|Home | About | Journals | Submit | Contact Us | Français|
Intricate interactions between the Wnt and Bmp signaling pathways pattern the gastrulating vertebrate embryo using a network of secreted protein ligands and inhibitors. While many of these proteins are expressed post-gastrula, their later roles have typically remained unclear, obscured by the effects of early perturbation. We find that Bmp signaling continues during somitogenesis in zebrafish embryos, with high activity in a small region of the mesodermal progenitor zone at the posterior end of the embryo. To test the hypothesis that Bmp inhibitors expressed just anterior to the tailbud are important to restrain Bmp signaling we produced a new zebrafish transgenic line, allowing temporal cell-autonomous activation of Bmp signaling and thereby bypassing the effects of the Bmp inhibitors. Ectopic activation of Bmp signaling during somitogenesis results in severe defects in the tailbud, including altered morphogenesis and gene expression. We show that these defects are due to non-autonomous effects on the tailbud, and present evidence that the tailbud defects are caused by alterations in Wnt signaling. We present a model in which the posteriorly expressed Bmp inhibitors function during somitogenesis to constrain Bmp signaling in the tailbud in order to allow normal expression of Wnt inhibitors in the presomitic mesoderm, which in turn constrain the levels of canonical and non-canonical Wnt signaling in the tailbud.
Proper patterning of a vertebrate embryo depends on the coordinated activity of multiple intercellular signaling factors, including proteins and small molecules. These signals are carefully regulated temporally and spatially with areas of active signaling fine-tuned by inhibitors and activators, as in the case of signaling by Bone morphogenetic proteins (Bmps, reviewed in De Robertis and Kuroda, 2004; Heasman, 2006; Kimelman, 2006; Little and Mullins, 2006; Niehrs, 2004; Schier and Talbot, 2005). As demonstrated in fish and frogs, a gradient of Bmp activity is formed by the onset of gastrulation with the highest activity at the ventral pole (Jones and Smith, 1998; Tucker et al., 2008). This gradient patterns ventral fates and represses dorsal genes, and it directs cell migration towards the dorsal side during gastrulation in zebrafish (von der Hardt et al., 2007). Bmp inhibitors such as Chordin, Noggin and Follistatin establish the gradient and are required to limit its spread as Bmp signaling induces transcription of bmp genes during gastrulation.
Importantly, after the onset of gastrulation the role of Bmp signaling changes dramatically. Transgenic zebrafish have been used to demonstrate that mid-gastrula and later Bmp signaling no longer regulates large scale patterning, and instead it controls formation of the ventral tail fin, the cloaca and the notochord in the posterior of the embryo (Connors et al., 2006; Esterberg et al., 2008; Pyati et al., 2006; Pyati et al., 2005; Stickney et al., 2007), and in the anterior, establishment of the white blood cell population and the liver as well as regulating the asymmetry of the heart (Chocron et al., 2007; Hogan et al., 2006; Shin et al., 2007; Smith et al., 2008).
Wnt signaling also plays a major role during the gastrula stage, and, like Bmp signaling, is also regulated by secreted inhibitors (reviewed in De Robertis and Kuroda, 2004; Kawano and Kypta, 2003). Canonical (β-catenin dependent) signaling is essential for restricting the size of the dorsal organizer in fish and frogs, whereas non-canonical Wnt signaling is required for normal morphogenetic movements during gastrulation (reviewed in Heasman, 2006; Schier and Talbot, 2005; Tada et al., 2002). Intriguingly, the Bmp and canonical Wnt pathways interact (Hoppler and Moon, 1998; Marom et al., 1999; Ramel et al., 2005; Szeto and Kimelman, 2004), with canonical Wnt signaling able to prolong the duration of a Bmp signal (Fuentealba et al., 2007). Recently we demonstrated a post-gastrula role for canonical Wnt signaling in maintaining the mesodermal progenitors that form the posterior end of the zebrafish embryo (Martin and Kimelman, 2008).
Whereas the anterior trunk forms from cells that ingress during gastrulation (Kinder et al., 1999), more posterior somites form from a group of progenitors that reside in the most posterior end of the embryo in a region called the tailbud, which continuously contributes cells to the presomitic mesoderm during somitogenesis (Cambray and Wilson, 2002; Cambray and Wilson, 2007; Davis and Kirschner, 2000; Griffin and Kimelman, 2002; Kanki and Ho, 1997). The maintenance of the mesodermal progenitors is regulated by the T-box transcription factor Brachyury (reviewed in Naiche et al., 2005; Showell et al., 2004). In zebrafish, the Brachyury function is divided between two genes, no tail and brachyury, which act to sustain canonical Wnt signaling within the mesodermal progenitors (Martin and Kimelman, 2008). How the maturation and differentiation of these progenitors is regulated is still not well understood.
Analyzing post-gastrula embryos we found a small region within the tailbud mesodermal progenitor zone of active Bmp signaling, bordered by a region expressing several Bmp inhibitors. To test the post-gastrula role of the Bmp inhibitors, we generated a novel transgenic line to bypass their function by inducing cell-autonomous Bmp signaling. Ectopic activation of Bmp signaling during early somitogenesis caused specific gene expression and morphogenetic defects within the tailbud. Our analysis of the tailbud phenotype demonstrates the major role of the Bmp inhibitors is to sustain the expression of several Wnt inhibitors, which are necessary in turn to regulate gene expression and morphogenesis within the tailbud by limiting the activity of canonical and non-canonical Wnt signaling.
The Tg(HScaBmpR, GFP) line was created by placing a constitutively active mutant Bmp receptor (Macias-Silva et al., 1998) on one side of a multimerized heat shock promoter (Bajoghli et al., 2004) with a Green Fluorescent Protein (GFP) gene on the opposite side. This was flanked by two Tol2 elements (Fig. 2A and used to generate stable transgenics in the WIK/AB background according to Kawakami et al., (1998). Heat shocks were 39.5°C for one hour. Tg(HSwnt5a) fish (Stoick-Cooper et al., 2007), were heat shocked as described.
Single probe whole-mount in situ hybridization was performed as described (Griffin et al., 1995). Fluorescent in situ hybridization and antibody staining were performed according to Jülich et al. (2005). Phosphorylated-Smad 1,5 and 8 antibody (Cell Signaling Technology) was used at a 1:100 dilution, and detected with an Alexa-568-conjugated secondary antibody (Molecular Probes). Some embryos were counterstained with DAPI, which is shown pseudocolored red. Embryos stained with fluorescent markers were imaged on an FV-1000 Confocal Microscope (Olympus). Resulting images were deconvolved using Huygens Essential (Scientific Volume Imaging) and analyzed with ImageJ (NIH).
Donor embryos obtained from an outcross of Tg(HScaBmpR, GFP) hemizygotes to WIK/AB wild-types were injected with 1% rhodamine dextran at the 1-cell stage. Transplants were from donors at sphere stage into hosts at shield stage, targeted to the ventral mesoderm. Hosts were heat shocked at the 3-somite stage, photographed at 36 hpf, then fixed and stained with anti-GFP antibody (Roche) to identify transgenic cells.
Lineage labeling was performed according to Mara et al., (2007). An Axiovert 200M microscope (Zeiss) was used to perform the photoconversion, using a 40× objective with the aperture constricted. The illuminated region went from just posterior to the notochord to the tip of the tailbud.
After the completion of gastrulation, bmp transcripts are expressed in restricted regions of the posterior and anterior of the zebrafish embryo, with bmp4 and bmp2a broadly expressed at the most posterior end of the embryo (Martinez-Barbera et al., 1997). To determine where Bmp signaling is active, we performed whole-mount immunohistochemistry with an antibody specific to phosphorylated (activated) Smads 1 and 5 in post-gastrula embryos. This assay revealed strong Bmp signaling in the tailbud during somitogenesis. Combining this assay with fluorescent in situ hybridization (FISH) using a no tail probe that marks the tailbud mesodermal progenitors, demonstrated that the active Bmp signaling was in the middle of the mesodermal progenitor region (Fig. 1A, B, S1).
To better understand how Bmp signaling is regulated in this tissue, we examined expression of the Bmp inhibitors chordin, noggin1 and noggin2 (Dal-Pra et al., 2006; Furthauer et al., 1999; Miller-Bertoglio et al., 1997; Schulte-Merker et al., 1997). These transcripts were restricted to the notochord and presomitic mesoderm, just anterior to the region containing activated Smads (Fig. 1C, D; see also Fig. S5). Similarly, follistatin-like 1b is restricted to the notochord and presomitic region in a very similar pattern (see fstl1b gene expression at http://zfin.org). These results suggest that the Bmp inhibitors expressed in the presomitic and axial mesoderm might play an essential role in restricting Bmp signaling to the most posterior end of the embryo, within the mesodermal progenitor domain.
The Bmp inhibitors Chordin, and members of the Noggin and Follistatin family, are required to properly establish the dorsal-ventral axis. Reducing the levels of these proteins in early fish or frog embryos produces an extremely severe ventralization, precluding analysis of the role of the Bmp inhibitors during later stages of development (Dal-Pra et al., 2006; Khokha et al., 2005). Repressing a single Bmp inhibitor results in a milder phenotype but retains the problem of affecting the development of cells that will contribute to the tailbud at a later time. To circumvent the problems with early inhibitor activation we created a new transgenic zebrafish line allowing temporal activation of Bmp signaling. Cell autonomous activation was chosen to entirely bypass the effects of extracellular inhibitors. In addition, a cell autonomous activator permits cell transplantation studies since it does not affect Bmp levels in surrounding cells. We tested constitutively active mutants of several Type I Bmp receptors as well as a constitutively active Smad5 by microinjecting mRNA encoding these proteins into zebrafish embryos and scoring for a ventralized phenotype (not shown). A constitutively active mutant of the murine Alk6 (ca-BmpRIb) (Macias-Silva et al., 1998) produced the strongest ventralized phenotype at the lowest dose, and thus this was used for further studies.
The ca-BmpR1b gene was cloned into a vector containing a multimerized minimal heat shock promoter (Fig. 2A), which drives transcription bidirectionally in response to a heat shock such that GFP is expressed simultaneously with the ca-BmpR1b (Bajoghli et al., 2004). This allows embryos expressing the transgene to be sorted from non-transgenic embryos without requiring the GFP to be fused to the receptor, which could potentially change its activity. These elements were flanked by Tol2 sites so that Tol2-mediated transgenesis could be utilized to enhance integration efficiency (Kawakami et al., 1998). 11 transgenic founders were produced using this construct, and screened by heat-shocking their progeny; the line producing the strongest ventralized phenotype after heat shock at the late blastula stage (dome stage) was used for future studies. The embryonic phenotype produced from this line (Fig. 2B) was phenotypically indistinguishable from that produced by injecting bmp RNA (Kishimoto et al., 1997). The ventralization phenotype was further confirmed by examining the expression of eve1 (Fig. 2C,D), which expands when Bmp levels increase (Mullins et al., 1996; Nikaido et al., 1997).
In order to assess the later roles of Bmp activation, which functionally inhibits the activity of the Bmp inhibitors, transgenic embryos were heat shocked at the 3-somite stage and photographed at subsequent time points to determine if morphological changes resulted. In these and all subsequent experiments, embryos were collected from hemizygous transgenic adults outcrossed to wild-type adults. Half of the embryos are transgenic, which is detected by the presence of GFP following heat shock. This experimental paradigm ensures that the transgenic and non-transgenic embryos are treated identically.
The first visible phenotype is disruption of posterior somite morphology and altered shape of the embryo's posterior end. While the defects were only apparent in the posterior of the embryo, examination of heat shocked transgenic embryos one hour after the heat shock demonstrated ubiquitous Smad activation (not shown). Initially the posterior is shorter and wider than in controls, which persists through mid-somitogenesis (Fig. 3A-D). Eventually the posterior narrows and elongates to a normal size, although the resulting embryo is clearly aberrant (Fig. 3E, F). Somite boundaries remained irregular and the somites never attained their normal size (Fig. S2). In addition, the epidermal fin folds, which extend from the dorsal and ventral sides of the embryo, were greatly reduced. The recovery in the tailbud is likely to be due to a lack of perdurance of the activated Bmp signal since we observed that the ectopic Smad expression peaked around 2.5 hours after the end of the heat shock (Fig. S3). In contrast, the epidermal defects did not recover (Fig. 3E, F), indicating that epidermal cells require tight regulation of Bmp signaling during early somitogenesis.
The wide, rounded tailbud observed at the 14-somite stage morphologically resembled the posterior of a spadetail mutant embryo (Kimmel et al., 1989). The spadetail mutant phenotype arises from a genetic block in maturation of mesoderm progenitors in the tailbud, causing them to accumulate at the most posterior end of the embryo where they continue to express the no tail gene (Griffin et al., 1998; Griffin and Kimelman, 2002; Ho and Kane, 1990). We hypothesized that a similar block in maturation might be caused by ectopic Bmp signaling. In situ hybridization using a no tail probe indicated that the mesodermal progenitor zone is enlarged in response to ectopic Bmp signaling (Fig. 4A, B; n=25). This expansion was detectable much earlier than the morphological changes were visible (9-somite stage versus the 14-somite stage). Thus, one reason Bmp signaling is tightly restricted to the tailbud is to prevent the improper expansion of the mesodermal progenitor zone.
To further understand the morphological changes resulting from ectopic Bmp signaling we performed lineage labeling experiments. Embryos were injected with RNA encoding the fluorescent protein Kikume with a nuclear localization signal (NLS-Kikume). Kikume is a green fluorescent protein that can be permanently converted to a red fluorescent state by a brief irradiation with UV light (Ando et al., 2002). Embryos were injected with NLS-kikume RNA and heat shocked at the 3-somite stage, then cells in the tailbud were labeled by photoconverting NLS-Kikume to red fluorescence. After 8-9 hours the embryos were imaged to determine the contribution of labeled cells to the presomitic mesoderm. Labeled tailbud cells in wild-type embryos are observed along almost the entire extent of the presomitic mesoderm, visible over 250 μm from the posterior end of the embryo (Fig. 4C, E). Labeled cells in transgenic embryos contribute to presomitic mesoderm, although they do not extend as far anteriorly (Fig. 4D, E). The aberrant cell movements combined with the broader tailbud observed in transgenic embryos (Fig. 4D) suggests a defect in convergence/extension cell movements that are necessary to extend the body axis during somitogenesis. Thus, ectopic Bmp activation caused changes in both gene expression and morphogenesis within the tailbud. Heat shock activation of the transgene at the 17-somite stage resulted in a similar enlargement of the mesoderm progenitor zone and disruption of somite morphology (data not shown). As expected, the smaller progenitor pool at this stage did not enlarge as dramatically as at earlier stages, and only the most posterior somites were affected.
The enhanced Bmp signaling could either cell autonomously affect the behavior of single cells in the tailbud, or it could change the behavior of the entire tailbud through cell non-autonomous effects. To distinguish between these possibilities we used cell transplantation, taking advantage of the fact that the activated Bmp receptor activates Bmp signaling cell autonomously. One possible caveat to these experiments would be if activation of the Bmp signaling pathway induced Bmp ligand expression as occurs in an autoregulatory loop in the early gastrula embryo (Kondo, 2007). However, activation of the activated Bmp receptor at the 3-somite stage does not enhance expression of bmp2a or bmp4, two bmp genes with strong expression within the tailbud (Fig. S4).
Twenty to 30 donor cells were taken from transgenic embryos or controls injected with rhodamine-dextran as a lineage tracer. Transplants were targeted to the ventral margin to maximize contribution to tailbud mesoderm (Fig. 4F). Hosts were allowed to develop normally until the 3-somite stage, then subjected to a heat shock to activate the transgene. We observed that transplanted cells from transgenic and non-transgenic embryos contributed to the same tissues and to the same anterior extent within the embryo (Fig. 4G, H; n = 18 each). Within the somites, donor cells from both transgenic and non-transgenic embryos adopted the characteristic morphology of differentiated muscle cells and contributed to somites throughout the tail and most posterior trunk. Thus, inappropriate Bmp signaling did not cell autonomously alter morphogenesis of cells in a wild type host, suggesting that Bmp caused non cell-autonomous changes within the tailbud.
Our recent work has demonstrated a critical role for canonical Wnt signaling in maintaining the mesodermal progenitor cells, specifically as part of an autoregulatory loop with no tail and the related gene brachyury (Martin and Kimelman, 2008). Since ectopic Bmp signaling caused non-cell autonomous defects and upregulated no tail expression, we hypothesized that it may be acting via a modulation of Wnt signals. In situ hybridization with probes to the primary Wnt ligands present in the tailbud during somitogenesis was performed. The two canonical wnt genes, wnt8 and wnt3a, retained wild-type expression patterns in response to ectopic Bmp signaling, while expression of the non-canonical wnt gene, wnt5a, expanded slightly (not shown). These results suggested that the major effect of Bmp activation was not on wnt gene expression.
We next considered the alternate possibility that the levels of Wnt ligands were increased when Bmp signals were raised due to changes in the levels of Wnt inhibitors. Intriguingly, Wnt inhibitors are expressed within the presomitic mesoderm abutting the mesodermal progenitor region in a pattern similar to that of the Bmp inhibitors (Hsieh et al., 1999; Pezeron et al., 2006; Tendeng and Houart, 2006). Ectopic Bmp activation markedly reduced the levels of sfrp1a and wif1 (Fig. 5B, D; n=25 for both; also see Fig. S5), whereas expression of the Bmp inhibitors chordin, noggin1 and noggin2 was essentially unchanged or increased (Fig. S5). Both of these secreted inhibitors are capable of binding and inactivating canonical and non-canonical Wnts in the extracellular space (Kawano and Kypta, 2003). Intriguingly, only the posterior expression of sfrp1a was reduced, demonstrating that the regulation of sfrp1a by Bmp signaling is specific to this region of the embryo.
Reduced expression of the secreted Wnt inhibitors suggested that Wnt signaling may act at a greater distance than in a normal embryo. We performed two experiments to examine the role of canonical and non-canonical Wnt signaling in the tailbud during early somitogenesis. Canonical Wnt signaling was assayed directly by staining embryos with an antibody to β–catenin. Transgenic and control embryos were heat shocked at the 3-somite stage and fixed at the 9-somite stage, by which time no tail expression has expanded (Fig. 4 A,B). The β-catenin localization was detected with the antibody and a fluorescent secondary antibody, and the embryos were counterstained with DAPI, then flat-mounted and imaged on a confocal microscope. Nuclear accumulation of β-catenin indicates active canonical Wnt signaling in a cell, but it is difficult to resolve this staining specifically as membrane localized β-catenin is highly abundant in all cells. To identify the domain of Wnt signaling we used the computer program ImageJ to define the region of each nucleus based on the DAPI staining. β-catenin staining that fell outside the nuclear domains was excluded from analysis, as was staining in the dorsal-most (epithelial) cell layer. Examining nuclear β-catenin staining in this manner revealed a significant expansion of active canonical Wnt signaling in response to ectopic Bmp signaling (Fig. 5E-L, n=5 transgenic, 3 wild-type; see also Fig. S6 for higher magnification).
Currently there is no method to detect active non-canonical Wnt signaling within an embryo. However, Stoick-Cooper et al. (2007) recently described a transgenic zebrafish line expressing wnt5a under the control of the heat shock promoter, which was used to analyze limb regeneration in zebrafish. To determine if expanded non-canonical Wnt signaling could induce the altered morphology we observed, we performed lineage labeling as described above following a heat shock at the 3-somite stage. The observed results matched those resulting from ectopic Bmp signaling (Figure 4), with reduced contribution to anterior presomitic mesoderm in transgenic embryos compared to non-transgenic controls (Fig. 6). These results support the hypothesis that enhanced Bmp signaling down-regulates the Wnt inhibitors, resulting in expanded canonical and non-canonical Wnt signaling, which in turn results in expanded no tail expression (due to increased canonical Wnt signaling) and aberrant morphogenesis of the tailbud (due to increased non-canonical Wnt signaling).
A large body of work has shown that Bmp and Wnt signals cooperate to maintain and pattern mesoderm during gastrulation (De Robertis and Kuroda, 2004; Esterberg et al., 2008; Heasman, 2006; Kimelman, 2006; Niehrs, 2004; Schier and Talbot, 2005). Ventral mesoderm is continuously exposed to high levels of both signals during this time, and the cells receiving the greatest signal are set aside as a progenitor pool. These cells migrate directly to the ventral pole, forming part of the tailbud (Myers et al., 2002) and subsequently contribute to somites, forming much of the tissue in the posterior trunk and tail (Dubrulle and Pourquie, 2004; Holley and Takeda, 2002; Tam et al., 2000).
Like the early ventral mesoderm, the progenitor population continues to express Bmps and Wnts. The roles and requirements for these signals at this late stage of development are only now becoming clear with the advent of transgenic approaches that allow these signaling pathways to be turned on and off during the post-gastrula stages. The canonical Wnt signal, for example, was recently shown to be required for maintaining no tail and brachyury expression and for sustaining the progenitor state of the mesodermal progenitors (Martin and Kimelman, 2008). Bmp signaling, in contrast, is required post- early gastrula in the posterior of the embryo to form the ventral fin, suppress the formation of a secondary tail, for notochord growth and for the formation of the cloaca (Connors et al., 2006; Esterberg et al., 2008; Pyati et al., 2006; Pyati et al., 2005; Stickney et al., 2007).
We show here that the zone of active Bmp signaling is restricted to a small zone within the mesodermal progenitors, similar to what has been observed in Xenopus (Beck et al., 2001). As seen in gastrula stages, restriction of Bmp signaling is not just transcriptional, but is also regulated by Bmp inhibitors, which we show are expressed within the presomitic mesoderm and notochord directly anterior to the progenitor zone. Revealing a necessity for the embryo to limit Bmp signaling, we demonstrate that ectopic Bmp signaling results in malformations of the tail. Interestingly, the posterior tissues are the most severely affected, suggesting that the primary role of Bmp signaling during early somitogenesis is to pattern this region. The transgenic zebrafish line expressing an inducible cell autonomous Bmp signaling activator used in this work will be very useful for examining more subtle effects of activating Bmp signaling post-gastrula, and it will be freely available at the Zebrafish International Resource Center.
The defects caused by expanding Bmp signaling involved changes in no tail expression as well as a failure of cells to undergo normal convergence and extension movements, which we demonstrate is due to a cell non-autonomous effect. Importantly, we show that ectopic Bmp signaling inhibits expression of the secreted Wnt inhibitors sfrp1a and wif1 within the posterior region. Down-regulation of the Wnt inhibitors would be expected to allow Wnts produced in the tailbud to act at a greater distance. This fits well with the observation that no tail is not activated ectopically in anterior regions but instead its expression expands beyond its normal tailbud domain. Intriguingly, double and triple knockdown of the mouse sfrp genes (sfrp1, sfrp2 and sfrp5) causes convergence extension defects in the mouse tailbud, suggesting that the mechanisms of tailbud formation observed here are conserved (Satoh et al., 2006; 2008). Moreover, the Bmp-type Smad interacting protein Sip1 directly regulates sfrp1 expression in the mouse hippocampus (Miquelajauregui et al., 2007), suggesting that the mechanism of Bmp signaling regulating Wnt inhibitor expression is conserved in tissues outside of the tailbud.
Based on our results, we propose a model for the role of the Bmp inhibitors during somitogenesis (Fig. 7). The Bmp inhibitors, chordin, noggin1, noggin2 and follistatin-like 1b are expressed in the presomitic mesoderm and notochord, constraining active Bmp signaling to a small region of the mesodermal progenitor zone where it acts to promote ventral fin formation and cloaca development. It is important to constrain Bmp signaling in order to allow the normal expression of the Wnt inhibitors sfrp1a and wif1, which in turn regulate the levels of canonical and non-canonical Wnt signaling. This may explain the result described by Lin and Slack (2008) that inactivation of Bmp signaling results in a loss of Wnt-driven tail regeneration in Xenopus.
While embryos heat shocked at the 3-somite stage show very pronounced defects at the 14-somite stage that continue through later somitogenesis, it is intriguing that axis extension does continue. While we ascribe this partly to the transient nature of the ectopic Bmp signaling, our results demonstrate that the morphogenesis of the embryo is surprisingly robust. Potentially the embryos have a mechanism to correct morphogenetic defects that occur during embryogenesis in order to correct naturally occurring defects.
Our results reveal both similarities and differences between the gastrula and tailbud. In the gastrula, Bmp and canonical and non-canonical Wnt signals are expressed in overlapping regions in the ventral-lateral regions, opposed by Bmp and Wnt inhibitors expressed together on the dorsal side of the embryo. Similarly, Bmp and Wnt signals are expressed posteriorly in the tailbud in an overlapping pattern, opposed by Bmp and Wnt inhibitors expressed together anteriorly (Fig. 7). As in the tailbud, overexpression of the non-canonical Wnts causes convergence extension defects in fish, frogs and chick (Hardy et al., 2008; Heisenberg et al., 2000; Moon et al., 1993; Ungar et al., 1995), demonstrating that the levels of these ligands must be very tightly regulated for normal morphogenesis.
By contrast, Wnt signaling plays a critical role in regulating the size of the organizer region in the early gastrula (Christian and Moon, 1993; Erter et al., 2001; Lekven et al., 2001; Ramel and Lekven, 2004), whereas in the tailbud its role is to sustain the progenitor fate (Martin and Kimelman, 2008). Bmp signaling in the early gastrula is essential for establishing the tail fate (Agathon et al., 2003; Szeto and Kimelman, 2006), regulating the movement of cells (von der Hardt et al., 2007), and limiting the size of the organizer (reviewed in De Robertis and Kuroda, 2004; Heasman, 2006; Stickney et al., 2002), whereas in the tailbud it does not have any of these roles and instead has very different functions. Thus some aspects of the early gastrula are retained in the tailbud whereas other roles have changed. Bmp has transitioned from a potent ventralizing signal to a ligand required in different ways for a number of distinct tissues in the anterior and posterior regions including the cloaca, notochord, liver, heart, white blood cells and ventral fin. These differences underscore the importance of using precise temporal regulation of signal activators and inhibitors to understand the roles of dynamic signaling during embryogenesis (see also Tucker et al., 2008).
Previous authors have argued that tail development is essentially a continuation of gastrulation as shown in Xenopus and chick embryos (Gont et al., 1993; Knezevic et al., 1998). We demonstrate here evidence that developmental modules necessary for gastrulation are also required to pattern the elongating tail even as the precise role of these modules is changing. Moreover, the tissues expressing the factors are also changing. While the notochord, which expresses some of the inhibitors, is derived from the early gastrula organizer region, the presomitic mesoderm expression of the inhibitors is novel; in fact, the cells in the presomitc mesoderm originate in the mesodermal progenitor region and only turn on the Bmp and Wnt inhibitors after they leave the progenitor zone. Thus the transcriptional regulation of these inhibitors must differ from that occurring during the early gastrula. It will be of great interest to understand whether or not this later regulation uses the same types of elements as are used in the early gastrula embryo.
A number of elegant studies have focused on the roles and precise regulation of Bmp and Wnt signaling in the early frog and fish embryo. There is cross-talk between these pathways, and multitudes of factors are involved in regulating the precise levels of the signaling factors. It will be interesting to use this knowledge to determine if this is specific to the gastrulating embryo when so many critical patterning decisions and morphogenetic events are occurring, or if the same complex regulation occurs during somitogenesis. While the tailbud is much smaller than the gastrula embryo, with emerging technologies this type of approach is now becoming feasible.
Fig. S1. Bmp signaling is active in the mesoderm progenitor zone. Active Bmp signaling (red) is principally active in the dorsal-posterior region of the tailbud, where expression of no tail (green) marks the mesoderm progenitor zone. The strongest ntl expression is in the notochord (see Fig. 1A). Shown is a confocal section at the midline of a 3-somite tailbud with dorsal at top and posterior at right.
Fig. S2. Ectopic Bmp signaling disrupts somite morphogenesis. Wild-type (top) and transgenic (bottom) embryos were heat shocked at the 3-somite stage and fixed at approximately 32 hours post fertilization. DAPI staining shows the nuclei aligned along clear borders in the chevron-shaped somites of the wild type embryo. Anterior somites in the transgenic embryo have indistinct or nonexistent borders. Posterior somites show partially normal segmentation but do not form the correct shape.
Fig. S3. Ectopic Bmp signaling is transiently induced by heat shock. Wild-type (left) and transgenic (right) embryos were heat shocked at the 3-somite stage and fixed at the indicated intervals. All were stained for the presence of phosphorylated Smads 1 and 5. In wild type embryos p-Smad is restricted to the tailbud and developing somites. Transgenic embryos show high levels of p-Smad throughout all tissues within 90 minutes of heat shock activation. This appears to peak approximately 2.5 hours after the heat shock ends.
Fig. S4. Ectopic Bmp signaling minimally affects expression of bmp2a and bmp4. Wild-type (left) and transgenic (right) embryos were heat shocked at the 3-somite stage and fixed at the 9-somite stage, then stained for expression of bmp2a or bmp4 by WISH. No change in the expression of bmp2a is observed. The region of bmp4 expression in the tailbud is enlarged very slightly, and a small number of epidermal cells express this ligand ectopically.
Fig. S5. Inhibitors of Bmp and Wnt signaling respond to ectopic Bmp signal. Transgenic and control embryos were heat shocked at 3 somites and fixed at 9 somites. In-situ hybridization demonstrates that ectopic Bmp signaling upregulates expression of noggin1 and noggin2. Expression of chordin may be very slightly downregulated. wif1 expression is reduced but not eliminated. It is shown in combination with pax2.1 staining, which does not change under these conditions.
Fig. S6. Nuclear β-catenin can be measured by comparison to DAPI staining. Optical section of a tailbud imaged laterally, stained with DAPI and antibody against β-catenin. The high levels of β-catenin at the cell membrane obscure other staining within the cell. Using the computer program ImageJ to define the area of the nuclei based on DAPI staining allows for specific exclusion of non-nuclear staining. The resulting composite clearly shows regions of active canonical Wnt signaling.
We thank Ben Martin and Doug Weiser for their valuable comments on the manuscript, and we greatly appreciate Jeff Wrana and Liliana Attisano, Thomas Czerny, David Parichy, Koichi Kawakami, Jen-Chieh Hsieh and Corrine Houart for providing plasmids. We are indebted to Scott Holley for providing his fluorescent in situ protocol along with valuable advice, and to Christi Stoick-Cooper and Randall Moon for providing Tg(HSwnt5a) embryos. This work was supported by the NSF (IBN-00783030) and NIH (GM079203) grants to DK. R.R. was supported by NSRA Grant T32 HD007183-26A1 from the NIH-NICHD.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Richard Row, Department of Biochemistry, University of Washington, Seattle, WA 98195-7350.
David Kimelman, Department of Biochemistry, University of Washington, Seattle, WA 98195-7350.