Betaglycan Is Required for the Establishment of Nephron Endowment in the Mouse

Betaglycan is an accessory receptor for the transforming growth factor-β (TGFβ) superfamily, many members of which play key roles in kidney development. The purpose of this study was to define the role of this co-receptor on fetal murine kidney development. Stereological examination of embryonic and adult betaglycan heterozygous kidneys revealed augmented nephron number relative to littermate controls. Fetal heterozygous kidneys exhibited accelerated ureteric branching, which correlated with augmented nephron development at embryonic day (e) 15.5. In contrast, betaglycan null kidneys exhibited renal hypoplasia from e13.5 and reduced nephron number at e15.5. Quantitative real-time PCR analysis of e11.5–e14.5 kidneys demonstrated that heterozygous kidneys exhibited a transient decrease in Bmp4 expression at e11.5 and a subsequent cascade of changes in the gene regulatory network that governs metanephric development, including significant increases in Pax2, Eya1, Gdnf, Ret, Wnt4, and Wt1 expression. Conversely, gene expression in null kidneys was normal until e13.5, when significant reductions were detected in the expression of Bmp4 as well as other key metanephric regulatory genes. Tgfb1 and Tgfb2 mRNA expression was down-regulated in both nulls and heterozygotes at e13.5 and e14.5. The opposing morphological and molecular phenotypes in betaglycan heterozygote and null mutants demonstrate that the levels of betaglycan must be tightly regulated for optimal kidney development.


Introduction
The development of the metanephros and the subsequent establishment of nephron endowment begins when an outgrowth from the Wolffian duct known as the ureteric bud, extends towards and invades the metanephric mesenchyme [1]. Reciprocal interactions between the epithelial ureteric bud and the surrounding metanephric mesenchyme result in branching and extension of the ureteric tree, which generates the framework of the collecting duct system of the adult kidney. At the tips of the ureteric tree branches, mesenchymal condensation and mesenchymal-to-epithelial transition occur giving rise to nephrons [2].
The many processes involved in metanephric development are tightly regulated by complex molecular regulatory networks [3]. Members of the TGFb superfamily of signalling molecules have previously been shown both in vitro [4,5,6] and in vivo to regulate key aspects of metanephric development [7,8,9,10,11,12]. At embryonic day (e) 11.5, transcripts for the three TGFb isoforms (Tgfb1-3), the type I (Tgfbr1) and II (Tgfbr2) TGFb signalling receptors as well as transcripts for many other members of the TGFb superfamily are present in the mouse kidney [11]. Of the TGFb isoform null mouse lines, only Tgfb2 mutant mice exhibit defects in kidney development [6,12,13,14]. These include dysplastic ureteric tree formation, impaired glomerulogenesis and renal agenesis [6,12]. In contrast, Tgfb2 heterozygous mice exhibit accelerated metanephric development, which results in augmented nephron endowment at postnatal day (PN) 30 [6]. These data indicate that TGFb2 plays complex, non-redundant roles in the developing metanephros, and further understanding of the factors that control TGFb2 action in the fetal kidney is required.
Recent evidence suggests that the efficacy of TGFb isoforms in tissue development depends on the presence or absence of specific TGFb superfamily co-receptors such as the type III TGFb receptor (TGFBR3), commonly referred to as betaglycan [15,16]. Betaglycan presents ligand to the type II TGFb receptor, an action of particular importance for TGFb2, which does not bind to the type II TGFb receptor with high affinity alone [15,17,18]. Additional studies have highlighted a more complex role for betaglycan, which is now known also to directly bind inhibins and certain BMPs through its core, while indirectly influencing BMP and activin function through the regulation of inhibins [19,20].
To define the role of this co-receptor in metanephric development, we studied betaglycan heterozygous (betaglycan +/2 ) and null (betaglycan 2/2 ) mouse kidneys at the morphological and molecular levels. We show that betaglycan mutant kidney phenotypes are similar to the previously characterized Tgfb2 mutant phenotypes. Furthermore, we demonstrate that the opposing phenotypes in the betaglycan 2/2 and betaglycan +/2 kidneys arise from distinct alterations in the positive and negative gene regulatory networks which govern fetal kidney formation. Collectively, our data indicate that the TGFb2/betaglycan signalling pathway has dosage-dependent actions within the fetal kidney, and that this pathway must be tightly controlled for normal branching morphogenesis and nephrogenesis to occur.

Results
Betaglycan is expressed in the developing mouse kidney At e11.5, betaglycan was localised to the ureteric epithelium, with no specific immunostaining detected in the metanephric mesenchyme ( Figure 1A-B). By e14.5, betaglycan was detected in developing pre-glomerular structures while remaining strongly expressed in ureteric epithelium ( Figure 1C-E). At this developmental stage, expression was also observed in the interstitium ( Figure 1D) but was absent from the mesenchyme in the nephrogenic zone ( Figure 1E). At e16.5, betaglycan immunolocalization was similar to that seen at e14.5 ( Figure 1F-G). Betaglycan expression was not detected in the negative control, e14.5 betaglycan null kidney ( Figure 1H-J).

Metanephric signalling networks are disrupted in betaglycan mutant mice
Early Ureteric Branching Morphogenesis. BMP4 and GDNF, both members of the TGFb superfamily, play important roles in regulating ureteric budding and ureteric branching morphogenesis. Formation of the ureteric bud at approximately e11.0 and subsequent ureteric branching morphogenesis both require secreted GDNF from the metanephric mesenchyme to bind with its cognate receptor RET, as well as inhibition of BMP4 at the ureteric bud site and around the ureteric tips. In accordance with the increased ureteric branching observed in wholemount urogenital ridges, betaglycan +/2 mice exhibited significantly decreased Bmp4 expression at e11.5 (p,0.01) while Gdnf expression was significantly upregulated at e12.5 in comparison to age-matched wildtype metanephroi ( Figure 3A-B). These changes in betaglycan +/2 metanephroi were accompanied by consistent increases in Ret expression from e12.5 ( Figure 3C). In contrast, betaglycan 2/2 metanephroi displayed wildtype-like mRNA expression profiles for Bmp4, Gdnf and Ret at e11.5 ( Figure 3A-C). However, from e12.5-e14.5, significant decreases in the expression of both the inhibitory (Bmp4) and stimulatory pathways (Gdnf/Ret) controlling renal branching morphogenesis were detected in betaglycan 2/2 kidneys ( Figure 3A-C).

Regulation of the Ret and GDNF Expression
Profile. GDNF signalling through RET is viewed as the critical positive factor in branching morphogenesis [21,22]. Multiple studies have identified upstream transcriptional regulators of Gdnf such as Pax2, Eya1 and Six1 as integral to the induction of GDNF expression and in turn early metanephric development [23,24,25,26]. The most important of these factors appears to be Pax2, as Pax2 is not only critical for the initiation of Gdnf expression but also in the enhancement of Ret expression in the developing ureteric tree [23]. At e11.5, Pax2 expression, but not Eya1 or Six1 expression, was significantly increased in betaglycan +/2 mutants ( Figure 3D-F). At e12.5 and e13.5, betaglycan +/2 metanephroi displayed significant increases in both Pax2 and Eya1 mRNA compared with age-matched wildtype, with expression for both genes returning to wildtype levels at e14.5 ( Figure 3D-E). Betaglycan 2/2 metanephroi displayed wildtype-like expression profiles of Pax2 and Eya1 until e13.5, after which time significant decreases in expression of these genes was observed in betaglycan null metanephroi compared to age-matched wildtype metanephroi.
TGFb Isoforms. All three TGFb isoforms were detected in wildtype metanephroi from e11.5, with an upregulation in Tgfb1 and Tgfb2 expression detected at e13.5-e14.5 ( Figure 4). This upregulation in Tgfb1 and Tgfb2 mRNA was not observed in betaglycan mutant metanephroi, with both betaglycan +/2 and betaglycan 2/2 mutants displaying decreased expression levels of these genes compared to wildtype littermates (p,0.01; Figure 4A-B). Interestingly, the levels of Tgfb2 and betaglycan mRNA expression appeared to be directly correlated, as Tgfb2 expression at e13.5 and e14.5 was less in betaglycan +/2 mutants than in wildtype mice, and less again in betaglycan 2/2 mutants ( Figure 4B). Tgfb3 expression was similar in all three genotypes at all ages examined ( Figure 4C).
Body weight (g) 16 while BMP/GDF signalling predominantly utilizes SMAD1/5/8 [29,30]. As betaglycan can bind both TGFb and BMP ligands, the altered expression of the co-receptor has the potential to disrupt both pathways [20,31]. To determine whether TGFb or BMP signal transduction was altered in betaglycan mutants, phospho (p)SMAD1 and pSMAD3 immuno-stained cells were counted in e13.5 metanephroi from betaglycan wildtype and mutant embryos. This analysis illustrated that far fewer cells exhibited pSMAD1 than pSMAD3 in all three genotypes. However, no differences in the percentages of cells containing activated pSMAD1 or pSMAD3 were found between genotypes ( Figure 5A-B).

Discussion
Despite considerable functional redundancy amongst TGFb superfamily members in the developing kidney [11], morphological and molecular defects were detected in betaglycan +/2 and betaglycan 2/2 kidneys. Tightly controlled betaglycan expression is essential for normal metanephric development as partial or complete ablation of expression results in opposing renal pheno-types. In the betaglycan +/2 kidney, changes in gene expression were observed as early as e11.5. Notably, a decrease in Bmp4 expression at this time was observed in betaglycan +/2 kidneys compared with wildtype and betaglycan 2/2 kidneys. Although BMP4 appears to have multiple roles in kidney development [32], at the time of ureteric bud outgrowth, BMP4 activity in the sleeve of metanephric mesenchyme surrounding the Wolffian duct must be antagonized by an endogenous inhibitor, gremlin 1, for budding to occur [10,33]. While the timing of ureteric budding appears similar in the three genotypes, BMP4 can also act as an inhibitor of GDNF/RET/WNT-mediated branching morphogenesis [10]. Therefore, the reduced expression of Bmp4 at e11.5 is in accordance with the observed increase in ureteric branching at this time. Interestingly, the level of renal Bmp4 mRNA expression in betaglycan +/2 mice after this time was not different to that of age-matched wildtype mice. However, a cascade of changes in the gene regulatory network governing metanephric development followed in the betaglycan +/2 kidney, including significant increases in expression levels of Pax2, Eya1, Gdnf, Ret, Wnt4, and Wt1. Of these genes, Pax2 was the first to be upregulated in the betaglycan +/2 kidney at e11.5, in keeping with its role in inducing expression of downstream mediators of branching morphogenesis such as Gdnf and Ret [23,34].
In contrast to betaglycan +/2 kidneys, the rate of early branching morphogenesis in betaglycan 2/2 metanephroi resembled that of wildtypes. Accordingly, no differences in levels of gene expression were found between null and wildtype kidneys at e11.5. At e12.5, betaglycan 2/2 kidneys exhibited only small decreases in Ret expression compared to wildtypes. However, at e13.5, expression levels of several integral nephrogenic genes were significantly reduced in betaglycan 2/2 kidneys compared to wildtype, and these reductions correlated with the reduced kidney volume observed in betaglycan 2/2 embryos at this time. These findings indicate that a complete loss of betaglycan results in uncoupling of the reciprocal inductive signalling pathways that operate between the metanephric mesenchyme and ureteric epithelium to drive the progression of branching morphogenesis and nephrogenesis. In particular, genes that drive the Gdnf/Ret regulatory pathway, such as Pax2, fail to be upregulated in the betaglycan 2/2 kidney, which may be a possible mechanism for the development of the hypoplastic renal phenotype at e13.5.
Importantly, the expression of Six1, an upstream regulator of Gdnf, of similar hierarchal order to Eya1, remained similar between all genotypes through out all developmental stages analysed. Previously, direct interactions between Eya1 and Six1 have been shown to be essential for the induction and upregulation of Gdnf expression in the MM [26,35]. As Eya1, and not Six1 expression was altered in betaglycan mutant metanephroi, it is possible to suggest that the influence of betaglycan on downstream Gdnf expression may be elucidated through perturbations to Eya1 expression and not that of Six1. It is also possible that the conserved Six1 expression in betaglycan mutants may be responsible for sufficient Gdnf expression induction to prevent the manifestation of more severe renal phenotypes such as renal agenesis in betaglycan 2/2 embryos [35].
Finally, loss or reduction of betaglycan expression resulted in decreased Tgfb1 and Tgfb2 expression in the fetal kidney. This may be due to reduced TGFb sensitivity in the betaglycan mutant kidneys as TGFbs are well-recognized to auto-induce their own expression by both transcriptional and post-transcriptional mechanisms [36,37]. A loss or reduction in betaglycan on the cell surface may lead to decreased sensitivity to the TGFbs, which could in turn reduce the level of TGFb auto-induction and compound the effects of the loss of betaglycan. Indeed, in betaglycan 2/2 kidneys, the collective reduction of several TGFb superfamily inputs (TGFb1, TGFb2, BMP4) coupled with the complete absence of betaglycan, a key determinant of cellular TGFb sensitivity, may result in sub-threshold levels of TGFb superfamily signalling, which may explain why betaglycan 2/2 kidneys exhibit hypoplasia rather than an even greater rate of kidney development than seen in betaglycan +/2 mice.
Is betaglycan essential as a TGFb co-receptor in the fetal kidney?
As fetal metanephroi express several potential ligands of betaglycan, the specific mechanisms by which diminished betaglycan expression impacts kidney development are not yet clear. However, several pieces of evidence suggest that betaglycan is serving an essential role as a TGFb2 co-receptor during early kidney development. Firstly, betaglycan expression is coincident with the previously published expression pattern of TGFb2 in murine metanephroi [4,11,38]. Betaglycan was localized in the ureteric epithelium from the earliest stages of ureteric bud formation, and from e14.5 was also detected in metanephric mesenchyme and the epithelium of developing nephrons. TGFb2 is similarly expressed by the ureteric epithelium, with expression also detected in neighbouring tissues after e14.5 [38]. The expression of betaglycan in the ureteric epithelium at e11.0-e11.5 and  the increased branching of the ureteric tree at e11.5-e12.0 suggests that TGFb2 initially acts directly upon ureteric epithelium to inhibit branching. Subsequent expression of betaglycan in interstitial cell populations from e14.5 suggests that after this age, TGFb2/betaglycan signalling is involved in the development of both the epithelial and interstitial cell populations of the metanephros.
Secondly, betaglycan and Tgfb2 mutant renal phenotypes share several similarities. Like betaglycan +/2 , Tgfb2 heterozygotes develop normally and possess more nephrons than wildtype mice at PN30, displaying an even greater enhancement of nephron number than betaglycan +/2 mice [6,39]. This suggests that TGFb2/betaglycan signalling is required to negatively regulate nephrogenesis, possibly through the regulation of ureteric branching morphogenesis. This observation is in agreement with previous work that suggests abrogation of endogenous betaglycan in vitro attenuates endogenous autocrine and/or paracrine TGFbmediated negative regulation of lung branching morphogenesis [40]. In contrast to the renal phenotype in betaglycan +/2 mice, in vivo stereological analysis and in vitro culture of betaglycan 2/2 metanephroi revealed a hypoplastic renal phenotype that closely resembled the defective development of Tgfb2 null explants [6]. These contrasting phenotypes in betaglycan +/2 and betaglycan 2/2 kidneys suggest that the TGFb2/betaglycan signalling pathway plays dual roles in vivo, both limiting the growth of the ureteric tree while supporting the maintenance of the newly-formed nephrons.
The contrasting and complex renal phenotypes in betaglycan +/2 and betaglycan 2/2 mice suggest that TGFb2/betaglycan signalling is required at more than one stage of kidney development and in more than one cell type, with specific developmental processes in the kidney having higher thresholds for TGFb2 signalling than others. Specifically, the data suggest that greater TGFb2 signalling is necessary to limit early branching of the ureteric tree, and that betaglycan is an essential component of the TGFb2 signalling complex that allows critical thresholds of TGFb2 signalling to be met during this process. In contrast, TGFb2-mediated survival of ureteric epithelium during late gestation appears to occur at a lower threshold of TGFb2 signalling. This conclusion is based on the observation that newly-formed epithelial structures are maintained when the level of either the ligand or the accessory receptor is reduced, as in Tgfb2 heterozygotes or betaglycan +/2 mice, respectively. Conversely, ablation of TGFb2 signalling in Tgfb2 null metanephroi results in degeneration of these newly-formed structures, an effect that could be hypothesized to be due to insufficient pro-survival signalling attributed to TGFb2 [12,41,42]. As degeneration of renal tissues was not observed in betaglycan 2/2 metanephroi, this suggests that the relatively low, but still present TGFb2 signalling in betaglycan 2/2 metanephroi is sufficient to satisfy a threshold necessary for cell survival. This further suggests that the processes of ureteric branching morphogenesis and cell/ tissue survival have differing TGFb2 concentration-dependent signalling thresholds.
The betaglycan mutant phenotypes largely resemble their Tgfb2 mutant counterparts, suggesting that a major role for betaglycan in the fetal murine kidney is as a determinant of TGFb2 efficacy. Furthermore, the opposing phenotypes in both Tgfb2 and betaglycan, null and heterozygote kidneys indicate that the TGFb2/betaglycan signalling pathway has dose-dependent actions and that activity of this pathway must be tightly controlled for normal branching morphogenesis and nephrogenesis to occur. Molecular analysis of betaglycan mutant kidneys indicated that the distinct morphological phenotypes in the betaglycan +/2 and betaglycan 2/2 kidneys were mirrored at the molecular level. In particular, specific temporal changes in the expression of certain TGFb superfamily members (for example Bmp4 during the initiation of ureteric branching morphogenesis in betaglycan +/2 metanephroi) may underlie the contrasting phenotypes. These findings may be of considerable importance to human health in that renal hypodysplasia is a recognized cause of paediatric renal failure and a risk factor for hypertension in adults [43].

Animal Models
All animal handling and experimental procedures were approved by the Monash Medical Centre Animal Ethics Committee (Study approval ID B08/26). Wildtype, betaglycan +/2 and betaglycan 2/2 embryos were obtained from timed matings of betaglycan +/2 mice, and genotyped as previously described [15,44]. Embryo development was verified via Theiler staging criteria and the total number of tail somites present from tail to hind limb [45]. Kidneys were dissected from e11.5-e16.5 embryos, processed and fixed as described below. To assess postnatal renal morphology, kidneys were collected from wildtype and betaglycan +/2 PN4, PN5, PN6 and PN30 mice, immersion-fixed in 10% neutral buffered formalin, processed and embedded in paraffin wax. Sections (5 mm) were stained with haematoxylin and eosin, periodic acid Schiff's, Masson's trichrome or silver methenamine histological stains as indicated, and assessed in a blind manner.

Estimation of nephron number in developing kidneys
To determine the total number of developing nephrons in embryonic kidneys, metanephroi collected from e15.5 mouse embryos were immersion fixed in 4% PFA for 30 min, embedded in paraffin wax and exhaustively sectioned at 5 mm. All sections were blocked in 2% bovine serum albumin (BSA) in PBS with 0.3% Triton-X for 30 mins at room temperature. Sections were then digested with 0.1% neuraminidase (Sigma Aldrich) in 1% CaCl 2 in PBS, and incubated overnight at 4uC with biotinylated peanut agglutinin (PNA; Sigma-Aldrich) diluted at 20 mg/ml in 2% BSA/ PBS and 0.3% Triton-X with 1 mmol/L CaCl 2 /MnCl 2 /MgCl 2 . Following incubation, sections were rinsed twice with PBS. The biotinylated PNA was visualized using the Elite streptavidin/biotin amplification ABC kit (Vector Laboratories Inc., Burlingame, CA, USA). The ABC solution was applied for 1 hour at room temperature, after which slides were washed and the stain developed with diaminobenzidine (DAB; Sigma Aldrich)/0.01% H 2 O 2 /PBS. Sections were counterstained with haematoxylin.
Section pairs were used to estimate PNA-positive developing nephrons using the physical disector/fractionator combination [49]. PNA-positive structures (which included all developing nephrons from early S-shaped bodies to developing glomeruli) were marked in the first (n) section. The nth+5 section was then viewed, with positive structures present in nth and not the nth+5 section counted. Those counted in the nth+5 section and not in the nth section were also counted to double the efficiency of the technique. This process was repeated with the first nominated nth+5 section denoted the nth section and a new nth+5 section selected. Total number of developing nephrons was determined as the total number of PNA-positive structures observed (Cullen-McEwen et al., submitted).

Estimation of ureteric epithelial volume
Ureteric epithelial volume was determined in e13.5 metanephroi. Paraffin-embedded metanephroi were exhaustively sectioned at 5 mm, after which all sections were dewaxed, rehydrated and blocked in 2% BSA in PBS with 0.3% Triton-X for 30 min at room temperature. Sections were then incubated overnight at 4uC with biotinylated dolichos biflorus (DBA; Sigma-Aldrich). DBApositive structures were visualized using the Elite streptavidin/ biotin amplification ABC kit (Vector Laboratories Inc, CA, USA). Unbiased stereology was then used to determine total and relative epithelial volumes, as well as total metanephric volumes [46].

Estimation of total nephron number at PN30
Total nephron number was determined in female betaglycan heterozygous and wildtype kidneys at PN30. Body weights were recorded and kidneys dissected, cleaned of fat and blood vessels, decapsulated and weighed. Kidneys were immersion-fixed in 10% neutral buffered formalin, and processed for embedding in glycolmethacrylate. Kidney volume, total glomerular number, mean and total glomerular and mean and total renal corpuscle volumes were estimated as described previously [6,49,50], with total nephron number estimated using the unbiased physical disector/fractionator combination.

Quantitative real-time PCR
Total RNA was isolated from pooled e11.5-e14.5 metanephroi and complementary DNA was synthesized as previously described [51]. All findings were observed across four separate, independently collected pools of kidneys for each genotype and age (n = 4-6/pool). Real-time PCR was performed using the Applied Biosystems ABI 7900 HT Fast real-time machine with all reactions performed in triplicate. Yields were converted to femtograms and normalized to the 18S mRNA level per sample, as previously described [51]. Primers and their optimized annealing tempera- tures are listed in Table 3. Cycling conditions comprised 10 min denaturation at 95uC, then 55 cycles of denaturing (95uC for 15 sec), annealing (5 sec) and elongation (72uC for 8 sec and 79uC for 8 sec). PCR products were detected at 16-30 cycles. All PCR products were initially verified by sequencing.

Statistical analysis
Unpaired Student's t-tests, one-way or two-way ANOVAs, and Tukey's post-hoc analyses were conducted using GraphPad Prism TM 5 (Software Inc, 1992-2007. Values are mean 6 SD. A probability of 0.05 or less was accepted as statistically significant.