Conserved regulation and role of Pitx2 in situs-specific morphogenesis of visceral organs

Although recent studies have provided insight into how left-right (LR)asymmetry is generated during vertebrate development(Hamada et al., 2002; Levin, 2005; Tabin, 2005), knowledge of this process remains limited. One of many important questions still unanswered concerns the mechanism by which situs-specific organogenesis is executed.

Situs-specific morphogenesis in the mouse begins soon after the loss of asymmetric expression of Nodal on the left side of the lateral plate mesoderm (LPM). The looping of the heart tube toward the right side is followed by asymmetric lobe formation in the lungs, rotation of the digestive tract and remodeling of the vascular system. This asymmetric morphogenesis occurs in response to Nodal, which functions as a left-side determinant. Organ primordia that have received the Nodal signal thus adopt a left-side morphology, whereas they adopt a right-side morphology in the absence of this signal. At the cellular level, the left and right sides of organ primordial may differ with regard to rate of cell proliferation, apoptosis or migration. However, the cellular basis of the generation of morphological asymmetry is poorly understood.

The bicoid-type homeobox transcription factor Pitx2 is thought to play a major role in asymmetric morphogenesis(Campione et al., 1999; Logan et al., 1998; Ryan et al., 1998; Yoshioka et al., 1998). The Pitx2 gene encodes three isoforms: Pitx2a, Pitx2b and Pitx2c. Pitx2a and Pitx2b mRNAs are generated by alternative splicing and are expressed bilaterally, while Pitx2c mRNA, a transcript from an alternative promoter, is expressed asymmetrically(Kitamura et al., 1999; Liu et al., 2002; Schweickert et al., 2000). Pitx2 knockout mice manifest LR defects in many organs, including the typical right isomerism of the lungs (Gage et al., 1999; Kitamura et al.,1999; Lin et al.,1999; Lu et al.,1999). However, the exact role of Pitx2 in situs-specific organogenesis is unknown.

Asymmetric expression of Pitx2 begins in the left LPM concomitantly with that of Nodal, but it persists after Nodal expression disappears. Pitx2 expression is thus still apparent in the LPM-derived mesenchyme of various visceral organs at the late somite stage. Asymmetric expression of mouse Pitx2 is conferred by an enhancer (ASE) that contains three binding sites for the transcription factor Foxh1, a target of Nodal signaling, as well as a binding site for the homeobox transcription factor Nkx2 (Shiratori et al., 2001). The Foxh1 binding sites are essential for the initiation of asymmetric Pitx2 expression, whereas the Nkx2-binding site is dispensable for such initiation but necessary for maintenance of late-stage expression. The left-sided expression of Pitx2 is thus initiated by Nodal signaling and maintained by Nkx2. Consistently, asymmetric Pitx2 expression is lost in mutant mice lacking any of the Nodal signaling components, such as Foxh1(Yamamoto et al., 2003) and cryptic [Cfc1 - Mouse Genome Informatics; a co-receptor for Nodal(Shen and Schier, 2000; Yan et al., 1999)]. The LR asymmetric expression pattern of Pitx2 is conserved among all vertebrates examined.

We have now studied the regulation and role of Pitx2 in asymmetric organogenesis by examining the asymmetric enhancer ASE of this gene. Our data suggest that asymmetric Pitx2 expression in vertebrates other than the mouse is also regulated by this highly conserved enhancer. Generation of mice that lack the Pitx2 ASE confirmed its essential role in regulation of Pitx2 and revealed the precise role of Pitx2 in situs-specific morphogenesis. We also tested the importance of two-step regulation of Pitx2 expression by rescue experiments with a Pitx2 transgene.

Genomic DNA libraries for rat, human and zebrafish (Stratagene) or for chicken (Clontech) were screened by hybridization with mouse Pitx2cDNA as a probe. DNA fragments containing the last intron of Pitx2were obtained from the isolated clones by PCR with a pair of primers corresponding to exons 4 and 5. DNA fragments containing the whole intron 5 of vertebrate Pitx2 genes were ligated to the 5′ end of an hsp68-lacZ reporter construct(Kothary et al., 1988). Seven tandem repeats of the Nkx2-binding sequence of mouse Pitx2 were ligated to the 5′ end of a 380 bp DNA fragment containing the ASE of mouse Lefty2 (Saijoh et al.,1999) and to the hsp68-lacZ construct. A Foxh1-lacZ transgene was constructed from a 234 kb bacterial artificial chromosome (BAC) clone (RP23-156P23) containing mouse Foxh1; an internal ribosomal entry site (IRES)-lacZ cassette was thus inserted into the 3′ untranslated region of Foxh1upstream of the poly(A) addition sequence in the BAC clone. Transgenic mice were generated by injection of these various lacZ constructs into the pronucleus of fertilized eggs as described previously(Saijoh et al., 1999). They were then collected at embryonic day (E) 8.2 or 10.5 and stained with theβ-galactosidase substrate X-gal.

A targeting vector was designed to delete the 0.6 kb ASE region of Pitx2 (see Fig. 4A). To delete the PGK-neo sequence, we subjected targeted embryonic stem(ES) cells to electroporation with a CAG-Cre vector(Sakai and Miyazaki, 1997). ES cell culture and injection of blastocysts obtained by crossing B6C3F1 males and C57BL/6J females were performed by standard methods(Hogan et al., 1994). Male chimeras were bred with C57BL/6J females to yield heterozygous F₁offspring, which were then mated with each other to produce Pitx2^ΔASE/ΔASEhomozygotes. Offspring and embryos were genotyped by PCR; primers for the wild-type Pitx2 allele were 5′-TCACTGAAATCCCTGGGGAGA(ΔASE55) and 5′-AGAAGCAAGCCTCACGCACTT (ΔASE53), and those for the targeted allele wereΔASE55 and 5′-AGCCACAGAAGCCATCAACTT[ΔASE53(2)]. The former primers yielded a 134 bp product and the latter a 370 bp product.

Whole-mount in-situ hybridization was performed with standard protocols and probes specific for Pitx2(Yoshioka et al., 1998) or cryptic (Shen et al., 1997)mRNAs. For histological analysis, some embryos that were subjected to X-gal staining or whole-mount in-situ hybridization were embedded in paraffin and sectioned at a thickness of 8 μm. For scanning electron microscopy, E12.5 mouse embryos were processed using the standard procedure(Nonaka et al., 2005) and were observed using a scanning electron microscope (S-2600N, Hitachi, Japan).

Two transgenes that confer asymmetric expression of Pitx2c were designed as follows. A 0.9-kb DNA fragment containing the wild-type ASE of Pitx2 or a mutant ASE (NmASE) in which the Nkx2 binding site was mutated was linked to a 1.3-kb fragment of the Pitx2c promoter, Pitx2c cDNA, and an IRES-lacZ cassette. The resulting constructs were microinjected into the pronucleus of fertilized mouse eggs obtained by crossing Pitx2^ΔASE/+heterozygous B6/129 males and wild-type C57BL/6J females. Thirteen and 10 lines were established for the wild-type and mutant ASE constructs,respectively. Expression of each transgene was monitored by X-gal staining. Three transgenic lines (Tg39 and Tg55 for the wild-type ASE,and Tg50 for the mutant ASE) that showed similar levels of transgene expression at E8.2 were studied. Pitx2^ΔASE/+ males harboring each transgene were mated with Pitx2^ΔASE/+ females to generate Pitx2^ΔASE/ΔASEhomozygotes with the transgene.

Two permanent transgenic mouse lines, Pitx2 17-lacZ and Pitx2-Cre, were established. The Pitx2 17-lacZ and Pitx2-Cre transgenes consist of a 1.3 kb fragment of the Pitx2c promoter and either lacZ or Cre followed by a 17 kb fragment containing exons 1c, 4 and 5, as well as the ASE and 3′flanking region of Pitx2(Shiratori et al., 2001). These transgenes were first introduced into Pitx2^ΔASE/+ heterozygotes. To study the lineages of left LPM cells expressing Pitx2, we crossed Pitx2^ΔASE/+ males harboring the Pitx2-Cre transgene with Pitx2^ΔASE/+ females harboring the CAG-CAT-lacZ transgene (Sakai and Miyazaki, 1997). The resulting embryos were collected at E9.5 and stained with X-gal.

Asymmetric expression of mouse Pitx2 is controlled by an enhancer(ASE) located in the last intron, intron 5(Shiratori et al., 2001). The ASE contains three Foxh1-binding sequences and an Nkx2-binding sequence(Fig. 1). The former sites are required for initiation of Pitx2 expression in left LPM at E8.2(Fig. 2A), whereas the latter is responsible for maintenance of asymmetric expression from E9.5 to 10.5(Fig. 2F,K,P,U)(Shiratori et al., 2001). We therefore examined Pitx2 genes of other vertebrates to determine whether this transcriptional regulatory mechanism for achieving asymmetric expression is conserved.

We isolated Pitx2 genes from human, rat, chicken and zebrafish. For chicken Pitx2, a 16 kb region containing the last intron, which is located between the exons encoding the homeodomain in all vertebrates examined, was tested for ASE activity with a transgenic mouse assay. ASE activity was initially localized to a 9.0 kb region and was subsequently mapped to a 2.0 kb region containing the last intron. An hsp68-lacZreporter construct driven by the 9.0 kb fragment of chicken Pitx2thus gave rise to asymmetric X-gal staining not only in left LPM at E8.2(Fig. 2C) but also in various primordial organs, including the common atrial chamber, lung bud, septum transversum and gut dorsal mesentery at E10.5(Fig. 2H,M,R,W). Nucleotide sequencing of the 2.0 kb region containing intron 5 revealed the presence of two Foxh1-binding sequences and an Nkx2-binding sequence in the last intron(Fig. 1). For the Pitx2 genes of human, rat and zebrafish, DNA fragments containing the entire last intron were isolated by PCR and sequenced. Each fragment contained two or three Foxh1-binding sequences and an Nkx2-binding sequence(Fig. 1). The fragment of human Pitx2 showed typical ASE activity, with the corresponding hsp68-lacZ transgene giving rise to left-sided X-gal staining in LPM at E8.2 (Fig. 2B) as well as in various organs at E10.5 (Fig. 2G,L,Q,V). The DNA fragment containing the last intron of zebrafish Pitx2 conferred asymmetric expression of the reporter construct in left LPM at E8.2 (Fig. 2E) but not in primordial organs at E10.5(Fig. 2J,O,T,Y). Moreover,expression of the transgene in left LPM was apparent only in the anterior portion (Fig. 2E). As we showed previously (Shiratori et al.,2001), the ASE of frog Pitx2 was active in left LPM at E8.2 (Fig. 2D) as well as in the common atrial chamber and gut dorsal mesentery but not in the lung bud or septum transversum at E10.5 (Fig. 2I,N,S,X). These results thus showed that an enhancer that regulates asymmetric expression of Pitx2 (ASE) is evolutionarily conserved among vertebrates. In all species examined, the ASE contains multiple Foxh1-binding sites and an Nkx2-binding site, suggesting that asymmetric expression of vertebrate Pitx2 is initiated by Nodal signaling and maintained by Nkx2.

The role of the Nkx2-binding sequence in the Pitx2 ASE was examined further by linking it to the asymmetric enhancer (ASE) of mouse Lefty2. The Lefty2 ASE contains two Foxh1-binding sequences but no Nkx2-binding sequence, and it gives rise to asymmetric expression in left LPM at E8.2 (Fig. 3A)(Saijoh et al., 1999) but not in organ primordia at E10.5 (Fig. 3C,E,G) (Shiratori et al.,2001). However, the addition of seven copies of the Nkx2-binding sequence of mouse Pitx2 to the Lefty2 ASE(N₇-Lefty2 ASE) (Fig. 3B) resulted in asymmetric expression of the corresponding hsp68-lacZ reporter construct in various visceral organs at E10.5(Fig. 3D,F,H). An Nkx2-binding sequence is thus not only essential but also sufficient for maintenance of asymmetric gene expression.

To establish the role of the ASE in Pitx2 regulation, we generated a mutant Pitx2 allele that lacks the ASE(Pitx2^ΔASE)(Fig. 4). We first examined how asymmetric Pitx2 expression was affected by ASE deletion. Pitx2^ΔASE/ΔASEembryos developed normally until the early somite stage. As expected,asymmetric Pitx2 expression in left LPM was abolished at the early somite stage, whereas bilateral Pitx2 expression in the head mesenchyme remained unaffected (Fig. 5A,B). At later stages (E9.0 to 9.5), left-sided Pitx2expression in various primordial organs also did not develop in Pitx2^ΔASE/ΔASEembryos (Fig. 5D,E,G,H,J,K,M,N,P,Q). These results thus showed that asymmetric expression of Pitx2 was lost as a result of specific deletion of the ASE, confirming the role of the ASE in such expression.

Pitx2^ΔASE/ΔASEnewborn mice manifested various LR defects, including those previously reported for Pitx2 null mice(Kitamura et al., 1999; Lin et al., 1999; Lu et al., 1999) and mutant mice specifically lacking Pitx2c(Liu et al., 2002). Thus, Pitx2^ΔASE/ΔASEmice showed right isomerism in the lungs(Fig. 6A,B; 18/18 mice examined), reversed positioning of the great arteries(Fig. 6C,D; 18/18), reversed positioning of the heart apex toward the right(Fig. 6C,D; 5/18), bilateral inferior vena cava (Fig. 6C,D;5/18), right isomerism in the atrium (Fig. 6E,F; 18/18), and an endocardial cushion defect, as well as a common atrioventricular (AV) valve with a ventricular septal defect (VSD) and atrial septal defect (ASD) (Fig. 6G,H; 4/4), and double-outlet right ventricle (DORV). To know the cellular basis of the endocardial cushion defects, we also examined the AV cushions at E12.5 by scanning electron microscope(Fig. 6I,J) and by sagittal sectioning (Fig. 6K,L). The AV cushions consist of two portions, the superior AV cushion (SAVC) and inferior AV cushion (IAVC). The SAVC and IAVC are fused with each other at E12.5 in the wild-type embryo (Fig. 6I,K),but they remained separated in the Pitx2^ΔASE/ΔASEembryo (Fig. 6J,L). In the mutant embryo, SAVC was hypoplastic while IAVC appeared relatively normal in size (Fig. 6J,L). These results suggest that a hypoplastic SAVC is responsible for the common AV valve with VSD and ASD in Pitx2^ΔASE/ΔASEmice. This is consistent with the fact that Pitx2 is expressed in the myocardium adjacent to SAVC but is largely absent in the myocardium adjacent to IAVC (Kitamura et al.,1999).

We also noted previously unrecognized LR defects in the Pitx2^ΔASE/ΔASEmice, including reversed positioning of the azygos vein (12/18) and aorta(9/18) on the right side of the thorax(Fig. 6M,N), reversed relation of the aorta and vena cava in the abdomen(Fig. 6U,V; 4/16), abnormal location of the portal vein on the ventral side of the duodenum(Fig. 6O,P; 1/11), abnormal rotation of the gut, such as aberrant looping of the duodenum(Fig. 6Q,R; 12/12), and the absence of a cross of the duodenum and colon(Fig. 6S,T; 11/12), and mal-location of the pancreas on the ventral or right side of the duodenum(Fig. 6R; 12/12).

These morphological defects of Pitx2^ΔASE/ΔASEmice resemble those of cryptic knockout mice(Yan et al., 1999) and of conditional mutant mice that lack Foxh1 expression in the lateral plate (Yamamoto et al., 2003). Unlike the cryptic or Foxh1 mutant animals, however, some laterality-dependent events remained unaffected in Pitx2^ΔASE/ΔASEmice. In particular, the directions of heart looping and embryonic turning were normal in all (12/12) Pitx2^ΔASE/ΔASEembryos examined. Furthermore, the stomach was always located on the left side(12/12) and the position of the spleen was normal (12/12), although the latter was reduced in size (12/12), as has been described for Pitx2 knockout mice (Lin et al., 1999; Lu et al., 1999). These events and organ positions are thus probably regulated by a mechanism dependent on Nodal-cryptic-Foxh1 signaling but independent of the Pitx2 ASE.

In Pitx2^ΔASE/ΔASEmice, asymmetric expression of Pitx2 was not detected in left LPM at E8.2 (Fig. 5B) and most asymmetric expression domains normally apparent at E9.0 to 9.5 were also lost(Fig. 5E,H,K,N,Q). A reduced but substantial level of asymmetric Pitx2 expression was detected in the common cardinal vein and vitelline vein of the mutant embryos at E9.0 to 9.5, however (Fig. 5E,H,N,Q). To determine the mechanism by which this ASE-independent expression of Pitx2 is regulated, in particular whether or not it is induced by Nodal, we examined cryptic^-/- mice(Yan et al., 1999) and Foxh1 conditional mutant mice(Yamamoto et al., 2003) for Pitx2 expression in these veins. In the cryptic^-/-embryos, asymmetric expression of Pitx2 was lost in all organs,including the common cardinal vein and vitelline vein, at E9.0 and 9.5(Fig. 5F,I,L,O,R). These expression domains were also lost in the Foxh1 mutant(Yamamoto et al., 2003). The frequencies of an abnormal phenotype for the inferior vena cava in the thorax and for the relation between the portal vein and duodenum were 28 and 9%,respectively, in Pitx2^ΔASE/ΔASEmice, compared with corresponding values of 68 and 47% for cryptic^-/- mice (data not shown). The inferior vena cava and portal vein are derived from the common cardinal vein and vitelline vein,respectively, in both of which a reduced level of asymmetric Pitx2expression remained in Pitx2^ΔASE/ΔASEembryos. These results suggest that the Pitx2 expression in these veins of Pitx2^ΔASE/ΔASEmice is also induced by Nodal signaling.

We also examined whether cryptic, Foxh1 and Nodal are expressed in the ASE-independent expression domains of Pitx2. cryptic expression, as revealed by whole-mount in-situ hybridization, was apparent in LPM at E8.2 (Fig. 7A) but not in the common cardinal vein or vitelline vein at E9.0 or 9.5(Fig. 7D,G,J,M,P,S). To detect expression of Foxh1 or Nodal, we examined the Foxh1-lacZ BAC transgenic mice (see Materials and methods) and Nodal^lacZ/+ mice(Collignon et al., 1996),respectively. The Foxh1-lacZ BAC transgene was expressed in left LPM at E8.2 (Fig. 7C), consistent with the distribution of Foxh1 mRNA revealed by whole-mount in-situ hybridization (Saijoh et al.,2000). Expression of the Foxh1-lacZ transgene was apparent in the heart at E9.0 and 9.5, where Pitx2 is expressed, but it was absent from the common cardinal vein and vitelline vein(Fig. 7F,I,L,O,R,U). Similarly, Nodal expression was apparent in left LPM at E8.2(Fig. 7B), but it was detected at only a low level at E9.0 (Fig. 7E,H,K) and not at all at E9.5(Fig. 7N,Q,T) in the common cardinal vein and vitelline vein.

These results suggest that asymmetric expression of Pitx2 in the common cardinal vein and vitelline vein is induced by Nodal signal and is subsequently maintained in the absence of Nodal signal, which is consistent with our previous observations that asymmetric expression in various organs is maintained by Nkx2 in the absence of Nodal signal(Shiratori et al., 2001).

To study the fates of Pitx2-expressing cells, we established transgenic mice that express Cre under the control of a 17 kb region of Pitx2 that contains the ASE(Shiratori et al., 2001). These Pitx2-Cre mice were crossed with CAG-CAT-lacZ mice,which harbor a Cre-responsive lacZ transgene(Sakai and Miyazaki, 1997),and the resulting embryos were stained with X-gal. Stained cells were specifically located in the left LPM at E8.2 and on the left side of various visceral organs at E9.5 (data not shown). Given that the ASE is continuously active in left LPM-derived cells between E8.2 and 9.5, the Pitx2-Cretransgene would be expected to detect all the cells in which the ASE was once active.

The X-gal staining pattern at E9.5 obtained with the Pitx2-Cretransgene was compared with that obtained with Pitx2 17-lacZ, which consists of lacZ under the control of the 17 kb region of Pitx2 and would be expected to mark only those cells in which the ASE was active at the time examined. Although the two staining patterns were highly similar, differences were detected in the heart ventricle and AV canal. A large portion of the ventricle and myocardium adjacent to the IAVC of the AV canal was X-gal negative for Pitx2 17-lacZ(Fig. 8A-D) but X-gal positive for Pitx2-Cre (Fig. 8I-L). This differentially stained region thus represents a domain in which Pitx2 expression is induced at E8.2 but is repressed by E9.5, earlier than the remaining expression domains are repressed.

The X-gal staining pattern yielded by Pitx2 17-lacZ in Pitx2^ΔASE/ΔASEembryos at E9.5 (Fig. 8E-H),however, was virtually identical to that conferred by Pitx2-Cre in wild-type embryos (Fig. 8I-L). The large portion of the ventricle and AV canal in which Pitx2 17-lacZ was not expressed in wild-type embryos was thus X-gal positive in Pitx2^ΔASE/ΔASEembryos harboring this transgene. Such expansion of X-gal-positive cells in Pitx2^ΔASE/ΔASEembryos might result from either an abnormal contribution of cells in which the ASE was once active or from the loss of negative feedback by Pitx2 itself. To distinguish between these possibilities, we examined the pattern of X-gal staining conferred by Pitx2-Cre in Pitx2^ΔASE/ΔASEembryos (Fig. 8M-P). The staining pattern was indistinguishable from that of wild-type embryos harboring the same transgene (Fig. 8I-L), which favors the latter possibility. These results are in principle consistent with the fate mapping data in Xenopus(Ramsdell et al., 2005) but suggest a negative feedback loop in a portion of the AV canal. However, our findings require further work because a transgene does not always recapitulate correct expression patterns of a gene [as illustrated in Table 1 and the previous reports; for example, Mortlock et al.(Mortlock et al., 2003)].

Whereas asymmetric expression of Nodal is transient, asymmetric Pitx2 expression induced in LPM at E8.2 is maintained in LPM-derived cells of visceral organs for an additional 2 days. We next examined whether the earlier expression of Pitx2 in LPM is sufficient for situs-specific organogenesis of some organs or whether continuous expression is necessary for all organs.

To address this issue, we generated permanent transgenic mouse lines that express Pitx2 either transiently or continuously. These mice thus harbor a Pitx2 transgene driven either by a mutant ASE (NmASE) that lacks the Nkx2-binding site or by the wild-type ASE(Fig. 9A). In the former transgenic mice (line Tg50), the transgene was expressed in LPM at E8.2 (Fig. 9D) but not in visceral organs at E9.5 or 10.5 (Fig. 9G,J,M; data not shown), as expected. In the latter transgenic mice (lines Tg39 and Tg55), asymmetric expression of the transgene was apparent in LPM at E8.2 (Fig. 9B,C) and was maintained at E10.5. However, transgene expression was maintained at E10.5 only in subsets of visceral organs; it was highly expressed in the truncus arteriosus, weakly in the gut dorsal mesentery and not in lung bud in Tg39 mice (Fig. 9E,H,K), whereas it was highly expressed only in the gut dorsal mesentery in Tg55 mice (Fig. 9F,I,L). A high level of Pitx2 expression in some domains may be nonpermissive because we were unable to establish transgenic mouse lines that express the transgene in many visceral organs.

The transgenes in the Tg50, Tg39 and Tg55 lines were then examined for their ability to rescue the LR defects of Pitx2^ΔASE/ΔASEmice. The transgene of line Tg39, which was highly expressed in the truncus arteriosus (Fig. 9E),was able to rescue the morphological defect of the great vessels (11/13 mice,85%) (Fig. 10A) but not the internal defects of the heart (including ASD, VSD and DORV) or other defects(Fig. 10D,G; data not shown)of Pitx2^ΔASE/ΔASEmice. The transgene of line Tg55, which was highly expressed in the gut dorsal mesentery (Fig. 9L),was able to rescue the abnormal rotation of the gut (7/12, 58%) and aberrant positioning of the pancreas (7/12, 58%)(Fig. 10H, data not shown) but not other defects (Fig. 10B,E)of Pitx2^ΔASE/ΔASEmice. Thus, defects were rescued for the organs in which the transgene was highly expressed. Finally, the transgene of line Tg50 was unable to rescue any of the morphological defects of Pitx2^ΔASE/ΔASEmice (Fig. 10C,F,I). These results demonstrate that situs-specific morphogenesis requires continuous expression of Pitx2 on the left side of primordial organs.

Pitx2 is expressed asymmetrically for a long period during organogenesis and executes LR asymmetric morphogenesis in vertebrates(Campione et al., 1999; Logan et al., 1998; Ryan et al., 1998; Yoshioka et al., 1998). Mouse Pitx2 is regulated by an ASE that contains three Foxh1-binding sequences and an Nkx2-binding sequence, both of which are essential for asymmetric expression (Shiratori et al.,2001). The phenotype of Pitx2^ΔASE/ΔASE mice described in the present study now establishes an essential role for the ASE in Pitx2regulation. Addition of multiple copies of the Nkx2 binding sequence derived from the ASE of Pitx2 to the ASE of Lefty2 transformed the transient action of the latter into a longer lasting one, confirming the essential role of the Nkx2-binding sequence in the ASE of Pitx2. We have also now shown that Pitx2 of all vertebrates (mouse, rat, human,chicken, frog, zebrafish) examined possesses an ASE in the last intron with a similar organization, namely two or three Foxh1-binding sequences and one Nkx2-binding sequence. Asymmetric Pitx2 expression in vertebrates is thus regulated by a highly conserved enhancer, ASE.

The ASEs of Pitx2 from each of the various vertebrates examined showed largely similar activities in mouse embryos. The ASE of mouse, human,or chicken Pitx2 showed activity in the left LPM at E8.2 as well as in many organs, including the common atrial chamber, lung bud, septum transversum and gut dorsal mesentery, at E10.5. By contrast to the ASE of Pitx2 from other vertebrates, however, that of zebrafish Pitx2 was active only in the anterior portion of left LPM at E8.2 and was inactive in all organs examined at E10.5. It is thus possible that the Nkx2-binding sequence in the ASE of zebrafish Pitx2 is recognized by zebrafish Nkx2 but not by mouse Nkx2. Alternatively, the Nkx2-binding sequence of the ASE of zebrafish Pitx2 alone may not be sufficient for maintenance of Pitx2 expression.

The Pitx2^ΔASE/ΔASEmice manifested LR defects in many organs. LR asymmetry in some organs,however, remained normal in the Pitx2^ΔASE/ΔASEmice. This latter finding is not due to residual Pitx2 expression,which was detected only in the common cardinal vein and vitelline vein. In mutant mice that lack expression of cryptic(Yan et al., 1999) or Foxh1 (Yamamoto et al.,2003), however, those organs that remain normal in Pitx2^ΔASE/ΔASEmice are abnormal. The asymmetric morphogenesis of such organs thus appears to be regulated in a manner independent of Pitx2 but dependent on Nodal signaling.

LR asymmetric morphogenesis is achieved by three mechanisms: (1)directional looping of a tube; (2) differential lobation, as in the lungs or liver; and (3) one-sided regression, as in blood vessels(Hamada et al., 2002). LR asymmetric events regulated in a Pitx2-independent manner include heart looping, embryonic turning and looping of the duodenum-stomach, all of which correspond to the first pattern of morphogenesis. The corresponding organs form initially as a straight tube at the midline that subsequently undergoes looping or turning. Directional looping of the developing heart, for example,may be achieved by physical forces intrinsic to the heart, such as those generated by changes in the arrangement of intracellular actin bundles(Itasaki et al., 1991; Itasaki et al., 1989), changes in myocardial cell shape (Manasek et al.,1972) or differential rates of cell proliferation. Embryonic turning may also involve LR asymmetric rates of cell proliferation in LPM(Miller and White, 1998). Alternatively, the force for heart looping may be provided externally, such as by the adjacent splanchnopleure (Voronov et al., 2004). Although most situs-specific organogenesis depends on Pitx2, the mechanism of Pitx2 action remains unknown. Pitx2-dependent events include the development of asymmetries in lung lobation, blood vessel remodeling and atrial shape, and it remains to be determined how Pitx2 regulates such seemingly different cellular processes.

Pitx2 is expressed asymmetrically for a long period and is regulated in two steps: initiation by Nodal signaling and maintenance by Nkx2. In the present study, we examined whether the early phase of Pitx2expression is sufficient or whether continued expression is necessary for situs-specific organogenesis. Our data obtained from transgenic rescue experiments demonstrate that the persistent expression of Pitx2 is required. It is possible that Pitx2 regulates various cellular events in organs undergoing LR asymmetric morphogenesis. If so, what is the role of the early phase of Pitx2 expression in left LPM? The early-phase expression may play a positive role in LR morphogenesis by activating target genes that are essential for morphogenesis in the late phase. Alternatively,the early-phase expression may simply be necessary for the late-phase expression; although Pitx2 might play a role only during the late phase, its expression may need to be initiated at the early phase in response to Nodal signaling. Distinguishing between these possibilities will require the generation of transgenic mice that asymmetrically express Pitx2 only at the late phase.

Supplementary material for this article is available at http://dev.biologists.org/cgi/content/full/133/15/3015/DC1

We thank E. Robertson for Nodal^lacZ mice, N. Copeland for sharing the BAC recombination system, S. Gong for advice on BAC transgenesis, and H. Hashiguchi-Jo, K. Yamashita, S. Ohishi, K. Mochida, and Y. Ikawa for technical assistance. This work was supported by a grant from CREST (Core Research for Evolutional Science and Technology) of the Japan Science and Technology Corporation (to H.H.) and by a grant from the NIH (to M.M.S.).