1. Developmental Biology
Download icon

Single nuclei RNA-seq of mouse placental labyrinth development

  1. Bryan Marsh
  2. Robert Blelloch  Is a corresponding author
  1. The Eli and Edythe Broad Center of Regeneration Medicine and Stem Cell Research, Center for Reproductive Sciences, University of California, San Francisco, United States
  2. Department of Urology, University of California, San Francisco, United States
Research Article
  • Cited 0
  • Views 2,038
  • Annotations
Cite this article as: eLife 2020;9:e60266 doi: 10.7554/eLife.60266

Abstract

The placenta is the interface between mother and fetus in all eutherian species. However, our understanding of this essential organ remains incomplete. A substantial challenge has been the syncytial cells of the placenta, which have made dissociation and independent evaluation of the different cell types of this organ difficult. Here, we address questions concerning the ontogeny, specification, and function of the cell types of a representative hemochorial placenta by performing single nuclei RNA sequencing (snRNA-seq) at multiple stages of mouse embryonic development focusing on the exchange interface, the labyrinth. Timepoints extended from progenitor-driven expansion through terminal differentiation. Analysis by snRNA-seq identified transcript profiles and inferred functions, cell trajectories, signaling interactions, and transcriptional drivers of all but the most highly polyploid cell types of the placenta. These data profile placental development at an unprecedented resolution, provide insights into differentiation and function across time, and provide a resource for future study.

Introduction

The placenta links maternal tissues and the embryo, providing necessary support and instructing the development of the embryo. The placenta performs many essential functions including transport of nutrients and exchange of gases and waste between maternal and fetal blood, production and uptake of hormones, and regulation of the maternal immune system (Maltepe and Fisher, 2015; Woods et al., 2018). The mouse placenta is separated into three main regions – the decidua, the junctional zone (JZ), and the labyrinth. Select fetal trophoblast invade into the maternal decidua and remodel and line maternal arteries, facilitating maternal blood to flow into the fetal portion of the placenta (Adamson et al., 2002; Maltepe and Fisher, 2015). The JZ lies directly beneath the decidua and consists of parietal trophoblast giant cells, spongiotrophoblast, and glycogen cells that largely function in hormone secretion and metabolism (Simmons et al., 2007; Woods et al., 2018). Beneath the JZ, closest to the embryo, is the labyrinth, which is specially designed to maximize surface area for gas and nutrient exchange between maternal and fetal blood (Figure 1ASimmons et al., 2008a; Soncin et al., 2015; Woods et al., 2018).

Figure 1 with 1 supplement see all
Nuclear isolation and snRNA-seq of mouse placental cells (E9.5-E14.5).

(A) Schematic showing the main regions of the placenta – Labyrinth, Junctional Zone, Parietal TGC, and Decidua. Removal of the decidual stroma and the allantois is marked by scissors and cut lines. Sections of placentas at E9.5–14.5 stained with DAPI to label nuclei and reveal tissue architecture. The dotted line is drawn at the interface between the junctional zone (left of the line) and the maternal decidual stroma (right) to demonstrate the growth occurring during this time span. (B) Schematic outlining nuclei isolation from tissue, purification using FACS, transcriptome capture, and downstream analysis. (C) Number of nuclei collected at each gestational age and in total. (D) Visualization of the 27,326 nuclei included in the analysis plotted in two dimensions by transcriptome similarity using uniform manifold approximation and projection (UMAP). Each dot represents one nucleus colored according to assignment by clustering analysis (see Materials and methods). Dotted lines encircle clusters with common properties. (E) Expression of Xist in each sample. Female placenta samples express Xist in all nuclei. Male placental samples express Xist only in maternally derived nuclei. The main cell type groupings from D are outlined here also.

Figure 1—source data 1

Marker genes for clusters in complete snRNA-seq data (E9.5-E14.5).

https://cdn.elifesciences.org/articles/60266/elife-60266-fig1-data1-v2.csv

The development of the labyrinth structure begins with fusion of the allantois with the chorion at E8.5 (Cross et al., 2003). This begins a process of invasion and branching morphogenesis resulting in an expansive surface area for gas and nutrient exchange by E10.5 (Soncin et al., 2015). The gas exchange interface in the mouse contains three layers of differentiated trophoblast separating maternal blood from the endothelial cells of the fetal vasculature (Simmons et al., 2008a; Maltepe and Fisher, 2015; Coan et al., 2005). Sinusoidal trophoblast giant cells (S-TGC) reside in the maternal blood space, have demonstrated exocrine functions, and are a source of placental lactogens (Simmons et al., 2008b). S-TGC are attached to the outermost syncytiotrophoblast layer (SynTI) but are perforated allowing maternal blood to contact SynTI (Coan et al., 2005; Woods et al., 2018). Next, are the two SynT layers which together transport gasses and nutrients from the maternal blood to the fetal vasculature (Georgiades et al., 2002; Woods et al., 2018). SynTI is the outer layer in direct contact with maternal blood forming the barrier between maternal blood and fetal tissue. Beneath, is SynTII which functions as an intermediary between SynTI on its apical membrane and the fetal endothelium on its basal side. Proper patterning of each cell type ensures efficient transport of nutrients and waste between maternal and fetal blood, and is necessary for embryonic growth, development, and viability.

Investigating the cells of the placental labyrinth is difficult due to their size and an expansion in the number or ploidy of nuclei. SynT cells are multinucleate and are not amenable to standard dissociation and isolation procedures. As a reference, human SynT are 80–200 µM in diameter and contain 5–15 nuclei meaning their size and fragility do not allow for profiling by FACS or passage through microfluidic channels for droplet-based single-cell RNA sequencing (Liu et al., 2018). Therefore, questions remain about how each cell type is specified, how essential placental functions are distributed between cell types, and how each cell type changes throughout developmental time.

Single-cell RNA sequencing has become the preferred method for understanding the composition of a complex tissue at the transcriptional level. Several studies in both mouse (Nelson et al., 2016; Home and Ghosh, 2019) and human (Vento-Tormo et al., 2018; Suryawanshi et al., 2018) have profiled the placenta using scRNA-seq, however, syncytiotrophoblast are greatly underrepresented in these data. Studies in mouse have either avoided profiling the labyrinth trophoblast, or are unable to distinguish between the two SynT layers (Nelson et al., 2016; Home and Ghosh, 2019). Recently, application of droplet-based transcriptome profiling to nuclei has been shown to provide high quality data and results which recapitulate those from whole cell scRNA-seq (Habib et al., 2017; Lake et al., 2018; Bakken et al., 2018; Ding et al., 2020). To circumvent complications with single-cell dissociation and capture of placental cells, we applied single nuclei RNA-seq (snRNA-seq) to the mouse placenta.

In this study, we used snRNA-seq to profile the murine placental development at four time points spanning placental cell type specification, differentiation, and maturation (E9.5, E10.5, E12.5, and E14.5). We resolve all populations of the labyrinth at single-nucleus resolution and identify distinct markers and transcript signatures for each population. We are able to separate each SynT layer, furthering the understanding of the independent roles of each cell type at the gas exchange interface. Mapping differentiation from stem/progenitor cells to terminal differentiation, we capture novel intermediates and identify putative regulators. We predict signaling interactions between populations of the labyrinth to understand crosstalk between these populations, revealing factors guiding concerted development of SynT and fetal endothelium. Finally, we identify transcription factor (TF) driven gene regulatory networks active in each placental population, uncovering candidate regulators of fate and function in the placenta. By separating and characterizing placental cell types, specifically those in the labyrinth, these data provide a resource for understanding development in vivo, generation of molecular tools in vitro and in vivo, and for future modeling of prenatal pathologies.

Results

Nuclear isolation and snRNA-seq of mouse placental cells (E9.5-E14.5)

To gain a better understanding of the constitution of cell types and developmental trajectories in the mouse placenta, we performed single nuclei droplet-based sequencing of their transcriptomes at four embryonic time points: E9.5, E10.5, E12.5, and E14.5. Given the largely syncytial nature of the placenta, we reasoned that sequencing of nuclei would bypass the problems inherent in using a single-cell RNA-sequencing approach. To decrease contribution from maternal tissues, the placenta was carefully dissected at the boundary between the fetal junctional zone and the maternal decidua (Figure 1A). On the labyrinth side of the placenta, the placenta was separated from embryo proper at the point of attachment with the allantois on the basal side (embryo facing). Nuclei were isolated and sorted for DNA content by FACS enabling the removal of doublets and contaminating cytoplasmic RNA (Figure 1B). Isolated nuclei were subjected to droplet-based sequencing using the 10x Genomics platform.

A total of 27,326 nuclei passed quality control (between 500 and 4000 unique genes identified and fewer than 0.25% mitochondrial reads), ranging from 4124 to 9485 nuclei across the embryonic days (Figure 1C, Supplementary file 1). Clustering of nuclei using the Seurat package identified 26 clusters, then projected in UMAP expression space allowing for an understanding of the relationship among populations as visualized by proximity (Figure 1D, Figure 1—figure supplement 1A–CSatija et al., 2015; Butler et al., 2018). Based on general markers, these populations were separated into five broad groups: trophoblast, decidual stroma, blood cells, endothelial cells, and fetal mesenchyme.

The maternal decidual stroma was identified by the absence of Xist expression in male embryonic tissue (Figure 1E). Marker genes further subdivided the five broad groups into subgroups (Figure 1—figure supplement 1D, Figure 1—source data 1, Supplementary file 2). The fetal mesenchyme group consisted of two distinct subgroups. One (clusters 6, 10, and 24) expressed markers of pericytes (e.g. Acta2) and several growth factors including Wnt5b and Pdgfrb that are critical for development and maintenance of vascular populations (Eaton et al., 2020). The other (clusters 15 and 19) expressed Gata4, Kit, and Pdpn. The nature of the cells in this subgroup is unclear, but immunofluorescence for PDPN showed the cells to be interstitial with long projections that appear to make contact with PECAM positive fetal endothelial cells (Figure 1—figure supplement 1E). The endothelial group consisted of cells expressing markers of vascular endothelium (Cluster 3 - Pecam1, Kdr, Tek), lymphatic endothelium (cluster 26 - Flt4/Vegfr3), and pro-angiogenic and lymphogenic lineages (Clusters 20 and 25 - Vegfc and Igf1, respectively) (Cao et al., 1998; Björndahl et al., 2005). The identification of cells expressing lymphatic markers is intriguing because the placenta is not thought to contain lymphatics. However, lymphatic endothelium has been discovered in human and it is possible a similar cell type exists in mouse (Pique-Regi et al., 2019). The blood cell group consisted of putative erythrocytes (Cluster 13 - Hbb-y, Hba-x), macrophages (Cluster 17 - Mrc1+; Martinez and Gordon, 2014), B-cells (Cluster 21 - Bank1, Btla; Aiba et al., 2006), T-cells (Cluster 22 – Gzmc, Slamf1, Cd84; Griewank et al., 2007, Veillette, 2006), and natural killer cells (Cluster 23 Cd244; Lee et al., 2004). Together, these data show that snRNA-seq has the sensitivity and specificity to identify the multitude of cell types in the developing placenta.

Sub-clustering identifies the trophoblast subpopulations of the labyrinth and junctional zone

Trophoblast cells are unique to the placenta and perform the majority of its specialized functions. Unsurprisingly, they also make up the largest of the five broad groups of nuclei (Figure 1D). To gain a deeper understanding of the trophoblast populations, we used Seurat to cluster only trophoblast nuclei. Nuclei collected at each individual timepoint were analyzed separately and those assigned a trophoblast identity were then integrated and visualized on a single UMAP. This analysis resulted in 13 clusters consisting of 16,386 nuclei, with clusters ranging in size from 159 to 2906 nuclei (Figure 2A, Figure 2—figure supplement 1A). The clusters formed five appendages arising from a central body.

Figure 2 with 3 supplements see all
Sub-clustering of the trophoblast nuclei identifies the cell types of the labyrinth and junctional zone.

(A) UMAP showing the 16,836 nuclei included in the trophoblast subset, clustered and plotted according to transcriptome similarity. Clusters were annotated according to canonical marker genes and named below. Inset shows the number of nuclei collected at each gestational age. (B) Dot plot showing average expression and percent of nuclei in each cluster expressing canonical and novel marker genes identified for each cluster. Genes listed on the x-axis and clusters on the y-axis. (C–E) Nuclei along the differentiation from LaTP to SynTII, SynTI, and S-TGC were ordered by pseudotime using Slingshot. The nuclei included along each pseudotime axis shown at top. The expression of select genes (y-axis) representing each differentiation are shown in each heatmap. Each column represents a nucleus organized in pseudotime proceeding from left to right along the x-axis. (F) Expression of genes unique to several trophoblast populations projected in UMAP space (top) and localization of corresponding protein in E12.5 mouse placenta sections by immunofluorescence staining (Middle – 20x, Scale Bar = 200 µM; Bottom – high magnification inset area shown by white dashed line). Separation of labyrinth (La) and junctional zone (JZ) shown by the orange dashed line.

Figure 2—source data 1

Marker genes for clusters in subclustered trophoblast nuclei.

https://cdn.elifesciences.org/articles/60266/elife-60266-fig2-data1-v2.csv
Figure 2—source data 2

Expression of the top 200 variable genes along the pseudotime ordering of cells from LaTP to SynTII (Figure 2C).

https://cdn.elifesciences.org/articles/60266/elife-60266-fig2-data2-v2.csv
Figure 2—source data 3

Expression of the top 200 variable genes along the pseudotime ordering of cells from LaTP2 to SynTI (Figure 2D).

https://cdn.elifesciences.org/articles/60266/elife-60266-fig2-data3-v2.csv
Figure 2—source data 4

Expression of the top 200 variable genes along the pseudotime ordering of cells from LaTP2 to S-TGC (Figure 2E).

https://cdn.elifesciences.org/articles/60266/elife-60266-fig2-data4-v2.csv

The central body consisted of 4 clusters. One of these clusters highly expressed the receptor tyrosine kinase Met (Figure 2B), which was previously described as a marker of labyrinth trophoblast progenitor (LaTP) cells with trilineage potential: SynTI, SynTII, and S-TGC (Ueno et al., 2013). The snRNA-seq data places these cells as most closely related to SynTII. Notably, the LaTP shared expression of the Wnt signaling pathway members Ror2, Lgr5, and Tcf7l1 with presumptive SynTII precursors. Interestingly, the clustering identified a second subpopulation of cells in the central body that shared relatively high expression of Tcf7l1 and the long non-coding RNA Pvt1 with LaTPs, but expressed low levels of Met, Ror2, and Lgr5 and high levels of Egfr (Figure 2B). These cells were more closely related to SynTI and S-TGC. We designated this subpopulation as LaTP2. Correlative analysis of RNA levels for Met and Egfr in individual cells confirmed that for the most part these markers were uniquely expressed in LaTP and LaTP2 cells, respectively (Figure 2—figure supplement 1B). Previously described LaTPs were also shown to express high levels of the adhesion protein EPCAM (Ueno et al., 2013). EPCAM mRNA and protein was mostly limited to the MET+ LaTP cells, although rare EPCAM/EGFR double positive cells were also present (Figure 2—figure supplement 1C–D). Positionally, MET expressing LaTP cells were more basal than the EGFR expressing LaTP2 cells, suggesting that LaTP may be giving rise to LaTP2 (Figure 2—figure supplement 1D). Consistent with this interpretation, immunofluorescence staining of E8.5 embryos identified a large number of EPCAM/MET double positive cells, but few EGFR positive cells in the developing chorion plate (Figure 2—figure supplement 1E). Together, these data show that LaTPs can be separated into two populations based both on mRNA and protein expression. The developmental relationship between these two populations will require future lineage tracing.

The remaining two clusters within the central body appeared to represent precursors of the differentiated cells of the junctional zone – glycogen cells (GCs) and spongiotrophoblasts (SpT) [Junctional zone precursors (JZPs) 1 and 2 (Figure 2A)]. JZP1 exclusively expressed high levels of Cdh4, whereas Prune2 and Ncam1 showed a continuum of expression, increasing from JZP 1 to 2 and finally specified GCs (Figure 2B and Figure 2—figure supplement 2A). Markers such as Igfbp7, Pla2g4d, Plac8, and Pcdh12 marked later stages of GC specification. In contrast to GCs, the SpT arm separated from late JZP2/early GCs, with increasing expression of Mitf, Flt1, Prl8a9, and Slco2a1 as they transitioned from precursors to differentiated cells. The relationship between LaTPs and JZPs is unclear as they shared few markers. One shared marker was Pvt1, a target and enhancer of expression of the oncogene Myc and, therefore, may reflect the shared proliferative nature of these two populations, explaining their juxtaposition in UMAP space (Tseng et al., 2014).

The arms extending from the central body showed a transition from progenitors through precursors to fully differentiated cells representing five previously described cell types of the mouse placenta: SynTI, SynTII, S-TGCs, GCs, and SpTs. SynTI was easily identified based on the expression of known markers of SynTI function including the transferrin receptor Tfrc and monocarboxylase transporter Mct1 (Walentin et al., 2016), but also expressed previously unknown markers such as the transcription factor (TF) Glis1 and the retinol receptor Stra6 (Figure 2B). SynTI precursors also expressed these markers, albeit at lower levels, while additionally expressing high levels of the signaling molecules Epha4 and Tgfa, which decreased with further differentiation. SynTII was annotated by expression of the TF Gcm1 and fusion gene Synb (Maltepe and Fisher, 2015; Simmons et al., 2008a), but also expressed many previously unknown cell-specific genes such as the glucagon receptor Gcgr and vascular endothelial growth factor Vegfa. SynTII precursors showed elevated expression of Ror2 and Lgr5 relative to the LaTP. In contrast, Tcf7l1 decreased throughout the process of differentiation from LaTP to SynTII. S-TGCs were easily identified by the expression of the canonical marker Ctsq (Simmons et al., 2007; Simmons et al., 2008a), but also uniquely expressed the leptin receptor Lepr and the Nitric Oxide Synthase one adaptor Nos1ap among other genes. Marker genes for all populations are included in supplementary information (Figure 2—source data 1).

We further probed the transition from multipotent LaTP through precursors to each of the differentiated labyrinth cell types (SynTII, SynT1, and S-TGC) using pseudotime analysis. The pseudotime analysis program Slingshot aligned cells along a 2D trajectory from the most transcriptionally similar LaTP population (LaTP-SynTII; LaTP2-SynTI; LaTP2-STGC) to each of the mature populations (Street et al., 2018; Figure 2C–E, Figure 2—source datas 24). Pseudotime analysis expanded upon the marker genes for each population by defining the timing of expression of genes through the differentiation process and identifying distinct signaling pathways in each lineage. Expression of Wnt regulated TF Tcf7l1 along with Wnt receptor Ror2 were maintained from LaTP through commitment to a SynTII precursor state. Upon commitment to the SynTII lineage, the TF Zfhx3 and growth factor receptor Igf1r are newly expressed while canonical markers Gcm1 and Synb are among the later transcript changes (Figure 2C). SynTI Precursor cells maintain Pvt1 expression but uniquely upregulate the ephrin receptor Epha4, the Egf-related ligand Tgfa, and the Egf receptor substrate Eps8. Only mature SynTI express the kinase Prkce and clathrin coat assembly protein Snap91 (Figure 2D). S-TGC Precursor cells activate the receptors Pparg and Lifr as well as the hypoxia induced TF Epas1 upon lineage commitment. Expression of the canonical marker Ctsq occurs only in mature S-TGC along with Lepr and the glucocorticoid signaling effector Tsc22d3 (Figure 2E). These data are consistent with distinct signaling pathways (e.g. SynTII – WNT, IGF; SynTI – EGF, EPH; S-TGC – LIF, PPAR) driving differentiation of LaTP into the three main cell types of the placental labyrinth. Interestingly, the S-TGC appear to differentiate from the junction of LaTP2 and JZP1 (Figure 2A), potentially suggestive that S-TGC arise from ectoplacental cone cells (Simmons et al., 2007; Simmons et al., 2008a) as well as from chorion derived LaTP (Ueno et al., 2013). It is also possible the location of the S-TGC near JZP1 reflects the shared prolactin expression between S-TGC and other junctional zone populations. Again, lineage tracing will be required to address their developmental relationships.

As described above, the remaining two arms were consistent with differentiation of the JZ populations: SpT and GC cells. Immunofluorescence of the representative markers NCAM1 and SLCO2A1 confirmed their expression in the JZ (Figure 2F). Furthermore, staining of juxtaposed sections confirmed their mutually exclusive expression (Figure 2—figure supplement 2B). Co-staining with the known GC marker PCDH12 (Rampon et al., 2005), verified the GC identity of the NCAM1+ cells (Figure 2—figure supplement 2C). Previous work has identified differential expression of prolactin genes between GC and SpT (Simmons et al., 2008a). Therefore, we analyzed the expression of these genes in our dataset. Expression analysis of prolactin genes showed both distinct and overlapping expression of this large gene family between the SpT and GC cells, and confirmed the identities of these clusters (Figure 2—figure supplement 3A and B). Of important note, parietal TGC nuclei were not identified in our snRNA-seq data, likely due to their very large size precluding their passage through the FACS pre-filter, which removes nuclei greater than 35 microns (Figure 2—figure supplement 2D). Our data also did not identify invasive trophoblast giant cells (spiral artery and canal TGCs) that invade the decidua, which was dissected away. Still, this analysis of the trophoblast snRNA-seq was able to clarify the cell types and their relationships for a predominance of the lineages of the mouse placenta identifying novel intermediate states and markers of the different cell populations.

Developmental time course and trajectory inference reveal details of lineage dynamics and commitment

Next, we asked how the trophoblast cell populations changed over developmental time by separating the conglomerate data into its individual timepoints (Figure 3A). This visualization clearly showed the rapid diminishment of the LaTP populations over time, resulting in few progenitors remaining by E14.5 (Figure 3A and B, Supplementary file 3). The precursor populations also decreased over time, presumably maturing into terminally differentiated cells, which dramatically increased during the time course (Figure 3A and B). To determine which populations were primarily responsible for the expansion of the number of cells in the placenta, we evaluated expression of the cell cycle marker Mki67 which is lost when cells exit the cell cycle (Gerdes et al., 1984). The majority of Mki67 expressing nuclei were among the progenitor and precursor cell populations (Figure 3—figure supplement 1A). However, at E9.5 and E10.5, a substantial number of nuclei at the distal tips of each arm also expressed Mki67, suggesting ongoing proliferation even after cells have transitioned to a more mature-like expression state. Very few nuclei expressed Mki67 at E14.5 consistent with the expansion in placental cell number happening prior to E14.5, with later growth occurring through cellular hypertrophy rather than cell division (Ueno et al., 2013; Paikari et al., 2017).

Figure 3 with 1 supplement see all
Developmental time course and trajectory inference reveal details of lineage dynamics and commitment in trophoblast.

(A) UMAP projection of all nuclei captured at each gestational age. (B) Quantification of the proportion of each cluster captured at each developmental time point. (C) RNA velocity vectors of the trophoblast nuclei at each time point. Arrows show the estimated magnitude and direction of each nucleus in pseudotime. (D) The magnitude of velocity vectors shown in D summarized as violin plots, with nuclei split by cluster identity and developmental time point. (E) Violin plots showing the log2 fold change of differentially expressed genes between nuclei from S-TGC at E9.5 vs. each other timepoint. (F) Immunofluorescence staining of LEPR (top) and PODXL (bottom) in the placental labyrinth at each developmental timepoint at 63x. Insets, magnification of area surrounding a S-TGC indicated by dotted line, DAPI stain removed to highlight protein expression.

Figure 3—source data 1

Average RNA velocity magnitude split by developmental stage and cluster (Figure 3D).

https://cdn.elifesciences.org/articles/60266/elife-60266-fig3-data1-v2.xlsx
Figure 3—source data 2

S-TGC Differential Expression (E9.5 v E10.5).

https://cdn.elifesciences.org/articles/60266/elife-60266-fig3-data2-v2.csv
Figure 3—source data 3

S-TGC Differential Expression (E9.5 v E12.5).

https://cdn.elifesciences.org/articles/60266/elife-60266-fig3-data3-v2.csv
Figure 3—source data 4

S-TGC Differential Expression (E9.5 v E14.5).

https://cdn.elifesciences.org/articles/60266/elife-60266-fig3-data4-v2.csv

Cellular dynamics can be tracked not only by collecting samples at different developmental time points, but also by leveraging splicing information inferred through comparisons of intron vs. exonic reads of genes within individual cells, termed RNA velocity. We performed RNA velocity using the scVelo package to produce vectors representing both the direction and speed of cellular maturation (Bergen et al., 2019). In the scVelo visualization, each nucleus is identified by an arrow representing the inferred direction and rate of cellular change projected in UMAP space. The data was visualized separately for each different developmental timepoint to discern the dynamics of cell transitions and how these dynamics change over time (Figure 3C). In addition, the velocity magnitude was summarized for each cell population at each developmental timepoint (Figure 3D, Figure 3—source data 1).

These analyses uncovered a number of interesting properties associated with placental development. Progenitor vectors (LaTP and LaTP2) showed little directionality and small magnitudes, likely reflecting a maintenance of multipotency and self-renewal. In contrast, the cells within each precursor population showed uniform directionality and large magnitudes, especially at E10.5 and E12.5, the period of greatest placental expansion. Surprisingly, the differentiated cell populations also showed vectors of great magnitude and uniform directionality at the early timepoints, suggesting ongoing transcriptional maturation, before diminishing by E14.5. To evaluate ongoing changes in the mature populations, we selected the S-TGC population and performed differential expressional analysis between each timepoint and E9.5. Consistent with the vectors, the number of differentially expressed genes at each consecutive timepoint increased from 143 to 700 genes (adj. p-value<0.05; Figure 3E, Figure 3—figure supplement 1B, Figure 3—source datas 24).

Gene ontology analysis showed that genes up at E9.5 were highly enriched for ontology groups concerning cell division, while those at E14.5 were enriched for Ras GTPase signaling and various developmental processes (Figure 3—figure supplement 1C, Figure 3—figure supplement 1—source data 1). Examples of genes that increased over developmental time include Ctsq, the Cathepsins (Ctsj, Ctsr, Ctsm, Cts3, and Cts6), and several genes involved in hormone signaling and response (Lepr, Tsc22d3, and Ghrh) (Figure 3—figure supplement 1B,D,E,F). In contrast, Podxl is expressed from the very early stages of S-TGC specification through E14.5. Podxl is a highly charged membrane protein which, in a different context, has been implicated in luminogenesis in early embryos (Shahbazi et al., 2017). While the formation of maternal blood spaces in the placental is distinct from the early embryo, the early expression of Podxl in S-TGCs might function in maintaining open maternal blood spaces within the placental labyrinth. The differential dynamics of PODXL and LEPR in S-TGCs was confirmed by immunohistochemistry (Figure 3F). Similar ongoing maturation processes were seen in the SynTII, SynTI, GC, and SpT populations (Figure 3—figure supplement 1G). Together these data support a model of early expansion by self-renewing progenitor cells and committed precursors followed by ongoing maturation of all five differentiated cell types of the placenta proper.

Defining distinct roles of the trophoblast subtypes at the gas exchange interface

The labyrinth is the site of transport between fetal and maternal blood in the mouse placenta. Fetal and maternal blood are separated by a thin membrane consisting of four cell layers starting with the fetal endothelial cells, then SynTII, followed by SynTI, and finally S-TGCs (Simmons et al., 2008aFigure 4A). A basal lamina separates the fetal endothelial cells from SynTII cells. The expression programs and functions of the three trophoblast cell layers have been difficult to dissect due to the syncytial nature of the SynTI and SynTII cells. Isolation of nuclei followed by snRNA-seq overcomes this barrier enabling the determination of the unique expression programs of each trophoblast layer. Immunohistochemistry for the marker genes PECAM1, IGF1R, STRA6, and PODXL demonstrate each cluster corresponds to endothelial, SynTII, SynTI, and S-TGC populations respectively in vivo (Figure 4A). Each layer could be distinguished by high- resolution image analysis with nucleated erythrocytes residing in the PECAM1+ fetal vessels and large polyploid S-TGC residing in the maternal blood sinuses. PODXL+ cytoplasmic extensions of the S-TGCs spread over the perimeter of the maternal blood space forming a permeable lining (Figure 4A).

Figure 4 with 1 supplement see all
Defining distinct roles of the trophoblast subtypes at the gas exchange interface.

(A) Schematic showing the relative location of the cell types at the gas and nutrient exchange interface of the labyrinth – S-TGC, SynTI, SynTII, (sometimes referred to as trophoblast layer I, II, and III, respectively) and the fetal endothelium (top). Immunofluorescence staining resolving expression in each pair of neighboring cell types – Endothelial/SynTII, SynTII/SynTI, and SynTI/S-TGC in E12.5 mouse placenta sections. (B) Heatmap of all differentially expressed genes by log fold change in SynTI, SynTII, and S-TGC clusters. (C–E) Network diagrams showing the top 5 Gene Ontology (Biological Process) enriched in marker genes (adj. p-value<0.05, logFC >0.5) in each cluster (Yu G. 2018). Gene ontology categories are nodes and associated marker genes are annotated. Adjusted p-values for enrichment are listed in the table insets.

Figure 4—source data 1

Differential expression between interface clusters (SynTI, SynTII, and S-TGC).

https://cdn.elifesciences.org/articles/60266/elife-60266-fig4-data1-v2.csv
Figure 4—source data 2

Gene Ontology results for differentially expressed genes between interface clusters (SynTI, SynTII, and S-TGC).

https://cdn.elifesciences.org/articles/60266/elife-60266-fig4-data2-v2.xlsx

SynTII, SynTI, and S-TGCs were distinguished from one another based on their transcript profiles (Figure 4B, Figure 4—source data 1). Furthermore, gene ontology analysis of differentially expressed genes between populations suggested highly distinct functions (Figure 4C–E and Figure 4—figure supplement 1, Figure 4—source data 2). SynTII differentially expressed genes were highly enriched for cell–cell and cell–matrix interactions such as those encoding cell junction and cell adherens proteins consistent with a role as a basal permeability barrier between maternal and fetal blood. SynTII also expressed collagens and laminin subunits (Figure 4E), which likely form the basal lamina separating SynTII from the fetal endothelium. Also identified in SynTII differentially expressed genes was the stem cell population maintenance GO category, which included several genes known to be important for placental function though not specific to stem cells (Bcl9, Gata2, and Tbx3).

In contrast, SynTI differentially expressed genes were enriched for intracellular transport and cell signaling including genes encoding for vesicular organization and transport, protein localization to membranes, and Ras signaling. These data are consistent with SynTI layer playing a critical role in the transport of materials between maternal and fetal blood. SynTI is also enriched for genes involved in neuron projection development, driven by expression of canonical axonal pathfinding genes (Plxna2, Plxnd1) and synapse genes (Snap91). The exact function of these genes in SynTI is unclear and would be of interest for future study.

Although more generally understood as an exocrine cell type, the S-TGC were enriched for GTPase and Ras signaling and receptors including Kras, Grb2, and Pdgfrb, suggesting these cells also transduce signals from maternal blood. S-TGC differentially expressed genes were strikingly enriched for those encoding for migration factors. Such expression could explain the long thin cellular extensions from these cells, which form the only non-syncytial trophoblast at the interface (Coan et al., 2005). S-TGCs are connected to SynTI by desmosomes, which are known to undergo constant remodeling and have additional functions in signaling and migration (Simmons et al., 2008a; Johnson et al., 2014). These S-TGC-specific programs along with the high expression of Podxl, likely underlie a central role for these cells in the function of the maternal sinuses. S-TGC are also enriched for the placental development category which includes several genes important for giant cell function including Hand1 (regulation of giant cell lineage), Hsd17b2 (steroid hormone production), and E2F7 (endoreduplication).

Given that the labyrinth represents the site of exchange of nutrients, minerals, metabolites, ions, gases, and regulatory factors, between the mother and fetus, we next wanted to better understand how these functions are distributed across the different trophoblast layers. We focused on two large functional categories: solute transporter and cholesterol/vitamin transport. All genes in these categories that were found to be differentially expressed between one of the interface populations (SynTI, SynTII, S-TGC) compared to all other trophoblast populations were included in this analysis. The results suggested a striking separation of functions between the layers (Figure 4—figure supplement 1B and C). For example, genes responsible for carboxylate transport were highly enriched in SynTII, while organic anion transporters were enriched in SynTI. Even within specific transport pathways, functions were often separated; for example, factors important in Zinc transport were distributed across, yet distinctly expressed between the populations. Cholesterol transporters are enriched in SynTII (Scarb1, Vldlr) and S-TGC (Hdlbp, Ldlr), while Folate receptors (Folr1, Folr2) are uniquely expressed in SynTI. Together, these data show a highly evolved separation of functions between the three trophoblast layers that separate maternal from fetal blood.

Predicting cell signaling within the placental labyrinth

Single-cell RNA-seq has been used to predict signaling between different cell types of a tissue or organ (Efremova et al., 2020; Vento-Tormo et al., 2018). Here, we asked whether we could do the same with snRNA-seq to provide insight into the developmental signals driving mouse placental development. We used the package CellPhoneDB (Efremova et al., 2020) which applies a data base of annotated receptor-ligand pairs to the single-cell expression data evaluating all possible combinations of cell types (i.e. identified clusters). We focused our analysis on the populations of the developing labyrinth (LaTP, LaTP2, SynTI Precursor, SynTI, SynTII, S-TGC Precursor, S-TGC, and fetal endothelium). Given that CellPhoneDB receptor-ligand database was developed for human cells, we only included orthologous genes from our mouse snRNA-seq data. The results are summarized as a heatmap of the predicted strength (average log 2 expression of ligand-receptor pair) in Figure 5—figure supplement 1 with receptor–ligand interactions shown on the rows and each cluster pair shown as columns (full data in Figure 5—figure supplement 1—source data 1). A total of 139 unique ligand–receptor interactions reaching significance (adj. p-value<0.05) were uncovered.

CellPhoneDB predicted an interaction between Rspo3 in the endothelium with Lgr5 in the LaTP/SynTII lineage (Figure 5A). Rspo3 is required in vivo for the maintenance of Gcm1 expression (Kazanskaya et al., 2008; Aoki et al., 2007) and canonical Wnt signaling is essential for differentiation of SynTII (Zhu et al., 2017; Matsuura et al., 2011). Predicted signaling in the opposite direction included Vegfa secreted by SynTII to the receptors Kdr, Flt1, Nrp1, Nrp2 located only on endothelial cells. Immunostaining confirmed SynTII and SynTI as the sources of VEGFA in the labyrinth (Figure 5B, Figure 5—figure supplement 5–S2A). These data suggest a feedback loop where Rspo3-Lgr5 supports SynTII differentiation and function, while Vegfa-Kdr/Flt1 drives vascular maintenance and remodeling, thereby coordinating lineage differentiation with development of tissue structure.

Figure 5 with 2 supplements see all
Predicting Cell Signaling within the placental labyrinth.

(A) Heatmap showing all Wnt signaling interactions (top) and Vegf signaling (bottom) between the endothelium and labyrinth trophoblast populations. Left: schematics summarizing signaling interactions. (B) Immunofluorescence staining for VEGFA and IGF1R demonstrating colocalization in SynTII in E12.5 mouse placenta sections (63x). Fetal Vessels are outlined in yellow dashed line. (C) Heatmap showing all Egfr signaling interactions between labyrinth populations. Schematic showing secretion of Egf ligands Hbegf, Tgfa, and Nrg1 from SynTI to Egfr expressed on SynTII, LaTP, and S-TGC. (D) Immunofluorescence staining for HBEGF and EGFR demonstrating colocalization of the ligand and receptor SynTII, but only HBEGF expression in SynTI (E12.5 mouse placenta sections - 63x) (E) Heatmap showing signaling interactions of Bmp8a (top) and Bmp7 (bottom), secreted by the LaTP and SynTII or LaTP only, respectively. (F) Heatmap showing predicted signaling interactions between Bmp6 secreted by the endothelium to receptors on LaTP, SynTI, SynTII and S-TGC. (G) Transcript expression for the Smad TFs which are differentially regulated (Smad1/3/5/6).

Analysis of marker genes suggested Egf signaling as a regulator of LaTP differentiation to SynTI (Figure 2D). CellPhoneDB analysis was consistent with that interpretation as it predicted an interaction between the EGFR ligands Tgfa, Hbegf, and Nrg1 produced by SynTI with Egfr on both LaTP2 and SynTI precursor populations. However, it also predicted an interaction with Egfr on LaTP, SynTII, and S-TGCs (Figure 5C). We confirmed protein expression HBEGF alone in SynTI, and colocalization of HBEGF and EGFR in SynTII (Figure 5D; Figure 5—figure supplement 2B and C). Therefore, SynTI appears to be a hub of EGF signaling that impacts the majority of the cell populations in the labyrinth. Loss of Egfr in vivo results in decreased placental size, disorganized labyrinth, and a reduction in the SpT layer (Du et al., 2004; Miettinen et al., 1995; Strunk et al., 2004). Depletion of Egf ligands yields a variety of phenotypes (Liu et al., 2019; Luetteke et al., 1993). Taken together, these data suggest a complex regulation of Egfr signaling being coordinated by SynTI that is critical for proper development of the labyrinth.

Numerous ligand–receptor interactions between Bmp8a and several receptors, including Acvr1, Bmpr1, and Bmpr2 were predicted in the labyrinth. The source of Bmp8a is LaTP and SynTII and the receptors are located on all labyrinth populations, including LaTP and SynTII (Figure 5E). Additional Bmp 60A subfamily members, Bmp6 and Bmp7, were highly secreted. Bmp7 is secreted by LaTP and Bmp6 by the endothelium (Figure 5E and F). In addition to BMP ligand and receptors, the pathway’s downstream effectors, the Smad TFs, also showed striking differential expression between the cell types of the labyrinth. Smad2/4/7 were relatively uniformly expressed in trophoblasts, while Smad1/3/5/6 displayed varied levels of expression across trophoblast populations (Figure 5G and Figure 5—figure supplement 2D). Previous work has shown that while Bmp8a is largely redundant with Bmp8b, loss of Smad1, and combined knockout of Bmp5/7 exhibit severe placental defects (Zhao and Hogan, 1996; Tremblay et al., 2001; Solloway and Robertson, 1999). Cell-type-specific perturbation of elements of the Bmp pathway has not been performed; however, ligand–receptor modeling with CellPhoneDB combined with snRNA-seq expression data provides insight into how the pathway likely functions in vivo.

Modeling transcription factor regulon activity identifies new candidate regulators of SynTII

Transcription factors (TFs) are the most direct regulators of cell fate and can be used to reprogram cells to new cell fates (Kubaczka et al., 2015). Our snRNA-seq data contained many well-studied TFs, but expression alone is not necessarily a good indicator of function. TFs typically act cooperatively with other TFs and numerous factors influence their activity. Therefore, we sought to identify active TFs based on both expression of the TF and its downstream targets within the same cells. To do so, we used the SCENIC package (Aibar et al., 2017) which uses a database of TFs and their annotated motifs to identify enrichment of a TF motif within or near the promoter regions of co-expressed genes within each cell to define ‘regulon activity’. As such SCENIC infers activity of a transcriptional network rather than considering expression of the TF alone.

Application of SCENIC to the trophoblast dataset recovered 200 transcription factor regulons with activity in at least one cluster (Figure 6—source data 1). To ensure that the regulon activity data maintained the variation of the transcript expression data, we projected the nuclei in UMAP space using regulon activity transcript expression as the underlying variable. The resulting plot was color coded based on a cluster identity as defined by Seurat analysis of transcript expression (Figure 6A). There was overall strong agreement in the two approaches to separate nuclei by cell type both in UMAP space and hierarchical clustering of Pearson correlations between populations (Figure 2A vs. Figure 6A, Figure 6—figure supplement 1A). These data show that TF regulon activity provides an alternative means to identify the different cell types of the placenta.

Figure 6 with 1 supplement see all
Modeling transcription factor regulon activity identifies new candidate regulators of SynTII.

(A) UMAP projection derived from regulon activity predicted by SCENIC. The clusters are colored according to Seurat clustering of transcript data. The cluster identities of each of the five arms of the UMAP are annotated. (B) Heatmap of the scaled regulon activity for select regulons expressed in each trophoblast population. The number of genes predicted to be regulated by each transcription factor is included in parenthesis. (C) Comparison of transcript expression of the TF (left) and the regulon activity for that TF (right) at the single-cell resolution projected in UMAP space for Gcm1, Pax8, Ets2, and Esrrg. The number of genes predicted to be regulated by each transcription factor is denoted in parentheses. (D) Quantification of the number active regulons with a lethal phenotype in mouse according to the Jackson Laboratory MGI (Embryonic/Prenatal - 78, Perinatal/Neonatal - 24, Postnatal - 11, Pre-weaning - 9, lethality at weaning - 2, no developmental lethality - 76) and the number with known placental phenotypes in vivo (n = 17). (E) Heatmap showing expression of a network of regulons enriched for expression in SynTII (left). Gene regulatory network of high confidence predicted interactions from GRNboost2 (right). Unidirectional regulation denoted by blue arrows and bidirectional regulation by yellow. Predicted targets of Gata1 and Tcf7l2 expressed in SynTII and important in SynTII differentiation or function are annotated. (F) Immunofluorescence staining for GATA1 and IGF1R showing GATA1 expressing nuclei in IGF1R SynTII cells (E12.5 mouse placenta sections - 63x). GATA1+ SynTII nuclei – white arrowheads, GATA1- putative SynTI nucleus – yellow arrowhead.

Figure 6—source data 1

SCENIC scaled regulon activity in trophoblast nuclei.

https://cdn.elifesciences.org/articles/60266/elife-60266-fig6-data1-v2.csv
Figure 6—source data 2

SCENIC regulon predicted binding targets for all active regulons.

https://cdn.elifesciences.org/articles/60266/elife-60266-fig6-data2-v2.txt

The identified regulons with differential activity across the populations included both known and unknown TF regulators of placental development. Known factors include Gcm1 in SynTII specification, Grhl2 in S-TGC, and Ets2 in JZ development (Figure 6BAnson-Cartwright et al., 2000; Walentin et al., 2015; Yamamoto et al., 1998). Examples of the many unknowns include Tbx15 in LaTP, Gata1 in SynTII, Pax2 in SynTI, and Meis1 in S-TGC. The power of the regulon approach is further supported by comparing the distribution of TF expression alone versus that of their regulon (Figure 6C). While expression of Gcm1 and Pax8 correlated the distribution of their regulon activity, Ets2 and Esrrg did not. Ets2 was broadly expressed, yet its regulon activity was highly enriched among differentiating S-TGCs and GCs. Esrrg, on the other hand, was lowly expressed throughout, yet showed strong regulon active among SpT cells. Esrrg is not well studied in the mouse placenta, but has known roles in steroid hormone production and metabolism in human trophoblast, functions also important in the mouse placenta (Poidatz et al., 2012; Luo et al., 2014).

Given that most of the TFs underlying cluster-specific regulons have not been studied in the placenta, we next researched the Mouse Genome Informatic database for any phenotypes associated with the knockout of TFs driving the uncovered regulons. Only 17 of the 200 TFs had annotated placental phenotypes, while 124 are lethal during development (defined as until weaning) (Figure 6D). Many of these developmental lethal phenotypes likely are associated with placental defects given that 68% of 103 intrauterine lethal and subviable mouse knockout lines also have placental defects (Perez-Garcia et al., 2018). Unfortunately, few of knockouts evaluated in the Perez-Garcia et al paper were for transcription factors, making comparisons to our regulon data impossible. However, we did see robust and cell-type-specific expression of several factors uncovered in their paper as having placental phenotypes, including Fryl in GCs, Atp11a in the SynTI lineage, and Adcy9 in SynTII (Figure 6—figure supplement 1B and C). Together, these analyses suggest likely important roles for the uncovered regulons in placental development.

Next, we asked if we could uncover networks among the TFs themselves focusing on the SynTII lineage. In particular, the regulon data was evaluated for evidence of coexistence of paired regulon activity with associated high confidence annotations (defined by either direct annotation or inferred by orthology) of binding sites for one or both of the TFs in the other TF’s promoter region (Figure 6E, Figure 6—source data 2). Gcm1, the master regulator of the SynTI lineage, and Zhfx3 have reciprocal high confidence motif in their promoter regions, suggestive of a positive feedback loop. Zfhx3 and Tcf7l2 also show mutual regulation of one another, and Tcf7l2 also has a high confidence binding site in the promoter of Gcm1. Thus, these 3 TFs establish what appears to be an autoregulatory loop. Zfhx3 (Atbf1) does not have known functions in the placenta. However, two independent genetic disruptions of Zfhx3 result in embryonic to pre-weaning lethality with surviving pups exhibiting severe growth defects (Sun et al., 2012; Parsons et al., 2015). While Tcf7l2 knockout mice do not have a placental phenotype, we identify high confidence motifs in the promoter regions of many genes important for SynTII differentiation and function including Lgr5, Vegfa, Slc43a2, and Gcm1 (Perez-Garcia et al., 2018; Kazanskaya et al., 2008; Guetg et al., 2015). It is possible that redundancy with Tcf7l1 underlies the lack of Tcf7l2 knockout placental phenotype (Nguyen et al., 2009).

Zfhx3 also appears to be in a positive regulatory loop with the two TFs Zmiz1 and Mxi1. Zmiz1 and Mxi1 in turn both have high confidence binding sites in the promoter of Gata1. While Gata1 has not been studied in the context of the mouse placenta, transcript expression, regulon activity, and high confidence motifs suggest it regulates multiple genes with important functions in SynTII including Vegfa, Slc40a1 (ferroportin), and the basal epithelial transcription factor Trp63. Immunofluorescence showed the colocalization of GATA1 with IGFR1, thereby confirming its expression in SynTII (Figure 6F). In particular, strong GATA1 protein expression was visible in the SynTII nuclei abutting the basal membrane (Figure 6F and Figure 6—figure supplement 1D – white arrow heads), but not in SynTI nuclei closest to maternal blood spaces (Figure 6F – yellow arrowhead). Strong GATA1 staining was also obvious in the nuclei of fetal erythrocytes with vascular lumen. Therefore, GATA1 appears to be expressed exclusively in SynTII population of trophoblasts and likely has important regulatory functions specific to SynTII.

Discussion

This study profiles the development of the mouse placenta by transcriptome capture of nuclei, rather than whole cells, resulting in a high-resolution map of labyrinth differentiation and allowing for the first independent analyses of the two SynT populations. In contrast to previous studies where SynT comprise fewer than 10% of cells identified, almost 30% of trophoblast nuclei captured in this study originate from SynT cells. Profiling four developmental stages from E9.5 to E14.5 allowed for the capture of stem/progenitor and terminally differentiated populations including previously unknown intermediate states and new putative regulators of cell identity. Modeling receptor–ligand interactions predicted how SynTII and endothelial cells coordinate spatial proximity and development through growth factor secretion. In addition, network inference predicted gene regulatory networks defined by TFs and feedback loops among the TFs, including novel candidate regulators of placental development. All together, these data represent a high-resolution map of placental labyrinth development providing insights into the cell autonomous and non-autonomous molecular events underlying the formation and function of this remarkable organ.

Understanding the development of any tissue begins with the progenitor populations. Previous work uncovered a population of EPCAM and MET expressing progenitors that give rise to the three trophoblast populations of the labyrinth (SynTI, SynTII, and S-TGC) (Ueno et al., 2013). Clustering analysis of the snRNA-seq data separated these cells into two closely related clusters, LaTP and LaTP2. Both expressed Epcam, but only LaTP expressed Met. The relationship between these populations is unclear. Both LaTP populations had low RNA velocities and there was no coordinated directionality to indicate differentiation from one population to the other. However, the relative positions of the two populations in the developing labyrinth and the earlier presence of EPCAM/MET double positive cells (i.e. at E8.5) suggest that LaTP2 arise from LaTP. Alternatively, the two populations may arise from progenitors that separate earlier in development resulting in distinct lineage biases. LaTP is more similar to SynTII, while LaTP2 is more similar to SynTI and S-TGC. Understanding the specification and heterogeneity of labyrinth progenitors will be an important undertaking given that early defects are likely to have large effects in proportion of terminally differentiated trophoblast resulting in embryonic defects and ultimately reduced viability.

Tracking differentiation from the LaTP populations identified previously unknown precursor populations along differentiation to SynTI and S-TGC. Profiling differentiation to SynTII confirmed previous work demonstrating the role of Wnt signaling in specification and efficient differentiation of SynTII, both in vivo and in vitro (Zhu et al., 2017; Matsuura et al., 2011). However, such detailed information and derivation protocols did not exist for SynTI or S-TGC. SynTI precursor cells strongly upregulated Egf signaling members Tgfa and Eps8. Additionally, expression of Egf ligands by mature SynTI and Egfr by LaTP 2, SynTII, and S-TGC underscore the centrality of Egf signaling to labyrinth development and function. While Egfr loss or mutation in vivo causes labyrinth disorganization, the deciphering of mechanism has been confounded by defects in junctional zone cells and strain-specific variation (Du et al., 2004; Strunk et al., 2004; Dackor et al., 2009; Lee and Threadgill, 2009). In human trophoblast, EGF has been shown to stimulate differentiation to SynT in vitro and decreased EGFR phosphorylation is associated with intrauterine growth restriction (Morrish et al., 1987; Fondacci et al., 1994). Given the redundancy of Egf ligands, a careful temporal and cell-type-specific dissection of the loss of Egfr will be necessary to separate the multiple functions of Egf signaling in placental development and health.

Pseudotemporal ordering along the S-TGC trajectory implicated Lif-Stat3 signaling evidenced by increasing expression of Lifr and supported by high activity of the predicted Stat3 regulon by SCENIC. Loss of Lifr results in disorganization of the labyrinth and dramatically enlarged non-functional maternal blood sinuses lined by multinuclear aggregates, and not S-TGC (Ware et al., 1995). While Lif-Stat3 is also active in the junctional zone, this striking labyrinth phenotype is likely explained by a necessity for Lif signaling in the differentiation, patterning, and/or function of S-TGC. Deciphering how Lifr is necessary for S-TGC and the regulation of maternal blood spaces will require genetic manipulation specific to S-TGC. This will be important to understand establishment and maintenance of functional maternal blood spaces which are at the core of common placental diseases like preeclampsia, fetal growth restriction, and diabetic complications (Ware et al., 1995; Roberts and Escudero, 2012).

This study represents the first independent transcript profiling of each SynT layer in vivo. This was achieved by the isolation of nuclei prior to transcriptome capture. These data demonstrate the utility of snRNA-seq in profiling syncytial placental cells without damaging and lengthy enzymatic digestion or single-cell dissociation. Mouse is the most common model for human placentation, even though the species differ in organ morphology, especially with respect to the gas exchange interface. Human placentas have villi that float in a pool of maternal blood. The villi are covered by a single syncytial layer and have no clear S-TGC equivalent. Comparing the transcript profiles of the interface populations in mouse and human will help to understand how the functions of SynTI, SynTII, and S-TGC are distributed across the fewer trophoblast populations of the human placenta. Such information is also likely to provide important insight into how the different structures arose during evolution and which human developmental processes can be effectively modeled in mouse.

In our study of the mouse labyrinth, specific marker genes identified in each layer provide insight into the functions of SynTI versus SynTII. Gene Ontology analysis of SynTI indicates enrichment for membrane associated proteins, including numerous transporters, receptors, and GTPase signaling factors. Many of these proteins, like STRA6, are localized to the apical surface where they transduce signals and uptake nutrients from maternal blood. However, SynTII also expressed many classes of solute transporters and signaling receptors, reflecting the export of molecules to the fetal endothelium. SynT layer-specific expression indicates which transporters are likely involved in uptake from maternal blood (SynTI enriched), export to fetal blood (SynTII enriched), or both (SynTI and SynTII enriched). Transcript data represent the first step in understanding the chain of nutrient transport and inform the critical cell type for targeted therapies for nutrient deficiencies.

Surprisingly, SynTII display unique enrichment for adhesion and junctional proteins, likely necessary to support frequent gap junctions on the border with SynTI and robust attachment to the basal lamina, respectively. Notably, several laminin, integrin, and collagen subunits are highly expressed. Electron microscopy has demonstrated only one basal lamina exists, which is between SynTII and fetal endothelial cells (Coan et al., 2005). Endothelial cells express a complementary set of extracellular matrix protein subunits (Col4a1, Col5a2, Col18a1), suggesting deposition of this membrane is a coordinated process between these two cell types. Coordinated development between SynTII and the endothelium is also found in a positive feedback loop of necessary growth factors (Wnt and Vegf). These data support an additional role for SynTII as a vascular support cell and likely cell type of origin for many placental vascular abnormalities.

Recent work identified that a surprising proportion of developmental lethal mouse mutants had unidentified placental phenotypes (Perez-Garcia et al., 2018). At least 200 TF defined regulatory networks are active in trophoblast during development. Over 62% of these TFs exhibit developmental lethal phenotypes upon homozygous loss of function but only 8.5% have published placental phenotypes (according to MGI). Candidate regulators were identified in all cell types, including Zfp110 in LaTP, Pax2 in SynTI, Gata1 in SynTII, Meis1 in S-TGC, and Etv6 in GCs. For some TFs with strong phenotypes in other organ systems, a methodical analysis of placental morphology may not have been performed or published. For example, Zfhx3 is predicted to form an autoregulatory loop with Gcm1 and regulate several genes important to SynTII function. While loss of Zfhx3 results in developmental lethality and severe growth defects, its role in placental development has not been examined (Sun et al., 2012; Parsons et al., 2015). Several TFs may have not been identified due to overlapping or compensatory roles with other proteins. One such example is Tcf7l1 and Tcf7l2. While highly active in the SynTII, Tcf7l2 exhibits no placental abnormalities upon knockout. Tcf7l1, which is also highly active in LaTP, has demonstrated redundancy with Tcf7l2 in the skin and it is likely each factor compensates for loss of the other (Nguyen et al., 2009). As the knockout of both factors is lethal prior to the genesis of the labyrinth, conditional deletion in only extraembryonic tissues would be required to investigate their function in the placenta. While conditional or temporal control over deletion may be necessary for other factors inducing early lethality (Klf5) or confounding defects in other organ systems (Gata1), genetic deletions for almost all identified TFs exist and represent a large resource for immediate study of their effects in placental development.

In summary, this study represents a compendium of mouse placental labyrinth development from E9.5 through E14.5. These data provide a lens through which to better understand congenital defects, environmental and nutrient deficiencies, and species-specific differences in development and function.

Materials and methods

Key resources table
Reagent type
(species) or resource
DesignationSource or referenceIdentifiersAdditional information
Strain (Mus musculus)C57BL/6J Micehttps://www.jax.org/strain/0006640006646–12 weeks old
Commercial assay or kitNuclei isolation Kit: Nuclei EZ PrepSigma-AldrichNUC101-1KT
AntibodyE-cadherin Monoclonal antibody (ECCD-2)Thermofisher Scientific13–1900IF (1:250)
AntibodyHuman/mouse NCAM-1/CD56 Polyclonal antibodyR and D SystemsAF2408-SPIF (1:25)
AntibodyAnti-EpCAM Polyclonal antibodyAbcamab71916IF (1:1000)
AntibodyAnti-Stra6 Polyclonal antibodySigma-AldrichABN1662IF (1:100)
AntibodyAnti-Igf1r Polyclonal antibodyR and D SystemsAF305IF (1:50 w/ antigen retrieval)
AntibodyAnti-Slco2a1 Polyclonal antibodyAtlas AntibodiesHPA013742IF (1:25 w/ antigen retrieval)
AntibodyAnti-Lepr Polyclonal antibodyR and D SystemsAF497IF (1:200)
AntibodyAnti-Pcdh12 Polyclonal antibodyAbcamab113720IF (1:25 w/ antigen retrieval)
AntibodyAnti-Podxl Polyclonal antibodyR and D SystemsAF1556IF (1:25)
AntibodyAnti-Pecam1 Polyclonal antibodyAbcamab23864IF (1:50)
AntibodyAnti-Vegfa Polyclonal antibodyAbcamab51745IF (1:50)
AntibodyAnti-Met Polyclonal antibodyR and D SystemsAF276IF (1:100)
AntibodyAnti-Egfr Polyclonal antibodyR and D SystemsAF1280IF (1:100)
AntibodyAnti-Gata1 Monoclonal antibodyCell Signaling3535IF (1:100)
AntibodyAnti-Hbegf Polyclonal antibodyR and D SystemsAF8239IF (1:20)
Software, algorithmRhttps://www.r-project.org/
Software, algorithmImageJImageJ (http://imagej.nih.gov/ij/)
Software, algorithmSeurat (3.1.3)https://satijalab.org/seurat/
Software, algorithmcellranger (3.0.2)https://support.10xgenomics.com/single-cell-gene-expression/software/pipelines/latest/feature-bc
Software, algorithmCellPhoneDBhttps://www.cellphonedb.org/
Software, algorithmSCENIChttps://github.com/aertslab/SCENIC
Software, algorithmClusterProfilerhttps://guangchuangyu.github.io/software/clusterProfiler/
Software, algorithmSlingshothttps://github.com/kstreet13/slingshot
Software, algorithmscVelohttps://github.com/theislab/scvelo
Software, algorithmFlowJohttps://www.flowjo.com

Animal husbandry

Request a detailed protocol

All mice were maintained according to the UCSF guidelines and the Laboratory Animal Resource Center standards. All experiments were reviewed and approved by the UCSF Institutional Animal Care and Use Committee. To generate all placentas C57Bl/6J males between 3 and 6 months of age were bred to C57Bl/6J females between 6 and 12 weeks of age (Jackson Laboratory, Bar Harbor, ME). Developmental staging was assessed by days post coitum (dpc) and confirmed by embryo morphology.

Placental dissection

Request a detailed protocol

Each placenta sample was dissected in 1x PBS at 4°C. The yolk sac and allantois were removed from the basal chorionic plate using forceps. The outermost layers of the decidua were also removed, taking care to not disrupt the cells of the labyrinth or junctional zone of the fetal placenta. Sample passing these criteria were used in nuclei isolation below.

Nuclei isolation

Request a detailed protocol

Whole dissected placental samples were placed in a 40 mm plastic dish with 1 mL of Nuclei EZ lysis buffer (Sigma-Aldrich, St.Louis, MO - NUC-101) and reduced to small pieces of tissues (<1 mm) using a razor blade. The solution containing the tissue was then transferred to a dounce homogenizer and 1 mL of Nuclei EZ lysis buffer added. The nuclei were lysed into solution by 10 strokes of the loose pestle (A) and 10 strokes of the tight pestle (B). Using a wide bore pipette the solution was transferred to a 15 mL falcon tube, 2 mL of Nuclei EZ lysis buffer added, and incubated on wet ice for 5 min with gentle mixing with a pipette. The solution was then passed through a 35 µM filter and centrifuged at 500 g for 5 min at 4°C. The resulting supernatant was removed, the pellet gently resuspended in 1.5 mL of Nuclei EZ lysis buffer and incubated on ice for 5 min, then centrifuged at 500 g for 5 min at 4°C. The supernatant was removed and 1 mL of Nuclei Wash and Resuspension Buffer (1x PBS, 1.0% BSA, 0.2 U/μL RNase Inhibitor) added without mixing or resuspension and incubated on ice for 5 min. After the incubation an additional 1 mL of Nuclei Wash and Resuspension Buffer was added and the pellet gently resuspended. The solution was then centrifuged at 500 g for 5 min at 4°C (in a swinging bucket rotor to prevent nuclei adhering to the sides of the tube), the supernatant discarded, and the nuclei resuspended in 1 mL of Nuclei Wash and Resuspension Buffer. This final nuclei solution was passed through a 35 µM filter and nuclei visually inspected on a hemocytometer to assess, morphology, damage, and aggregation. Finally, 1 μL of DAPI stock solution (10 μg/mL) was added to the nuclei solution in preparation for FACS.

FAC-sorting of nuclei into 10x genomic reagents for transcriptome capture

Request a detailed protocol

All cell sorting was performed on a FACS Aria II (BD Biosciences, San Jose, CA) using a 70 µM nozzle. Nuclei sorting was performed according to steps outlined here - https://www.protocols.io/view/frankenstein-protocol-for-nuclei-isolation-from-f-3fkgjkw/abstract.

Nuclei were sorted in the nuclei wash and Resuspension Buffer (described above) plus DAPI. 17,500 nuclei yielding a strong DAPI signal and DNA content profile were sorted into one well of a 96-well plate containing 10x Genomics V3 GEM mastermix without RT Enzyme C (20.0 μL RT Reagent Mix, 3.1 μL Template Switch Oligo, 2.0 μL Reducing Agent B, 27.1 μL nuclease free H2O). The volume of 17,500 sorted nuclei was previously established to be 19.5 μL. Immediately post sorting 8.3 μL of RT Enzyme C (10x Genomics, Pleasanton, CA) was added to the sample, and transcriptome collection on the 10x Chromium V3 platform performed.

Single-cell RNA sequencing and analysis

Request a detailed protocol

To capture the transcriptome of individual nuclei we used the Chromium Single Cell 3' Reagent V3 Kit from 10X Genomics. For all samples 17,500 nuclei were loaded into one well of a Chip B kit for GEM generation. Library preparation including, reverse transcription, barcoding, cDNA amplification, and purification was performed according to Chromium 10x V3 protocols. Each sample was sequenced on a NovaSeq 6000 S4 to a depth of approximately 20,000–30,000 reads per nucleus. The gene expression matrices for each dataset was generated using the CellRanger software (v3.0.2 - 10x Genomics). A custom premRNA GTF was generated to create an intron–exon reference according to 10x Genomics recommendations and all reads aligned to mm10 using STAR. (https://support.10xgenomics.com/single-cell-gene-expression/software/pipelines/latest/advanced/references). The counts matrix was thresholded and analyzed in the package Seurat (v3.1.3). Nuclei with fewer than 500 or greater than 4000 unique genes, as well as all nuclei with greater than 0.25 percent mitochondrial counts, were excluded from all subsequent analyses. For each sample, counts were scaled and normalized using ScaleData and NormalizeData, respectively, with default settings and FindVariableFeatures used to identify the 2000 most variable genes as input for all future analyses. PCA was performed using RunPCA and significant PCs assessed using ElbowPlot and DimHeatmap. For individual analysis of each sample the number of PCs and the resolution parameter of FindClusters can be found in Supplementary file 1. Dimensionality reduction and visualization using UMAP was performed by RunUMAP. Differentially expressed genes were identified using FindAllMarkers. Integration of each timepoint into one dataset was performed using FindIntegrationAnchors and IntegrateData, both using 20 dimensions (after filtering each dataset for number of genes, mitochondrial counts, and normalizing as described above). Data scaling, PCA, selection of PCs, clustering and visualization proceeded as described above using 20 PCs and a resolution of 0.6. Each cluster was analyzed and clusters containing abnormal UMI and co-expression of multiple cell-type-specific genes were determined to be enriched for doublets and removed from downstream analysis. The number of nuclei excluded represented an estimated doublet rate of ~3.9%. To generate the trophoblast only dataset, data from each timepoint was subset for only trophoblast clusters using the function SubsetData based upon annotations from marker genes identified by FindAllMarkers. Integration of the trophoblast only dataset was performed using the same integration methods above (using 20 dimensions for FindIntegrationAnchors and IntegrateData, and 20 PCs and a resolution of 0.6 for FindClusters and RunUMAP). Differentially expressed genes for each integrated dataset were identified using FindAllMarkers.

Slingshot

Request a detailed protocol

We used the Slingshot package to order cells for each SynTI, SynTII, and S-TGC lineage in pseudotime. Each lineage was subset and the 1000 most variable genes identified and used as input into the generalized linear model (GAM package). Expression is plotted for the top 200 most variable genes along each lineage and included in Supplementary data.

scVelo

View detailed protocol

RNA velocity analysis was applied to the entire conglomerate dataset using Velocyto to generate spliced and unspliced reads for all cells. This dataset was then subset for the trophoblast dataset introduced in Figure 2. The scVelo stochastic model was run with default settings and subset by each timepoint. The magnitudes were calculated from the UMAP cell embeddings.

Differential expression between timepoints

Request a detailed protocol

To account for the different number of nuclei recovered within a cluster at each timepoint, we downsampled to the least number of cells collected and performed differential expression using FindMarkers function and repeated this for 100 random permutations of nuclei. Only genes found to be significant (adj. p-value<0.05) in all 100 permutations were retained in the final list of differentially expressed genes. To retain the power in the larger clusters we did not downsample all clusters to the same size. Therefore the number of differentially expressed genes between timepoints cannot be compared between clusters.

Gene ontology analysis

Request a detailed protocol

Gene Onotology analysis was performed with clusterProfiler enrichGO function. The simplify function within this package was used to consolidate hierarchically related terms using a cutoff of 0.5. Terms were considered significantly enriched with an adjusted P-value of less than 0.05.

CellPhoneDB

Request a detailed protocol

Cluster annotations and counts data were used as input for CellPhoneDB with default settings (minimum of 10% of nuclei in a cluster expressing a gene, a p-value cutoff of 0.05, and 10 permutations). The databases of receptor–ligand interactions were generated for human proteins, not for mouse. As such, we performed this analysis using orthogonal genes between species (a total of 724 genes in the CellPhoneDB receptor–ligand databases). However, this does not account for differences in receptor–ligand interactions that may be different between mouse and human. Raw data were analyzed and heatmaps generated in R using modified scripts from CellPhoneDB. For heatmaps, the mean values of each significant receptor–ligand pair are shown, and we do not consider non-significant interactions, these have been given the mean value of zero.

Immunofluorescence

Request a detailed protocol

Placentas for cryosectioning were fixed in 4% PFA at 4°C for 8 hr, washed in 1x PBS, then submerged in 30% sucrose overnight at 4°C prior to embedding in OCT medium. Placentas were sectioned at 10 µM for all conditions. In brief, slides were washed in 1x PBST (1x PBS, 0.05% Tween-20), blocked for 1 hr (1x PBS 5% donkey serum + 0.3% TritonX), incubated in primary antibody diluted for 3 hr at room temperature (or overnight at 4°C), washed in 1x PBST, incubated in secondary antibody (Alexafluor 488, 594, and 680) for 1 hr at room temperature, incubated in DAPI for 10 min at room temp, washed in 1x PBST, and mounted and sealed for imaging. Any antigen retrieval was performed prior to the blocking step by heating the slides in a 1x citrate buffer with 0.05% Tween-20 at 95C for 30 min. All antibodies and the dilutions are listed in the key resources table. All immunofluorescence staining was performed in n = 3 biological replicates (three distinct placentas) and representative image is shown.

Data availability

Sequencing data have been deposited in GEO under accession code GSE152248. Processed data as R objects are available at figshare (https://figshare.com/projects/Single_nuclei_RNA-seq_of_mouse_placental_labyrinth_development/92354).

The following data sets were generated
    1. Marsh BP
    2. Blelloch RH
    (2020) NCBI Gene Expression Omnibus
    ID GSE152248. Single nuclei RNA-seq of mouse placental labyrinth development.

References

    1. Gerdes J
    2. Lemke H
    3. Baisch H
    4. Wacker HH
    5. Schwab U
    6. Stein H
    (1984)
    Cell cycle analysis of a cell proliferation-associated human nuclear antigen defined by the monoclonal antibody Ki-67
    Journal of Immunology 133:1710–1715.
    1. Roberts JM
    2. Escudero C
    (2012) The placenta in preeclampsia
    Pregnancy Hypertension: An International Journal of Women's Cardiovascular Health 2:72–83.
    https://doi.org/10.1016/j.preghy.2012.01.001
    1. Solloway MJ
    2. Robertson EJ
    (1999)
    Early embryonic lethality in Bmp5;Bmp7 double mutant mice suggests functional redundancy within the 60A subgroup
    Development 126:1753–1768.
    1. Tremblay KD
    2. Dunn NR
    3. Robertson EJ
    (2001)
    Mouse embryos lacking Smad1 signals display defects in extra-embryonic tissues and germ cell formation
    Development 128:3609–3621.
    1. Ware CB
    2. Horowitz MC
    3. Renshaw BR
    4. Hunt JS
    5. Liggitt D
    6. Koblar SA
    7. Gliniak BC
    8. McKenna HJ
    9. Papayannopoulou T
    10. Thoma B
    (1995)
    Targeted disruption of the low-affinity leukemia inhibitory factor receptor gene causes placental, skeletal, neural and metabolic defects and results in perinatal death
    Development 121:1283–1299.

Decision letter

  1. Didier YR Stainier
    Senior Editor; Max Planck Institute for Heart and Lung Research, Germany
  2. Steve Charnock-Jones
    Reviewing Editor; University of Cambridge, United Kingdom
  3. Steve Charnock-Jones
    Reviewer; University of Cambridge, United Kingdom

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

This paper uses single nuclei sequencing of the developing mouse placenta. It gives particular insight into the differences in the syncytial layers of the labyrinth. It adds considerably to our knowledge of the sinusoidal trophoblast giant cells. The molecular detail provided is a substantial advance.

Decision letter after peer review:

Thank you for submitting your article "Single nuclei RNA-seq of the mouse placenta uncovers mechanisms of differentiation" for consideration by eLife. Your article has been reviewed by three peer reviewers, including Steve Charnock-Jones as the Reviewing Editor and Reviewer #1, and the evaluation has been overseen by Didier Stainier as the Senior Editor.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

As the editors have judged that your manuscript is of interest, but as described below that additional experiments are required before it is published, we would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (https://elifesciences.org/articles/57162). First, because many researchers have temporarily lost access to the labs, we will give authors as much time as they need to submit revised manuscripts. We are also offering, if you choose, to post the manuscript to bioRxiv (if it is not already there) along with this decision letter and a formal designation that the manuscript is "in revision at eLife". Please let us know if you would like to pursue this option. (If your work is more suitable for medRxiv, you will need to post the preprint yourself, as the mechanisms for us to do so are still in development.)

Summary:

This manuscript describes the use of single nuclei RNA sequencing of the mouse placenta. The authors performed snRNA-seq at four different stages of mouse embryonic development E9.5, E10.5, E12.5 and E14.5 and performed various computational analyses to identify distinct cells types of the developing placenta, especially, the progenitor and differentiated cell types of the placental labyrinth.

The use of separated nuclei allows analysis of the transcripts in the syncytial layers of the placenta to be analysed for the first time. This is a very significant advance and opens up a new dimension of analysis. The study provides information about possible new transcriptional regulatory mechanisms in placental cell types. The reviewers are therefore enthusiastic about this manuscript. As with other single nuclei or single cell papers this manuscript contains a very large amount of data. The quality and depth of these data will be extremely valuable to those working in the field. However, there are some areas of the manuscript that require revision and/or clarification.

Essential revisions:

1) In several places are there are claims that are too bold – they are not fully supported by the data and need to be toned down. For example: subsection “Developmental time course and trajectory inference reveal details of lineage dynamics and commitment” please change "luminogenesis" ->" lumenogenesis in early embryos". The inference of the role of Podxl is too strong. The lumenogenesis described by Shahbazi et al., is the formation of a lumen in the centre of a rosette of primitive endoderm whereas the lumen of the maternal blood spaces in the labyrinthine layer is formed from the invagination of the fused chorion and allantois by finger-like projections of extra-embryonic mesodermal cells. These processes are quite likely to be regulated differently. Similarly, the inference in subsection “Defining distinct roles of the trophoblast subtypes at the gas exchange interface” is too strong.

2) The authors seem to describe LaTP2 cells as "novel intermediate states". This is a bit of a stretch given the lack of information about the direct lineage relationship between the precursors and the differentiated cell types, or about how LaTP and LaTP2 might be related. In the absence of direct lineage analysis this text needs to be toned down to avoid overstating the findings (but please note the comment below about immunolocalization).

3) The discovery of LaTP2 cells is potentially very interesting and these cells seem to be largely defined by the absence of Met, Ror2, Lgr5. Can the authors rule out that this subpopulation is standing out simply because of technical dropout of these gene markers? The Uena et al., study showed that c-Met (and its ligand Hgf) sustain the LaTP population. Please comment on whether you think that the LaTP2 cells could be controlled through alternative pathways.

4) The analyses of hematopoietic cells using GATA4 and cKit is problematic. Multiple previous studies have shown that placental hematopoiesis starts at ~E10.5 as the hematopoietic stem cells emerge at that developmental stage and maximal number of HSCs are detected around E12.5. Thus, it is surprising that so many hematopoietic cells are detected at E9.5 and E10.5 and fewer at E12.5. Please reanalyze the data using additional hematopoietic markers.

5) The study does not provide information of Parietal giant cells and invasive trophoblast population and also contains incomplete information on other junctional zone cells. There are published reports that Junctional Zone cells, like glycogen cells undergo major changes in gene expression and function after E14.5 leading to the expansion of the junctional zone. (Coan et al., 2006). Also, Figure 3B clearly shows strong presence of junctional zone progenitors at E14.5, indicating incomplete differentiation. Thus, the study is not a complete picture of mouse placental development and authors should be more cautious about this aspect in the description of the title and within the body of the manuscript. Their description should be specific to their study, which is mostly focusing on the placental labyrinth zone involving SynT and S-TGC development and function.

6) Similarly, the authors need to explicitly comment on the cells that are poorly represented in their analysis. Specifically, the analyses of spongiotrophoblast vs. glycogen cells are also questionable. Ectoplacental cone derived cells expressing genes like Tpbp and Ascl2 are missing. These are well known genes with respect to the development of the spongiotrophoblast and progenitors of the invasive glycogen cells. The authors should include analyses with respect to these genes. In addition, some of their markers are also questionable. For example, Prl8a9 is highly expressed in Glycogen cells (Simmons et al., 2008). Also, authors should include data or comment about specific Glycogen cell markers like Connexin 31, IGF2R.

7) Subsection “Defining distinct roles of the trophoblast subtypes at the gas exchange interface” – While you can make the argument this cell layer functions as if it is a maternal endothelium I don't think it is correct to state "it forms a maternal endothelium". After all it is not composed of endothelial cells and does not have the receptors for the angiogenic growth factors made by SynTI and II as illustrated in Figure 5B. A description of the endothelial like properties of these cells would be warranted – indeed desirable. For example, do they have (nor not have) RNAs encoding anti- and adhesive surface markers (ICAM for example).

8) The section describing the likely functional differences in the trophoblast layers in the labyrinth requires additional detail and clarification (subsection “Defining distinct roles of the trophoblast subtypes at the gas exchange interface” and Figure 4). There seem to be three groups of "differentially expressed" genes – one for each of the trophoblast types (S-TGCs, SynTI and SynTII). It is not clear from the methods whether this is from all cells (albeit down-sampled) at all time points. I assume that as the FindMarkers function was used then the comparison is the group of interests (S-TGCs for example) vs all other groups. It would be helpful to have this detail clarified. Also, the authors should comment on why there appear not to be any genes that are lower in any group (so were only positive markers selected?) and what criteria were used to define a "differentially expressed" gene. In Supplementary file 14 there are apparently significantly regulated transcripts with a fold change of ~1.2 so are the differently expressed genes defined solely by the Padj? – but there is clearly some additional filtering (as shown in Figure 4C-E). Were all genes listed in Supplementary file 14 used for the ontogeny analysis? – the numbers of genes listed would suggest not (Supplementary file 14, 2104 genes; Supplementary file 15 686+483+842=1910). I presume not all the genes map to GO terms but the methods for the GO analysis are not described so it is not possible to know this.

9) The GO analysis yields a very large number of significant terms with the top 5 clusters for each cell type presented in Figure 4C-E. The rationale for choosing the "top 5" is not clear – especially as clusters (ie GO terms) are hierarchically related. This is exemplified in 4D where cluster D (GO:0072659) is actually a child of term B (GO:1990778). Finally, the large number of terms reported is hard to understand as they are from different levels in the GO term hierarchy and report the same pathway (eg hippo signalling and regulation of hippo signalling).

10) There is very little said about the possible function and development of the S-TGCs. These cells (as their name implies) have previously been described as being similar to giant cells. The UMAP shown in Figure 2A suggests they arise from a population of cells related to the junctional zone precursors. Indeed, this would fit with the prolactin expression profiles (see Simmons et al., 2008). Notably, the prolactins, which are present in the S-TGCs are not in found in the labyrinthine syncytial cells but in EPC derived cells lineage. This requires some discussion.

11) The analysis of likely ligand – receptor interactions, and the transcription factor regulons, raise interesting new leads, but very few are validated.

12) As authors are proposing presence of a new progenitor population, it is important to provide more information about the origin, timing of emergence as well as spatial distribution of the two LaTP progenitor lineages. These aspects need more analyses, especially at an earlier developmental time point (E7.5-E9). As the authors have identified probable unique marker genes, they should include additional studies with placental sections during early time points (E7.5-E9) to better characterize the two LaTP populations.

To address these concerns (11-12), we request some additional immunolocalization studies be added. Of particular interest are transcription factors and markers that differentiate between the LaTP and LaTP2 lineages. We appreciate this this work will be constrained by the availability of suitable antibodies.

13) The two syncytial layers for a continuous barrier between the maternal and fetal compartments. Nutrient, gas and waste products have to cross or be transported across this barrier – ie they need to cross both layers. The section describing this (subsection “Defining distinct roles of the trophoblast subtypes at the gas exchange interface”) does not address this. Indeed, the analytical approach seems flawed – by focusing on the genes that are "enriched" in one layer, overlooks that fact that a transport network may exist across both layers. Of course, it is quite possible that one mechanism may be used to transport something across one layer and a different mechanism operate for the other layer, nonetheless the nutrients needs to traverse both layers. Describing the transporter RNAs that are present in both SynT layers is essential. This is an opportunity that has been missed as this study has the resolution to evaluate this for the first time.

https://doi.org/10.7554/eLife.60266.sa1

Author response

Summary:

This manuscript describes the use of single nuclei RNA sequencing of the mouse placenta. The authors performed snRNA-seq at four different stages of mouse embryonic development E9.5, E10.5, E12.5 and E14.5 and performed various computational analyses to identify distinct cells types of the developing placenta, especially, the progenitor and differentiated cell types of the placental labyrinth.

The use of separated nuclei allows analysis of the transcripts in the syncytial layers of the placenta to be analysed for the first time. This is a very significant advance and opens up a new dimension of analysis. The study provides information about possible new transcriptional regulatory mechanisms in placental cell types. The reviewers are therefore enthusiastic about this manuscript. As with other single nuclei or single cell papers this manuscript contains a very large amount of data. The quality and depth of these data will be extremely valuable to those working in the field. However, there are some areas of the manuscript that require revision and/or clarification.

We thank the reviewers for their interest and insightful suggestions to improve this manuscript. We have added additional analyses and validation of findings with immunofluorescence staining in the main and supplementary figures. Additionally, we have incorporated the reviewers’ feedback and modified the text for clarity and to better reflect the data presented. The specific changes are detailed below.

Essential revisions:

1) In several places are there are claims that are too bold – they are not fully supported by the data and need to be toned down. For example: subsection “Developmental time course and trajectory inference reveal details of lineage dynamics and commitment” please change "luminogenesis" ->" lumenogenesis in early embryos". The inference of the role of Podxl is too strong. The lumenogenesis described by Shahbazi et al. is the formation of a lumen in the centre of a rosette of primitive endoderm whereas the lumen of the maternal blood spaces in the labyrinthine layer is formed from the invagination of the fused chorion and allantois by finger-like projections of extra-embryonic mesodermal cells. These processes are quite likely to be regulated differently. Similarly, the inference in subsection “Defining distinct roles of the trophoblast subtypes at the gas exchange interface” is too strong.

The sentences concerning the formation of lumen structures in the early embryo and the maternal blood sinuses in the placenta have been amended to reflect that these are different processes. Podxl likely does not function the same way in both contexts. We have changed the text to clarify that the expression and charge of Podxl may function to maintain open blood sinuses, not generate sinuses de novo.

2) The authors seem to describe LaTP2 cells as "novel intermediate states". This is a bit of a stretch given the lack of information about the direct lineage relationship between the precursors and the differentiated cell types, or about how LaTP and LaTP2 might be related. In the absence of direct lineage analysis this text needs to be toned down to avoid overstating the findings (but please note the comment below about immunolocalization).

“Novel intermediate States” is meant to reflect the SynTI Precursor, SynTII Precursor, and S-TGC Precursor populations, which have distinct gene expression from both progenitors and their respective differentiated states. The text has been modified to reflect this. In concert with Essential revision point 12, we have provided additional evidence for gene expression differences between LaTP and LaTP2. Between the LaTP populations, we identify Met as largely restricted to LaTP and Egfr largely restricted to LaTP2 (Figure 2—figure supplement 1C). We colocalize each protein with EPCAM, an established marker of LaTP (previously described in Ueno et al., 2013) which through sparse clonal labeling has been shown to give rise to SynTI, SynTII, and S-TGC. We identify at E9.5 that MET and EPCAM have matching domains of expression, while EGFR and EPCAM are largely non-overlapping, although rare double positive cells do exist (Figure 2—figure supplement 1D). Similar to mRNA, EGFR and MET protein appear to be mutually exclusive. The MET/EPCAM positive cells appear more basally localized than EGFR positive cells suggestive that the former gives rise to the later. The new analysis and immunofluorescence data have been added to the supplementary figures and the text has been changed to reflect these findings.

3) The discovery of LaTP2 cells is potentially very interesting and these cells seem to be largely defined by the absence of Met, Ror2, Lgr5. Can the authors rule out that this subpopulation is standing out simply because of technical dropout of these gene markers? The Uena et al., study showed that c-Met (and its ligand Hgf) sustain the LaTP population. Please comment on whether you think that the LaTP2 cells could be controlled through alternative pathways.

The data in Figure 1—figure supplement 1B and C and Figure 2- 1A demonstrate that LaTP and LaTP2 do not differ drastically in the number of unique genes identified, the number of total transcripts identified, or the percent of mitochondrial reads. Thus, we believe the differences in these populations is not due to technical dropout or low information cell driving the clustering. We have identified Egfr as a strong marker of the LaTP2 population (Figure 2—figure supplement 1C and D). We also confirm strong MET protein expression in the LaTP population. Because LaTP2 do not express Met or Wnt receptors, Ror2 and Lgr5, it is unlikely this population is sustained by HGF-MET or Wnt signaling. However, given the expression of Egfr (both transcript and protein) in LaTP2 and the expression of Egf ligands, Tgfa and Hbegf, by SynTI we hypothesize that Egf signaling is important for control of LaTP2. The text and figures (Figure 2—figure supplement 1B-E) have been updated to include these new data.

4) The analyses of hematopoietic cells using GATA4 and cKit is problematic. Multiple previous studies have shown that placental hematopoiesis starts at ~E10.5 as the hematopoietic stem cells emerge at that developmental stage and maximal number of HSCs are detected around E12.5. Thus, it is surprising that so many hematopoietic cells are detected at E9.5 and E10.5 and fewer at E12.5. Please reanalyze the data using additional hematopoietic markers.

We included clusters 15 and 19 in the Fetal Mesenchyme group due to their mesenchymal features, but did note these cell expressed some HSC markers but lacked other important markers. We analyzed the gene expression in these clusters according to markers found in McKinney-Freeman et al., 2013. No cluster identified in our data matches with HSCs, although other hematopoietic lineages express some of these markers. Recent, single cell sequencing data in the mouse identified HSCs to be only 0.05% of total cells. By these numbers we would only expect 13 HSC nuclei in our data. Likely, the lack of HSCs in our data is due to the rarity of this population.

We have edited the text to clarify the annotation of these cells in mesenchyme (not trophoblast and not HSC). Additionally, we performed immunofluorescence staining for another shared marker of clusters 15 and 19, PDPN. This shows expression in interstitial cells with long processes which seem to make contact with fetal endothelial cells. These data are now included in Figure 1—figure supplement 1E and mentioned in text.

5) The study does not provide information of Parietal giant cells and invasive trophoblast population and also contains incomplete information on other junctional zone cells. There are published reports that Junctional Zone cells, like glycogen cells undergo major changes in gene expression and function after E14.5 leading to the expansion of the junctional zone. (Coan et al., 2006). Also, Figure 3B clearly shows strong presence of junctional zone progenitors at E14.5, indicating incomplete differentiation. Thus, the study is not a complete picture of mouse placental development and authors should be more cautious about this aspect in the description of the title and within the body of the manuscript. Their description should be specific to their study, which is mostly focusing on the placental labyrinth zone involving SynT and S-TGC development and function.

The title and descriptions of the study have been changed to reflect the focus on the development of the placental labyrinth and aspects not included in this study (maturation of the junctional zone after E14.5) are explicitly listed.

6) Similarly, the authors need to explicitly comment on the cells that are poorly represented in their analysis. Specifically, the analyses of spongiotrophoblast vs. glycogen cells are also questionable. Ectoplacental cone derived cells expressing genes like Tpbp and Ascl2 are missing. These are well known genes with respect to the development of the spongiotrophoblast and progenitors of the invasive glycogen cells. The authors should include analyses with respect to these genes. In addition, some of their markers are also questionable. For example, Prl8a9 is highly expressed in Glycogen cells (Simmons et al., 2008). Also, authors should include data or comment about specific Glycogen cell markers like Connexin 31, IGF2R.

We used Pcdh12 as a marker of glycogen cell identity (Simmons et al., 2008). In our data Pcdh12 is expressed in increasing levels from JZP1 to JZP2 to differentiated glycogen cells. We thank the reviewers for suggesting Simmons et al., BMC Genomics 2008 as a resource for determining Glycogen Cell vs. Spongiotrophoblast specific expression of prolactin genes. We analyzed the expression of the prolactin genes highlighted in this paper in our dataset. This data is now included in Figure 2-figure supplement 3A and B, alongside the expression of Pcdh12 and the other markers suggested by the reviewers (Tpbpa, Ascl2, Connexin31/Gjb3, and Igf2r). Our data is largely in agreement with the in situ hybridization data from Simmons et al. 2008, and this is summarized in Figure 2-figeru supplement3. While it is clear that there are several common markers of the SpT and Glycogen trophoblast and that prolactin expression in the placenta is quite complex, we feel the annotation we have provided are the best possible considering the current understanding of the field.

7) Subsection “Defining distinct roles of the trophoblast subtypes at the gas exchange interface”- While you can make the argument this cell layer functions as if it is a maternal endothelium I don't think it is correct to state "it forms a maternal endothelium". After all it is not composed of endothelial cells and does not have the receptors for the angiogenic growth factors made by SynTI and II as illustrated in Figure 5B. A description of the endothelial like properties of these cells would be warranted – indeed desirable. For example, do they have (nor not have) RNAs encoding anti- and adhesive surface markers (ICAM for example).

We analyzed these cells for specific endothelial markers and do not see them highly expressed. The language has been changed to reflect that these cells form a lining but lack canonical endothelial properties.

8) The section describing the likely functional differences in the trophoblast layers in the labyrinth requires additional detail and clarification (subsection “Defining distinct roles of the trophoblast subtypes at the gas exchange interface” and Figure 4). There seem to be 3 groups of "differentially expressed" genes – one for each of the trophoblast types (S-TGCs, SynTI and SynTII). It is not clear from the methods whether this is from all cells (albeit down-sampled) at all time points. I assume that as the FindMarkers function was used then the comparison is the group of interests (S-TGCs for example) vs all other groups. It would be helpful to have this detail clarified. Also, the authors should comment on why there appear not to be any genes that are lower in any group (so were only positive markers selected?) and what criteria were used to define a "differentially expressed" gene. In Supplementary file 14 there are apparently significantly regulated transcripts with a fold change of ~1.2 so are the differently expressed genes defined solely by the Padj? – but there is clearly some additional filtering (as shown in Figure 4C-E). Were all genes listed in Supplementary file 14 used for the ontogeny analysis? – the numbers of genes listed would suggest not (Supplementary file 14, 2104 genes; Supplementary file 15 686+483+842=1910). I presume not all the genes map to GO terms but the methods for the GO analysis are not described so it is not possible to know this.

For these analyses the cells in the SynTI, SynTII, and S-TGC clusters were subset from the trophoblast dataset containing all timepoints. FindAllMarkers was used to identify positive differentially expressed genes among these three populations (one population against the other two, for each population). Differentially expressed genes are determined using a Wilcoxon rank sum test considering genes expressed in at least 25% of cells in a cluster and those with log fold change of 0.25 or greater.

All markers were included in the GO analysis using the package clusterProfiler. The list of genes was first converted to Entrez IDs. In each cluster approximately 1% of genes did not match to Entrez IDs (SynTI – 7/715; SynTII – 8/507; S-TGC – 12/882). Transcripts that did not match represent non-coding RNAs and predicted genes. Those genes with Entrez IDs were then subject to GO analysis and for each cluster with 95-96% of genes mapping to at least one GO categories (SynTI – 686/708; SynTII – 482/499; S-TGC – 842/870). Therefore, the analysis considers 2010 out of a possible 2104 marker genes (95.5%). The methods have been updated with a thorough description of the GO analysis.

9) The GO analysis yields a very large number of significant terms with the top 5 clusters for each cell type presented in Figure 4C-E. The rationale for choosing the "top 5" is not clear – especially as clusters (ie GO terms) are hierarchically related. This is exemplified in 4D where cluster D (GO:0072659) is actually a child of term B (GO:1990778). Finally, the large number of terms reported is hard to understand as they are from different levels in the GO term hierarchy and report the same pathway (eg hippo signalling and regulation of hippo signalling).

We thank the reviewers for pointing out the redundancy. We have repeated this analysis, this time using a cutoff to consolidate hierarchically related terms with greater similarity than 0.5. We still show the top 5 GO terms ranked by adjusted p-value for clarity of presentation. However, the top 10 GO terms are included in Figure 4—figure supplement 1A and all significantly enriched GO terms are included in Supplementary Table 15. If reviewers are aware of alternative means/packages for consolidating related terms, we would be happy to apply them. For now, we present our latest analysis in Figure 4C-E and Figure 4—figure supplement 1A have updated the text accordingly.

10) There is very little said about the possible function and development of the S-TGCs. These cells (as their name implies) have previously been described as being similar to giant cells. The UMAP shown in Figure 2A suggests they arise from a population of cells related to the junctional zone precursors. Indeed, this would fit with the prolactin expression profiles (see Simmons et al., 2008). Notably, the prolactins, which are present in the S-TGCs are not in found in the labyrinthine syncytial cells but in EPC derived cells lineage. This requires some discussion.

The reviewer’s raise an interesting point concerning the origin of S-TGCs. Simmons et al., 2007 and Simmons et al., 2011 provide evidence for these cells arising from either ectoplacental cone or chorion trophoblast. Ueno et al. 2013 demonstrate S-TGC can arise from Epcamhi LaTP (a chorionic population). Our data place the S-TGC as sprouting from between the LaTP and JZP populations. Unfortunately, this does not conclusively provide an answer to the origin of S-TGC, but supports idea that this population may arise from multiple locations. Additional discussion of the genesis has been included in the text describing Figure 2, and additional discussion of function has been added to the discussion of Figure 4.

11) The analysis of likely ligand – receptor interactions, and the transcription factor regulons, raise interesting new leads, but very few are validated.

In addition to the validation of VEGFA protein expression in SynTII cells (Figure 5B and Figure 5—figure supplement 2A), we also validated the expression of HBEGF in SynTI through colocalization with SynTI marker STRA6 (Figure 5—figure supplement 2B). We further confirm Egf signaling between SynTI and SynTII and colocalized the ligand HBEGF and the receptor EGFR. These experiments reveal only HBEGF protein in SynTI but co-expression of HBEGF and EGFR in SynTII, suggesting functional signaling in this layer (Figure 5D and Figure 5—figure supplement 2C).

To validate the expression of a SCENIC predicted TF regulon, we selected GATA1. Staining for GATA1 revealed expression in trophoblast to be restricted to SynTII, although GATA1 is strongly expressed by several hematopoietic lineages in the fetal vessels (Figure 6F and Figure 6—figure supplement 1D). Colocalization of GATA1 with SynTII marker IGF1R demonstrates GATA1 positive nuclei abutting the basal membrane of SynTII marked by IGF1R.

12) As authors are proposing presence of a new progenitor population, it is important to provide more information about the origin, timing of emergence as well as spatial distribution of the two LaTP progenitor lineages. These aspects need more analyses, especially at an earlier developmental time point (E7.5-E9). As the authors have identified probable unique marker genes, they should include additional studies with placental sections during early time points (E7.5-E9) to better characterize the two LaTP populations.

To address these concerns (11-12), we request some additional immunolocalization studies be added. Of particular interest are transcription factors and markers that differentiate between the LaTP and LaTP2 lineages. We appreciate this this work will be constrained by the availability of suitable antibodies.

The reviewers raise important points concerning the emergence and location of the two LaTP populations. To address this question, we have now done additional immunostaining as described above. To recap: first, we identify Met as a specific marker of LaTP, and Egfr as a highly enriched marker in LaTP2 (Figure 2—figure supplement 1B and C). We then colocalized each of these proteins with EPCAM, the canonical LaTP marker identified by Ueno et al., 2013 (Figure 2—figure supplement 1D). At E9.5 EPCAM and MET have overlapping domains of expression. This is in agreement with the transcript data and suggests that the LaTP cluster is the LaTP population as defined by Ueno et al., 2013. In contrast, only the rare EGFR+ cell showed coexpression with EPCAM protein (Figure 2—figure supplement 1D). Interestingly the EGFR+ cells were juxtaposed to the MET+ and positioned as such to suggest that the later may be giving rise to the former.

In addition to the staining at E9.5, we also added staining at E8.5 (Figure 2—figure supplement 1E). This timepoint is coincident with chorioallantoic fusion. We found many EPCAM/MET positive cells in the chorion, but few to no EGFR expressing cells. These data would be consistent with interpretation that EPCAM/MET precede and may even give rise to the EGFR+ LaTP2 cells. Along with figures noted above, we have added text to discuss these findings.

13) The two syncytial layers for a continuous barrier between the maternal and fetal compartments. Nutrient, gas and waste products have to cross or be transported across this barrier – ie they need to cross both layers. The section describing this (subsection “Defining distinct roles of the trophoblast subtypes at the gas exchange interface”) does not address this. Indeed, the analytical approach seems flawed – by focusing on the genes that are "enriched" in one layer, overlooks that fact that a transport network may exist across both layers. Of course, it is quite possible that one mechanism may be used to transport something across one layer and a different mechanism operate for the other layer, nonetheless the nutrients needs to traverse both layers. Describing the transporter RNAs that are present in both SynT layers is essential. This is an opportunity that has been missed as this study has the resolution to evaluate this for the first time.

We apologize for the lack of clarity concerning this analysis. The analysis of transport genes includes all genes found to differentially expressed in any of the interface populations (SynTI, SynTII, and/or S-TGC) compared to all other trophoblast populations. However, since this is distinct from the differentiation expression shown in Figure 4B and used in the GO analysis in Figure 4C-E. The text has been changed to explicitly state which genes were analyzed. This information is also included in the figure legend.

https://doi.org/10.7554/eLife.60266.sa2

Article and author information

Author details

  1. Bryan Marsh

    1. The Eli and Edythe Broad Center of Regeneration Medicine and Stem Cell Research, Center for Reproductive Sciences, University of California, San Francisco, San Francisco, United States
    2. Department of Urology, University of California, San Francisco, San Francisco, United States
    Contribution
    Resources, Data curation, Formal analysis, Investigation, Visualization, Methodology, Writing - original draft
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-4979-5233
  2. Robert Blelloch

    1. The Eli and Edythe Broad Center of Regeneration Medicine and Stem Cell Research, Center for Reproductive Sciences, University of California, San Francisco, San Francisco, United States
    2. Department of Urology, University of California, San Francisco, San Francisco, United States
    Contribution
    Conceptualization, Supervision, Funding acquisition, Writing - review and editing
    For correspondence
    robert.blelloch@ucsf.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1975-0798

Funding

National Institute of Child Health and Human Development (P50 HD055764)

  • Robert Blelloch

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank the members of the University of California, San Francisco National Center of Translational Research in Reproduction and Infertility for helpful comments during the design, execution, and publication of this project. A special thanks to Susan Fisher, Alex Pollen, and Peter Sudmant for insightful comments during data analysis and preparation on the manuscript. We would also like to thank all members of the Blelloch Lab for their comments and support, especially Brian DeVeale for help with data analysis and proofreading the manuscript. We would also like to acknowledge our funding source, the NIH Eunice Kennedy Shriver National Institute for Child Health and Human Development P50 HD055764.

Ethics

Animal experimentation: This study was performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. All of the animals were handled according to approved institutional animal care and use committee (IACUC) (Approval number: AN173513) of the University of California, San Francisco.

Senior Editor

  1. Didier YR Stainier, Max Planck Institute for Heart and Lung Research, Germany

Reviewing Editor

  1. Steve Charnock-Jones, University of Cambridge, United Kingdom

Reviewer

  1. Steve Charnock-Jones, University of Cambridge, United Kingdom

Publication history

  1. Received: June 22, 2020
  2. Accepted: October 30, 2020
  3. Accepted Manuscript published: November 3, 2020 (version 1)
  4. Version of Record published: November 16, 2020 (version 2)

Copyright

© 2020, Marsh and Blelloch

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 2,038
    Page views
  • 364
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Developmental Biology
    Markus Frederik Schliffka et al.
    Research Article

    During the first days of mammalian development, the embryo forms the blastocyst, the structure responsible for implanting the mammalian embryo. Consisting of an epithelium enveloping the pluripotent inner cell mass and a fluid-filled lumen, the blastocyst results from a series of cleavages divisions, morphogenetic movements and lineage specification. Recent studies identified the essential role of actomyosin contractility in driving the cytokinesis, morphogenesis and fate specification leading to the formation of the blastocyst. However, the preimplantation development of contractility mutants has not been characterized. Here, we generated single and double maternal-zygotic mutants of non-muscle myosin II heavy chains (NMHC) to characterize them with multiscale imaging. We find that Myh9 (NMHC II-A) is the major NMHC during preimplantation development as its maternal-zygotic loss causes failed cytokinesis, increased duration of the cell cycle, weaker embryo compaction and reduced differentiation, whereas Myh10 (NMHC II-B) maternal-zygotic loss is much less severe. Double maternal-zygotic mutants for Myh9 and Myh10 show a much stronger phenotype, failing most attempts of cytokinesis. We find that morphogenesis and fate specification are affected but nevertheless carry on in a timely fashion, regardless of the impact of the mutations on cell number. Strikingly, even when all cell divisions fail, the resulting single-celled embryo can initiate trophectoderm differentiation and lumen formation by accumulating fluid in increasingly large vacuoles. Therefore, contractility mutants reveal that fluid accumulation is a cell-autonomous process and that the preimplantation program carries on independently of successful cell division.

    1. Developmental Biology
    Naokazu Inoue et al.
    Short Report

    To trigger gamete fusion, spermatozoa need to activate the molecular machinery in which sperm IZUMO1 and oocyte JUNO (IZUMO1R) interaction plays a critical role in mammals. Although a set of factors involved in this process has recently been identified, no common factor that can function in both vertebrates and invertebrates has yet been reported. Here, we first demonstrate that the evolutionarily conserved factors dendrocyte expressed seven transmembrane protein domain-containing 1 (DCST1) and dendrocyte expressed seven transmembrane protein domain-containing 2 (DCST2) are essential for sperm–egg fusion in mice, as proven by gene disruption and complementation experiments. We also found that the protein stability of another gamete fusion-related sperm factor, SPACA6, is differently regulated by DCST1/2 and IZUMO1. Thus, we suggest that spermatozoa ensure proper fertilization in mammals by integrating various molecular pathways, including an evolutionarily conserved system that has developed as a result of nearly one billion years of evolution.