Determination of left/right asymmetric expression of nodal by a left side-specific enhancer with sequence similarity to a lefty-2 enhancer

Results

A node-specific enhancer in the upstream region of nodal

At the early somite stage in mice, nodal is expressed in two domains: the LPM and the node (Zhou et al. 1993; Collignon et al. 1996). Expression of nodal in LPM shows a strict L-R specificity, being confined to the left half (Collignon et al. 1996;Lowe et al. 1996). Expression in the node also shows a subtle L-R asymmetry (Collignon et al. 1996). In this study we focus primarily on transcriptional regulatory elements that are required for expression ofnodal in the node and left LPM at the early somite stage.

We first measured the transcriptional activity of the 5-kb region upstream of the nodal coding sequence (from −5 kb to −50 bp with respect to the translational initiation codon). Construct5′-1, in which the 5-kb upstream fragment was linked to a promoterless lacZ gene (Fig. A), was injected into the pronuclei of fertilized mouse embryos at the one-cell stage. The embryos were allowed to develop in utero until embryonic day 8.2 (E8.2), when they were recovered and the expression of thelacZ transgene was examined by staining with X-gal (5-bromo-4-chloro-3-indoyl-β-d-galactoside). (Other constructs described below were tested similarly unless otherwise mentioned.) The 5′-1 construct failed to give rise to substantial X-gal staining at E8.2 (Fig. B), suggesting that the 5-kb upstream region is not sufficient for nodal expression. The observation that 5′-1 reproducibly yielded low-intensity staining in a few cells of the left LPM (Fig. B) suggested the possible existence of a weak cis-regulatory element that contributes to asymmetric expression.

Next we examined the effect of construct 5′-2, which contained the upstream region of nodal from −11.5 kb to −50 bp (Fig. A). This construct gave rise to X-gal staining in the node (Fig. B), suggesting the presence of a node-specific enhancer (NDE) between −11.5 and −5 kb. To examine the specificity of the NDE, we linked this 6.5-kb region (from −11.5 to −5 kb) to a heterologous promoter, that of the hsp68 gene. The resulting construct (NDEhsp) yielded X-gal staining in the node but also in ectopic sites (Fig. B), indicating that both NDE and the proximal promoter region of nodal are required for strictly node-specific expression.

To locate the NDE more precisely, we first divided the 6.5-kb region between −11.5 and −5.0 kb into two and attached each portion to the 5-kb proximal promoter region. Construct NS1.5, which contained the 1.5-kb 3′ portion of the upstream region, did not show expression in the node, whereas SN5.0, which contained the 5.0-kb 5′ portion of the upstream region, possessed NDE activity (Fig. ). Then we prepared two sets of deletion mutants from SN5.0 and examined their activity (Fig.) Among the 5′ deletion mutants tested,Δ1–33 and Δ1–70retained NDE activity, whereas Δ1–21 did not. Of the 3′ deletion mutants, Δ2–1 andΔ2–77 showed X-gal staining in the node, whereas Δ2–37 did not. These results localized the NDE to a 0.8-kb region between −9.5 and −8.7 kb.

An ASE in intron 1 of nodal

Among five constructs (3′-1 to 3′-5) containing the 5-kb proximal promoter region and various portions ofnodal downstream region, construct 3′-1, which contained a 3.5-kb downstream fragment including intron 1, gave rise to marked X-gal staining in the left LPM (Fig. ), whereas the remaining constructs failed to yield any staining in E8.2 embryos. To confirm the specificity of the 3.5-kb fragment, we linked it to the hsp68 promoter, yielding construct3′-1hsp. This construct also gave rise to X-gal staining in the left LPM, but the intensity of the staining was much greater than that obtained with 3′-1. Transverse sections confirmed that both 3′-1 and 3′-1hsp induced left-sided expression in LPM (including the splanchnopleure and somatopleure) (Fig. B). X-gal staining in the PFP was not apparent with 3′-1, but it was observed with 3′-1hsp. Transverse sections of embryos transgenic for 3′-1hspindicated that the staining in the PFP was predominantly left-sided (Fig. B). These results suggested the presence of an ASE in intron 1 of nodal.

To verify further the presence of a left side-specific enhancer in intron 1, we generated a permanent mouse line (A8) harboring3′-1 and two such lines (H2 and H13) harboring3′-1hsp. In all three transgenic lines, X-gal staining was observed in left LPM at the early somite stage (Fig. ; data not shown). In embryos of lines H2 and H13, left-sided staining was also apparent in the PFP (Fig. A). These X-gal staining patterns were indistinguishable from those observed by ‘transient’ transgenic assay (Fig. ), further confirming the presence of a left side-specific enhancer in intron 1.

ASE is essential and sufficient for L-R asymmetric expression

To delineate further the ASE responsible for asymmetric expression of nodal, we prepared two sets of deletion mutants from3′-1hsp and examined their transcriptional activity in transgenic embryos (Fig. ). Among the 5′ deletion mutants, left-sided lacZ expression in LPM and PFP was observed with 5′Δ2–30 and5′Δ1–20, but not with5′Δ1–12,5′Δ2–8, or5′Δ2–11. Of the 3′ deletion mutants,3′Δ2–1 and3′Δ107 gave rise to asymmetric expression, whereas 3′Δ2–7 did not. These results localized the ASE to a 340-bp region of intron 1. Among the deletion mutants examined, left PFP was positive for X-gal staining whenever left LPM was positive. Thus, this 340-bp region alone was able to direct asymmetric expression in both the LPM and PFP.

Internal deletion of the 340-bp region containing ASE from construct3′-1, yielding 3′-1ΔASE, abolished the asymmetric expression of X-gal in LPM (Fig.), indicating that ASE is absolutely required for such expression. When the 340-bp region was linked to the 5-kb proximal promoter region of nodal, the resulting construct (5′-1ASE) yielded left-sided X-gal staining in LPM (Fig.). Finally, a construct (501/301hsp) in which the 340-bp region was linked to the hsp68 promoter rather than to the nodal promoter, gave rise to X-gal staining in both left LPM and left PFP (Fig. ). Together, these results suggested that ASE is both essential and sufficient for the asymmetric expression ofnodal.

Sensitivity of ASE to iv, inv, and lefty-1 mutations

Expression of nodal is affected in situs mutants such asiv/iv, inv/inv, andlefty-1 ^−/− mice; the pattern of expression is randomized (left sided, right sided, or bilateral) iniv/iv mutants, reversed in inv/inv mutants, and bilateral inlefty-1 ^−/− mutants (Collignon et al. 1996; Lowe et al. 1996; Meno et al. 1998). These observations indicate that nodal acts downstream of the iv,inv, and lefty-1 genes. We therefore examined whether expression of nodal–lacZ transgenes was affected byiv, inv, or lefty-1 mutations (Fig.). Mice of transgenic lines H2 and H13, which harbor3′-1hsp, were mated with iv/iv mice, and thelacZ transgene was transferred to the iv/iv background. Theiv/iv embryos harboring the transgene showed three different patterns of X-gal staining in LPM (Fig. A): left-sided (2/8 embryos), right-sided (2/8 embryos), or bilateral (4/8 embryos). The pattern of X-gal staining in PFP was similarly affected; for example, staining in one PFP was reversed when the staining in LPM was reversed (Fig. A).

The same transgene was also transferred to theinv/inv background by mating mice of lines H2 and H13 with inv/+ mice (inv/inv homozygotes are lethal). After crossing [inv/+, 3′-1hsp] mice with [inv/+] mice, ∼25% of thelacZ ⁺ embryos, which were inv/invhomozygotes, showed aberrant X-gal staining in the LPM; the staining pattern in such embryos was either right-sided (1/10 embryos), right-side dominant (4/10 embryos), or bilaterally equal (5/10 embryos) (Fig. B). Again, the pattern of X-gal staining in PFP was affected similarly to that in LPM. These patterns of X-gal staining observed in theiv/iv and inv/inv mice were almost identical to those of the distribution of nodalmRNA in these mutants (Collignon et al. 1996; Lowe et al. 1996). A minor discrepancy was that, among inv/invembryos, the frequency of bilateral X-gal staining was somewhat higher than that of bilateral nodal mRNA distribution. The 3.5-kb region of nodal contained in 3′-1hsp, which includes ASE, was thus able to reproduce the response of the intact gene to the iv and inv mutations.

In lefty-1 ^−/− mutant mice,nodal (as well as lefty-2) expression in the LPM becomes bilateral (Meno et al. 1998). We examined whether the 3.5-kb fragment of nodal containing ASE was able to respond to thelefty-1 mutation. The 3′-1hsp transgene was transferred to the lefty-1 ^−/−background by mating mice of lines H2 and H13 with lefty-1mutant animals. At the two- to four-somite stages, the transgene was expressed on the left side oflefty-1 ^−/− embryos. However, at the four- to eight-somite stages, alllefty-1 ^−/− embryos showed bilateral expression of lacZ in the LPM (Fig. C). Thus,lacZ expression in the right LPM was apparent later than that in the left LPM. At the four- to six-somite stages, X-gal staining in the right LPM extended posteriorly (to a greater extent than doeslefty-2 expression). In addition, X-gal staining in the PFP was up-regulated and bilateral. These patterns of X-gal staining inlefty-1 ^−/− embryos are consistent with the distribution of nodal mRNA observed previously in these animals (Meno et al. 1998). The lefty-1mutant mice were also mated with animals of the A8 line containing the3′-1 transgene. As expected, this transgene was expressed bilaterally in the LPM of embryos with thelefty-1 ^−/− background. Expression of lacZ in the PFP, which was not observed in the wild-type background (Figs. and ), was apparent in thelefty-1 ^−/− background (Fig. D) . These results suggest that the 3.5-kb region of nodal, and most likely ASE, is able to respond not only to the iv andinv mutations but also to thelefty-1 ^−/− mutation.

ASE-directed transgene expression in the epiblast and endoderm of gastrulating embryos

nodal is known to exhibit dynamic expression patterns during development; it is expressed in the epiblast at E5.5 to E6.0, the visceral endoderm at E6.0 to E6.5, and the posterior ectoderm at the early-streak stage (E6.5) (Conlon et al. 1994; Collignon et al. 1996; Varlet et al. 1997). We therefore examined whether ASE conferred similar expression patterns on lacZ transgenes. Embryos were recovered at various developmental stages either from permanent line A8 (harboring the 3′-1 transgene) or from lines H2 and H13 (harboring the 3′-1hsp transgene) and were subjected to X-gal staining (Fig. ). In embryos containing the 3′-1hsptransgene, X-gal staining was apparent throughout the epiblast at E5.5. At E6.5, lacZ expression was restricted to the posterior epiblast. The visceral endoderm also became transiently positive for X-gal staining at E6.5. The expression of lacZ was restricted to the posterior ectoderm, before appearing in left LPM and left PFP at E8.0. Embryos harboring the 3′-1 transgene showed a similar pattern of X-gal staining in the epiblast (with the exception that the staining was restricted to the proximal epiblast at E6.5). Again, the anterior visceral endoderm was positive for X-gal staining at E6.0 to E6.5. These results suggest that ASE is responsible not only for asymmetric expression of nodal at the early somite stage but also for the expression in the epiblast and visceral endoderm at earlier stages.

We also generated a permanent transgenic line (N1) containing the5′-2 construct and examined the expression of the transgene in gastrulating embryos. At E5.5, marked X-gal staining was apparent in the proximal epiblast. The expression of lacZbecame undetectable at E6.5 (both the ectoderm and the endoderm were negative), but it appeared in the node at E7.5 and was maintained until the early somite stage (E8.0 to E8.2). At E8.0, weak X-gal staining was also detected in the left LPM [consistent with the observation by transient transgenic assay that construct 5′-1 gave rise to faint X-gal staining in left LPM (Fig. )]. The observation thatlacZ expression was induced in the anterior visceral endoderm by 3′-1 and 3′-1hsp, but not by5′-2, suggests that nodal expression in the visceral endoderm is regulated by the 3.5-kb intron region, most likely by ASE.

Sequence conservation between lefty-2 ASE andnodal ASE

The lefty-2 and nodal genes exhibit similar expression patterns; both are asymmetrically expressed in the left LPM at the same developmental stage (two- to five-somite pair stage). We recently showed that L-R asymmetric expression of lefty-2 is controlled by a left side-specific enhancer (Saijoh et al. 1999). Our present data suggest that, like that of lefty-2, the expression of nodal is also regulated by such an ASE element. We therefore compared the nucleotide sequence of the nodal ASE (340 bp) with that of the lefty-2 ASE (350 bp). Several blocks of sequences were found to be conserved between the two enhancers (Fig.).

To identify the conserved sequence motifs required for activity of thenodal ASE, we dissected the 340-bp region further. Various deletion mutants were linked to the hsp68 promoter and the expression of the resulting constructs was examined (Fig.). Deletion of the 340-bp region from the 3′ end revealed that the 501/302hsp construct was able to induce expression both in the left LPM and the PFP, although the extent of expression was less marked than that apparent with501/301hsp. The 501/303hsp and 501/304hspconstructs gave rise to expression in the LPM but not in the PFP. Construct 501/305hsp exhibited no ASE activity. Deletion of the 340-bp region from the 5′ end revealed that502/301hsp and 503/301hspretained the ability to induce expression in both left LPM and PFP, whereas 504/301hsp had lost ASE activity.

These results suggest that a 10-bp conserved sequence, CAATCCACAT (shown in pink in Fig. ), is essential for expression in the left LPM as well as in the PFP. Although two versions of this sequence are present in the 340-bp region, the 3′ version appears more important for asymmetric expression than does the 5′ version. This 10-bp sequence is located between the deletion points of5′Δ1–20 (which was active) and5′Δ1–12 (which was inactive) (Figs. and). However, additional sequences are also necessary for maximal ASE activity, because activity gradually decreased on deletion of the 340-bp region from either end. These data also suggest that the 3′ region of ASE is necessary for expression in the PFP. In particular, the difference in activity between 501/302hspand 501/303hsp suggests that a 7-bp conserved sequence, CCCTGCC (shown in yellow in Fig. ), may be essential for ASE activity in the PFP.

Discussion

Role of ASE in the asymmetric expression of nodal

Previously, we have shown that L-R asymmetric expression oflefty-2 is attributable to ASE, whereas that oflefty-1 is mediated by a combination of bilateral enhancers and a right side-specific silencer (Saijoh et al. 1999). We have now shown that expression of nodal, like that of lefty-2, is regulated by a left side-specific enhancer (Fig.). Similar results have been obtained by others (Norris and Robertson 1999).

In our transgenic assays, ASE appeared to be both essential and sufficient for the asymmetric expression of nodal. In general, transgenic or transfection assays often overestimate the role of acis-regulatory element. However, targeted deletion of a 600-bp region spanning the nodal ASE almost abolished the left-sided expression of nodal, confirming the critical role of ASE in the asymmetric expression of this gene (Norris and Robertson 1999).

It is not absolutely clear whether ASE alone is responsible for the asymmetric expression of nodal. Although targeted deletion of ASE in embryonic stem cells largely prevented the asymmetric expression of nodal, weak left-sided expression was still apparent in the left LPM (Norris and Robertson 1999). Furthermore, in our transient transgenic assay, 5′-1, a construct lacking ASE, repeatedly gave rise to faint X-gal staining in a few cells in the left LPM. The permanent line N1, harboring the 5′-2 transgene, also showed similar weak lacZ expression in the left LPM. Thus, although ASE appears to have the major role, an additional enhancer (LSE, for left side-specific enhancer) in the proximal promoter region (between −5 and 0 kb) likely also contributes to the asymmetric expression of nodal. Although ASE and LSE might each act independently as a left side-specific enhancer in our transgenic assays, they may function synergistically to establish asymmetric expression of nodal in vivo.

The expression of nodal in the PFP has not been described previously. However, the nodal ASE was able to direct left-sided expression of lacZ in the PFP. This effect was most apparent when the nodal ASE was linked to the hsp68promoter. For several reasons, we believe that the ASE-induced expression in the PFP is not an artifact of the transgenic assay, but that it rather reflects expression of nodal in the PFP in vivo: (1) X-gal staining in the PFP was repeatedly observed with3′-1hsp and its deletion mutants in transient as well as permanent transgenic embryos; (2) X-gal staining in the PFP showed an apparent L-R specificity, being observed predominantly in a few cells on the left side, similar to the expression patterns oflefty-1 and lefty-2; and (3) X-gal staining in the PFP, like that in LPM, responded to iv, inv, andlefty-1 mutations. We attempted to detect nodal mRNA in the PFP by whole-mount in situ hybridization; however, although weak signals were detected occasionally in the PFP, we were unable to obtain convincing results (data not shown). It is likely that the extent ofnodal expression in the PFP is not sufficient to be detected by this method. A similar reason may also explain why X-gal staining in the PFP was not obvious with 3′-1 (the hsp68promoter appeared more efficient than did the nodal promoter, given that the intensity of X-gal staining in the left LPM induced by3′-1hsp was greater than that induced by3′-1). An alternative possibility is that although ASE possesses an intrinsic ability to induce expression in the PFP,nodal expression is normally repressed in PFP by a negative regulatory element. nodal expression in the PFP may be negatively regulated by Lefty-1 because 3′-1 transgene gave rise to X-gal staining in the absence of lefty-1.

The 3.5-kb intronic fragment containing ASE responded to iv,inv, and lefty-1 mutations. Expression of the3′-1hsp (and 3′-1) transgene was affected in a manner similar to that of nodal itself, as determined previously by in situ hybridization (Collignon et al. 1996; Lowe et al. 1996; Meno et al. 1998). It is therefore likely that signals derived from iv, inv, and lefty-1 converge within the ASE region of nodal.

Our data suggest that the expression of nodal in the node is regulated by a separate enhancer (NDE) located in the far-upstream region. Although nodal expression in the node exhibits subtle L-R specificity (Collignon et al. 1996), constructs 5′-2and SN5.0, as well as deletion mutants derived fromSN5.0, failed to show L-R specificity of expression in the node. Thus, NDE itself does not appear to possess L-R specificity. Such asymmetric expression of nodal in the node may be achieved by a combination of NDE and an additional cis-regulatory element such as ASE.

Similar transcriptional regulatory mechanisms underlying expression of nodal in the epiblast, endoderm, and left LPM

nodal is known to exhibit very dynamic expression patterns during development; it is expressed throughout the epiblast at E5.5, in the visceral endoderm at E6.0 to E6.5, in the node between E7.5 and E8.0, and in the left LPM at E8.0 (Zhou et al. 1993; Collignon et al. 1996; Lowe et al. 1996; Varlet et al. 1997). Expression ofnodal in the visceral endoderm is required for specification of the anterior axis during gastrulation (Varlet et al. 1997). In our transgenic assays, ASE was able to induce expression not only in the left LPM, but also in the epiblast and the visceral endoderm between E5.5 and E6.5. Targeted deletion of a 600-bp region that includes ASE also affected nodal expression in the epiblast, visceral endoderm, and left LPM (Norris and Robertson 1999). Therefore,nodal expression in the epiblast, visceral endoderm, and left LPM appears to be controlled by the same enhancer, ASE, suggesting that similar transcriptional regulatory mechanisms underlie nodalexpression in these domains.

The 6.5-kb region including NDE induced expression not only in the node at the early somite stage (E8.0) but also in the proximal epiblast at E5.5 (Fig. A) (Norris and Robertson 1999). A similar proximodistal (P-D) graded expression in the epiblast was also observed at E6.5 in a permanent transgenic line harboring 3′-1, consistent with the notion that the epiblast is already specified along with the P-D axis at this early stage (Beddington and Robertson 1999). Thus, the proximal epiblast, which moves posteriorly and later gives rise to the primitive streak, appears to express nodal at a higher level than does the distal epiblast, although previous in situ hybridization analysis failed to detect such a P-D expression gradient. A higher level of nodal expression may be required for the posterior movement of proximal epiblast cells.

Similarities between nodal ASE and lefty-2 ASE

The ASEs of lefty-2 and nodal exhibit several common features: (1) Both enhancers direct left-sided expression not only in the LPM and but also in the PFP; (2) they both respond toiv, inv, and lefty-1 mutations (Fig. ;Saijoh et al. 1999); and (3) with both enhancers, directed expression in the PFP can be partially separated from that in the LPM. One difference between the two enhancers is that whereas thelefty-2 ASE (350-bp region) can be clearly divided into at least two subdomains (one for the anterior LPM, the other for the posterior LPM), no such subdivision was apparent for nodal ASE (340 bp region). Finally, several conserved sequence motifs are apparent in the nucleotide sequences of the two enhancers. Thus, the asymmetric expression of nodal and of lefty-2 may be regulated by similar transcriptional mechanisms.

Dissection of the 340-bp region of the nodal ASE suggested that most (if not all) of the conserved motifs are required for maximal enhancer activity. However, a 7-bp conserved sequence (CCCTGCC) located in the 3′ portion of this region may be particularly important for expression in the PFP. Moreover, a 10-bp conserved sequence (CAATCCACAT), which exist twice in nodal ASE andlefty-2 ASE, appears essential for expression in both left LPM and PFP. However, database searches (TRANSFAC) failed to identify any previously known transcription factor binding sites that corresponded to these sequences. Clarification of the mechanism by which signals derived from upstream genes, such as the inv and ivgenes, induce the asymmetric expression of nodal will require identification of the transcription factors that bind to these conserved sequences.

Materials and methods

lacZ constructs

The nodal gene was isolated from a genomic library derived from 129/Sv mice. The 3′ end of the nodalupstream fragment (SalI–HindIII, 5 kb) is located in the 5′ untranslated region (+80 bp relative to the putative TATA box, −50 bp relative to the translation initiation codon). This upstream fragment (from −5 kb to +80 bp) was linked to thelacZ fragment (HindIII–BamHI, 4 kb) derived from pCH110 (Pharmacia) and was then subcloned in Bluescript or pGEM, yielding construct 5′-1. To generate construct5′-2, we subcloned a 6.5-kb SalI fragment (−11.5 to −5 kb) into the SalI site of5′-1. The NDEhsp construct was created by subcloning the blunt-ended 6.5-kb SalI fragment (−11.5 to −5 kb) into the SmaI site of hsp68–lacZpA(Kothary et al. 1989). For generation of constructs 3′-1to 3′-5, the indicated fragments of nodal (Fig.) were subcloned into 5′-1 at the 3′ end oflacZ. For generation of 3′–1hsp, the 3.5-kb intron fragment was subcloned into the SmaI site ofhsp68–lacZpA in either orientation. For the deletion mutants derived from 3′-1hsp, deletion proceeded from theSalI site. For all constructs, lacZ fragments free of vector sequences were isolated by gel electrophoresis before microinjection.

Transient transgenic assay

Transient transgenic assays were performed essentially as described previously (Saijoh et al. 1999). Transgenic embryos were generated by injection of lacZ constructs (4 ng/1 μl) into pronuclei of fertilized eggs obtained from crosses between (C57BL/6 × C3H) F1 mice (Hogan et al. 1994; Sasaki and Hogan 1996). The injected embryos were transferred into pseudopregnant recipients and allowed to develop in utero until E8.2. The E8.2 embryos were examined for the presence of the transgene (PCR) and for lacZ expression (by X-gal staining) according to standard protocols. The amount of β-galactosidase activity was estimated from the extent of X-gal staining after incubation in the staining buffer for 2, 8, and 24 hr. When a given construct was active, most of X-gal-positive embryos showed the staining in restricted sites without ectopic lacZ expression, whereas some embryos had variable ectopic expression as well. This was the case for all of the active constructs used in this study except for NDEhsp, which always gave rise to X-gal staining in ectopic sites as well as in the node. For each construct, a representative embryo showing a typical staining pattern is presented in the figures. Primers used to amplify the lacZ sequence were 5′-CTCAAACTGGCAGATGCACGGT-3′ and 5′-CGTTGCACCACAGATGAAACGC-3′.

Permanent transgenic lines

For some lacZ constructs, the transgenic embryos were allowed to develop to term and permanent transgenic lines were established. The presence of the transgene was examined by Southern blot analysis with the lacZ fragment as a probe. ThelacZ transgenes were introduced into the background ofiv/iv or inv/+ by first mating transgenic mice with iv/iv orinv/+ mice, respectively. The genotype of the resulting offspring was determined by PCR or Southern blot analysis. [iv/iv, lacZ] or [inv/+, lacZ] mice thus obtained were mated, and transgene expression was examined at E8.2.The ivalleles were genotyped by PCR with primers 5′-GCCAGCAATGAACGAGTGGCCCTCAAACCT-3′ and 5′-AGCTCTGGAAACCGTGGCTGGTGTGGCTGT-3′, followed by TaqI digestion (the wild-type allele is sensitive to TaqI, yielding 50-bp fragments; the mutant allele is resistant to TaqI, yielding a 100-bp fragment) (Supp et al. 1997).