BMP antagonism is required in both the node and lateral plate mesoderm for mammalian left-right axis establishment

The first morphological indication of mammalian left-right (L-R) asymmetry occurs during somitogenesis stages, in the direction of heart looping. Later,heart, lungs, liver, spleen, intestines and the vascular system all display asymmetry about the L-R axis. Morphological abnormalities and misalignments of organs resulting from laterality defects can cause disease or death. In humans, clinically significant laterality defects occur in at least 1 in 10,000 births (Peeters and Devriendt,2006).

This L-R asymmetric organ morphogenesis depends on earlier L-R axis formation. In mouse, this is established around the node, the location of the Spemann organizer at the end of gastrulation, and is propagated from the midline to the lateral plate mesoderm (LPM)(Shiratori and Hamada, 2006). Nodal, a member of the transforming growth factor beta (TGFβ) superfamily of secreted ligands, is a crucial left-side determinant. Nodalexpression occurs peripheral to the node and then in the left LPM(Collignon et al., 1996). Perinodal Nodal expression is required for Nodal expression in the left LPM (Brennan et al.,2002; Saijoh et al.,2003). An initially low level of Nodal in the LPM can induce Nodal expression itself, via a positive-feedback mechanism(Saijoh et al., 2000). Nodal also activates downstream target genes in the left LPM, such as Lefty2, an inhibitor of Nodal itself(Saijoh et al., 2000), and Pitx2, which regulates left-side-specific morphogenesis(Logan et al., 1998; Yoshioka et al., 1998). Inappropriate Nodal expression in the LPM thus leads to severe laterality defects, underscoring the importance of determining how the asymmetric expression of Nodal is regulated.

Bone morphogenetic proteins (BMPs), another class of the TGFβsuperfamily, play an important role in regulating Nodal expression in the LPM. In chick embryos, a Cerberus-like factor, Caronte, is expressed in the left paraxial mesoderm and left LPM, where it promotes Nodal expression; it appears to do so by inhibiting Bmp2, Bmp4 and Bmp7 in the LPM, suggesting a negative role for BMP in regulating Nodal (Yokouchi et al.,1999; Rodriguez-Esteban et al., 1999). However, beads coated with BMP in the right LPM activated Nodal, whereas beads coated with the BMP antagonist noggin placed in the left LPM blocked Nodal expression(Piedra and Ros, 2002). Thus,data from the chick system have suggested both positive and negative roles for BMP in regulating asymmetric Nodal expression.

How these results from the chick model relate to the roles of BMP in mammalian L-R formation is unclear. Molecular regulation of L-R axis formation may differ between chick and mouse (Meyers and Martin, 1999), and there does not appear to be a mammalian Caronte ortholog. In mouse, Bmp2 and Bmp4 are expressed symmetrically in the LPM(Fujiwara et al., 2002) and are present at the right time and place to influence LPM Nodalexpression. Experiments in mouse embryos have also suggested either positive or negative roles for BMP signals in regulating Nodal. Without embryonic Bmp4, Nodal is not expressed, and embryos cultured with the BMP antagonist noggin (Nog) do not express Nodal in the LPM(Fujiwara et al., 2002). These results suggest a positive role. Nevertheless, embryos lacking the BMP signaling component genes Smad5 or Alk2 (Acvr1 -Mouse Genome Informatics) show bilateral Nodal expression in the LPM(Chang et al., 2000; Kishigami et al., 2004),suggesting a negative role in the mouse. This conclusion is consistent with data from other vertebrates; for example, exogenous introduction of a constitutively active BMP receptor causes diminished Nodal expression in the Xenopus embryo (Ramsdell and Yost, 1999). Similarly, ectopic expression of bmp2bthroughout the zebrafish embryo causes diminished Nodal expression(Chocron et al., 2007). Overall, it is clear that BMP signaling plays an important role in regulating asymmetric Nodal expression during mammalian development, but both its function in doing so and the manner in which it is regulated remain unresolved.

The organizer BMP antagonists Nog and chordin (Chrd) regulate BMP signaling prior to ligand-receptor binding (Balemans and Van Hul, 2002). Chrd;Nog double-mutant mouse embryos exhibit defects in the formation of all three embryonic axes(Bachiller et al., 2000), but there has been no analysis of the role of these factors in establishing the L-R axis. Here we use genetics and molecular embryology to study the roles of BMP signaling and BMP antagonism in regulating Nodal expression during mammalian L-R axis establishment.

Wild-type embryos were generated from random outbred ICR stock (Harlan). Nog^9e and Chrd^tm1Emdr mutations were maintained in an outbred, ICR background as described(Anderson et al., 2002). Chrd is homozygous viable in this genetic background, without the phenotypes of the specific backgrounds previously reported(Bachiller et al., 2003). Nodal^tm1Rob (Collignon et al., 1996) and Bmp4^tm2Blh(Lawson et al., 1999)heterozygotes were maintained in an ICR genetic background. Triple and quadruple mutants were generated by crossing Chrd^-/-;Nog^+/-;Bmp4^+/-with Chrd^-/-;Nodal^+/-.

Embryos were collected at E8.0-8.5 or E9.0-9.5, or after culture, fixed at 4°C overnight in 4% paraformaldehyde in PBS, washed in PBS and stored at-20°C in methanol. After gene expression analysis, genomic DNA was prepared from each embryo and genotyped by PCR as described: Nog and Chrd (Anderson et al.,2002); Bmp4(Stottmann et al., 2006); Nodal (Collignon et al.,1996).

Whole-mount in situ hybridization (WMISH) was performed according to standard procedures (Yamamoto et al.,2004) with the following probes: eGFP, Lefty2, Lefty1(Nakamura et al., 2006); Lfng (Kume et al.,2001); Cryptic (Shen et al., 1997); Nodal(Collignon et al., 1996); Pitx2c (Liu et al.,2001); Nog (McMahon et al., 1998); and Chrd(Klingensmith et al., 1999). lacZ staining was performed as described(Stottmann et al., 2001). Quantitative (q) RT-PCR used total RNA prepared from 20 pieces of left or right 4-5s LPM using Trizol (Invitrogen) and glycogen carrier (Ambion), with reverse transcription employing a Taqman RT-PCR Kit (Applied Biosystems) after DNase I treatment (Ambion). qPCR was performed using the MyiQ Real-Time PCR System and SYBR Green mix (Bio-Rad) with the following primers (forward and reverse): Nog, 5′-TTTTGGCCACGCTACGTGAA-3′ and 5′-CTAGCAGGAACACTTACACT-3′; Chrd,5′-TTCCCAGAGAATCAGAGCTG-3′ and 5′-TCTGGAAGGGTTCTAGTCTC-3′; Nodal,5′-ACTTTGCTTTGGGAAGCTGA-3′ and 5′-ACCTGGAACTTGACCCTCCT-3′; Bmp4,5′-AGACCCTAGTCAACTCTGTT-3′ and 5′-CTCTACCACCATCTCCTGAT-3′; β-actin,5′-AAGAGCTATGAGCTGCCTGA-3′ and 5′-CACAGGATTCCATACCCAAG-3′. Whole-mount immunostaining was performed as described with anti-phospho-Smad1/5/8 antibody (Cell Signaling)(Yang and Klingensmith, 2006)or anti-acetylated tubulin (Sigma) (Nakaya et al., 2005). Node areas were calculated by NIH Image software and statistical treatment used the χ² test.

For whole embryo culture, headfold-stage embryos were isolated and parietal endoderm removed. Liposomes composed of expression vectors and Lipofectamine 2000 (Invitrogen) were injected between the endoderm and LPM (see Fig. S3 in the supplementary material). Expression vectors were eGFP, with the eGFP gene in the pCAGGS vector(Okabe et al., 1997), and Nog, Bmp4 or Nodal, in which each coding region was inserted into pEF-BOS (Mizushima and Nagata,1990). For control embryos, GFP vector was injected alone. For experimental embryos, equal concentrations of eGFP and secreted product vectors were used. Liposome solution was prepared as follows: 12.5μl OPTI-MEM (Invitrogen) was mixed with 1 μl Lipofectamine 2000 or with 1 μg expression vector(s). These solutions were combined and incubated for 5 minutes at room temperature; 0.1 μl solution was injected. Embryos were cultured with rotation for 14 hours in a humidified atmosphere of 5%CO₂/95% air at 37°C in Dulbecco's Modified Eagle's Medium(DMEM) with 50% rat serum. Embryos with direct eGFP fluorescence in the desired regions were selected for WMISH with probes for GFP and a relevant marker. For Nodal signaling experiments, 3s ICR embryos were cultured with DMSO (0.1% in DMEM) or SB431542 (Sigma) in 0.1% DMSO (100 μg/ml) for 3 hours until 5s.

Node explants (including peripheral tissues) were isolated from 1-3s ICR embryos and cultured for 4 hours at 37°C and 5% CO₂ in DMEM containing 10% FCS. Control culture was performed with BSA (200 ng/ml). Experimental culture was performed with recombinant human BMP2 (200 ng/ml)(R&D Systems). Twenty explants from either treatment were used to isolate total mRNA. qRT-PCR was performed using the following primers (forward and reverse): Lfng, 5′-CACCATTGGCTACATTGTAG-3′ and 5′-CAAACATGCCATAGCTTCAGG-3′; Dll1,5′-AAGTGCCAGTCACAGAGCTC-3′ and 5′-TGCAGACAGAACATACACCG-3′; Notch1,5′-AGTCAGGCAGATGTACAACC-3′ and 5′-AGGAACTGGGTAGTGGTCAT-3′; Rbpjk,5′-ACCTTCACCTACACACCAGA-3′ and 5′-GACGATGTGACACTGGTAGA-3′.

Given the ambiguity concerning the roles of BMP signaling in regulating Nodal expression during mouse L-R axis formation, we wanted to know where BMP signaling was active in the relevant spatiotemporal context. Nodal is expressed in the left LPM at the 3- to 5-somite stages(3-5s) (Collignon et al., 1996; Nakamura et al., 2006). Accordingly, we assayed the spatiotemporal distribution of BMP signaling activity during early somite stages. We used immunohistochemistry to visualize binding of antibodies to phosphorylated Smads 1, 5 and 8 (p-Smad1/5/8; Smad 8 is also known as Smad9 - Mouse Genome Informatics) in mouse embryos. Activation of the BMP receptor complex by BMP ligand binding causes Smad1/5/8 phosphorylation and subsequent signal transduction(Kishigami and Mishina, 2005). As previously shown (Yang and Klingensmith, 2006), the distribution of active BMP signaling during gastrulation and neurulation largely coincides with expression of Bmp4 and other BMP genes (Furuta et al., 1997; Solloway and Robertson, 1999). At 2s, p-Smad1/5/8 expression was observed bilaterally in all embryos. At 3s, 75% of embryos (9/12) still expressed bilaterally, but 25% (3/12) showed weaker expression in left LPM. Strikingly asymmetric p-Smad1/5/8 distribution was observed in all embryos at 4-5s, with elevated levels in the right LPM relative to the left(Fig. 1A-C). By contrast, Bmp4 is expressed bilaterally in the LPM at 4-5s(Fig. 1D,E)(Fujiwara et al., 2002). Expression of Bmp2 is bilateral like Bmp4(Fujiwara et al., 2002), and Bmp7 is in the node and midline(Solloway and Robertson,1999). In summary, we observed higher levels of BMP signaling activity in the right LPM than in the left, although none of the known BMP ligands active at this time shows a similarly biased expression pattern. This left-sided decrease in BMP signaling occurs at or just before the time when Nodal begins to be expressed in the left LPM.

Our results suggest that asymmetric activation of BMP signaling in the mammalian LPM does not depend on BMP gene expression patterns. It might instead be created by asymmetric BMP antagonist activity. We therefore examined expression of Chrd and Nog in this context. At embryonic day 8 (E8.0), Nog is expressed robustly in the node,notochord and lateral neural folds(McMahon et al., 1998), and at lower levels in heart, allantois and LPM. At 4-5s, elevated expression was observed in the left LPM relative to the right(Fig. 1F-H). Chrd was expressed broadly at low levels, with higher levels in node and notochord;however, we also detected increased Chrd hybridization in left LPM relative to the right (Fig. 1I-K). Embryo sections confirmed that these expression domains are in the LPM itself, rather than in adjacent germ layers (see Fig. S1 in the supplementary material). We independently assayed BMP antagonist expression by quantitative RT-PCR (qPCR) using RNA isolated from left or right LPM at 4-5s,finding that both Nog and Chrd are expressed at higher levels in the left LPM (Fig. 1L). Together, these results indicate that the BMP antagonists Nog and Chrd are expressed at elevated levels in left LPM(Fig. 1M), implicating them in producing asymmetric BMP signaling activity during early L-R axis formation.

Consistent with the hypothesis that Chrd and Nog are involved in mammalian L-R axis formation, the double-null mutant(Chrd^-/-;Nog^-/-) has defects in directionality of heart looping (Bachiller et al.,2000) (Fig. 2A-D;see Table S1 in the supplementary material). By contrast, Chrd^-/-;Nog^+/- and Chrd^-/-;Nog^+/+ embryos all exhibited normal heart looping (Fig. 2D). These results suggest that the BMP antagonists Nog and Chrd function redundantly in L-R axis formation. We therefore investigated the expression of Nodaland other asymmetrically transcribed genes in Chrd^-/-;Nog^-/- embryos. Nodalexpression was greatly diminished or absent in 82% of embryos (n=11)(Fig. 2E,F). A few embryos (9%)expressed Nodal in right LPM (Fig. 2I). Expression of Lefty2 and Pitx2, targets of Nodal in the left LPM (Hamada et al., 2002), was also diminished or absent in most Chrd^-/-;Nog^-/- embryos(Fig. 2F,H,I; see Fig. S2 in the supplementary material). Expression of Cryptic (Cfc1 -Mouse Genome Informatics), a Nodal signaling component, was unchanged (see Fig. S2 in the supplementary material). These results suggest that Nog and Chrd are required for mammalian L-R axis establishment by promoting Nodal expression in the left LPM.

A key factor in initiating Nodal expression in the left LPM is Nodal expression around the node(Brennan et al., 2002; Saijoh et al., 2003). Given that Chrd and Nog are expressed in the node(Klingensmith et al., 1999),and that LPM expression of Nodal is absent in most Chrd^-/-;Nog^-/- mutants, we also examined perinodal expression of Nodal. This assay revealed three populations:∼29% of Chrd^-/-;Nog^-/- mutants (7/24)expressed perinodal Nodal at essentially the same level as in wild type (denoted ++); ∼58% of mutants (14/24) showed greatly reduced expression (+); and 12% showed no detectable Nodal expression around the node (-) (Fig. 3A-E). These data reveal that Chrd and Nog play an important role in promoting perinodal Nodal expression, but are not necessarily essential for its expression.

We correlated the level of Nodal expression around the node with the status of Nodal expression in the LPM in the Chrd^-/-;Nog^-/- mutants, revealing two populations. Mutants showing essentially normal levels of Nodalaround the node (the `++' class) showed weak but detectable expression of Nodal in the LPM (3/7). By contrast, embryos showing weak (+) or absent (-) Nodal expression around the node (17/17) showed no expression of Nodal in the LPM(Fig. 3F). Thus, Chrdand Nog promote Nodal expression around the node and subsequent expression in LPM.

These results imply that ectopic BMP activity around the node would downregulate local Nodal expression. We observed ectopic anti-p-Smad1/5/8 staining around the nodes of Chrd^-/-;Nog^-/- mutants relative to wild-type embryos at the stages we assayed, including presomitic bud-stage embryos (data not shown) and 4-5s embryos (see Fig. S3 in the supplementary material). This indicates that BMP signaling is indeed increased perinodally in the absence of Chrd and Nog. To directly test the consequences of ectopic BMP around the node on Nodal expression, we cultured explants isolated from 1-3s wild-type embryos with recombinant human BMP2 protein. Whereas the control carrier protein BSA had no affect, ectopic BMP markedly decreased perinodal Nodal expression (see Fig. S3 in the supplementary material).

Notch signaling is a positive regulator of perinodal Nodalexpression (Krebs et al.,2003; Raya et al.,2003). To assess Notch signaling in Chrd^-/-;Nog^-/- embryos, we assayed expression of lunatic fringe (Lfng), a positive transcriptional target in the presomitic paraxial mesoderm (Morales et al., 2002). We observed sharply reduced levels of Lfngexpression just lateral and anterior to the node at the early somite stage in the mutants (see Fig. S3 in the supplementary material), suggesting that BMP reduces the activity of the Notch pathway. To test this, we cultured explants from the node region of wild-type embryos with BMP2. Expression of the Notch signaling targets Dll1 and Lfng was reduced (see Fig. S3 in the supplementary material). We also observed dysmorphic nodes in Chrd^-/-;Nog^-/- embryos, of abnormal shape and with fewer cilia (Fig. 3G-I). Along with L-R heart looping defects, similarly dysmorphic nodes and absent perinodal Nodal expression have been observed previously in embryos lacking the Notch signaling components Dll1(Przemeck et al., 2003) and Baf60c (Smarcd3 - Mouse Genome Informatics)(Takeuchi et al., 2007). Collectively, these findings indicate that BMP antagonism in the node via Chrd and Nog is necessary for expression of Nodal around the periphery of the node, and suggest that the loss of this expression in Chrd^-/-;Nog^-/- embryos results from decreased activity of the Notch pathway.

Some Chrd^-/-;Nog^-/- mutants with significant perinodal Nodal expression nevertheless lacked Nodalexpression in the left LPM (Fig. 3B,F). Moreover, in wild-type embryos, both Chrd and Nog are elevated in left LPM, whereas BMP signal transduction through Smad1/5/8 is reduced there relative to the right LPM. These findings raise the possibility that beyond promoting Nodal expression around the node, Nog and Chrd have an additional direct function to promote Nodal expression in the LPM. Accordingly, we investigated BMP signaling activity in the LPM of Chrd^-/-;Nog^-/-embryos by assaying p-Smad1/5/8 staining. We observed increased staining, with bilaterally equivalent levels in left and right LPM(Fig. 4A,B), indicating that Chrd and Nog are required for the left-sided reduction in BMP signaling.

To directly test whether BMP antagonism in the left LPM per se can promote Nodal expression therein, we introduced a Nog expression vector into the left LPM of Chrd^-/-;Nog^-/-embryos at the late headfold stage (schematized in Fig. S4 in the supplementary material). GFP was included with Nog to mark transfected cells. In control injections, we introduced the GFPvector alone. Embryos were then cultured to 5-6s and assayed by in situ hybridization for Nodal and GFP. The results are summarized in Fig. 4F. Injection of Nog and GFP vectors into the left LPM of Chrd^-/-;Nog^-/- embryos resulted in Nodal expression in left LPM in 40% of embryos (10/25), whereas introduction of GFP vector alone resulted in 4% of mutants (1/23)expressing Nodal in the left LPM(Fig. 4Fa), a highly significant difference (P<0.001).

We then considered how the expression of Nodal in the left LPM related to the expression of Nodal around the node in the mutants injected with Nog and/or GFP vectors(Fig. 4Fb). Among embryos showing robust Nodal expression around the node (++), 4/6 also showed Nodal expression in the LPM when Nog was ectopically expressed there. By contrast, only 1/5 of such embryos expressed Nodal in the left LPM when only GFP was injected. In Chrd^-/-;Nog^-/- embryos with low perinodal Nodal expression (+), 6/15 showed Nodal in the left LPM when Nog was injected into this tissue(Fig. 4E). None (0/14)expressed Nodal in the left LPM when only GFP was introduced(Fig. 4D). This difference is significant (P=0.007). When the mutant embryos showed no detectable Nodal around the node, however, ectopic Nog in the left LPM never resulted in Nodal expression there (0/4). These data indicate that BMP antagonism in the left LPM promotes Nodal expression in this tissue, but BMP antagonism might not be sufficient for Nodalexpression in the left LPM. Instead, it is likely to depend on synergistic influences from the node region, such as Nodal itself. Thus, BMP antagonism appears to create a permissive environment for Nodalexpression in the left LPM.

The direct, positive effect of BMP antagonist expression on Nodalexpression in the left LPM of the Chrd^-/-;Nog^-/- double-null implies that local BMP activity in the LPM inhibits Nodal expression. We tested this by introducing a Bmp4 expression vector into the left LPM of wild-type embryos. About 80% (16/20) of such embryos showed absent Nodalexpression in the left LPM, versus 4/18 when the vector contained GFPalone (Fig. 4G,H,K), a highly significant increase (P<0.001). This manipulation increased left-side BMP signaling (Fig. 4G,H). We created a similar imbalance by expressing ectopic Nog in the right LPM of wild-type embryos. This resulted in ectopic right-sided Nodal expression in several embryos: 4/15 showed bilateral Nodal expression in the LPM, and one showed only right-sided expression in the LPM (Fig. 4J,L). When the control GFP vector alone was transfected into the right LPM, none (0/16) showed right-sided Nodal expression(Fig. 4I,L), again a significant difference (P=0.023). These results indicate that BMP signaling in the LPM represses Nodal expression there, whereas local BMP antagonism in the LPM derepresses Nodal transcription in this tissue.

Nodal promotes its own expression in the left LPM via a positive-feedback loop (Shiratori and Hamada,2006), and our results indicate that BMP antagonism also promotes Nodal expression in the LPM. To genetically assess the relevance of our embryo culture experiments to Nodal regulation, we produced embryos mutant for combinations of Chrd, Nog, Nodal and Bmp4null alleles, then assayed Nodal expression. Embryos of the control genotypes Nodal^+/-, Chrd^-/- and Chrd^-/-;Nog^+/- were indistinguishable from wild type, showing normal robust Nodal expression in the node and left LPM (Fig. 5A,B,C,F). By contrast, Chrd^-/-;Nodal^+/- and Chrd^-/-;Nog^+/-;Nodal^+/-embryos fell into two distinct classes (see Table S2 in the supplementary material). One had robust Nodal expression around the node, as in wild-type embryos (denoted ++). Among these, 66% of Chrd^-/-;Nodal^+/- embryos and only 12.5% of Chrd^-/-;Nog^+/-;Nodal^+/-embryos showed Nodal expression in the left LPM(Fig. 5D,G). Embryos in the second class showed markedly reduced perinodal Nodal expression (+),of which ∼42% of Chrd^-/-;Nodal^+/-embryos expressed Nodal in the left LPM but at reduced levels,whereas the rest showed no expression (Fig. 5E). Strikingly, no Chrd^-/-;Nog^+/-;Nodal^+/-embryos showed any expression of Nodal in the LPM(Fig. 5H). Thus Chrd,Nog and Nodal have a positive genetic interaction that promotes robust Nodal expression in the left LPM.

Bmp4 is expressed bilaterally in the LPM(Fig. 1D,E)(Fujiwara et al., 2002),raising the possibility that it is an endogenous BMP ligand antagonized by Chrd and Nog in the left LPM. We therefore reduced the gene dosage of Bmp4 in the triple Chrd^-/-;Nog^+/-;Nodal^+/-mutants to see whether the phenotypic defects were lessened. Indeed, many quadruple mutant Chrd^-/-;Nog^+/-;Nodal^+/-;Bmp4^+/-embryos showed rescue of Nodal expression in the left LPM, with 42%expressing robust levels of perinodal Nodal (++) and 25% expressing low levels (+) (Fig. 5J,K). By contrast, when Bmp4 dosage was normal, only 13% of Chrd^-/-;Nog^+/-;Nodal^+/- embryos showed Nodal expression in LPM when perinodal expression was robust(Fig. 5G), and none showed such expression when perinodal expression was weak(Fig. 5H). These results suggest that endogenous left-side elevated Chrd and Nogexpression permits activation of a Nodal positive-feedback loop in the left LPM by antagonizing local Bmp4 activity.

Given the significance of Nodal as a left-side determinant, we considered whether the asymmetry in BMP antagonist expression is initially established by left-side-specific Nodal signaling. Nog expression in the left LPM correlates spatially and temporally with Nodal expression (see Fig. S6 in the supplementary material); moreover, we never observed asymmetric Nog expression in the LPM of Chrd^-/-;Nog^-/- embryos(Fig. 6B,D), in contrast to wild type (Fig. 1) and Nog^-/- homozygotes(Fig. 6A,C). Also, Nodal is expressed in the left LPM of Nog^-/- (see Fig. S5 in the supplementary material) but not of Chrd^-/-;Nog^-/- embryos(Fig. 2F,I). These data suggest that elevated BMP antagonist expression in the left LPM might be a product of left-side-specific Nodal signaling.

To explore this possibility, we cultured wild-type embryos with a specific inhibitor of Nodal signaling, SB431542(Inman et al., 2002). It inhibits activation of TGFβ signaling by blocking the phosphorylation ability of the activin type I receptors Alk4, 5 and 7 (Acvr1b, Tgfbr1 and Acvr1c, respectively - Mouse Genome Informatics) without affecting BMP signaling activation (Inman et al.,2002). SB431542 can block both endogenous and exogenous signaling activation via Smad2 phosphorylation in embryos(Ho et al., 2006). We first evaluated expression of the Nodal target gene Lefty2 in embryos subjected to SB431542 or to the carrier, DMSO. Embryos cultured with DMSO alone showed normal Lefty2 expression in left LPM(Fig. 6E), whereas embryos cultured with SB431542 lacked Lefty2 expression(Fig. 6F). Thus, Nodal signaling is inhibited by SB431542 treatment in this assay. The DMSO-treated embryos showed normal elevation of Nog expression in the left LPM(Fig. 6G). By contrast, embryos cultured with SB431542 lacked left-side elevation of Nog expression,it being reduced to approximately the same basal level as in the right LPM(Fig. 6H). We also assayed BMP antagonist expression by qPCR in left and right LPM of 4-5s embryos treated with DMSO or SB431542. Both Nog and Chrd expression levels in the left LPM were dramatically decreased by SB431542 treatment (see Fig. S7 in the supplementary material).

Lastly, we investigated the consequences of ectopic Nodal in the right LPM on BMP antagonist expression. Control embryos injected with the GFP expression vector showed normal Lefty2 expression in left LPM and left-side elevation of Nog expression(Fig. 6I,K). As expected from previous results (Nakamura et al.,2006), embryos injected with the Nodal vector into the right LPM showed reversed, right-side expression of Lefty2(Fig. 6J). Thus, these conditions created a L-R reversed Nodal signaling context. In such embryos, Nog expression was elevated on the right side rather than on the left(Fig. 6L). Altogether, these results reveal that the left-side-specific elevation of Nogexpression is produced by Nodal-dependent left-side determination, and suggest that increased BMP antagonist expression in the left LPM is induced by Nodal signaling.

Here we define for the first time the functions of endogenous BMP antagonism in the establishment of the L-R axis of mammalian embryos. In addition to elucidating crucial roles for BMP signaling attenuation by Chrd and Nog, we demonstrate directly the controversial role of BMP activity in regulating Nodal expression in the LPM. We found that endogenous levels of BMP signaling activity are higher in the right LPM than in the left;by contrast, Chrd and Nog are both expressed at higher levels on the left. In vivo, these BMP antagonists are necessary to create this asymmetry of BMP activity in the LPM. In Chrd;Nog mutants, Nodal expression in the left LPM is reduced or absent. This implies that BMP in the LPM represses Nodal expression, and that BMP antagonists function there directly to protect Nodal expression. We confirmed these roles by manipulating BMP signaling and antagonism specifically in the LPM of cultured embryos. Further embryo culture experiments showed that Nodal activity is responsible for this left-side-specific upregulation of BMP antagonists. In addition to promoting Nodal activity in the left LPM, Chrd and Nog also promote Nodalexpression around the node. Consistent with our embryological results, genetic crosses suggested that Chrd, Nog and Nodal act synergistically in L-R patterning, whereas Bmp4 acts antagonistically to these factors.

Thus, our data indicate that Chrd and Nog function together to facilitate at least two major aspects of L-R axis establishment. First, they are required in the node for normal node morphology and for perinodal expression of Nodal. They are then necessary in the left LPM to diminish BMP signaling activity in this domain, and thus inhibit the repressive effect of BMP on the Nodal autoregulatory loop. These findings lead us to propose a model for the mechanisms by which Chrd and Nog function in the establishment of the L-R axis (Fig. 7).

A crucial leftward flow created by rotating monocilia in the node leads to the asymmetric distribution of extracellular cues, with subsequent transfer of asymmetry cues to the left LPM (Shiratori and Hamada, 2006). Perinodal expression of Nodal is required for Nodal expression in left LPM(Brennan et al., 2002; Saijoh et al., 2003), and may itself be a component of the laterality signal transferred from the node to the LPM (Oki et al., 2007). Chrd;Nog mutant embryos form nodes, but of abnormal morphology and reduced cilia density. They have a variable reduction of Nodalexpression around the node, ranging from nearly normal to none detectable. These data indicate that BMP antagonists promote perinodal expression of Nodal but are not essential for it.

The initial L-R defects in Chrd;Nog mutants might be explained by one or more of the following mechanistic possibilities. The sparse cilia or dysmorphic shape in Chrd;Nog nodes might be inadequate to generate sufficient levels of essential leftward cues. Or, an unknown BMP-like molecule might directly repress perinodal Nodal expression, and the lack of local BMP antagonism in the mutants would leave it unopposed. A third possibility is that Chrd;Nog mutants lack sufficient activity of a positive regulator of Nodal expression in this domain. Data suggest this might be the Notch pathway, which is known to promote Nodalexpression around the node (Krebs et al.,2003; Raya et al.,2003). We observed reduced Lfng and Dll1expression, genes that are positively regulated by Notch signaling, in the vicinity of the node. The phenotypes we observed of heart looping defects,dysmorphic nodes and reduced perinodal Nodal expression are similar to those of embryos lacking the Notch signaling components Dll1(Przemeck et al., 2003) and Baf60c (Takeuchi et al.,2007). These findings imply an antagonistic relationship between BMP and Notch signaling in L-R patterning at the node. A precedent for such a relationship has been documented in the cerebellar rhombic lip(Machold et al., 2007).

We observed higher levels of endogenous BMP signaling activity in the right LPM, i.e. in the side lacking Nodal expression. Our site-specific manipulation experiments demonstrated that BMP signaling in the LPM leads to local repression of Nodal expression. Introduction of a Bmp4expression vector into portions of the left LPM resulted in a coincident reduction in Nodal expression, whereas exogenous Nogsimilarly introduced into the right LPM caused ectopic Nodalexpression in the right LPM. These results strongly support the conclusion that the right-sided elevation of endogenous BMP signal activity functions to repress Nodal expression in the right LPM.

Additional evidence from Xenopus, zebrafish, chick and mouse studies also supports a negative regulatory role for BMP signaling on Nodal expression in the LPM (see Introduction). Nevertheless, some data suggest a positive role for BMP signaling in regulating Nodalexpression in the LPM. Of greatest relevance is a previous report in mouse that Bmp4 is a positive regulator of Nodal expression in the left LPM(Fujiwara et al., 2002), which is the opposite conclusion to ours. Fujiwara et al. found that when Bmp4 was absent in extraembryonic as well as embryonic tissues, node morphology was abnormal and Nodal expression was absent. When only embryonic expression was missing, node morphology was restored but Nodal was still absent, in both node and LPM. However, because Bmp4 is never expressed in the node or its vicinity, it seems very unlikely that Bmp4 has a local role in regulating Nodal expression around the node. It is more likely that the node was functionally abnormal owing to defects in primitive streak or mesodermal development, despite looking morphologically intact.

To determine whether there might be a positive role for BMP signaling in the expression of Nodal in the LPM, Fujiwara et al.(Fujiwara et al., 2002)cultured wild-type embryos with Nog after Nodal expression was established around the node; this resulted in a lack of Nodal in the LPM. By contrast, our data demonstrate that BMP activity has a negative role directly in the LPM to downregulate Nodal. Moreover, our genetic crosses demonstrate that Bmp4 per se has an antagonistic relationship with Chrd, Nog and Nodal in expression of Nodal in the LPM. We speculate that the discrepancy between the results of these studies might be owing to experimental differences in the timing or specificity of manipulations. For example, because the exposure to Nog recombinant protein in these cultured embryos was ubiquitous rather than specific to the left LPM, we suggest that the lack of Nodalexpression might have been due to an indirect effect of defective production or transport of the laterality signal from the node to the LPM, rather than to a direct effect on the LPM.

Our analysis revealed asymmetric Nog and Chrd expression in the LPM. Elevation of expression of these genes on the left side is consistent with the endogenous right-side elevation of BMP signal activity we observed in wild-type embryos. By contrast, Chrd;Nog embryos showed bilaterally equivalent BMP signal distribution in the LPM. Introduction of exogenous Nog directly into the LPM of Chrd;Nog embryos rescued Nodal expression in the LPM in some embryos. Thus, Nog and Chrd promote Nodal expression in both the node and LPM, and function synergistically to help establish the L-R axis sequentially in the node and LPM.

The functional significance of this asymmetric BMP antagonist distribution was further supported by the genetic interaction of Chrd, Nog and Nodal. Chrd^-/-;Nog^+/-;Nodal^+/-embryos showed diminished Nodal expression in LPM even in those embryos having robust Nodal expression around the node. Removal of one functional allele of Bmp4 in this compound mutant partially rescued Nodal expression in the left LPM, implying that endogenous Nog and Chrd function in the left LPM to protect Nodal expression by antagonizing local Bmp4.

Robust Nodal expression in the LPM is established by a positive-feedback loop mechanism dependent on Nodal itself(Saijoh et al., 2000). An early step toward Nodal expression in the left LPM appears to be the transfer of a small amount of Nodal from the node(Oki et al., 2007). We showed that asymmetric Nog expression in LPM is induced by Nodal. Accordingly, the physiological significance of the endogenous asymmetric BMP antagonist expression in the LPM might be to maintain this Nodal positive-feedback loop. Meanwhile, left-side expression of Nodal is suggested to suppress Nodal expression on the right side through induction of Nodal inhibitors (Nakamura et al., 2006). Our finding of right-side elevated BMP signal distribution and its repressive effect on Nodal expression might function synergistically with this mechanism.

We thank Dr H. Hamada for rat serum and expression vectors; Drs E. Robertson and B. Hogan for the Nodal and Bmp4 mouse mutants;Drs B. Hogan, E. Robertson, M. Shen, H. Hamada and J. Martin for plasmids; T. Sitzman and K. Carmody for mouse husbandry; and Dr B. Hogan and L. Custer for comments on the manuscript. This work was supported by a fellowship grant from the TOYOBO Foundation to N.M. and by awards from NIH to J.K.