Role of fibroblast growth factor receptor 2 in kidney mesenchyme

Abstract

Conditional deletion of murine fibroblast growth factor receptors (Fgfrs) 1 and 2 in metanephric mesenchyme leads to renal agenesis with unbranched ureteric buds; however, there are occasionally 2 buds per nephric duct. Our goal was to determine whether conditional deletion of Fgfr1 or Fgfr2 alone resulted in multiple ureteric bud induction sites. While deletion of Fgfr1 alone results in no abnormalities, loss of Fgfr2 often leads to multiple ureteric buds and anomalies including renal aplasia, misshaped kidneys, partially duplicated kidneys, duplicated ureters, and obstructed hydroureter. Deletion of Fgfr2 did not change expression domains of glial cell line-derived neurotrophic factor (Gdnf), Robo2, bone morphogenetic protein 4, or Sprouty1, all of which regulate ureteric bud induction. Cultured Fgfr2 mutant nephric ducts were also not more sensitive to exogenous GDNF than controls. Whole mount in situ hybridization revealed that in mutant embryos, Fgfr2 was deleted from stromal cells around the nephric duct and ureteric bud base, which correlates well with the ureteric bud induction abnormalities. Thus, Fgfr2 is critical in ensuring that there is a single ureteric bud from the nephric duct. The plethora of later stage defects in Fgfr2 conditional knockouts is reminiscent of many human cases of genetic urogenital anomalies.

Fibroblast growth factor receptors (Fgfrs) are receptor tyrosine kinases with 4 known signaling members and 22 ligands in mammals (1). Fgfrs are expressed throughout embryogenesis in many tissues including the kidney (1).

The metanephric kidney arises from two embryonic tissues, the metanephric mesenchyme and ureteric bud. At embryonic day (E) 10.5 in the mouse and the 5^th week of gestation in humans, the metanephric mesenchyme induces a single ureteric bud from the Wolffian (nephric) duct near the hind limb (2). Subsequently the ureteric bud elongates and branches within the metanephric mesenchyme, giving rise to the collecting ducts, pelvis, and ureter (2). At its tips, the ureteric bud induces local metanephric mesenchyme to condense and differentiate into nephron epithelia. Regions of stromal mesenchyme also surround the developing nephrons.

Many studies have documented diverse actions of Fgfrs in developing kidneys. Addition of Fgf2 to isolated rat metanephric mesenchymal tissues prevented apoptosis (3, 4) and promoted condensation (4). Transgenic mice with a dominant negative Fgfr fragment developed renal agenesis/severe dysgenesis (5). Mice null for Fgf7, Fgf10, or Fgfr2-IIIb (the receptor isoform for Fgf7 and Fgf10), have small kidneys with normal-appearing nephrons (6-8). Mice with conditional deletion of Fgf8 from the metanephric mesenchyme, however, have interrupted nephron development (9, 10).

While Fgfr1 and Fgfr2 null mice are early embryonic lethal (11-14), conditional knockout approaches have revealed roles for these receptors in kidney development. We observed that deletion of Fgfr2 from the ureteric bud results in aberrant ureteric bud and renal stromal patterning (15). We also demonstrated that loss of both Fgfr1 and Fgfr2 from the metanephric mesenchyme (Fgfr1/2^Mes-/-) leads to renal agenesis (16). Although Fgfr1/2^Mes-/- embryos had unbranched ureteric buds, there are occasionally two buds from one nephric duct (16). Our goal was to determine whether there were ureteric bud induction abnormalities in either Fgfr1^Mes-/- or Fgfr2^Mes-/- mice and subsequent developmental consequences.

Methods

Fgfr2^Mes-/- and Fgfr2^UB-/- conditional knockout mice

This study was approved by the Research Institute at Nationwide Children’s Hospital Institutional Animal Care and Use Committee. The conditional knockout mice have been described previously (15, 16). Briefly, transgenic mice with the Pax3 promoter driving cre recombinase expression in metanephric mesenchyme (Pax3cre^Tg/+) (gift from Dr. Jon Epstein, University of Pennsylvania, Philadelphia) were crossed with mice possessing lox-p sites that flank a critical portion of Fgfr2 (Fgfr2^Lox/Lox) (gift from Dr. David Ornitz, Washington University, St. Louis). Transgenic mice with a portion of the HoxB7 promoter driving cre recombinase expression (Hoxb7cre^Tg/+) were also crossed with Fgfr2^Lox/Lox mice. Progeny with cre and 2 lox-p sites had functional deletion of Fgfr2 from the metanephric mesenchyme (Pax3cre^Tg/+/Fgfr2^Lox/Lox) or the nephric duct and ureteric bud (Hoxb7cre^Tg/+/Fgfr2^Lox/Lox), respectively. Littermates with all other genotypes were phenotypically normal and used as controls.

Genotyping

Tail clippings and/or embryonic heads were used to extract DNA for polymerase chain reaction genotyping, using primers (see Table 1) as described (15, 16).

Table 1.

PCR primers used for genotyping

Gene	Primers	Size (base pairs)
Pax3cre	5′- AATCTTATGGTCACCTGAGTGTTAAATGTCCAATTTAC-3′	230
	5′- CATCTTCAGGTTCTGCGGG-3′
Fgfr2	5′-GTCAATTCTAAGCCACTGTCTGCC-3′	307 (wt)
	5′-CTCCACTGATTACATCTAAAGAGC-3′	373 (floxed)

Histology

E16.5 Fgfr2^Mes-/- and control littermates were dissected, fixed in 4% paraformaldehyde in phosphate buffered saline overnight at 4°C, and then embedded in paraffin. 4-10 μm sections were stained with hematotxylin and eosin. Whole-mount in situ hybridization was performed on E10.5 to E11.5 Fgfr2^Mes-/- and control embryos and/or dissected kidneys as described previously (16). Digoxigenin-UTP-labeled antisense and sense RNA probes were generated against multiple targets (Table 2). The Sprouty1 probe template was published previously (17). Some whole mount in situ hybridization-stained tissues were re-embedded in paraffin and sectioned at 10 μm. In addition, radioactive in situ hybridization was performed on E11.5 Fgfr2^Mes-/- and control embryos with probes against Fgfr2 as described (16).

Table 2.

Probe templates used for in situ hybridization

Probe	GenBank Accession #	Nucleotide #’s
Ret	BC059012	3205-3688
Gdnf	NM010275	2894-3428
Pax2	NM011037	450-937
Fgfr2	X55441	1594-1941
BMP4	D14814	8378-8605

Organ cultures

E10.5 Fgfr2^Mes-/- and control littermate nephric ducts and surrounding mesenchyme were dissected and grown on nucleopore filters floating on Dulbecco’s modified Eagle medium with 10% fetal bovine serum at 37°C for 3 days. In addition to serum only (N=24 control and N=15 nephric ducts), recombinant human GDNF (R&D Systems) was added at 25 ng/ml (N=16 control, N=8 mutant) and 50 ng/ml (N=18 control and N=10 mutant) as described (18). Tissues were fixed in methanol and immunofluorescence was performed against Calbindin-D_28k (Sigma Chemical Co.-Aldrich) to visualize nephric ducts and ureteric buds, as described (19).

Microscopy and photography

Gdnf and Robo2 expression domains were measured for long axis length and planar surface area, and Gndf domains were assessed for maximum width and length of posterior limit of signal to midpoint of hind limb using ImageJ (version 1.32j from Wayne Rasband, NIH, USA). All images were captured on an AxioCam digital camera (Zeiss, Thornwood, NY) mounted on a DM LB microscope (Leica, Wetzlar, Germany) and then converted to Adobe Photoshop files.

Results

Fgfr2^Mes-/- mice have ureteric bud induction abnormalities resulting in many aberrant phenotypes

To determine whether deletion of Fgfr1 or Fgfr2 from the metanephric mesenchyme causes ureteric bud induction abnormalities, we performed whole mount in situ hybridization on E10.5-E11.5 embryos for Ret, a marker of the nephric duct and ureteric bud. We detected no abnormalities in Fgfr1^Mes-/- mice (not shown) or in controls (Figure 1a, c). In contrast, 67% (8/12) of the Fgfr2^Mes-/- embryos had two ureteric buds from one nephric duct (50% unilateral and 17% bilateral) (Figure 1b, d, e). No Fgfr2^Mes-/- mice had more than 2 ureteric buds per side. At E10.5, the affected Fgfr2^Mes-/- mice possess one ureteric bud at or near the normal position and one that is anterior (cranial) along the nephric duct (Figure 1b). By E11.5, one of the two ureteric buds on one side had elongated and branched within the metanephric mesenchyme, while the other was shorter with blunted branching, and/or had an inappropriate trajectory (Figure 1d, e). Thus, a majority of Fgfr2^Mes-/- embryos had two ureteric buds from one nephric duct, although only one of the two made good contact with the metanephric mesenchyme.

Figure 1. Whole mount in situ hybridization for Ret in Fgfr2^Mes-/- and control embryos.

a. Control E10.5 embryo with nephric duct (nd) and single ureteric bud (ub) at its normal position. b. Fgfr2^Mes-/- E10.5 embryo with two ureteric buds from one nephric duct, including an anterior bud (asterisk) and a more properly positioned bud. c. Control E11.5 embryo with a single normally positioned ureteric bud that has elongated and branched. d. Fgfr2^Mes-/- E11.5 embryo with elongation and branching of the anterior ureteric bud (asterisk) and minimal elongation and no branching of the more normally-positioned bud. e. Fgfr2^Mes-/- E11.5 embryo with a right-sided (R) single, properly-positioned ureteric bud, but a left-sided (L) anterior bud that has elongated and branched (asterisk) and a more normally-positioned bud has elongated less and has not appreciably branched (also note that the trajectory of the left anterior bud appears more appropriate than the more normally-positioned bud). Scale bars = 100 μm.

We then examined E16.5 embryos to determine how the ureteric bud induction abnormalities affected later stages of renal development. Approximately 37% (4 of 11) Fgfr2^Mes-/- embryos examined had abnormalities (50% bilateral and 50% unilateral). Phenotypes included duplex ureters (N=2 complete, N=1 partial), hydroureter (N=3), partial duplex kidneys (N=2), misshaped kidneys (N=2), and renal agenesis (N=1) (Figures 2, 3 and not shown). Fgfr2^Mes-/- mice with bilateral abnormalities usually had different urogenital phenotypes on either side (Figures 2 and not shown). To determine whether mutants with hydroureter had aberrant bladder insertion, we obtained serial cross sections of ureters from the renal pelvis to the bladder. One female Fgfr2^Mes-/- embryo had a hydroureter (Figure 3a, arrowhead) and an ipsilateral more normal ureter that were both attached to a partially duplex kidney (the hydroureter to the small, abnormally-positioned “kidney” and the normal ureter to appropriately contoured “kidney”). In caudal sections (Figure 3b), the hydroureter ends in the paramesonephros while the normal ureter inserts into the bladder (not shown). Thus, at later ages, a number of Fgfr2^Mes-/- embryos display a wide range of urogenital abnormalities.

Figure 2. Transverse H & E sections of E16.5 control and Fgfr2^Mes-/- kidneys.

a, b. Control embryo with normal caliber ureters (a, ur) that empty into a renal pelvis with a normal papilla (b, arrowhead). c, d. Fgfr2^Mes-/- embryo with a massive hydroureter on one side (c, ur-one line) that empties into a flattened renal pelvis (d, concave arrowhead) and duplicated ureters on the other side (c, ur-two lines) that coalesce into a single ureter entering the kidney (d, ur). Scale bar = 250 μm

Figure 3. Transverse H&E sections of an E16.5 Fgfr2^Mes-/- female embryo from the renal pelvis (a) and the bladder (b).

a. Partial duplex kidney with a hydroureter (arrowhead) draining a smaller, misshapen kidney moiety (k’) fused to a more normal-appearing kidney (k). b. The hydroureter ends blindly in a fibrotic mass in the paramesonephros near the bladder (bl). Scale bar = 250 μm.

Gdnf and Robo2 expression is unaltered in Fgfr2^Mes-/- mutants

Some mouse models have multiple ureteric bud induction sites from shifted and/or expanded expression domains of glial cell line-derived neurotrophic factor (Gdnf) in the metanephric mesenchyme (20, 21). Since Fgfr2 is deleted from the metanephric mesenchyme in Fgfr2^Mes-/- mice (16), it is plausible that altered Gdnf expression could explain the abnormalities in Fgfr2^Mes-/- mice. Thus, we performed whole mount in situ hybridization for Gdnf in Fgfr2^Mes-/- and control E10.5 embryos. Both controls and mutants had “tear-shaped” Gdnf expression patterns with antisense probes (Figure 4a, b, respectively), and neither had signal with sense probes (not shown). To compare the expression domains of Gdnf in control and Fgfr2^Mes-/- embryos, we calculated the maximal width (A), area (B), long axis length (C), and distance from the posterior expression limit to the midpoint of the hind limbs (D) (Figure 4c). We found no differences in the size or position of the Gdnf signal between control (N=10) and Fgfr2^Mes-/- mutants (N=12) (Table 3). Thus, Gdnf expression domain is unaltered in Fgfr2^Mes-/- embryos.

Figure 4. Whole mount in situ hybridization for Gdnf in E10.5 Fgfr2^Mes-/- and control embryos.

a-b. Control embryo (a) and Fgfr2^Mes-/- embryo (b) have similar Gdnf expression in the metanephric mesenchyme (mm) and position relative to the midpoint of the hind limb (lines). c. Diagram showing the measurements obtained for the Gdnf expression domain, including: area (A), long axis length (B), maximal width (C), and length of posterior limit of signal to midpoint of hind limb. Scale bar = 250 μm.

Table 3.

Measurements of Gdnf expression domains in Fgfr2^Mes-/- and control embryos

Gdnf measurement	Control (μm)	Mutant (μm)	p-value
A. Maximal width (mean ± SD)	174 ± 18.5	164 ± 31.3	0.45
B. Area (mean ± SD)	81.5 ± 15.2	79.1 ± 20.5	0.78
C. Long axis length (mean ± SD)	613 ± 129	622 ± 127	0.87
D. Posterior to mid hindlimb (mean ± SD)	1377 ± 206	1335 ± 156	0.62

Given that alterations in Robo2 expression have been reported in models of aberrant ureteric bud induction, we performed whole mount in situ hybridization against Robo2 in Fgfr2^Mes-/- and control E10.5 embryos (20). As with Gdnf, we detected no differences between control and mutant long axis length (766μm ± 95.8μm vs. 737μm ± 28.1μm, p=0.32) or surface area (86.778μm² ±13.6μm² vs. 83.9μm² ± 15.4μm², p=0.60) (Figure 5).

Figure 5. Whole mount in situ hybridization for Robo2 in E10.5 Fgfr2^Mes-/- and control embryos.

Control (a) and Fgfr2^Mes-/- embryos (b) have similar Robo2 expression patterns in the metanephric mesenchyme (mm). Scale bar = 250 μm.

Fgfr2 is deleted from stromal mesenchymal cells adjacent to nephric ducts and main ureteric bud trunks in Fgfr2^Mes-/- mice

Other mouse lines develop multiple ureteric bud induction sites secondary to perturbations in the stromal cells surrounding the nephric duct and the main ureteric bud trunk (22). Thus, we determined whether Fgfr2 is normally expressed in these stromal cells and whether expression is deleted in Fgfr2^Mes-/- mice by whole mount in situ hybridization in isolated E11.5 kidney tissues. In controls (N=5 embryos), we detected Fgfr2 in a wide band that appears to be present in the nephric duct and ureteric bud trunk epithelia, and the surrounding stromal mesenchymal cells (Figure 6a). In Fgfr2^Mes-/- tissues (N= 5 embryos), however, Fgfr2 is expressed in a much more thin band, that appears to be localized to the nephric duct and ureteric bud trunk epithelium only (and absent from the stromal mesenchymal cells) (Figure 6b). We next examined Fgfr2 expression in Hoxb7cre^Tg/+/Fgfr2^Lox/Lox kidney tissues (N= 8 embryos), which selectively deletes Fgfr2 from the ureteric bud and nephric duct epithelium (15). In Hoxb7cre^Tg/+/Fgfr2^Lox/Lox tissues, we detected linear Fgfr2 expression only in stromal cells on either side of the nephric duct and ureteric bud epithelia (Figure 6c). There was no signal in tissues incubated with sense probes (not shown).

Figure 6. Whole mount in situ hybridization for Fgfr2 in E11.5 control, Fgfr2^Mes-/- and Fgfr2^UB-/- isolated kidneys.

a. Control with a broad band of Fgfr2 expression that appears to be within the epithelium in and adjacent stromal cells near in the nephric duct (nd) and the ureteric bud (ub) at the induction site. b. Fgfr2^Mes-/- kidney with a narrow band of Fgfr2 signal that appears confined to the epithelial cells of the tissues at the induction site. c. Fgfr2^UB-/- kidney with linear Fgfr2 expression in stromal cells on either side of the nephric duct and ureteric bud near the induction site. Scale bar = 100 μm.

To confirm the whole mount data, we performed radioactive in situ hybridization for Fgfr2 on E10.5 control and Fgfr2^Mes-/- tissue sections. In controls, Fgfr2 is expressed in metanephric mesenchyme, ureteric bud ampullae, the ureteric trunk, and in stromal tissue surrounding ureteric trunk epithelium (Figure 7a, a’, arrow). In Fgfr2^Mes-/- embryos, however, Fgfr2 is present in ureteric epithelium within the metanephric mesenchyme and the early trunk exiting the mesenchyme as well as the nephric duct, but not in metanephric or stromal mesenchyme) (Figure 7c, c’). Near the ureteric trunk and the nephric duct junction, Fgfr2 is expressed in the epithelia and the stromal cells between the epithelia in controls (Figure 7b, b’, arrowhead), whereas it is present only in the epithelial tissues in mutants and not the stromal strip (Figure 7d, d’, arrowhead). Taken together, Fgfr2 is normally expressed in stromal cells adjacent to the nephric duct and main ureteric bud trunks and is deleted from these mesenchymal cells in Fgfr2^Mes-/- mice.

Figure 7. In situ hybridization for Fgfr2 in E11.5 control and Fgfr2^Mes-/- tissue sections.

a, a’. Brightfield (a) and darkfield (a’) images of controls with Fgfr2 expression in metanephric mesenchyme (MM), ureteric bud (UB) ampulla and trunk and in stromal cells adjacent to the trunk (arrow). b, b’. Control near junction of ureteric bud trunk and nephric duct (ND) showing Fgfr2 signal in both epithelium and stromal tissue between epithelia (arrowhead). c, c’. Fgfr2^Mes-/- revealing Fgfr2 is expressed in the ureteric bud within metanephric mesenchyme and early trunk leaving the kidney and in nephric duct, but excluded from metanephric and stromal mesenchyme. d, d’. Fgfr2^Mes-/- embryo near junction of ureteric bud trunk and nephric duct shows Fgfr2 signal is present in the epithelium, but excluded from the stromal cells in between. Scale bar = 25 μm.

Bmp4 expression is unaltered in Fgfr2^Mes-/- mice

Since Fgfr2 appears to be deleted in the aforementioned mesenchymal cells in Fgfr2^Mes-/- mice, we examined whether the mutants had altered expression of stromal Bmp4, a known antagonist of ureteric bud induction (22). In E11.5 kidney tissues, both control (N=19 embryos) and Fgfr2^Mes-/- (N=8 embryos) displayed linear Bmp4 signal in the stromal cells adjacent to the nephric duct and the ureteric bud (Figure 8), including mutants with 2 ureteric buds arising from one nephric duct (not shown). We observed no signal in tissues incubated with sense probe (not shown). Thus, Bmp4 expression appears to be unaltered in Fgfr2^Mes-/- embryonic kidneys.

Figure 8. Whole mount in situ hybridization for Bmp4 in E11.5 control and Fgfr2^Mes-/- isolated kidneys.

a. Control kidney with linear Bmp4 expression in stromal cells around the nephric duct (nd) and ureteric bud main trunk (ub). b, c. Lower magnification (b) and higher magnification (c) of Fgfr2^Mes-/- kidneys with similar linear Bmp4 signal in stromal mesenchyme adjacent to the nephric duct and ureteric bud. d. Diagram of Bmp4 expression in kidneys. Scale bar = 100 μm.

Sprouty1 expression is not altered in Fgfr2^Mes-/- mice

As in Fgfr2^Mes-/- mice, Sprouty1 null mice develop multiple ureteric bud induction sites without alterations in either Gdnf or Bmp4 expression (17). Given that Sprouty1 transcription is also stimulated by Fgfr signaling, we performed whole mount in situ hybridization for Sprouty1 on isolated kidney tissues from E11.5 control (N=17) and Fgfr2^Mes-/- (N=10) embryos. In both controls and mutants, Sprouty1 signal appears to be present in the stromal cells surrounding the nephric duct and main ureteric bud trunk (Figure 9a, b). To determine whether there was differential expression in the kidney itself, we sectioned and examined the stained tissues. In both control and Fgfr2^Mes-/- kidneys, we observed intense Sprouty1 signal in the ureteric bud tips, and weaker signal in the surrounding metanephric mesenchyme (Figure 9c, d). We detected no signal with sense probe (not shown). Thus, Sprouty1 expression appears to be unaltered in Fgfr2^Mes-/- mice.

Figure 9. Whole mount in situ hybridization for Sprouty1 in E11.5 control and Fgfr2^Mes-/- isolated kidneys.

a, b. Control (a) and Fgfr2^Mes-/- kidney (b) both with linear Sprouty1 expression in stromal cells around the nephric duct (nd) and ureteric bud main trunk (ub). c, d. Sections through whole mount stained control (c) and Fgfr2^Mes-/- kidneys (d) both with strong Sprouty1 expression in ureteric bud ampullae (ub) and weaker expression in surrounding metanephric mesenchyme (arrowheads). Scale bars = 100 μm.

Fgfr2^Mes-/- cultured nephric ducts are not more sensitive to exogenous GDNF

Since others have reported mutant mouse lines with increased sensitivity of cultured nephric ducts to exogenous GDNF (17), we examined effects of GDNF on E10.5 kidney primordia of control and Fgfr2^Mes-/- mice. Both control and mutant nephric ducts had similar dose-dependent responses to GDNF. At 25ng/ml, controls had means of 2.73 ureteric buds/mm ± 1.12 buds/mm vs. mutants with 2.29 buds/mm ± 1.31 buds/mm (p=0.41) (not shown) and at 50 ng/ml, controls averaged 3.97 buds/mm ± 1.83 buds/mm vs. mutants with 3.83 buds/mm ± 1.03 buds/mm (p=0.91) (Figure 10). Thus, Fgfr2^Mes-/- nephric ducts do not appear to be more sensitive to GDNF than controls.

Figure 10. Response of control and Fgfr2^Mes-/- E10.5 cultured nephric ducts to exogenous GDNF for 3 days.

a-b. Both control (a) and Fgfr2^Mes-/- (b) nephric ducts show single branched ureteric buds from nephric ducts with serum only. c-d. Control (c) and Fgfr2^Mes-/- (d) nephric ducts develop similar numbers of ectopic ureteric buds in response to 50 ng/ml of exogenous GDNF. Scale bar = 25 μm.

Discussion

Previous reports show that the Fgfr family functions in different stages and lineages of renal development (3-10). Among the Fgf receptors, Fgfr2 regulates ureteric bud and renal stromal patterning (from its expression in the ureteric bud) and in early metanephric mesenchymal development (secondary to its actions with Fgfr1 in the renal mesenchyme) (15, 16). This report documents another role for Fgfr2 in the metanephric mesenchyme; i.e., ensuring that there is a single, properly positioned ureteric bud from the nephric duct. Moreover, Fgfr2^Mes-/- mice frequently develop multiple ureteric bud induction sites resulting in urinary tract anomalies including renal aplasia, misshaped kidneys, partially duplicated kidneys, duplicated ureters, and hydroureter with obstructed insertion into the ductus deferens or paramesonephros.

Fgfr2^Mes-/- mice are different from many other models with multiple ureteric buds in that there is such a wide spectrum of later urinary tract abnormalities and that the incidence of multiple ureteric buds (67%) appears to be greater than the later renal anomalies (37%). Regarding the latter point, it may be that with higher numbers of Fgfr2^Mes-/- embryos analyzed, the incidence of early and later phenotypes would be the similar. Another plausible explanation for both findings, however, is that the two ureteric buds “compete” for contact with the metanephric mesenchyme. If the “normally positioned” ureteric bud made contact and the anterior bud regressed, there may be no apparent phenotype at later stages of renal development. Conversely, if the anterior ureteric bud alone made contact with the metanephric mesenchyme, chances are high that the resulting ureter would insert blindly into the ductus deferens or paramesonephros, causing obstruction and hydroureter. Such phenotypes have been described in other mutants with ureteric buds that originate too anterior off the nephric duct (23). The renal agenesis seen in Fgfr2^Mes-/- mice likely resulted from neither ureteric bud contacting the mesenchyme. When both ureteric buds do make contact, only one appears to induce a normal kidney, while the other is attached to a smaller misshaped kidney. Thus, it appears that only one of the two ureteric buds makes good or any contact with the metanephric mesenchyme in affected Fgfr2^Mes-/- mice, which could explain the incidence and variety of later stage renal defects.

The reason for multiple ureteric bud induction sites is not clear. As shown, Gdnf and Robo2, two key molecules expressed in metanephric mesenchyme that control ureteric bud induction sites, were not altered in Fgfr2^Mes-/- mutants. Thus, another potential explanation for the ureteric bud induction abnormalities would be perturbations in stromal mesenchymal cells surrounding the nephric duct and the base of the ureteric bud. Moreover, molecule(s) expressed in these cells appear to inhibit the interaction between Gdnf with its receptor Ret, thus restricting the nephric duct to one site for ureteric bud induction (22). We speculated that if Fgfr2 were expressed in this cell population, that the Pax3cre transgenic line (used to make the Fgfr2^Mes-/- mice) should delete the gene, based on previous cre recombinase expression data (16). By comparing whole mount in situ hybridization in control, Pax3cre-driven, and Hoxb7cre-driven Fgfr2 conditional knockout embryonic kidneys, we showed that Fgfr2 is expressed in stromal cells around the nephric duct and ureteric bud near its induction site and that it is deleted from these cells in Fgfr2^Mes-/- mice. We confirmed these findings by examining Fgfr2 expression in Fgfr2^Mes-/- and control embryonic tissue sections by radioactive in situ hybridization. Thus, the role of Fgfr2 in preventing ectopic ureteric bud formation is more likely secondary to its presence in the stromal cells adjacent to the nephric duct and ureteric bud base.

Despite the potential role of Fgfr2 in stromal cells, we saw no changes in Bmp4 or Sprouty1 (which is expressed in ureteric epithelium, metanephric mesenchyme and stromal mesenchyme). Furthermore, we saw no differences in the sensitivity of cultured Fgfr2^Mes-/- nephric ducts versus controls to exogenous GDNF (although the in vitro culture system may not be sensitive enough to detect subtle differences). Novel downstream targets of Fgfr2 activity in the metanephric and/or stromal mesenchyme responsible for control of ureteric bud induction sites will be explored in the future.

Finally, the urinary tract anomalies in the Fgfr2^Mes-/- mice may have clinical relevance. Activating mutations of FGFR2 can cause urogenital abnormalities in humans including hydroureter and solitary kidney (24, 25). Human studies have also shown that specific structural renal anomalies seen in Fgfr2^Mes-/- mice, such as renal agenesis, often occur in familial clusters (26, 27). Other studies have shown that patients with renal agenesis/dysgenesis have relatives with a much higher rate of urinary tract anomalies than the general population, with phenotypes including duplicated ureters and unilateral renal agenesis, much as the Fgfr2^Mes-/- mice (28, 29). Finally, there have been a few reports showing how a known single genetic defect in humans may result in many renal malformations, often within the same family (30, 31). Thus, FGFR2 or its ligands may be good candidate genes to screen for mutations in families that have a wide spectrum of renal anomalies.

Abbreviations

Bmp4

Bone morphogenetic protein 4

E

Embryonic

Fgfr

Fibroblast growth factor receptor

Fgfr^Mes-/-

conditional deletion of Fgfr from the metanephric mesenchyme

Fgfr^UB-/-

conditional deletion of Fgfr from the ureteric bud

Gdnf

Glial cell-line derived neurotrophic factor