Sustained Bmp signaling is essential for cloaca development in zebrafish

Bone morphogenetic protein (Bmp) signaling plays crucial roles in forming and patterning ventral and posterior tissues during vertebrate embryogenesis(reviewed by Dale and Jones,1999; Hammerschmidt and Mullins, 2002; Kishigami and Mishina, 2005). Zebrafish mutants in the Bmp pathway, such as swirl (bmp2b) and snailhouse (bmp7), have severe deficits in ventral and posterior mesoderm, along with expansion of dorsal and anterior tissues, such as the brain and notochord(Mullins et al., 1996). Conversely, analysis of mutants with increased Bmp signaling, combined with overexpression experiments in Xenopus, have shown that Bmp signaling is also sufficient for the formation of blood, vascular and kidney tissues,which are derived from the ventral mesoderm in zebrafish(Kimmel et al., 1990). For example, overexpression of Bmp RNAs in Xenopus animal caps is sufficient to induce the expression of red blood cell markers(Zhang and Evans, 1996), and chordino, a zebrafish mutant with excess Bmp signaling, forms extra blood, kidney and vascular cells(Hammerschmidt et al., 1996; Leung et al., 2005). Thus, Bmp signaling is important for the induction of ventral mesodermal derivatives during development.

Despite the crucial requirement of Bmp signaling for ventral mesoderm formation and differentiation, the temporal basis of this role is not well understood. Studies to date have examined the effects of manipulating Bmp signaling only at the earliest stages of development, and have not examined later effects on the differentiation of ventral mesodermal derivatives. Furthermore, early loss of Bmp signaling results in aberrant formation of the ventral mesoderm, prohibiting the subsequent analysis of patterning within this domain. Because Bmps are expressed at high levels in the ventral region of the embryo until the end of somitogenesis(Martinez-Barbera et al.,1997), it is important to determine whether Bmp signaling plays any role in the ventral mesodermal derivatives beyond their initial formation.

To analyze the temporal roles of Bmp signaling in the patterning of ventral mesodermal tissues, we used an inducible transgenic zebrafish line that we recently generated (Pyati et al.,2005). This line, containing the Hsp70 promoter driving a dominant-negative Bmp receptor fused to GFP, allows stage-specific Bmp inhibition upon heat shock of transgenic embryos. Initially, we used these transgenic fish to describe the roles of Bmp in tail development. We showed that during the early gastrula stages, Bmp signaling is crucial for the development of the primary tail, but, by the mid-gastrula stage, Bmp signaling is important only for ventral tail fin formation and for preventing the development of secondary tail structures(Pyati et al., 2005). Thus,the roles of Bmp in the posterior mesoderm change over time.

In this study, we describe the patterning roles of Bmp in ventral mesodermal tissues, including blood, vascular and kidney precursors, after the earliest phase of gastrulation. Although Bmp signaling is needed at the early gastrula stage for these ventral derivatives to form, we find, surprisingly,that inhibition of Bmp signaling just two hours later at the mid-gastrula stage causes an expansion of the blood and vascular precursors into the extreme ventral embryonic domain. Lineage labeling of these cells shows that they normally contribute to ventral tail tissues, including the developing cloaca (the common opening of the gut and kidneys), and that the development of these structures is deficient in transgenic embryos. We find that patterning and function of the excretory system is dependent on early and sustained Bmp signaling, and we examine, for the first time, the morphogenesis of the presumptive cloaca. Finally, we show that the T-box transcription factor HrT is an important downstream mediator of Bmp signaling in excretory system development. These results reveal a novel role for Bmp signaling in posterior organogenesis beyond early ventral mesoderm specification, and they implicate misregulation of this major developmental signaling pathway as a possible cause of anorectal malformations during human embryogenesis.

Heat shocking and sorting of embryos were performed essentially as described (Pyati et al.,2005), except that pre-warmed 40°C media was used for 30-minute heat shocks in a 40°C air incubator. This produced robust transgene expression (marked by GFP) within 30 minutes.

In situ hybridization was performed as previously described(Griffin et al., 1998). Coloration reactions were stopped simultaneously for transgenic embryos and wild-type siblings, and, for these experiments, all images shown are representative of 100% of embryos examined with each probe(n≥30).

kaede mRNA (100 pg) (Ando et al., 2002) was injected into one-cell-stage embryos from a transgenic outcross and heat shocked at the mid-gastrula stage (as described above). Embryos were then immobilized in 1% low melting point agarose on slides at the 10-somite stage and UV light at 405 nm was focused on the ventral tail bud and underlying cells for 5 minutes for photoconversion. At 24 hpf, wild-type and transgenic embryos were sorted based on loss of the ventral tail fin (100% penetrant in transgenic embryos), and visualization of red labeling was performed using a compound fluorescent microscope. Labeling was performed in a total of 25 embryos.

Fluorescent rhodamine dextran (Molecular Probes) was injected into the anterior gut of wild-type and transgenic embryos at 4 dpf, when the gut was fully formed in wild-type embryos. Dye excretion was visualized and photographed using a Zeiss Axioplan 2 compound fluorescent microscope. Injections were performed in at least five larvae per genotype, in two separate experiments.

Embryos were incubated for 1 hour in BODIPY TR methyl ester dye(Cooper et al., 2005), then filmed using a Bio-Rad MRC-600 confocal microscope for 4 dpf experiments. For time-lapse microscopy of cloacal opening, a Zeiss LSM Pascal confocal microscope was used. The time-lapse movie was made by taking a z-series through an msxb-gfp transgenic embryo once every 10 minutes, then using one optical section for each timepoint.

Exon 1, intron 1 and 5.5 kb upstream of the 5′UTR (∼7.5 kb total)of msxb were PCR-amplified and cloned upstream of EGFP-SV40polyA. Embryos were injected with 80 ng/μl of linearized plasmid, grown to adulthood, and the progeny were screened for EGFP fluorescence. Embryonic EGFP expression faithfully recapitulates endogenous msxb expression, with the exception of the early neural crest domain(Akimenko et al., 1995).

Examination of cell death using acridine orange was performed as previously described (Clements and Kimelman,2005). Forty embryos per genotype were stained at 24-somites to image dying cells in the developing cloaca.

hrT translation blocking or mismatch morpholino (2 ng)(Szeto et al., 2002) was injected into one-cell-stage embryos, and cloaca formation was observed at 24-48 hpf.

Bmp signaling has widely been described as being crucial for the formation of ventral mesoderm, including blood, vasculature and kidney cells, during development. However, the roles of Bmp signaling in regulating the relative proportions of each ventral cell type within the mesodermal domain have not been addressed. Previously, we had observed that a reduction of Bmp signaling at the early gastrula stages (approximately 50-70% epiboly) caused severe deficits in ventral mesoderm specification, phenocopying mutants in the Bmp pathway (Pyati et al., 2005). However, in this earlier study, we did not address the requirement for Bmp signaling to regulate the differentiation of ventral mesoderm after its initial formation. To determine whether Bmp plays any role in regulating the development of ventral mesodermal tissues beyond the early phase of gastrulation, we heat shocked hs::dnBmpr-GFP embryos and wild-type siblings at the mid-gastrula stage (80% epiboly, when gastrulation was just over halfway completed). In the same experiment, we heat shocked embryos at dome stage (about 2 hours before gastrulation) to compare the effects of inhibiting early ventral mesoderm development. We next performed whole-mount in situ hybridization to examine the expression of gata1 [red blood cell precursors (Detrich et al.,1995)], flk1 [vascular precursors(Sumoy et al., 1997); kdr–Zebrafish Information Network], and pax2.1 [kidney precursors (Pfeffer et al.,1998)] at the 10-somite stage. These genes were expressed in adjacent stripes of cells within the lateral posterior mesoderm of wild-type embryos heat shocked at either stage (Fig. 1A,D,G). By contrast, heat shock of transgenic embryos before the start of gastrulation (dome stage) resulted in a loss of all three ventral mesodermal markers (Fig. 1B,E,H), consistent with an early gastrula role for Bmp signaling in specifying the ventral mesoderm (Pyati et al., 2005). Surprisingly, reduction of Bmp signaling at the mid-gastrula stage led to an expansion of gata1- and flk1-positive cells into the mesoderm ventral to the tail bud(compare regions marked with arrows in Fig. 1C,F with regions in Fig. 1A,D). This expansion of blood and vascular progenitors was also confirmed by the expanded expression of the early blood and vascular marker fli1 (Brown et al.,2000; Thompson et al.,1998) into the same domain ventral to the tail bud (data not shown). pax2.1, which was normally expressed at low levels around the tail bud (Fig. 1G), was also upregulated in this domain in transgenic embryos(Fig. 1I), but to a much lesser extent than the blood and vascular markers. Thus, kidney cell fates are not dramatically altered at the expense of blood and vascular gene expression. Because transgenic embryos heat shocked at the mid-gastrula stage normally displayed C1 dorsalization (loss of the ventral tail fin) without obvious cell movement defects during epiboly, we conclude that reduction of Bmp signaling during mid-gastrulation caused the extreme ventral mesoderm to undergo a fate change to blood and vascular progenitors.

The expansion of blood and vascular cells into the region below the tail bud when Bmp signaling was inhibited suggested that zebrafish have a domain that is more ventral than blood and vascular cells, and that the fate of these cells is dependent on Bmp signaling. As the fate of these ventral cells was unknown, we performed lineage labeling in wild-type and transgenic siblings heat shocked at mid-gastrula using Kaede, a photoconvertible fluorescent protein (Fig. 2A)(Ando et al., 2002). Kaede can be photoconverted from green to red fluorescence with UV light (∼400 nm),allowing the selection of well-injected embryos for lineage labeling. We injected one-cell-stage embryos from a clutch of outcrossed transgenic fish with kaede RNA, and then heat shocked the embryos at 80% epiboly. At 10-somites, which is the stage when we observed ectopic gata1 and flk1 expression in these cells(Fig. 1), we photoconverted the Kaede with a pinhole focusing UV light on the tissue below the tail bud. As a consequence of the size of the pinhole used, we also labeled the ventral portion of the tail bud. As expected, we observed red fluorescence in somitic tissue, which is derived from ventral tail bud cells in both transgenic and wild-type siblings (Fig. 2C,D). Strikingly, cells in the ventral tail (not shown) and those lining the ventral yolk extension were also labeled red because of the labeling of cells below the tail bud. Transgenic embryos had a severe reduction in these tissues compared with wild-type siblings (compare bracketed regions in Fig. 2C,D). We also observed labeled cells in the presumptive cloaca at the posterior end of the yolk extension, and transgenic embryos consistently had cyst-like swellings in this tissue (boxed region in Fig. 2D). This swelling, as well as the reduction in ventral tail tissues, could be clearly visualized using Nomarski brightfield optics(compare boxed regions in Fig. 2E,F). Thus, we conclude that the extreme ventral cells at the 10-somite stage contribute to ventral tail tissue, presumptive cloaca, and a population of cells lining the ventral part of the yolk extension. Additionally, we find that mid-gastrula Bmp signaling is crucial for the formation of these tissues.

The reduction of ventral tissues in transgenic embryos prompted us to determine which cell fates were lost as a result of reducing Bmp signaling at the mid-gastrula stage using markers of ventral tissues. The T-box transcription factor gene hrT (tbx20–Zebrafish Information Network) was previously shown to be expressed in cells ventral to the tail bud at the 10-somite stage, and later in cells ventral to the yolk extension (as well as in the forming heart field)(Griffin et al., 2000). Intriguingly, morpholino knockdown of hrT causes expansion of gata1-expressing cells into the ventralmost mesoderm, similar to the effect of reducing Bmp signaling during the mid-gastrula stages(Szeto et al., 2002). Because this result suggested a possible link between the loss of Bmp signaling and the expansion of gata1 expression ventrally, we examined hrTexpression in transgenic embryos heat shocked at mid-gastrula. We observed a complete loss of hrT transcripts in transgenic embryos compared with wild-type siblings at both the 10-somite stage in the ventral mesoderm below the tail bud, (compare Fig. 3A-C with E-G) and at the 18-somite stage in ventral mesodermal cells lining the yolk extension (compare Fig. 3D with 3H). The expression of hrT in the heart field, however,was not affected (Fig. 3A,E). These results show that mid-gastrula Bmp signaling is necessary for maintaining the fate of the extreme ventral mesoderm as marked by hrTexpression.

The expression of hrT in ventral mesodermal cells near the presumptive excretory system, coupled with defects in these tissues in transgenic embryos, led us to hypothesize that there were defects in the development of the urogenital and/or anorectal systems when Bmp signaling was reduced at the mid-gastrula stage. Indeed, ventral/caudal mesoderm is thought to be a signaling center for excretory system development, because expanded ventral mesoderm can increase the size of the proctodeum, an ectodermal group of cells that contributes to the anal tissue(Isaacs et al., 1994). RNA encoding the T-box transcription factor Tbx2b marks the excretory system and ventral border of the yolk extension in zebrafish(Dheen et al., 1999). After a mid-gastrula stage heat shock, we observed a clear reduction of tbx2bexpression in cells ventral to the yolk extension and at the extreme end of the developing excretory system (the proctodeum) in transgenic embryos(compare Fig. 3I with 3M). This result shows that Bmp signaling is important for normal patterning of excretory system tissues and reveals that the regulation of tbx2bexpression in the excretory system is similar to its Bmp-dependent regulation in the developing heart (Yamada et al.,2000). We next examined the expression of hoxd13, which is expressed in the developing proctodeum(van der Hoeven et al., 1996). hoxd13 expression was greatly reduced in the proctodeal region of transgenic embryos when compared with wild-type siblings, even though the posterior expression of hoxd13 expanded in the tail bud (compare Fig. 3J with 3N). These results show that a reduction of Bmp signaling at the mid-gastrula stage perturbs the development of proctodeal and ventral yolk extension domains.

Because the terminus of the pronephric ducts ultimately connects to the urogenital system and opens to the outside of the embryo(Drummond et al., 1998), we examined whether there were defects in the formation or positioning of the kidney terminus in transgenic embryos compared with wild type siblings. As shown by the expression of gata3(Neave et al., 1995), the pronephric terminus was specified normally, but it was not properly positioned at the periphery of the embryo in transgenics compared with controls at the 20-somite stage (compare Fig. 3K with 3O).

Next, we examined the expression of vox, a ventral gene that is known to be a direct target of Bmp signaling(Imai et al., 2001; Kawahara et al., 2000; Melby et al., 2000). vox is expressed in ventral tail tissues, including the ventral tail fin and developing proctodeal region, in wild-type embryos. In accordance with reports showing that vox becomes dependent on Bmp signaling by mid to late gastrulation (Melby et al.,2000; Ramel and Lekven,2004), we observed a total absence of vox expression in the ventral domain of transgenic embryos when compared with wild-type siblings at the 20-somite stage (Fig. 3L,P).

Finally, we assayed for loss of tissue specification in the cloacal region by analyzing the expression of two cloaca markers at 24 hpf, prdm1(Wilm and Solnica-Krezel,2005) and evx1(Thaeron et al., 2000). The cloaca expression of both markers is absent in transgenic embryos when compared with wild-type siblings (Fig. 3Q-T), supporting the hypothesis that loss of Bmp signaling beyond the early gastrula stages causes a loss of cloacal cell fate. Overall, these results clearly show that mid-gastrula Bmp signaling is necessary for the expression of genes in the distal excretory system and the proper positioning of the kidney terminus.

We wished to analyze how long during embryogenesis Bmp signaling was required for cloaca specification. After a 40°C heat shock, the dominant-negative Bmp receptor is present for 10-15 hours (as indicated by GFP fluorescence; data not shown), meaning that transgenic embryos heat shocked at mid-gastrula would still have the dominant-negative receptor present throughout much of somitogenesis. Thus, it was important to determine a window of time that Bmp signaling was required for cloaca specification. To address this question, we heat shocked transgenic and wild-type sibling embryos at 80%epiboly, bud, 3-, 5-, 8-, 10-, 12- and 14-somites, then fixed 20 embryos of each genotype per stage of heat shock for analysis of hoxd13expression and left the remainder to score cloaca phenotypes at 24 hpf. As shown in Table 1, cloacal defects can be induced in transgenic embryos strongly between 80% epiboly and 8-somites, then to a lesser extent at 10-somites. However, heat-shocks after this stage produced no effects on cloaca development. Absence of hoxd13 expression closely parallels the morphological scoring,although, by 10-somites, there is no obvious reduction in the expression of this gene. This is likely to reflect a more subtle loss of proctodeal cells in transgenic embryos heat shocked at 10-somites. Taken together, these results show that sustained Bmp signaling is essential from 80% epiboly through the early somite stages for proctodeal specification and cloaca development.

Because the alterations described above suggested that excretory system function may be defective when Bmp signaling is inhibited beyond early gastrulation, we developed an assay to analyze whether transgenic embryos could properly excrete waste through their cloaca at 4 days post-fertilization(dpf), when the gut begins to function[Fig. 4A; based on a similar strategy to examine digestive physiology(Farber et al., 2001)]. We injected fluorescent rhodamine dextran into the anterior gut of wild-type and transgenic sibling embryos that had been heat shocked at the mid-gastrula stage. Upon injection, we could visualize peristalsis of the gut as dye was forced posteriorly in both wild type and transgenic embryos (data not shown). However, upon peristalsis of the dye into the hindgut, we observed severe defects in transgenic larvae. Contrary to wild-type siblings, in which dye was excreted through the cloaca (Fig. 4B,C), transgenic animals could not eliminate dextran from their gut cavity, and they developed an obvious swelling of their hindgut as a consequence (Fig. 4D,E). These results are consistent with our molecular analysis and show that cloacal opening is dependent on proper Bmp signaling during development.

Having established that there were defects in excretory system patterning and function upon Bmp inhibition, we next wanted to determine how the defects in the excretory system developed. These processes have not been described thoroughly in zebrafish, affording us an opportunity to document the morphogenesis of the excretory tissues for the first time. Initially, we used the kidney terminus marker gata3 to monitor the aberrant modeling of the urogenital system between the 22-somite and 28-somite stages. In wild-type embryos, the kidney terminus had reached the ventral limit of the tail by the 22-somite stage (Fig. 5A). Subsequently, the kidney terminus in wild-type embryos was remodeled to form an opening (which we call the presumptive cloaca, as the embryos do not yet have an anus) by the 28-somite stage. Interestingly, gata3 expression was progressively reduced in the distal tip of the presumptive cloaca(Fig. 5A-D). This reduction could be due either to downregulation of gata3 in the mesoderm of the kidney terminus or to the inclusion of epidermal cells, which do not normally express gata3. Examination of p63 protein expression (data not shown)coupled with time-lapse microscopy using an msxb-gfp transgenic line(see below) support the latter possibility. In transgenic embryos, the level of gata3 expression was normal throughout the time course, but the terminus of the kidney did not properly extend to the ventral limit of the tail (Fig. 5E-H). Instead,these embryos displayed an obvious swelling of the kidney terminus, no remodeling at the ventral tail limit and no reduction of gata3expression in the distal kidney tip, over the course of the experiment. The defects in kidney extension and remodeling were also observed in living embryos (Fig. 5I-L). There, we could observe acellular gaps between the kidney terminus and the ventral epidermis of transgenic embryos, probably as a result of the early loss of proctodeal cells.

Because defective cloacal morphogenesis could be due in part to changes in cell death, we used Acridine Orange staining to assay for dying cells in both wild-type embryos and transgenic siblings during the period when the presumptive cloaca was forming. We discovered that a low level of cell death occurs in the presumptive cloacal region of wild-type embryos, and that this cell death is absent in transgenic siblings(Fig. 6A,B). These data suggest that developmentally regulated cell death may play an important role in cloacal morphogenesis.

Next, we took advantage of the optical transparency of the zebrafish embryo to further dissect the mechanism of cloacal opening. For these experiments, we used a transgenic zebrafish line that labels cells of the ventral epidermis with GFP under the control of the msxb promoter. Endogenous msxb is normally strongly expressed in cells of the ventral epidermis during the stages of cloaca opening(Akimenko et al., 1995), making this transgenic line very useful for our studies. As shown by detailed time-lapse filming of a wild-type msxb-gfp embryo starting at 24 somites (Fig. 6C-F, see also Movie 1 in the supplementary material), cloacal opening involves massive cellular rearrangements within the proctodeum and kidney terminus. A single vacuolated cell within the proctodeum migrates ventrally and forms a pore within the epidermis (labeled in green by msxb-gfp), possibly by undergoing cell death. The kidney terminus connects to this pore and undergoes a morphogenetic transformation to adopt a columnar appearance. The newly formed presumptive cloaca thus consists of at least two cell types: kidney cells in the dorsal aspect and epidermal cells at the ventral terminus.

Finally, we wished to examine the architecture of the mature excretory system in living animals. Although the kidney connects to the presumptive cloaca during late somitogenesis, the gut tube does not connect to the presumptive cloaca until 4 dpf. Using BODIPY TR methyl ester(Cooper et al., 2005), we labeled all the tissues of 4-day old larvae for confocal microscopic analysis. This labeling allowed us to assay the entire kidney and excretory system for defects that occurred in transgenic larvae, and it provided a unique glimpse into the morphology of the excretory system during development. Wild-type larvae showed a clear opening of the gut into the cloaca, next to the kidney opening (Fig. 6G). Consistent with the lack of cloacal function that we observed earlier, transgenic larvae failed to form a cloacal opening, and the kidneys and gut failed to open to the outside of the body, resulting in large swellings in both structures(compare Fig. 6H). This result shows that development of both the urogenital and gut openings are dependent on sustained Bmp signaling, and that the failure of the presumptive cloaca to develop normally leads to major malformations of the excretory tissues.

The T-box transcription factor HrT was an attractive candidate to be a downstream mediator of Bmp signaling in the ventral mesoderm involved in forming the excretory system. As shown above, hrT is normally expressed in the extreme ventral mesoderm at the 10-somite stage, and its expression is lost in transgenic embryos heat shocked at the mid-gastrula stage (Fig. 3A-H). Furthermore,we previously observed that HrT is important for preventing gata1expression in the extreme ventral mesoderm(Szeto et al., 2002),mirroring the results observed here with a loss of Bmp signaling at the mid-gastrula stage. To test whether HrT might be important for development of the presumptive cloaca, we injected either a morpholino oligonucleotide (MO)that blocks hrT translation or a four-base mismatch MO and examined the structure of the presumptive cloaca at 24-48 hpf. As shown in Fig. 7A, hrT morphants had dysmorphic cloaca development (55%, n=128) compared with mismatch control-injected embryos (Fig. 7B; 8%, n=118). However, unlike embryos with mid-gastrula reductions in Bmp signaling, hrT morphants had normal ventral tail fin formation (data not shown). Furthermore, the expression of tbx2b(n=43), hoxd13 (n=41) and gata3(n=69) was normal in morphant embryos(Fig. 7C-H), suggesting that HrT is involved in only a subset of the roles of Bmp within the developing excretory system. We conclude that, in addition to its role in heart development, HrT has an additional important role as a Bmp-regulated gene that is required to allow the excretory system to form normally.

Our studies reveal a novel role for the Bmp pathway in formation of the zebrafish excretory system. Upon inhibition of Bmp signaling at any point between the mid-gastrula and 8-10 somite stages, we observe defects in the presumptive cloaca at 24 hours post-fertilization, demonstrating that the inductive events leading to proper formation of the cloaca take place well before the presumptive cloaca actually forms between the 20- and 28-somite stages. In the absence of sustained Bmp signaling, the kidney tubules form normally, but they fail to make a connection to the outside of the body at approximately one day of development. A similar defect is observed with the gut tube at 4 days post-fertilization, when the gut normally connects to the cloaca. In both cases, we observed large swellings forming at the end of the kidney and gut tubes, which we think is likely to be due to hydrostatic pressure. Indeed, in our excretion assay, we observed peristaltic movements in the guts of heat-shocked transgenic larvae that were the same as in wild-type animals, but the larvae failed to excrete the rhodamine tracer.

Our data suggest that the failure to form a normal cloaca in transgenic animals results from a combination of defects, including altered cell fates within the ventral region of the embryo, revealed by aberrant expression of the marker genes tbx2b, hoxd13, vox, prdm1 and evx1. Additionally, there is a loss of developmentally regulated cell death in the proctodeal region of transgenic embryos. Interestingly, zebrafish mutants with increased Bmp signaling have excessive cell death in the proctodeal region(Hammerschmidt et al., 1996),indicating that precise Bmp signaling levels may be important for regulating cell death in the developing cloaca. Specification of the kidney terminus is normal in transgenic embryos, as revealed by gata3 expression,supporting the hypothesis that defective proctodeal development is the primary cause of excretory system defects when sustained Bmp signaling is compromised. Overall, we suggest that the misspecification of extreme ventral cells results in an inability of the kidney ducts and gut tube to migrate to the ventral surface and interact with the proctodeum and epidermis to form the cloaca.

One of the key mediators of sustained Bmp signaling is the T-box transcription factor HrT. hrT is expressed specifically in the ventral mesodermal cells that normally do not express blood and vascular genes, and reduction of Bmp signaling at the mid-gastrula stage caused a loss of hrT expression in this region. Importantly, we had previously observed that inhibition of HrT function with MOs caused the expression of gata1 in the most ventral mesoderm(Szeto et al., 2002), exactly as is observed when Bmp signaling is inhibited at the mid-gastrula to early somitogenesis stages. We now show that inhibition of HrT function causes a defect in cloacal development, similar to that observed with inhibition of Bmp signaling. However, the defects caused by inhibiting HrT function are much less severe than those observed when Bmp signaling is inhibited, with no loss of ventral tail fin or reduction of cells under the yolk extension, and no loss of ventral gene expression. These results indicate that hrT is one of several Bmp-regulated genes involved in regulating the formation of the cloaca.

Our data suggest a modification to the Bmp gradient model, which posits that the highest levels of Bmp signaling at the ventral limit of the mesoderm are required for formation of the blood and vascular cells(Dosch et al., 1997). Our results reveal an extreme ventral mesodermal domain that is normally devoid of blood and vascular cells, and which requires continuous Bmp signaling at early gastrulation and beyond to maintain its fate. These findings raise two possible models for the specification of the extreme ventral mesoderm, and both involve sustained Bmp signaling. In the first model, the extreme ventral mesoderm requires both the highest and most sustained Bmp signaling to be specified, thus representing the highest point in the Bmp gradient. In the second model, all of the ventral mesodermal derivatives require the same level of Bmp signaling at the early gastrula stage, but only the extreme ventral mesoderm requires sustained Bmp signaling. In this scenario, the extreme ventral mesodermal cells would still need to encounter the highest effective levels of Bmp signaling, as they would integrate moderate levels of signaling over an extended time. In both scenarios, the extreme ventral mesoderm requires the most Bmp signaling of any tissue, but the first model fits more closely with the classical idea of a spatial gradient, as Bmp levels would be highest where the extreme ventral mesoderm would form. Future experiments using low, moderate and high levels of Bmp inhibition at the early gastrula stage should distinguish between these two models. In either scenario, our studies show that only the extreme ventral mesoderm is dependent on Bmp signaling from the mid-gastrula stage. Moreover, we find that the role of Bmp signaling changes dramatically between the early gastrula and mid-gastrula stages; while Bmp signaling is required during the earliest phase of gastrulation for blood, vascular and kidney cell specification, after this stage, Bmps limit the domain of expression of blood and vascular markers.

Based on the results discussed above, we propose the following model for Bmp signaling in development of the excretory system(Fig. 8). Beyond the early gastrula stage, sustained Bmp signaling is required to specify the most ventral region of the embryo that subsequently forms the presumptive cloacal region, ventral tail fin and ventral lining of the yolk extension. One of the crucial targets of the sustained Bmp signaling is hrT, which acts to suppress early blood gene expression in the ventral mesodermal cells, and is necessary for the morphogenesis of the presumptive cloaca. The ventral cells at the end of the yolk extension undergo a complex remodeling that allows the proctodeum to interact with the kidney terminus, possibly by emitting a signal that allows the kidney tubules to migrate to the most extreme ventral region of the embryo. Our data suggest that in response to a loss of Bmp signaling,the misspecified ventral cells fail to spread underneath the tail as the tail bud extends, thus leading to severe deficiencies in ventral tissues, including the proctodeum. As a result, the extension of the kidney ducts is stalled, and they fail to connect to the proctodeum and open to the outside of the embryo. A similar morphogenesis is repeated three days later, when the gut tube needs to connect to the outside of the embryo. As these events can be readily visualized in living embryos by confocal microscopy, the zebrafish provides an excellent system for examining the cellular dynamics that underlie these important biological processes.

The formation of the human cloacal region involves a complex set of morphological transformations that connect the internal organs to the outside of the body. Approximately one in 5,000 live births have anorectal malformations (Smith, 1988),which range from surgically simple defects such as an imperforate anal membrane to far more difficult problems such as failure of the anus and/or rectum to form properly, aberrant connections between the rectum and urinary tracts, and malformations within the cloaca(Durston et al., 1989; Nixon, 1988; Pena et al., 2004). Notably,cloacal malformations, particularly the more severe type, typically involve other malformations within the body, suggesting that these defects are not the failure of a single morphogenetic event but instead involve a more widespread alteration during early development (Smith and Saeki, 1988). The most common malformations associated with cloacal defects involve alterations in the vertebral column, although alterations in the kidneys and urinary tract, genitals and lower limbs are also observed (Duhamel, 1961; Smith and Saeki, 1988). Based on the observation that these defects often involve the more posterior parts of the body, Duhamel (Duhamel,1961) proposed that the cloacal defects were part of a spectrum of caudal defects in which the most extreme form manifests itself as sirenomelia,in which the legs fuse into a single structure.

Intriguingly, a recent study in the mouse has shown that a loss of Bmp7 combined with a half dose or complete loss of twisted gastrulation (Tsg) results in sirenomelia(Zakin et al., 2005). Bmp signaling defects in caudal mesoderm have been implicated by clinical researchers as one possible cause of analogous human malformations(Jain and Weaver, 2004). Zebrafish develop a ventral tail fin instead of a true set of hindlimbs, and loss of the ventral tail fin has been regarded as a readout for defects in ventral/caudal mesoderm (Tucker and Slack,2004). Thus, we hypothesize that a reduction of Bmp signaling after the early gastrula stages results in a global loss of the caudal mesoderm, affecting the ventral tail fin and cloaca in zebrafish, and the lower limbs and cloaca in humans. Based on the studies reported here, we suggest that very subtle alterations in Bmp signaling, due to either genetic or environmental perturbations, may be one cause of human cloacal malformations.

We wish to thank Ashley Webb for critical comments on this manuscript,David Raible and his laboratory for helpful advice and use of their lab space,Dave White and the UW Zebrafish facility for fish care and maintenance, Daniel Szeto, Vladimir Korzh, Carl Neumann, Thomas Wilm and Lilianna Solnica-Krezel for probes, Lalita Ramakrishnan and Muse Davis for help with lineage-labeling experiments, Hillary McGraw for help with time-lapse microscopy, and Lance Buesa for help with in situ hybridizations. This work was supported by grants from the NIH (GM65469) to D.K. and an NSF grant (IBN-0212258) to M.S.C. U.J.P. was supported by NIH training grant GM07270.