Tbx6 is a member of the T-box gene family. Studies of knockout mice indicate that Tbx6 is involved in somite differentiation. In the present study, we cloned Tbx6 from another vertebrate species, namely Xenopus laevis, and studied its roles in development. The expression of Tbx6 in Xenopus started from the early gastrula stage, reached a peak during the late gastrula to neurula stages and then declined. Initial expression of Tbx6 was observed in the paraxial mesoderm during the gastrula stage. The Tbx6-expressing region spread anteriorly and ventrally in the neurula stage. In the tailbud stage, the area of expression shrank caudally and was finally restricted to the tip of the tailbud. Overexpression of Tbx6 mRNA in dorsal blastomeres caused atrophy of the neural tube and inhibited differentiation of the notochord. Animal cap explants overexpressing Tbx6 or Tbx6VP16 mRNA, but not Tbx6EnR mRNA, differentiated mainly into ventral mesodermal tissues. This suggests that Tbx6 is a transcriptional activator. Higher doses of Tbx6 or Tbx6VP16 mRNA caused hardly any muscular differentiation. However, coinjection of Tbx6 mRNA with noggin mRNA elicited marked muscle differentiation. These results suggest that Tbx6 is implicated in ventral mesoderm specification but is involved in muscle differentiation when acting together with the dorsalizing factor noggin.
Mutant animals that have deficiencies in certain organs have been widely analyzed to unravel the mechanisms of animal development. One example is the mouse T (Brachyury) mutant. Heterozygous mutant mice (T/+ or t/+) have a short tail or no tail (reviewed by Goodfellow 1990). In homozygous mice (T/T), cell migration becomes abnormal and mesodermal cells decrease (Yanagisawa et al. 1981), the notochord is almost absent (Chesley 1935; Grüneberg 1958) and somites are much reduced (Chesley 1935). The T gene has been cloned (Herrmann et al. 1990) and was shown to be expressed first in the node and primitive streak and later in the notochord (Wilkinson et al. 1990). The orthologous genes were subsequently cloned from Xenopus (Xbra) and zebrafish (Zf-T;Smith et al. 1991; Schulte-Merker et al. 1992, 1994). Studies of these vertebrate T genes have suggested that the T gene has a conserved expression pattern among vertebrates and that its function is concerned with notochord and somite formation (reviewed by Smith 1999). However, overexpression of Xbra mRNA does not induce the notochord in Xenopus animal caps (Cunliffe & Smith 1992). When overexpressed together with cooperative factors, such as noggin (Cunliffe & Smith 1994) or pintallavis (O’Reilly et al. 1995), Xbra does induce the notochord in animal caps.
The T protein is composed of two domains, a conserved motif called the T-box (Bollag et al. 1994), which has DNA-binding activity, and a transcriptional activation domain (Kispert et al. 1995). In the mouse, many T-box-containing genes, including Tbx1–6 (Agulnik et al. 1996), have been cloned. One such gene, Tbx6, is expressed in paraxial mesoderm in the gastrula and, later, in the tailbud (Chapman et al. 1996). In zebrafish, Tbx6 is expressed in a similar fashion and is transcribed in response to mesoderm induction by basic fibroblast growth factor (bFGF) or activin (Hug et al. 1997). In Tbx6-knockout mice, somitomeres are transformed into neural-tube-like structures, which indicates that Tbx6 is necessary for normal somite differentiation (Chapman & Papaioannou 1998). However, little is known about the mechanism by which Tbx6 is involved in somite formation.
Noggin has been cloned from a cDNA library of LiCl-dorsalized Xenopus embryos as a gene that can rescue ultraviolet (UV)-ventralized embryos (Smith & Harland 1992). In contrast, bone morphogenetic protein-4 (BMP-4) has ventralizing activity when overexpressed in dorsal regions of Xenopus embryos (Köster et al. 1991; Dale et al. 1992; Jones et al. 1992). Noggin turned out to be a binding protein for BMP (Zimmerman et al. 1996) and the balance between BMP and noggin is a key factor in dorsoventral patterning of mesoderm in early Xenopus embryos (Dosch et al. 1997). In the chick, when a noggin-expressing cell line is implanted near the somite, the myo-D-expressing area expands and supernumerary segmented somites occur (Hirsinger et al. 1997; Reshef et al. 1998; Tonegawa & Takahashi 1998). These results suggest that the balance between BMP-4 and noggin is also important in mesodermal patterning in the chick embryo.
We isolated Tbx6 cDNA from Xenopus laevis embryos and tested its muscle-inducing activity. Animal caps overexpressing Tbx6 mRNA differentiated into ventral mesodermal tissues, but not muscle. However, when noggin was coexpressed with Tbx6, the animal caps differentiated into well-developed muscle. These results suggest that Tbx6 acts together with a dorsalizing factor, such as noggin, in Xenopus somite formation.
Materials and Methods
Injection and histology
Fertilized eggs of Xenopus laevis were obtained from a pair of adults that were injected with 750 IU of human chorionic gonadotropin into the dorsal lymph sac. Embryos were dejellied with 3% cysteine in Steinberg’s solution (pH 7.8), washed and cultured in Steinberg’s solution containing 50 mg/L kanamycin sulfate and staged according to Nieuwkoop & Faber (1956). Synthetic mRNA was injected into two blastomeres (5 nL per blastomere) at the two- to four-cell stage either near the animal pole or at the marginal zone. When mRNA was injected near the animal pole, the animal caps were excised at stage 8 with tungsten needles, transferred to normal amphibian medium (NAM)/2 (Godsave et al. 1988) and cultured at 20°C. Non-injected animal caps at stage 8 were also excised and cultured at 20°C in NAM/2 containing human recombinant activin (gift from Dr Y. Eto, Ajinomoto Co., Kawasaki, Japan) or human recombinant bFGF (Progen, Heidelberg, Germany). For dorsal or ventral injections, embryos were reared at 14°C. Animal caps or embryos were fixed in Bouin’s fluid for 30 min, passed through an ethanol series and xylene, and embedded in Histoprep (Wako Chemicals, Osaka, Japan). Serial sections (7 μm) were cut and stained with Mayer’s hematoxylin and eosin.
In vitro transcription
Coding sequences of Tbx6, Tbx6EnR, Tbx6VP16 and Antipodean were polymerase chain reaction (PCR) amplified (see below), cloned into plasmid pCS2+ and digested with NotI. Noggin Δ5′ in plasmid pGEM5Zf(–) (Smith & Harland 1992) was digested with NotI. Complete Xbra cDNA in pSP73 (pXT1; Smith et al. 1991) was linearized with HincII to remove approximately 600 bp of the 3′-untranslated region (UTR). These templates were extracted with phenol/chloroform and precipitated with ethanol. The mRNA was synthesized at 37°C for 2 h in a 25 μL reaction mixture containing 1 μg template, 12 mM MgCl2, 10 mM dithiothreitol (DTT), 2 mM ATP, 2 mM cytidine triphosphate (CTP), 2 mM thymidine triphosphate (TTP), 0.5 mM guanine triphosphate (GTP), 1.5 mM Cap Analog (G5′ppp5′G; New England Biolabs, Beverly, MA, USA), 0.04% acetylated BSA (Gibco-BRL, Rockville, MD, USA), 40 mM Tris (pH 8.0), 2 mM spermidine, 25 mM NaCl, 1 μL SP6 polymerase (TaKaRa, Kyoto, Japan), 0.5 μL placental RNasein (Toyobo, Osaka, Japan) and diethyl pyrocarbonate (DEPC)–water. Following digestion by 20 U DNase I for 15 min at 37°C, RNA was extracted with phenol/chloroform and precipitated in 2 M ammonium acetate and 70% ethanol. The RNA was dissolved in DEPC–water, determined by A260 and stored at –80°C. The mRNA was diluted with autoclaved and filtrated MilliQ water (Millipore Inc., Bedford, MA, USA) and used for microinjection. Bone morphogenetic protein-4 in pUC19 (Xbr23; Nishimatsu et al. 1992) was excised with EcoRI, subcloned into pBluescript II KS(–) and digested with BamHI. The mRNA was synthesized as above except that T3 polymerase (Stratagene, La Jolla, CA, USA) and 5 mM DTT were used.
RNA isolation and cDNA synthesis
The RNA was extracted from embryos or animal caps according to the methods of Evans & Kay (1991), except that proteinase K digestion was omitted. Following overnight LiCl precipitation, 10 μg total RNA was digested for 1 h at 37°C in 25 μL solution composed of 10 U DNase I (Gibco-BRL), 0.5 μL RNasein, 1 × transcription buffer (Gibco-BRL), 2 mM DTT and DEPC–water. Following extraction with phenol/chloroform, RNA was precipitated with ethanol. The RNA (2 μg) was mixed with 10 pmol oligo-dT primer and heat denatured. The cDNA was synthesized by using avian myeloblastosis (AMV) reverse transcriptase XL (Life Sciences, St Petersburg, FL, USA) according to the manufacturer’s instructions.
Amplification of the Tbx6 fragment
Degenerate primers encoding all possible codons for amino acid sequences Y(I/V)HPDSP (forward primer; 5′-CGAATTCTAYRTXCAYCCIGAIWSXCC-3′) and VTAYQN (reverse primer; 5′-CCTGCAGRTTYTGRTAXGCXGTXAC-3′) were synthesized (Amersham Pharmacia Biotech, Shinjuku, Tokyo, Japan), where X = ATGC, R = AG, Y = CT, W = AT, S = CG and I = inosine. Polymerase chain reaction amplification was performed with gastrula cDNA and AmpliTaq Gold polymerase (Perkin Elmer, Branchburg, NJ, USA). The PCR program was 94°C for 9 min, 40 cycles of 94°C for 30 s, 45°C for 30 s and 72°C for 30 s, 72°C for 10 min and storage at 4°C. Amplified products were precipitated with ethanol and digested thoroughly with AlwNI, BclI and NsiI. The T-vector was produced from pBluescript II KS(–). Ten micrograms of the plasmid was digested with EcoRV, precipitated with ethanol and incubated in 200 μL reaction mixture containing 2 mM dTTP, 10 U Taq polymerase (Boehringer Mannheim, Mannheim, Germany) and 1 × buffer for 2 h at 70°C. The reaction was extracted with phenol/chloroform and precipitated with ethanol. The T-box fragments of approximately 250 bp were gel-purified and subcloned into the above T-vector and were ligated with the TaKaRa ligation kit version 2. The DNA sequence was determined using the BigDye terminator kit and 310 Genetic Analyzer (ABI, Foster City, CA, USA), and analyzed by GeneWorks 2.5.1 software (Oxford Molecular Group, Oxford, UK). One clone had a sequence resembling that of mouse Tbx6.
5′ Rapid amplification of cDNA ends of Tbx6 and screening of cDNA library
The cDNA was synthesized from stage 13 total RNA by using a 5′-phosphorylated gene-specific primer (5′-pCGTCGGCTGAA-3′) and AMV reverse transcriptase XL, and purified with glassmilk (Bio101, Vista, CA, USA). Following ligation using T4 RNA ligase (New England Biolaboratories), circularized cDNA was used as a template for two rounds of inverse PCR with LaTaq (TaKaRa). In the first PCR (forward primer, 5′-CACTCCATGCACAAGTACCAGC-3′; reverse primer, 5′-TCTGGTCCAGAGTGTTGTRTGGTC-3′), cDNA was amplified for 30 cycles. The reaction was diluted 1 : 50 for the second PCR (forward primer, 5′- GCTTCCATATTGTCCGTGCC-3′; reverse primer, 5′-TGATCTTGTGGAAGGAAATGGG-3′), which ran for 30 cycles. An amplified fragment of approximately 800 bp was ligated into the T-vector, from which approximately 600 bp was excised with BamHI and labeled with [32P]-dCTP using the Multiprime DNA-labeling system (Amersham Pharmacia). Approximately 800 000 plaques of a Lambda ZAPII (Stratagene) cDNA library derived from embryos from stages 22–25 were blotted onto Hybond-N membrane (Amersham Pharmacia), hybridized with the probe in ExpressHyb hybridization solution (Clontech, Palo Alto, CA, USA) and washed under stringent conditions.
Constructs of the template for mRNA synthesis
The Tbx6Eco fragment (M1 to E311) was amplified by PCR and cloned into the T-vector to generate Tbx6Eco/pBS. Similarly the Tbx6Bgl fragment (M1 to E311) was amplified and cloned into the T-vector to generate Tbx6Bgl/pBS. The repressor domain of Drosophila Engrailed (A2 to S298; Conlon et al. 1996) was amplified by PCR (forward primer, 5′-GAATTCTGGCCCTGGAGGATCGC-3′; reverse primer, 5′-AGATCTTCAGGATCCCAGAGCAGATTTCTCTGG-3′) from a whole Engrailed plasmid and cloned into the T-vector to generate EnR/pBS. The EnR domain was excised from EnR/pBS by EcoRI and ligated into the EcoRI site of Tbx6Eco/pBS to generate Tbx6EnR/pBS. The VP16 activation domain (360 bp) was excised by BglII and BamHI from the CRF2 plasmid, and was ligated to Tbx6Bgl/pBS, which was digested with BglII and BamHI to generate Tbx6VP16/pBS. The whole Tbx6-coding sequence (M1 to M506) was also amplified by PCR and inserted into the T-vector. In the above PCR, template plasmids were amplified for fewer than eight cycles. The whole Antipodean coding sequence was amplified from stage 13 cDNA by PCR (forward primer, 5′-AGGAACATGCATTCTCTGC-3′; reverse primer, 5′-CTGAGGTATTGGCATGGATGG-3′) for 40 cycles and cloned into the T-vector. In all PCR runs, LaTaq (TaKaRa) was used. Coding sequences of Tbx6, Tbx6EnR, Tbx6VP16, and Apod were excised from pBS with XhoI and XbaI and inserted into pCS2+. DNA sequence analysis confirmed correct connections and no mutations in the above constructs. The Apod protein had several mutations from the published sequence (Stennard et al. 1996): P82 to S, A120 to S, V295 to A, A345 to P, H416 to R and S419 missing.
In situ hybridization and northern blotting
In situ hybridization was performed according to the methods of Sive et al. (2000). A probe was synthesized according to the manufacturer’s instructions by using T7 RNA polymerase (Gibco-BRL) and DIG RNA labeling mix (Boehringer Mannheim) from a template of full-length Tbx6 cDNA in a pBluescript SK– phagemid linearized with XbaI. Probes were alkali hydrolyzed. After hybridization, embryos were treated with RNaseA and RNaseT1. Embryos were finally stained with nitroblue tetrazolium (NBT)/5-bromo-4-chloro-3-indolyl phosphate (BCIP) and cleared in benzyl alcohol (BA)/benzyl benzoate (BB). After this, some embryos were transferred to xylene and embedded in paraffin; 15–20 μm sections were cut, mounted on a slide glass, deparaffinized and observed under Nomarski optics.
For northern blotting, LiCl-precipitated total RNA from 50 embryos was purified by using Oligotex-dT30 (TaKaRa) according to the manufacturer’s instructions. PolyA(+) RNA from 50 embryos was electrophoresed in 1.5% agarose gel containing 2% formaldehyde and transferred to Hybond-N+ membrane (Amersham Pharmacia). A 1.2 kb PvuII-fragment from the Tbx6 coding sequence was labeled with [α-32P]-dCTP using a Megaprime kit (Amersham Pharmacia) and was hybridized to the membrane and washed as described elsewhere (Sambrook & Russell 2001). The membrane was exposed to an X-ray film (X-OMAT AR; Kodak, Rochester, NY, USA).
Reverse transcription–polymerase chain reaction
For the detection of gene expression, cDNA from the animal cap was amplified by ExTaq (TaKaRa). The following primers were used: Xbra (forward, 5′- TTACACCAAGGATCGTTATCACC-3′; reverse, 5′-GTGGTTATTTAGGCAGCAAGTAGC-3′, 353 bp, annealing temperature 60°C, 35 cycles), Tbx6 (forward, 5′-ATTGATTGGCAGGTACAGCACG-3′; reverse, 5′-AGTTCATGGAGAAACCTCTG-CC-3′, 329 bp, 62°C, 35 cycles), Eomes (forward, 5′-TACCAACTTCAGCTCCTCGC-3′; reverse, 5′-TCAGATCCTCCACTCCATCC-3′, 813 bp, 60°C, 45 cycles), Apod (forward, 5′-CAGCACAAACAGTAGGTTGCC-3′; reverse, 5′-CACAGCGTTAGTCTCATAGTCAGC-3′, 500 bp, 61°C, 35 cycles), Xwnt8 (forward, 5′-CAAGGCATATCTGACAT-ACTCAGC-3′; reverse, 5′-TTCCTCTTGTCCATCTCAAGC-3′, 573 bp, 60°C, 35 cycles), ornithine decarboxylase (ODC; forward, 5′-GAACAGCTTCAGCAATGACG-3′; reverse, 5′-TGAGTCATCAGTTGCTATGCG-3′, 478 bp, 60°C, 35 cycles), cardiac actin (forward, 5′-GGTCAGGTCATCACCATTGG-3′; reverse, 5′-GGTACTCTGTATATGTTGCTTGGAGG-3′, 436 bp, 60°C, 35 cycles), XlHbox8 (forward, 5′-GACATCTCAGGCTATGAAGTGC-3′; reverse, 5′-GCCAGCTCTACTCTCCTTGG-3′, 365 bp, 62°C, 35 cycles), T4 globin (forward, 5′-CCGCTTCTCAAGATGACTCTG-3′; reverse, 5′-AGCTGTCCTTGCTGATACCTTC-3′, 468 bp, 62°C, 35 cycles). Samples (10 μL) of the amplified solution were electrophoresed on an agarose gel, stained with ethidium bromide and photographed on 667 film (Polaroid, Cambridge, MA, USA).
Cloning of Tbx6 from Xenopus laevis
We searched for new T-box genes from Xenopus laevis and performed PCR from gastrula cDNA using degenerate primers. The fragments obtained were heterogeneous in length, spanning approximately 250–270 bp These fragments were cloned into pBluescript II KS(–) vector and several clones were sequenced. The results indicated that the clones contained known T-box sequences, such as Xbra, eomesodermin (eomes) and VegT/Antipodean (Apod), at a high frequency. Therefore, PCR fragments were digested with AlwNI to remove eomes, NsiI to remove Apod and BclI to remove Xbra, and the remaining intact fragments were cloned. One such clone had homology with mouse Tbx6. The 5′ upstream sequence of this fragment of approximately 700 bp was obtained by 5′ rapid amplification of cDNA ends (RACE). We obtained a cDNA clone from a neurula cDNA library by using the BamHI fragment from the 5′ RACE product as a probe.
The isolated clone was 2.0 kb long and had an open reading frame coding for 506 amino acids (Fig. 1A,B). The 5′ end of the clone had only approximately 30 bp upstream from the first ATG, in which no in-frame stop codon was present. We performed 5′ RACE again from the 5′ terminal 300 bp of the cDNA clone, but obtained no further upstream cDNA sequence. Therefore, we deduce that the cDNA clone obtained is full length and that the translation starts from the first or the second ATG, which is 45 nucleotides downstream. Both sequences are not quite consistent with the Kozak sequence. The 3′ UTR of this clone was approximately 480 bp long and had a non-standard putative polyA signal (AUUAAA).
The coding region had a T-box sequence and an adjacent putative nuclear localization signal (NLS) consisting of a cluster of basic amino acids (Fig. 1B). Similar to eomes, this coding region has several SPXX sequences, as shown by underlines (Fig. 1B). When the whole amino acid sequence was subjected to a BLASTP homology search, mouse Tbx6 (46% identity, 168 amino acids of 345) and human Tbx6 (43%, 182 amino acids of 419) were the most similar proteins, followed by members of the Tbx6 family of proteins including zebrafish spadetail, Chick Tbx6L and Xenopus VegT, followed by other T-box members. A dendrogram of the T-box amino acid sequences of the Tbx6 family proteins (Fig. 1C) again shows that the T-box of this clone is most homologous to mouse and human Tbx6 and we think that it is an ortholog of mouse Tbx6. Outside the T-box domain, this clone has no homologous proteins except that its C-terminal sequence is weakly homologous with SAP-1, an Ets-related transcription factor (Dalton & Treisman 1992).
Temporal and spatial pattern of Tbx6 expression
Northern blot analysis showed a faint band of approximately 2 kb from the early gastrula (stage 10; Fig. 2A). The signal was strong in the middle and late neurula (stages 16 and 20), declined sharply at stage 30 and was not detected at stage 40. In reverse transcription– polymerase chain reaction (RT-PCR) experiments, the maternal transcript was not detected, and the expression began in the early gastrula (stage 10). The expression reached its peak from late gastrula to middle neurula (stages 13–16) and decreased until stage 30 (Fig. 2B).
Whole-mount in situ hybridization showed that, in the gastrula, Tbx6 was expressed around the blastopore, except at the dorsal midline (Fig. 3A). As gastrulation proceeded, the expression area extended anteriorly and was widest at late gastrula to early neurula stages (Fig. 3B,C). At the tailbud stage, the area retracted to the tip of the tailbud (Fig. 3D,E).
A section of the early neurula embryo showed that Tbx6 was expressed in the presumptive somite and the lateral plate mesoderm. The latter includes both the splanchnic and somatic mesoderm (Fig. 3F). A section of the caudal part of the tailbud embryo showed that Tbx6 was expressed in the presomitic mesoderm and the whole lateral plate mesoderm except around the proctodeum (Fig. 3G). Sagittal sections of gastrula to neurula stage embryos showed that the presomitic mesoderm is positive, while segmented somites are negative (Fig. 3H–J).
Overexpression of Tbx6 mRNA in the embryo
To uncover the function of Tbx6, we performed an overexpression experiment. Synthetic Tbx6 mRNA was injected into the dorsal marginal zone of embryos at the four-cell stage. Approximately half the embryos displayed a gastrulation defect. However, the rest of the embryos completed gastrulation. When control embryos reached the tailbud stage (stage 31), most had no apparent head region (Fig. 4A). Sections of such embryos showed that the neural tube was pyknotic, the notochord was poorly differentiated and the somite was enlarged compared with that of normal embryos (Fig. 4E,F). Injection to the ventral marginal zone also caused gastrulation defects in approximately half the embryos, among which, in case of the weak phenotype, the anal region did not close completely (Fig. 4B–D). The rest developed to tailbud embryos with no apparent defects.
Overexpression of Tbx6 in animal caps
Tbx6 was truncated after the putative NLS and was fused to either the activation domain of the Herpes virus VP16 gene or the repression domain of the Drosophila Engrailed gene (Fig. 5A). Wild-type Tbx6 or fusion mRNA (400 pg) was injected into two-cell stage embryos near the animal pole and the animal caps were explanted at the blastula stage. When control embryos reached stage 13, the control caps were spherical (Fig. 5B) but the Tbx6-injected caps were elongated slightly (Fig. 5C). Noggin-injected caps did not close completely and elongated a little (Fig. 5D). Caps coinjected with Tbx6 and noggin mRNA were elongated markedly (Fig. 5E).
When control embryos reached stage 35, caps injected with 400 pg Tbx6EnR mRNA developed into atypical epidermis, similar to control caps (Fig. 5F,I). Caps injected with Tbx6 mRNA swelled and became spherical. These caps differentiated into mesenchyme and mesothelium, as did caps injected with Tbx6VP16 mRNA (Fig. 5G,H). Noggin-injected caps differentiated into cement gland and neural-like cells characterized by high cell density and well-stained nuclei with hematoxylin (Fig. 5J). When 400 pg noggin mRNA and 400 pg Tbx6 mRNA were coinjected, the animal cap differentiated into well-developed muscle (Fig. 5K). Results of histological analyses of stage 35 animal caps are summarized in Table 1.
Table 1. Frequencies of induction of each tissue
Noggin + Tbx6
200 + 400 pg
Each value represents percentage frequency.
Reverse transcription–polymerase chain reaction analysis of animal cap explants
To determine whether the expression of Tbx6 is elicited by mesoderm-inducing factors, we treated animal caps with activin or bFGF. When control embryos reached stage 13, the caps were harvested and analyzed with RT-PCR. Both factors induced Tbx6 expression similarly well at doses used (0.5–10 ng/mL activin and 1–10 ng/mL bFGF). We also injected the CSKA-Xwnt8 plasmid or BMP-4 mRNA at the animal pole of two-cell embryos and explanted the animal cap. Both these animal caps expressed Tbx6 transcript (Fig. 6A). Expression of Tbx6 was also induced efficiently by injection of Apod mRNA, or Tbx6 mRNA itself, and less efficiently by Xbra mRNA. The expression of Apod, Xbra, Eomes and Xwnt8 by various mesoderm- inducing substances or T-box genes was also examined. These genes also responded to a number of factors.
In contrast, injection of Tbx6 mRNA induced expression of Apod, Xbra, Eomes and Xwnt8 mRNA. When control embryos reached stage 35, marked T4 globin expression and weak cardiac actin expression were detected in Tbx6-injected animal caps. In coinjection of Tbx6 and noggin mRNA, the expression of cardiac actin was considerably augmented, coincidentally with histological analysis. In coinjected animal caps, expression of T4 globin decreased, but expression of XlHbox8 was detected (Fig. 6B). In noggin mRNA-injected caps, these mesodermal and endodermal markers were not detected.
Characterization of Xenopus Tbx6
Genes in the Tbx6 and VegT/Apod (Stennard et al. 1999) group have relatively low homology outside the T-box, so we classified Tbx6 based only on the T-box structure. Because the clone had higher homology with mouse and human Tbx6 than with Xenopus VegT/Apod, we concluded that it was an ortholog of mouse and human Tbx6. Because the clone does not have very high homology with zebrafish Tbx6 (Hug et al. 1997), it is unclear whether zebrafish Tbx6 also falls into the same group, as does ascidian As-T2 (Yasuo et al. 1996). However, the temporal and spatial pattern of expression of zebrafish Tbx6 is very similar to that of Xenopus Tbx6 (Hug et al. 1997). The temporal pattern of Tbx6 expression shows that its peak occured at the late gastrula to early neurula stages, later than that of Apod or Eomes, the expression peaks of which occur in the early to middle gastrula stages (Lustig et al. 1996; Ryan et al. 1996; Stennard et al. 1996; Zhang & King 1996; Horb & Thomsen 1997).
From the present study, Xenopus Tbx6 is suggested to be a transcriptional activator, but the activity seems to be moderate. At 10 pg, Tbx6VP16 mRNA was toxic to the animal cap, as was 1 ng Tbx6 mRNA, and 2 pg Tbx6VP16 mRNA had similar mesoderm-inducing activity to 400 pg of Tbx6 mRNA. Therefore, Tbx6VP16 was roughly 100-fold as potent as native Tbx6 at inducing mesoderm. Xbra (Kispert et al. 1995), eomes (Ryan et al. 1996) and Apod (Stennard et al. 1996) are also activators, but tbx2 (Carreira et al. 1998) and tbx3 are repressors (He et al. 1999).
In the animal cap assay, Tbx6 mRNA induced ventral mesodermal tissues. When we injected higher doses of mRNA, many of the animal caps became disaggregated and, finally, necrotic. We also tested higher doses of Tbx6VP16 mRNA, but the result was the same as with Tbx6 mRNA. We think that it is difficult to induce muscle differentiation solely by overexpressing Tbx6 under the conditions used in our experiments. Although cardiac actin was weakly expressed in Tbx6-injected animal caps, we never found muscle by histology in these caps. In contrast, Xbra can induce muscle at high doses (Cunliffe & Smith 1992; Rao 1994; O’Reilly et al. 1995). VegT/Apod is also capable of inducing muscle at high doses (Zhang & King 1996; Stennard et al. 1996; Horb & Thomsen 1997). Brat/VegT can also induce endodermal markers such as XlHbox8 and IFABP (Horb & Thomsen 1997). In contrast, Xlhbox8 expression was not induced by Tbx6 alone. VegT induced the secondary axis when overexpressed in ventral blastomeres (Zhang & King 1996), but ventral overexpression of Tbx6 did not have such an action. Therefore, Xbra and VegT/Apod/Brat can induce a wider spectrum of tissues than Tbx6. Xlhbox8 was expressed when we coinjected Tbx6 with noggin mRNA. Because the expression of Tbx6 rapidly retracted from the trunk region caudally after the late neurula stage, we think it rather unlikely that Tbx6 is involved in liver or pancreas differentiation. It is possible that another T-box gene, such as VegT, is involved in the development of these endodermal tissues in combination with BMP-suppressing factors, such as noggin.
Regulation of Tbx6 expression
Xenopus Tbx6 mRNA was detected in caps treated with bFGF, BMP-4, CSKA-Xwnt8 or activin. Therefore, Tbx6 seems to be induced by a variety of mesoderm-inducing signals. In case of bFGF and activin, the expression of Tbx6 was apparently not affected by different concentrations of the inducing factors. This pattern is similar to that of Xwnt8 and contrasts with Eomes, which was dose-dependently affected by inducing factors.
In t/t homozygous mice, the start of Tbx6 expression is not inhibited, but the expression disappears after 8.5 d.p.c., when developmental abnormalities appear. This suggests that T is concerned with the maintenance of Tbx6 expression (Chapman et al. 1996). Similarly, the expression of Tbx6 was not inhibited in Wnt3a-knockout mice up to the two-somite stage, but it disappeared at the six-somite stage (Yamaguchi et al. 1999). In zebrafish, ventrolateral expression of Tbx6 started in no tail (ntl) mutants, but its expression in the tailbud was missing (Hug et al. 1997). In spadetail (spt) mutants, expression of Tbx6 was not detected in the ventrolateral region of embryos, but it was retained in presomitic mesoderm or in the tailbud (Griffin et al. 1998). Therefore, VegT/Apod seems to direct Tbx6 expression in the ventrolateral region, whereas T/Brachyury directs its expression in the tailbud. In support of this, the expression of Tbx6 is totally undetectable in ntl/spt double mutants (Griffin et al. 1998). Although zebrafish Tbx6 may not be an ortholog of the present Xenopus Tbx6, it seems plausible that several factors control the expression of Xenopus Tbx6. In fact, Xbra, Apod or Xwnt8 can induce expression of Xenopus Tbx6. Taking the temporal patterns of their expression into consideration, all these factors are candidates for regulators of Tbx6 expression.
Xenopus Tbx6 induced its own expression, similar to Xbra (Tada et al. 1997) and VegT (Zhang & King 1996). Tbx6 also induced the expression of Apod, Xbra and Xwnt8. There are many positive loops of expression between Tbx6 and Apod, Xbra or Xwnt8. There is also a positive loop between Apod and Eomes (Stennard et al. 1996). Eomes and Brat/Apod induce the expression of Xbra and Xwnt8 (Stennard et al. 1996; Horb & Thomsen 1997). Therefore, many positive loops exist among these mesodermal inducers and T-box genes. There are, however, several differences between Tbx6 and Eomes. Eomes seems to be little or not at all induced by Xbra, BMP-4 and CSKA-Xwnt8, whereas Tbx6 was moderately induced by these factors.
Roles of Tbx6 in somite formation
Overexpression of VegT mRNA in the dorsal blastomeres inhibits forebrain development (Zhang & King 1996). A similar phenotype was induced by overexpression of Tbx6 in dorsal blastomeres. In these embryos, the trunk neural tube had regressed and the notochord did not differentiate well. Degeneration of the neural tube may be a secondary effect of insufficient formation of the notochord, but Tbx6 may have directly inhibited neural tube differentiation. Another possibility is that Tbx6 acted via Apod/VegT or Xwnt8 (Christian & Moon 1993). To avoid inhibition of axial development, it seems that the expression of Tbx6 is clearly repressed in notochord cells from the onset of its expression.
In Tbx6-knockout mice, the somite is not formed caudally to the forelimb bud and, instead, neural tube-like structures arise (Chapman & Papaioannou 1998). From this result, we deduced that Xenopus Tbx6 is also involved in the formation of somites. However, the overexpression of Tbx6 mRNA in animal caps did not induce muscle differentiation. Other factors have possible implications in somite formation. For instance, in T mutant mice, somites are greatly reduced (Chesley 1935). In zebrafish no tail mutants, somites become misshapen and lack muscle pioneer cells (Halpern et al. 1993). However, muscle differentiation does occur in these embryos. From the result that transplantation of wild-type cells into the midline region of no tail mutants restores not only the notochord and but also correctly shaped somites made up of surrounding no tail cells, it is expected that no tail (T) participates mainly in notochord formation and that the notochord secretes some factors that support somite differentiation (Halpern et al. 1993). One such candidate is noggin.
When noggin-expressing tissue culture cells are implanted near the somite of chick embryos, the myo-D-expressing area expands and additional somites occur (Hirsinger et al. 1997; Reshef et al. 1998; Tonegawa & Takahashi 1998). In noggin-knockout mice, somite or myotomal differentiation is very much inhibited (McMahon et al. 1998). Therefore, we injected noggin mRNA together with Tbx6 mRNA in the animal cap. The result that the overexpression of the two mRNAs elicited muscular differentiation supports the view that Tbx6 and noggin are also involved in somitic muscle differentiation in Xenopus. In zebrafish, the spadetail mutant has abnormal somites similar to no tail mutants in the tail region (Griffin et al. 1998). However, the expression of both Xbra and Apod disappears from dorsolateral mesodermal regions early in development. At the late gastrula stage (stage 13), the expression of Apod becomes trivial and later expression is seen only in Rohon–Beard cells and in the tip of the tailbud (Lustig et al. 1996; Stennard et al. 1996; Zhang & King 1996). In addition, the expression of Xbra after gastrulation is restricted to notochord and the tip of the tailbud (Smith et al. 1991). Therefore, the role of Xbra or Apod in muscle differentiation may be indirect and it is possible that Xbra and Apod act via Tbx6, whose expression peaks between stages 13 and 16 in presomitic and lateral plate mesoderm. Involvement of Tbx6 in somitic muscular differentiation may be evolutionarily conserved because the ascidian-related gene As-T2 is expressed in cells of the muscular lineage (Yasuo et al. 1996) and this gene has muscle-inducing activity (Mitani et al. 1999), although, in this case, ectopic muscle differentiation occured following injection of As-T2 mRNA alone.
We are grateful to Dr N. Mizuno of the ERATO project (Osaka University, Osaka, Japan) for his gift of the neurula cDNA library. We are also very grateful to Dr T. Kornberg (University of California, San Francisco, CA, USA) for the Engrailed plasmid, to Dr S. J. Triezenberg (Michigan State University, East Lansing, MI, USA) for the CRF2 plasmid, to Dr S. Nishimatsu (Kawasaki Medical School, Kurashiki City, Okayama, Japan) for the BMP-4 plasmid, to Dr W. C. Smith (University of California, Berkeley, CA, USA) for the noggin plasmid and Dr J. C. Smith (National Institute for Medical Research, London, UK) for the Xbra plasmid. We also thank Mr T. Hayata (Tokyo University, Komaba, Tokyo, Japan) for his help in DNA sequencing. This work was supported by grants to H. U. from the Ministry of Education, Science, Sport and Culture of Japan, from the Kihara Memorial Foundation, and from the Foundation for Promotion of Science and Education in Yokohama.