GATA6 is a crucial factor for Myocd expression in the visceral smooth muscle cell differentiation program of the murine ureter

In mammals, efficient removal of the urine from the kidneys to the outside relies on the peristaltic activity of a pair of tubular organs: the ureters. Structural basis of the contractile behavior of these straight tubes are smooth muscle cells (SMCs) that form concentric layers in the outer mesenchymal wall. In the mouse, these SMCs arise together with ensheathing fibrocytes in a precisely orchestrated manner from a homogenous pool of mesenchymal progenitors. This pool surrounds the distal aspect of the ureteric bud, an epithelial outgrowth of the nephric duct, at embryonic day (E)11.5. At E12.5, the mesenchymal cells adjacent to the ureteric epithelium (UE) transit from a slender fibroblastic to an enlarged rhomboid shape. At E14.5, they start to express Myocd: the key regulator of SMC differentiation (Wang and Olson, 2004). Cells in the direct vicinity of the UE become devoid of Myocd expression and differentiate from E16.5 onwards into fibrocytes of the lamina propria. The ones further away maintain Myocd expression and activate in a stepwise fashion the expression of different SMC structural genes until E18.5. As a consequence, they form a functional tunica muscularis that engages in peristaltic activity shortly after the onset of urine production in the fetal kidney at E16.5 (Bohnenpoll et al., 2017a; Bohnenpoll and Kispert, 2014).

Failure to activate the SMC differentiation program in the fetal ureter is detrimental to the integrity of the upper urinary system, i.e. the ureters and kidneys (Bohnenpoll et al., 2017c; Trowe et al., 2012). The urine is no longer propelled to the bladder and therefore accumulates in the ureter and renal pelvis, driving dilatation of these structures that finally results in destruction of the renal parenchyma. Importantly, ureter dilatations are frequently observed in human newborns, and, in most cases, initial dilatation recedes over time. However, in some cases the condition remains unresolved, manifesting as severe uro- and nephropathy (Chiodini et al., 2019; Ek et al., 2007; Herthelius et al., 2020).

A combination of embryological and genetic analyses in the mouse uncovered an interconnected system of signals and transcription factor activities that impinges on the proliferation, patterning and subsequent differentiation of the ureteric mesenchyme (UM) into SMCs and fibrocytes (Bohnenpoll and Kispert, 2014). Sonic hedgehog (SHH) originating from the UE prevents apoptosis in the outer UM, and induces proliferation and SMC differentiation of mesenchymal cells adjacent to the UE. In the latter cells, SHH signaling is required for expression of the transcription factor gene Foxf1, which, in turn, induces expression of the gene encoding the signaling molecule BMP4. FOXF1 synergizes in an unknown fashion with BMP4 in activation of Myocd expression and SMC differentiation (Bohnenpoll et al., 2017c; Mamo et al., 2017; Yu et al., 2002). WNTs from the UE activate the T-box transcription factors TBX2 and TBX3 to confine the adventitial fate to the outer UM (Aydogdu et al., 2018; Trowe et al., 2012). This could aid in directing the inner UM to the SMC differentiation pathway. Retinoic acid (RA) emanating from the UM and/or the UE delays SMC differentiation possibly by counteracting WNT signaling (Bohnenpoll et al., 2017b).

The GATA transcription factors are a highly conserved family of zinc-finger proteins that mediate tissue-specific gene expression. In mammals there are six family members that based on structure and function were grouped into two subfamilies (Tremblay et al., 2018). We have previously shown that GATA2, a member of the GATA1/2/3 subfamily acts at least partly as a feedback inhibitor of RA signaling to time the activation of Myocd and, hence, of SMC differentiation in ureter development (Weiss et al., 2019). Here, we set out to study the functional significance of GATA6, a member of the GATA4/5/6 subfamily, in this process. Of note, GATA6 has been implicated in patterning and differentiation of cardiac neural crest cells into vascular SMCs (Kodo, 2009; Lepore et al., 2006; Losa et al., 2017), as well as the maintenance of the quiescent phenotype of these cells (Mano et al., 1999; Perlman et al., 1998; Tremblay et al., 2018). GATA6 is also expressed in bladder SMCs but an in vivo requirement for the differentiation of these or other visceral SMCs has not been reported (Kanematsu et al., 2007; Morrisey et al., 1996).

Here, we demonstrate that Gata6 is expressed in the undifferentiated UM and that its conditional loss from this tissue leads to hydroureternephrosis in fetal and adult mice. Further molecular analyses suggest that GATA6 activates Myocd expression and thereby assures timely activation of the SMC differentiation program in the ureter.

To obtain a detailed profile of Gata6 expression in the development of the upper urinary tract, we performed RNA in situ hybridization on whole-kidney and ureter rudiments as well as on proximal ureter sections of E11.5 to E18.5 wild-type mouse embryos. Strong Gata6 expression was found in the entire UM at E11.5 and E12.5. At E14.5 and E16.5, expression at much reduced levels was confined to the inner region of the UM (Fig. 1A,B). Gata6 expression also occurred in the renal stroma from E12.5 to E18.5 (Fig. S1). GATA6 protein expression in the ureter recapitulated the pattern of Gata6 mRNA in this organ. The protein was confined to the nucleus at all analyzed stages (Fig. 1C).

To determine whether Gata6 expression in the UM depends on one of the signaling systems involved in development of this tissue, we analyzed mutants in which key cellular signaling components were conditionally deleted. We used a Tbx18^cre line, which mediates recombination in the entire UM starting from E11.5 (Airik et al., 2010; Bohnenpoll et al., 2013), and floxed alleles of the unique mediator of SHH signaling, Smo (Long et al., 2001), of canonical WNT signaling, Ctnnb1 (Brault et al., 2001), and of Bmp4 (Kulessa and Hogan, 2002), as previously reported (Bohnenpoll et al., 2017c; Mamo et al., 2017; Trowe et al., 2012). Loss of Smo and Ctnnb1 had no effect on Gata6 expression in the UM at E12.5. In contrast, Gata6 expression was strongly reduced in this region in Tbx18^cre/+;Bmp4^fl/fl embryos (Fig. 1D). Moreover, in E11.5 kidney rudiments cultured for 18 h with 10 µg/ml of the BMP antagonist noggin (Zimmerman et al., 1996), Gata6 expression was very weak in the UM, whereas application of 100 ng/ml BMP4 resulted in enhanced and ectopic Gata6 expression in the renal stroma. Application of 1 µM RA or 1 µM of the pan-RA receptor (RAR) antagonist BMS493 (Chazaud et al., 2003) did not affect Gata6 expression in this culture system (Fig. 1E). We conclude that expression of Gata6 in the undifferentiated UM depends on BMP4 signaling.

Development of Gata6-null embryos arrests during gastrulation as a consequence of defects in extra-embryonic endoderm function (Koutsourakis et al., 1999; Morrisey et al., 1998). We therefore used a conditional gene inactivation approach with a floxed allele of Gata6 and our Tbx18^cre line to analyze Gata6 function specifically in the UM (Airik et al., 2010; Sodhi et al., 2006). The tissue-specific inactivation of Gata6 in Tbx18^cre/+;Gata6^fl/fl (Gata6cKO) ureters was confirmed by severe downregulation of Gata6 mRNA in the UM at E11.5 and of GATA6 protein at E12.5 (Fig. S2). In matings of Tbx18^cre/+;Gata6^fl/+ males with Gata6^fl/fl females, Gata6cKO embryos were recovered at the expected ratio at E18.5 and all other analyzed stages (Fig. S3A). The external appearance of E18.5 mutant embryos was normal, but their urogenital system was frequently (∼80%) and sex-independently characterized by dilatation of the ureter with the proximal aspect being more affected. The severity of this phenotypic defect ranged from weak unilateral to strong bilateral hydroureter (Fig. 2A, Fig. S3B-E). Testis and epididymis were invariably laterally tethered to the kidney (cryptorchidism) and the adrenals were drastically reduced in size (Fig. 2A). The latter phenotype is likely to relate to the reported requirement of Gata6 in adrenogonadal progenitors (Tevosian et al., 2015), in which Tbx18^cre also mediates recombination (Häfner et al., 2015). In Tbx18^cre/+;Gata6^fl/+ littermates, the adrenals appeared unchanged, the testis was descended but weak hydroureter formation occurred sex independently in ∼60% of the cases (Fig. S3B-E), indicating a certain degree of haploinsufficiency.

Histological analysis showed that in E18.5 Gata6cKO embryos, hydroureter was associated with a dilation of the renal pelvis and a reduction of the renal papilla (Fig. 2B). To exclude the possibility that ureter dilatation is caused or contributed by physical obstruction along the ureter and/or of the vesico-ureteric junction, we injected ink into the renal pelvis and observed its flow upon mild hydrostatic pressure. In all mutants, the ink drained smoothly into the bladder, indicating that a functional rather than a physical obstruction underlies the hydroureteronephrosis (Fig. 2C). To further test this hypothesis, we analyzed proximal ureter sections of E18.5 Gata6cKO embryos for presence of SMCs. Markers of this differentiated cell type (ACTA2, TAGLN, Tagln, Tnnt2 and Myh11) were severely downregulated, as was Aldh1a2, a marker for fibrocytes of the inner lamina propria. Markers for adventitial fibrocytes (Dpt and Col1a2) were still expressed in the outer loose mesenchyme, albeit more weakly (Fig. 2D,E). Mesenchymal defects did not affect cyto-differentiation of the urothelium, as revealed by normal expression of KRT5, ΔNP63 and UPK1B, which combinatorially mark B cells (KRT5⁺ΔNP63⁺UPK1B⁻), I cells (KRT5⁻ΔNP63⁺UPK1B⁺) and S cells (KRT5⁻ΔNP63⁻UPK1B⁺) (Bohnenpoll et al., 2017a) (Fig. 2F). As increased hydrostatic pressure affects cyto-differentiation in the mesenchymal compartment, we also analyzed weakly dilated proximal ureters and non-dilated distal ureters. Expression of SMC markers was also reduced in these settings but the degree of reduction varied from very strong (Tnnt2, ACTA2 and TAGLN) to moderate (Tagln and Myh11). Fibrocyte and epithelial cyto-differentiation was unaffected (Figs S4,S5). We therefore conclude that SMC differentiation is reduced in E18.5 Gata6cKO ureters and that pressure-mediated dilatation aggravates the effect.

As hydroureter formation is frequently observed in newborn babies but often resolves for unknown reasons (Dudley et al., 1997), we wondered whether the dilative nephro-/uropathy in Gata6cKO mice persists into postnatal (P) stages. To answer this question, we analyzed urogenital systems of Gata6cKO mice at P21, when the animals showed a normal external appearance and behavior. We found undescended testes, a strongly dilated ureter and pelvis, absence of the renal papilla, and a severe reduction of the renal parenchyma (Fig. 3A-C). Expression of SMC markers was strongly reduced, whereas urothelial differentiation appeared unaffected (Fig. 3D-J). This shows that hydroureternephrosis does not resolve after birth in Gata6cKO animals.

To define both the onset and progression of urogenital malformations in Gata6cKO embryos, we analyzed urogenital systems at E14.5 to E16.5, i.e. shortly before and after onset of urine production in the kidney. At the morphological level, the mutant was distinguished by adrenal hypoplasia starting from E14.5, and by hydroureter formation and absent testicular descent at E16.5 (Fig. 4A). Histological staining of proximal ureter sections revealed a less condensed inner mesenchymal layer compared with the control at E15.5, and a strongly dilated ureter at E16.5 (Fig. 4B). In control embryos, expression of Myh11 started robustly at E14.5, and expression of Tagln, Tnnt2 and ACTA2 started at E15.5 in the UM. In Gata6cKO ureters, expression of SMC structural genes/proteins occurred at dramatically reduced levels at all stages analyzed (Fig. 4C,D). In contrast, markers for the lamina propria (Aldh1a2) and the tunica adventitia (Dpt, Fbln2 and Col1a2) were either not expressed or were unchanged (Fig. S6). Expression of the epithelial cyto-differentiation markers ΔNP63, UPK1B and KRT5 occurred normally, but expression of ΔNP63 (at E14.5 and E15.5) and of UPK1B (E15.5) was weaker, and stratification was delayed in Gata6cKO ureters (Fig. 4E).

The TUNEL assay did not detect apoptotic bodies in the mutant UM at E12.5 and E14.5 (Fig. S7A). Moreover, cell proliferation, as studied by the BrdU incorporation assay, was not changed in the mesenchymal and epithelial compartments of the mutant ureter at either stage (Fig. S7B,C; Table S1). Hence, Gata6 is crucial for the initiation of the SMC program but does not affect survival, proliferation and patterning of the UM.

Although SMC differentiation was dramatically reduced in Gata6cKO ureters at E14.5 and E15.5, the program may be compromised from E16.5 onwards by pressure-induced ureter dilatation. To analyze ureter development in the absence of urine load, we explanted ureters at E14.5, i.e. before urine formation, and cultured them for 8 days in medium with 10% FCS, monitoring daily for morphological changes and peristaltic activity. In the control, peristaltic movements started after 2 days and reached their frequency peak after 6 days. In Gata6cKO ureters, contractions started very weakly only after 4 days, exhibited approximately half of the wild-type frequency after 6 days, but reached the control levels after 8 days of culture (Fig. 5A,B; Table S2). The contraction intensity was strongly decreased at day 4 at the proximal, medial and distal positions analyzed, but reached control levels during the following days, except proximally, where it continued to be significantly reduced (Fig. 5C; Table S3). After 6 days of culture, expression of Myh11 appeared normal; Tagln/TAGLN and ACTA2 expression was weakly reduced and Tnnt2 was strongly reduced in the proximal ureter region (Fig. 5D). When explants of E14.5 ureters with kidneys were cultured for 8 days in serum-free medium, we also observed a delayed onset and a subsequent partial recovery of the peristaltic activity of Gata6cKO ureters, indicating that neither the presence of serum nor absence of kidney accounts for either aspect of the changed peristaltic profile (Fig. S8; Table S4).

Dilated ureters explanted from E18.5 Gata6cKO embryos regained frequent peristaltic contractions after 2 to 4 days in culture (Fig. 5E,F; Table S5). Contraction intensity was strongly decreased at days 0 and 2. After 4 and 6 days, control levels were observed medially; at the proximal and distal position, the contraction intensity remained strongly and weakly reduced, respectively (Fig. 5G, Table S6). At day 6 of culture, expression of SMC markers appeared unaffected (Myh11), weakly (Tagln/TAGLN and ACTA2) and strongly reduced (Tnnt2) (Fig. 5H).

Components of the ureteric excitation/conduction system were unchanged in E18.5 Gata6cKO embryos, as revealed by normal expression of HCN3, a marker for pelvic pacemaker cells (Hurtado et al., 2010), and of KIT, a marker for interstitial Cajal-like cells (David et al., 2005) (Fig. S9).

These data show that expression of most SMC structural genes/proteins is delayed but not permanently lost in Gata6cKO ureters, and that SMC differentiation and peristaltic activity can be partly recovered in ex vivo culture conditions even after a dilatation has occurred.

To identify in an unbiased fashion the molecular changes that may cause delayed and reduced SMC differentiation in Gata6cKO ureters, we performed microarray-based gene expression profiling of E14.5 Gata6cKO and control ureters. Using an intensity threshold of 100 and fold changes of at least 2 in the two individual arrays, we detected 61 genes with reduced and 57 genes with increased expression in mutant ureters (Fig. 6A-C; Tables S7, S8).

Functional annotation using the DAVID software tool (Huang da et al., 2009) revealed an enrichment of gene ontology (GO) terms related to ‘neuron’ in the pool of upregulated genes (Table S9) but RNA in situ hybridization did not detect changes of expression of selected candidate genes (Cartpt, Nefm, Phox2b and Hand2) in E14.5 Gata6cKO ureters (Fig. S10A). Interestingly, two genes relating to RA synthesis, Aldh1a3 (+5.5) and Aldh1a2 (+2.2) were the most upregulated genes. Targets of RA signaling in the UM (Bohnenpoll et al., 2017b), including Wt1 (+2.4) and Ecm1 (+1.6), were also increased (Fig. 6B, Table S7). RNA in situ hybridization confirmed increased expression of Aldh1a2 and Wt1 in the outer UM, of Aldh1a3 in the UE, and of Ecm1 in the inner UM of Gata6cKO ureters at E14.5. However, the direct target of RA signaling, Rarb (Mendelsohn et al., 1991), appeared unchanged in the UM (Fig. 6D).

Functional annotation revealed an enrichment of GO terms related to ‘epithelial differentiation’ in the pool of downregulated genes (Table S10). Genes associated with these terms included the S-cell markers Upk1b (-4.3), Upk2 (-3.2) and Upk1a (-3.0), the regulator of S-cell differentiation, Grhl3 (-2.3) (Yu et al., 2009), and the regulator of epithelial stratification, Trp63 (-2.5) (Weiss et al., 2013). In situ hybridization was not sensitive enough to detect expression changes of these candidate genes in mutant ureters (Fig. S10B). The list of downregulated genes also included Car3 (-3.2) and the WNT antagonist Shisa2 (-3.1), two genes previously shown to be expressed in the UM (Airik et al., 2010; Aydogdu et al., 2018), and most notably Myocd (-2.8): the master regulator of SMC differentiation (Wang and Olson, 2004). RNA in situ hybridization confirmed reduced expression of Car3 and Shisa2 in the UM of E14.5 Gata6cKO embryos; Myocd expression was almost absent (Fig. 6E).

Myocd expression and SMC differentiation in the developing ureter depends on a number of signaling pathways and transcription factor activities (Bohnenpoll and Kispert, 2014). Our microarray and in situ hybridization analyses did not detect changes in expression of Shh or of the direct target genes of SHH signaling, Ptch1 and Gli1 (Bohnenpoll et al., 2017c; Ingham and McMahon, 2001); of Wnt7b and Wnt9b; of Axin2, a direct target of this pathway (Jho et al., 2002; Trowe et al., 2012); of Bmp4 (Bohnenpoll et al., 2017c) and the direct targets of its activity, Id2 and Id4 (Hollnagel et al., 1999; Mamo et al., 2017); and of the transcription factor genes Foxf1 (Bohnenpoll et al., 2017c), Gata2 (Weiss et al., 2019), Sox9 (Airik et al., 2010), Tbx18 (Airik et al., 2006), Tcf21 (Airik et al., 2006), Tshz3 (Caubit et al., 2008), Tbx2/TBX2 and Tbx3/TBX3 (Aydogdu et al., 2018) in Gata6cKO ureters at E14.5 (Figs S11,S12).

RT-qPCR analysis confirmed that Myocd expression was strongly reduced in E14.5 Gata6cKO ureters, whereas expression of Foxf1 (an activator of Myocd expression), of the WNT target gene Axin2, of the BMP4 target Id2 and of the mesenchymal RA target Rarb was unchanged (Fig. 6F, Table S11). Expression of Myocd remained strongly reduced at E15.5 and E16.5. In E18.5 ureters and in E18.5 ureter explants cultured for 6 days Myocd expression was weakly reduced, indicating that Myocd expression is regained to a significant degree after E16.5 (Fig. 6G). Hence, Gata6 is required for activation of Myocd expression at E14.5, independently of transcription factors and signaling activities previously implicated in the regulation of this gene.

Although our assays did not detect increased expression of the direct RA target gene Rarb in the UM, increased expression of RA synthesizing enzymes and some RA-dependent genes may point to a functional implication of enhanced RA signaling in the delayed onset of SMC differentiation, as previously reported (Bohnenpoll et al., 2017b; Weiss et al., 2019). We therefore tested whether reduction of RA signaling ameliorates the peristaltic changes of Gata6cKO ureters. Treatment of control E13.5 ureter explants with the pan-RAR antagonist BMS493 (1 μM) did not affect the onset of contractions but lowered the contraction frequency from day 4 onwards. In Gata6cKO ureters, BMS493 treatment further delayed the peristaltic onset and reduced the contraction frequency (Fig. S13; Table S12). We conclude that increased RA signaling does not contribute in a major fashion to the delayed onset of SMC differentiation and peristaltic activity in Gata6cKO ureters.

We recently reported that, in mice with conditional loss of Gata2 in the UM, SMC differentiation was delayed and RA signaling was increased (Weiss et al., 2019). Given the phenotypic similarities, we wondered whether GATA2 and GATA6 would cooperate in ureteric SMC differentiation. We tested this by generating mice that were compound heterozygous for Gata2 and Gata6. Urogenital systems of Tbx18^cre/+;Gata2^fl/+;Gata6^fl/+ embryos neither exhibited enhanced ureter dilatation nor presented with cryptorchidism compared with Tbx18^cre/+;Gata2^fl/+ and Tbx18^cre/+;Gata6^fl/+ embryos, indicating that Gata2 and Gata6 control different molecular programs (Fig. S14).

To further analyze the role of GATA6 in Myocd activation, we employed a gain-of-function approach in a cellular system. Transfection of a Gata6 overexpression construct into NIH3T3 cells led to a marginal activation of Myocd transcription. In contrast, Foxf1, a known regulator of SMC differentiation in the UM (Bohnenpoll et al., 2017c), dose dependently and strongly activated Myocd expression in these cells. Intriguingly, co-transfection of Gata6 and Foxf1 overexpression constructs activated Myocd expression to levels clearly exceeding those achieved with individual transfections (Fig. 7A, Table S13).

Given the cooperativity of GATA6 and FOXF1 in Myocd activation in this cellular system, we wondered whether forced expression of FOXF1 would rescue the peristaltic defects of Gata6-deficient ureters. For this, we combined an inducible cre driver line (Axin2^creERT2) with the floxed allele of Gata6 (Gata6^fl) and a Foxf1 misexpression allele (Hprt^Foxf1) (Bohnenpoll et al., 2017c), and monitored the peristaltic activity of E12.5 ureter explants after administration of 4-hydroxytamoxifen in vitro. Axin2^creERT2/+;Gata6^fl/fl ureters showed a delayed onset of peristaltic activity and reduced contraction frequency compared with the control. Misexpression of Foxf1 led to premature activation of peristaltic activity irrespective whether Gata6 was present (in Axin2^creERT2/+;Gata6^fl/+;Hprt^Foxf1/y ureters) or lost (in Axin2^creERT2/+;Gata6^fl/fl;Hprt^Foxf1/y ureters) (Fig. 7B, Table S14). We conclude that FOXF1 and GATA6 both impinge on Myocd activation but that enhanced FOXF1 activity can override the loss of Gata6 (Fig. 7C).

Here, we have identified GATA6 as a novel regulator of visceral SMC differentiation in the ureter. GATA6 acts downstream of BMP4 and cooperates with FOXF1 in Myocd activation and, thus, SMC differentiation, possibly as pioneer (GATA6) and lineage-determining factor (FOXF1) (Fig. 7C). Our work confirms that a delayed and reduced onset of SMC differentiation results in ureter dilatation at birth, and that relief from the resulting hydrostatic pressure may aid in regaining ureter peristalsis.

Previous work described expression of Gata6 in various progenitor populations during murine development, including the visceral endoderm, the (pre-)cardiac mesoderm, and the neural crest and the endoderm of the primitive gut tube (including the bronchial tree), and also in mature SMCs of the aorta, large arteries and the bladder. These studies also reported Gata6 expression in the urogenital ridge at E13.5 but did not further analyze expression during urinary tract development (Freyer et al., 2015; Morrisey et al., 1996; Nemer and Nemer, 2003). Our work characterizes the undifferentiated UM as an additional expression domain of Gata6/GATA6 in mouse development. We found downregulation of Gata6/GATA6 expression with onset of mesenchymal cyto-differentiation in the ureter, which contrasts with the pattern in the adjacent bladder primordium, where Gata6 expression is maintained into adult stages (Freyer et al., 2015; Morrisey et al., 1996).

The conditional loss of Gata6 in the UM did not affect survival, proliferation and patterning of these progenitors but led to a failure to timely activate the SMC program in the fetal ureter. This finding is in line with previous reports that Gata6 is crucial for early differentiation of progenitors of various endodermal, mesodermal and ectodermal sources (Morrisey et al., 1998; Tevosian et al., 2015; Zhao et al., 2008, 2005) (for a review see (Tremblay et al., 2018)).

Our study stresses the significance of Gata6 and its related family members (Gata4 and Gata5) as regulators of muscle cell differentiation. Although the role of Gata6 (in combination with Gata4) in early differentiation of cardiomyocytes has been appreciated for long (Zhao et al., 2008), the role in vascular, and especially in visceral, SMC differentiation has remained less clear. Some in vitro studies initially suggested that GATA6 induces and maintains the contractile phenotype of vascular SMCs (Abe et al., 2003; Mano et al., 1999; Wada et al., 2002), whereas others questioned such a role (Lepore et al., 2005; Yin and Herring, 2005). However, recent in vivo loss- and gain-of-function studies provided strong evidence that Gata6 is both required and sufficient for the differentiation of cardiac neural crest cells into SMCs that surround the large vessels of the cardiac outflow tract (Losa et al., 2017).

Knockdown of endogenous GATA6 in primary human bladder SMCs led to decreased mRNA levels of some SMC structural genes, suggesting a role in maintaining the differentiated phenotype of these visceral SMCs (Kanematsu et al., 2007). However, to our knowledge, an in vivo requirement for Gata6 in the establishment or maintenance of the SMC phenotype in the bladder has not been reported.

It is important to note that Gata6 expression has not been observed during visceral SMC development of the respiratory system, the urethra and the gastrointestinal tract (Morrisey et al., 1996). Moreover, a role of Gata6 has not been reported for vascular SMCs that do not derive from neural crest cells, such as epicardium-derived coronary SMCs. We cannot exclude the possibility that low levels of expression of Gata6 or of other Gata family members have escaped detection and/or that redundancy of several Gata genes has precluded functional insights. However, it is also conceivable that Gata6 acts only in some of the SMC differentiation programs during embryonic development. This is in line with the finding that both vascular and visceral SMCs arise from a multitude of progenitors, and that diverse signals from adjacent epithelial and endothelial primordia (and also from within the SMC progenitors) are implicated in activation of the SMC differentiation program (Donadon and Santoro, 2021; Mack, 2011).

Our in vivo and ex vivo experiments have shown that BMP4 signaling is both required and sufficient for Gata6 expression in the UM. Given the findings that Gata4 and Gata6 are co-expressed in many mesodermal and endodermal progenitors, that BMP4 is upstream of Gata4 in the precardiac mesoderm (Schultheiss et al., 1997), endoderm (Rossi et al., 2001) and lateral plate mesoderm (Rojas et al., 2005), and that neural crest cell induction requires BMP4 signaling (Baker and Bronner-Fraser, 1997), it is tempting to speculate that BMP4 signaling-regulated expression of Gata6 (and/or Gata4) defines a crucial axis in the differentiation of both cardiomyocytes and a subset of vascular and visceral SMCs.

Our molecular characterization of Gata6cKO ureters revealed a dramatic reduction of Myocd expression in the fetal ureter. MYOCD acts a transcriptional co-activator that complexes with the DNA-binding protein serum response factor (SRF) in the activation of genes that harbor binding sites for SRF, so called CArG boxes in their promoters (Norman et al., 1988; Sun et al., 2006; Wang et al., 2001; Wang and Olson, 2004). Given the fact that Myocd is required for SMC differentiation, downregulation of Myocd expression is the likely cause for SMC differentiation and peristalsis defects in Gata6cKO ureters.

As Myocd is sufficient to induce SMC differentiation when ectopically expressed (van Tuyn et al., 2005; Wang et al., 2003), temporal and spatial control of Myocd expression underlies the regionalized programs of visceral and vascular SMC differentiation. As mentioned above, activation of Myocd and, hence, of the SMC program occurs in response to a multitude of signals that are released from endothelial and epithelial primordia, as well as in response to signals secreted from the SMC progenitors (Donadon and Santoro, 2021; Mack, 2011). In the murine ureter, SHH and WNTs have been characterized as epithelial signals from the ureteric bud that maintain UM proliferation and induce Myocd expression and SMC differentiation. SMO-dependent SHH signaling and CTNNB1-dependent WNT signaling induce and maintain, respectively, BMP4 expression in the UM. In turn, BMP4 is required for UM proliferation, Myocd expression and SMC differentiation (Bohnenpoll et al., 2017c; Mamo et al., 2017; Trowe et al., 2012). Importantly, we neither detected changes in expression of the ligand genes Shh, Wnt and Bmp4 nor alterations in their signaling activities in the UM of Gata6cKO ureters. Furthermore, we found unaltered expression of transcription factors, including Foxf1, that impinge on Myocd expression downstream of the activity of these signaling pathways in the UM. These data show that all known positive molecular inputs on Myocd expression in the UM are unaffected by loss of Gata6. This is notable as several studies suggested that Gata6 acts as a regulator of Bmp4 expression in some developmental settings (Nemer and Nemer, 2003; Peterkin et al., 2003; Whissell et al., 2014; Zeng and Childs, 2012).

Our Gata6cKO analysis detected increased expression of RA synthesizing enzymes in the early ureter, and slightly increased expression of some RA-dependent genes in the UM. Pharmacological inhibition of RA signaling did not alleviate the SMC defects but increased them, making it unlikely that RA, which is known to inhibit SMC differentiation in the UM (Bohnenpoll et al., 2017b), causes or contributes to the lack of Myocd activation in Gata6cKO UM. Increased RA signaling may, however, contribute to the slight delay in epithelial stratification and differentiation, and/or occurrence of a neuronal expression profile in the mutant ureter, as suggested by some studies (Bohnenpoll et al., 2017b; Lakard et al., 2007; Sidell and Horn, 1985).

Based on these results, we posit that GATA6 presents a novel and direct input on Myocd activation in the UM. This assumption receives indirect but strong support from a recent report that Gata6 is required for Myocd expression in cardiac neural crest cells, and that it occupies a binding site in the Myocd locus (Losa et al., 2017).

It is important to note that GATA6 expression occurs in the undifferentiated UM from E11.5 to E14.5, i.e. much earlier than that of Myocd, which can be detected from E14.5 onwards. Moreover, GATA6 was not sufficient to induce Myocd expression in NIH3T3 cells but enhanced the activity of FOXF1 in doing so. This parallels our earlier finding that BMP4 signaling, i.e. the activator of Gata6 expression, is not sufficient to activate the SMC program in the ureter but synergizes with FOXF1 (Bohnenpoll et al., 2017c). Given recent reports on the molecular function of GATA6 as a chromatin remodeling protein and pioneer factor in cardiac and endoderm development (Heslop et al., 2021; Sharma et al., 2020), and the occurrence of a binding site in the Myocd locus in neural crest cells (Losa et al., 2017), we hypothesize that GATA6 opens chromatin in and around the Myocd locus in the UM to ease subsequent binding of FOXF1 and to ensure precise temporal activation of Myocd transcription.

FOXF1 was not only sufficient to induce Myocd expression in NIH3T3 cells, it also led to premature activation of the SMC program upon forced overexpression in the UM. As the latter activity was independent of Gata6, it shows that FOXF1 acts as the dominant lineage-determining factor. Occurrence of Myocd expression and SMC differentiation in Gata6cKO ureters at E18.5 and at later stages in ex vivo cultures, may indicate that FOXF1 expression had sufficiently increased to open the Myocd locus and activate its transcription.

Interestingly, with the exception of Tnnt2, all of the SMC structural genes evaluated returned to normal expression levels in extended ex vivo cultures of Gata6cKO explants. As GATA6 binds and transactivates a cardiac-specific enhancer element in the Tnnc1 gene (Morrisey et al., 1996), it is conceivable that Tnnt2 presents an additional direct target of GATA6 activity in the UM.

We have recently reported that Gata2 is expressed in the UM and that its conditional loss leads to a delayed onset of SMC differentiation, ureter dilatation in fetal life and hydroureternephrosis after birth (Weiss et al., 2019). Given the co-expression of Gata6 with Gata2 in the UM, and the strong similarity of phenotypic changes of the upper urinary tract upon individual loss in the UM, it appeared possible that Gata2 and Gata6 act in a common pathway for Myocd activation. However, our analyses showed that the two genes are differentially regulated (Gata6 depends on BMP4, Gata2 on RA signaling), and that the loss of the genes in the UM leads to different molecular changes at E14.5. Although GATA2, at least partially, acts as a feed-back inhibitor for RA signaling (Weiss et al., 2019), deregulation of RA signaling is irrelevant for the SMC defects in Gata6cKO ureters, and direct regulation of Myocd seems likely. Moreover, we did not find genetic interaction when combining loss-of-function alleles of both genes. Last but not least, GATA2 and GATA6 belong to different subfamilies, the members of which preferentially interact with themselves (Tremblay et al., 2018).

Similar to Gata2-deficient ureters, Gata6cKO ureters had a normal excitation and/or conduction machinery, and regained considerable peristaltic performance when relieved from urinary pressure in an ex vivo culture setting. Although this confirms that increased hydrostatic pressure exacerbates the SMC defects in vivo, it highlights that a temporary artificial bypass in vivo may provide an opportunity for the mesenchymal coat to (re-)differentiate contractile SMCs and regain peristaltic activity. Irrespective of future therapeutic options for congenital forms of hydroureter formation in humans, Gata6 and Gata2 are candidates to include in mutational screens for monogenetic causes of this disease entity.

Mice with conditional inactivation of Gata6 (Gata6^tm2.1Sad, synonym Gata6^fl/fl) (Sodhi et al., 2006) were obtained from Joerg Heineke (previously Medizinische Hochschule Hannover, now Medical Faculty Mannheim, Germany) after permission from Steve Duncan (Medical University of South Carolina, Charleston, USA); mice with conditional inactivation of Bmp4 (Bmp4^tm3Blh, synonym Bmp4^fl) (Kulessa and Hogan, 2002) were a gift from Rolf Zeller (University of Basel, Switzerland) after permission from Brigid Hogan (Duke University Medical Center, Durham, NC, USA). Rolf Kemler (Max-Planck-Institute for Immunobiology and Epigenetics, Freiburg, Germany) provided mice with conditional inactivation of β-catenin (Ctnnb1) (Ctnnb1^tm2Kem, synonym Ctnnb1^fl) (Brault et al., 2001). Mice with conditional inactivation of smoothened (Smo) (Smo^tm2Amc, synonyms Smo^fl and JAX 004526) (Long et al., 2001) were purchased from the Jackson Laboratory, as were mice with a creERT2 knock-in allele of Axin2 [Axin2^{tm1(cre/ERT2)Rnu}, synonyms Axin2^creERT2 and JAX #018867] (van Amerongen et al., 2012). The cre driver line Tbx18^tm4(cre)Akis (synonym Tbx18^cre) and the Foxf1 misexpression allele [Hprt^{tm1(CAG−Foxf1,−EGFP)Akis}, synonym: Hprt^Foxf1] were previously generated in house (Airik et al., 2010; Bohnenpoll et al., 2017c).

All mouse lines were maintained on an NMRI outbred background. NMRI wild-type embryos were used for expression analysis. Embryos for phenotypic analyses were derived from matings of males double heterozygous for Tbx18^cre line and the floxed loss-of-function allele with females homozygous for the same floxed loss-of-function allele. Littermates without the Tbx18^cre allele were used as controls. For timed pregnancies, vaginal plugs were checked on the morning after mating; noon was designated to be E0.5. Embryos and urogenital systems were dissected in PBS. Specimens were fixed in 4% PFA/PBS, transferred to methanol and stored at −20°C prior to further processing. PCR genotyping was performed on genomic DNA prepared from yolk sacs or liver biopsies. The experiments were approved by the local Institutional Animal Care and Research Advisory Committee and permitted by the Lower Saxony State Office for Consumer Protection and Food Safety (reference number 42500/1H).

Ureters (experiments in Figs 5 and 7, Fig. S13) and ureters with kidneys (experiment in Fig. S8) were dissected from the embryo, explanted on 0.4 µm polyester membrane Transwell supports (3450, Corning) and cultured at the air-liquid interface with DMEM/F12 supplemented with 1% of concentrated stocks of penicillin/streptomycin, sodium pyruvate, glutamax and non-essential amino acids (21331020, 15140122, 11360070, 35050038 and 11140035, Thermo Fisher Scientific) with 10% FCS (S0115, Biochrom) (experiments in Fig. 5) or without FCS (experiment Fig. S8), or with IST-G (insulin-transferrin-selenium, 41400045, Thermo Fisher Scientific) (experiments in Fig. 7). To induce recombination with the Axin2^creERT2 line, 4-hydroxytamoxifen (H7904, Sigma-Aldrich) was used at a final concentration of 500 nM. Recombinant mouse BMP4 (5020-BP, R&D Systems) was dissolved in 4 mM HCl to a final concentration of 100 ng/ml. The BMP4 inhibitor noggin (ZO3205, Biozol/GenScript) was dissolved in water to 10 µg/ml. BMS493 (3509, Tocris BioScience) or RA (0695, Tocris BioScience) was dissolved in DMSO and added to the medium at a final concentration of 1 µM. The culture medium was replaced every second day. Contralateral kidneys were used as controls. For video documentation, the cultures were acclimatized to room conditions and then imaged in a bright-field channel for 1 min at a rate of 5 frames per second. Measurement of the contractions per minute and the peristaltic intensity was carried out either manually or via computational Fiji Multi-Kymograph analysis (Schindelin et al., 2012). Therefore, the length of the ureter was subdivided into 25 (proximal level), 50 (medial level) or 75 (distal level) percentiles. One contraction was set to 100 frames representing 20 s in real time. Kymograph grey values were divided by the maximum grey value and ratios were plotted with Microsoft Excel.

Embryos and urogenital systems were embedded in paraffin and sectioned at 5 µm. Hematoxylin and Eosin staining was performed according to standard procedures. Ink injection experiments were performed as described previously (Airik et al., 2006).

Section in situ hybridization on 10 µm paraffin wax-embedded sections was performed as previously described (Moorman et al., 2001). Whole mount in situ hybridization of whole-kidney rudiments or kidney cultures followed a standard procedure with digoxigenin-labeled antisense riboprobes (Wilkinson and Nieto, 1993).

Total RNA was isolated from three pools of 10 ureters each of E14.5 control and Gata6cKO embryos using TRIzol (15596-018, Thermo Fisher Scientific). cDNA was synthesized from 2.5 µg total RNA applying RevertAid H Minus reverse transcriptase (EP0452, Thermo Fisher Scientific) as described previously (Thiesler et al., 2021). The NCBI tool Primer3 version4.1 was used to design specific primers (Table S15). Reverse transcription-quantitative polymerase chain reaction (RT-qPCR) was performed in 10 µl 1:2 diluted BIO SyGreen Lo-ROX mix (PCR Biosystems) with 400 nM primers and 1 ng/µl cDNA applying a QuantStudio3 PCR system fluorometric thermal cycler (Thermo Fisher Scientific). Each of the three biological replicates represents the average of four technical replicates. Data were processed by QuantStudio data analysis software (version1.5.1, Thermo Fisher Scientific) using the comparative threshold cycle (ΔΔC_T) method with Gapdh (Werneburg et al., 2015) and Ppia as reference genes.

Two pools of 10 E14.5 ureters, from male and female control embryos and from Gata6cKO embryos, were collected. Total RNA was extracted using peqGOLD RNApure (product 732-3312, order 30-1010, PeqLab Biotechnologie) and subsequently sent to the Research Core Unit Transcriptomics of Hannover Medical School where RNA was Cy3 labeled and hybridized to Agilent Whole Mouse Genome Oligo v2 (4×44 K) microarrays (G4846A, Agilent Technologies). To identify differentially expressed genes, normalized expression data were filtered using Excel based on an intensity threshold of 100 and a more than 1.9-fold change in all pools. Microarray data have been deposited in GEO under accession number GSE174614.

For immunofluorescent staining, 5 µm paraffin wax-embedded sections were stained as previously described (Bohnenpoll et al., 2017a). Primary antibodies and dilutions were as follows: goat-anti-GATA6 (1:50, AF1700, R&D Systems), rabbit-anti-GATA6 (1:200, 5851, Cell Signaling Technology), mouse-anti-ACTA2 (1:500, A5228, Sigma-Aldrich), rabbit-anti-TAGLN (1:200, ab14106, Abcam), rabbit-anti-ΔNP63 (1:250, 619001, BioLegend), rabbit-anti-KRT5 (1:250, PRB-160P, BioLegend), mouse-anti-UPK1B (1:200, WH0007348 M, Sigma-Aldrich), rabbit-anti-TBX2 (1:200, 07-318, Merck Millipore), goat-anti-TBX3 (1:500, sc-31656, Santa Cruz Biotechnology), rabbit-anti-HCN3 (1:200, ab5506, Abcam) and rabbit-anti-KIT (1:200, ab5506, Abcam). (The last two antibodies are no longer available.) Secondary antibodies and dilutions were as follows: goat-anti-rabbit Alexa488 (1:250, A11034, Invitrogen), donkey-anti-mouse Alexa488 (1:250, A21202, Invitrogen), goat-anti-mouse Alexa555 (1:500, A21422, Invitrogen), biotinylated goat-anti-rabbit (1:200, 111065033, Dianova) and biotinylated donkey-anti-goat (1:200, 705-065-003, Dianova), Biotin-SP-conjugated Fab fragment goat anti-rabbit IgG (H+L) (1:400, 111-067-003, Jackson ImmunoResearch Labs, ordered via Dianova) and streptavidin HRP (1:500, 434323, Thermo Fisher Scientific). For amplification of the antibodies, the TSA Tetramethylrhodamine Amplification Kit (1:100, NEL701001KT, PerkinElmer) was used. After antigen retrieval (H-3300, Antigen Unmasking Solution, Vector Laboratories; 15 min, 100°C), labeling of primary antibodies was performed overnight at 4°C. Labeling of secondary antibodies was performed for 1 h at room temperature in blocking solution.

Cell proliferation rates were investigated by the detection of incorporated BrdU on 5 µm paraffin sections according to published protocols (Bussen et al., 2004). For the analysis, 12 sections per specimen (n=3) and genotype of the proximal ureter were stained. The BrdU-labeling index was defined as the number of BrdU-positive nuclei relative to the total number of nuclei, as detected by counterstaining of nuclei with 4′,6-diamidino-2-phenylindole (DAPI, 6335.1, Carl Roth) in histologically defined compartments of the ureter. Apoptosis in tissues was assessed by the TUNEL assay using the ApopTag Plus Fluorescein In Situ Apoptosis Detection Kit (S7111, Merck) on 5 μm paraffin wax-embedded sections.

The Gata6 open reading frame was amplified from a pCI vector (a kind gift from Joerg Heineke) and subcloned as a NcoI-NdeI fragment into a pSP64 in vitro transcription vector containing globin leader and trailer sequences and a MYC or HA tag, respectively (Farin et al., 2007). Tagged Gata6 was then shuttled as a HindIII-EcoRI fragment into pcDNA3 for eukaryotic expression. Foxf1 pCMV6-entry expression vector was obtained from Origene (MR225056).

NIH3T3 cells were seeded on six-well plates and grown to 70-80% confluency. For each experiment, three independent six-well plates were transfected with increasing amount of plasmids with Lipofectamine 2000 Transfection Reagent (11668027, ThermoFisher) following the manufacturer's recommendations. After 2 days, cells were harvested, total RNA was extracted with peqGOLD RNApure (30-1020, PeqLab) and first-strand cDNA synthesis was performed with RevertAid reverse transcriptase (EP0441, ThermoFisher). To quantify Myocd expression in NIH3T3 cells, a semi-quantitative PCR was established with Myocd-specific primers (forward, TGAGACTCACCATGACACTCC; reverse, TGGCGGTATTAAGCCTTGGT), and adjustment of the number of cycles to the mid-logarithmic phase. Annealing was performed at 56°C for 20 s, extension at 72°C for 20 s with 25-30 cycles. Quantification was performed with ImageJ (Schneider et al., 2012). Gapdh was used for normalization (Tanimizu and Miyajima, 2004).

For statistical analysis, the two-tailed Student's t-test was performed and the data were expressed as mean±s.d. For relative analyses, wild-type values were set to 1. Differences were considered significant at *P<0.05, highly significant at **P≤0.005 and extremely significant at ***P≤ 0.0005.

Sections and organ cultures were photographed using a DM5000 microscope (Leica Camera) with Leica DFC300FX digital camera or a Leica DM6000 microscope with Leica DFC350FX digital camera. Urogenital systems were documented using a Leica M420 microscope with a Fujix HC-300Z digital camera (Fujifilm Holdings). Whole-mount in situ hybridization of cultured kidney rudiments were documented using a Leica Z6 APO microscope with a Leica DFC420C digital camera. Figures were prepared with Adobe Photoshop CS3 and CS4.

We thank Rolf Kemler, Rolf Zeller and Brigid Hogan for mice; Swagata Goswami, Patricia Zarnovican and Imke Peters for help; Rita Gerardy-Schahn for lab space; and the Research Core Unit Transcriptomics of Hannover Medical School for microarray analysis. Some data of this study form part of a PhD thesis (J.K., Gottfried Wilhelm Leibniz Universität Hannover, 2021).