Skip to main content
PMC home page
Proceedings of the National Academy of Sciences of the United States of America logoLink to Proceedings of the National Academy of Sciences of the United States of America
. 1998 Aug 4;95(16):9366–9371. doi: 10.1073/pnas.95.16.9366

Differential rescue of visceral and cardiac defects in Drosophila by vertebrate tinman-related genes

Maijon Park *, Carol Lewis *, David Turbay †, Amy Chung ‡, Jau-Nian Chen §, Sylvia Evans , Roger E Breitbart ‡,‖, Mark C Fishman §, Seigo Izumo †,**, Rolf Bodmer *,‡‡
PMCID: PMC21344  PMID: 9689086

Abstract

tinman, a mesodermal NK2-type homeobox gene, is absolutely required for the subdivision of the early Drosophila mesoderm and for the formation of the heart as well as the visceral muscle primordia. Several vertebrate relatives of tinman, many of which are predominately expressed in the very early cardiac progenitors (and pharyngeal endoderm), also seem to promote heart development. Here, we show that most of these vertebrate tinman-related genes can readily substitute for Drosophila tinman function in promoting visceral mesoderm-specific marker gene expression, but much less in promoting cardiac-specific gene expression indicative of heart development. In addition, another mesodermal NK2-type gene from Drosophila, bagpipe, which is normally only needed for visceral mesoderm but not heart development, cannot substitute for tinman at all. These data indicate that the functional equivalence of the tinman-related subclass of NK2-type genes (in activating markers of visceral mesoderm development in Drosophila) is specific to this subclass and distinct from other homeobox genes. Despite the apparent overall conservation of heart development between vertebrates and invertebrates, the differential rescue of visceral mesoderm versus heart development suggests that some of the molecular mechanisms of organ formation may have diverged during evolution.

tinman (tin) and bagpipe (bap) are two mesodermal NK2 class homeobox genes that are closely linked in the Drosophila genome (1, 2). tin is expressed in the early mesoderm (3), where it appears to confer competence to a field of cells to assume a fate necessary for cardiac and visceral mesoderm development. In contrast, the initial bap expression is confined to the gut muscle progenitors, apparently under the control of tin (2). In tin mutants, the absence of mesodermal subdivision results in a failure of the heart and gut muscle formation, whereas in bap mutants only the visceral component is affected (1, 2).

Despite the difference in mature heart morphology, the early embryology of vertebrate heart development is not unlike that of Drosophila (4). Moreover, six members of the tin-related subclass (5–14) and two members of the bap-related subclass (14, 15) of Nkx genes (the vertebrate equivalents of the Drosophila NK2 genes) have been isolated thus far in various species. There they are also predominantly expressed in the cardiac and/or visceral primordia (reviewed in refs. 4 and 8). The distinction between tin versus bap relationship of the Nkx genes has not been straight forward (thus providing additional motivation for the present study): The homeodomains of the vertebrate tin-related genes are very similar to each other (80–90%) and clearly form a distinct subclass from those of the vertebrate bap-related genes, to which they are only 55–65% identical. However, the vertebrate tin-related homeodomains are equally similar (about 65%) to both Drosophila tin and bap homeodomains (refs. 4 and 8; see also Fig. 1A). By contrast, vertebrate bap-related homeodomains are significantly more similar to those of bap (70–80%) than to those of tin (50–60%) (14, 15). Moreover, each of the tin-related genes apparently is expressed in the developing heart, although not exclusively and not in every species (8). The most convincing argument for the tin versus bap relationship of the vertebrate Nkx genes has been the discovery of a closely linked tin/bap-related pair of genes in the mouse genome (11, 14), which suggests that the common ancestor of vertebrates and invertebrates already had a tin- and a bap-like gene.

Figure 1.

Figure 1

Open in a new tab

(A) List of the cDNAs used in this study: Drosophila tin (D Tin), Nkx2–5 from mouse (M Nkx2–5/Csx) and zebrafish (Z Nkx2–5), Nkx2–3 from Xenopus (X Nkx2–3) and zebrafish (Z Nkx2–3), and Nkx2–7 from zebrafish (Z Nkx2–7). Approximate gene structures and sequence identities of the Nkx genes to tin and bap are indicated. TN, Tin/Nkx-specific domain of 11 amino acids (4, 8); HD, homeodomain; NK2-SD, NK2-specific domain (8). (B–G) Immunocytochemical staining. Eve expression (B–D) in a subset of cardiac progenitors along the dorsal mesodermal border (arrowheads) and FasIII expression (E–G) in the visceral mesoderm (arrowheads) of stage 12 wild-type embryos (B and E), homozygous Hstin,tinGC14 embryos (C, D, F, and G) heat shocked for 30 min between 3.5 and 4.5 hr of development (stage 9, C and F) or between 5 and 6 hr of development (early stage 11, D and G). Asterisks indicate the absence of marker gene expression in the presumptive heart (D) and visceral mesoderm (G) when heat shocked later. (H) Graph represents the amount of marker gene expression in presumptive cardiac (Eve expression in red) or visceral mesoderm (FasIII expression in blue) as a consequence of heat shock induction of the tin transgene. Each column represents the mean of 30 embryos or more. Anterior in all micrographs is to the left and dorsal is up.

It has been suggested that basic molecular–genetic mechanisms of heart (and perhaps also visceral) mesoderm development may be conserved between vertebrates and invertebrates (4, 8). In particular, it may be that the vertebrate tin-related genes, and perhaps even Drosophila bap, are functionally interchangeable with tin function in Drosophila. As a test for this hypothesis, we wanted to determine whether or not the vertebrate tin-related genes and/or Drosophila bap can substitute for a loss-of-tin-function in transgenic flies.

MATERIALS AND METHODS

Constructs and Fly Stocks.

Full-length Nkx2–3 (12, 13), 2–5 (6, 11), 2–7 (13), and bap (2) cDNAs were inserted behind the hsp70 heat shock promoter at the KpnI site of the WH1 vector or the XbaI site of hsCasPeR (16) and transgenic flies were generated as for Hstin in ref. 1. At least two independent insertions of each construct were crossed into a tin null mutant background (tinEC40 and tinGC14, as in ref. 1) and examined for restoration of cardiac and visceral mesodermal marker gene expression indicative of heart and/or visceral mesoderm formation. A bap null mutant was generated and kindly provided by M. Frasch (Brookdale Center for Developmental and Molecular Biology, Mt. Sinai School of Medicine, New York, NY): the cytological deficiency Df(3R)eD7, which deletes both tin and bap genes, had been recombined with a transgenic insertion of a 10.7-kb genomic BamHI fragment that reportedly rescues the tin mutant phenotype (Df(3R)eD7,P[tin-CasPeRe28]; see ref. 2).

Temperature Shift Treatments.

Embryos were collected on plates with shallow grape agar at 2-hr intervals at 18°C and aged at 18°C until the embryos reached the desired developmental stage. Staged embryo-containing plates were covered and submerged once in a water bath at 37.5°C for 30–40 min or twice for 20–25 min with 1 hr of incubation at 18°C in between. The embryos were then aged at 18°C until fixation. Ages indicated in the text were adjusted for a standard at 25°C (development was about half as fast at 18°C).

Chimeric Constructs.

tin:Nkx2–5 chimeric constructs were made as follows: a mouse Nkx2–5/Csx fragment [310 bp (ref. 11)] containing the homeodomain and the NK2-SD were inserted into the full-length tin cDNA in which the homeodomain 38-aa 5′ and 30-aa 3′ to the homeodomain were deleted [bp 1028–1459 (ref. 3)]. The zebrafish TN-homeodomain fragment was made with PCR exactly from the beginning of the TN-domain to the end of the homeodomain [bp 82–591 (ref. 6)] was inserted at the equivalent location in the tin cDNA (bp 393-1361). The chimeric cDNAs ware then inserted into the XbaI site of hsCasPeRWH1.

Immunocytochemistry.

Antibody staining and tissue in situ hybridization of whole-mount embryos were performed as described in ref. 1. Anti-Eve (17) was used at 1:10,000 and anti-FasIII (18) and an antibody that marks the differentiating pericardial cells (obtained from T. Volk, Weizmann Institute, Rehovot, Israel) were used at 1:10. Homozygous mutant embryos were identified by the lack of reporter gene expression present on the balancer chromosomes that were used (as in ref. 1).

RESULTS

We used hsp70 heat shock promoter constructs to drive expression of the following Nkx-type genes in flies (Fig. 1A): Drosophila tin itself (3), Xenopus and zebrafish Nkx2–3 (12, 13), mouse and zebrafish Nkx2–5 (6, 11), zebrafish Nkx2–7 (13), and Drosophila bap (2). Transgenic flies harboring these conditional expression constructs were recombined with a tin null mutation and assayed for restoration of heart and visceral mesoderm marker gene expression (1). All of the transgenes were expressed ubiquitously and at high levels after induction (data not shown). If tin expression is induced after gastrulation, but before the mesoderm subdivides, markers of heart (Fig. 1 B and C) and visceral mesoderm formation in tin mutant embryos (Fig. 1 E and F) are well restored. Similar results were obtained with an early (Eve; ref. 17) and a late cardiac marker gene (data not shown; see Materials and Methods). Similarly, not only the marker gene FasIII (18) is activated in the prospective visceral mesoderm, but also the characteristic palisade morphology of the forming visceral mesoderm epithelium seems to be restored (Fig. 1 E and F). This suggested that appropriate cardiac and visceral mesoderm cell types have, at least in part, been induced as a consequence of tin transgene expression during mid-embryogenesis (although a functional heart is not formed with this protocol of transgene induction). In contrast, if tin is induced at a later time, the tin mutant phenotype as assayed with Eve and FasIII is progressively less rescued (Fig. 1 D, G, and H). Thus, it seems that tin function is first required in the early mesoderm to allow the specification of heart and visceral mesoderm progenitor tissues (1, 2).

We used this experimental paradigm to examine the rescue capabilities of the tin-related genes of vertebrates (Fig. 1A). We first examined the Nkx2–5 gene, because it is primarily expressed in the early heart (but also in the anterior visceral endoderm) and because it has been shown to be, in part, necessary and sufficient for heart development (5–7, 9). When mouse Nkx2–5/Csx is induced at the optimal time for rescue (Fig. 1H), expression of the visceral mesoderm marker, FasIII, seemed to be significantly restored, but heart markers were not (Fig. 2 A–D). The same experiment carried out with Nkx2–5 derived from zebrafish also gave robust rescue of the visceral mesoderm marker but only minimal rescue of heart markers (data not shown). Moreover, analysis of several independent transgenic insertions or when two (instead of one) heat shocks were applied resulted in essentially the same observations: 60–80% rescue of visceral mesoderm marker gene expression but only minimal rescue of heart markers (<10%). Thus, Nkx2–5 is capable of efficiently initiating, directly or indirectly, visceral mesoderm- but not cardiac-specific gene expression. Although unlikely to be the sole reason, it is also possible that quantitative differences in transcriptional activation, mRNA, or protein stability are contributing to this differential activation of visceral mesoderm versus heart markers.

Figure 2.

Figure 2

Open in a new tab

Stage 12 embryos stained for FasIII (A, B, and E) and Eve (C, D, and F), compare with Fig. 1 B and E for wild-type patterns. (A–D) Induction of full-length MNkx2–5 in homozygous tin mutant background (HsMNkx2–5,tinEC40). Without heat shock (A and C) no visceral mesoderm (A, asterisks) nor cardiac progenitors (C, asterisks) form, which is typical of tin mutants. In contrast, heat shock at 3.5–4.5 hr of development (stage 9) restores marker gene expression for visceral mesoderm considerably (B, arrowheads), but does not restore heart development markers (D, asterisks, compare with Fig. 1 B and C). (E) Induction of full-length XNkx2–3 in homozygous tin mutant background (HsXNkx2–3,tinEC40) restores visceral mesoderm (arrowheads) but not heart marker gene expression (data not shown). The same result was obtained with ZNkx2–7 (data not shown). When two consecutive heat shocks were given to each of these three transgenes, they did not result in more rescue. (F) Induction of full-length ZNkx2–3 in homozygous tin mutant background (HsZNkx2–3,tinEC40) restores not only visceral mesoderm (data not shown) but also cardiac markers (arrowheads).

We then examined the rescue abilities of Nkx2–7 from zebrafish, because its expression pattern appears to prefigure the expression of Nkx2–5 in the heart and also that of Nkx2–3 in the anterior endoderm (13). As it is the case with the Nkx2–5 genes, Nkx2–7 induction in fly embryos only rescues efficiently marker gene expression in the presumptive visceral mesoderm (data not shown). Thus, Nkx2–5 and Nkx2–7 can substitute for tin with respect to activating visceral mesoderm markers but not (or only minimally) with respect to markers of heart development. This suggests that the direct target genes of tin required for cardiac development are likely to be distinct from those required for visceral mesoderm development, and perhaps the regulation of cardiac targets has diverged more extensively between vertebrates and invertebrates.

Next, we examined the rescue activities of Nkx2–3, which in Xenopus has an expression pattern similar to that of Nkx2–5 (12) and also produces enlarged hearts when overexpressed (5). As with Nkx2–5, expression of Xenopus Nkx2–3 (Xnkx2–3) activates visceral mesoderm- (Fig. 2E) but not cardiac-specific gene expression (data not shown). In contrast, however, Nkx2–3 from zebrafish (ZNkx2–3), whose normal expression domain includes primarily the presumptive anterior endoderm and does not seem to overlap extensively with the heart progenitors after gastrulation (13), can rescue not only visceral mesoderm markers (data not shown) but also cardiac-specific gene expression in Drosophila (Fig. 2F). Thus, it seems that all three of these tin-related Nkx genes have similar activities in promoting Drosophila gene expression specific for visceral mesoderm in place of tin, but only ZNkx2–3 has significant levels of both tin activities. Sequence comparisons between tin and ZNkx2–3 or all other known Nkx genes, however, did not reveal particularly striking similarities outside the TN and the homeodomain between tin and ZNkx2–3 (see Discussion).

Since all of the NK2-type genes examined thus far were capable of restoring the activation of visceral mesoderm markers, we wondered if NK2-type genes in general are functionally promiscuous with respect to visceral mesoderm-specific gene expression when present in the mesoderm at the time when tin is normally required. We chose to examine the rescue ability of bap because it is the closest relative of tin in Drosophila. When tested in a bap null mutant background, in which no visceral mesoderm forms (as in Fig. 3A; M. Frasch, personal communication), ubiquitous induction of transgenic bap activity by heat shock rescues visceral mesoderm marker gene expression considerably (Fig. 3 A and B). However, induction of bap expression in tin mutant embryos does not activate heart or visceral mesoderm-specific gene expression (Fig. 3 C and D). Thus, in contrast to the tin-related Nkx genes, bap is clearly not able to substitute for any of the tin functions during mesodermal subdivision. This is particularly remarkable, since tin, but not bap, seems to have lost its NK2-specific domain, and the homeodomains of the tin-related Nkx genes are as similar to those of bap as to those of tin (Fig. 1A). Thus, because only Nkx2–3, 2–5, and 2–7 but not bap exhibit (partial) tin-like functions in our Drosophila assay, it is suggested that this group of Nkx genes is evolutionarily more closely related to tin than to bap. This suggestion of a tin-related subclass of NK2-type genes is further supported by the fact that the vertebrate bap-related homeodomains are considerably more similar to that of bap than to that of tin (11, 14).

Figure 3.

Figure 3

Open in a new tab

Stage 12 embryos stained for FasIII (A–C) and Eve (D), compare with Fig. 1 B and E for wild-type patterns. (A and B) Induction of full-length bap in homozygous bap mutant background (Hsbap,Df(3R)eD7,P[tin-CasPeRe28]; see ref. 2). Without heat shock no visceral mesoderm forms (A, asterisks), which is typical of bap mutants. In contrast, heat shock at 3.5–4.5 hr of development (stage 9) restores visceral mesoderm marker considerably (B, arrowheads). (C and D) Induction of full-length bap (same transformant as in A and B) in homozygous tin mutant background (Hsbap,tinEC40) with a heat shock at stage 9 does not restore marker gene expression for either visceral mesoderm (C) or cardiac progenitors (D). The same results were obtained with other transgene insertions or when two heat shocks were applied. Tissue in situ hybridization after heat shock with a bap antisense RNA probe shows high and ubiquitous levels of bap expression (data not shown).

Since most of the tin-related Nkx genes do not rescue heart but do rescue visceral mesoderm markers, we wondered which structural differences between tin and these genes could account for this lack of cardiogenic activity. Obvious candidates are the NK2-specific domain (not present tin), the difference in length and sequence of the region from TN to homeodomain, and the difference in homeodomain sequence (see Fig. 1A). In a first attempt to distinguish between these possibilities, we made chimeric tin/Nkx2–5 cDNA constructs (Fig. 4A) and examined their ability to rescue cardiac markers in tin mutant embryos.

Figure 4.

Figure 4

Open in a new tab

(A) Schematic of the chimeric constructs between tin and Nkx2–5. In one construct, the tin homeodomain was replaced by the MNkx2–5 homeodomain and NK2-specific domain (tin:2.5HD-NK2SD). In the other construct, the region from the TN domain to the homeodomain of tin was replaced by the same region from ZNkx2–5 (tin:2.5TN-HD). (B and C) Immunocytochemical staining of a monoclonal antibody specific for pericardial cells of the heart at late stages of development (obtained from T. Volk). Wild-type (B) and Hstin:2.5HD-NK2SD,tinEC40 embryos heat shocked at 3.5–4.5 hr of development (stage 9, C). Note the presence of many pericardial cells (arrowheads) after induction of this chimeric tin gene, although in a disorganized pattern. Without induction, pericardial cells do not form at all (data not shown). The same result was obtained with antibodies against Eve. Visceral mesoderm marker is also restored in heat-shocked embryos of this genotype (data not shown). (D) Early heat shock (3.5–4.5 hr of development) of Hstin:2.5TN-HD,tinEC40 embryos restores visceral mesoderm marker (data not shown), but as with the full-length Nkx2–5 genes, heart markers are absent (asterisks), except for an occasional cell (arrowhead indicates Eve expressing cell). Eve expression in the central nervous system is not affected (arrow).

When the tin homeodomain (including some flanking sequences) was replaced by the Nkx2–5 homeo- and NK2-specific domain (Fig. 4A), heart-specific markers were significantly restored (Fig. 4 B and C). These data suggest that the homeodomains of tin and Nkx2–5 are functionally interchangeable in this Drosophila assay. This implies that the target specificity of the tin gene product for cardiac-specific gene expression is unlikely to be encoded exclusively by the homeodomain and flanking sequences (in contrast what seems to be the case for some of the hox genes). Consistent with such an interpretation is the finding that in vitro both tin and Nkx2–5 recognize the same consensus binding site (19; T. V. Venkatesh and R.B., unpublished data). In addition, the presence of the NK2-specific domain (in conjunction with the Nkx2–5 homeodomain) does not appear to be interfering with the cardiac rescue ability of the tin transgene.

We also replaced the region between (and including) the TN domain and the homeodomain of tin with the equivalent region of Nkx2–5 (Fig. 4A). When examined in transgenic tin mutant embryos, this chimeric protein rescues FasIII expression in the presumptive visceral mesoderm (data not shown), as the full-length Nkx2–5 genes do, but heart development markers were not appreciably restored (Fig. 4D). The cardiogenic activity of tin is likely to involve the region between the TN domain and the homeodomain. This region is much larger (255 amino acids) in tin than in the tin-related Nkx genes (100–135 amino acids) or in bap (145 amino acids; see Fig. 1A). Either tin contains a Drosophila heart-specific domain (not present or conserved in the other genes) or these vertebrate Nkx genes have less overall cardiogenic activity, simply because their TN to homeodomains are shorter than those of tin. These results support the idea that tin in insects or its ancestor has diverged more than the vertebrate tin-related genes from their postulated ancestor. It will be interesting to find out whether or not, in a converse experiment, Drosophila tin is capable of substituting for Nkx2–5 and restoring cardiac differentiation of Nkx2–5 mutant hearts (7).

DISCUSSION

The data we present here provide strong evidence that a subset of the NK2-type genes, the postulated tin-related Nkx genes, can substitute for tin function with respect to the restoration of markers of visceral mesoderm development but not with respect to those of heart development (with the exception of zebrafish Nkx2–3). This rescue activity seems to be specific to the tin-related subclass of Nkx genes, which excludes the Drosophila NK2-type gene bap and probably also its vertebrate counterparts. We propose that tin in flies has diverged significantly from its ancestor after the vertebrate/invertebrate split during evolution (thereby losing its NK2-SD). During this process, tin may have adopted a new fly- or insect-specific cardiac function, thereby changing its spectrum of interactions and targets. Alternatively, or in addition, the ancestor of the vertebrate tin-related genes may have lost some of its old cardiogenic functions and perhaps adopted additional functions during evolution. Consistent with this view is the finding that the different Nkx genes have little similarities between the TN and the homeodomain (e.g., ref. 13; the case of Nkx2–3 is discussed, see below). We presently do not know whether the tin-specific cardiogenic activity is encoded in a discrete domain N-terminal to homeodomain or whether this activity is distributed along much of the coding region.

Nkx2–3 of zebrafish (as tin of Drosophila) is capable of initiating cardiac-specific gene expression, but the other tin-related Nkx genes including Nkx2–3 from Xenopus are not. A possible reason for this observation could be that ZNkx2–3 has low levels of sequence similarities to tin that have initially not been obvious (13) and that are not present in the other genes. Indeed, there is a 54-aa stretch 5′ to the homeodomain of ZNkx2–3 (amino acids 83–136; ref. 13) that is 23% identical (with only one 2-aa gap) to a 56-aa region in between the TN and the homeodomain of tin (amino acids 200–255; ref. 3). The same region in any of the other Nkx genes does not show any similarity to tin. Further domain-swap experiments are needed to determine whether this region is of functional significance for heart-specific gene expression in the Drosophila assay. It is possible that the postulated common ancestor of vertebrate and insect tin-related genes already contained this 54/56-aa region, but because of the extremely low level of sequence identity, this remains speculative.

The fact that most vertebrate tin-related Nkx are expressed predominantly in the developing heart but rescue only visceral mesoderm markers in flies, and that ZNkx2–3 is not expressed in heart tissue but does rescue heart markers suggests that molecular mechanisms of organ development are interchangeable, within limits, between different organs during the course of evolution. It may be that tin in flies acts in heart and visceral organ development whereas its relatives in vertebrates have adopted distinct and regionally more localized functions in the development of either one of these organs. For example, XNkx2–3 may have adopted a function in heart development during recent vertebrate evolution, whereas its counterpart in zebrafish may have assumed a function in pharyngeal endoderm development.

With these considerations in mind, we propose that the spectrum of developmental functions that a tin-related gene may be able to assume is likely to be restricted to a limited set (e.g., heart and visceral mesoderm and endoderm), delineated by the spectrum of functions of the postulated ancestor. The fact that in Drosophila tin is still required for both heart and visceral mesoderm is consistent with this hypothesis: Although tin in Drosophila may have retained much of its developmental requirement for the spectrum of organs specified by the postulated ancestor, its amino acid sequence may have diverged significantly and adopted fly heart-specific functions that may be distinct from those of most of the tin relatives in vertebrate heart development.

Acknowledgments

We thank M. Frasch for the bap mutant fly stocks prior to publication, the bap cDNA and Even-skipped antibodies. We thank T. Volk for the antibody against against pericardial cells. M. Park was supported by a fellowship from the Organogenesis Center at the University of Michigan. This work was supported by grants from the National Institutes of Health (to R.B. and S.I.), and a grant from the American Heart Association (to R.B.). R.E.B. was supported by grants from the Boston Children’s Heart Foundation, the National Institutes of Health, and by an Established Investigatorship from the American Heart Association.

Note Added in Proof

The ability of Nkx2–5 to rescue mutant defects in Caenorhabditis elegans and Drosophila has recently been reported by Haun et al. (20) and Ranganayakulu et al. (21), respectively.

References

  • 1.Bodmer R. Development. 1993;118:719–729. doi: 10.1242/dev.118.3.719. [DOI] [PubMed] [Google Scholar]
  • 2.Azpiazu N, Frasch M. Genes Dev. 1993;7:1325–1340. doi: 10.1101/gad.7.7b.1325. [DOI] [PubMed] [Google Scholar]
  • 3.Bodmer R, Jan L Y, Jan Y-N. Development. 1990;110:661–669. doi: 10.1242/dev.110.3.661. [DOI] [PubMed] [Google Scholar]
  • 4.Bodmer R. Medicine (Baltimore) 1995;5:21–27. [Google Scholar]
  • 5.Cleaver O B, Patterson K P, Krieg P A. Development. 1996;122:3549–3556. doi: 10.1242/dev.122.11.3549. [DOI] [PubMed] [Google Scholar]
  • 6.Chen J-N, Fishman M C. Development. 1996;122:3809–3816. doi: 10.1242/dev.122.12.3809. [DOI] [PubMed] [Google Scholar]
  • 7.Lyons I, Parson L M, Hartley L, Li R, Andrews J E, Robb L, Harvey R. Genes Dev. 1995;9:1654–1666. doi: 10.1101/gad.9.13.1654. [DOI] [PubMed] [Google Scholar]
  • 8.Harvey R P. Dev Biol. 1996;178:203–216. doi: 10.1006/dbio.1996.0212. [DOI] [PubMed] [Google Scholar]
  • 9.Fu Y, Izumo I. Roux’s Arch Dev Biol. 1995;205:198–202. doi: 10.1007/BF00357766. [DOI] [PubMed] [Google Scholar]
  • 10.Newman, C. S. & Krieg, P. A. Dev. Genet. (Amsterdam)22, 230–239. [DOI] [PubMed]
  • 11.Komuro I, Izumo S. Proc Natl Acad Sci USA. 1993;90:8145–1949. doi: 10.1073/pnas.90.17.8145. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Evans S M, Yan W, Murillo M P, Ponce J, Papalopulu N. Development. 1995;121:3889–3899. doi: 10.1242/dev.121.11.3889. [DOI] [PubMed] [Google Scholar]
  • 13.Lee K-H, Xu Q, Breitbart R E. Dev Biol. 1996;180:722–731. doi: 10.1006/dbio.1996.0341. [DOI] [PubMed] [Google Scholar]
  • 14.Tanaka M, Kasahara H, Bartunkova S, Schinke M, Komuro I, Inagaki H, Lee Y, Lyons G E, Izumo S. Dev Genet (Amsterdam) 1998;22:239–249. doi: 10.1002/(SICI)1520-6408(1998)22:3<239::AID-DVG6>3.0.CO;2-7. [DOI] [PubMed] [Google Scholar]
  • 15.Newman C S, Grow M W, Cleaver O B, Chia F, Krieg P A. Dev Biol. 1997;181:223–233. doi: 10.1006/dbio.1996.8416. [DOI] [PubMed] [Google Scholar]
  • 16.Thummel C S, Boulet A M, Lipshitz H D. Gene. 1988;74:445–456. doi: 10.1016/0378-1119(88)90177-1. [DOI] [PubMed] [Google Scholar]
  • 17.Frasch M, Hoey T, Rushlow C, Doyle H, Levine M. EMBO J. 1987;6:749–759. doi: 10.1002/j.1460-2075.1987.tb04817.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Patel N, Snow P M, Goodman C S. Cell. 1987;48:975–988. doi: 10.1016/0092-8674(87)90706-9. [DOI] [PubMed] [Google Scholar]
  • 19.Chen C Y, Schwartz R J. J Biol Chem. 1995;27:15628–15633. doi: 10.1074/jbc.270.26.15628. [DOI] [PubMed] [Google Scholar]
  • 20.Haun C, Alexander J, Stainier D, Okkema P G. Proc Natl Acad Sci USA. 1998;95:5072–5075. doi: 10.1073/pnas.95.9.5072. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Ranganayakulu, G., Elliott, D. A., Harvey, R. P. & Olson, E. N. (1998) Development, in press. [DOI] [PubMed]

Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences

RESOURCES