The organized array of smooth muscle cells (SMCs) and fibroblasts in the walls of visceral tubular organs arises by patterning and differentiation of mesenchymal progenitors surrounding the epithelial lumen. Here, we show that the TBX2 and TBX3 transcription factors have novel and required roles in regulating these processes in the murine ureter. Co-expression of TBX2 and TBX3 in the inner mesenchymal region of the developing ureter requires canonical WNT signaling. Loss of TBX2/TBX3 in this region disrupts activity of two crucial drivers of the SMC program, Foxf1 and BMP4 signaling, resulting in decreased SMC differentiation and increased extracellular matrix. Transcriptional profiling and chromatin immunoprecipitation experiments revealed that TBX2/TBX3 directly repress expression of the WNT antagonists Dkk2 and Shisa2, the BMP antagonist Bmper and the chemokine Cxcl12. These findings suggest that TBX2/TBX3 are effectors of canonical WNT signaling in the ureteric mesenchyme that promote SMC differentiation by maintaining BMP4 and WNT signaling in the inner region, while restricting CXCL12 signaling to the outer layer of fibroblast-fated mesenchyme.
Tubes are fundamental structures in organs that engage in the transport of gases, fluids and solids in metazoans. Besides the luminal epithelial lining, they are characterized by outer layers of fibro-muscular material that provide rigidity, flexibility and contractile activity to the tube (Iruela-Arispe and Beitel, 2013). The mammalian ureters are straight tubes that propel urine from the renal pelvis to the urinary bladder, and represent a simple system in which to analyze molecular pathways that control the differentiation and organization of fibroblasts and smooth muscle cells (SMCs) from a pool of mesenchymal progenitors.
In the mouse, these progenitors are first identified at embryonic day (E) 11.5 as a group of Tbx18-expressing fibroblast-like cells that surround the stalk region of the ureteric bud, an epithelial outgrowth of the nephric duct (Airik et al., 2006). At E12.5, Tbx18+ cells in proximity to the epithelium [hereafter referred to as cells of the inner layer of the ureteric mesenchyme (UM)] acquire a rhomboid shape and are densely compacted. They initiate the SMC program at E14.5 as evidenced by expression of the SMC regulatory gene Myocd. The majority of these SMC progenitors become terminally differentiated, contractile SMCs but some regain a fibroblast-like character and populate the subepithelial space. Tbx18+ cells further distal to the epithelium (hereafter referred to as cells of the outer layer of the UM) maintain their initial fibroblast-like character and differentiate from E13.5 on into adventitial fibroblasts (Bohnenpoll et al., 2017a). Shortly after onset of urine production in the kidney at E16.5, the ureteric wall has acquired a three-layered organization: the fibrous lamina propria on the inside, the peristaltically active tunica muscularis in the middle, and the fibrous tunica adventitia on the outside (Velardo, 1981) (Fig. S1, for a scheme of ureter development).
Tissue recombination and genetic experiments indicate that survival, patterning and differentiation of UM progenitors depend on SHH and WNT signals from the adjacent ureteric epithelium (UE) (Baskin et al., 1996; Bohnenpoll and Kispert, 2014; Cunha et al., 1991; Trowe et al., 2012; Yu et al., 2002). At E11.5, SHH activates a SMO-dependent pathway in the UM that is required for survival of cells in the outer layer, for proliferation and SMC differentiation of cells of the inner layer, and also for urothelial proliferation and differentiation. The proliferation and differentiation functions of SHH signaling are mediated by the forkhead transcription factor FOXF1, which, in turn, induces expression of and synergizes with BMP4 in SMC differentiation (Bohnenpoll et al., 2017b; Mamo et al., 2017). At E12.5, WNT7B and WNT9B activate the canonical WNT pathway in cells of the inner mesenchymal layer, disruption of which results in reduced proliferation and failed SMC differentiation of these cells, and expansion of the adventitial fate to the inner layer (Trowe et al., 2012). The transcription factors that mediate WNT signaling downstream of β-catenin (CTNNB1) in the UM are unknown.
Tbx2 and Tbx3 are two closely related members of the evolutionarily conserved family of T-box transcription factor genes (Papaioannou, 2014). Both encode transcriptional repressors that regulate proliferation, patterning and differentiation programs in a variety of developmental contexts, in some cases redundantly (Douglas and Papaioannou, 2013; Lu et al., 2010; Lüdtke et al., 2016; Singh et al., 2012; Zirzow et al., 2009).
Expression of TBX2 in the human fetal and adult kidney (Campbell et al., 1995; Law et al., 1995), of Tbx3 in the bladder urothelium (Ito et al., 2005), of Tbx2 and Tbx3 in the nephric duct and in the mesenchymal core of the developing urethra in the mouse (Chapman et al., 1996; Douglas et al., 2012), and urinary tract abnormalities in human patients with heterozygous loss of TBX3 (Gonzalez et al., 1976; Pallister et al., 1976) indicate that both genes also contribute to multiple subprograms in the development of different organs of the mammalian urinary tract. Here, we demonstrate essential roles for these factors in the differentiation of the mesenchymal components of the mouse ureter. We show that both genes are expressed in the inner layer of the UM in a Ctnnb1-dependent manner, and that they mediate a subset of canonical WNT signals in this tissue.
Tbx2 and Tbx3 expression in the UM depends on canonical WNT signaling
To determine the expression of Tbx2 and Tbx3 in ureter development, we performed in situ hybridization analysis on transverse sections of the trunk region of E12.5-E18.5 wild-type embryos. At all stages, Tbx2 and Tbx3 transcripts were abundant in the epithelial compartment of the ureter and also present in adjacent mesenchymal cells both at the proximal (kidney) level (Fig. 1A) as well as distally, i.e. close to the bladder (Fig. S2A). Immunofluorescence analysis confirmed expression of TBX2 and TBX3 protein in the UE from E12.5 to E18.5. Expression in the UM was prominent at E12.5 and E14.5, and weaker at E16.5 (Fig. 1B; Fig. S2B).
Because expression of both genes was restricted to the inner layer of the UM, we questioned whether their expression depends on SHH or WNT signals emanating from the UE (Trowe et al., 2012; Yu et al., 2002). To address this question, we used a conditional pathway deletion approach, with a Tbx18cre 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), and of canonical WNT signaling, Ctnnb1 (Brault et al., 2001), as previously reported (Bohnenpoll et al., 2017b; Trowe et al., 2012). Loss of Smo had no effect on Tbx2 and Tbx3 expression in the UM at E12.5 (Fig. 1C). In contrast, expression of both genes was lost from this region in Tbx18cre/+;Ctnnb1fl/fl embryos. Moreover, misexpression of a stabilized version of CTNNB1 in the entire UM (mimicking activated WNT signaling; Tbx18cre/+;Ctnnb1ex3(fl)/+) (Harada et al., 1999; Trowe et al., 2012) resulted in increased and expanded expression of Tbx2 and Tbx3 in this region at E12.5 (Fig. 1C). Together, this analysis indicates that expression of Tbx2 and Tbx3 in the inner layer of the UM depends on canonical WNT signaling and is independent of the SHH pathway.
Loss of Tbx2 and Tbx3 in the UM leads to reduced SMC differentiation
To investigate the role of Tbx2 and Tbx3 in the UM, we used the Tbx18cre line and floxed alleles of Tbx2 (Wakker et al., 2010) and Tbx3 (Frank et al., 2013). The validity of this approach was confirmed by absence of TBX2 and TBX3 protein expression specifically in the mesenchymal compartment of Tbx18cre/+;Tbx2fl/fl;Tbx3fl/fl (Tbx2/3cDKO) ureters at E12.5 (Fig. S3).
At E18.5, whole-mount preparations of urogenital systems of mice with conditional loss of two or three alleles of Tbx2 and/or Tbx3 appeared grossly normal (Fig. S4A). Kidneys and ureters were histologically undistinguishable from the controls (Fig. S4B). Expression of the SMC regulatory gene Myocd and of the SMC structural genes Myh11, Tagln and Tnnt2 was normal or appeared slightly reduced (Tnnt2 in the Tbx2 and Tbx3 single mutants). The adventitial marker genes Col1a2, Fbln2 and Dpt, and the lamina propria marker Aldh1a2 were normally expressed in the mutants (Fig. S4C), arguing that patterning of the mesenchymal compartment of the ureter is undisturbed, and SMC differentiation is minimally affected in triple allele mutants.
We next analyzed compound homozygous double mutants. At E18.5, urogenital systems of Tbx2/3cDKO embryos appeared morphologically unaffected (Fig. 2A). We did not detect histological changes in the kidney, but the mutant ureter lacked a clear distinction between a condensed inner and a more loosely organized outer mesenchymal layer. Instead, the entire UM was loosely organized with excessive extracellular space (Fig. 2B). Expression of SMC markers (Myocd, Myh11, Tagln, Tnnt2) was reduced to small patches of cells. The lamina propria marker Aldh1a2 was restricted to a cell layer directly underneath the epithelium, as in the control. Some markers of the outer adventitial layer appeared unchanged (Dpt, Postn) whereas others were strongly expanded to the inner mesenchymal region (Col1a1, Col1a2, Fbln2) (Fig. 2C). Differentiation of urothelial cell types was unaffected in the mutant as revealed by normal expression of KRT5, ΔNP63 and UPK1B, which combinatorially marked basal cells (KRT5+ΔNP63+UPK1B−), intermediate cells (KRT5−ΔNP63+UPK1B+) and superficial cells (KRT5−ΔNP63−UPK1B+) (Fig. 2D) (Bohnenpoll et al., 2017a).
To profile the molecular and cellular changes in E18.5 Tbx2/3cDKO ureters in an unbiased fashion, we compared their transcriptome with that of control ureters by microarray analysis. Using a threshold of 1.5-fold change and an expression intensity robustly above background (>100), we detected 405 genes with reduced expression and 327 with increased expression in Tbx2/3cDKO ureters (Fig. 3A; Tables S1, S2). Functional annotation using the DAVID software tool revealed strong enrichment of the terms extracellular matrix (ECM) and collagen in the pool of upregulated genes whereas the pool of downregulated genes was strongly enriched for various terms relating to structure and function of SMCs (Fig. 3B,C; Tables S3, S4). Hence, loss of Tbx2 and Tbx3 in the UM leads to a reduced SMC phenotype and a gain of ECM (mostly collagen and fibulin) deposition in the inner layer.
Loss of Tbx2 and Tbx3 in the UM disrupts ureter peristalsis
Tbx2/3cDKO ureters did not present the dilatation phenotype described in other mutants with reduced SMC investment (Bohnenpoll et al., 2017b; Mamo et al., 2017; Trowe et al., 2012), so we questioned whether peristalsis was affected. We isolated E18.5 ureters and cultured them for 6 days in a transwell setting, monitoring contraction frequency and intensity daily (Fig. 3D-F). After 1 day of culture both wild-type and mutant ureters contracted approximately three times per minute. Whereas the contraction frequency of the wild-type ureters decreased to 1.3 at day 6, that of mutant ureters increased to 4.6 (Fig. 3E). Interestingly, the contraction intensity of mutant ureters was significantly lower at all analyzed time points (Fig. 3F; Movies 1, 2). Together, this argues that reduced SMC investment results in reduced contraction intensity, which is counteracted by an increased contraction frequency, thus preventing dilatation at least up to this stage.
Early onset of ureter defects in Tbx2/3cDKO embryos
To define both the onset as well as the progression of mesenchymal defects in Tbx2/3cDKO ureters, we analyzed earlier embryonic stages. Histological analysis revealed a clear division of the UM into an inner layer with rhomboid-shaped condensed cells and an outer layer with loosely organized fibroblast-like cells at E14.5, E15.5 and E16.5 in the wild type. In the mutant, the UM was similarly subdivided at E14.5 but the inner layer appeared progressively less condensed at the subsequent stages (Fig. 4A). In the wild type, onset of Myocd expression at E14.5 was followed by that of Myh11, Tagln and Tnnt2 1 day later (Fig. 4B). Col1a2 and Fbln2 expression was homogeneous in the UM at E14.5, but was downregulated in the inner layer at E16.5. In the mutant ureter, Myocd and SMC structural genes were not activated until E16.5 and then only weakly. In contrast, Col1a2 and Fbln2 expression was found throughout the mutant UM at all stages (Fig. 4C). This SMC differentiation defect was accompanied by a delayed onset of peristaltic activity. Wild-type ureters explanted at E14.5 commenced contractions after 2 days in culture whereas mutant ureters started to contract after 4-6 days and had reduced contraction intensity and SMC investment (Fig. 4D,E; Fig. S5; Movies 3, 4).
A terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay did not detect apoptotic cells at E12.5 and at E14.5 in the mutant UM nor did a 5-bromo-2′-deoxyuridine (BrdU) incorporation assay reveal changes of proliferation in this tissue or the adjacent epithelium indicating that delayed SMC differentiation is not due to changes in either of these cellular programs (Fig. S6).
Reduced activity of a Foxf1-BMP4 module precedes SMC differentiation defects
To identify the molecular causes of the SMC differentiation defect and the de-repression of ECM genes in the inner layer of the UM in Tbx2/3cDKO embryos, we screened expression of a panel of genes previously implicated in the early development of the UM and the initiation of the SMC program (Bohnenpoll and Kispert, 2014). In E12.5 wild-type ureters, expression of Ptch1 (target of SHH signaling; Ingham and McMahon, 2001), Bmp4, Axin2 (target of WNT signaling; Jho et al., 2002) and Rarb (target of retinoic acid signaling; Mendelsohn et al., 1991), and of the transcription factor genes Tbx18, Sox9, Tcf21 and Tshz3 were expressed in the UM with highest levels in cells adjacent to the epithelium. Expression of all genes appeared unchanged in Tbx2/3cDKO ureters (Fig. S7).
In E14.5 wild-type ureters, the genes listed above and Id2 (BMP target; Hollnagel et al., 1999) were expressed in SMC progenitors. Sox9 was an exception as it was decreased. In Tbx2/3cDKO ureters, expression of Ptch1, Axin2, Rarb, Tbx18, Tshz3 and Sox9 was unchanged (Fig. S8). In contrast, Foxf1 was reduced. Expression of Bmp4 was strongly increased, whereas its target gene Id2 showed strongly reduced expression in the UM and unaltered expression in the epithelium (Fig. 4F). Thus, despite increased Bmp4 expression, the data indicate decreased BMP signaling in the UM. This, in combination with reduced expression of the SMC regulatory gene Foxf1 (Bohnenpoll et al., 2017b), likely contributes to the delayed and reduced SMC differentiation observed in Tbx2/3cDKO ureters.
Ectopic expression of TBX2 in the UM interferes with mesenchymal patterning and differentiation
Expression of TBX2 and TBX3 is confined to the inner layer of the UM from E12.5 to E16.5. To explore the significance of this temporo-spatial restriction and to gain further insight into the molecular function of these two transcription factors in the UM, we utilized a Cre/loxP-based misexpression approach.
For this, we employed the Tbx18cre line, and a HprtTBX2 line harboring an integration of a bicistronic transgene-cassette containing the human TBX2 open-reading frame followed by IRES-GFP in the ubiquitously expressed X-chromosomal Hprt locus (Singh et al., 2012). Owing to random X-chromosome inactivation, Tbx18cre/+;HprtTBX2/+ [female gain of function (GOF)] embryos have mosaic TBX2 expression whereas Tbx18cre/+;HprtTBX2/y (male GOF) embryos expressed the transgene in a uniform manner in the UM (Fig. S9).
Transgene expression did not interfere with viability of embryos at E18.5 but did affect the integrity of the urogenital system. Male GOF embryos invariably developed uretero-pelvic junction obstruction (Fig. 5A-C) and females had bilateral hydroureter with associated hydronephrosis (Fig. S10A-C). Expression of SMC genes was strongly reduced in male and female GOF ureters (Fig. 5D-G; Fig. S10D-G) as was the thickness of the adventitial layer at this stage (Fig. 5H,I; Fig. S10H,I).
Analysis of GOF ureters at E14.5 to E16.5 showed that SMC differentiation was delayed and reduced (Fig. 5J-M; Fig. S10J-M). Markers of the ECM were variably affected: Col1a2 was minimally affected whereas Fbln2 was markedly reduced in the outer layer of the UM (Fig. 5N,O; Fig. S10N,O).
Analysis of E14.5 male GOF ureters revealed that ectopic TBX2 did not alter expression of Ptch1, Axin2, Bmp4, Id2, Rarb, Tshz3, Tbx18 and Sox9 in the UM (Fig. S11). In contrast, expression of Foxf1 was markedly decreased; expression of Postn and Foxd1 in the outer layer of the UM was undetectable (Fig. 5P). Hence, ectopic expression of TBX2 interferes with mesenchymal patterning and SMC differentiation in the ureter, as does the combined loss of Tbx2 and Tbx3.
Comparative transcriptome analysis of Tbx2 loss- and gain-of-function ureters identifies genes regulated by TBX2/TBX3
As Tbx2 and Tbx3 often function as transcriptional repressors, many of their target genes should be activated under loss- and repressed under gain-of-function conditions. To identify such genes in an unbiased fashion, we performed microarray-based gene expression profiling. We compared wild-type and Tbx2/3cDKO ureters at E14.5, and wild-type and male GOF ureters at E13.5 to obtain a 2-day window of transcriptional changes in both loss- and gain-of-function conditions. Because the inner layer of the UM (where TBX2 and TBX3 function) represents only a fraction of the entire ureter, we employed a relatively low fold-change filter of 1.2. Using an intensity threshold of 100 as additional filter, we identified 238 genes that were consistently upregulated and 260 genes that were downregulated in Tbx2/3cDKO ureters (Fig. 6A; Tables S5, S6). Functional annotation found an enrichment of ECM terms with the pool of upregulated genes (Tables S7, S8). In male GOF ureters, expression of 298 genes was down- and of 442 genes upregulated (Fig. 6B; Tables S9, S10). The downregulated transcripts were enriched for the functional annotation terms muscle, ECM and WNT (Tables S11, S12).
The microarrays quantitatively confirmed the results of our in situ hybridization analysis of expression of SMC regulatory genes and pathways in the mutant ureters (Fig. 4F and Fig. 5P). Additionally, Id4, another direct target of BMP4 signaling (Liu and Harland, 2003), was decreased (−1.5). In male GOF ureters, expression of Foxf1 was decreased (−1.5) as was that of Tbx18 (−1.4) (Tables S5, S6, S9, S10).
We found 30 genes at the intersection of upregulated genes in Tbx2/3cDKO ureters and downregulated genes in male GOF ureters. Ten of these 30 genes were also upregulated in their expression in the microarray from E18.5 Tbx2/3cDKO ureters. Cxcl12 was slightly below the threshold in the E14.5 Tbx2/3cDKO pools (1.2-fold upregulated). We included this gene in the analysis because it was changed in Ctnnb1-deficient ureters (Trowe et al., 2012) (Fig. 3A and Fig. 6C,D; Table S13).
We used in situ hybridization to validate the expression of these genes in E14.5 control, Tbx2/3cDKO and male GOF ureters (Figs S12, S13). We found six genes with differential expression in any two of the three genotypes: Bmper, which encodes a BMP antagonist (Moser et al., 2003); Cxcl12, which encodes a chemokine (Harris et al., 2013); Dkk2 and Shisa2 encoding WNT antagonists (Glinka et al., 1998; Yamamoto et al., 2005); the vascular cell adhesion molecule gene Vcam1; and Fam129a, a gene without known protein function (Figs S12, S13). In the wild type, Bmper and Cxcl12 were uniformly expressed in the entire UM at E13.5 but subsequently restricted to the outer layer. In Tbx2/3cDKO ureters, Bmper and Cxcl12 expression in the inner layer of the UM remained high from E13.5 onwards whereas in GOF ureters it was reduced. Dkk2 was weakly expressed in the inner region of the UM of E13.5 wild-type embryos but subsequently downregulated. In Tbx2/3cDKO ureters, there was ectopic Dkk2 expression in the inner layer of the UM at E14.5. Shisa2 expression was confined to this region at E13.5 and E14.5 and expression in Tbx2/3cDKO ureters was increased at E14.5 and further augmented at E18.5. In GOF mutants, Shisa2 expression was reduced at E14.5 (Fig. 6E). Vcam1 was largely confined to the outer layer of the UM in the wild type. In Tbx2/3cDKO embryos, expression was slightly increased in the inner layer of the UM. In male GOF mutants, expression appeared slightly reduced in the entire UM (Fig. S13). Finally, Fam129a was expressed in the outer layer of the UM of wild-type embryos at E14.5. Expression in this domain was unaltered in the loss- but decreased in the gain-of-function condition at this stage (Fig. S12).
The expression patterns of Bmper, Cxcl12, Dkk2 and Shisa2 in the Tbx2/Tbx3 loss- and gain-of function conditions are consistent with these genes being direct targets of TBX2/TBX3-mediated transcriptional repression in the UM. Because embryonic ureters represent a tiny source of chromatin, we did not attempt a genome-wide occupancy chromatin immunoprecipitation (ChIP) experiment. Instead, we interrogated a previously generated data set of TBX3-bound chromatin from the embryonic lung, in which TBX2 and TBX3 are also functional in the undifferentiated mesenchyme (Lüdtke et al., 2016). We found TBX3 binding peaks associated with all four genes (Fig. S14, Table S14).
To validate whether TBX2 binds to genomic sequences harboring these peak regions, we performed ChIP on wild-type E16.5 ureters using anti-TBX2 antibody (Fig. 6F; Table S14). Enrichment was detected for the one peak found in Shisa2, and for one of the four and six tested peaks in Cxcl12 and Dkk2, respectively. Multiple peaks were enriched in Bmper (Fig. 6F,G). Hence, Bmper, Dkk2, Shisa2 and Cxcl12 may be direct targets of TBX2/TBX3 activity in the UM.
TBX2 and TBX3 are members of a closely related subfamily of T-box transcription factors that regulate a diverse array of developmental programs. To date, the molecular functions of these genes in the development of the mammalian urinary system were not characterized. Here, we identified a crucial role for these transcriptional repressors in the development of SMCs, the cell type essential for the contractile peristaltic activity of the ureter tube. Our results suggest that TBX2/TBX3 mediate specific aspects of canonical WNT signaling function in this tissue by restricting outer adventitial programs and supporting SMC differentiation in the inner region. Molecularly, this is achieved by regulation of BMP, and possibly of WNT and CXCL12 signaling (Fig. 7).
TBX2 and TBX3 mediate part of the function of canonical WNT signaling in the UM
SMCs provide the main support for the structure and the contractile activity of many tubular organs. Expression of many of the genes that characterize the SMC phenotype is under control of serum response factor (SRF), which recognizes cognate binding sites in the promotors of these genes and activates gene expression in combination with strong transcriptional activators such as MYOCD (Coletti et al., 2016; Miano, 2015). Expression and activity of SRF and MYOCD is regulated by a variety of intrinsic and extrinsic signals that reflect the heterogeneous developmental origin of this cell population. In the vascular system, NOTCH, TGFβ and platelet-derived growth factor have been characterized as main drivers of SMC differentiation, whereas in visceral tubes, SHH, BMPs and WNTs are predominant (Cohen et al., 2009; Itäranta et al., 2006; Mack, 2011; Shi and Chen, 2016; Trowe et al., 2012). The molecular targets of these signals, particularly of WNTs, as well as their interactions have remained poorly understood. Our previous work showed that WNT signaling is necessary and sufficient to subdivide the homogenous UM in a radial fashion by inducing cells in the vicinity of the epithelium (i.e. close to the source of the signal) to restrict to the SMC lineage (Trowe et al., 2012). Our expression analysis, in combination with conditional loss- and gain-of-function experiments, suggest that Tbx2 and Tbx3 are targets of this pathway and mediate some of its patterning and differentiation functions in the UM.
Our expression analysis showed that Tbx2 and Tbx3 are co-expressed in the inner layer of the UM from E12.5 to around E16.5. This pattern coincides with that of Axin2, a read-out of canonical WNT signaling (Jho et al., 2002; Trowe et al., 2012). Loss of Ctnnb1 abolished the mesenchymal expression of Tbx2 and Tbx3 in the ureter whereas a stabilized version of CTNNB1 was sufficient to induce ectopic expression of the two genes, collectively indicating that canonical WNT signaling regulates (co-)expression of Tbx2 and Tbx3 in the UM. It is noteworthy that expression of Tbx2 and Tbx3 at other embryonic sites is regulated by other signals, including BMPs (Behesti et al., 2006; Ma et al., 2005; Zirzow et al., 2009) and SHH (Lüdtke et al., 2016), indicating that the regulatory landscape for ureter expression of Tbx2 and Tbx3 is tissue specific.
Combined loss of Tbx2 and Tbx3 in the UM did not affect proliferation and resulted in the formation of a ureter of normal length showing that the pro-proliferative aspect of WNT signaling on mesenchymal progenitors is independent of these two factors. This comes as a surprise because in other in vivo and in vitro settings TBX2 and TBX3 repress cell cycle inhibitors to ensure proliferative expansion of progenitors (Jacobs et al., 2000; Lüdtke et al., 2013; Prince et al., 2004). It is conceivable that WNT signaling directly regulates cell cycle activators, such as Ccnd1 and Myc, as previously shown (Trowe et al., 2012), or employs as yet unknown transcription factors in the regulation of cell cycle progression.
Tbx2/3cDKO ureters displayed an expansion of expression of genes encoding ECM proteins, particularly collagens and fibulins, into the inner SMC layer of the ureteric wall, and ectopic expression of TBX2 was sufficient to repress expression of some of these genes in the outer layer. However, a clearly demarcated outer tunica adventitia with tangentially oriented fibroblast-like cells that expressed markers such as Dpt and Postn was established in Tbx2/3cDKO ureters. Hence, TBX2/TBX3 do not mediate all patterning functions of WNT signaling in the UM, but are required to suppress a particular adventitial subprogram, namely the deposition of ECM in the inner layer of the UM.
Possibly connected with this phenotypic difference is the finding that SMCs are not absent in Tbx2/3cDKO ureters but are dispersed, severely reduced in number and abnormally differentiated shortly before birth. Importantly, inner mesenchymal cells acquired a typical rhomboid shape, appeared condensed and expressed markers such as Foxf1 and Axin2 in the mutant at E14.5, compatible with the notion that SMC progenitors have been established. These SMC progenitors are likely to maintain their ability to produce ECM while being at least partially able to initiate the SMC differentiation program, similar to myofibroblasts associated with fibrotic disease states. Consequently, the Tbx2/3cDKO ureter can still withstand the hydrostatic pressure of the urine and does not dilate. In contrast, in Ctnnb1-deficient ureters SMC progenitors are not established leading to complete absence of mature SMCs at birth and dilatation after onset of urine production in the fetal kidney (Trowe et al., 2012). Expansion of the adventitial fates to the inner mesenchymal region in these mutants may therefore simply reflect the default state of differentiation in the complete absence of SMC progenitors.
Although other (direct) targets of canonical WNT signaling in the UM have not been identified, molecular marker analysis of Ctnnb1-deficient ureters provided evidence for additional WNT-dependent factors possibly involved in SMC specification. Expression of Tbx18 and Sox9, transcription factor genes required for the specification and SMC differentiation of the UM (Airik et al., 2006, 2010), respectively, were absent at E12.5, whereas at E14.5 expression of Gata2, Tcf21 and Tshz3 was lost. Moreover, expression of the SHH target gene, Ptch1 and its mediator Bmp4 was completely extinguished at this stage (Trowe et al., 2012). This indicates that WNT signaling maintains a set of crucial transcriptional regulators as well as the SHH-FOXF1-BMP4 regulatory axis independently of Tbx2/Tbx3, and that this could at least partly account for the lack of SMC specification and the more severe phenotype associated with its loss in the ureter.
Although misexpression of a stabilized version of CTNNB1 in the UM was sufficient to induce ectopic formation of SMC progenitors (Trowe et al., 2012), widespread and premature expression of TBX2 performed here did not; this is further evidence that TBX2/TBX3 acts downstream in a subprogram of SMC differentiation once progenitors are induced. In fact, misexpression of TBX2 resulted in a lack of SMC differentiation, revealing crucial temporal regulation for onset of expression after the progenitors are established. Although ectopic TBX2 did not affect the major signaling pathways, we found reduced expression of Foxf1, an essential downstream mediator of SHH signaling in SMC differentiation (Bohnenpoll et al., 2017b). Intriguingly, the Foxf1/Foxf2 locus harbors a regulatory element bound and activated by TBX5 (Hoffmann et al., 2014). Because TBX2 and TBX3 can bind to the same DNA elements as TBX5 (Habets et al., 2002), it is conceivable that overexpressed TBX2 directly represses transcription of Foxf1 and Foxf2 upon binding to this element, thus compromising SMC differentiation.
Ideally, we would further substantiate our conclusion that TBX2/TBX3 mediate part of WNT signaling in the UM with a genetic rescue experiment. Unfortunately, this experiment is technically unfeasible at present because Tbx18cre-mediated misexpression of TBX2 in the UM leads to SMC inhibition. An Axin2creERT2 line would permit TBX2 expression in SMC progenitors but, as this line is itself under control of WNT signaling (Bohnenpoll et al., 2017b; van Amerongen et al., 2012), it cannot be activated when WNT signaling is inhibited as would be required for the rescue experiment.
TBX2 and TBX3 regulate specific signaling pathways in the UM
Our global analysis of transcriptional changes identified four genes that were both upregulated when Tbx2/Tbx3 were inactivated and downregulated when enhanced TBX2 expression was forced into the UM. Compatible with the notion that they represent direct targets of TBX2/TBX3 transcriptional repression, they also featured productive TBX2/TBX3 binding peaks in their regulatory regions. The nature of the encoded proteins in combination with the observed molecular changes suggest that TBX2/TBX3 regulate SMC development largely by repressing Bmper to maintain BMP signaling. Repression of CXL12 signaling and maintenance of WNT signaling may present additional mediators of its function to permit functional SMCs.
Dkk2 encodes a member of a small family of secreted glycoproteins that inhibit WNT signaling by binding to the WNT co-receptors LRP5/6 and KREMEN (Mao et al., 2002, 2001). Shisa2 encodes a member of a family of transmembrane proteins that trap WNT and FGF receptors in the endoplasmatic reticulum and prevent their maturation (Yamamoto et al., 2005). Both Dkk2 and Shisa2 were expressed in the inner layer of the UM in wild-type embryos at E13.5 and were subsequently downregulated (Dkk2) or maintained at low levels (Shisa2) compatible with the notion that they are involved in lowering the levels of WNT and possibly FGF signaling in this region at early stages. Dkk2 was upregulated in Tbx2/3cDKO ureters at E14.5, but not at E18.5. In contrast, Shisa2 was maintained at high levels even at E18.5. Surprisingly, at E14.5 expression of Axin2, a bona fide target of canonical WNT signaling, and of the FGF target Etv4 (Mao et al., 2009), was unchanged in Tbx2/3cDKO embryos. This suggests that changes in the activities of these pathways are not present or are too small to be detected at this stage. Alternatively, Axin2 and Etv4 may not be faithful read-outs of the activity of these pathways in this context.
In a recent report on the molecular function of Tbx2 and Tbx3 in the pulmonary mesenchyme, Frzb, a secreted frizzled-related protein that competitively inhibits WNT binding to frizzled receptors, and Shisa3 were identified as direct functional targets of TBX2/TBX3 in this tissue (Lüdtke et al., 2016). This adds to the notion that in some developmental contexts TBX2 and TBX3 maintain WNT signaling at high levels by repressing members of various families of WNT antagonists.
Although our data do not clearly indicate altered WNT signaling in developing Tbx2/3cDKO ureters, expression of the bona fide targets of BMP signaling Id2 and Id4 (Hollnagel et al., 1999; Liu and Harland, 2003) was clearly downregulated in the inner mesenchymal layer of Tbx2/3cDKO ureters at E14.5. This correlated with increased expression of the BMP antagonist gene Bmper (Moser et al., 2003), arguing that ectopic BMPER accounts for reduced BMP4 signaling. Interestingly, expression of Bmp4 was strongly upregulated indicating a compensatory feedback mechanism. We found that Foxf1 was also severely downregulated in the Tbx2/3cDKO ureters, suggesting a possible role of BMP4 signaling as an activator of Foxf1 transcription. Alternatively, reduced SHH or WNT signaling input may have lowered Foxf1 expression (Bohnenpoll et al., 2017b). As FOXF1 and BMP4 are both independently required for SMC differentiation in the ureter (Bohnenpoll et al., 2017b; Mamo et al., 2017), we suggest that their reduced expression and activity, respectively, largely accounts for the delayed and reduced onset of SMC differentiation in Tbx2/3cDKO ureters.
Notably, BMP4 signaling in the UE was not affected in Tbx2/3cDKO embryos. The expression level of Id2 was normal and urothelial differentiation was unchanged. It is possible that urothelial differentiation requires lower levels of BMP4 than the mesenchyme (Bohnenpoll et al., 2017b). Alternatively, reduced expression of Foxf1 and of BMP4 signaling may combinatorially enhance the mesenchymal defects, as discussed.
Our analysis also showed that TBX2/TBX3 are both required and sufficient to repress expression of Cxcl12 in the inner layer of the UM. CXCL12 is a chemokine that mediates its effects by binding to CXCR4 and CXCR7 (ACKR3) (Balabanian et al., 2005; Burns et al., 2006). CXCL12 has a well-established role as a chemoattractant for numerous cell types, most prominently immune cells, but also circulating fibroblasts (Guyon, 2014). The latter may contribute in some organ settings to fibrosis by depositing collagens (Phillips et al., 2004). A number of reports indicate a more direct involvement of CXCL12 in control of ECM deposition, possibly by conversion of resident fibroblasts to myofibroblasts (Jackson et al., 2017; Rodriguez-Nieves et al., 2016; Tan et al., 2017). Although the role of CXCL12/CXCR4/7 signaling in the ureter has not been analyzed, the above reports suggest that expanded CXCL12 signaling in Tbx2/3cDKO ureters contributes in some way to the ectopic deposition of ECM in the inner layer of the UM.
Together, our analysis reveals a crucial role for the T-box transcriptional repressors TBX2 and TBX3 in regulating the temporal and spatial activity of at least three different signaling pathways to assure the progression of SMC progenitors to fully differentiated contractile SMCs in the ureter (Fig. 7). Whether TBX2 and TBX3 play a similar role in the development of other visceral or vascular SMCs remains to be explored.
MATERIALS AND METHODS
Mouse strains and husbandry
The mouse alleles employed have all previously been described: a loss-of-function allele of Tbx18 generated by insertion of the cre gene into the start codon [Tbx18tm4(cre)Akis; synonym: Tbx18cre] (Trowe et al., 2010); floxed loss-of-function lines for Tbx2 (Tbx2tm2.2Vmc; synonym: Tbx2fl) (Wakker et al., 2010), Tbx3 (Tbx3tm3.2Moon; synonym: Tbx3fl) (Frank et al., 2013), β-catenin (Ctnnb1tm2Kem; synonym: Ctnnb1fl) (Brault et al., 2001) and Smo (Smotm2Amc; synonym: Smofl) (Long et al., 2001); a floxed gain-of-function allele of β-catenin [Ctnnb1tm1Mmt; synonym: Ctnnb1(ex3)fl] (Harada et al., 1999); the reporter line Gt(ROSA)26Sortm4(ACTB-tdTomato,-EGFP)Luo (synonym: Rosa26mTmG) (Muzumdar et al., 2007); and an allele with insertion of the human TBX2 gene at the Hprt locus [Hprttm2(CAG-TBX2,-EGFP)Akis; synonym: HprtTBX2] (Singh et al., 2012). All were maintained on an outbred (NMRI) background.
Tbx18cre/+;Tbx2fl/fl;Tbx3fl/fl embryos were generated by mating Tbx18cre/+;Tbx2fl/+;Tbx3fl/+ males with Tbx2fl/fl;Tbx3fl/fl;R26mTmG/mTmG females. To generate embryos conditionally misexpressing human TBX2 or stabilized Ctnnb1, Tbx18cre/+ males were mated with HprtTBX2/TBX2 or Ctnnb1(ex3)fl/(ex3)fl females, respectively. Cre-negative littermates were used as controls. For timed pregnancies, vaginal plugs detected in the morning after mating were designated as E0.5 at noon.
All animal work conducted for this study was approved by the Niedersächsisches Landesamt für Verbraucherschutz und Lebensmittelsicherheit (permit number AZ33.12-42502-04-13/1356) and was performed at the central animal laboratory of the Medizinische Hochschule Hannover according to European and German legislation (2010/63/EU and TierSchG).
Organ cultures and video documentation
Ureters for explant cultures were dissected in L-15 Leibovitz medium (Biochrom). Ureters isolated from the embryo were explanted on 0.4 µm polyester membrane Transwell supports (Corning) and cultured at the air-liquid interface with DMEM/F12 (Gibco) supplemented with 10% fetal calf serum (Biochrom), 1× Penicillin/Streptomycin, 1× Pyruvate and 1× Glutamax (all from Gibco) in a humidified incubator with 5% CO2 at 37°C. Medium was refreshed every second day.
To document the contractile behavior of ureter explants, culture plates were removed from the incubator and imaged for 1 min at room temperature using a Leica DMI6000B microscope. The contraction frequency was expressed as the number of full proximal-to-distal contractions per minute. Individual contraction intensities were defined as the difference between relaxed and contracted width divided by the relaxed width of the ureter. The overall contraction intensity was defined as the average of proximal, medial and distal contraction provided as a percentage. All movies were processed and analyzed using ImageJ software (Schneider et al., 2012).
Histological and immunohistochemical analyses
Embryos, urogenital systems and ureter explants were fixed in 4% paraformaldehyde, paraffin-embedded and sectioned at 5 µm. Sections were stained with Hematoxylin and Eosin according to standard procedures.
Immunofluorescence staining was performed on 5-µm-thick paraffin sections using the following primary antibodies: polyclonal rabbit-anti-TBX2 (1:500; 07-318, Millipore), polyclonal goat-anti-TBX3 (1:500; sc-31656, Santa Cruz), polyclonal rabbit-anti-KRT5 (1:250; PRB-160P-100, Covance), polyclonal rabbit-anti-ΔNP63 (1:250; 619001, BioLegend), monoclonal mouse-anti-UPK1B (1:250; WH0007348M2, Sigma-Aldrich), polyclonal rabbit-anti-TAGLN (1:200; ab14106, Abcam), polyclonal FITC-conjugated rabbit anti-ACTA2 (1:200; F3777, Sigma-Aldrich) and monoclonal mouse-anti-BrdU (1:250; 1170376, Roche).
Fluorescent staining was performed using the following secondary antibodies: biotinylated goat anti-rabbit IgG (1:250; 111065033, Dianova), biotinylated donkey anti-goat IgG (1:250; 705-065-003, Dianova), biotinylated goat-anti-mouse IgG (1:250; 115-065-003, Jackson ImmunoResearch), Alexa 488-conjugated goat anti-rabbit IgG (1:500; A11034, Molecular Probes) and Alexa 555-conjugated goat anti-mouse IgG (1:500; A21422, Molecular Probes). The signals of TBX2, TBX3, ΔNP63 and BrdU were amplified using the Tyramide Signal Amplification system (Perkin Elmer). For antigen retrieval, paraffin sections were deparaffinized, pressure-cooked for 20 min in antigen unmasking solution (Vector Laboratories), treated with 3% H2O2/PBS for blocking of endogenous peroxidases, washed in PBST (0.05% Tween-20 in PBS) and incubated in TNB Blocking Buffer (Perkin Elmer). Sections were then incubated with primary antibodies at 4°C overnight. Nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI). At least three specimens of each genotype were used for each analysis.
In vivo cell proliferation rates of E12.5 and E14.5 cre-negative (control) and Tbx18cre/+;Tbx2fl/fl;Tbx3fl/fl ureters were assayed by detection of incorporated BrdU on 5-µm-thick sections (Bussen et al., 2004). Twelve sections of each specimen (n=5) were analyzed. The BrdU labeling index was defined as the number of BrdU-positive nuclei relative to the total number of nuclei detected by DAPI counterstaining in histologically defined compartments of the ureter. Statistical analysis was performed using the two-tailed Student's t-test. Values are indicated as mean±s.d. P<0.05 was considered significant. Apoptosis in tissues was assessed by TUNEL assay using the ApopTag Plus Fluorescein In Situ Apoptosis Detection Kit (Chemicon) on 5-µm-thick paraffin sections.
In situ hybridization analysis
Non-radioactive in situ hybridization analysis of gene expression was performed on 10-µm-thick paraffin sections of the proximal and distal ureter using digoxigenin-labeled antisense riboprobes as described previously (Moorman et al., 2001). At least three specimens of each genotype were used for each analysis.
Two independent pools each of control and mutant ureters were used for microarray analysis. Pool sizes were as follows: 20 ureters from E14.5 cre-negative and Tbx18cre/+;Tbx2fl/fl;Tbx3fl/fl embryos; 12, 24, 12 and 18 ureters from E18.5 male and female cre-negative, male and female Tbx18cre/+;Tbx2fl/fl;Tbx3fl/fl embryos, respectively; and 44 and 53 ureters from E13.5 male cre-negative and male Tbx18cre/+;HprtTBX2/y embryos, respectively. Total RNA from each pool was extracted using peqGOLD RNApure (PeqLab) and subsequently processed by the Research Core Unit Transcriptomics of Hannover Medical School. Whole Mouse Genome Oligo v2 (4×44K) Microarrays were used for E18.5 Tbx18cre/+;Tbx2fl/fl;Tbx3fl/fl and E13.5 Tbx18cre/+;HprtTBX2/y microarray analysis. 048306On1M_V2 microarrays were used for E14.5 Tbx18cre/+;Tbx2fl/fl;Tbx3fl/fl transcriptome analysis. Normalized expression data were filtered using Excel. Functional enrichment analysis for up- and downregulated genes was performed with DAVID 6.8 web-software (david.ncifcrf.gov) using default settings, and terms were selected based on P-value. P-values for the overlap of different gene sets were calculated using Fisher's exact test.
Chromatin immunoprecipitation (ChIP) analysis
ChIP was performed using the SimpleChIP Plus Enzymatic Chromatin IP Kit (Magnetic Beads) (9005, Cell Signaling). Approximately 100 dissected E14.5 ureters were treated according to the manufacturer's instructions. Single-cell suspensions were generated using a Minilys homogenizer with mixed 1.4/2.8 mm ceramic beads (Peqlab). The DNA-containing supernatants were incubated overnight with a rabbit anti-TBX2 (1:25; 07-318, Merck Millipore) antibody and collected on ProteinG Magnetic Beads (9006, NEB) with a magnetic MiniMACS Separator (130-042-102, Miltenyi Biotec). Quantification of ChIP-enrichment over 2% input control was performed by semi-quantitative PCR according to the manufacturer's instructions.
Sections were imaged using a Leica DM5000 microscope with Leica DFC300FX digital camera or a Leica DMI6000B microscope with Leica DFC350FX digital camera. All images were then processed in Adobe Photoshop CS4.
We thank Rolf Kemler for the Ctnnb1 floxed mouse line. Microarray data used in this publication were generated by the Research Core Unit Genomics (RCUG) at Hannover Medical School.
Conceptualization: N.A., A.K.; Methodology: N.A., C.R., M.-O.T., A.K.; Software: M.-O.T.; Validation: M.-O.T., T.H.L.; Formal analysis: N.A., C.R., M.-O.T., M.K., T.H.L.; Investigation: N.A., C.R., M.-O.T., M.K., T.H.L., A.K.; Resources: M.M.T., V.M.C., A.M., A.K.; Data curation: N.A., M.-O.T., A.K.; Writing - original draft: N.A., A.K.; Writing - review & editing: C.R., M.-O.T., M.K., T.H.L., M.M.T., V.M.C., A.M., A.K.; Visualization: N.A., M.-O.T., M.K., T.H.L., A.K.; Supervision: C.R., A.K.; Project administration: A.K.; Funding acquisition: A.K.
This work was supported by grants from the Deutsche Forschungsgemeinschaft (DFG KI728/8-2, KI728/10-1 to A.K.). N.A. was supported by the Hannover Biomedical Research School (HBRS) and the MD/PhD program Molecular Medicine.
Microarray data have been deposited in GEO under accession number GSE122561.
The authors declare no competing or financial interests.