We have isolated a novel basic helix–loop–helix (bHLH) gene homologous to the Drosophila proneural gene atonal, termed ATH‐3, from Xenopus and mouse. ATH‐3 is expressed in the developing nervous system, with high levels of expression in the brain, retina and cranial ganglions. Injection of ATH‐3 RNA into Xenopus embryos dramatically expands the neural tube and induces ectopic neural tissues in the epidermis but inhibits non‐neural development. This ATH‐3‐induced neural hyperplasia does not require cell division, indicating that surrounding cells which are normally non‐neural types adopt a neural fate. In a Xenopus animal cap assay, ATH‐3 is able to convert ectodermal cells into neurons expressing anterior markers without inducing mesoderm. Interestingly, a single amino acid change from Ser to Asp in the basic region, which mimics phosphorylation of Ser, severely impairs the anterior marker‐inducing ability without affecting general neurogenic activities. These results provide evidence that ATH‐3 can directly convert non‐neural or undetermined cells into a neural fate, and suggest that the Ser residue in the basic region may be critical for the regulation of ATH‐3 activity by phosphorylation.
In Drosophila, neural development is controlled positively or negatively by multiple basic helix–loop–helix (bHLH) genes (Campos‐Ortega and Jan, 1991; Jan and Jan, 1993; Jarman et al., 1993, 1994). For example, achaete–scute complex (AS‐C) and atonal are proneural genes, and the former is required for external sensory organ development while the latter is required for photoreceptor and chordotonal organ development. In contrast, hairy and Enhancer of split [E(spl)] inhibit neural development by antagonizing the proneural genes.
As in Drosophila, vertebrate neurogenesis is also controlled positively or negatively by multiple bHLH genes (Guillemot et al., 1993; Ferreiro et al., 1994; Ishibashi et al., 1994, 1995; Turner and Weintraub, 1994; Kageyama et al., 1995; Lee et al., 1995; Tomita et al., 1996). XASH‐3, a Xenopus bHLH gene homologous to the Drosophila proneural gene complex AS‐C (Zimmerman et al., 1993), converts ectodermal cells to a neural fate and therefore acts as a proneural gene (Ferreiro et al., 1994; Turner and Weintraub, 1994). Mash‐1, a mammalian bHLH gene homologous to AS‐C, regulates early stages of neuronal differentiation, and NeuroD, another vertebrate bHLH gene, functions in terminal differentiation of neurons (Johnson et al., 1990; Guillemot et al., 1993; Lee et al., 1995). Thus, multiple bHLH genes positively regulate vertebrate neurogenesis at different stages. In contrast, HES‐1, a mammalian bHLH gene homologous to Drosophila hairy and E(spl), is expressed in neural precursor cells and acts as a negative regulator of neurogenesis (Akazawa et al., 1992; Sasai et al., 1992). Forced expression of HES‐1 blocks neuronal differentiation in the brain (Ishibashi et al., 1994) and retina (Tomita et al., 1996). Conversely, a HES‐1‐null mutation accelerates neuronal differentiation and results in severe anomalies of the brain (Ishibashi et al., 1995) and eye (Tomita et al., 1996). Thus, HES‐1 prevents premature neurogenesis and regulates brain and eye morphogenesis.
The exact mechanism by which HES‐1 prevents premature neurogenesis is unclear, but it is likely that HES‐1 antagonizes bHLH genes that positively regulate neurogenesis, as in the case of Drosophila. Interestingly, expression of Mash‐1 is up‐regulated in HES‐1‐null mice (Ishibashi et al., 1995), suggesting that Mash‐1 up‐regulation may contribute to premature neurogenesis. However, null mutation of Mash‐1 does not cause any apparent abnormalities in the central nervous system (CNS) (Guillemot et al., 1993) and, based upon the expression patterns, XASH‐3 or its mammalian equivalent and NeuroD are unlikely to compensate for Mash‐1. Therefore, additional bHLH genes may be required for the CNS development.
We have isolated from Xenopus and mouse a novel bHLH gene, termed ATH‐3, which is homologous to the Drosophila proneural gene atonal (Jarman et al., 1993, 1994). In Xenopus, ATH‐3 expression starts in the presumptive CNS during neural induction and persists at a high level in the brain and retina. In addition, ATH‐3 is able to promote neural development at the expense of adjacent non‐neural tissues. In the animal cap assay, ATH‐3 can directly convert ectodermal cells into neurons expressing anterior markers. Interestingly, a Ser→Asp mutation in the basic region, which mimics phosphorylation of Ser, impairs anterior marker‐inducing abilities without losing the general neurogenic activity. These results indicate that ATH‐3 can act as a vertebrate proneural gene and that ATH‐3 activity may be regulated by phosphorylation of Ser in the basic region.
Structural analysis of Xenopus and mouse ATH‐3
PCR with degenerate oligonucleotide primers was carried out to search for a novel bHLH gene expressed in the developing nervous system. We obtained a PCR fragment from a bHLH gene, which was termed ATH‐3 because of structural similarity to the Drosophila proneural gene atonal. This PCR fragment was used to screen a mouse genomic library to obtain the mouse ATH‐3 gene (Isaka et al., 1996). By using this ATH‐3 genomic clone as a probe, Xenopus and mouse cDNA libraries were screened to determine the full‐length coding sequences.
Xenopus and mouse ATH‐3 consisted of 315 (Figure 1A) and 330 amino acid residues (Figure 1B), respectively, and shared 93% identity in the bHLH domain (Figure 1, underlined, and Figure 2B). The structural similarity extended to the upstream and downstream regions of the bHLH domain, with an overall identity of 67% (Figure 2A). In addition, ATH‐3 showed significant sequence homology to MATH‐1 (Akazawa et al., 1995), MATH‐2/NEX‐1 (Bartholomä and Nave, 1994; Shimizu et al., 1995), NeuroD/BETA2 (Lee et al., 1995; Naya et al., 1995), NDRF/KW8 (Kume et al., 1996; Yasunami et al., 1996) and Atonal (Jarman et al., 1993) in the bHLH domain (Figure 2B), suggesting that these factors share a recent common ancestral gene. However, unlike the latter factors, ATH‐3 contained serine (amino acid residue 89 in Xenopus) or threonine (98 in mouse) in the basic region, which forms a potential phosphorylation site. It has been shown that myogenic bHLH factors contain a threonine residue in the basic region and that its phosphorylation inactivates the myogenic activity (Li et al., 1992). Thus, the activity of ATH‐3 could also be regulated by phosphorylation and dephosphorylation of the basic region. Another structural feature common to Xenopus and mouse ATH‐3 is the acidic region (Glu–Asp stretch; amino acid residues 38–56 of Xenopus and 49–65 of mouse) and the basic region (Lys–Arg stretch; amino acid residues 57–72 of Xenopus and 66–81 of mouse), which are located upstream of the bHLH domain. These acidic and basic regions are also present in MATH‐2, NeuroD and NDRF, and could be involved in transcriptional activity.
Spatial and temporal distribution of ATH‐3
The spatiotemporal expression patterns of ATH‐3 were determined by in situ hybridization. In Xenopus, ATH‐3 expression was first detected weakly at stage 12 in two stripes within the presumptive neural plate (Figure 3A, arrowheads). At stage 14, ATH‐3 was expressed in three stripes on both sides of the midline of the neural plate (Figure 3B, arrowheads). Assessed by their positions, cells in these stripes may correspond to primary neuronal precursors, which differentiate into motor and sensory neurons and interneurons. At stages 18 and 21, strong ATH‐3 expression appeared in the cranial ganglions (Figure 3C and D, arrowheads) while weaker expression remained in the spinal cord. Later, ATH‐3 was expressed strongly in the eye, forebrain and cranial ganglions, where active neuronal differentiation occurs (Figure 3E and F). Thus, ATH‐3 initially may be expressed in neural progenitor cells and later in differentiating post‐mitotic neurons of the anterior nervous system in Xenopus embryos.
In mice, weak expression of ATH‐3 initially was detected widely in the neural tube around embryonic day (E) 8.5, before the process of turning (Figure 4A). By E9.5, ATH‐3 expression had become prominent in the ventral part of the brain and spinal cord (Figure 4B) and initiated weakly in the retina (data not shown). At E10.5, ATH‐3 was expressed at high levels in the trigeminal (Figure 4C, arrow) and dorsal root ganglions (Figure 4D) and the ventral part of the midbrain and hindbrain (data not shown). At E12.5, ATH‐3 expression was detected in the mantle layer of the brain and spinal cord, trigeminal ganglion, Rathke's pouch, dorsal root ganglion and the ventricular zone of the neural retina (Figure 4F and H). At E16.5, the expression persisted at a high level in the diencephalon (Figure 4J) and the neural retina (Figure 4L) but decreased to quite low levels in other regions. Thus, ATH‐3 expression initially occurred widely in the nervous system but then became restricted to the anterior region in mouse embryos, indicating that Xenopus and mice show somewhat similar patterns of spatiotemporal expression.
Northern blot analysis demonstrated that a high level of ATH‐3 expression continued after birth to adulthood in mouse retina (Figure 5A). In situ hybridization analysis showed that ATH‐3 was expressed in the outer half of the ventricular zone at postnatal day (P) 0 (Figure 5B). During P3–P5, the inner (INL) and outer nuclear layers (ONL) develop in the retina and, after this stage, ATH‐3 expression continued in the INL (Figure 5B), indicating that mature retinal neurons also expressed ATH‐3. In adults, ATH‐3 was not expressed in any other tissues that we examined (Figure 5A and data not shown). Thus, ATH‐3 expression persisted in the neural retina from E9.5 onwards, suggesting that ATH‐3 may be involved not only in the development but also in the maintenance of retinal cells.
Neural hyperplasia by injection of ATH‐3 RNA
To assess the function of ATH‐3 in neural development, in vitro generated Xenopus ATH‐3 RNA was injected into Xenopus embryos. When 50 or 100 pg of ATH‐3 RNA was injected into one cell of two‐cell stage Xenopus embryos, the most notable and frequent phenotype was expansion of the CNS on the injected side (Table I). Co‐injection of β‐galactosidase RNA allowed us to confirm the injected side by X‐gal staining (data not shown).
Immunological analyses with the pan‐neural marker NEU‐1 (Itoh and Kubota, 1989) showed that the neural tube was significantly enlarged laterally at stages 25 and 30 (Figure 6A and B, arrowheads). The optic cup was deformed and displaced on the injected side (Figure 6B, arrow). This ATH‐3 phenotype of neural tube enlargement is different from that of NeuroD, which does not cause neural tube hyperplasia (Lee et al., 1995), but seems quite similar to that of XASH‐3, which expands the neural tube (Ferreiro et al., 1994; Turner and Weintraub, 1994). In addition to the expanded neural tube, ectopic NEU‐1 or neural cell adhesion molecule (N‐CAM) staining was detected in the epidermis on the injected side (Figure 6B, G and H). Ectopic neurogenesis was also induced in the epidermis when ATH‐3 RNA was injected into a ventral cell of four‐cell stage embryos (Figure 6E). Ectopic NEU‐1‐positive cells in the epidermis showed the morphology of neurons with multiple processes (Figure 6E). This ATH‐3 phenotype is quite similar to that of NeuroD, which ectopically converts neural crest and epidermal cells into neurons (Lee et al., 1995). These results suggest that ATH‐3 cannot only expand the neural tube like XASH‐3 but can also induce ectopic neurogenesis in the epidermis like NeuroD.
At the tadpole stage, the brain was significantly enlarged (Figure 6I, arrowhead), and the eye was deformed and displaced ventrally on the injected side. Histological analyses revealed that the forebrain dramatically expanded laterally and anteriorly on the injected side (Figure 6J, arrowheads). As a result, the development of the surrounding regions was disturbed. The hindbrain was also expanded by overexpression of ATH‐3 (Figure 6K, arrowheads). Because the injected ATH‐3 RNA was likely to have disappeared by this time, the above results indicate that neural hyperplasia remained permanently by transient expression of ATH‐3.
The observed neural hyperplasia could be due to either cell proliferation or conversion of surrounding non‐neural cells to a neural fate. To distinguish between these possibilities, cell division of ATH‐3 RNA‐injected embryos was blocked by hydroxyurea/aphidicolin (HUA) treatment at mid‐gastrulation. It has been shown that embryos treated with HUA at mid‐gastrulation stop cell division but develop almost normally until tailbud stages (Harris and Hartenstein, 1991; Turner and Weintraub, 1994). As shown in Figure 6C, ATH‐3 overexpression expanded the CNS and induced ectopic neurogenesis in the epidermis on the injected side even in the presence of HUA. Thus, cell proliferation is not necessary for neural hyperplasia induced by ATH‐3, suggesting that overexpression of ATH‐3 converts adjacent non‐neural cells to a neural fate.
Suppression of non‐neural tissue development by ATH‐3
If the surrounding non‐neural cells are converted into a neural fate, development of non‐neural tissues should be reduced in ATH‐3 RNA‐injected embryos. To test this possibility, we next determined expression of non‐neural markers in the injected embryos. On the ATH‐3 RNA‐injected side, expression of twist, a marker for non‐neural types of neural crest cells (Hopwood et al., 1989; Turner and Weintraub, 1994), was severely reduced (Figure 7B and C), suggesting that ectopic ATH‐3 expression decreased the population of neural crest cells with the potential to differentiate into non‐neural cells. These cells may instead become ectopic neurons, which were observed in the epidermis of ATH‐3 RNA‐injected embryos (see Figure 6E, G and H).
In the injected embryos, expression of keratin, an epidermal marker (Jonas et al., 1985), was also significantly reduced (Figure 7E and F). In addition, somite formation was also severely blocked, as revealed by impaired myosin expression (Figure 7H, arrows). These results support the hypothesis that ATH‐3 expands the neural tissues at the expense of adjacent non‐neural cells.
Activation of neural gene expression in animal caps by ATH‐3
To assess more clearly the ATH‐3 function in neurogenesis, additional markers were examined in the injected embryos. The ATH‐3 RNA was injected into both cells of two‐cell stage Xenopus embryos, and subsequently the animal caps were isolated from the injected embryos at stage 9. These explants were cultured for an additional 3 h (equivalent to stage 11), 1 day (stages 25–30) or 3 days (stages 35–40). Uninjected animal caps are known to become atypical epidermis and, therefore, neural and mesodermal markers were not expressed (Figure 8, lane 2). In contrast, injection of ATH‐3 RNA induced significant expression of the pan‐neural marker N‐CAM (Kintner and Melton, 1987) and the neuronal markers type‐II β‐tubulin (Good et al., 1989) and neurofilament‐M (NF‐M) (Sharpe, 1988) in stage 25–30 animal caps (Figure 8, lane 3), indicating that ATH‐3 can induce neuronal differentiation. In addition, expression of the anterior neural markers XANF‐1 (Zaraisky et al., 1992) and XIF‐3 (Sharpe et al., 1989) and the retinal marker opsin (Saha and Grainger, 1993) was also significantly induced in ATH‐3 RNA‐injected animal caps (Figure 8, lane 3). However, other markers such as the posterior neural marker XlHbox6 (Wright et al., 1990) and the floor‐plate marker F‐spondin (Ruiz i Altaba et al., 1993) were not induced by ATH‐3 (Figure 8, lane 3). Thus, ATH‐3 can promote development of neurons with anterior features, suggesting that ATH‐3 may have a role in specification of anterior neuronal types. This neural induction by ATH‐3 was also observed when the culture was continued for 3 days (equivalent to stages 35–40) (data not shown), indicating that the activation of neural gene expression by ATH‐3 is stable. This is in sharp contrast to the action of XASH‐3, which only transiently induces neural gene expression in animal caps (Ferreiro et al., 1994).
In ATH‐3 RNA‐injected animal caps, expression of mesodermal markers was not detected; Xbra (Smith et al., 1991) and goosecoid (Blumberg et al., 1991) were not expressed in the animal cap explants that had been cultured for 3 h (data not shown). In addition, S‐actin (Stutz and Spohr, 1986) was not expressed in the injected animal caps that had been cultured for 1 day (equivalent to stages 25–30) (Figure 8 lane 3). Thus, ATH‐3 promoted neuronal differentiation without inducing mesoderm, suggesting that ATH‐3 may directly convert ectodermal cells into neurons.
Modification of ATH‐3 activities by a single amino acid change in the basic region
Among the vertebrate neural bHLH factors that have been characterized, ATH‐3 has a unique structural feature; a serine or threonine residue in the basic region (Figure 2B), which forms a potential phosphorylation site. As an initial step in relating the possible phosphorylation to the neurogenic activity of ATH‐3, a Ser89→Asp mutation was introduced into Xenopus ATH‐3 (S89D), which mimics the phosphorylation of Ser. Another mutation was also introduced to change Ser89 to Asn (S89N), which could represent a non‐phosphorylated form. On the animal cap analysis, S89D was able to induce N‐CAM, type II β‐tubulin and NF‐M expression (Figure 8, lane 4), indicating that S89D keeps the general neurogenic activities. However, it failed to induce the anterior neural markers XANF‐1 and opsin (Figure 8, lane 4). In contrast, S89N induced anterior neural gene expression as well as general neural markers, like wild type ATH‐3 (Figure 8, lane 5). These results indicate that Ser89 is critical for the regulation of ATH‐3 activity by phosphorylation and that modification of a single amino acid residue in the basic region can regulate some of the neurogenic activities of a bHLH factor.
ATH‐3 may function in both determination and differentiation steps of Xenopus neural development
In this study, we showed that overexpression of ATH‐3 can induce ectopic neurons and hyperplasia of the CNS but suppress development of non‐neural tissues in Xenopus embryos. Interestingly, ATH‐3‐induced neural hyperplasia does not require cell division, indicating that the adjacent non‐neural lineage cells adopt a neural fate. In addition, ATH‐3 can induce neurons without inducing mesoderm in the animal cap assay. These results provide evidence that ATH‐3 can directly convert non‐neural or undetermined cells into neurons.
It has been proposed that there are at least two separate developmental choices for generation of neurons in vertebrates; first, whether or not to adopt a neural lineage, and second, whether or not to differentiate as a neuron (Ferreiro et al., 1994; Turner and Weintraub, 1994; Lee et al., 1995). The first step involves the initial decision between neural and epidermal fates in the ectoderm by a proneural gene, while the second step is the subsequent process of neuronal differentiation. It has been shown that the Xenopus bHLH gene XASH‐3 can expand the neural tube at the expense of adjacent non‐neural ectoderm and therefore acts as a proneural gene (Ferreiro et al., 1994; Turner and Weintraub, 1994). We showed that ATH‐3 can induce similar effects in Xenopus embryos, i.e. expansion of the neural tube and disturbance of development of the surrounding non‐neural tissues. These results suggest that, like XASH‐3, ATH‐3 can function as a proneural gene in Xenopus embryos. The early onset of ATH‐3 expression before the neural plate appears is also consistent with the notion that ATH‐3 is a proneural gene. Interestingly, whereas in animal cap assays XASH‐3 can only transiently induce neural gene expressions and requires the neural inducer noggin for stable induction (Ferreiro et al., 1994), ATH‐3 can stably induce neural gene expression without noggin. Thus, ATH‐3 seems to have a stronger neurogenic activity than XASH‐3, or ATH‐3 may be less susceptible to inhibitory signals such as Notch and HES‐1.
Another bHLH gene, NeuroD, regulates the terminal differentiation step of neural development. It has been shown that NeuroD can induce ectopic neural tissues in the epidermis of Xenopus embryos. ATH‐3 also induces ectopic neural tissues in the epidermis, like NeuroD, suggesting that ATH‐3 and NeuroD have similar neurogenic functions. Furthermore, ATH‐3 is expressed at a high level in the nervous system during neuronal differentiation stages, suggesting that ATH‐3 can also act as a NeuroD‐like differentiation gene. Thus, ATH‐3 may function in both proneural and neuronal differentiation stages in Xenopus embryos.
In the case of mice, the ATH‐3 function remains to be determined. Because ATH‐3 expression starts around E8.5 in the neural tube, after the neural lineage is mainly determined, ATH‐3 may function primarily in neuronal differentiation stages.
ATH‐3 activity may be regulated by phosphorylation of Ser (or Thr) in the basic region
ATH‐3 has a serine or threonine residue in the basic region, which could serve as a phosphorylation site. Other neural bHLH factors contain asparagine or arginine in the corresponding position, thus indicating that the amino acid residue in this position is not conserved among the neural bHLH factors (Figure 2B). Interestingly, the myogenic bHLH factors contain a threonine residue in the basic region, and it has been shown that its phosphorylation inactivates the myogenic activities (Li et al., 1992). Change from Thr to Asp, which mimics the phosphorylation in the basic region of myogenin, causes loss of DNA binding and myogenic activities (Brennan et al., 1991). In the case of Xenopus ATH‐3, mutation of Ser89 to Asp (S89D) maintains a general neurogenic activity but severely impairs anterior marker‐inducing abilities in the Xenopus animal cap assay. In contrast, S89N, which could represent a non‐phosphorylated form, can induce both general and anterior neural markers. These results thus point to the importance of the possible phosphorylation of Ser/Thr in the basic region of ATH‐3, which can modify the anterior‐specific neurogenic activities, although it remains to be determined whether or not the basic region of ATH‐3 is phosphorylated in vivo.
The mechanism by which S89D, which loses anterior neurogenic activity, can induce general neural markers is quite interesting, and we can offer two possibilities. One is that S89D could lose the DNA binding activity and titrate negative regulators by forming a non‐DNA binding heterodimer complex, thereby activating gene expression. The other possibility is that the DNA binding specificity of S89D could be changed by addition of a negative charge in the basic region; it could bind to the promoter of general neural genes but not of anterior genes. Determination of ATH‐3 binding sequences will be necessary to address these problems.
ATH‐3 in retinal development
It is striking that ATH‐3, expressed at a high level in the retina, can induce opsin expression in the animal cap assay, raising the possibility that ATH‐3 may be involved in retinal specification. However, induction of anterior neural markers by ATH‐3 could reflect a general property of Xenopus ectoderm, since neural inducers like noggin, chordin and follistatin also induce a similar spectrum of anterior markers (Lamb et al., 1993; Hemmati‐Brivanlou and Melton, 1994; Sasai et al., 1995). Nevertheless, we speculate that the opsin‐inducing activity may be an intrinsic property of ATH‐3, because S89D loses the opsin‐inducing ability but maintains general neurogenic activities in the animal cap assay.
Eye morphogenesis is known to be regulated by the eye master control gene Pax‐6 (Halder et al., 1995), and it is possible, therefore, that ATH‐3 expression is regulated by Pax‐6. However, ATH‐3 was similarly expressed in the optic vesicles of wild‐type and Small eye mutant mice, which have a Pax‐6 mutation (Hill et al., 1991; Walther and Gruss, 1991), indicating that ATH‐3 expression does not depend upon Pax‐6 (our unpublished data). Thus, ATH‐3 and Pax‐6 may constitute different genetic pathways for retinal development. Because only a part of the Pax‐6 expression domains becomes retina, it has been suggested that another factor not regulated by Pax‐6 may be required for retinal specification (Macdonald and Wilson, 1996). Thus, our findings that ATH‐3 can induce opsin expression suggest that ATH‐3 may have an independent function in the process of retinal specification within Pax‐6‐expressing regions.
Materials and methods
Isolation of the mouse ATH‐3 gene and cDNA
Oligo(dT)‐primed cDNA was prepared from E9.5 mouse CNS and subjected to PCR with fully degenerate oligonucleotide primers corresponding to the amino acid sequences NARER and TLQMA of Atonal, as previously described (Akazawa et al., 1995; Shimizu et al., 1995). We subcloned and sequenced the amplified fragments, and obtained the ATH‐3 sequence. The PCR clone of ATH‐3 was labelled with [α‐32P]dCTP by a random primer method and used as a probe for screening a mouse genomic library (Stratagene) (Takebayashi et al., 1994). Sequence analysis indicated that an ∼7.5 kb BglII fragment of the ATH‐3 gene contained the open reading frame (Isaka et al., 1996). Subsequently, a 685 bp HincII–PstI fragment from the mouse ATH‐3 gene which contained most of the coding region was used as a probe for screening the P0–P10 mouse retina cDNA library. Approximately one million plaques were screened, and 200 positive clones were obtained. Thirty clones were sequenced, and all contained the ATH‐3 sequence.
Isolation of Xenopus ATH‐3 cDNA
The 685 bp HincII–PstI fragment of the mouse ATH‐3 gene was used as a probe for screening the Xenopus stage 30 embryo cDNA library (Stratagene). Four positive clones were isolated and sequenced, and all contained ATH‐3 cDNA.
In situ hybridization and Northern blot analysis of ATH‐3
In situ hybridization experiments were performed, essentially as described previously (Harland, 1991; Hatada et al., 1995; Takebayashi et al., 1995). Digoxigenin‐labelled antisense RNAs corresponding to the 658 bp PvuII–HindIII mouse ATH‐3 genomic fragment and the 634 bp XbaI–EcoRI Xenopus ATH‐3 cDNA fragment were synthesized in vitro. These probes were hybridized to whole mouse embryos, 10 mm cryostat sections or Xenopus albino embryos.
For Northern blot analysis, 15 μg of total RNA was electrophoresed on a formamide/1.2% agarose gel and transferred to a nylon membrane. The filter was hybridized at 42°C with the 32P‐labelled HincII–EcoRI fragment of mouse ATH‐3 cDNA and washed, as previously described (Shimizu et al., 1995).
Analysis of RNA‐injected Xenopus embryos
The full‐length coding region of the Xenopus ATH‐3 cDNA was subcloned into pSP64T vector, and capped ATH‐3 RNA was produced in vitro as described before (Krieg and Melton, 1984). The ATH‐3 and β‐galactosidase RNAs were injected into one cell of two‐cell or four‐cell stage Xenopus embryos. At various stages, the injected embryos were subjected to histological study with haematoxylin‐eosin (HE) staining, immunological study with NEU‐1 monoclonal antibody (Itoh and Kubota, 1989) and anti‐myosin monoclonal antibody MF 20 (Bader et al., 1982) or in situ hybridization of N‐CAM, twist and keratin. The probe regions for N‐CAM, twist and keratin are nucleotide residues 1466–2020 (Accession nos. M76710), 155–652 (M27730) and 615–1171 (M11940), respectively. To identify the injected side, X‐gal staining was performed.
For HUA treatment, injected embryos were transferred to 20 mM hydroxyurea–0.15 mM aphidicolin at stage 10.5, as previously described (Harris and Hartenstein, 1991).
Xenopus animal cap assay
Mutations were introduced into ATH‐3 cDNA by hybridizing oligonucleotides 5′‐AAGTCCATGCATTCTATCACGCTCCCTGGCATTG GC‐3′ and 5′‐AAGTCCATGCATTCTGTTACGCTCCCTGGCATTGGC‐3′ with heat‐denatured ATH‐3 cDNA for S89D and S89N, respectively. Each mutant was cloned into pSP64T and sequenced.
In vitro synthesized ATH‐3, S89D or S89N RNA was injected into both cells of two‐cell stage Xenopus embryos (1 ng/embryo). Animal caps that had been either injected or uninjected with RNA were explanted at the late blastula stage (stage 9) and cultured in Steinberg's solution for 3 h, 1 day or 3 days. Total RNA was extracted from animal caps by using the acidic guanidine thiocyanate method. Reverse transcription–PCR (RT–PCR) analysis for XANF‐1 (Zaraisky et al., 1992), EF‐1α (Krieg et al., 1989), N‐CAM (Kintner and Melton, 1987), XlHbox6 (Wright et al., 1990), NF‐M, F‐spondin (Ruiz i Altaba et al., 1993), XIF3 (Sharpe et al., 1989) and S‐actin (Stutz and Spohr, 1986) was done, as previously described (Hemmati‐Brivanlou and Melton, 1994; Hatada et al., 1995; Sasai et al., 1995). In addition, we performed RT–PCR by using the following primers; opsin (Saha and Grainger, 1993), 5′‐TTCGGATGGTCCAGATACATCC‐3′ and 5′‐GGTGGTAAGAGATTCCTGTTGC‐3′ (36 cycles); and type II β‐tubulin (Good et al., 1989), 5′‐ATTAACAAGTCGTGGCAGCC‐3′ and 5′‐TCTGGACATTGCATCTACC‐3′ (36 cycles).
We would like to thank Dr S.R.Nash for reading the manuscript, Dr K.Tomita for technical assistance, Dr H.Y.Kubota for NEU‐1 antibody, Dr T.Obinata for anti‐myosin antibody MF 20 and Dr H.Ohkubo for communicating his results before publication. This work was supported by research grants from the Ministry of Education, Science and Culture of Japan, Special Coordination Funds for Promoting Science and Technology, and the Yamanouchi Foundation to R.K. K.T. is a research fellow of the Japan Society for the Promotion of Science.
- Copyright © 1997 European Molecular Biology Organization