Dynamic WT1 expression during gastrulation specifies peritoneal smooth muscle fate independently of mesothelial fate Open Access

The embryonic origin of vascular smooth muscle cells (vSMCs) is relatively diverse as different regions of the vasculature arise from different progenitors, including neural crest cells, lateral plate mesoderm (LPM) and somatic mesoderm (Donadon and Santoro, 2021). Differentiation of vSMCs and tissue formation in vertebrates are driven by a series of canonical cell signalling events and mechanical forces that regulate progenitor cell recruitment, differentiation and maturation (del Monte et al., 2011; Qi et al., 2021; Stratman et al., 2020). Platelet-derived growth factor receptor beta (PDGFRB) signalling is a key regulatory pathway in vSMC myogenesis since disruption of PDGFRB signalling reduces vSMC proliferation and recruitment (Hellström et al., 1999; Stratman et al., 2020). Specification of vSMCs is marked by expression of smooth muscle cell markers, such as transgelin (TAGLN), alpha-smooth muscle actin (ACTA2), smooth muscle myosin heavy chain (MYH11), desmin and others (Assinder et al., 2009; Muhl et al., 2022; Smyth et al., 2018). However, the time point at which these genes start to be expressed in embryonic development, and their roles during the early stages of vSMC specification and differentiation, remain unclear.

Wilms' tumor protein 1 (WT1) is a transcription factor involved in the development of several tissues and organs in normal embryogenesis. The earliest WT1 expression was identified in the intermediate mesoderm (IM) and LPM in embryonic day (E) 9.5 mouse embryos (Armstrong et al., 1993). Its most prominent roles are in the development of the kidneys, arising from the IM, and of the heart, where the WT1-expressing epicardium contributes to coronary vessel formation (Kreidberg et al., 1993; Rackley et al., 1993). The epicardium is the mesothelial layer of the heart, which is continuous with the pericardial mesothelium of the pericardial cavity. WT1 is also expressed in the pleural and peritoneal mesothelium, lining the cavities and covering the organs housed within, from about E9.5 onwards and throughout adulthood (Armstrong et al., 1993; Moore et al., 1999, 1998; Que et al., 2008; Wilm et al., 2005, 2021).

Previously, studies showed that the epicardium can give rise to vSMCs of the developing coronary vasculature (Guadix et al., 2006; Zhou et al., 2008). Genetic lineage-tracing experiments using a transgenic mouse line expressing Cre recombinase under control of regulatory elements of the human WT1 gene [Tg(WT1-cre)AG11Dbdr; abbreviated as WT1-Cre] in combination with the Rosa26^lacZ/lacZ reporter mouse revealed that a population of WT1-expressing serosa and pleural mesothelium cells give rise to vSMCs in the peritoneal and lung cavities (Que et al., 2008; Wilm et al., 2005). These results suggested that the mesothelium and vasculature of the intestine and lungs may have a developmental relationship similar to that of the epicardium and the coronary vasculature. Similar studies using mWt1/IRES/GFP-Cre (Wt1cre) transgenic mice crossed with the Rosa26R^EYFP reporter mouse also found a contribution of WT1-expressing mesothelial cells (MCs) to the vasculature of the developing intestine (Carmona et al., 2013).

In order to determine when MCs become committed to give rise to intestinal vascular smooth muscle, we previously conducted pulse-chase studies of postnatal and adult mice (Wilm et al., 2021) using a tamoxifen (Tam)-inducible, temporally controlled and Wt1-driven Cre system [Wt1^{tm2(cre/ERT2)Wtp}, Wt1^CreERT2; Zhou et al., 2008]. Our results showed that WT1-expressing peritoneal cells maintained the mesothelium over the intestine and peritoneal wall (and contributed to visceral fat; Chau et al., 2011), but failed to give rise to intestinal vasculature components. However, WT1-expressing epicardium was observed to contribute towards formation of coronary vessels postnatally and in adult mice (Wilm et al., 2021). These findings suggested that the capacity of WT1-expressing MCs within the peritoneal cavity to differentiate into vSMCs must be restricted to developmental stages.

We therefore propose that there is an embryonic contribution of WT1-expressing cells towards the vSMC fate in the peritoneum and intestine, of which the temporo-spatial resolution and molecular mechanisms have remained unresolved so far. Here, to address this, we used the Tam-inducible Wt1^CreERT2 mouse line in combination with Rosa-based reporter mice for lineage-tracing analysis to determine the contribution of WT1-expressing cells towards intestinal vasculature in the mouse embryo. Our results indicate that, during embryonic development, two distinct populations of WT1-expressing cells emerge in the peritoneum, one that defines MCs, and one that defines SMC fate. Lineage-traced cells were found not only as vSMCs but also as visceral SMCs (visSMCs) in the intestinal wall, when Tam had been given at stages around E7.5 to E8.5, while predominantly MCs were labelled when Tam had been given from approximately E9.5 onwards. Importantly, both vSMCs and visSMCs, as well as the general parietal peritoneal wall tissue including the mesothelium, is derived from the LPM.

We provide, for the first time, a detailed analysis of WT1 transcript and protein expression in E6.5-E8.5 mouse embryos. In silico analysis of single-cell RNA sequencing (scRNAseq) data sets revealed that Wt1 is co-expressed with the emerging mesoderm marker brachyury (Bra; T) during gastrulation, and later with genes representing an SMC signature in E6.5-E8.5 mouse embryos. We then describe expression of WT1 in the E7.5 and E8.5 mouse embryo in the primitive streak and the emerging neuro-mesodermal progenitor cells as well as LPM using cutting-edge confocal microscopy techniques. WT1-derived cells within the mesentery of the developing intestine aligned with the developing endothelial vascular network and co-expressed PDGFRB, while inactivation of WT1 affected vascularisation of the developing intestine. Together, these data reveal a previously unappreciated developmental differentiation pathway of visceral and vascular smooth muscle cells influenced by WT1 expression that is independent of mesothelium formation.

To determine when during embryogenesis Wt1-expressing cells within the peritoneal cavity became fated to develop into vSMCs, we combined the Wt1^CreERT2/+ lineage driver with either the Rosa26R (lacZ) or the Rosa26^mTmG (GFP-fluorescent) lineage reporter in mice, gave Tam once at different embryonic stages (pulse), and performed endpoint analysis at stages around E18.5 (chase; see Fig. 1A, Fig. S1A, Table S1). We obtained similar results with both reporter systems: Tam administration at E9.5 or up to E13.5, followed by endpoint whole-mount analysis, revealed almost exclusive labelling of cells covering the surface of the mesentery and intestine at mid-intestinal level (Fig. 1B-E, Fig. S1B-F, Table S1). Based on their position and shape, we concluded that these were MCs. When Tam was given to pregnant dams at E7.5 or E8.5 followed by endpoint analysis, we observed instead labelled cells surrounding the developing vessels of the mesentery, and in what appeared to be the circular visceral smooth muscle of the intestine at mid-intestinal level, but barely in the mesothelium (Fig. 1F-J, Fig. S1G-K). Littermates carrying no Wt1^CreERT2 allele showed no labelled cells after Tam administration (Fig. S1L,M).

We also analysed the stomach with attached spleen and the colon of foetuses under this experimental design, and observed that Tam dosing at E7.5 resulted in labelling of cells with circular visceral smooth muscle appearance but not surface-covering MCs. By contrast, foetuses after Tam dosing at E8.5 showed a mixed population of labelled cells in the stomach, mostly consisting of what appeared to be circular visceral smooth muscle; in the colon, there were only labelled cells with circular visceral smooth muscle appearance. In foetuses that had received Tam at E9.5 up to E14.5, cell labelling occurred predominantly in surface-covering MCs (Fig. S2, Table S1). However, after Tam dosing at E9.5 and up to E11.5 we observed in a few foetuses occasional lacZ-expressing cells that appeared to be circular visceral smooth muscle, in the stomach and colon.

Interestingly, the spleen, which is also an LPM-derived organ, was not labelled when Tam was administered at E7.5 (Fig. S2A), but was strongly labelled in foetuses with Tam dosing from E8.5 onwards (Fig. S2B-F). This indicates that WT1 was not yet expressed at E7.5 in the cells that later will become splenic precursors forming the putative splenic mesenchyme (Hecksher-Sørensen et al., 2004).

Immunofluorescence (IF) staining on sections through intestine and mesentery of Wt1^CreERT2/+;Rosa26^mTmG/+ foetuses (Fig. 2J) confirmed that after inducing Tam-based lineage tracing at E8.5, visSMCs and vSMCs co-expressed ACTA2 and GFP and were found close to the CD31 (PECAM1)-positive vascular endothelial cells, while hardly any MCs (pan-cytokeratin, WT1) were GFP positive (Fig. 2A-E′). By contrast, after Tam dosing at E11.5, MCs were GFP-labelled but no visSMCs or vSMCs (Fig. 2F-I′). These results suggest that, during embryonic development, two distinct populations of WT1-expressing cells emerge: one that defines MCs from approximately E9.5 onwards, and one that defines visceral and vascular SMC fate predominantly at stages around E7.5 to E8.5.

An additional observation in the mice after Tam dosing at E7.5 and E8.5 was that labelled visSMCs and vSMCs could only be found in a short fragment of the mid-intestinal region including mesentery (Fig. S1G). This suggests a short pulse-like or transient period of WT1 expression during which Tam exposure allowed the recombination and labelling of future visSMCs and vSMCs cells in the midgut region. However, the labelled visSMCs in the stomach and colon additionally indicate contribution along fore- and hindgut regions not limited to initiation at E7.5-E8.5.

Next, we analysed expression of Wt1 transcripts during mouse gastrulation stages E6.5 to E8.5 using a published scRNAseq data set (Pijuan-Sala et al., 2019). We found that Wt1 was expressed in mouse embryos throughout this developmental period in 0.65-1.66% of total embryonic cells (Fig. S3A). Mean expression for Wt1 was between 0.955 and 1.13 with cells considered as expressing Wt1 if levels were above 0 [expression levels were log-normalised (count per million)] (Fig. S3B). Wt1 expression was highly dynamic, with the majority of Wt1-expressing cells between E6.5 and E7.25 identified as epiblast, and from E7.5 onwards shifting towards a majority in cells described as ‘mesenchyme’ (from E8.25-E8.5; Fig. S3C). Since visSMCs and vSMCs have their origin in the LPM but the dataset annotation lacked a clear indication which of the mesoderm tissues were LPM, we characterised in a candidate approach somatic, paraxial, nascent, mixed and intermediate mesoderm together with primitive streak, notochord, epiblast and mesenchyme for classical LPM and other mesodermal markers, confirming that ‘mesenchyme’ was representing the LPM population (Fig. S3E,F).

Next, we defined a set of 15 genes each that we considered as an SMC signature (Muhl et al., 2022) and as an MC signature (Kanamori-Katayama et al., 2011) (Fig. S3G). We assessed within all Wt1-expressing cells the enrichment of the SMC signature genes or MC signature genes across the three developmental stages E6.5, E7.5 and E8.5, and found that at E6.5 and E8.5 the SMC gene set was expressed in a higher proportion of cells than the MC gene set, and at E8.5 the SMC gene set had a higher relative expression than the MC gene set (Fig. 3A). Correlation analysis within total Wt1-expressing cells of individual SMC signature or MC signature genes with Wt1 expression across different embryonic time points revealed a positive and highly significant correlation with several genes of the respective sets (Fig. 3B,C). The correlation of Wt1 expression with SMC genes was higher at earlier embryonic stages, but was higher with MC genes at later stages; generally, the analysis suggested a highly dynamic change in correlation across the gene sets (Fig. 3B). Wt1-expressing cells that significantly co-expressed at least three genes (the threshold) of the SMC gene set were found from E6.5 in the epiblast, at E7.0 also in the primitive streak, and from E7.5 in the LPM (Fig. 3D), while Wt1-expressing cells that significantly co-expressed with the MC gene set above the threshold were found from E7.25 in the LPM (Fig. 3E). Next, we explored how the Wt1-expressing cells that expressed the SMC and/or MC gene signatures above the threshold segregated. At E7.5, Wt1-expressing cells separated into two distinct populations of 11 SMC signature cells and ten MC signature cells; ten of the SMC signature cells were in the LPM, while one of the MC signature cells was in the LPM (Fig. 3F,H). By contrast, at E8.5, there were 11 cells with both SMC and MC signatures above the threshold in addition to two distinct populations of 58 SMC signature cells and 33 MC signature cells (Fig. 3G). In the SMC signature group, 45 cells were localised in the LPM (78%), while 24 MC signature cells were localised in the LPM (72%) (Fig. 3I). Our analysis further revealed that Wt1 expression levels in the SMC signature cells were generally lower than in MC signature cells, at both E7.5 and E8.5 (Fig. 3H,I, Fig. S3H,I). Out of the 11 cells with both SMC and MC gene signatures, nine were found in the LPM (Fig. S3J).

Taken together, these data suggest that a consistent population of Wt1-expressing cells arises from the epiblast, via the primitive streak to the emerging mesoderm and the LPM. The Wt1-expressing cells divided into two populations with either smooth muscle or mesothelial fate between E7.0 and E7.5, emerging predominantly in the LPM. Onset for the SMC fate in Wt1-expressing cells appeared to be slightly earlier and more prominent than the MC fate, with respect to number of cells, correlation of expression and significance. Furthermore, Wt1 expression levels were generally lower in cells with the SMC signature than in cells with the MC signature.

The HAND1 and HAND2 transcription factors are key regulators of vascular development (Barnes et al., 2010, 2011; Vincentz et al., 2021, 2017). HAND1-expressing cells emerge from the LPM, similarly to WT1, and both HAND1 and HAND2 have been shown to signify precursors of the MC and SMC lineage in mouse and zebrafish (Maska et al., 2010; Prummel et al., 2022; Reynolds et al., 2021; Yin et al., 2010). Therefore, we explored the correlation of Hand1 or Hand2 expression with the SMC or MC gene sets and in the context of Wt1 expression. Total Hand1-expressing cells showed slightly stronger correlation and significance with the SMC gene sets from E6.5 to E7.25 than with the MC gene sets (Fig. S4A). Hand2-expressing cells were overall more strongly correlated with SMC and MC signature genes than were Hand1-expressing cells (Fig. S4B). There was significant correlation with Wt1 expression in Hand2-expressing cells from E7.5 onwards, but not in Hand1-expressing cells (Fig. S4A,B). We further analysed the correlation of Hand2 expression in Wt1-expressing cells with SMC and MC signature genes, and found limited correlation and significance with either gene set (Fig. S4C,D). The majority of Wt1- and Hand2-co-expressing cells were found in the LPM, starting from E7.0 onwards (Fig. S4E).

These results suggest that Hand2 expression during gastrulation stages correlated with both SMC and MC fate more strongly than did Hand1 expression. By contrast, Hand2 expression in Wt1-expressing cells was not strongly correlated with either SMC or MC fate even though co-expressing cells were identified in the LPM.

Next, we determined whether WT1 protein could be detected in the tail bud of E8.5 wild-type mouse embryos using confocal IF microscopy of whole-mount stained embryos. Our analysis revealed that WT1 was expressed in emerging mesodermal cells, as well as extra-embryonic tissues, including the yolk sac and allantois (Fig. 4A-D). Co-immunostaining for BRA and confocal imaging clearly identified WT1-expressing cells in the BRA-domains of primitive streak and IM/LPM region (Fig. 4A,C,D-G; compare with Fig. S6 based on the EMAP eMouse Atlas project; Richardson et al., 2014). Co-expression analysis of individual focal planes revealed that in the most posterior region around the primitive streak WT1 expression was found in some BRA-positive cells emerging from the node (Fig. 4E), the mesenchyme including neuromesodermal progenitor cells and IM/LPM, the neural plate, primitive blood cells and allantois (Fig. 4D-G, and insets). Quantification of co-expression of WT1 and BRA (Fig. 4E,F, insets) revealed that cells with strong WT1 expression were also positive for strong BRA expression (Fig. 4H), confirming that WT1 was co-expressed in BRA-positive cells emerging from the mesoderm. Of note, our analysis of the scRNAseq dataset suggested that very few cells contained transcripts of both Wt1 and Bra, whereas protein for both genes seemed to be detectable in more cells.

At E7.5, the co-expression domain between WT1 and BRA was more restricted, with WT1 mostly expressed in the epiblast, where it was co-expressed with BRA, and in the primitive streak (Fig. S7B). Further confocal image analysis of triple whole-mount IF for WT1, BRA and TAGLN in E8.5 mouse tailbuds demonstrated protein co-expression of WT1 with TAGLN in the mesoderm/paraxial mesoderm and LPM (Fig. 4I-N, Fig. S7A). These data support the observation that co-expression of WT1 and TAGLN in the emerging LPM define the smooth muscle population. Analysis of HAND2 and WT1 expression in the emerging LPM of wild-type mouse embryos at E8.5 showed that there were cells that either co-expressed these proteins, or expressed them individually in a mutually exclusive manner (Fig. S5).

Taken together, these data show that WT1 is dynamically expressed in the gastrulating embryo in cells of the primitive streak and emerging mesoderm, including the neuromesodermal progenitor cells and LPM, where it is co-expressed with HAND2 and the SMC marker TAGLN.

To obtain a better temporal resolution of the cells that had expressed WT1 at gastrulation stages for lineage tracing of SMCs, we administered Tam at E7.5 or E8.5 and observed the range and localisation of GFP-positive cells at E9.5 (Fig. S8A).

Tam dosing at E7.5 resulted in very few GFP-positive cells, mostly found in the pro-epicardium (Fig. S8B-E). In the three litters analysed, we found that all 15 embryos had between two and 20 GFP-positive cells in the pro-epicardium, while GFP-positive cells (range 1-12 cells) in the posterior trunk region were observed only in nine out of 15 embryos (Table S2, Fig. S11). Importantly, we also detected GFP-positive cells in the neural tube of embryos after Tam at E7.5 (Fig. S8F-I) and E8.5 (Movie 1), in line with our observations in Figs 3 and 4 that WT1 was also expressed in the emerging neuroectoderm. By contrast, in embryos after Tam administration at E8.5 we observed abundant GFP-positive cells localised to the pro-epicardium at the caudal side of the forming heart (Fig. S8J-M), and to the urogenital ridge (Fig. S9B,C). We observed GFP-positive cells consistently towards the ventral midline, including the mesentery, in sections, and by using light-sheet microscopy of the trunk region of the E9.5 embryo (Figs S9B,C, S12, Movie 1, Table S3).

Together, these results demonstrate that the Wt1^CreERT2/+ lineage-tracing system detected cells that expressed WT1 at around E7.5 and E8.5 within a similar spectrum of cell progeny, but GFP-positive cell numbers were smaller when Tam was given at E7.5. This was in agreement with the scRNAseq data analysis (Fig. 3) and distribution of WT1-expressing cells by whole-mount IF (Fig. 4). Moreover, we could identify GFP-positive cells that appeared in the region of the mesentery as potential precursors for visceral and vascular smooth muscle of the developing intestinal tract.

To elucidate the developmental path and fate of the vSMCs and visSMCs using the emergence of the GFP-positive cells, we analysed embryos at E10.5 after Tam dosing at E8.5. At this stage, GFP-positive cells were predominantly seen in the region of the urogenital ridge and the heart (Fig. S9A,D-F′). In tissue sections, GFP-positive cells localised in the region between the dorsal aorta and the ventral mesentery, and within the urogenital ridge (Fig. S9E,E′). Some of the cells within the region between dorsal aorta and ventral mesentery appeared to be situated close to endothelial cells (Fig. S9F,F′).

In embryos dosed with Tam at E8.5 and analysed at E12.5 (Fig. 5A), we found GFP-positive cells in a spotty appearance within the mesentery of the developing small intestinal segment of the intestinal tract (Fig. 5B,B′). IF analysis on sections through this region revealed that GFP-positive cells were localised within the mesentery and in the developing intestinal wall near the forming vascular plexus (Fig. 5C,D). GFP-expressing cells were distinct from endothelial cells but closely associated (Fig. 5D). Furthermore, our analysis showed that some of the GFP-labelled cells within the mesentery and the intestinal wall co-expressed PDGFRB, suggesting that they had started the smooth muscle differentiation programme (Fig. 5E-F″). Occasionally, we found a few GFP-positive cells in the mesothelial covering of the mesentery (Fig. 5E,F). Together, these results indicate that Wt1 lineage-traced cells, labelled at around E8.5, appeared in the vicinity of the developing vasculature and co-expressed PDGFRB, a key marker of vascular smooth muscle progenitor cells during stages of intestinal patterning and angiogenesis. When endpoint analysis was performed at E15.5, we detected GFP-positive cells within the visceral smooth muscle and surrounding mesenteric vessels (Fig. 5G,H), similar to findings from the endpoint analysis at around E18.5 (Fig. 1I). We also explored the Wt1 lineage in embryos after Tam administration at E6.5 and analysis at E14.5, but observed only very few GFP-labelled cells near the intestine and gonad in one out of five analysed embryos (from three litters; Fig. S10).

We next examined how deletion of Wt1 during gastrulation stages would affect vascularisation of the developing intestine and mesentery by administering Tam at E7.5 to three litters of Wt1^CreERT2/fl;Rosa26^mTmG/mTmG ([email protected]) and Wt1^fl/+;Rosa26^mTmG/mTmG (control) embryos (Fig. 6A). Cre activity could be demonstrated in the [email protected] embryos by virtue of GFP expression throughout the embryos, most prominently in the heart (Fig. 6B), suggesting that WT1 had been inactivated in at least some of the cells expressing Wt1 at E7.5. In the intestines of [email protected] embryos, we found higher variability in the density of endothelial CD31 expression, and in the vascular network formation, when compared to control embryos (Fig. 6C-J). Images captured by 3D confocal microscopy were tested for batch effects, which could be neglected (Fig. 6F). Our analysis revealed that the variability in the density of CD31 expression was higher in the [email protected] group than in the controls (Fig. 6E,G), and the coefficient of variation for most parameters of vascular network formation was much higher in the [email protected] embryos than in controls (Fig. 6J), again revealing higher variability. Furthermore, we could detect ACTA2-expressing cells around CD31-expressing vessels within the developing intestine, and observed that in the [email protected] embryos the ACTA2-expressing cells appeared less aligned and attached to the endothelial cells than in the controls (Fig. 6K).

Collectively, these data indicate that Tam administration led to disruption of WT1 function in cells that expressed Wt1 at around E7.5, affecting vessel formation in the developing intestine and mesentery at later stages, in both endothelial and mural cells.

In this study, we have for the first time determined the temporal and spatial relationship between the emerging intestinal vascular and visceral smooth muscle cells and the expression of the transcriptional regulator WT1 during gastrulation and post-gastrulation stages of mouse embryonic development. Our analysis revealed that WT1 is expressed in a progenitor cell pool in the emerging mesoderm of the gastrulating embryo. Some of these cells differentiate into smooth muscle cells in the intestine and mesentery. Our data indicate that this process is temporally independent of the formation of the mesothelium covering these tissues, which also expresses WT1. Thus, our data shed new light onto smooth muscle development as a process independent of mesothelium development, and the temporospatial involvement of WT1 in this process (Fig. 7).

Using Tam-dependent, inducible, genetic lineage tracing, we analysed the embryonic contribution of Wt1-expressing cells to visSMCs and vSMCs of the intestine. In previous studies, we had shown that Wt1-expessing cells give rise to vSMCs in the mesentery and intestine using a non-inducible Wt1-based lineage-tracing system (Wilm et al., 2005). Similarly, by using the same or an alternative Wt1-driven genetic reporter system, it had been demonstrated that Wt1-expressing progenitor cells give rise to visSMCs and vSMCs not only in the intestine but also in the lungs (Cano et al., 2013; Carmona et al., 2013; Que et al., 2008); however, the temporal involvement and spatial significance of Wt1 expression during smooth muscle specification and differentiation had not been determined. In a follow-on study, we determined that Wt1-expressing cells in postnatal stages fail to give rise to intestinal visSMCs and vSMCs, by making use of the same Tam-dependent temporal activation process for lineage analysis as utilised here (Wilm et al., 2021). We concluded that the source of intestinal and mesenteric SMCs had to be cells expressing Wt1 during embryonic development; however, it was not clear whether these cells arise from the overlying mesothelium, which also expresses WT1 during embryonic development, as had been our previous hypothesis (Wilm et al., 2005).

To address this question, we administered Tam between E7.5 and E13.5 using the Wt1^CreERT2/+ driver in combination with either the Rosa26^lacZ or Rosa26^mTmG reporter systems, and analysed the presence of lacZ-expressing and GFP-labelled cells in foetuses just before birth (E17.5, E18.5 or E19.5). Our analysis showed that Tam dosing at E7.5 or E8.5 resulted in labelling of vSMC and visSMCs, which had originated from Wt1-expressing cells. When Tam was given at E9.5 or later stages, the labelled cells were confined to the mesothelial covering of the mesentery and intestine. This finding demonstrates a clearly defined mechanism for the emergence of the mesothelial lineage that is separate from that of the vascular/visceral smooth muscle lineage, indicating a role for WT1 in distinct mechanisms.

Other studies have previously used the Tam-inducible Wt1^CreERT2/+ driver in combination with reporter systems to elucidate the lineage and origin of smooth muscle cells in the lungs. Contribution to airway smooth muscle, vascular smooth muscle and parenchyma was observed in late-stage mouse foetuses when Tam had been given between E9.5 and E11.5 (Dixit et al., 2013; Moiseenko et al., 2017). Time-lapse experiments of ex vivo cultured lungs from lineage-traced E12.5 embryos after Tam dosing at E10.5 revealed a direct contribution of mesothelial cells towards the deeper lung parenchyma (Dixit et al., 2013). These results suggest that the smooth muscle of the airways and lung vasculature may have a contribution via the mesothelium at later developmental stages, in a mechanism different to that described here for the intestinal tract.

In the heart, WT1-based lineage tracing has demonstrated a contribution of epicardial cells to coronary vessel formation (Quijada et al., 2020; Rudat and Kispert, 2012; Wilm et al., 2005; Zhou et al., 2008). Based on in situ hybridisation and IF on sections of E8.5 embryos, WT1 is not expressed in the developing heart at this stage. By contrast, non-inducible Wt1^CreEGFP;Rosa26^mTmG embryos at E8.5 have a number of GFP-expressing cells in the forming heart, indicating an early contribution of Wt1-expressing cells (Rudat and Kispert, 2012). However, WT1 is also expressed in non-epicardial cells in the heart, including myocardium, which makes interpretation of lineage experiments more challenging (Rudat and Kispert, 2012). In addition, our own postnatal lineage-tracing experiments using the Tam-inducible Wt1^CreERT2/+ driver had shown that after birth Wt1-expessing cells contribute to coronary SMCs (Wilm et al., 2021), suggesting a different mechanism of coronary angiogenesis and maintenance of coronary vasculature than in the intestinal tract and mesentery.

The Tam-inducible lineage-tracing system is dependent on Tam-driven recombination activity, which then irreversibly switches on the reporter gene expression. Several studies have analysed the temporal delay between Tam administration and reporter expression, indicating that, depending on route of administration (oral gavage or intraperitoneal injection), a temporal delay of 12-18 h could be observed in other genetic systems (Nguyen et al., 2009; Zhu et al., 2008). This temporal delay needs to be factored in when interpreting the results of the current study, suggesting that about 0.5 embryonic days may have to be added to the stages indicated for the recombination initiation.

The Tam-inducible lineage-tracing system is based on the Wt1^CreERT2/+ driver. We and others have previously shown that the Wt1^CreERT2/+ driver fails to provide 100% recombination efficiency, leading to inefficient cell labelling of between 14.5% and 80% (Chen et al., 2014; Li et al., 2013; Wilm et al., 2021). A range of reasons could contribute to this observation, including inefficiency of the recombination event and/or insufficiency of Tam delivery and physiological distribution, as discussed previously (Wilm et al., 2021). These deficiencies inherent to the Tam- and Cre-driven lineage-tracing/gene-ablation system need to be taken into account when interpreting the distribution and coverage of labelled cells and functional disruption of genes within tissues.

One of the surprising findings of this study emerged from our analysis of scRNAseq data (Pijuan-Sala et al., 2019) showing that, from E6.5 onwards, Wt1 is expressed in the epiblast, primitive streak and emerging mesoderm, and in the LPM where it is co-expressed with key markers of SMC development. Analysis of WT1 expression in early-stage mouse embryos had been previously performed by in situ hybridisation showing Wt1 transcripts in the intermediate mesoderm/LPM in E9.0 mouse embryos (Armstrong et al., 1993). Because further insights into Wt1 expression or function at gastrulation stages have not been published yet, our results are the first to shed light onto WT1 gastrulation-stage expression or engagement. We have identified Wt1 expression in epiblast and primitive streak cells in embryos up to about E7.25, and subsequently a shift to specifically identified mesodermal cells in embryos from E7.5 to E8.5 where they are predominantly present in the LPM. Wt1-expressing cells with an SMC signature emerged slightly earlier and more robustly from E6.5 onwards than Wt1-expressing cells with a MC signature; however, interestingly, there were a few Wt1-expressing cells that have both SMC and MC signatures. We confirmed co-expression of WT1 protein with BRA in the primitive streak/emerging mesoderm, and with the key SM marker TAGLN in the LPM.

These results demonstrate that WT1-expressing cells unexpectedly emerge already in gastrulation-stage embryos. Our findings indicate that the SMCs labelled by Wt1 lineage tracing are not derived from Wt1-expressing MCs arising first and then differentiating into visSMCs and vSMCs. Instead, the slightly earlier onset and prevalence of the Wt1-expressing SMC signature cells during gastrulation appears to provide a direct link to the labelled SMCs many days later in the embryo.

The peritoneal mesothelium emerges due to separation of the LPM into somatic and splanchnic mesoderm, which surround the peritoneal cavity; the splanchnic mesoderm forms the mesodermal component of the intestinal tract (Funayama et al., 1999). Detailed analysis of the emergence of the mesothelium has been provided in chick and quail embryo models (Thomason et al., 2012; Winters et al., 2012), demonstrating that the bilateral splanchnopleure, consisting of splanchnic mesoderm and endoderm, turns into a trilaminar tissue with a mesenchymal layer emerging between an outer mesodermal epithelium and the endoderm. In the chick embryo, the outer epithelium is situated on a basal lamina already at Hamburger–Hamilton stage (HH) 13 (∼19 somites, 48 h), whereas cytokeratin expression, as indicator for epithelium, starts only at HH19 (∼40 somites, 71 h; roughly corresponding to mouse embryos at Theiler stage 17 or E10.5), and WT1 is not expressed at this stage yet (Winters et al., 2012). The outer epithelium is defined as mesothelium only from HH29 (6 days) onwards (chick and quail), when it has organised into a simple squamous epithelium (Thomason et al., 2012; Winters et al., 2012).

In zebrafish, hand2 expression has been presented as a marker for mesothelial progenitor cells emerging from the lateral most region of the LPM at tailbud stage (i.e. at late gastrulation stages) (Prummel et al., 2022). Co-expression with Tagln in the LPM-emerging cells, and a more limited expression of wt1a/b, restricted to the cells of the anterior visceral peritoneum, and absent posteriorly and in the parietal mesothelium, suggested that hand2 was a more encompassing marker for arising mesothelium. Furthermore, hand2 mutant zebrafish had disorganised LPM-derived cells, in particular a lack of parietal mesothelium, indicating an upstream regulatory role of hand2 in mesothelium formation (Prummel et al., 2022).

In the mouse embryo, the peritoneal mesothelium has been shown to arise between E10.5 and E11.5 since at E11.5, a continuous layer of cells expressing cytokeratin and WT1 is present (Wilm et al., 2005). Further information on mesothelial precursor populations and lineage in mouse embryonic development is limited: non-inducible Hand2- or Hand1-based lineage tracing showed labelling of mesothelium over the liver in E14.5 or E15.5 mouse embryos but no temporal analysis (Prummel et al., 2022). A non-inducible Hand1 lineage has also been shown to label the visceral smooth muscle in mouse embryos at E14.5 (Barnes et al., 2010), and Hand1 deletion in the LPM affected the formation of the visceral smooth muscle (Maska et al., 2010).

Our analysis of the correlation of SMC or MC signature gene expression in Hand1- or Hand2-expressing cells during gastrulation, in comparison to that in Wt1-expressing cells, revealed that Hand2 expression is more highly correlated with both SMC and MC signature gene expression than is Hand1, and overlaps with Wt1-expressing cells. In addition, both HAND2 and WT1 are present in an overlapping domain in the posterior LPM in the E8.5 embryo. These data indicate that during gastrulation and in the emerging mesoderm Wt1- and Hand2-expressing cells have similar expression profiles and WT1 and HAND2 similar expression domains.

Visceral smooth muscle arises from around E13 in the mouse embryo based on ACTA2 IF (Walton et al., 2016; Wilm et al., 2005), while the vascular smooth muscle in the intestine and mesentery can be detected from E12.5 during angiogenic remodelling, with mature vessels covered by mural cells, detectable from E15.5 onwards (Hatch and Mukouyama, 2015; Wilm et al., 2005). A complex signalling network including the myocardin/serum response factor (SRF) interaction controls SMC differentiation by regulating expression of cytoskeletal smooth muscle proteins, including ACTA2, TAGLN and smooth muscle myosin heavy chain. Furthermore, PDGFRB signalling regulates vSMC proliferation, migration and differentiation (Donadon and Santoro, 2021). In this study, we have used a signature of the key regulators and markers of SMC fate and differentiation to determine their co-expression with Wt1 already from E7.5 onwards in the LPM. In E12.5 embryos, Wt1 lineage-traced cells in the developing intestinal/mesenteric vascular network co-expressed PDGFRB after activation of the reporter at around E8.5. Importantly, Tam-induced knock out of Wt1 at E7.5 affected vascularisation of the intestine and mesentery in E12.5 mutant embryos compared to controls: the endothelial network was more irregular based on the higher variability in cell density and branching, and the forming mural/vSMCs were less aligned with the vascular networks. These results strongly indicate a role for WT1 in smooth muscle development, originating in cells of the emerging mesoderm. Furthermore, our analysis of Wt1 lineage tracing at earlier stages (E9.5, E10.5) of development suggest that the Wt1 lineage-traced cells originating at E8.5 may migrate into the mesentery and towards the developing vascular network.

Taken together, our data clearly demonstrate that the visSMCs and vSMCs that arise from WT1-expressing cells at around E7.5 or E8.5 are independent in their specification and differentiation from the mesothelial cells of the peritoneum. These can only be tagged after initiation of WT1-controlled recombination from around E9.5 onwards. Our results therefore indicate an early emergence of the visSMC and vSMC lineage linked to WT1 expression in the gastrulating embryo, and an uncoupling from the mesothelial lineage. A detailed mechanistic role of WT1 in both processes remains to be elucidated.

The following compound mutants were used in this study and bred in-house: Wt1^CreERT2/+;Rosa26^lacZ/lacZ [Wt1^{tm2(cre/ERT2)Wtp};Gt(ROSA)26Sor/J] (Soriano, 1999; Zhou et al., 2008), Wt1^CreERT2/+;Rosa26^mTmG/+ [Wt1^{tm2(cre/ERT2)Wtp};Gt(ROSA)26Sor^{tm4(ACTB-tdTomato,-EGFP)Luo}/J] (Muzumdar et al., 2007) and Wt1^CreERT2/fl;Rosa26^mTmG/+ [Wt1^{tm2(cre/ERT2)Wtp};Wt1^fl; Gt(ROSA)26Sor^{tm4(ACTB-tdTomato,-EGFP)Luo}/J] (Martínez-Estrada et al., 2010). CD1 female mice (at least 10 weeks old, Charles River, Harlow, UK) were time-mated with Wt1^CreERT2/+;Rosa26^lacZ/lacZ or Wt1^CreERT2/+;Rosa26^mTmG/+ males, and noon of the day on which a vaginal plug was detected was considered as E0.5. CD1 female mice time-mated by Charles River were ordered in. Mice were housed in individually ventilated cages under a 12-h light/dark cycle, with ad libitum access to standard food and water. All animal experiments were performed under a Home Office licence granted under the UK Animals (Scientific Procedures) Act 1986 and were approved by the University of Liverpool AWERB committee. Experiments reported are in line with the ARRIVE guidelines.

For Wt1 lineage tracing in embryos, pregnant CD1 mice after mating with Wt1^CreERT2/+;Rosa26^lacZ/lacZ or Wt1^CreERT2/+;Rosa26^mTmG/+ male mice were dosed with 100 µg/g body weight of Tam [T5648, Sigma-Aldrich; 40 mg/ml, in corn oil (C8267, Sigma-Aldrich)] via oral gavage once on various days and analysed between E9.5 and E19.5. Pregnant dams were monitored for their wellbeing including daily weight checks. For the Wt1 knockout, pregnant Wt1^fl/fl;Rosa26^mTmG/mTmG female mice mated with Wt1^CreERT2/+;Rosa26^mTmG/mTmG male mice were given 100 µg/g body weight of Tam via oral gavage at E7.5.

Dissected embryos or tissues were fixed in 2% paraformaldehyde (PFA)/0.25% glutaraldehyde in PBS for 1-1.5 h. Whole-mount XGal staining was performed overnight by incubating the embryos/tissues in XGal solution [5 mM potassium hexacyanoferrate(II), 5 mM potassium hexacyanoferrate(III), 1× PBS, 2 mM MgCl₂, 0.02% NP-40, 0.01% sodium deoxycholate, 1 mg/ml XGal (5-bromo-4-chloro-3-indolyl-beta-D-galactopyranoside dissolved in N,N dimethyl formamide) in the dark, followed by overnight post-fixation in 4% PFA (in PBS) at 4°C (Wilm et al., 2005; 2021). All reagents were from Sigma-Aldrich.

The following primary antibodies were used: anti-WT1 rabbit polyclonal (1:200 to 1:500; clone C-19, sc-192, Santa Cruz Biotechnology), anti-WT1 mouse monoclonal (1:50; clone 6F-H2, M3561, Dako), anti-ACTA2 mouse monoclonal (1:100 to 1:200; clone 1A4, A2547, Sigma-Aldrich), anti-ACTA2 rabbit polyclonal (1:200; ab5694, Abcam), anti-Pecam/CD31 rat monoclonal (1:50 to 1:200; 550274, BD Pharmingen), anti-GFP rabbit or goat polyclonal (1:5000; ab6556 or ab6673, Abcam) (Wilm et al., 2021), anti-brachyury goat polyclonal (1:200; PA5-46984, Invitrogen/Thermo Fisher Scientific), anti-PDGFRB rat monoclonal (1:100; clone APB5, eBioscience), anti-TAGLN/SM22 mouse monoclonal (1:100; 60213-1-Ig, Proteintech). The anti-ACTA2 and anti-TAGLN/SM22 antibodies were directly labelled using the Zenon direct labelling kit (Zenon™ Alexa Fluor® 647 Conjugation Kit - Lightning-Link®, Z24107, Invitrogen/Thermo Fisher Scientific) according to the manufacturer's instructions. Secondary antibodies were Alexa fluorophore-coupled (donkey anti-rabbit 488, A-21206; donkey anti-goat 594, A-11058; donkey anti-rat 594, A-21209; Invitrogen/Thermo Fisher Scientific) and were used at a dilution of 1:1000. DAPI (D9542, Sigma-Aldrich) was used at 1:250-1:1000 for nuclear counterstaining.

Tissues were fixed in 4% PFA for 30-90 min, protected in 30% sucrose overnight, placed in Cryomatrix (Thermo Fisher Scientific) and snap-frozen. Frozen sections were generated at 8-10 µm on a Thermo Scientific HM525 NX cryostat. IF staining was performed following standard protocols (Wilm et al., 2005). Two bleaching steps of 20 min each in 3% H₂O₂ in methanol were included for embryos or their tissues carrying the Rosa26^mTmG/+ mutation in order to remove the tdTomato fluorescence (Lua et al., 2015). Subsequently to completed IF, sections were coverslipped with Fluoro-Gel (with Tris buffer; Electron Microscopy Sciences), sealed and stored in the dark at 4°C.

Whole tissues/embryos were fixed in 4% PFA for 60 min on ice and stored in 0.0025% Triton X-100 (Sigma-Aldrich) in PBS at 4°C until used. Tissues/embryos were permeabilised using 0.25% Triton X-100 (PBS) for 60 min on ice. IF staining was performed using previously published protocols (Baillie-Johnson et al., 2015; Beccari et al., 2018; Turner et al., 2017). E7.5 or E8.5 IF-stained embryos were mounted on coverslips within SecureSeal imaging spacers (Grace Bio-Labs) and embedded in RapiClear 1.52 solution (SunJin Lab) overnight. E12.5 IF-stained embryos were mounted on coverslips within 1.0 mm iSpacer (SunJin Lab) in RapiClear 1.52 solution for at least two nights. Samples were stored in the dark at 4°C before imaging.

IF sections were imaged on a Leica DM 2500 upright microscope with a Leica DFC350 FX digital camera and a Leica EL6000 fluorescence light source supported by the Leica Application Suite software package (LAS, version 3 or 4; Leica Microsystems).

XGal-stained and GFP-positive embryos and tissues were imaged using a Leica MZ 16F dissecting microscope equipped with a Leica DFC420 C digital camera and a Leica EL6000 fluorescence light source supported by LAS software.

Imaging of whole-mount embryos was performed using an Andor Dragonfly spinning disc confocal microscope (Andor, Oxford Instruments) mounted on an inverted Leica DMi8 base using either a 10× (0.45NA) or a 25× water-dipping objective (NA0.95). The z-steps for the 10× and 25× objectives were 1.32 μm and 0.35 μm, respectively. DAPI, Alexa 488, 568 and 647 were sequentially excited using 405 nm, 488 nm, 561 nm and 637 nm laser diodes, respectively, and emitted light reflected through 450/50 nm, 525/50 nm, 600/50 nm and 700/25 nm bandpass filters, respectively. An iXon 888 EM-CCD camera was used to collect emitted light, and data were captured using Fusion version 5.5 (Bitplane) with subsequent image analysis performed using either Fiji (Schindelin et al., 2012) or Imaris (Bitplane). Images were manually quantified by circling visible nuclei within a single z-plane to generate regions of interest (ROIs) using the ROI manager tool in Fiji. The average grey values were then measured for the selected regions for each fluorescent channel.

The unfixed trunk area of E9.5 Wt1^CreERT2/+;Rosa26^mTmG/+ embryos dosed with Tam at E8.5 was mounted in 1% low melting point agarose (ROTI Garose, Carl Roth), transferred into a capillary (Size 4, Brand GmbH) and placed in the imaging chamber of a Zeiss light-sheet Z.1 microscope with one 5× detection objective and two 5× illumination objectives supported with Zen software (Carl Zeiss). Imaging was performed using the 488 nm channel to capture GFP-positive cells, and the 561 nm channel to image membrane-bound dTomato to provide tissue context. Images were adjusted to colours using Imaris software and the movie edited to highlight the shape of the selected area by using the surface tool, and by labelling GFP-positive cells with coloured spots to indicate their different locations.

Data analysis from a previously published dataset (Pijuan-Sala et al., 2019) was performed using R (version 4.3.2). Data were accessed through Bioconductor ‘MouseGastrulationData’ package (version 1.16.0), and only processed data were extracted for downstream analysis. Extracted data were normalised using the ‘logNormCounts’ function from the ‘scuttle’ package (version 1.12.0). Normalisation of cell count per sample was carried out by dividing the total number of cells by the sample size and rounding up to the nearest integer. Signature scoring was performed using the AUCell package (version 1.26.0) (Aibar et al., 2017) and threshold was manually selected from suggested thresholds calculated by the package. All plots were created using the ‘ggplot2’ package (version 3.4.3).

For whole-mount CD31 staining analysis, regions of interest (indicated by white boxes) were pre-processed by background subtraction, median filtering and manual thresholding due to photobleaching. Each ROI was then: (1) inputted into 3Dsuite to measure the total volume of the vasculature of respective regions; and (2) 3D-skeletonised and the resulting branching information was analysed using AnalyzeSkeleton.

We acknowledge funding by the BBSRC Alert13, grant number BB/L014947/1, for the Zeiss Z.1 light-sheet microscope. We are also indebted to the University of Liverpool's Centre for Cell Imaging (CCI) facility for provision of state-of-the-art imaging equipment funded through grants awarded to the University of Liverpool's CCI by the MRC (MR/K015931/1) and the BBSRC (BB/M012441/1, BB/R01390X/1), and for excellent technical support, assistance and training. We express our thanks to the Biomedical Services Unit at the University of Liverpool, for expert support in mouse maintenance, breeding and timed mating. We acknowledge Sumaya Dauleh and Thomas P. Wilm for technical assistance.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.204332.reviewer-comments.pdf