Development 128, 2153-2161 (2001) 2153 Printed in Great Britain © The Company of Biologists Limited 2001 DEV4528

Hoxa11 and Hoxd11 regulate branching of the ureteric bud in the developing kidney

Larry T. Patterson, Martina Pembaur and S. Steven Potter* Division of Nephrology and Hypertension and Division of Developmental Biology, The Children’s Hospital Research Foundation, Cincinnati, OH 45229, USA *Author for correspondence (e-mail: [email protected])

Accepted 23 March 2001

SUMMARY

Hoxa11 and Hoxd11 are functionally redundant during marker of the metanephric kidney, show that the kidney development. Mice with homozygous null mutation branching defect was not simply the result of homeotic of either have normal kidneys, but double mutants transformation of metanephros to mesonephros. Absent have rudimentary, or in extreme cases, absent kidneys. We Bf2 and Gdnf expression in the midventral mesenchyme, have examined the mechanism for renal growth failure in findings that could by themselves account for branching this mouse model and find defects in ureteric bud defects, shows that Hoxa11 and Hoxd11 are necessary for branching morphogenesis. The ureteric buds are either normal in the ventral mesenchyme. unbranched or have an atypical pattern characterized by Attenuation of normal gene expression along with the lack of terminal branches in the midventral renal cortex. absence of a detectable proliferative or apoptotic change in The mutant embryos show that Hoxa11 and Hoxd11 the mutants show that one function of Hoxa11 and Hoxd11 control development of a dorsoventral renal axis. By in the developing renal mesenchyme is to regulate immunohistochemical analysis, Hoxa11 expression is differentiation necessary for mesenchymal-epithelial restricted to the early metanephric mesenchyme, which reciprocal inductive interactions. induces ureteric bud formation and branching. It is not found in the ureteric bud. This suggests that the branching defect had been caused by failure of mesenchyme to Key words: Organogenesis, Kidney, Renal, Branching, Development, epithelium signaling. In situ hybridizations with Wnt7b, a Hox, Metanephros, Mouse

INTRODUCTION absence of mesenchyme (Qiao et al., 1999). The mechanism does require the presence of more than one mesenchymal cell Renal organogenesis has served as a model system for the type. With loss of the winged helix Bf2 study of multiple developmental mechanisms such as (Foxd1 – Mouse Genome Informatics) from the stromal branching morphogenesis and mesenchyme-to-epithelial mesenchyme progenitors, normal ureteric bud growth and conversion. These processes depend upon the successful branching is impaired (Hatini et al., 1996). The loss of α8 integration of inductive interactions between tissues. Inductive integrin from the mesenchyme immediately adjacent to the bud events in the kidney are functionally well described (Saxen, also impairs ureteric bud branching morphogenesis (Muller et 1987) and remain of great interest at the molecular level. al., 1997). One distinguishing feature in renal development is Ureteric bud induction from the Wolffian duct and its the complex process of mesenchymal morphogenesis. After subsequent growth and branching to form the collecting system induction of the bud by the mesenchyme, the bud provides occurs in response to signaling from nephrogenic and stromal signals that promote mesenchyme survival and induce progenitor mesenchyme. The mesenchyme secreted factors, nephronogenesis. The induced mesenchyme condenses, GDNF (Pichel et al., 1996) and amphiregulin (Lee et al., 1999) becomes polarized epithelium and forms vesicles. The vesicles are two of probably several soluble morphogens that regulate then elongate and differentiate to form the mature glomeruli ureteric bud development. In addition, mutation of an enzyme and tubules of the metanephric kidney. In the absence of involved in proteoglycan synthesis (Bullock et al., 1998) and ureteric bud derived signals, the mesenchyme becomes inhibition of extracellular matrix sulfation (Davies et al., 1995) apoptotic. The processes of branching morphogenesis and within the nephrogenic mesenchyme demonstrate that normal nephronogenesis must conform to a set pattern of development. ureteric bud branching morphogenesis requires extracellular Hox are a well described family of clustered genes matrix as well. The signaling mechanism for ureteric bud that encode homeodomain-containing transcription factors. growth and branching does not require mesenchymal-epithelial Relatively little is known about the role of Hox genes during cell contact, as demonstrated by ureteric bud growth in the organogenesis. We have described morphological changes 2154 L. T. Patterson, M. Pembaur and S. S. Potter within the reproductive tract that suggest the Hox genes Douglass, 1992 (Dressler and Douglass, 1992). The antibody to regulate positional information during organogenesis (Hsieh- Hoxa11 has previously been described (Gendron et al., 1997). Li et al., 1995; Gendron et al., 1997). During neurogenesis, Polyclonal rabbit antibody from immune and preimmune serum was Hox genes have been shown to control both dorsoventral and purified from a A column. Sections (6 µm) were fixed for 5 anteroposterior patterning of the hindbrain (Davenne et al., minutes with 3% paraformaldehyde in PBS. The sections were then 1999). In the developing gut, it has been demonstrated that permeabilized with 0.1% Triton X-100 in phosphate-buffered saline (PBS) for 10 minutes and washed in PBS 0.05% Tween (PBST). The Hoxd13 regulates regionally restricted inductive signaling primary antibodies were diluted 1:100 to 1:200 in 2% goat serum (Roberts et al., 1998). Over 15 Hox genes are expressed in the PBST and incubated with the sections for 1 hour. The slides were developing kidney (Davies and Brandli, 1997 – The Kidney washed in PBST and incubated with rhodamine- or fluorescein- Development Database http//mbisg2.sbc.man.ac.uk/kidbase/ conjugated secondary antibody diluted in goat serum PBST. kidhome.html and http://golgi.ana.ed.ac.uk/kidhome.html.). Dissected embryonic urogenital blocks from double heterozygous These genes may control cell proliferation, inductive signaling Hoxa11/Hoxd11 matings were processed for whole-mount antibody pathways and/or anteroposterior-dorsoventral patterning of the detection of cytokeratin. The tissue was fixed in methanol and incubated kidney or its multiple components. We have shown that two 4-5 hours in 2% goat serum PBST with 1:100 dilution of anti-pan members of the family, Hoxa11 and Hoxd11, exhibit functional cytokeratin (Sigma C2562). After washing, the tissue was incubated redundancy in formation of the kidney, as well as the forelimb with FITC-conjugated anti-mouse IgG secondary antibody (Sigma F2012) in PBST (2% goat serum). Photographic slides were taken with and vertebra (Davis et al., 1995). Homozygous mutant mice for an Olympus BHS model microscope with a reflected light fluorescence either Hoxa11 or Hoxd11 have normal kidneys; however, the attachment. The slides were scanned into Adobe Photoshop. kidneys of the double homozygous mutants are absent or rudimentary. Lectin staining To better understand the function of Hoxa11 and Hoxd11 in Mutant and normal urogenital blocks from E13.5 embryos were kidney development, we defined their expression domains stained with fluorescein-conjugated Dolichos biflorus agglutinin and examined early renal developmental anomalies of (DBA, Vector). The tissue was dissected and fixed for 10 minutes in Hoxa11/Hoxd11 double mutant mice. Although Hoxa11 and 2% paraformaldehyde in PBS, treated with 3% bovine serum albumin o Hoxd11 expression was limited to the mesenchyme in the early (BSA) in PBS 0.05% Triton X-100 for 1 hour at 37 C, and incubated developing kidney, the most striking mutant defect was in with DBA 1:40 in PBS Triton for 1 hour. After washing, the tissue was transferred to a slide for photography. So as not to distort the the pattern of ureteric bud branching morphogenesis. This structures, care was taken not to compress the tissue with a coverslip. establishes a role for Hoxa11/Hoxd11 in mesenchyme for patterning ureteric bud branching morphogenesis. Expression Whole-mount in situ hybridization analysis of the mutant renal mesenchyme demonstrated an Mutant and normal urogenital blocks were dissected and hybridized altered mesenchymal character. There was loss of expression with riboprobes as described previously (Wilkinson, 1992; Hogan, of genes that are crucial for ureteric bud morphogenesis, as 1994). Wnt11 probe was obtained from a 782 bp PvuII fragment of well as diminished expression of genes that are normally EST 349486 (Research Genetics). The pCMV Pax2 construct, kindly induced in the mesenchyme by the ureteric bud. These findings provided by Gregory Dressler, was used to obtained a 575 bp support the hypothesis that Hox genes control pattern XbaI/BamHI Pax2 riboprobe (Dressler et al., 1993). A 399 bp cDNA formation in kidney development by promoting mesenchyme- fragment was amplified using tgtccagtgtggagaactttactg and ctctacacctcaaaaagggcttag primers, cloned and used for a Bf2 epithelial reciprocal inductive signaling. riboprobe. The Wnt7b EST clone 334147 contained a 530 bp fragment that was used as a riboprobe. The Wt1 and Gdnf riboprobes were kind gifts from Andreas Schedl (from Buckler et. al., 1991) and Frank MATERIALS AND METHODS Costantini (Srinivas et. al., 1999), respectively. A 504 bp EcoRI/SmaI fragment was subcloned from pSHlox Hoxd11 and used for Hoxd11 Breeding and genotyping Hoxa11/Hoxd11 mutant mice riboprobe (a gift from M. Todd Valerius). Clones for Hoxd10 and Hoxa11 and Hoxd11 mutant mice have been previously described Hoxd12 riboprobes were generously provided by Denis Duboule (Small and Potter, 1993; Davis et al., 1995). The homozygous mice (Dolle et al., 1989), and those for Hoxc10 and Hoxc11 by Alexander are sterile and heterozygotes have significantly reduced fertility. The Awgulewitsch (Peterson, 1994). Whole-mount tissues were post-fixed colony has been maintained on a mixed genetic background of three in paraformaldehyde, embedded in paraffin, sectioned and strains of mice (C57, C3H and CF1). For this study, 66 mutants were counterstained with nuclear Fast Red (Vector Laboratories). generated by matings between compound heterozygous parental mice. The day when the vaginal plug was observed was considered to be Embryonic proliferation and apoptosis embryonic day (E) 0.5. Embryos with at least one wild-type allele for Day 13.5 urogenital blocks were snap frozen and parasagital sections each gene were used as control normal littermates. The genotypes of stained. At least four sections were stained for proliferation or for all embryos were determined by PCR as before for Hoxa11 (Small apoptosis. Antibody to phosphorylated histone H3 was used to detect and Potter, 1993). PCR genotyping for wild type and mutant Hoxd11 mitotic cells (Correia and Conlon, 2000; Wei et al., 1999). Tissue was alleles was performed using the oligonucleotide primers: also stained with antibody to cytokeratin and with DAPI to identify cgctgtccctacaccaagtaccagatccgc, tccagtgaaatattgcagacggtccctgtt and the epithelium of the branching ureteric bud and all nucleated cells, gtttcagcagtgttggctgtattttcccac. respectively. Digital images were obtained on an Olympus BX60 microscope for each marker and were merged using Photoshop before Immunostaining counting nuclei. Between 1800 and 3000 nucleated mesenchymal Embryonic day 10.5, 11.5 and 13.5 embryos from CD-1 and mutant cells per section were counted and the number of mitotic mice were snap frozen in embedding media and sectioned on a mesenchymal cells expressed as a fraction of the total number of cells. cryostat. Serial sections were stained with antibodies to Hoxa11, The proportion of proliferating cells on the dorsal and ventral halves Pax2, cytokeratin (Sigma) and phosphorylated histone H3 (Upstate of the kidney was also determined. Sections for apoptosis were first Biotechnology 06-570), following the procedure of Dressler and stained with antibody to cytokeratin as above, then post-fixed in 4% Hoxa11/d11 function in renal development 2155

PFA for 15 minutes. Following the in situ cell death detection kit with expression restricted to the caudal segment of the protocol (Roche), the tissue was permeabilized for 2 minutes at 4°C intermediate mesoderm. with 0.1% Triton and 0.1% sodium citrate. After washing, apoptotic Hoxd11 exhibited a pattern of expression similar to Hoxa11. cells were fluorescein labeled using terminal transferase for one hour At E10.5 Hoxd11 was expressed in the intermediate mesoderm o at 37 C. The slides were then stained with DAPI. Pyknotic cells, adjacent to the hindlimb buds (Fig. 3D). At E11.5 it was ureteric bud epithelium and all nucleated cells could then be detected expressed in the mesenchyme surrounding the ureteric bud and counted by fluorescence microscopy. Similar to mitotic cells, the proportion of apoptotic cells was determined. (Fig. 3E) and at E13.5, Hoxd11 expression was maintained in the mesenchyme surrounding the termini of the branching ureteric bud (Fig. 3F,G). We observed no dorsoventral RESULTS asymmetry in expression of either Hoxa11 or Hoxd11 in the developing kidneys on days E11.5 and E13.5. Early defects of the Hoxa11/Hoxd11 double mutant kidney Wnt7b expression in Hoxa11/Hoxd11 mutant kidneys We previously reported the newborn kidney phenotype of mice Hoxa11 and Hoxd11 were expressed in the metanephric homozygous mutant for Hoxa11, Hoxd11, or both. The single mesenchyme of the developing kidney, while the primary mutant kidneys appeared morphologically normal, whereas observed morphological defect in Hoxa11/Hoxd11 double double mutant kidneys were severely reduced in size. These mutants was in the branching of the ureteric bud. This mutant kidneys contained both glomeruli and tubules (Davis et suggested that the mutant mesenchyme did not properly signal al., 1995). the ureteric bud. Three different models could explain this To better understand the origin of the double mutant altered metanephric mesenchyme-ureteric bud interaction. phenotype, we examined ureteric bud formation and branching First, as observed in the adjacent paraxial segmented morphogenesis in mutant kidneys during early organogenesis. mesoderm, these Hox genes may pattern the anterior-posterior In total, 66 double homozygous mutants were examined. At axis of the intermediate mesoderm. In this case, loss of Hoxa11 E11.5, mutants, like control littermates, showed a single and Hoxd11 would cause a homeotic transformation of outgrowth from the caudal segment of the Wolffian duct (Fig. segment identity resulting in anteriorization or conversion of 1A,B). At E13.5, however, all double mutants exhibited defects the metanephros into mesonephros. Second, Hoxa11 and in branching morphogenesis. The severity was variable, with Hoxd11 may simply regulate proliferation of the mesodermal less affected mutants (Fig. 2B,C) showing long intervals segment that gives rise to the primordium of the metanephric between the initial and subsequent branches. In the severely mesenchyme. Reduced mass of mesenchymal tissue would affected mutants, the outgrowths remained unbranched though then perhaps result in reduced signaling. Finally, these Hox they were more elongated than the initial E11.5 bud and they genes may simply be viewed as upstream regulators of still maintained a rounded tip. The mesenchyme surrounding mesenchyme differentiation or of specific morphogens that are these outgrowths of the nephric duct was not condensed around required for normal bud growth and branching. the tip and was easily fragmented (Fig. 2D). Thus, the earliest The metanephros and mesonephros are morphologically identified renal phenotype was impaired branching distinct. Features that distinguish the normal anterior primitive morphogenesis of the ureteric bud while ureteric bud growth mesonephros and the posterior mature metanephros are the or elongation remained intact. number of outgrowths from the Wolffian duct, their length and the subsequent branching of those outgrowths. The Hoxa11 and Hoxd11 expression in the developing mesonephros contains multiple short outgrowths from the duct kidney that do not branch, while the metanephros has a single bud that To better understand the apparent function of the Hoxa11 and arborizes. As shown earlier, there was a single outgrowth from Hoxd11 genes in promoting ureteric bud branching we defined their expression patterns in the early developing kidney. We found that expression of these genes was restricted early in wild type development to the caudal end of the intermediate mesoderm. At E10.5, Hoxa11 was present in the primordial metanephric blastema, adjacent and dorsomedial to the Wolffian duct expansion and at the level of the hindlimb buds (Fig. 3A). On the adjacent section, Pax2 expression was strong in the Wolffian duct and was initiating in the metanephric mesenchyme undergoing induction (Fig. 3B). Also, whereas Pax2 was expressed in the mesonephros, Hoxa11 was absent (data not shown). Later, at E11.5, Hoxa11 was expressed within the metanephric mesenchyme surrounding the branching ureteric bud (Fig. 3C) and at E13.5 it persisted in the nephrogenic mesenchyme around the tips Fig. 1. Normal initiation of ureteric bud in Hoxa11/Hoxd11 mutant of the bud and to a lesser extent in the stromal mesenchyme. embryos. Dissected E11.5 urogenital blocks were stained with Consistent with transcriptional function, Hoxa11 antibody to cytokeratin to identify the Wolffian duct and its immunostaining was predominately nuclear. Thus, Hoxa11 outgrowths. Single outgrowths (arrows) from the Wolffian duct (wd) was a specific early marker of the progenitors of the were identified in the normal littermate (A) and in the double mutant metanephric mesenchyme prior to ureteric bud outgrowth, embryo (B). Scale bar, 0.2 mm. 2156 L. T. Patterson, M. Pembaur and S. S. Potter

Fig. 2. Disrupted ureteric bud branching and growth in the Hoxa11/Hoxd11 mutant embryos. Dissected E 13.5 urogenital blocks from mutant embryos (B- D) and normal littermates (A) were stained with cytokeratin (A-C) or DBA lectin (D) to identify the ureteric bud (u) and its branches. An abnormal branching pattern with elongated primary branches is seen in mutants (B,C) compared with the normal pattern (A). A severely affected mutant kidney (D) showed absence of branching in a bud (arrows) dissected free of the nephric duct. Scale bar, 0.2 mm. the Wolffian duct in the mutant mice that sometimes branched. Proliferation and apoptosis The absence of branching or rudimentary branching in the Altered mesenchymal growth could explain the mutant mutant was consistent with a mesonephric-type outgrowth phenotype. Immunostaining for phosphorylated histone H3, from the caudal end of the Wolffian duct. Therefore, we a marker of mitotic cells, was examined in mutant and characterized this outgrowth using a molecular marker specific normal E13.5 kidneys. We found normal numbers of mitotic for the metanephric bud. cells in mutant kidneys (Fig. 5, Table 1). Additionally, Wnt7b is expressed in the caudal segment of the Wolffian we found the mitotic cells were evenly distributed between duct and the ureteric bud of the metanephros, but is not the dorsal and ventral halves of the kidneys. Likewise, detectable in the mesonephros. In normal E13.5 mice, Wnt7b expression was detected in the ureter leading to the metanephric kidney and in the branches within the developing kidney (Fig. 4A). In situ hybridizations of double homozygous mutant littermates showed Wnt7b expression in the caudal outgrowth from the Wolffian duct, suggesting that the duct maintained metanephric identity (Fig. 4B). The ureteric bud of the mutant again elongated without branching and appeared more tortuous than normal. This molecular characterization argues that mutation of the Hoxa11 and Hoxd11 genes did not result in homeotic transformation of the metanephros into mesonephros.

Fig. 3. Early renal expression patterns of Hoxa11 and Hoxd11. E10.5 (A,B) and E11.5 (C) embryos were sectioned transversely and stained with polyclonal antibody to Hoxa11 (A,C) or Pax2 (B). Hoxa11 is expressed in the limb bud (lb) and in the posterior intermediate mesoderm (arrowhead), but not in the Wolffian duct (arrow), which stains for Pax2. On E11.5 (C), Hoxa11 can only be detected in the mesenchyme surrounding the ureteric bud (ub) and not in the bud itself. E10.5 (D), E11.5 (E) and E13.5 (F) embryos were stained by whole-mount in situ hybridization with a riboprobe to Hoxd11. At E10.5, staining was found in the posterior hindlimb bud (lb) field and in the intermediate mesoderm (im). The dorsal view of the tissue was taken after removal of the overlying neural tube. An oblique view of an E11.5 (E) whole-mount urogenital block showed expression posteriorly in the metanephros (arrow) and the very posterior tip of the developing gonad (go). At E13.5 (F), expression in the metanephros (m) was consistent with mesenchymal expression that overlaps with Hoxa11 expression. Sections of the Hoxd11 stained kidneys show expression in the mesenchyme (G) around a branch of the ureteric bud (arrow). ad, adrenal; nt, neural tube. Scale bar, 0.1 mm Hoxa11/d11 function in renal development 2157

normal numbers of apoptotic cells were identified by TUNEL assay in mutant kidneys (Fig. 5, Table 1). The apoptotic cells were also evenly distributed between the dorsal and ventral halves. It remains possible that very subtle changes in proliferation or apoptosis, not detected in these studies, contributed to reduction in renal size. The diminished branching of the ureteric bud, however, was probably the main cause of attenuated renal growth during this early stage before obvious nephron development. Further Fig. 4. The outgrowth from the Wolffian duct of mutant embryos reduced renal size at subsequent stages of development would maintained Wnt7b expression that is characteristic of the be expected because of the reduced number of ureteric bud tips metanephros. By in situ hybridization, Wnt7b was not detectable in needed for nephron induction and later growth of the kidney. the mesonephros (data not shown). (A) Wnt7b was expressed in the Because we did not detect a change in proliferation or posterior Wolffian duct, in the ureter (arrow), the branches and not apoptosis, we argue that a change in early mesenchymal the tips of the bud in the normal E13.5 metanephros (m). (B) In the growth alone cannot account for the defect in branching E13.5 mutant kidney, Wnt7b was expressed in the unbranched morphogenesis. outgrowth (arrows) of the Wolffian duct. Scale bars, 0.2 mm.

Fig. 5. TUNEL assay (B,D) and immunostaining for phosphorylated histone H3 (A,C) showed similar rates of apoptosis and mitosis, respectively, in Hoxa11/Hoxd11 mutant (A,B) and normal (C,D) kidneys. Digital photographs for phosphorylated histone H3 (red), cytokeratin (a marker of ureteric bud epithelium, green), and DAPI (a marker for all nuclei, blue) were merged (A,C). Likewise, photographs of the TUNEL assay (green), cytokeratin (red) and DAPI (blue) were merged (B,D). Scale bar, 0.1 mm.

Table 1. Proliferation and apoptosis PHH3* TUNEL-positive cells Total % Ventral % Dorsal % Total % Ventral % Dorsal % ‡ Mutant1 2.2 2.2 (166/7457) 2.2 (173/7734) 1.2 1.5 (69/4751) 0.9 (59/6394) ‡ Mutant2 1.6 1.6 (103/6405) 1.5 (93/6046) 0.9 0.9 (39/4336) 0.9 (39/4255) Wild type1 2.4 2.4 (122/5050) 2.3 (105/4492) 1.3 1.4 (52/3824) 1.2 (48/3970) Wild type2 1.6 1.6 (80/4992) 1.6 (84/5281) 0.6 0.6 (39/6373) 0.5 (33/6392)

*α-phosphorylated histone H3-positive cells. ‡Hoxa11/Hoxd11 double mutant mouse. The numbers in parentheses indicate the number of stained cells/total mesenchymal cells. 2158 L. T. Patterson, M. Pembaur and S. S. Potter Abnormal pattern of ureteric bud branching and metanephric growth in Hoxa11/Hoxd11 double mutants We further examined the pattern of ureteric bud branching and the distribution of branch termini in Hoxa11/Hoxd11 double mutants by in situ hybridizations with riboprobes for Wnt11, Ret and Emx2. In the normal kidney, Wnt11 and Ret expression is restricted to branch tips, and Emx2 is expressed in a slightly broader domain that extends from the most distal bifurcation to the branch tips. In the most severely affected Hoxa11/Hoxd11 double mutant kidneys, with no branching of the ureteric bud, riboprobes for Wnt11, Emx2 and Ret did not hybridize with the unbranched tip (data not shown). In the less severely affected Hoxa11/Hoxd11 mutant kidneys, however, the branch tips were reduced in number, but still hybridized to Wnt11, Emx2 and Ret. Surprisingly, the mid-ventral E13.5 mutant kidney was devoid of branch termini (Fig. 6C), while the poles and dorsum (Fig. 6B) appeared more normal. Sections confirmed the absence of organized epithelial structures of ureteric bud branches within this field of the kidney (Fig. 7A). This was a reproducible pattern in the less severely affected kidneys, with the ventral surface showing the most dramatic branching defect. We also observed an apparent defect in position of the mutant kidneys. The kidneys of the double mutants were located more caudal and medial than normal, and had close approximation of the inferior poles. The less severely affected E13.5 kidneys often showed a somewhat dumb-bell shape, consistent with the reduced bud growth and branching, and nephron induction in the mid-ventral region (Fig. 6B,C). These results showed that the Hoxa11/Hoxd11 mutations did influence overall renal growth that again can partly be explained by the diminished number of ureteric bud branches. Mesenchyme analysis We further studied the nature of the mutant mesenchyme Fig. 6. Altered dorsoventral pattern and gene expression in the kidneys using molecular markers for stromal precursor of Hoxa11/Hoxd11 mutant mice. E13.5 kidneys from Hoxa11/Hoxd11 mesenchyme (Bf2; Hatini et al., 1996), induced double mutants (B,C,E,F,H,I,K,L,N,O) and normal littermates mesenchyme (Pax2; Dressler and Douglass, 1992), early (A,D,G,J,M) were hybridized with riboprobes for Wnt11 (A-C), Bf2 undifferentiated mesenchyme (Wt1; Armstrong et al., (D-F), Pax2 (G-I), Wt1 (J-L) and Gdnf (M-O). The dorsal (not shown) 1993) and a mesenchyme signaling factor (Gdnf; Pichel, and ventral regions of the normal kidneys showed an even distribution 1996). The early metanephric mesenchyme gives rise to of ureteric bud termini (Wnt11), stromal mesenchyme (Bf2), induced mesenchyme (Pax2), uninduced mesenchyme (Wt1) and Gdnf two distinct cell lineages, the nephrogenic lineage, which expression. The dorsal region of the mutant kidneys showed near forms the nephrons, and the stromal lineage, which gives normal distribution of bud termini (B) and stromal mesenchyme (E). In rise to the renal interstitial mesenchyme. In situ contrast, the dorsum showed patchy Pax2 (H) and reduced Wt1 (K) hybridization with Bf2 probe was used to determine if the expression. The mid-ventral region was more severely affected in the ventral cells differentiated into stromal precursor cells. mutants with loss of ureteric bud branch termini (C, arrow) and stromal Surprisingly, there was no Bf2 expression on the mid- mesenchyme (F, arrow). Areas of induced mesenchyme (I) and ventral surface of the mutant kidney (Figs 6F, 7B). expression of Wt1 (L) were diminished in the mutant ventral Therefore, not only is the nephrogenic mesenchyme not metanephric mesenchyme. A patch of induced mesenchyme (white replaced by stromal precursor mesenchyme, but arrow) was found on the surface of the left kidney (I). Also of note Hoxa11/Hoxd11 function is required for normal stromal were the lateral entry and elongated primary branches (black arrow) of the ureteric bud in the mutant. The expression of Pax2 in the cell differentiation on the ventral surface of the developing branches of the bud in the interior of the kidney stands out because of kidney. Whether this is due to absence of bud tips or the lack of Pax2 expression in the overlying mesenchyme (I). Gdnf directly to loss of Hoxa11 and Hoxd11 is not known. expression was significantly decreased on the ventral surface of the Pax2 is normally expressed in the ureteric bud, its mutant kidney (O). ad, adrenal; go, gonad; gu, gut; m, metanephros. branches and within the induced mesenchyme. Normally, Scale bars, 0.2 mm. cortical mesenchyme Pax2 expression obscures underlying Hoxa11/d11 function in renal development 2159

Fig. 7. Sections of Hoxa11/Hoxd11 mutant kidneys showed the mid-ventral region consists of mesenchyme and contains no organized epithelium of ureteric buds. Longitudinal sections through E13.5 kidney hybridized with the Wnt11 probe (A) showed bud tips (arrow) and central mesenchyme (arrowhead). Transverse section through E13.5 kidney hybridized with Bf2 (B) showed Bf2 expression on the dorsal Fig. 8. Anteroposterior or dorsoventral gradients of expression were (d) but not on the ventral (v) surface. not found in the kidney for other Hox genes. Wild-type E13.5 kidneys were hybridized with riboprobes to Hoxc10 (A), Hoxc11 (B), Hoxd10 (C) and Hoxd12 (D). The intensity of staining for these Pax2-positive ureteric bud branches after whole-mount in situ genes on the ventral surface (shown) was equivalent to the dorsal hybridization. Hybridizations with Pax2-specific probe showed surface and varies with distance from the termini of the branches. limited induction in mutant metanephric mesenchyme. The Note also that the expression of Hoxc11 is globally very weak. Scale E13.5 mutant kidney ventral surface shown in Fig. 6I, for bar, 0.1 mm. example, contained only a single area of induced mesenchyme. This in situ further demonstrated the abnormal branching mutant phenotype in Hoxa11/Hoxd11 double mutants. The pattern of the ureteric bud in mutants. By E13.5 the ureter major conclusions are that Hoxa11/Hoxd11 regulate should insert into the kidney medially; however, a lateral metanephric mesenchyme-ureteric bud inductive interactions insertion was found in the mutant kidneys. In addition, the during patterning of the kidney. The mechanism for control of primary branches failed to form the normal initial series of these interactions appears to be independent of metanephric dichotomous branches. It should also be noted that areas of mesenchyme growth. induced mesenchyme were greatly reduced even on the dorsal surface of the mutant kidney, despite evidence of ureteric bud Defective branching morphogenesis of the ureteric branching (Fig. 6H). Thus, in the absence of Hoxa11/Hoxd11, bud in mutant kidneys reciprocal inductive interactions are reduced. The mesenchyme Hoxa11 and Hoxd11 showed similar patterns of expression, to bud signaling was deficient, resulting in defective bud present in the early metanephric mesenchyme, later branching and the mesenchyme induction in response to bud surrounding the invading and branching bud, and down- signaling was poor, as measured by Pax2 expression. These regulating during mesenchyme-to-epithelia conversion. The results provide evidence that the Hoxa11/Hoxd11 genes control most striking developmental alteration in the double mutant more than just proliferation of the mesenchyme during kidney kidney, however, was defective branching morphogenesis of development. The expression of Wt1 was globally reduced in the ureteric bud, which did not express Hoxa11/Hoxd11. This the mutant kidney, on both dorsal and ventral surfaces (Fig. suggested defective signaling from the mesenchyme to the 6K,L), further illustrating the altered nature of the mutant ureteric bud. metanephric mesenchyme. Not unexpectedly, decreased Gdnf The function of Hox genes in vertebrate development has expression was seen on the ventral surface of the mutant kidney been a controversial subject. In Drosophila, however, it is and could explain the branching defect (Fig. 6J). generally agreed that Hom-C (Drosophila Hox) genes have an important patterning function. Null mutations in Drosophila Dorsoventral renal patterning Hox genes often result in anteriorizations of segment identity, In an attempt to explain the dorsoventral pattern, we examined while misexpression mutations often convert structures to a the normal mesenchymal expression patterns of several other more posterior identity (for a review see Manak and Scott, Hox genes in the kidney at E13.5. We found approximately 1994). The roles of Hom-C genes during Drosophila equal ventral and dorsal expression of Hoxc10, Hoxc11, organogenesis have also been examined. For example, they Hoxd10 and Hoxd12 (Fig. 8). Thus, we were unable to explain pattern the gut along the anterior-posterior axis. Spatially the apparent dorsal compensation for loss of both Hoxa11 and restricted expression of Hom-C genes within the gut mesoderm Hoxd11. is required for mesoderm-endoderm interactions and the formation of constrictions between the four chambers of the midgut (Reuter et al., 1990; Immergluck et al., 1990; Capovilla DISCUSSION et al., 1994). The Hom-C genes of the mesoderm appear to control the expression of the transforming growth factor β In this study, we examined the expression patterns of Hoxa11 family member dpp, which then serves to induce labial and Hoxd11 in the developing kidney, and defined the kidney expression in the endoderm. 2160 L. T. Patterson, M. Pembaur and S. S. Potter

In vertebrates, however, it has been suggested that the Hox ventral domain of the mutant kidney argues against a genes function in a manner that is surprisingly different from proliferation restricted function for Hoxa11/Hoxd11. The that observed in flies. Duboule has proposed that all developing mutant mesenchyme was not simply reduced in mammalian Hox genes regulate cell proliferation and not cell size. Some regions of the mutant mesenchyme remained able identity (Douboule, 1995). In its extreme form this model to induce budding, while others did not. These results contrast states that all mammalian Hox genes control identical or with other reports such as the description of the mutant functionally equivalent downstream targets that are involved in embryos kidneys, where the kidneys were hypoplastic, had the regulation of cell proliferation. This model is consistent reduced proliferation, but maintained normal structure (Moens with the observations that Hox have similar in vitro et al., 1993; Bates et al., 2000). In this case, Myc was thought DNA target binding specificities, single Hox gene knockout to play a role in proliferation without disturbing pattern. mice have relatively mild phenotypes, and double knockout mice sometimes have reduced or absent structures (Davis et al., Additional defects in mutant metanephric 1995). It is also consistent with the results of a number of mesenchyme studies that suggest at least some role for Hox genes in Perturbation of mutant mesenchyme differentiation was controlling cell proliferation rates (Goff and Tabin, 1997). detected by in situ hybridizations. First, the expression level To better understand the nature of the morphogenic defects of Wt1 was severely reduced in all regions of mutant in the Hoxa11/Hoxd11 mutant kidneys additional gene mesenchyme, further indicating altered character and not just expression studies were performed. Even when unbranched, reduced mass. Along with the reduced Wt1 expression, we also the ureteric buds of mutant embryos did not lose metanephric found decreased Gdnf expression in the ventral mesenchyme. identity and continued to express Wnt7b, a marker of the Hoxa11/d11 appear to be upstream of very early events in metanephros. We were thus unable to detect evidence of a intermediate mesoderm differentiation. Pax2 expression in homeotic transformation of metanephros to mesonephros, one mesenchyme was also reduced, even in regions where potential result of loss of Hox gene function. relatively normal bud branching was observed. In the early Not only was there a defect in branching, but in situ developing wild-type kidney, Pax2 expression is found initially hybridization with ureteric bud tip specific riboprobes in the ureteric bud, which then induces Pax-2 expression in demonstrated that there was a distinct patterning defect. In the the metanephric mesenchyme. This reduced mesenchymal most severely affected kidneys the bud completely failed to expression of Pax-2 in mutants therefore suggests a defective branch. In the less severely affected kidneys, however, there response to ureteric bud induction. was a relatively normal distribution of bud tips on the dorsal We also examined the expression of the winged-helix surface and at the anterior and posterior poles, but an absence transcription factor Bf2, a stromal lineage marker, in the mutant of bud tips in the mid-ventral region of the kidney. This metanephric mesenchyme. There was no Bf2 expression in the demonstrates a biologically important difference between cells ventral domain of the mutant metanephric mesenchyme, along the dorsoventral axis of the kidney. In the ventral domain, indicating that the failure of this mesenchyme to induce which because of rotation was formerly the lateral domain, branching morphogenesis of the bud was not simply the result Hoxa11/Hoxd11 expression is rigidly required for branching to of a decision en masse to form stroma instead of nephrogenic occur. In contrast, in the dorsal domain, formerly the medial mesenchyme. Similarly, the nephrogenic mesenchyme did not domain, near normal branching can often take place, even in simply fail to proliferate, resulting in its replacement by the absence of Hoxa11/Hoxd11 function. To our knowledge, stromal mesenchyme. The reduced Bf2 expression in the this is the first reported observation of a dorsoventral axis in mutant mesenchyme is also of interest because of the the kidney. Review of the literature indicates that branching similarities between the Bf2 and Hoxa11/Hoxd11 mutant morphogenesis and tubulogenesis are normally synchronous in phenotypes. In each case there is a striking defect in branching all quadrants of the kidney. It is therefore unlikely that our morphogenesis (Hatini et al., 1996). These observations observations represent an accentuation of a lag in development suggest that Bf2 lies downstream of Hoxa11/Hoxd11 in kidney of the ventral aspect of the kidney. A functional difference also development, and that, in addition to altered Gdnf expression, has been reported in the cells along the anteroposterior axis of the altered Bf2 expression in Hoxa11/Hoxd11 mutants at least the kidney, as defined by differing responses to bone partly accounts for the observed kidney phenotype. morphogenetic protein 4 in organ culture (Raatikainen-Ahokas Finally, based on expression analysis, we found no candidate et al., 2000). With over 15 Hox genes expressed in the Hox gene that could account for compensation on the dorsal developing kidney, it is possible that dorsoventral or surface of the mutant kidney. In situ hybridization showed anteroposterior Hox gene gradients could control development roughly equivalent dorsoventral levels of expression for each of the renal axes and compensate for the absence of of the four Hox genes tested. Despite ventral expression of Hoxa11/Hoxd11 in the dorsal domain. several Hox genes, total compensation for loss of Hoxa11 and A second potential effect of Hoxa11/Hoxd11 loss of function Hoxd11 in the mutant did not occur. The overlapping is a diminished proliferative capacity of the mesenchyme. expression of these and perhaps other Hox genes raises a Because of the dorsoventral defect, we were not only able to provocative question: what is the function of each individual compare mutant with wild-type kidneys, but also severely Hox gene in renal development? As in C. elegans, the affected ventral mutant mesenchyme to the more normal dorsal individual Hox genes may serve multiple functions, including mesenchyme. Even with this advantage, we were unable to proliferation and differentiation (Salser and Kenyon, 1996). In detect a change in proliferation. Similarly, we were unable to aggregate, the Hox genes may control proliferation within the find a change in apoptosis. Absence of change in proliferation kidney thus accounting for the apparent preservation of or apoptosis and the observed qualitative difference in the mesenchymal proliferation in the mutant ventral mesenchyme. Hoxa11/d11 function in renal development 2161

Hoxa11 and Hoxd11 appear to have attained a separate and associated with maternal reproductive defects in Hoxa-11 null mice. Biol. distinct function during renal mesenchyme differentiation. As Reprod. 56, 1097-1105. a result, the reciprocal inductive interactions between the Goff, D. J. and Tabin, C. J. (1997). Analysis of Hoxd-13 and Hoxd-11 misexpression in chick limb buds reveals that Hox genes affect both bone metanephric mesenchyme and ureteric bud are greatly condensation and growth. Development 124, 627-636. attenuated in the kidneys of mutant embryos. Elucidation of Hatini, V., Huh, S. O., Herzlinger, D., Soares, V. C. and Lai, E. (1996). downstream effectors of these and other Hox genes will greatly Essential role of stromal mesenchyme in kidney morphogenesis revealed by aid in our understanding of their function. Additionally, targeted disruption of Winged Helix transcription factor BF-2. Genes Dev. 10, 1467-1478. detailed examination of kidneys in double knockouts of other Hogan, B. (1994) Manipulating the Mouse Embryo: A Laboratory Manual. paralogous Hox genes will enhance our understanding. 2nd edn. New York: Cold Spring Harbor Laboratory Press. Hsieh-Li, H. M., Witte, D. P., Weinstein, M., Branford, W., Li, H., Small, Many thanks to M. Todd Valerius for many insightful discussions K. and Potter, S. S. (1995). Hoxa 11 structure, extensive antisense and to D. Witte for assistance with the TUNEL assay. We gratefully transcription, and function in male and female fertility. Development 121, acknowledge G. Dressler for the Pax2 riboprobe, A. Schedl for the 1373-1385. Wt1 riboprobe, F. Costantini for the Gdnf riboprobe, D. Duboule for Immergluck, K., Lawrence, P. A. and Bienz, M. (1990). Induction across the Hoxd10 and Hoxd12 riboprobes, and A. Awgulewitsch for the germ layers in Drosophila mediated by a genetic cascade. Cell 62, 261- 268. Hoxc10 and Hoxc11 riboprobes. This work was supported by NIH Lee, S. B., Huang, K., Palmer, R., Truong, V. B., Herzlinger, D., Kolquist, grant HD32061 to S. S. P. K. A., Wong, J., Paulding, C., Yoon, S. K., Gerald, W., Oliner, J. D. and Haber, D. A. (1999). The Wilms tumor suppressor WT1 encodes a transcriptional activator of amphiregulin. Cell 98, 663-673. Manak, J. R. and Scott, M. P. (1994). A class act: conservation of REFERENCES homeodomain protein functions. Dev. 120, Suppl. 61-77. Moens, C. B., Stanton, B. R., Parada, L. F. and Rossant, J. (1993). Defects Armstrong, J. F., Pritchard-Jones, K., Bickmore, W. A., Hastie, N. D. and in heart and lung development in compound heterozygotes for two different Bard, J. B. (1993). The expression of the Wilms’ tumour gene, WT1, in the targeted mutations at the N-myc locus. Development 119, 485-499. developing mammalian embryo. Mech. Dev. 40, 85-97. Muller, U., Wang, D., Denda, S., Meneses, J. J., Pedersen, R. A. and Bates, C. M., Kharzai, S., Erwin, T., Rossant, J. and Parada, L. F. Reichardt, L. F. (1997). Integrin alpha8beta1 is critically important for (2000). Role of N-myc in the developing mouse kidney. Dev Biol. 222, epithelial-mesenchymal interactions during kidney morphogenesis. Cell 88, 317-325. 603-613. Buckler, A.J., Pelletier, J., Haber, D. A., Glaser, T. and Housman, D. E. Peterson, R. L., Papenbrock, T., Davda, M. M. and Awgulewitsch, A. (1991). Isolation, characterization, and expression of the murine Wilms’ (1994). The murine Hoxc cluster contains five neighboring AbdB-related tumor gene (WT1) during kidney development. Mol. Cell. Biol. 11, 1707- Hox genes that show unique spatially coordinated expression in posterior 1712. embryonic subregions. Mech. Dev. 47, 253-260. Bullock, S. L., Fletcher, J. M., Beddington, R. S. and Wilson, V. A. Pichel, J. G., Shen, L., Sheng, H. Z., Granholm, A. C., Drago, J., Grinberg, (1998). Renal agenesis in mice homozygous for a gene trap mutation in A., Lee, E. J., Huang, S. P., Saarma, M., Hoffer, B. J., Sariola, H. and the gene encoding heparan sulfate 2-sulfotransferase. Genes Dev. 12, Westphal, H. (1996). Defects in enteric innervation and kidney 1894-1906. development in mice lacking GDNF. Nature 382, 73-76. Capovilla, M., Brandt, M. and Botas, J. (1994). Direct regulation of Qiao, J., Sakurai, H. and Nigam, S. K. (1999). Branching morphogenesis decapentaplegic by Ultrabithorax and its role in Drosophila midgut independent of mesenchymal-epithelial contact in the developing kidney. morphogenesis. Cell 76, 461-475. Proc. Natl. Acad. Sci. USA 96, 7330-7335. Correia, K. M. and Conlon, R. A. (2000). Surface ectoderm is necessary for Raatikainen-Ahokas, A., Hytonen, M., Tenhunen, A., Sainio, K. and the morphogenesis of somites. Mech. Dev. 91, 19-30. Sariola, H. (2000). BMP-4 affects the differentiation of metanephric Davenne, M., Maconochie, M. K., Neun, R., Pattyn, A., Chambon, P., mesenchyme and reveals an early anterior-posterior axis of the embryonic Krumlauf, R. and Rijli, F. M. (1999). Hoxa2 and Hoxb2 control kidney. Dev. Dyn. 217, 146-158. dorsoventral patterns of neuronal development in the rostral hindbrain. Reuter, R., Panganiban, G. E., Hoffmann, F. M. and Scott, M. P. (1990). Neuron 22, 677-691. Homeotic genes regulate the spatial expression of putative growth factors in Davies, J., Lyon, M., Gallagher, J. and Garrod, D. (1995). Sulphated the visceral mesoderm of Drosophila embryos. Development 110, 1031- proteoglycan is required for collecting duct growth and branching but not 1040. nephron formation during kidney development. Development 121, 1507- Roberts, D. J., Smith, D. M., Goff, D. J. and Tabin, C. J. (1998). Epithelial- 1517. mesenchymal signaling during the regionalization of the chick gut. Davis, A. P., Witte, D. P., Hsieh-Li, H. M., Potter, S. S. and Capecchi, M. Development 125, 2791-2801. R. (1995). Absence of radius and ulna in mice lacking hoxa-11 and hoxd- Salser, S. J. and Kenyon, C. (1996). A C. elegans Hox gene switches on, off, 11. Nature 375, 791-795. on and off again to regulate proliferation, differentiation and morphogenesis. Dolle, P., Izpisua-Belmonte, J. C., Falkenstein, H., Renucci, A. and Development 122, 1651-1661. Duboule, D. (1989). Coordinate expression of the murine Hox-5 complex Saxen, L. (1987). Organogenesis of the Kidney. Cambridge: Cambridge -containing genes during limb pattern formation. Nature 342, University Press. 767-772. Small, K. M. and Potter, S. S. (1993). Homeotic transformations and limb Dressler, G. R. and Douglass, E. C. (1992). Pax-2 is a DNA-binding protein defects in Hox A11 mutant mice. Genes Dev. 7, 2318-2328. expressed in embryonic kidney and Wilms tumor. Proc. Natl. Acad. Sci. USA Srinivas, S., Wu, Z., Chen, C. M., D’Agati, V. and Costantini, F. (1999). 89, 1179-1183. Dominant effects of RET misexpression and ligand-independent Dressler, G. R., Wilkinson, J. E., Rothenpieler, U. W., Patterson, L. T., RET signaling on ureteric bud development. Development 126, 1375- Williams-Simons, L. and Westphal, H. (1993). Derregulation of Pax-2 1386. expression in transgenic mice generates severe kidney abnormalities. Nature Wei, Y., Yu, L., Bowen, J., Gorovsky, M. A. and Allis, C. D. (1999). 362, 65-67. Phosphorylation of histone H3 is required for proper Duboule, D. (1995). Vertebrate Hox genes and proliferation: an alternative condensation and segregation. Cell 97, 99-109. pathway to homeosis? Curr. Opin. Genet. Dev. 5, 525-528. Wilkinson, D. G. (1992). Whole mount in situ hybridization of vertebrate Gendron, R. L., Paradis, H., Hsieh-Li, H. M., Lee, D. W., Potter, S. S. and embryos. In In Situ Hybridization (ed. D. Wilkinson), pp. 75-83. Oxford: Markoff, E. (1997). Abnormal uterine stromal and glandular function IRL Press.