Isolation of MMc GFP-positive SRY-negative cells from mouse embryos
To label MMc cells, we utilized a transgenic mouse line that systemically expresses GFP. The transgenic mouse line was developed by Okabe et al.  and we refer to the GFP gene as “OkabeGFP” in the data analysis section. As shown in Fig. 1a, GFP-heterozygous female mouse was crossed with wild-type male mouse, and MMc cells were detected in the fetus with wild-type genotype as GFP-positive cells. The mother mice were obtained by crossing BALB/cByJJcl wild-type female mice with GFP-homozygous male mice to avoid detecting GFP-positive grand-maternal cells as MMc cells .According to previous studies, MMc cells are detected as early as at the E12.5–E13.5 mouse embryonic stage [23, 24]. Further, MMc cell migration increases throughout gestation, peaking at parturition . Hence, to clarify the initial cell type repertoire of MMc cell population, and to minimize bias from estimating cell types differentiated from MMc cells with multipotent stem potential, we analyzed E14.5 embryos in the current study (Fig. 1b). To obtain MMc cells from individual embryos, the embryos were excised, washed, digested, and the cells mesh-filtered for a dissociated-cell suspension, as previously reported by us . The cell suspension was then processed using magnetic cell sorting system to enrich MMc cells, followed by selection of live MMc cells using flow cytometric cell sorting, and scRNA-seq to obtain transcriptomic data for single cells. Estimation of cell types were done by combining the transcriptomic data for MMc cells with reference data from Tabula Muris database , and performed mixed-clustering analysis. The experimental workflow is shown in Fig. 1c (see Methods for details).
For 52 GFP-negative E14.5 fetuses processed, live GFP-positive cells were detected in 26 embryos. Following preparation of the sequencing library and quality checking, 210 qualified cells were obtained and sequenced using NovaSeq6000 platform (Illumina), with an average read depth of 7,974,354 reads/cell. In each cell, 6772 genes on average were detected [transcripts per million (TPM)>0, Supplementary Figure 1, Additional file 1]. After removing 19 cells with no GFP expression detected in the RNA-seq data, a set of 191 GFP-positive cells isolated from 26 embryos was obtained. Importantly, none of the cells expressed the Y-chromosome specific SRY gene (ENSMUSG00000069036), indicating that the isolated GFP-positive cells were of female origin. These observations suggest that the identified cells were most likely MMc cells. Nonetheless, as a caveat, it is possible that a minor proportion of these cells migrated from female GFP-positive siblings.
Immune-related cell is the major isolated MMc cell type
Before determining the type of the isolated MMc cells by clustering with the Tabula Muris dataset of female origin (20,586 scRNA-seq data for 20 organs and with 79 cell types annotated, Supplementary Figure 2, Additional file 2), we tested the optimal calculation parameters for classifying the Tabula Muris data. Briefly, we constructed an elbow plot representing the relationship between the explained variance and principal components (PCs) (Supplementary Figure 3a, Additional file 3). Two PC sets (PCs 1 to 11, and 1 to 20) were identified as candidate parameters for cell classification. While both PC sets classified the Tabula Muris cells well, cell type annotations using the 11 PC set matched the annotations in Tabula Muris better than those using the 20 PC set (Supplementary Figure 3, Supplementary Table 1, Supplementary Table 2, see also Methods). We therefore used the 11 PC set for cell type estimation of MMc cells. The cell type for each identified cluster was defined based on the most abundant cell type defined in the annotation file for the Tabula Muris project (see Methods). We performed mixed-clustering of the Tabula Muris cell data and the isolated MMc cell data. The isolated GFP-positive MMc cells were classified in 14 clusters (Fig. 2a,b) rather than falling into a single MMc cluster, suggesting that the mix-clustering approach works well for MMc cell type estimation and that MMc cells represent different cell types (Fig. 2c, Fig. 3a).
Cluster 8 (myeloid cell) contained the highest number of MMc cells, and 36% of MMc cells were classified therein. The second largest cluster was cluster 10 (granulocyte), with 27% of MMc cells classified therein. In addition to these two clusters, two clusters of immune-related cell types were identified: cluster 4 (B cell) and cluster 6 (immature T cell). Consistent with the mixed-clustering analysis, we confirmed the expression of immune-related genes in scRNA-seq data for the isolated MMc cells. For example, most of the GFP-positive cells in cluster 8 (myeloid cell) indeed expressed marker genes of dendritic cell (Itgax+, Cd24a+, and Cd68+) and macrophage (Ptprc+, H2-Eb1+, Cd86+, Selplg+, Cd14+, Cd3e–, Cd19–). Of note, cells in cluster 8 also expressed Foxp3, encoding a major transcription factor for regulatory T cell differentiation. While none of the GFP-positive cells in cluster 10 (granulocyte) expressed a full set of marker genes of granulocyte (Ltf+, Pglyrp1+, Lcn2+, Camp+, Mki67–, Stmn1–) or those of granulocyte–monocyte progenitor cells (Flt3+, Kit+, Mpeg1+, Itgb2+, Ahnak+, Pld4+, Cd68+, Hp–), the expression of many of these genes was detected, including one cell that expressed all granulocyte–monocyte progenitor marker genes except the Kit gene. Further, some cells in cluster 10 (granulocyte) expressed marker genes of dendritic cell, macrophage, or monocyte (Ly6c2+, Cx3cr1+, Cd14+, Csflr+, Mrc1+), which we expected to be included in cluster 8. Among the MMc cells in cluster 10 (granulocyte), 47% (24/51 cells) expressed marker genes of invading monocyte (Cd11b+, Csflr+, Ly6c2+, Cd14+; based on the Tabula Muris reference; Supplement–Detailed Discussion of Organ Cell Types). The two GFP-positive cells in cluster 4 (B cell) indeed expressed genes characteristic for circulating B cell (Cd79a+, Cd79b+, Cd74+, Cd19+), with the expression of mature (naive) B cell-related genes (Chchd10+, Cd79a+, Cd79b+, Cd19+, Ms4a1+, Cd74+, Mki67–, Stmn1–). While none of the GFP-positive cells in cluster 6 (immature T cell) expressed a complete set of marker genes for T cell (Ahnak+, Thy1+, Cd3e+, Cd8a+), some of the cells expressed the Cd3 gene, encoding a T-cell receptor component. Further, we have detected autoimmune regulator (Aire) gene expression in four cells (in clusters 14, 18, and 21). These cells showed the gene expression pattern of medullary thymic epithelial cell (Aire+, Cldn3+, Cldn4+) , suggesting that the MMc cells present cell-type–specific antigens, as well as different cell-specific antigens from the mother to the fetal cells.
In addition to the immune-related cell types, we also, unexpectedly, identified MMc cells clustered in several tissue-specific and terminally differentiated cell type clusters, such as cluster 0 (microglial cell, identified as Cx3cr1+, P2ry12+, and Tmem119+; no MMc cell in this cluster expressed all these genes), cluster 1 (fibroblast, identified as Dcn+ and Gsn+; 3 out of 9 cells in the cluster expressed all these genes), cluster 7 (basal cell of the epidermis, identified as Cd34+ and Itga6+; the two MMc cells in this cluster expressed these two genes), cluster 11 (astrocyte, identified as Aldh1l1+, Slc1a3+, and Aqp4+; no MMc cells in this cluster expressed all these genes; “blank (defined by the Tabula Muris annotation file)” cell type was the largest in this cluster), cluster 17 (endothelial cell, identified as Pecam1+; 4 out of 21 MMc cells in this cluster expressed this gene), cluster 18 (luminal epithelial cell of the mammary gland, identified as Krt8+, Krt18+, and Krt19+; the two MMc cells in this cluster expressed these genes), cluster 21 (epithelial cell of the proximal tubule, identified as Vil1+; 2 out of 6 MMc cells in this cluster expressed this gene), cluster 22 (hepatocyte, identified as Alb+, Ttr+, Apoal+, and Serpina1c+), and cluster 27 (epithelial cell of the lung, identified as Pecam1– and EpCAM+; the one cell in this cluster did not express the marker gene). Apart from immune-related or differentiated cell types, 16 cells were classified in cluster 14 (multipotent progenitor cell), suggesting that some MMc cells have stem cell-like cell phenotype, as has been reported previously [7, 18]. This was also supported by the detection of the expression of marker genes of hematopoietic stem cell (Kit+, Stmn1+, Mki67+) in 6 out of 12 cells, and a marker of cell proliferation, namely, Mki67, in 10 out of 12 cells in the cluster 14.
Although the number of identified MMc cells in different embryos differed widely, the presented results suggest that most MMc cells are immune cells, with the remaining cells either proliferating/stem cells, or terminally differentiated cells (Fig. 3b). Further, while a relatively small proportion of maternal stem cells was identified among the analyzed MMc cells (8% in Fig. 3b), frequent detection of these cells in GFP-positive embryos (46% of embryos, Table 1) suggests that different MMc cell types are characterized by different migration ability.
Proportions of immune-related and stem cell-like MMc cells differ in individual embryos
While MMc cells are thought to be present in all individuals, including in health, association of an increased MMc occurrence with a variety of physiological phenomena (including tolerance, regeneration, and fetal tissue damage or enhancement of certain congenital diseases) has been reported. These variable outcomes could be caused by environmental factors, such as pathogens, and/or major histocompatibility complex (MHC) compatibility between the mother and the fetus; however, they might also be explained by differences in the MMc cell type repertoire between individuals. In our previous study, we showed that the frequency of MMc cells largely differs among inter-individual embryos . However, the existence of inter-individual variations in MMc cell types remained to be clarified. Accordingly, in the current study, data for the MMc cells isolated from 26 embryos implied that the proportions of MMc cell types differ between individuals (Fig. 3c; Supplementary Table 3, Additional file 6). Since the potential role of MMc cells in the fetus and the neonate is often related to the immune system, such as immunological tolerance, with activation of cytotoxic profile upon depletion (Castellan F. et al., submitted), we also analyzed the potential variation in the proportion of immune-related MMc cells (defined by clustering analysis with Tabula Muris data) in different embryos. We found statistically significant differences in the proportion of immune-related MMc cells among individual embryos (Table 3, p=4.95´10–6, Fisher’s exact test, two-sided). Similarly, we detected a significant variation in the proportion of MMc proliferating/stem cells (Table 2, p=0.0398, by Fisher’s exact test, two-sided). Taken together, these observations suggest that the MMc cell type repertoire differs between embryos.
Most MMc cells express migration-related proteins
To shed light on the mechanism underpinning MMc cell migration from the mother to the fetus, we searched for genes that were commonly expressed in MMc cells (genes with TPM>1, and detected in more than 90% of MMc cells). After excluding housekeeping genes, such as ribosomal protein genes, we identified few genes that could be involved in MMc cell migration. One of them was interferon-induced transmembrane protein 2 (Ifitm2) gene, encoding a member of the interferon-induced transmembrane (IFITM) family. While studies focusing on this gene are scarce, Iftim3, encoding another member of this family, is expressed in migratory primordial germ cells, and reportedly regulates cell adhesion and differentiation . We also identified syndecan-binding protein (Sdcbp) gene, known to be involved in cell migration and invasion of tumor metastasis in human . As a third candidate, we detected a gene for macrophage migration inhibitory factor (MIF), an important mediator of the innate immune system . Although further functional studies are required, these results imply that these gene could be involved in the migration of MMc cells.