Cloning of the human twist gene: Its expression is retained in adult mesodermally-derived tissues

Gene 187 (1997) 83–92

Cloning of the human twist gene: Its expression is retained in adult mesodermally-derived tissues Sherry M. Wang, Vincent W. Coljee, Robert J. Pignolo, Mitch O. Rotenberg, Vincent J. Cristofalo *, Felipe Sierra Center for Gerontological Research, Allegheny University of the Health Sciences, 2900 Queen Lane, Philadelphia, PA 19129, USA Received 3 June 1996; revised 2 September 1996; accepted 9 September 1996; Received by A. Bernardi

Abstract We have previously reported on the isolation of several differentially expressed genes derived from young and senescent non growing WI-38 human fetal lung-derived fibroblasts. A 0.8-kb cDNA clone, isolate EPC-A2 (early population doubling cDNA-A2), encodes the 3∞ end of the human homolog of the Twist protein. Twist genes encode basic helix-loop-helix DNA-binding transcription factors that play crucial roles in mesoderm development. Here, we report the cloning and sequencing of the genomic human twist gene. It encodes a protein of 201 amino acids with 96% amino acid sequence identity to mouse Twist, and 100% sequence conservation in the DNA-binding region among all species in which it has been characterized. We further show that expression of human twist is retained in mesodermally-derived tissues and cell lines derived from adult donors. Keywords: Transcription factor; Genomic DNA; Tissue-specific expression; Helix-loop-helix; Aging

1. Introduction Normal human diploid fibroblasts exhibit a limited proliferative capacity in culture (Hayflick and Moorhead, 1961; Hayflick, 1965), and this phenomenon has been interpreted as an expression of aging at the cellular level. Although the phenomenon of in vitro replicative senescence has been observed in a variety of cell types, human diploid fibroblasts are the most commonly used model system for cellular aging studies. While fibroblasts display a variety of morphological and physiological changes as a function of in vitro aging, little is known about how gene expression changes as a function of in vitro life span, and the role these changes may play in the development and/or maintenance of the senescent phenotype. In an attempt to search for genes whose expression and regulation are selectively altered during in vitro senescence of WI-38 cells, we have prepared subtracted libraries from growth arrested young and senescent WI-38 cells (Doggett et al., 1992). Here, we report the

identification of EPC-A2, a gene overexpressed in quiescent young versus senescent cells, as a human homolog of twist. Twist genes have previously been characterized in different species including Drosophila, Xenopus and mouse. They encode basic helix-loop-helix DNA-binding transcription factors that play crucial roles in mesoderm development. In this manuscript, we report the cloning, molecular characterization and pattern of expression of the Human twist (H-twist) gene. The high degree of conservation at both nucleotide and amino acid sequence levels among species, and its high expression in placenta and other mesodermally derived tissues and cell types, suggest a coherence with the role of twist genes in embryonic development, especially in mesoderm formation.

2. Results and discussion 2.1. Cloning and sequencing of the H-twist gene

* Corresponding author. Tel. +1 215 9918460; Fax +1 215 8431192. Abbreviations: ribonuclease.

b-HLH, basic

helix-loop-helix

RNase;

RNase,

Previous work from our laboratory, based on subtractive hybridization of non-growing young and senescent cells has allowed the identification of several genes which

0378-1119/97/$17.00 Copyright © 1997 Elsevier Science B.V. All rights reserved PII S 03 7 8 -1 1 1 9 ( 9 6 ) 0 0 7 27 - 5

84

S.M. Wang et al. / Gene 187 (1997) 83–92

Fig. 1. Differential expression of H-twist in young and senescent WI-38 cells. Young [Population Doubling Level (PDL) 18] and senescent (PDL 62) WI-38 cells were grown for 7 days in Autopow MEM supplemented with BME vitamins (Flow Laboratories McLane, VA) and 10% FBS, as described by Cristofalo and Charpentier (1980), and then serum-starved in serum-free medium (MCDB-104, Gibco BRL) for 3 days. Twenty-five mg of total RNA were used for Northern blot analysis. Hybridization was performed with random primed 32P-labeled 0.8-kb H-twist cDNA. b -microglobulin was used as a control for RNA 2 loading and quantitation.

are differentially expressed during in vitro senescence of human diploid fibroblasts (Doggett et al., 1992). A 0.8-kb cDNA clone was isolated from the young specific library, and used as a probe for further characterization by Northern analysis. A low abundance mRNA of 1.6 kb was detected and found to be overexpressed in young quiescent cells as compared to senescent cells (Fig. 1). Sequence analysis of this partial clone indicated that it was derived from the 3∞ end of its cognate mRNA,

and showed remarkably high sequence homology with members of the twist gene family. Using a full-length mouse twist ( Wolf et al., 1991) cDNA to screen a genomic l library, we identified three positive clones from a screen of 6×105 plaques. From these, we further subcloned and characterized a 9.4-kb EcoRI fragment, which was present in all three clones. Restriction endonuclease analysis of this fragment allowed the identification of a 2.9-kb NcoI fragment which hybridized to the original 0.8-kb clone. Fig. 2A shows the results of further restriction endonuclease mapping with various enzymes, followed by Southern blot analysis. These studies led to a more complete characterization of this genomic fragment that contains the complete H-twist gene. The 2.9-kb NcoI fragment was then subcloned into pBluescript (SK ) and used for further sequencing, using partial Exo III deletions. Analysis and comparison of the cDNA and genomic sequences revealed the following characteristics of the gene: the first ATG (+317) is followed by an uninterrupted open reading frame of 603 nucleotides that ends at +920 with a TAG codon, coding therefore for a 201 amino acid protein ( Fig. 2B). One intron of 536 nucleotides was found at position +965 to +1500, downstream of the stop codon. The 3∞-end of the gene reveals two potential polyadenylation signals at positions +1565 and +1915. Two putative TATA boxes are present at positions −32 and −110, but analysis of the transcription initiation site using both RNase mapping ( Fig. 3A) and S nuclease digestion ( Fig. 3B) indicates 1 that only the former one is functional. Transcriptional initiation occurs at a single Adenosine (A) residue (position +1), 32 bp downstream of the proximal TATA box ( TATAA). The deduced protein encoded by this clone has a calculated molecular weight of 20 909 and appears to be basic, with a theoretical pI of 9.591. The carboxy terminal portion of the protein (from amino acids 108 to 164) contains a basic DNA binding and dimerization motif common to the ‘b-HLH proteins’ (Murre et al., 1989; Benezra et al., 1990). The protein is mostly hydrophilic in the region close to the NH -terminus, 2

Fig. 2. (A) Schematic map of an EcoRI genomic H-twist clone. A 9.4-kb EcoRI clone, obtained from the original l library, was analyzed for the location of the H-twist gene by restriction endonuclease mapping and subsequent Southern blot analysis. A 2.9-kb NcoI fragment was found to contain the H-twist sequence, and was mapped in further detail. (B) Nucleotide sequence of genomic H-twist and deduced amino acid sequence of the encoded protein. Nucleotides are numbered on the bottom. An asterisk and an underline denote the position of the transcription initiation site (+1). Arrows indicate the position of the intron (+965 to +1500). Two potential TATA boxes (−32, −110), and two potential polyadenylation sites (+1565, +1915) are shown in bold. The translation start and stop codons are bolded and underlined. Methods: The original 0.8-kb cDNA was isolated from a young G subtracted library of WI-38 cells (Doggett et al., 1992). Using a full-length M-twist cDNA ( Wolf et al., 1991), we 0 screened a genomic l FIX IIA library (from Stratagene) derived from human diploid fibroblast WI-38 cells. Three positive clones were obtained from a screen of 6×105 plaques. From these, a 9.4-kb EcoRI fragment was selected, and subcloned into pBluescript (SK ) vector for further examination. Southern blot analysis of restriction digest fragments, using 32P-labeled oligo nucleotide probes derived from the 0.8-kb H-twist cDNA clone, indicated that a 2.9-kb NcoI clone may contain the complete H-twist gene. This NcoI fragment was therefore subcloned into a pBluescript (SK ) for sequence analysis. Sequencing was performed by the dideoxy method (Sanger et al., 1977), using a 7-deaza-dGTP Sequenase 2.0 kit (from USB) on double-stranded DNA templates prepared by limited exonuclease III digestion (Henikoff, 1987). Terminal deoxytransferase was also used for parallel sequencing to solve sequencing compressions, which occurred mostly at the 5∞ end of the gene.

S.M. Wang et al. / Gene 187 (1997) 83–92

85

86

S.M. Wang et al. / Gene 187 (1997) 83–92

Fig. 2. Cont.

and appears to be more hydrophobic at the C-terminus where the helix-loop-helix domain is located (data not shown). A computer search of the complete nucleotide sequence of the 2.9-kb genomic human twist segment showed extensive homology with genes of the twist family in both the coding and non-coding regions (92% nucleotide sequence identity between mouse and human in the coding region). The deduced H-Twist protein shares 96% amino acid sequence identity to mouse

Twist. Therefore, we believe this is a true member of the twist family, and we have named it human twist (Htwist). Furthermore, mouse and human twist share 71% nucleotide sequence identity in their promoter region (up to approximately 300 bp upstream from the cap site identified in the human gene). The putative promoter sequences upstream of the transcription initiation site (up to −824 bp) were screened with Macvector for DNA consensus sequence

87

S.M. Wang et al. / Gene 187 (1997) 83–92

A

B

Fig. 3. Determination of the H-twist transcription initiation site. (A) Precise mapping of the transcription initiation site by RNAse protection. Lanes A, G, C and T represent dideoxynucleotide chain termination sequencing reaction with the primer complementary to H-twist from +98 to +79. A portion of the sequence of the sense strand is shown with the transcriptional start site indicated by a star. (B) S mapping of the 5∞ end 1 of the transcript. Methods: A 471-bp Fok I-digested fragment (from −377 to +94 of the genomic sequence) was first subcloned into the SmaI site of pBluescript (SK ). The plasmid DNA was then linearized with EcoRI, and an antisense RNA was synthesized with T RNA polymerase (Promega) 3 and [a-32P]UTP (Amersham) as described by Melton et al. (1984). RNase mapping was performed by using an RPA II kit from Ambion. 25 or 50 mg of total RNA isolated from IMR-90 cells was hybridized to 32P-labeled antisense RNA and digested with 0.5 units RNase A and 20 units RNase T at 37°C, for 30 min. The protected fragments were subjected to electrophoresis on an 8% polyacrylamide/7 M urea gel. A DNA sequencing 1 ladder was generated by using the primer 5∞-GGTCCCAAAAAGAAAGCGCC-3∞ (DNA International ), complementary to nucleotides +98 to +79 of the H-twist gene. For S1 mapping, this same primer was 5∞ end labeled with [c-32P]ATP (NEN ) using T4 polynucleotide kinase (Promega), and annealed to an EcoRI linearized and denatured double-strand plasmid DNA, containing the 471-bp H-twist sequence from −377 to +94. The end labeled oligo probe was then extended using the Klenow fragment of E. coli DNA polymerase I (Promega). The newly synthesized single strand, complementary to the mRNA, was purified from a 5% polyacrylamide/8 M urea gel. 5×104 cpm of the labeled probe were hybridized overnight at 42°C to 30 mg total RNA isolated from IMR-90 cells, or to 30 mg of yeast tRNA as a control. The hybridization mixture was then digested with 100 units of S nuclease (Ambion) at 42°C for 16 h. The S protected fragments were electrophoresed on an 8% polyacrylamide/7 M 1 1 urea gel.

88

S.M. Wang et al. / Gene 187 (1997) 83–92

A

B

Fig. 4. (A) Schematic map of the H-twist nucleotide sequences upstream of the transcription initiation site, showing consensus sequence motifs for known transcription factors. Consensus sequence motifs of known transcription factors were searched in the promoter region of H-twist using the Macvector program (Macvector). (B) Alignment of promoter sequences from mouse and human twist genes. The available sequences from mouse (above) and human (below) promoter sequences were aligned relative to the known transcription initiation site from the human sequences. The consensus sites capable of binding known transcription factors are underlined. Fig. 5. (A) Alignment of human, mouse, Xenopus and Drosophila twist amino acid sequences. The conserved residues are boxed, and the domains of DNA-binding as well as the Helix-Loop-Helix regions are indicated. (B) Schematic representation of putative motifs present in the H-twist protein. The predicted H-twist protein contains six potential PKC phosphorylation sites (P), two potential cAMP phosphorylation sites (C ), one potential Tyr phosphorylation site ( T ), and a Helix-Loop-Helix domain (HLH ). The analysis of the putative motifs present in H-twist protein was performed using software obtained through the University of Wisconsin Genetics Computer Group ( UWGCG).

S.M. Wang et al. / Gene 187 (1997) 83–92

A

B

89

90

S.M. Wang et al. / Gene 187 (1997) 83–92

motifs to which known regulatory proteins might bind (Fig. 4). Many consensus sequences for known transcription factors were found in this region. Of interest, we found potential binding sites for NF-kB (GGGRHTYYHC at position −89 to −99) and engrailed (HCWATHAAA at position −112 to −104). These sites are 100% conserved in the corresponding region of the mouse Twist gene ( Fig. 4). Other consensus sequences, such as those that bind AP-2, Sp1, Rb and ATF/CREB are all present in the 5∞ regulatory region of the H-twist gene. 2.2. Twist proteins are highly conserved among Drosophila, Xenopus, mouse and human The sequence of the Twist proteins from the four species analyzed thus far are aligned and shown in Fig. 5A. The amino acid sequences of the twist proteins consist of 201 (human), 206 (mouse), 166 (Xenopus), and 490 (Drosophila) residues. Regions of conserved amino acid sequences appear scattered throughout the proteins. The best alignment was observed in the basic region of the DNA-binding domain, which is 100% conserved among all species studied. This suggests that they may all bind to the same DNA sequence, called the E-box, and which is found in many cell type-specific genes (Lenardo and Baltimore, 1989; Steward et al., 1988; Ghosh et al., 1990; Kieran et al., 1990; Thisse et al., 1987, 1991). The Helix-Loop-Helix region of the protein dimerization domain shows 68% identity among all four species, whereas the ‘Loop’ region shows 100% identity. The carboxy terminal region, whose sequences are believed to be specific to ‘Twist family’ members ( Wolf et al., 1991), is 100% conserved among Xenopus, mouse, and human. Taken together, these results suggest that the gene we have characterized is the true human homolog of the Twist proteins described in other species. An analysis of the motif sequences of the predicted H-Twist protein was performed by using the GCG program, and several potential phosphorylation sites were found (Fig. 5B). Six putative PKC phosphorylation sites were observed in H-Twist, four of which are conserved among Xenopus, mouse and human. Two of these lie within the loop region of the protein dimerization domain, and are conserved among all species. Since it has been found that Twist proteins are highly conserved from Drosophila to humans ( Thisse et al., 1988; Hopwood et al., 1989; Wolf et al., 1991; this work), we wanted to further test this high level of phylogenic conservation. For this, we probed a Southern blot with a random primed 32P-labeled 0.8-kb H-twist cDNA fragment to detect the twist homolog from a variety of species. Twist-related sequences were detected in all the mammalian species tested. A strong signal was also detected in chicken, but not in yeast ( Fig. 6). If related sequences are present in Saccharomyces cerevis-

Fig. 6. Detection of twist related sequences in various species. Genomic DNA from the indicated species was digested with EcoRI (from Clontech), and subjected to Southern blot analysis. Hybridization was performed by using random primed 32P-labeled 0.8-kb H-twist cDNA as a probe and washed under low stringency conditions according to the manufacturer’s instructions.

iae, they may be divergent enough that no cross-hybridization was observed even at the low stringency used. 2.3. H-twist expression is tissue and cell type specific The pattern of expression of twist genes during embryonic development has been well studied in Drosophila, Xenopus, and mouse. Since human early developmental tissues are difficult to obtain, we have used various adult human tissues as well as several cell lines originated from different normal human tissues to examine the expression pattern of H-twist. Among the human tissues tested, the strongest expression of the 1.6-kb H-twist mRNA was observed in placenta, a tissue known to be a fetomaternal organ which contains two components: a large fetal portion and a small maternal portion. The fetal portion of the placenta develops from the chorionic sac, which is mostly derived from the mesoderm. Signals of intermediate strength were detected in adult heart and skeletal muscle, which are also mostly derived from mesoderm. In skeletal muscle, two cross-reacting mRNA species were observed, one is 1.6 kb, the other is much smaller. The pattern is not reminiscent of mRNA degradation, and probably reflects the presence of a true

S.M. Wang et al. / Gene 187 (1997) 83–92

A

91

B

Fig. 7. Tissue and cell type specific expression of H-twist. Northern blot analysis of (A) Two mg of poly A+RNA isolated from adult human tissues (from Clontech) and (B) 30 mg of total RNA derived from various cell lines of different origins. Methods: Total cellular RNA from various cell lines was isolated by guanidinium isothiocyanate extraction (Chirgwin et al., 1979), and purified by either CsCl ultracentrifugation (Glisin et al., 1974) or phenol extraction (Chomczynski and Sacchi, 1987). Total RNA was then electrophoretically fractionated on glyoxal gels (McMaster and Carmichael, 1977) and transferred to GeneScreen membranes (Dupont, NEN Research Products, Boston, MA). RNA was fixed by exposure to UV light and hybridized to the random primed 32P-labeled 0.8-kb H-twist cDNA fragment. Hybridization was performed by using random primed 32P-labeled 0.8-kb H-twist cDNA as a probe in 50% formamide at 42°C. Cell lines used for this study include AG 7086 (peritoneal mesothelial cells), AG2102 (endometrial fibroblasts), AG5337 (retinal pigmented epithelial cells), and AG11137 (mammary epithelial cells). All of these cell lines were obtained from the National Institute on Aging Cell Repository (Coriell Institute for Medical Research, Camden, NJ ). Human fetal astrocytes were a generous gift from M.D. West (Geron Corp.). T lymphocytes were from male donors between 24 and 28 years of age, isolated as described by Murasko et al. (1991). All of these cell lines were grown to confluency to achieve density-dependent growth arrest, and all the cultures had completed less than 50% of their in vitro replicative life span at the time of usage.

cross-hybridizing species. At present, we do not know the significance of this observation. Weak signals were found in kidney and pancreas, while no signal was observed in brain, which is ectoderm-derived, or lung and liver, which are primarily derived from the endoderm (Fig. 7A). These results indicate that in adult humans, twist is expressed preferentially in mesodermally derived tissues. We further examined H-twist expression in various cell types derived from different human tissues. H-twist was found to be expressed in WI-38 cells (fetal lungderived fibroblasts), as well as human peritoneal mesothelial cells and endometrial fibroblasts (both of which represent mesodermal cells derived from young adults) (Fig. 7B). However, H-twist expression was not detected in T lymphocytes, primary cells derived from a tissue of mesoderm origin. This is not surprising since both Xenopus and mouse twist have already been shown to be expressed only in a subset of mesodermal tissues (Hopwood et al., 1989; Wolf et al., 1991). No H-twist expression was observed in human fetal astrocytes (an

ectodermally derived cell line), or in either of the two epithelial cells tested, retinal pigmented epithelial and mammary epithelial cells (both are ectodermally derived cell lines) ( Fig. 7B). H-twist mRNA has also been detected in both fetal and adult human skin fibroblasts [ Wang et al., unpublished data].

3. Conclusions (1) We have cloned and molecularly characterized the human homolog of the Twist gene (H-twist), including the determination of the nucleotide sequence and the transcription initiation site of the gene. (2) The H-twist protein (201 amino acids) shares 96% amino acid sequence identity with mouse twist. The DNA-binding region of the twist proteins are 100% conserved among the four species analyzed thus far. (3) H-twist expression is tissue and cell type specific, it is preferentially expressed in mesoderm derived cell lines and tissues.

92

S.M. Wang et al. / Gene 187 (1997) 83–92

Acknowledgement We wish to thank Dr. Marie A. DiBerardino for her helpful discussion and comments. This work was supported by NIH grant AG 00378 to V.J.C.

References Benezra, R., Davis, R.L., Lockshon, D., Turner, D.L. and Weintraub, H. (1990) The protein Id: A negative regulator of helix-loop-helix DNA binding protein. Cell 61, 49–59. Chirgwin, J.M., Przybyla, A.E., MacDonald, R.J. and Rutter, W.J. (1979) Isolation of biologically active ribonucleic acid from sources enriched in ribonuclease. Biochemistry 18, 5294–5299. Chomczynski, P. and Sacchi, N. (1987) Single-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction. Anal. Biochem. 162, 156–159. Cristofalo, V.J. and Charpentier, R. (1980) A standard procedure for cultivating human diploid fibroblastlike cells to study cellular aging. J. Tissue Culture Methods 6, 117–121. Doggett, D.L., Rotenberg, M.O., Pignolo, R.J., Phillips, P.D. and Cristofalo, V.J. (1992) Differential gene expression between young and senescent quiescent WI-38 cells. Mech. Aging Dev. 65, 239–255. Ghosh, S., Gifford, A.M., Riviere, L.R., Tempst, P., Nolan, G.P. and Baltimore, D. (1990) Cloning of the p50 DNA binding subunit of NF-kB: Homology to rel and dorsal. Cell 62, 1019–1029. Glisin, V., CrKvenjaKov, R. and Byers, C. (1974) Ribonucleic acid isolated by cesium chloride centrifugation. Biochem 13, 2633–2637. Hayflick, L. and Moorhead, P.S. (1961) The serial cultivation of human diploid cell strains. Exp. Cell Res. 25, 585–621. Hayflick, L. (1965) The limited in vitro lifetime of human diploid cell strains. Exp. Cell Res. 37, 614–636. Henikoff, S. (1987) Unidirectional digestion with exonuclease III in DNA sequence analysis. Meth. Enzymol. 155, 156–165. Hopwood, N.D., Pluck, A. and Gurdon, J.B. (1989) A Xenopus mRNA related to Drosophila twist is expressed in response to induction in the mesoderm and the neural crest. Cell 59, 893–903. Kieran, M., Blank, V., Logeat, F., Vandekerckhove, J., Lottspeich, F., Le Bail, O., Urban, M.B., Kourilssky, P., Baeuerle, P.A. and

Isael, A. (1990) The DNA binding subunit of NF-kB is identical to factor KBF1 and homologous to the rel oncogene product. Cell 62, 1007–1018. Lenardo, M.J. and Baltimore, D. (1989) NF-kB: a pleiotropic mediator of inducible and tissue-specific gene control. Cell 58, 227–229. McMaster, G.K. and Carmichael, G.G. (1977) Analysis of single- and double-stranded nucleic acids on polyacrylamide and agarose gels by using glyoxal and acridine orange. Proc. Natl. Acad. Sci. USA 74, 4835–4838. Melton, D.A., Krieg, P.A., Rebagliati, M.R., Maniatis, T., Zinn, K. and Green, M.R. (1984) Efficient in vitro synthesis of biologically active RNA and RNA hybridization probes from plasmids containing a bacteriophage SP6 promoter. Nucleic Acid Res. 12, 7035–7056. Murasko, D.M., Nelson, B.J., Matour, D., Goonewardene, I.M. and Kaye, D. (1991) Heterogeneity of changes in lymphoproliferation ability with increasing age. Exp. Gerontol. 26, 269–279. Murre, C., McCaw, P.S., Vaessin, H., Caudy, M., Jan, Y.N., Cabrera, C.V., Buskin, J.N., Haushka, S.D., Lassar, A.B., Weintraub, H. and Baltimore, D. (1989) Interactions between heterologous helix-loophelix proteins generate a complex that binds specifically to a common DNA sequence. Cell 58, 537–544. Sanger, F., Nicklen, S. and Coulson, A.R. (1977) DNA sequencing with chain-terminating inhibitors. Proc. Natl. Acad. Sci. USA 74, 5463–5467. Steward, R., Zusman, S.B., Huang, L.H. and Schedl, P. (1988) The dorsal protein is distributed in a gradient in early Drosophila embryos. Cell 55, 487–495. Thisse, B., Stoetzel, C., Messal, M.E.I. and Perrin-Schmitt, F. (1987) Genes of the Drosophila meternal dorsal group control the specific expression of the zygotic gene twist in presumptive mesodermal cells. Genes Dev. 1, 709–715. Thisse, B., Stoetzel, C., Gorostiza-Thisse, C. and Perrin-Schmitt, F. (1988) Sequence of the twist gene and nuclear localization of its protein in endomesodermal cells of early Drosophila embryos. EMBO J. 7, 2175–2183. Thisse, C., Perrin-Schmit, F., Stoetzel, C. and Thisse, B. (1991) Sequence-specific transactivation of the Drosophila twist gene by the dorsal gene product. Cell 65, 1191–1201. Wolf, C., Thisse, C., Stoetzel, C., Thisse, B., Cerlinger, P. and PerrinSchmitt, F. (1991) The M-twist gene of Mus is expressed in subsets of mesodermal cells and is closely related to the Xenopus X-twi and the Drosophila twist genes. Dev. Biol. 143, 363–373.