Reversal of left-right asymmetry induced by aberrant Nodal signaling in the node of mouse embryos

Establishment of the left-right (L-R) axis is fundamental for morphogenesis of visceral organs (Hamada et al., 2002; Shiratori and Hamada, 2006). The L-R axis of the mouse embryo is established by successive processes that begin with a leftward flow of fluid on the ventral surface of the node (nodal flow). Asymmetric expression of Nodal and Cerl2 (Dand5 - Mouse Genome Informatics) subsequently develops in the crown cells of the node, with the level of Nodal expression being higher on the left side and that of Cerl2 expression being higher on the right. It has been suggested that Cerl2 binds Nodal and thereby inhibits its activity, with the result that more active Nodal probably emanates from the left side of the node than from the right side (Marques et al., 2004). The Nodal signal is then transmitted to the lateral plate mesoderm (LPM), where it induces Nodal expression (Oki et al., 2007). Positive and negative regulatory loops amplify the small difference in gene expression in the node to generate the robust expression of Nodal in the left LPM (Nakamura et al., 2006). After Nodal expression appears in the left LPM near the node, the positive regulatory loop extends the Nodal expression domain to the entire left LPM (Norris et al., 2002; Yamamoto et al., 2003; Nakamura et al., 2006). Conversely, Lefty1 and Lefty2, which are feedback inhibitors of the Nodal signal, prevent Nodal expression in the right LPM (Meno et al., 1998; Meno et al., 2001). Nodal in the left LPM then induces the expression of Pitx2, which encodes a transcription factor required for correct L-R morphogenesis (Logan et al., 1998; Piedra et al., 1998; Ryan et al., 1998; Lin et al., 1999; Lu et al., 1999; Shiratori et al., 2001).

Whereas this sequence of principal events in establishment of the L-R axis has been relatively well characterized, the initial molecular mechanism by which asymmetric gene expression develops at the node remains largely unknown. This lack of knowledge is exemplified by the lack of a clear explanation for the L-R phenotype of mice with the recessive inv (inversion of embryonic turning) mutation (Yokoyama et al., 1993). The inv mutation was detected in transgenic mouse lines and results in a phenotype characterized by situs inversus and cyst formation in the kidneys. In inv/inv embryos, Nodal and Pitx2 are expressed in the right LPM, in a reversal of the pattern seen in wild-type embryos, giving rise to the situs inversus (Collignon et al., 1996; Lowe et al., 1996; Ryan et al., 1998). The role of inv is evolutionarily conserved, with the corresponding human gene (INVS) having been shown to be responsible for infantile nephronophthisis (NPHP2) (Otto et al., 2003). The inversin (Inv) protein contains 15 tandem repeats of the ankyrin motif, two destruction boxes and two IQ motifs (Mochizuki et al., 1998; Morgan et al., 1998; Yasuhiko et al., 2001; Morgan et al., 2002). Although Inv has been shown to interact with various proteins, including calmodulin, nephrocystin (Nphp1 - Mouse Genome Informatics), Apc2, dishevelled, catenins and N-cadherin (cadherin 2 - Mouse Genome Informatics), its mechanism of action in L-R axis formation remains unknown (Yasuhiko et al., 2001; Morgan et al., 2002; Nurnberger et al., 2002; Otto et al., 2003; Simons et al., 2005). If the direction of nodal flow were rightward in the inv/inv embryo, the explanation for its L-R reversal phenotype would be simple. However, the direction of nodal flow in such embryos is normal, although the flow is slow and turbulent (Okada et al., 1999). The fact that L-R asymmetry is reversed despite the normal direction of nodal flow in inv/inv mice has been a challenge to the notion that nodal flow plays a key role in L-R determination.

We have now examined the mechanism by which L-R asymmetry of inv/inv mice is reversed. We show that a small disturbance of molecular signaling at the node of inv/inv embryos is responsible for the L-R phenotype in a manner dependent on nodal flow. Moreover, analysis of the regulation of Cerl2 expression suggests the presence of an innate mechanism for amplification and stabilization of L-R asymmetry at the node.

All mutant mice were maintained by backcrossing to the FVB/N strain. Data were obtained from mice of at least the N3 generation after confirmation that the pattern of Nodal expression in inv/inv embryos was identical to that apparent on the original FVB/N background. The number of backcrosses for each genotype studied was N5 to N11 for Cerl2 and inv;Cerl2 mutants, N5 to N10 for Lefty1 and inv;Lefty1 mutants, N4 to N9 for Cryptic (Cfc1 - Mouse Genome Informatics) and inv;Cryptic mutants and N6 to N7 for the inv;iv double mutant. Staining of embryos harboring the Lefty2-lacZ transgene with 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside (X-gal) was performed on the N3 generation. Genotyping of mice and embryos was performed by PCR analysis of genomic DNA isolated from the tail or yolk sac.

Whole-mount in situ hybridization was performed according to standard procedures. For two-color analysis, embryos were subjected to simultaneous hybridization with digoxigenin- and fluorescein-labeled probes. Both probes were successively detected with alkaline phosphatase-conjugated antibodies. The first color was developed with NBT/BCIP (Roche). After inactivation of alkaline phosphatase at 70°C, the second antibody was applied and the associated color was developed with INT/BCIP (Roche); this brownish red color can be eliminated by exposure to methanol. Mutant embryos were always processed with control embryos, and the color was developed with the same stop time for comparison. Photographs of the stained embryos were taken with a digital camera (Olympus D12 or Nikon DS-2Mv) attached to a dissecting microscope (Leica). For photographs of the node, the embryo was carefully oriented in 80% glycerol solution so that the crown cells on the left and right sides were positioned in the same horizontal plane.

Cerl2 expression detected by whole-mount in situ hybridization was subjected to quantitative analysis. Images were processed with the use of Photoshop (Adobe) software. The stained areas in the node were selected with the ‘Magic Wand’ tool and snapshots were taken with the use of ‘Grab’ on Mac OSX. The images were then converted to grayscale and inverted. Statistical information for the selected areas was displayed in ‘Histogram Palette’, and ‘Mean’ and ‘Pixels’ were recorded. ‘Mean’ represents the average intensity value. The ‘Mean’ of the background color in the center of the node was subtracted from that of Cerl2 staining in each sample. The product of ‘Mean’ and ‘Pixels’ was regarded as a measure of the relative level of expression. The value of R/L (when R>L) or L/R (when L>R) for staining in each embryo was plotted on the y-axis of a graph, with y=1 corresponding to L=R.

Histograms of two-color staining for Cerl2 and Nodal expression were also obtained with the use of Photoshop. In brief, the stained areas in the node were selected with the ‘Rectangular Marquee’ tool. Composite histograms of the RGB color channels displayed in ‘Histogram Palette’ were recorded with the use of ‘Grab’ on Mac OSX.

Embryos were collected at the late-bud to headfold stage, transferred to Hepes-buffered Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum, and cultured until the four-somite stage in 50 ml tubes containing 25% DMEM and 75% rat serum in the presence of 40 μM SB431542 (Sigma) or vehicle alone (0.15% dimethyl sulfoxide). The tubes were rotated in a humidified incubator containing 5% CO₂ and 95% air.

Nodal is expressed in the right LPM, rather than the left LPM, in inv/inv embryos (Collignon et al., 1996; Lowe et al., 1996). Given that Nodal expression in the LPM is induced by Nodal from the node (Brennan et al., 2002; Saijoh et al., 2003; Oki et al., 2007), we first examined the expression of Nodal and Cerl2, which encodes a Nodal inhibitor, in the node. In wild-type embryos, expression of Nodal and Cerl2 becomes asymmetric from the two-somite stage, with that of Nodal being stronger on the left side (Fig. 1A,B; see Fig. 3A,C,E) and that of Cerl2 stronger on the right (Fig. 1B) (Collignon et al., 1996; Lowe et al., 1996; Marques et al., 2004). In inv/inv embryos, however, Nodal expression was symmetric (22 of 25 embryos at the two- to seven-somite stages) (Fig. 1B; see Fig. 3B,D,F). Furthermore, Cerl2 expression was found to be equal on both sides until the two-somite stage but was reversed from the three-somite stage, being stronger on the left side (Fig. 1B; see Fig. 6E). These results suggested that, in inv/inv embryos, active Nodal is initially distributed symmetrically in the node, and that the crown cells on the right side subsequently produce more active Nodal than those on the left as a result of the reversed pattern of Cerl2 expression.

To test whether the reversal of Cerl2 expression is responsible for the right-sided expression of Nodal in the LPM of inv/inv embryos, we analyzed inv and Cerl2 double-mutant mice. If the balance between Nodal and Cerl2 at the node is a key determinant of the sidedness of Nodal expression in the LPM, the expression pattern of Nodal in the LPM of inv/inv;Cerl2^-/- embryos would be expected to be subject to the status of Nodal expression in the node. During the course of this study, we noticed that the phenotype of inv/inv appears to depend on the mouse background. For example, many inv/inv embryos (5/7) on the FVB/129 hybrid background expressed Pitx2, a direct target gene of Nodal signaling, in both left and right LPM (Fig. 2A). We therefore performed all our analyses on animals with the FVB/N background (the original background of inv). As shown previously (Marques et al., 2004), Cerl2^-/- embryos on the FVB/N background expressed Nodal in the left or both left and right LPM (Fig. 2A,B). By contrast, Nodal expression was completely randomized in inv/inv;Cerl2^-/- embryos (Fig. 2A). Consistent with this expression pattern, Pitx2 expression was also found to be randomized in inv/inv;Cerl2^-/- embryos (Fig. 2A,C). This randomization correlates with the symmetric expression of Nodal in the node of inv/inv embryos (see Fig. 7C for model). These results suggested that the reversal of L-R asymmetry in inv/inv embryos is probably attributable to the combination of the reversed expression pattern of Cerl2 and the symmetric expression of Nodal in the node.

Lefty1 and Lefty2 function as feedback inhibitors of Nodal (Meno et al., 1999; Meno et al., 2001) and contribute to initial L-R determination characterized by the unilateral expression of Nodal in the left LPM. We therefore examined the expression of both Lefty1 and Lefty2 around the node of inv/inv embryos. In wild-type embryos, Lefty1 expression in the floorplate extended anteriorly from the region abutting the node at around the three-somite stage, whereas its expression in the node was detected only after the appearance of Nodal expression in the left LPM (n=29 embryos at the one- to four-somite stages) (Fig. 1A; Fig. 3A,C). In inv/inv embryos, however, ectopic expression of Lefty1 was observed on the posterior side of the node, and it spread into the node before the onset of Nodal expression in the LPM (n=14 embryos at the one- to four-somite stages) (Fig. 3B,D). We also confirmed the ectopic expression of Lefty1 in inv/inv embryos on the B6/129 background (data not shown). Similar results were obtained for Lefty2. Whereas Lefty2 was expressed at a low level in the floorplate of wild-type embryos (n=7 at the two- to four-somite stages), its expression was markedly increased on the posterior side of the node of inv/inv embryos (n=10 at the one- to three-somite stages) (Fig. 3E,F). We next examined which cell types in the node region ectopically express Lefty genes in inv/inv embryos. Frontal sections perpendicular to the node region showed that Lefty1 was expressed in cells at the midline of the dorsal layer of the node, in the crown cells and endoderm cells at the posterior of the node, and in the mesoderm and ectoderm cells adjacent to the posterior of the node (Fig. 3G,H). Expression of Lefty2 was detected in the same cell types as Lefty1, although it was restricted to the posterior side of the node (Fig. 3I).

Lefty1/2 expression in the midline is induced by the Nodal signal (Bamford et al., 2000; Saijoh et al., 2000; Meno et al., 2001; Yamamoto et al., 2003). The ectopic expression of Lefty genes in inv/inv embryos therefore suggested that the level of the Nodal signal was increased in the corresponding regions. To confirm this suggestion, we performed two sets of experiments. First, we introduced a Lefty2-lacZ transgene (line E38) into inv mutant mice. This transgene detects the Nodal signal in the left LPM and floorplate (Saijoh et al., 1999). At the three- to four-somite stages, staining with X-gal revealed a low level of lacZ expression in the node of wild-type or inv/+ embryos (n=5) (Fig. 4A; data not shown). However, in inv/inv embryos at the same stages, the staining was increased, especially in the posterior region of the node (n=3) (Fig. 4B), indicative of an increase in the level of the Nodal signal. Second, we examined the effect of SB431542, a specific inhibitor of Alk4 (Acvr1b - Mouse Genome Informatics) (a type I receptor for Nodal), on Lefty1/2 expression in the node of inv/inv embryos in culture. The increased expression of Lefty1/2 in inv/inv embryos was also apparent in whole-embryo culture (Fig. 4C). Addition of SB431542 to the culture medium at various concentrations revealed that Lefty expression was substantially inhibited at 10 μM (4/4 embryos) whereas it was lost (3/5 embryos) or apparent in only a few cells (2/5 embryos) at 20 μM (data not shown). At 40 μM, SB431542 completely eliminated Lefty expression in inv/inv embryos (Fig. 4D). These results thus indicated that ectopic Lefty expression is dependent on the Nodal signal, which may be increased in the node of inv/inv embryos.

Theoretically, the increase in the level of the Nodal signal in inv/inv embryos might be attributable to upregulation of components of the Nodal signaling pathway or to downregulation of inhibitors of Nodal signaling. We first examined whether the expression of Cryptic or Gdf1, both of which encode components of the Nodal signaling pathway, might be increased. Both genes were found to be normally expressed in the node of inv/inv embryos (data not shown). We next focused on the early expression of Cerl2. At around the onset of somitogenesis in wild-type embryos, Cerl2 is expressed in a horseshoe pattern, with the expression level being lowest in the posterior domain (Fig. 1B; Fig. 4E) (Marques et al., 2004). In inv/inv embryos, however, Cerl2 expression in the posterior domain was slightly weaker than that in wild-type embryos until the two-somite stage (Fig. 1B; Fig. 4F). Double staining for Nodal and Cerl2 expression indicated that the expression of Nodal was markedly greater than that of Cerl2 in the posterior region of the node of inv/inv embryos but not in that of wild-type embryos (n=15, wild type or inv/+; n=11, inv/inv) (Fig. 4G,H). Composite histogram analysis of color components also showed that the luminance value of staining for Nodal expression was markedly higher in the posterior domain than in the lateral domain of the node in inv/inv embryos (Fig. 4H1,H2), whereas such was not the case for wild-type embryos (Fig. 4G1,G2). These results suggested that the local downregulation of Cerl2 expression might be responsible for increased Nodal activity and thereby for the ectopic expression of Lefty genes in inv/inv embryos. This notion was also supported by the pronounced expression of Lefty1 that was observed in and on the posterior side of the node in 10 of 11 Cerl2^-/- embryos compared with that in wild-type embryos (Fig. 4I,J). Together, these results suggested that a local decrease in the level of Cerl2 expression leads to the ectopic and increased expression of Lefty genes in inv/inv embryos.

To determine whether the ectopic expression of Lefty genes contributes to development of the L-R phenotype of inv/inv mice, we analyzed inv and Lefty1 double-mutant embryos. In inv/inv embryos, the onset of Nodal expression in the LPM was delayed and the sidedness of the expression was reversed compared with wild-type embryos (Fig. 5A,B,F,G). We previously showed that Nodal is ectopically expressed in the anterior right LPM of Lefty1^-/- embryos on the B6/129 hybrid background, resulting in left isomerism of visceral organs, with only one-quarter of the Lefty1^-/- embryos surviving to weaning age (Meno et al., 1998). However, we found that the phenotype of Lefty1^-/- mice depends on the genetic background. On the FVB/129 hybrid background, intercrosses of Lefty1^+/- mice gave rise to Lefty1^-/- animals in approximately the expected Mendelian ratio at weaning (20 of 85 animals, or 23.5%). After further backcrossing to the FVB/N background, Nodal expression in Lefty1^-/- embryos became normal (Fig. 5C,H), probably because upregulated expression of Lefty2 in the floorplate compensated for the loss of Lefty1 (Meno et al., 1998). Remarkably, most (14 of 17) inv/inv;Lefty1^-/- embryos showed bilateral Nodal expression in the LPM, despite the upregulation of Lefty2 expression (Fig. 5D,E,I). The expression on both sides was apparent in the distal portion at early somite stages and extended from the anterior to the posterior LPM, suggesting that it was bilaterally induced by Nodal from the node (Fig. 5D). Furthermore, the onset of Nodal expression in LPM of the double mutant was as early as that apparent in wild-type embryos (Fig. 5F,I), indicating that the delay in Nodal expression in inv/inv embryos is attributable to the ectopic expression of Lefty1 in the node. Given that the timing of Nodal expression in LPM was advanced in the double mutant compared with that in inv/inv embryos, it is likely that Nodal was bilaterally expressed in the LPM as a result of the symmetric expression of Cerl2 in the node by the two-somite stage (Fig. 1B; see Fig. 6E and Fig. 7A,B).

The observation that Lefty genes are ectopically expressed in inv/inv embryos prompted us to explore the possibility that Cerl2 expression is regulated by the Nodal signal. We cultured inv mutant embryos in the absence or presence of SB431542 and then examined Cerl2 expression. SB431542 had no effect on the expression of Cerl2 in wild-type or inv/+ embryos (Fig. 6A,C; data not shown). Unexpectedly, Cerl2 expression was decreased and symmetric in inv/inv embryos (4/4) cultured without the drug (Fig. 6B). Consistent with this symmetry of Cerl2 expression, Pitx2 was bilaterally expressed in the left and right LPM of inv/inv embryos (7/8, or 88%), whereas it was normally expressed in the left LPM of wild-type (5/5) or inv/+ (10 of 10) embryos (see Fig. S1 in the supplementary material). Despite this phenotypic change in inv/inv embryos, the addition of SB431542 restored Cerl2 expression in these embryos (5/7, or 71%) to a level similar to that in wild-type or inv/+ embryos (Fig. 6D), suggesting that Cerl2 expression is inhibited by the Nodal signal.

To ascertain whether Cerl2 expression is regulated by the Nodal signal in utero, we examined Cerl2 expression in inv/inv;Cryptic^-/- embryos. Cryptic functions as a cofactor in Nodal signaling, and Cryptic^-/- embryos lack Lefty expression as a result of loss of the Nodal signal (Fig. 6F,G) (Yan et al., 1999). To assess the difference in Cerl2 expression between the left and right sides, we performed a densitometric evaluation of photographs of embryos stained for Cerl2 transcripts. The pattern of Cerl2 expression in inv/inv;Lefty1^-/- embryos was the same as that in inv/inv embryos (Fig. 6E), possibly because Lefty2 expression was upregulated in the double mutant (Fig. 5E). Cerl2 expression in Cryptic^-/- embryos (8/8) was found to be normal (Fig. 6H). Importantly, Cerl2 expression in inv/inv;Cryptic^-/- embryos became almost symmetric (Fig. 6E,I), a pattern different from that apparent in wild-type or inv/inv embryos (see Fig. 7D for model), indicating that Cerl2 expression is regulated by the Nodal signal. These results thus suggested that the Nodal signal is probably inhibited in the crown cells on the left side in inv/inv embryos, thus resulting in the reversed asymmetry of Cerl2 expression.

Finally, we addressed whether leftward nodal flow contributes to the L-R reversal phenotype of inv/inv embryos by generating the inv/inv;iv/iv double mutant. The cilia on the ventral side of the node are immotile in iv/iv embryos, resulting in the inability of these embryos to generate nodal flow (Okada et al., 1999). The expression of Lefty genes was increased in inv/inv;iv/iv embryos (Fig. 6K), compared with that in inv/+;iv/iv embryos (Fig. 6J), as it was in inv/inv embryos. Notably, Pitx2 expression was randomized in inv/inv;iv/iv embryos (left-sided, right-sided, or bilateral in 2/12, 3/12 and 7/12 embryos, respectively) (Fig. 6L) as well as in iv/iv embryos (2/12, 4/12 and 6/12, respectively; data not shown) (Yoshioka et al., 1998), indicating that the inv phenotype is dependent on nodal flow.

On the basis of our results, we propose a mechanism for the generation of asymmetry in the node (Fig. 7A). In this model, nodal flow triggers the initial asymmetric expression of Nodal and Cerl2 in the crown cells of the node of wild-type embryos in a manner dependent on the Inv protein. Nodal in the node not only induces expression of its own gene (Norris et al., 2002) but also downregulates that of the gene for its inhibitor, Cerl2. Such a relation would result in amplification of an initial small difference and stabilize the asymmetric expression of Nodal and Cerl2 in the node, ensuring Nodal expression in the left LPM (Nakamura et al., 2006; Oki et al., 2007). In inv/inv embryos, the symmetric expression of Nodal and a local downregulation of Cerl2 expression allow the Nodal signal to induce Lefty1 expression in the node. The asymmetry of Cerl2 expression is then reversed, because the Nodal signal is probably inhibited on the left side in a manner dependent on nodal flow. Ectopic Lefty in the ventral layer of the node is a candidate mediator of the asymmetric inhibition of Nodal signaling, based on the assumption that it is transported by leftward nodal flow and is able to prevent Nodal from inhibiting Cerl2 expression of the left side of the crown cells (Fig. 7B). More active Nodal is therefore produced on the right side, which is sensed by the right LPM and results in its expression of Nodal before that in the left LPM.

Establishment of the L-R axis is mediated by sequential events that amplify a small difference into a robust one. The initial L-R cue generated by nodal flow results in asymmetric gene expression around the node and subsequent transfer of information to the left LPM. Nodal produced around the node is required for the induction of Nodal expression in the left LPM (Brennan et al., 2002; Saijoh et al., 2003). In addition, it has been suggested that Nodal diffuses from the node to the LPM and directly induces its own expression there (Oki et al., 2007). We recently revealed that Nodal-Lefty signaling contributes to the L-R decision on the side of the LPM that expresses Nodal (Nakamura et al., 2006). Such a system, which we designated ‘SELI’ (self-enhancement and lateral inhibition), explains how a slight difference is amplified into the robust and unilateral expression of Nodal in the LPM. In the present study, through an analysis of inv mutant mice, we have shown that the mechanism responsible for generation of the L-R difference in the amount of active Nodal around the node is key to an understanding of the initial events in L-R axis formation.

The amount and activity of Nodal produced by the crown cells of the node are thought to be regulated by several mechanisms. First, Nodal requires GDF1 as a binding partner to exert its full activity (Tanaka et al., 2007). Gdf1 is expressed in the crown cells of the node and the LPM, and Gdf1 mutant mice lack Nodal expression in the left LPM (Rankin et al., 2000). Second, Cerl2 produced by the crown cells of the node inhibits Nodal, probably by binding to it (Marques et al., 2004). Finally, the expression of Nodal and Cerl2 is differentially regulated on the left and right sides, with Nodal expression being stronger on the left side and Cerl2 expression stronger on the right (Collignon et al., 1996; Lowe et al., 1996; Pearce et al., 1999; Marques et al., 2004). Although a method to detect the activity of Nodal secreted from the node remains to be developed, such a pattern of regulation probably generates more active Nodal on the left side than on the right. This notion is supported by the observation that Lefty1 expression, which is induced by Nodal (Yan et al., 1999; Yamamoto et al., 2003), is left-sided in the dorsal layer of the node of wild-type embryos, whereas it is bilaterally upregulated in the node of Cerl2^-/- embryos.

The inhibition of Nodal by Cerl2 in the node region appears to be fundamental to L-R determination. About half of Cerl2^-/- embryos express Nodal and Pitx2 bilaterally in the LPM, suggesting that increased Nodal activity on the right side of the node induces Nodal in the right LPM (Marques et al., 2004). However, the remaining half of Cerl2^-/- embryos normally express Nodal and Pitx2 in the left LPM. Given that the asymmetry of Nodal expression in the node is maintained in Cerl2^-/- embryos (Marques et al., 2004), the subtle difference in the amount of Nodal is likely to be amplified by the SELI system to result in the expression of Nodal in the left LPM (Nakamura et al., 2006). The asymmetry of Cerl2 expression might be sufficient to determine the sidedness of Nodal expression in the LPM, even if Nodal is symmetrically expressed in the node. In inv/inv and inv/inv;Lefty1^-/- embryos, which express Nodal symmetrically in the node, Nodal expression in the LPM corresponded to the asymmetry of Cerl2 expression immediately before the onset of Nodal expression in the LPM. Furthermore, the absence of Cerl2 in inv/inv embryos gave rise to the randomized expression of Nodal and Pitx2, probably reflecting the symmetric expression of Nodal in the node. Mutant mice with a modified cis-element of Nodal as well as transgenic mice that express Nodal symmetrically in the node were found to manifest normal Nodal expression in the left LPM (Brennan et al., 2002; Saijoh et al., 2003). Although Cerl2 expression was not examined in these mice, their phenotype might be explained if the asymmetry of Cerl2 expression was normal.

Then, how does the balance between Nodal and Cerl2 expression in the node determine the side of the LPM that expresses Nodal? We have previously shown that Myc epitope-tagged Nodal is detected in the area immediately external to the apical and basolateral membranes of crown cells of the node expressing this protein, suggesting that Nodal is secreted from both membrane regions (Oki et al., 2007). We also showed that Nodal secreted from the basolateral membrane of the crown cells may diffuse along the internal route to the LPM. The LPM on the same side as the crown cells that secrete the larger amount of active Nodal may therefore first experience a level of the Nodal signal sufficient to induce Nodal expression. Once Nodal is expressed in the LPM, the induction of Lefty gene expression results in inhibition of Nodal expression on the contralateral side by the SELI system (Nakamura et al., 2006). Visualization of the diffusing Nodal and Lefty proteins in the embryo will be required for a full understanding of these processes, but as yet we have not been able to achieve such visualization.

If the generation of a subtle but consistent asymmetry in Nodal activity around the node is key to L-R patterning, then how are the expression patterns of Nodal and Cerl2 established? An enhancer located in the upstream region of Nodal drives bilateral expression in the node, which is activated by Notch and its ligand delta-like 1 (Dll1) (Krebs et al., 2003; Raya et al., 2003). It has been suggested that the left-biased expression of Nodal in the node is generated by a positive feedback loop mediated by Nodal and the intronic enhancer ASE (Brennan et al., 2002; Norris et al., 2002). This suggestion seems reasonable, because if there is less Cerl2 on the left side of the node then the expression of Nodal on the left is probably augmented by the Nodal positive loop. We have also now revealed that Cerl2 expression in the node of inv/inv embryos is suppressed by the Nodal signal. First, Cerl2 expression in inv/inv embryos was increased by exposure to SB431542, a specific inhibitor of Alk4. Second, the asymmetry of Cerl2 expression was lost in inv/inv;Cryptic^-/- embryos. Given that Cerl2 expression in inv/inv embryos is stronger on the left side, the Nodal signal would be expected to be lower in the crown cells on the left side in inv/inv embryos. The inhibition of Cerl2 expression by the Nodal signal would be expected to be a reliable mechanism for generating asymmetry in the net amount of active Nodal in wild-type embryos. After the slight asymmetry in the expression of Nodal and Cerl2 is established in the node, the difference is further amplified because the greater the extent of Nodal signaling on the left side, the more the expression of Cerl2 is suppressed. This pattern of regulation probably stabilizes the L-R difference in the node sufficiently to allow induction of Nodal expression in the left LPM by the SELI system.

With regard to Cerl2, cis-regulatory elements for expression of the gene in the node have not been identified to date. In the present study, we found that the loss of Inv resulted in downregulation of Cerl2 expression on the posterior side of the node, symmetry of Cerl2 expression at the two-somite stage, and reversal of the normal asymmetry of Cerl2 expression at later stages. The local diminution of Cerl2 expression at the posterior side of the node suggests that the Inv protein may contribute to a signaling pathway responsible for the upregulation of Cerl2 expression. Here, we propose that Cerl2 expression in the node of wild-type embryos is regulated by an Inv-dependent mechanism and the Nodal signal. The crown cells of the node on the left side may sense nodal flow by an unknown mechanism dependent on the Inv protein and downregulate Cerl2 expression in response. The Nodal signal on the left side further inhibits Cerl2 expression in a manner that does not require Inv. In inv/inv embryos, the Inv-dependent pathway that generates asymmetric Cerl2 expression is absent, but the inhibitory effect of the Nodal signal on Cerl2 expression is still operative. The Nodal signal is probably lower on the left side in inv/inv embryos, which eventually results in reversal of the asymmetric expression of Cerl2. This model explains why the regulation of Cerl2 expression by the Nodal signal appeared only in inv/inv embryos, as revealed by inhibition of the Nodal signal either by the addition of SB431542 or by the loss of Cryptic. Disturbance of the Nodal signal alone may not affect Cerl2 expression, because the Inv-dependent pathway is sufficient to establish the asymmetry of Cerl2 expression.

Our results indicate that the L-R reversal phenotype of inv/inv embryos is attributable to complex events that include a delay in Nodal expression in the LPM and the reversal of asymmetric Cerl2 expression. Ectopically expressed Lefty1 in the node was shown to delay the onset of Nodal expression in the LPM, probably by diffusing bilaterally and inhibiting the Nodal signal in the LPM. The delay in Nodal expression in the LPM seems to be a prerequisite for the L-R reversal phenotype of inv/inv, given that Nodal was expressed bilaterally in the LPM of inv/inv;Lefty1^-/- embryos, in which the timing of Nodal expression was normal. Ectopic expression of Lefty genes was detected in the crown cells on the posterior side of the node in addition to the floorplate and the dorsal layer of the node. Given that the crown cells face the yolk sac cavity, apically secreted Lefty may spread and be conveyed toward the left side by nodal flow, resulting in inhibition of the Nodal signal in the crown cells on the left side. This scenario would account for the timecourse of Cerl2 expression in inv/inv embryos. The prevention of asymmetric regulation of Cerl2 expression by the loss of Inv is followed by the inhibition of Nodal by Lefty on the left side, eventually resulting in the reversal of asymmetric Cerl2 expression. Determination of whether Lefty genes are actually required for reversal of the asymmetry of Cerl2 expression in inv/inv embryos will require the generation and characterization of inv mutants lacking both Lefty1 and Lefty2 expression (such as inv/inv;Lefty1^-/-;Lefty2^ΔASE/ΔASE embryos).

In conclusion, we have addressed the longstanding mystery of the inv phenotype. Our results suggest that the L-R reversal in inv/inv embryos results from an aberration of the mechanism responsible for generating molecular asymmetry in the node: Nodal expression thus becomes symmetric, and the asymmetry of Cerl2 expression is reversed at later stages, resulting in the production of more active Nodal on the right side than on the left. Furthermore, ectopic Lefty, produced as a result of a local downregulation of Cerl2 expression, suppresses Nodal expression in the LPM. The LPM on the right may therefore receive the Nodal signal at a level sufficient to override suppression by Lefty earlier than that on the left, resulting in Nodal expression in the right LPM.

We thank M. Shen for Cryptic mutant mice and S. Ohishi for genotyping and maintenance of mouse colonies. This work was supported by grants from the Ministry of Education, Culture, Sports, Science, and Technology of Japan, the CREST, the Kyushu University Interdisciplinary Programs in Education and Projects in Research Development, and the Naito Foundation.