This article is a US Government work and, as such, is in the public domain in the United States of America.
Nodal signaling is a critical regulator of multiple aspects of early vertebrate development including asymmetry along the left/right (LR) axis. To study Nodal function occurring specifically in the postgastrulation embryo, we have used Cre/loxP based conditional mutagenesis. A floxed allele of Nodal was generated and shown to have wild-type function. This allele was then used in conjunction with the T-Cre line, which expresses Cre recombinase broadly in the mesodermal and definitive endodermal lineages posterior to the cranial region. T-Cre activity leads to complete deletion of Nodal before its normal transient expression in the early somite stage lateral plate mesoderm, thereby causing severe LR developmental defects. No other abnormalities were found, suggesting that Nodal signaling has no additional essential functions in developmental patterning within the extensive mesodermal and endodermal domains marked by T-Cre activity. Developmental Dynamics 237:3591–3601, 2008. Published 2008 Wiley-Liss, Inc.
Nodal, a transforming growth factor (TGF) -β like secreted signaling molecule, plays multiple essential roles in axial patterning and germ layer specification during early vertebrate development (Ang and Constam,2004; Shen,2007). These various functions are reflected in the complex and dynamic expression pattern found for the Nodal gene. During early mouse development, expression occurs at several sites and at several developmental stages indicative of Nodal signaling being used and reused to progressively pattern the embryo. One of the most striking domains is the strictly left sided expression in the lateral plate mesoderm (LPM), occurring in a narrow developmental window at early somite stages. The ultimate outcome of Nodal signaling from the LPM, which is intimately tied into the action of its antagonist Lefty (reviewed by Tabin,2006), is the characteristic left/right (LR) asymmetric location and/or structure of the visceral organs (reviewed in Levin,2005).
Insight into Nodal function has come from the analysis of various mutant alleles. The null mutant arrests before gastrulation with defects in anterior–posterior patterning and without forming mesoderm or endoderm (Iannaccone et al.,1992; Conlon et al.,1994). More recently conditional mutagenesis using the Cre/loxP approach (Lewandoski,2001; Branda and Dymecki,2004) has been used to analyze Nodal expression specific to the epiblast (Lu and Robertson,2004). The introduction of loxP sites and selectable markers can often result in hypomorphic alleles that are informative even in the absence of Cre-mediated recombination (Meyers et al.,1998). Two alleles with reduced Nodal function and with less severe defects than the null mutant have been described (Lowe et al.,2001; Saijoh et al.,2003). Embryos carrying the Nodalfl1 hypomorphic allele, when compound heterozygous with the null allele, develop past the block at gastrulation seen in Nodal null homozygotes, but then display a variety of developmental abnormalities. Analysis of the least severe class of mutants verified the critical role Nodal signaling plays in patterning the LR body axis in mouse development (Lowe et al.,2001). Embryos homozygous for the Nodalneo allele undergo normal gastrulation but then develop LR defects due to loss of Nodal expression from the LPM (Saijoh et al.,2003). To dissect the various temporal and spatial domains of Nodal function further, we have generated a new conditional mutant allele, Nodalfl2. Here, we use this allele in conjunction with the recently described T-Cre transgenic line, which expresses Cre recombinase broadly in the mesodermal and endodermal lineages commencing soon after gastrulation initiates (Perantoni et al.,2005). We find that extensive deletion of Nodal affects only LR development, resulting in a suite of defects characterized by randomization and right isomerisms of the thoracic and abdominal viscera.
Development and Characterization of the Nodalfl2 Line
We previously reported that a floxed Nodal allele, Nodalfl1, originally developed for conditional mutagenesis, was hypomorphic. The overall reduced level of function of this allele in the absence of Cre-mediated recombination provided insight into several aspects of Nodal function in early mouse development (Lowe et al.,2001). To generate loss of Nodal function in specific tissues and developmental stages in an otherwise normal embryo, we created a new conditional allele, Nodalfl2. This new allele differs from wild-type only by the presence of loxP sites in the first intron and 3′ untranslated region (UTR; Fig. 1). Homozygous Nodalfl2 animals were derived at the expected frequency and were found to be normal and fertile. However, to determine whether Nodalfl2 is truly a wild-type allele, we examined embryos compound heterozygous for Nodalfl2 and the null allele, NodalΔ1 (Lowe et al.,2001). Using this approach in our previous study of Nodalfl1, we found numerous defects indicating it was hypomorphic (Lowe et al.,2001). Even in the small percentage of Nodalfl1/Δ1 embryos that appeared completely normal at embryonic day (e) 14.5, we found transpositions of the aorta and pulmonary trunk and right isomerisms of the lung. Here, a similar analysis of Nodalfl2/Δ1 embryos revealed none with transpositions of the great vessels or right lung isomerisms, or with any other overt defect (Table 1). Examination of embryos at earlier stages, beginning at e7.5, also revealed no abnormalities with only three embryos found resorbed (Table 2). While the latter could not be genotyped, the frequency is not significantly different from that seen in normal matings. In total, we examined almost 200 embryos derived from crosses of Nodalfl2/fl2 homozygotes and Nodal+/Δ1 heterozygotes, of which 52 were genotyped. Of these, 50% were Nodalfl2/Δ1 (the expected Mendelian frequency). Together, these data provide strong evidence that the Nodalfl2 allele is fully functional.
Table 1. LR Phenotypic Analysis of Nodalfl2/Δ1 vs. Nodalfl1/Δ1 Embryos
Table 2. Phenotypic Analysis of Embryos from Nodal+/Δ1 × Nodalfl2/fl2 Matings from e7.5–e16.5
Number Nodalfl2/Δ1 (total number genotyped)
Characterization of Early T-Cre Activity in Relation to Nodal Expression
The recently described T-Cre transgenic line, in which Cre recombinase is driven by regulatory elements of the T/Brachyury gene, has been shown to be activated in mesoderm at the primitive streak stage (Perantoni et al.,2005). This suggested that it would be an ideal genetic tool to examine postgastrulation Nodal function. To confirm that there is no earlier T-Cre activity, which might impact Nodal expression required at or before mesoderm formation, we evaluated the exact time of onset using the conditional reporter lines R26R (Soriano,1999) and R26R-EYFP (Srinivas et al.,2001). To complement this analysis, we used the Nodalfl1 line as a reporter. In embryos from this line, which carries an IRES βgeo cassette inserted just after the stop codon, β-galactosidase expression mirrors that of Nodal mRNA (Lowe et al.,2001). Cre-mediated deletion of Nodalfl1 eliminates βgeo. Thus, domains of Cre activity that overlap Nodal expression can be identified as regions undergoing loss of X-gal staining (Kumar et al.,2007).
The earliest activity detected in embryos derived from T-Cre × R26R matings was indeed at e6.5, in nascent mesoderm migrating extra-embryonically (Fig. 2A). This was confirmed by confocal analysis of EYFP fluorescence in embryos derived from T-Cre × R26R-EYFP matings (Fig. 2B). Evidence of recombination was then seen in the lateral mesodermal wings migrating out from the primitive streak (Fig. 2C,D). By the early headfold stage, T-Cre activity was apparent along the entire primitive streak (Fig. 2E–G). At primitive streak stages, Nodal is expressed in posterior embryonic ectoderm migrating into the primitive streak but is rapidly down-regulated as cells differentiate into mesoderm. X-Gal staining of late streak stage embryos derived from T-Cre × Nodalfl1 matings did not reveal any difference between T-Cre–negative and T-Cre–positive embryos (Fig. 3A,B), indicating there is minimal overlap in expression either spatially or temporally between Nodal and T-Cre at these stages.
As the primitive streak elongates, Nodal expression localizes to the very anterior of the streak and by headfold stages is found further anterior, in cells surrounding the indentation at the distal tip of the embryo. This region has been called the node, but a recent analysis (Blum et al.,2007) suggests it is more properly considered the posterior notochord (PNC) with the term node reserved for the dynamic population of cells at the anterior end of the primitive streak containing the midgastrula organizer (MGO; Robb and Tam,2004). Beginning at approximately the two- to three-somite stage (2–3 ss), Nodal expression around the PNC becomes stronger on the left side. At approximately the same stage, Nodal expression commences in the left LPM, persisting until the 8–9 ss (Lowe et al.,1996). Over these same stages, T-Cre activity detected using the R26R reporter was found throughout the lateral and somitic mesoderm (Fig. 2H–K). X-gal staining also began to be evident in the PNC and notochord by the 3–4 ss (Fig. 2K). Given that Cre is driven by primitive streak specific regulatory elements, this result probably reflects the relatively late onset of T-Cre activity in the node/anterior streak/MGO, from which the PNC and notochord are derived, rather than de novo expression. Analysis of T-Cre activity over these same stages using the Nodalfl1 line as a reporter revealed a progressive loss of X-gal–positive cells around the PNC from late headfold to early somite stages (Fig. 3D,F,H) and complete absence of X-gal staining in the LPM (Fig. 3H). Together these results show that T-Cre should completely delete the Nodalfl2 allele within the mesoderm before its normal expression in the LPM, but only partially delete at the PNC well after expression in this domain commences. We also confirmed the broad T-Cre activity in mesoderm and posterior gut endoderm reported for later stages (Perantoni et al.,2005). Although activity does not extend fully into the most anterior cranial region (Fig. 2L), T-Cre should lead to extensive deletion of Nodal throughout the majority of the embryo.
Conditional Mutagenesis of Nodal Using T-Cre
To determine the phenotypic and molecular consequences of T-Cre–mediated deletion of Nodal, we first produced animals heterozygous for both T-Cre and the null allele, NodalΔ1. T-Cre;Nodal+/Δ1 animals were then bred with homozygous Nodalfl2/fl2 animals, and embryos were collected at various stages for analysis and genotyping. We expected approximately 25% to be T-Cre;Nodalfl2/Δ1, the appropriate genotype for conditional mutagenesis resulting from Cre deletion of Nodalfl2. To confirm that T-Cre indeed deletes the Nodalfl2 allele and determine the extent of recombination, we carried out polymerase chain reaction (PCR) reactions that specifically amplify across the deletion site (Fig. 1D). We also assessed any reduction in the level of the normal Nodal transcript by performing whole-mount in situ hybridization (WMISH) with a probe specific for Nodal exon 2, which is removed in the deleted allele, NodalΔ2 (Fig. 1C,D). PCR genotyping of e7.5 embryos showed the expected NodalΔ2 product in T-Cre–positive embryos indicating successful recombination (not shown). As expected, WMISH with the Nodal exon 2 probe showed no significant difference between Nodalfl2 control and T-Cre;Nodalfl2/Δ1 embryos at the late streak stage (Fig. 3I,J). However, by the late headfold stage, there was reduced expression around the PNC (Fig. 3K,L), providing the first indication of recombination in Nodal-expressing cells. By the 1–2 ss, the reduction in Nodal expression was even more apparent (Fig. 3M,N). In addition, although Nodal expression in the left LPM begins by the 3 ss in control embryos (Fig. 3O), it was undetectable in T-Cre;Nodalfl2/Δ1 embryos at this and all later somite stages (Fig. 3P), indicating complete recombination in this lineage. These results indicate that T-Cre activity occurs early enough in the mesodermal lineage to cause complete deletion of Nodal from the LPM before its normal onset of expression. In contrast, a significant level of Nodal expression occurs around the PNC before its eventual elimination.
T-Cre–Mediated Deletion of Nodal Leads to LR Phenotypes Only
To determine the morphological consequences of T-Cre–mediated deletion of Nodal, we examined embryos from T-Cre;Nodal+/Δ1 × Nodalfl2/fl2 matings for any overt defects. We found no abnormalities at e7.5, indicating that T-Cre activity does not interfere with critical Nodal functions in the pre- and early gastrula stage embryo. However, there were morphological abnormalities seen at late e8.5 and e9.5, specifically in the direction of heart looping. Normally, the heart tube loops to the embryo's right side, but 15% (n = 97) of embryos derived from T-Cre;Nodal+/Δ1 × Nodalfl2/fl2 matings had reversed heart looping (Fig. 4A). Genotyping of a representative subset showed that all of these carried T-Cre and the NodalΔ1 allele. The 15% frequency corresponds to approximately half of the 25% of embryos hypothetically expected to have the appropriate genotype for conditional mutagenesis, indicating that, in embryos deleted for Nodal, the direction of heart looping is randomized (half normal, half reversed). At later stages, hearts from embryos genotyped as NodalΔ1 and positive for T-Cre showed ventricular septal defects as revealed by mixing of different colored dyes injected separately into the left or right ventricle (Fig. 4B,C). In addition, 26 of 63 embryos examined at e13.5 and later showed right lung isomerism (Fig. 4D,E) as well as transpositions of the great vessels and small or absent spleens (not shown). Seven of these embryos also had stomachs placed on the right side. All of these phenotypically abnormal embryos, but none of the apparently normal ones, genotyped as NodalΔ1 and positive for T-Cre (Table 3).
Table 3. LR Phenotypic Analysis of Embryos (n = 63) from T-Cre; Nodal+/Δ1 × Nodalfl2/fl2 Matings
Further analysis failed to detect abnormalities in other tissues or developmental processes, including the limbs and tail, or in kidney and urogenital development. Notably, none of the embryos examined in this study displayed midline or anterior/head morphological defects, which arise in the majority of Nodalfl1/Δ1 hypomorphic embryos (Lowe et al.,2001). These negative results indicate that T-Cre–mediated conditional mutagenesis of Nodal solely targets Nodal function in LR development.
Molecular Analysis of T-Cre–Deleted Embryos
The observed abnormalities in LR development no doubt arise due to the loss of Nodal from the left LPM. To determine the molecular consequences of T-Cre–mediated deletion of Nodal at early somites stages, we carried out WMISH with probes for genes that have been implicated as downstream targets of Nodal signaling, including Lefty-2, Pitx2, and Nodal in the left LPM, Lefty-1 in the left prospective floor plate (PFP) and Nodal around the PNC. As expected given the complete loss of Nodal from the LPM in Nodalfl2/Δ1 embryos positive for T-Cre, the normal expression of Lefty-2 in the LPM (Fig. 5A) was lost (Fig. 5B). In addition, the normal expression of Pitx2 in the LPM and heart (Fig. 5C) was not detected in T-Cre;Nodalfl2/Δ1 embryos (Fig. 5D). Similarly, expression of Lefty-1 in the PFP (Figs. 3O, 5A,E) also was not found (Figs. 3P, 5B,F). Although there is no direct evidence, the most posterior expression of Lefty-1 in the midline of the PNC (brackets in Figs. 3O, 5E) may be induced wholly or in part by Nodal signals coming from nearby PNC crown cells. If so, our results suggest that, by the 2–3 ss, Nodal levels in the PNC of T-Cre–deleted embryos are below the threshold critical for Lefty-1 induction.
To examine expression of the deleted Nodal allele, we used a probe specific for the still remaining exon 1 (Fig. 1D). This probe detects the undeleted allele as well, as seen by the normal WMISH pattern in wild-type control embryos (Fig. 5E). For T-Cre;Nodalfl2/Δ1 embryos an estimation of the amount of continued expression of the deleted allele was made by comparing the hybridization signal detected by the exon 1 probe to that from the exon 2 probe. Using the exon 1 probe, we were unable to detect any expression in LPM of 3–4 ss T-Cre–deleted embryos, but there were near normal levels around the PNC (Fig. 5F), compared with the weak and patchy hybridization signal detected using the exon 2 specific probe (Fig. 3N,P). This result indicates that the deleted allele continues to be transcribed around the PNC of T-Cre;Nodalfl2/Δ1-positive embryos. Importantly, by the 3–4 ss, this expression was stronger on the left side, thereby displaying the same asymmetric pattern seen in normal embryos (Fig. 5E,F).
We also examined the expression of Cerl-2/Dante, which is asymmetrically expressed around the PNC at early somite stages in a pattern opposite to that of Nodal (Fig. 5G) but is not known to be a target of Nodal signaling. Importantly, there was no loss of expression of Cerl-2/Dante in T-Cre;Nodalfl2/Δ1 embryos (Fig. 5H). In addition, right asymmetric expression was apparent at the same stages as in control embryos. As another indicator of PNC function, we asked if there was any alteration in the expression of two genes expressed within the PNC, Dnahc5 (Fig. 5I) and Lrd (not shown), encoding components of the dynein motor complex involved in ciliary beating and nodal flow (Hirokawa et al.,2006). Both were found to be expressed normally in T-Cre–deleted embryos (Fig. 5J and not shown). In addition, we found normal expression of the midline markers T/Brachyury, Shh, and Foxa2 (not shown). In summary, these data argue that T-Cre–mediated conditional mutagenesis of Nodal results in the complete loss of asymmetric signals from the LPM while having apparently no impact on gene expression patterns at and around the PNC.
As shown previously (Perantoni et al.,2005) and confirmed here, T-Cre activity in the primitive streak commences early enough and broadly enough to result in recombination throughout the mesodermal lineages of the developing embryo. In addition, activity within cells of the gastrula organizer at the anterior of the primitive streak leads to recombination also occurring in the endoderm and midline mesendoderm, although the timing is such that only the more posterior domains of these lineages are affected. Even so, it is clear that T-Cre activity can cause deletion of floxed alleles throughout an extensive portion of the embryo. Thus, utilization of this line in conjunction with our newly described conditional allele of Nodal provided a unique opportunity to assess the consequences of broad deletion of Nodal activity commencing after gastrulation. The fact that we find only LR defects, which presumably derive from loss of Nodal signaling from the e8.5 LPM, suggests that there are no other essential developmental functions for Nodal in mesodermal or endodermal tissues arising at later stages. In fact, there appear to be very few additional domains of Nodal expression beyond e8.5. The only well-documented expression is that found in the first branchial arch and forebrain roof starting at e9.5 (Andersson et al.,2006). We do not see evidence of T-Cre activity in either of these regions of the embryo, so an evaluation of Nodal function in these domains awaits development and/or utilization of appropriate Cre lines. Expression of Nodal receptors has been seen in the developing limb bud (Jornvall et al.,2004) although Nodal has not been detected, perhaps because levels are too low or transient. However, the lack of any limb defects in T-Cre–deleted embryos suggests that, if Nodal signaling is active in the limb bud, it is dispensable.
Two previous studies describing targeted deletion or disruption of the Nodal enhancer that specifically drives expression in the PNC have shown that this results in almost undetectable levels of Nodal expression at the PNC, and is accompanied by absence of expression in the left LPM (Brennan et al.,2002; Saijoh et al.,2003). These findings have led to the conclusion that the principal role for Nodal signaling originating from the crown cells of the PNC is to initiate expression of Nodal in the most proximal part of the LPM, where activity then subsequently amplifies and expands Nodal expression throughout the LPM and induces downstream target genes that ultimately control the LR asymmetric development of the viscera. One question not addressed is whether Nodal signaling at the PNC plays any additional role, in LR development or otherwise. Theoretically, this could be addressed using embryos lacking Nodal in PNC or LPM, but not both, and comparing phenotypes to embryos lacking Nodal in both domains. While it is technically challenging to generate embryos in which Nodal is removed from the PNC but retained in the LPM, T-Cre deletion of Nodal results in embryos that lack LPM expression but do retain, at the late headfold stage and perhaps beyond, a substantial level of expression at the PNC. The level we see is at least as high, if not higher, than that seen in certain allelic combinations reported by Brennan et al. and Saijoh et al., which were sufficient to initiate expression of Nodal in the LPM and thus produced no LR patterning abnormalities (Brennan et al.,2002; Saijoh et al.,2003). As a functional test of the residual levels of Nodal remaining in the PNC, we asked whether these could induce expression of the T-Cre–deleted Nodal allele in the LPM. However, we could not detect any transcription of the residual exon 1 fragment by WMISH, probably due to the absence of the Nodal-positive feedback loop that in normal embryos amplifies the level of Nodal transcripts in the LPM to robust levels. As an alternative readout of Nodal-dependent Nodal expression, we examined the PNC and found that although there is partial T-Cre deletion, Nodal activity at the PNC is sufficiently high to allow continued expression of the Nodal gene. Importantly, the T-Cre–deleted gene still develops the left asymmetric expression pattern seen in wild-type. This asymmetry may be dependent on continued Nodal signaling but is clearly independent of antagonism by Lefty-1, perhaps because of the unaltered activity of Cerl-2/Dante, another Nodal antagonist. In addition, this finding shows that Nodal asymmetry around the PNC does not require Nodal activity feeding back from the left LPM, although it may augment it in normal embryos. Together these results argue that T-Cre–deleted embryos at presomite and early somite stages, when symmetry is first broken, have an apparently functionally normal PNC with intact Nodal signaling. The fact that the LR defects we observed are identical to those found in embryos that lack both PNC and LPM Nodal signaling provides strong support for the idea that the sole contribution of Nodal expression at the PNC is to initiate LPM expression.
In summary, we find that the consequences of broad deletion of Nodal in the developing postgastrulation embryo can be wholly assigned to the loss of Nodal activity in the left LPM at early somite stages, arguing that essential developmental functions in the extensive mesodermal and endodermal domains marked by T-Cre activity are complete by e8.5. The question of whether Nodal signaling is essential for postgastrulation cranial development or has a redundant function in organogenesis is still open, as is whether there is any role in homeostasis in the adult or in disease states.
To generate the Nodalfl2 targeting vector, we began with intermediate constructs from the Nodalfl1 targeting vector containing only the 5′ loxP site (Lowe et al.,2001). Site-directed mutagenesis was used to modify four nucleotides within the 3′ untranslated region (UTR), and thereby generate two new restriction sites (Fig. 1A). These were used to directionally introduce a floxed PGKneo cassette from the ploxpneo vector (Yang et al.,1998). ES cells correctly targeted with this vector were transiently transfected with a Cre-containing expression plasmid and desired recombinants were identified using the PCR primer pair B/C (Fig. 1C). Primer B: 5′-ATGATTGTGGAGGAGTGTGGGTGC-3′; Primer C: 5′-TCTCTGGCTTGGCAGGTCTAAG-3′.
Mouse and Embryo Genotyping
Nodalfl2 animals were genotyped routinely using primer set B/C (Fig. 1C). To genotype embryos for the NodalΔ2 allele, primer set A/C was used (Fig. 1D). Primer A: 5′-AGCACTGGATGCTTGCTCTTCC-3′. Genotyping for the Nodalfl1 and NodalΔ1 alleles was described previously (Lowe et al.,2001). For genotyping T-Cre mice and embryos, we used primers wholly within the Cre sequence as described (Lowe et al.,2000).
Phenotypic and Molecular Analysis
Procedures for X-Gal staining of R26R and Nodalfl1 embryos and confocal microscopic analysis of R26R-EYFP were described previously (Kumar et al.,2007). WMISH was carried out as described (Lowe and Kuehn,2000), with modifications (Kumar et al.,2007). To detect the deleted Nodal transcript, an exon 1 containing EcoRI/BamHI fragment from the mouse Nodal cDNA was subcloned into the EcoRI and BamHI sites of pBluescriptSK. DIG labeled antisense RNA probe was made using T3 polymerase following digestion with EcoRI. The Nodal exon 2 probe specific for the undeleted wild-type transcript has been described (Zhou et al.,1993). Analysis of cardiac and vasculature malformations was done by colored latex injections into right and left ventricles as described previously (Oh and Li,1997).
We thank Linda Lowe for expert technical assistance in generating the Nodalfl2 targeting vector; the NCI CCR Gene Targeting Facility for generation of chimeric mice; Dr. Frank Costantini for R26R-EYFP mice; Dr. Philippe Soriano for R26R reporter mice; Dr. Alan O. Perantoni for advice; the NCI-Frederick Image Analysis Laboratory for help with confocal microscopy; and Drs. Ira Daar and Terry Yamaguchi for comments on the manuscript. This work was supported by the Intramural Research Program of the National Institutes of Health, National Cancer Institute, Center for Cancer Research.