglia-like cells. (A) Images of HBC-iMGs from SARS-CoV-2 positive cases (n=4) 1037 and negative controls (n=4), immunostained for microglial markers: IBA1, PU.1, P2RY12, TMEM119. Scale bar = 100 1038 µm. (B) Morphology-smoothed density plots (solidity vs. eccentricity) for SARS-CoV-2 negative and positive samples 1039 as indicated. Cells from negative controls exhibit a more ramified morphology than those from positive cases, 1040 suggestive of a less activated phenotype. Red color shows high density and blue is low density. Representative 1041 confocal microscopy images of amoeboid, bipolar, and ramified HBC -iMGs. Scale bar = 50 µm. ( C) Violin plots 1042 represent distribution of cell solidity (left) and eccentricity (right) measurements from SARS -CoV-2 negative controls 1043 (blue, n=5223 cells) and positive cases (orange, n=237 cells). Solid lines represent median values and dashed lines 1044 interquartile range. Group differences assessed by Mann-Whitney U Test. ****P<0.0001. ns = not significant. 1045 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint 52 1046 Figure 5. Impact of maternal SARS -CoV-2 on HBC -iMG synaptosome engulfment. HBC-iMGs: Hofbauer cells 1047 transdifferentiated toward microglia-like cells. (A) Representative image showing colocalization of pHrodo-red labeled 1048 synaptosomes (SYN) and IBA1 positive HBC-iMGs. Hoechst = nuclear stain. Scale bar = 100 µm. (B) Violin plots of 1049 phagocytic index of image fields from SARS-CoV-2 negative controls (blue, n=187 fields) and positive cases (orange, 1050 n=32 fields). Phagocytic index is calculated as synaptosome area in pixels divided by cell count per image field. Solid 1051 lines represent median values and dashed lines interquartile range. Group differences assessed by Mann-Whitney U 1052 Test. ****P<0.0001. 1053 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for thisthis version posted December 30, 2023. ; https://doi.org/10.1101/2023.12.29.23300544doi: medRxiv preprint