Department of Biochemistry and Genetics, The Medical School, University of Newcastle upon Tyne, Newcastle upon Tyne, NE2 4HH, UK

Correspondence to: Andrew D Sharrocks, Department of Biochemistry and Genetics, The Medical School, University of Newcastle upon Tyne, Newcastle upon Tyne, NE2 4HH, UK

^aCurrent address: Unilever Research, Colworth, Shambrook Bedford, MK44 ILQ, UK

^bCurrent address: School of Biological Sciences, Stoppard Building, University of Manchester, Oxford Road, Manchester. MI3 9PT, UK.

^cLA Brown and SH Yang contributed equally to this work

The ternary complex factor (TCF) subfamily of ETS-transcription factors represent key nuclear targets of the MAP kinase pathways. Members of this subfamily are classified by the presence of several conserved domains for DNA binding, interaction with SRF, interaction with MAP kinases and transcriptional activation. In this study we have isolated a further member of this subfamily (TCF-1) from zebrafish. The protein product of zebrafish TCF-1 (zTCF-1), shares sequence similarity with the mammalian TCFs in all four conserved domains, with highest overall similarity to SAP-1. Zebrafish TCF-1 is expressed throughout zebrafish embryonic development and exhibits typical TCF DNA binding characteristics, with the B-box being required for interaction with SRF. Of the mammalian TCFs, its DNA binding specificity resembles Elk-1. zTCF-1 is a target for both the growth factor/mitogen-activated and stress-activated MAP kinase cascades in vitro and in vivo. However, differential targeting occurs, with the profile of its activation closely resembling that of mammalian SAP-1. Together, our results demonstrate that the TCFs have been functionally conserved during vertebrate development.

ETS-domain; ternary complex factor; transcription factors; zebrafish

Introduction

The ternary complex factors (TCFs) represent a subfamily of the ETS-domain transcription factors (reviewed in Treisman, 1994). In mammals, there are three known TCFs, Elk-1, SAP-1 and SAP-2/ERP/Net. Members of this subfamily are characterized by a high degree of sequence conservation in their ETS DNA-binding domains and in three additional conserved domains; the B-box, D-domain and C-domain (see Figure 1 and 3a). The B-box is required for efficient ternary complex formation with SRF and the c-fos SRE (Dalton and Treisman, 1992; Janknecht et al., 1994; Hill et al., 1993; Price et al., 1995; Giovane et al., 1994; Lopez et al., 1994; Shore and Sharrocks, 1994). In Elk-1, the D-domain has been shown to act as a docking motif for the ERK and JNK MAP kinases (Yang et al., 1998a,b) and the C-domain is a MAP kinase-regulated transcriptional activation domain (TAD) (reviewed in Treisman, 1994).

In addition to forming ternary DNA-bound complexes with SRF, TCFs exhibit an autonomous, SRF-independent binding activity to higher affinity binding sites (reviewed in Treisman, 1994). In comparison to SAP-1, Elk-1 exhibits a more stringent DNA binding specificity (Shore and Sharrocks, 1995) whereas SAP-2 also differs and exhibits an intermediate specificity (Shore et al., 1996; ALB, unpublished data). Both autonomous and SRF-dependent DNA binding by TCFs are controlled by cis-acting negative regulatory domains. In Elk-1 and SAP-1, the B-box appears sufficient to inhibit DNA binding although further inhibitory effects are exerted by C-terminal amino acids (Dalton and Treisman, 1992; Janknecht et al., 1994). In the case of SAP-2, DNA binding appears to be inhibited by a combination of the C-terminal TAD and an additional domain, the net inhibitory domain (NID), which is located immediately C-terminal to the B-box (Giovane et al., 1994; Lopez et al., 1994; Price et al., 1995; Maira et al., 1996). However, ternary complex formation by SAP-2 is inefficient in comparison to Elk-1 and SAP-1 (Price et al., 1995). In the case of Elk-1 and SAP-1, phosphorylation of their C-terminal TAD by MAP kinases, leads to the relief of inhibition and the activation of their DNA binding activity (Sharrocks, 1995; Gille et al., 1992, 1995a,b; Whitmarsh et al., 1995, 1997).

All three mammalian TCFs can be phosphorylated and activated by MAP kinases in vitro and in vivo. However, the response of each TCF to individual growth factor-regulated (ERK) and cytokine/stress-inducible (JNK and p38) MAP kinase pathways differs. Elk-1 is efficiently phosphorylated in vitro and activated in vivo by ERK, JNK and p38, p38₂ and p38 (reviewed in Treisman, 1996; Price et al., 1996; Janknecht and Hunter, 1997; Whitmarsh et al., 1997; Enslen et al., 1998) although in the case of the p38 isoforms, targeting via the D-domain does not occur (Yang et al., 1998b). In contrast, SAP-1 is a good target for ERK and p38 but is a poor JNK substrate (Whitmarsh et al., 1995, 1997; Price et al., 1996; Strahl et al., 1996). SAP-2 also appears to only be a good substrate for all the ERK and p38 MAP kinase isoforms (Price et al., 1996).

The in vivo role of the TCFs is currently poorly understood although one of their functions is to regulate the expression of immediate-early genes in response to various inductive stimuli (reviewed in Treisman, 1994; Cahill et al., 1996). Elk-1 has been shown to be expressed in an active form in cells of the nervous system within the brain, suggesting a role in regulating neuronal gene expression (Sgambato et al., 1998). In this study, we have isolated a cDNA encoding a member of the TCF subfamily from zebrafish (TCF-1) which is expressed during early development. The high degree of sequence similarity and DNA binding properties of the protein product of zebrafish TCF-1 (zTCF-1) demonstrate that it is conserved at both the sequence and functional levels with its mammalian counterparts. Moreover, in common with mammalian TCFs, zTCF-1 is activated and phosphorylated in a differential manner by a series of MAP kinase pathways and thereby can respond to diverse signals mediated by growth factors, mitogens, cytokines and stress. The identification and characterization of a TCF homologue from zebrafish is an important step in understanding the role of this subfamily of ETS-domain transcription factors in mediating the action of MAP kinase pathways during vertebrate development.

Results

Cloning of a member of the TCF subfamily of ETS-domain proteins from zebrafish

A cDNA encoding part of the conserved ETS DNA-binding domain of a putative zebrafish TCF homologue (TCF-1) was amplified from an embryonic zebrafish cDNA pool using degenerate oligonucleotides and used to screen an embryonic cDNA library. One of the isolated cDNAs (»1.9 kb) was sequenced and shown to contain an open-reading frame of 1329 nucleotides encoding a protein of 443 amino acids with a predicted molecular mass of 47 kDa (Figure 1a).

The amino acid sequence of the protein product of zebrafish TCF-1 (zTCF-1) shares strong homology with the mammalian members of the TCF subfamily of ETS-domain proteins (Figure 1a). In particular, high amino acid sequence conservation is exhibited within the ETS-domain (82, 80 and 76% identical to SAP-1, SAP-2 and Elk-1 respectively) with further strong similarity within the other three functional domains conserved in this subfamily (the B-box and D- and C-domains). The mammalian TCFs are known targets of several MAP kinase cascades (reviewed in Treisman, 1996; Whitmarsh and Davis, 1996; Janknecht and Hunter, 1997; Price et al., 1996; Whitmarsh et al., 1997; Strahl et al., 1996). Five potential phosphoacceptor sites for MAP kinases are conserved amongst zTCF-1 and the mammalian family members (Figure 1a). All these sites in the mammalian TCFs have been shown to be phosphorylated by one or more of the MAP kinases in vitro and in vivo.

Overall, the protein product of zebrafish TCF-1 shows strongest identity to human SAP-1a (56% overall identity) with 47 and 38% overall identity observed between SAP-2 and Elk-1 respectively. Furthermore, phylogenetic analysis suggests that zebrafish TCF-1 is more closely evolutionarily related to SAP-1 than to the other two mammalian TCFs (Figure 1b). From primary sequence comparisons, zebrafish TCF-1 is therefore most likely to represent a homologue of mammalian SAP-1.

Expression of zebrafish TCF-1 during embryonic development

In order to demonstrate that zebrafish TCF-1 is expressed during early development, its expression was detected using RT - PCR (Figure 2). As a control, its expression profile was compared to EF-1 which is a ubiquitously expressed translation elongation factor. Initial experiments were carried out to ensure that the PCR reaction was in the linear range (Hair and Sharrocks, data not shown). Zebrafish TCF-1 transcripts could be detected at low levels during the earliest stages examined (0 - 2 h post fertilization (hpf)). The expression of TCF-1 increased substantially after 3 hpf and was maintained during the first 48 h of development (Figure 2, lanes 2 - 7). The relative expression levels of zebrafish TCF-1 at each stage closely mirrored that of EF-1, albeit at a lower level. In order to detect its spatial and temporal expression pattern, whole mount in situ hybridization experiments were carried out. However, no specific zebrafish TCF-1 expression pattern could be detected (data not shown; see Discussion).

DNA binding properties of zTCF-1

The mammalian TCFs Elk-1 and SAP-1 efficiently form ternary complexes with SRF and the c-fos SRE. Ternary complex formation requires a combination of protein-DNA interactions by the ETS-domain and binding to SRF via the B-box (reviewed in Treisman, 1994). SAP-2 also exhibits ternary complex forming ability but its efficiency is much reduced in comparison to SAP-1 and Elk-1 unless C-terminal truncations are made (Price et al., 1995). All the TCFs can bind to the high affinity E74-like binding motifs in an SRF-independent manner (Dalton and Treisman, 1992; Janknecht et al., 1994; Lopez et al., 1994; Giovane et al., 1994; Whitmarsh et al., 1995; Price et al., 1995).

The DNA binding properties of zTCF-1 were investigated on the c-fos SRE and the E74 site (Figure 3). A series of truncated zTCF-1 proteins were produced to analyse the contribution of the conserved domains to DNA binding (Figure 3a). Autonomous binding of the zTCF-1 proteins to the SRE in the absence of SRF could not be detected (data not shown). However, in the presence of SRF, ternary zTCF-1-SRF-SRE complexes could be detected (Figure 3b). Full-length zTCF-1 formed weak ternary complexes (not visible in Figure 3b; see Figure 5). However, a series of C-terminally truncated derivatives efficiently formed ternary complexes with the exception of zTCF1-92 in which the B-box region was removed (Figure 3b lanes 1 - 6). In the absence of SRF, all the zTCF-1 derivatives efficiently bound the E74 site (Figure 3c) except the full-length protein which bound this site with reduced affinity (not visible in Figure 3c; see Figure 5).

Together these results demonstrate that the ETS-domain is sufficient to mediate zTCF-1 binding to high affinity sites but that the B-box is required to form ternary complexes with SRF and the c-fos SRE. The C-terminal end of the protein appears to represent an inhibitory domain as its removal stimulates complex formation on both the E74 site and the SRE. zTCF-1 therefore shares many of the DNA binding properties exhibited by mammalian TCFs (see Discussion).

The mammalian TCFs exhibit different DNA binding specificities with Elk-1 showing a more stringent specificity than either SAP-1 or SAP-2 (Shore and Sharrocks, 1995; Shore et al., 1996). The binding of zTCF-1 to a series of sites containing ets-motifs was therefore tested. zTCF-1 bound to these sites with differing efficiencies (Figure 4a). In comparison to Elk-1 and SAP-1, the relative binding efficiencies of zTCF-1 closely mirrors those exhibited by Elk-1 (Figure 4b). SAP-1 exhibits a lower site-selectivity and binds all the tested sites better than either zTCF-1 or Elk-1 (Figure 4b; Brown et al., 1998a). The DNA binding specificity of zTCF-1 is therefore most similar to that of human Elk-1 (see Discussion).

In vitro phosphorylation and activation of zTCF-1 by MAP kinases

The mammalian TCFs are known targets of both the mitogen-activated (ERK) and stress-inducible (JNK and p38) MAP kinase cascades (reviewed in Treisman, 1996; Whitmarsh and Davis, 1996; Janknecht and Hunter, 1997; Price et al., 1996; Whitmarsh et al., 1997; Strahl et al., 1996). zTCF-1 was therefore tested as a potential substrate for representative members of the different classes of MAP kinases. Elk-1 is efficiently phosphorylated in vitro by ERK2, JNK2, p38 and p38 (Figure 5b, lane 1). However, in comparison, zTCF-1 and SAP-1 are only efficiently phosphorylated by ERK2 and p38 (Figure 5b lanes 2 and 3).

ERKs have been demonstrated to interact directly with Elk-1 (Rao and Reddy, 1994; Cano et al., 1995; Yang et al., 1998a) by binding to the D-domain (Yang et al., 1998a). The binding of ERK2 to the C-terminal domains of Elk-1, zTCF-1 and SAP-1 was investigated by incubation of the ERK2 with GST fusion proteins immobilised on glutathione-agarose beads. Specifically bound proteins were obtained by co-precipitation with the beads followed by removal of non-specifically bound proteins by extensive washing. Bound kinases were subsequently detected by incubation of the final precipitates with ³²P-ATP to detect phosphorylation of the GST fusion proteins. Binding of ERK2 was detected to Elk-1, zTCF-1 and SAP-1 (Figure 5c lanes 1 - 3). Peptide competition assays using a peptide derived from the Elk-1 D-domain have been used to demonstrate the binding of ERK2 to its substrate occurs via the D-domain (Yang et al., 1998b). Similarly, incubation of ERK2 with the D-domain peptide blocks its ability to efficiently phosphorylate zTCF-1 and SAP-1 (Figure 5d), indicating that targeting of ERK2 to zTCF-1 is required for efficient phosphorylation.

DNA binding by mammalian TCFs is activated by phosphorylation (Sharrocks, 1995; Gille et al., 1992, 1995a, b; Whitmarsh et al., 1995, 1997). In vitro translated full-length zTCF-1 binds DNA inefficiently (Figure 3), therefore we tested whether phosphorylation by MAP kinases stimulates its DNA binding activity. Phosphorylation of zTCF-1 by ERK2 stimulates its ability to bind autonomously to the high affinity E74 site (Figure 5e lanes 1 - 2) and to form ternary complexes with SRF and the c-fos SRE (Figure 5e lanes 3 - 4).

Together these results demonstrate that zTCF-1 is selectively phosphorylated by different classes of MAP kinases in vitro which leads to the activation of its DNA binding activity.

Activation of zTCF-1 by MAP kinase cascades in vivo

Mammalian TCFs also exhibit different responses to individual growth factor/mitogen-activated and stress-inducible MAP kinase cascades in vivo (reviewed in Treisman, 1996; Whitmarsh and Davis, 1996; Janknecht and Hunter, 1997, Price et al., 1996; Whitmarsh et al., 1997; Strahl et al., 1996). For example, JNKs only poorly phosphorylate and activate SAP-1 in comparison to Elk-1 (Whitmarsh et al., 1995; Gille et al., 1995b; Strahl et al., 1996; Price et al., 1996). The response of zTCF-1 to different classes of MAP kinases in vivo was therefore tested (Figure 6).

EGF activates the endogenous ERK pathway in COS7 cells and stimulates transcriptional activation by Elk-1 (Yang et al., 1998a). Similarly, both SAP-1 and zTCF-1 are activated by EGF stimulation (Figure 6a).

The stimulation of CHO cells with IL-1 leads to the activation of Elk-1 by the JNK pathway and SAP-1 by an unknown kinase (Figure 6b; Whitmarsh et al., 1997). zTCF-1 is also activated in response to IL-1 stimulation (Figure 6b). However, this activation is not inhibited by the JNK inhibitor protein JIP-1 (Figure 6b) or by a dominant negative form of MKK4 (data not shown), indicating that in common with SAP-1, zTCF-1 is activated by a novel IL-1-inducible kinase.

Elk-1 (Whitmarsh et al., 1997; Price et al., 1996) and SAP-1 (Price et al., 1996; Janknecht and Hunter, 1997) can also be activated by the p38 MAPK kinase cascades in vivo. Elk-1 is activated by both p38 and p38 when cotransfected with a constitutively active version of the upstream kinase MKK6 (Figure 6c). In contrast, zTCF-1 and SAP-1 are activated by p38 but neither is significantly activated by p38 (Figure 6c).

Collectively, these results demonstrate that zTCF-1 is differentially activated by different classes of growth factor/mitogen- and stress-activated MAP kinase cascades in vivo in a manner which closely resembles the responses of mammalian SAP-1 to these pathways.

Discussion

Members of the TCF subfamily of ETS-domain transcription factors are important nuclear mediators of the cellular responses to the activation of MAP kinase pathways (reviewed in Treisman, 1994, 1996; Whitmarsh and Davis, 1996; Sharrocks et al., 1997). In this study, we have isolated a cDNA encoding a further family member from zebrafish (TCF-1) that exhibits sequence similarity in four conserved domains which are characteristic of this subfamily. In common with its mammalian counterparts, the protein product of zebrafish TCF-1, zTCF-1, represents a target for diverse MAP kinase cascades. Both sequence and functional conservation of the TCFs points to a critical, fundamental role for these transcription factors during vertebrate development.

Conservation of TCF domain structure

Significant sequence similarity is observed between zTCF-1 and the mammalian TCFs, in all four of the conserved functional domains. Overall, the greatest sequence similarity is observed with SAP-1 (56% identity), whilst Elk-1 exhibits the least similarity (38%). This pattern is repeated within each conserved domain with SAP-1 exhibiting the greatest sequence identity in each case (ETS-domain, 82%; B-box, 81%; D-domain, 73%; C-domain, 87%). Furthermore, in addition to the five MAP kinase sites conserved in all the TCFs, five additional potential MAP kinase sites are conserved within the C-terminal domains of zTCF-1 and SAP-1. An additional region appears to be conserved amongst zTCF-1, SAP-1 and SAP-2 which acts as a negative inhibitory domain in Net (the murine SAP-2 homologue) and is thought to adopt a helix - loop - helix like structure (Maira et al., 1996). The high overall similarity to SAP-1 and within the important functional domains of the protein, suggests that zTCF-1 is most likely a homologue of mammalian SAP-1.

DNA binding properties of zTCF-1

In common with the mammalian TCFs, zTCF-1 can bind DNA either autonomously or in an SRF-dependent manner on weaker binding sites. Autonomous binding is mediated by the ETS-domain alone and the binding specificity of zTCF-1 closely resembles that of Elk-1 which exhibits a more stringent binding specificity than either SAP-1 or SAP-2 (Figure 4, Shore and Sharrocks, 1995; Shore et al., 1996; Brown et al., 1998a). It was previously demonstrated that the presence of amino acid D69 in the Elk-1 ETS-domain (V68 in SAP-1) is mainly responsible for generating this stringent binding specificity (Shore et al., 1996). Similarly, zTCF-1 contains the same residue at the equivalent position in the ETS-domain (D68; Figure 1).

Ternary complex formation with SRF and the c-fos SRE requires the presence of the B-box region of zTCF-1 (Figure 3b). However, it is interesting to note that the full-length protein inefficiently forms DNA-bound complexes both in the presence or in the absence of SRF. Such negative regulation has been observed in all the mammalian TCFs although different regions have been implicated in mediating this repressive effect (Dalton and Treisman, 1992; Janknecht et al., 1994; Price et al., 1995; Maira et al., 1996). In the case of zTCF-1, deletion of the C-terminal 60 amino acids is sufficient to relieve this inhibition. In contrast to SAP-2/Net, no further negative regulatory regions appear to be present towards the N-terminus of the protein. In the case of Elk-1 and SAP-1, MAP kinase-mediated phosphorylation activates their DNA binding (Sharrocks, 1995; Gille et al., 1992, 1995a,b; Whitmarsh et al., 1995, 1997) although the mechanism of SAP-2 activation is unclear. Similarly, phosphorylation of zTCF-1 stimulates the binding activity of the full-length protein (Figure 5e).

Activation of zTCF-1 by MAP kinase cascades

The mammalian TCFs are known targets of both the mitogen-activated (ERK) and stress-inducible (JNK and p38) MAP kinase cascades (reviewed in Treisman, 1996; Whitmarsh and Davis, 1996; Janknecht and Hunter, 1997; Price et al., 1996; Whitmarsh et al., 1997; Strahl et al., 1996). zTCF-1 is also an in vitro MAP kinase substrate and a target for these cascades in vivo (Figures 5 and 6). However, in common with the mammalian TCFs, zTCF-1 appears to be differentially targeted by different classes of MAP kinases. In comparison to Elk-1, zTCF-1 is efficiently phosphorylated by ERK2, and p38 but is a poor substrate for JNK2 and p38. Similarly, in vivo, zTCF-1 is only efficiently activated by ERK2 and p38. This activation profile closely resembles that of SAP-1. In addition, IL-1 activation of both zTCF-1 and SAP-1 takes place by a cascade which is distinct from the JNKs (Figure 6b; Whitmarsh et al., 1997). As the highest conservation of potential MAP kinase sites in and around the C-terminal activation domain is also observed with SAP-1, the signalling pathways that converge on these transcription factors and their mechanisms of activation are likely to be very similar. Peptide competition assays (Figure 5d) indicate that in common with Elk-1 (Yang et al., 1998b), phosphorylation of both SAP-1 and zTCF-1 by ERK is facilitated by the presence of the D-domain. Therefore docking of the kinase onto a site distinct from the phosphoacceptor motifs appears to be an evolutionarily common mechanism in the TCF subfamily.

zTCF-1 is a homologue of which mammalian TCF?

The sequence conservation and activation profile by MAP kinase cascades suggest that zTCF-1 represents a direct homologue of mammalian SAP-1. However, the DNA binding specificity more closely resembles that of Elk-1. It is possible that other mammalian TCFs exist which exhibit stronger sequence similarity to zTCF-1. In this regard, it is interesting to note that the other family members cloned from zebrafish (PEA3, 61% identity and Fli-1, 75% identity) (Brown et al., 1998a,b) exhibit substantially higher sequence similarity than zTCF-1 to their closest mammalian counterparts. The existence of other zebrafish TCFs is suggested from the sequence of cDNAs encoding part of the ETS-domain isolated using degenerate PCR primers (our unpublished data). These additional TCFs might represent homologues of other mammalian TCFs.

Potential role of zTCF-1 during development

To date, little is known about the roles of TCFs in regulating transcription during vertebrate development although one of their functions is thought to be to regulate the expression of immediate-early genes in response to various inductive stimuli (reviewed in Treisman, 1994; Strahl et al., 1996). In adults, mammalian TCFs appear to be ubiquitously expressed with each tissue expressing the individual TCFs to different relative amounts (Price et al., 1995; Magnaghi-Jaulin et al., 1996). Recently, Elk-1 has been shown to be expressed in an active form in cells of the nervous system within the brain (Sgambato et al., 1998). Zebrafish TCF-1 is expressed throughout early development. However, we could not detect a specific expression pattern by whole-mount in situ hybridization, indicating that its expression levels may be too low (data not shown). Similarly, mammalian TCFs are expressed at low levels (Price et al., 1995). As the relative expression levels of TCF-1 closely resembles that of the ubiquitously expressed EF-1 (Figure 2), TCF-1 may also be expressed ubiquitously in the zebrafish embryo, as demonstrated for its mammalian counterparts in a variety of cell lines. A low level ubiquitous expression pattern would explain our inability to detect specific expression patterns in the early embryo.

MAP kinase pathways and their nuclear substrates have been implicated in several important regulatory processes (reviewed in Treisman, 1996; Whitmarsh and Davis, 1996). Furthermore, several developmental events in lower eukaryotes have been shown to be regulated by MAP kinase cascades via ETS-domain transcription factors. However, the role of such pathways and transcription factors in vertebrate development has not been addressed. The identification of a zebrafish TCF homologue provides the basis for future studies to investigate the role of these transcription factors in transmitting signals from MAP kinase pathways into a transcriptional response during early vertebrate development.

Materials and methods

Cloning of zebrafish TCF

A random-primed Zap cDNA library, prepared from 20 - 40 h-old zebrafish embryos, was screened with a mix of two partial zebrafish cDNAs encoding the central region of the ETS DNA-binding domain of two potential TCF homologues as described elsewhere (Brown et al., 1998b). One clone was purified which contained a cDNA insert corresponding to zebrafish TCF-1. A clone encoding a second TCF was not isolated. pAS253 (encoding full-length zTCF-1) was created by excision of the pBluescriptSK⁺ vector by co-infection with helper phage. pAS256 is identical to pAS253 except that the zebrafish TCF-1 cDNA is inserted into the EcoRI site of pBluescriptSK⁺ in the opposite orientation.

RT - PCR

Embryos of appropriate stages were isolated and dechorionated before being frozen in liquid nitrogen. Total RNA was extracted using the RNeasy extraction kit according to the manufacturer's instructions (QIAGEN). In each case, 30 mg of tissue was used (representing 20 - 35 embryos). Samples were normalized based on the concentration of the 18S rRNA band and equal amounts (»0.5 g) were used in random hexanucleotide primed reverse transcription reactions followed by a 32 cycle PCR (Plant et al., 1996) using the following primer pairs: EF-1 [EFzebra1 (5'-TTCCGTCTGCCACTTCAGGAT-3' ) and EFzebra2 (5'-CAACGGTCTGCCTCATGTCA-3'). T_a=57°C, giving a product of 563 bp] and zTCF-1 [ADS603 (5'-CCAACCAACAAACATCTCC-3') and ADS604 (5'-CTCACTGTCAATCACCAGC-3'). T_a=52°C, giving a product of 526 bp]. PCRs were optimized to ensure that the amplification was in the linear range and the products visualized on an ethidium bromide stained 3% NuSieve agarose gel (Flowgen).

Plasmid constructs

pAS351 (encoding GST fused to zTCF-1 amino acids 311 - 443) was constructed by ligating a BamHI/XbaI cleaved PCR product (primers ADS267/326) into the same sites in pGEX-KG (Guan and Dixon, 1991). pAS466 (encoding C-terminally Flag-tagged full-length zTCF-1) was constructed by ligating a NcoI/XhoI cleaved PCR product (primers ADS293/325) into the same sites in pET-nef-PFH (Zhao and Narayan, 1993). pAS545 (encoding GST-Elk-310; Elk-1 amino acids 310 - 428) (Yang et al., 1998a) and pGEX-SAP-290 (encoding GST-SAP-290; SAP-1 amino acids 290 - 431) (Whitmarsh et al., 1995) have been described previously.

The following plasmids were prepared for producing proteins by in vitro transcription/translation. pAS316 (encoding zTCF-1 amino acids 1 - 92; zTCF1-92) and pAS317 (encoding zTCF-1 amino acids 1 - 162; zTCF1-162) were constructed by ligating NcoI/BamHI and NcoI/NotI cleaved PCR products (primers ADS318/319 and ADS309-319 respectively) into the same sites in pAS37 (Sharrocks et al., 1993a). All PCRs to generate zTCF-1-derived constructs were carried out using pAS253 as a template.

For transient transfection assays, the following constructs were used: pAS572 (encoding GAL-Elk-205 [Elk-1 amino acids 205 - 428] under the control of a CMV promoter) (Yang et al., 1998a). pAS855 (encoding GAL-SAP-290 [SAP-1 amino acids 290 - 431] under the control of a CMV promoter), was constructed by inserting a HindIII/XbaI fragment from pGAL-SAP-1 (Whitmarsh et al., 1997) into the same sites of pCMV5. pAS856 (encoding GST fused to zTCF-1 amino acids 310 - 443) was constructed by ligating a BamHI/XbaI cleaved PCR product (primers ADS267/268) into the same sites in pAS769. pAS857 (encoding the GAL4 DNA binding domain fused to zTCF-1 amino acids 310 - 443) was constructed by ligating a BamHI/XbaI fragment from pAS857 into the same sites in pSG424. pAS858 (encoding GAL-zTCF-310 - 443 driven by a CMV promoter) was constructed by inserting a HindIII/XbaI fragment from pAS857 into pCMV5.

pCMV5-F-p38 (Raingeaud et al., 1995), pCDNA3-F-p38 (Enslen et al., 1998), pCDNA3-F-MKK6(Glu) (encoding constitutive-active MKK6) (Whitmarsh et al., 1996), pCDNA3-F-DN-MKK4 (encoding dominant-negative MKK4) (Whitmarsh et al., 1995), pCDNA3-F-JIP-1(JBD) (encoding the JNK binding domain of JIP-1) (Dickens et al., 1997), pG5E1b-luc (Seth et al., 1992) have been described previously.

The nucleotide sequences of all the PCR-derived constructs were verified by di-deoxy sequencing.

Protein production

Purification and expression of GST-fusion and hexahistidine-tagged proteins were carried out essentially as described previously (Shore et al., 1995; Yang et al., 1998a). In vitro translated proteins were synthesized using the TNT coupled in vitro transcription/translation system (Promega) from either pBluescriptKS⁺-derived plasmids or from linear PCR products under the control of a T3 promoter. The primer pairs ADS238/ADS261, ADS238/265 and ADS238/274 were used on the template pAS253 to synthesize PCR products encoding the proteins zTCF1-213, zTCF1-383 and zTCF1-303 respectively. ³⁵S-labelled proteins were analysed by electrophoresis through 0.1% sodium dodecylsulphate (SDS)-12% polyacrylamide gels before visualization and quantification by autoradiography and phosphorimaging (BAS 1500 Phosphorimager and TINA 2.08e software, Fuji).

Gel retardation analysis

Gel retardation assays were performed as described previously (Sharrocks et al., 1993b) using 5% polyacrylamide gels cast in 0.5´Tris-borate-EDTA (TBE). DNA binding sites were derived from annealed synthetic oligonucleotides (E74 and SRE) (Shore and Sharrocks, 1994) or from cloned sites from binding site-selection procedures (Shore and Sharrocks, 1995; Brown et al., 1998a). Individual DNA-binding sites derived from the site-selection procedure were synthesized and labelled by PCR as described previously (Shore and Sharrocks, 1995). Binding reactions were set up to achieve <50% binding of the free DNA to allow quantification of relative DNA binding affinities. To account for slight differences in the amounts of each binding site added in each experiment, results were normalized by dividing by the total DNA present in the binding reaction.

Cell culture, transfection and extract preparation

COS-7 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (FBS) (Gibco BRL). CHO cells were maintained in F12 medium supplemented with 5% foetal bovine serum (FBS) (Gibco BRL). Transfection experiments were carried out using Superfect transfection reagent (Qiagen) as described previously (Yang et al., 1998a).

For reporter gene assays, GAL4-driven promoters were co-transfected with various vectors encoding GAL4-fusion proteins. The activities of the GAL4 DNA binding domain (amino acids 1 - 147), and GAL4-fusion proteins (50 ng of plasmid DNA) were measured in cotransfection assays in all cell lines using 1 g of reporter plasmid pG5E1bLuc. Transfection efficiencies were normalized by measuring the activity from a cotransfected plasmid (0.5 g) which expresses -galactosidase (pCH110, Pharmacia KB Biotechnology Inc.). Cell extracts, were prepared and luciferase and -galactosidase assays were carried out as described previously (Yang et al., 1998a).

Protein kinase assays

Recombinant JNK and p38 MAP kinases were prepared as described previously (Yang et al., 1998b). Recombinant active ERK2 was obtained from New England Biolabs (NEB). The kinase assays were routinely carried out in 20 l reaction volumes using GST-fusion proteins as described previously (Yang et al., 1998a). Full-length purified recombinant His-tagged zTCF-1 was used a substrate for subsequent use in gel retardation assays. The phosphorylation of substrate proteins was examined following SDS - PAGE by autoradiography, and quantified by phosphorimaging (Fuji BAS1500; TINA 2.08e software). Peptide competition experiments were carried out essentially as described above except for preincubation of 50 - 5000 pmoles of the peptide competitor (N-KGRKPRDLELPLSPSLLGGPGPE-C) with MAP kinases before the kinase reactions. Final peptide concentrations were 5 - 50 M (10 - 100-fold excess over GST - TCF substrates). Binding and phosphorylation assays were carried out as described previously (Yang et al., 1998a).

Note added in proof

The Accession number for the zTCF-1 cDNA sequence is AC, AJ249170; ID, DRE249170.

Acknowledgements

We thank Margaret Bell and Katherine Stewart for excellent technical and secretarial assistance, Bob Liddell for DNA sequencing and oligonucleotide synthesis. We are grateful to members of our laboratories for stimulating discussions and comments on the manuscript. We are also grateful to Alan Whitmarsh and Roger Davis for reagents. This work was supported by grants from the North of England Cancer Research Campaign, Cancer Research Campaign [CRC], the Wellcome Trust, a Newcastle University Research Committee Studentship to LA Brown and a Jeffcock Studentship to A Galanis. AD Sharrocks is a Research Fellow of the Lister Institute of Preventative Medicine.

Brown AL, Amores A, Schilling TF, Jowett T, Baert J-L, de Launoit Y and Sharrocks AD. (1998a). Oncogene 17, 93-104. MEDLINE

Brown AL, Schilling TF, Rodaway A, Jowett T, Ingham PW, Patient R and Sharrocks AD. (1998b). Mech. Dev. (Submitted).

Cahill MA, Janknecht R and Nordheim A. (1996). Curr. Biol. 6, 16-19. MEDLINE

Cano E, Hazzalin CA, Kardalinou E, Buckle RS and Mahadevan LC. (1995). J. Cell. Science 108, 3599-3609. MEDLINE

Dalton S and Treisman R. (1992). Cell 68, 597-612. MEDLINE

Dickens M, Rogers J, Cavanagh J, Raitano A, Xia Z, Halpern JR, Greenberg ME, Sawyers C and Davis RJ. (1997). Science 277, 693-696. Article MEDLINE

Enslen H, Raingeaud J and Davis RJ. (1998). J. Biol. Chem. 273, 1741-1748. MEDLINE

Gille H, Kortenjann M, Thomae O, Moomaw C, Slaughter C, Cobb MH and Shaw PE. (1995a). EMBO J. 14, 951-962. MEDLINE

Gille H, Sharrocks AD and Shaw PE. (1992). Nature 358, 415-417.

Gille H, Strahl T and Shaw PE. (1995b). Curr. Biol. 5, 1191-1200. MEDLINE

Giovane A, Pintzas A, Maira S-M, Sobieszcuk P and Wasylyk B. (1994). Genes Dev. 8, 1502-1513. MEDLINE

Guan K and Dixon JE. (1991). Anal. Biochem. 192, 262-267. MEDLINE

Hill CS, Marais R, John S, Wynne J, Dalton S and Treisman R. (1993). Cell 73, 395-406. MEDLINE

Janknecht R and Hunter T. (1997). EMBO J. 16, 1620-1627. MEDLINE

Janknecht R, Zinck R, Ernst WH and Nordheim A. (1994). Oncogene 9, 1273-1278. MEDLINE

Lopez M, Oettgen P, Akbarali Y, Dendorfer U and Liberman TA. (1994). Mol. Cell. Biol. 14, 3292-3309. MEDLINE

Magnaghi-Jaulin L, Masutani H, Lipinski M and Harel-Bellan A. (1996). FEBS Lett. 391, 247-251. MEDLINE

Maira S-M, Wurtz J-M and Wasylyk B. (1996). EMBO J. 15, 5849-5865. MEDLINE

Plant KE, Hair A and Morgan GT. (1996). Nucl. Acids Res. 24, 3514-3521.

Price MA, Cruzalegui FH and Treisman R. (1996). EMBO. J. 15, 6552-6563. MEDLINE

Price MA, Rogers AE and Treisman R. (1995). EMBO J. 14, 2589-2601. MEDLINE

Raingeaud J, Gupta S, Rogers JS, Dickens M, Han J, Ulevitch RJ and Davis RJ. (1995). J. Biol. Chem. 270, 7420-7426. MEDLINE

Rao VN, Huebner K, Isobe M, Ar-rusida A, Croce CM, Reddy ESP. (1989). Science 244, 66-70. MEDLINE

Rao VN and Reddy ESP. (1994). Oncogene 9, 1855-1860. MEDLINE

Seth A, Gonzalez FA, Gupta S, Raden DL and Davis RJ. (1992). J. Biol. Chem. 267, 24796-24804. MEDLINE

Sgambato V, Vanhoutte P, Pagès C, Rogard M, Hipskind R, Besson M-J and Caboche J. (1998). J. Neuroscience 18, 214-226.

Sharrocks AD. (1995). FEBS Lett. 368, 77-80. MEDLINE

Sharrocks AD, Brown AL, Ling Y and Yates PR. (1997). Int. J. Biochem. Cell Biol. 29, 1371-1387. MEDLINE

Sharrocks AD, Gille H and Shaw PE. (1993b). Mol. Cell. Biol. 13, 123-132.

Sharrocks AD, Von Hesler F and Shaw PE. (1993a). Nucl. Acids Res. 21, 215-221.

Shore P and Sharrocks AD. (1994). Mol. Cell. Biol. 14, 3283-3291. MEDLINE

Shore P and Sharrocks AD. (1995). Nucl. Acids Res. 23, 4698-4706.

Shore P, Bisset L, Lakey J, Waltho JP, Virden R and Sharrocks AD. (1995). J. Biol. Chem. 270, 5805-5811. MEDLINE

Shore P, Whitmarsh AJ, Bhaskaran R, Waltho J, Davis RJ and Sharrocks AD. (1996). Mol. Cell. Biol. 16, 3338-3349. MEDLINE

Strahl T, Gille H and Shaw PE. (1996). Proc. Natl. Acad. Sci. USA 93, 11563-11568. MEDLINE

Treisman R. (1994). Curr. Opin. Genet. Dev. 4, 96-101. MEDLINE

Treisman R. (1996). Curr. Opin. Cell. Biol. 8, 205-215. MEDLINE

Whitmarsh AJ and Davis RJ. (1996). J. Mol. Med. 74, 589-607. MEDLINE

Whitmarsh AJ, Shore P, Sharrocks AD and Davis RJ. (1995). Science 269, 403-407. MEDLINE

Whitmarsh AJ, Yang S-H, Su MS-S, Sharrocks AD and Davis RJ. (1997). Mol. Cell. Biol. 17, 2360-2371. MEDLINE

Yang S-H, Whitmarsh AJ, Davis RJ and Sharrocks AD. (1998b). EMBO J. 17, 1740-1749. Article MEDLINE

Yang S-H, Yates P, Whitmarsh AJ, Davis RJ and Sharrocks AD. (1998a). Mol. Cell. Biol. 18, 710-720. MEDLINE

Zhao L-J and Narayan O. (1993). Gene 137, 345-346. MEDLINE

Figure 1 (a) Alignment of the amino acid sequences of zebrafish TCF-1 and human Elk-1 (Rao et al., 1989), SAP-1a (Dalton and Treisman, 1992) and SAP-2 (Price et al., 1995). Residues which are identical in all four proteins are indicated by asterisks. The highly conserved B-box and ETS-, D-, and C-domains are boxed. An additional region conserved amongst zTCF-1, SAP-1 and SAP-2 is boxed and shaded. Potential MAP kinase phosphorylation sites conserved amongst all four proteins are highlighted. (b) Clustal analysis (PC/GENE) of the phylogenetic relationship amongst zebrafish TCF-1, and mammalian Elk-1, SAP-1 and SAP-2

Figure 2 Expression of zebrafish TCF-1 and EF-1 during early embryonic development. RT - PCR was carried out with specific primer pairs for zebrafish TCF-1 (top panel) and EF-1 (bottom panel) with RNA samples from embryos at preMBT (0 - 2 h), blastula (3 h), gastrula (6 h), segmentation1(10 - 12 h), segmentation (12 - 14 h), 24 h, 48 h. The times given are hours post fertilization (hpf). The negative control in lane 8 was carried out in the absence of RNA

Figure 3 DNA binding by zTCF-1. (a) Schematic illustration of the domain structure of the full-length and truncated proteins used in (b) and (c). Numbers indicate the positions of residues with respect to the full-length proteins. (b) and (c) Gel retardation analysis of wild-type and truncated zTCF-1 proteins bound to the c-fos SRE (b) or E74 (c) binding sites. Binding to the SRE was carried out in the presence of core^SRF. The position of binary SRF - SRE complexes is indicated (2°). Arrows represent ternary zTCF-1-SRF-SRE (b) and binary zTCF-1-E74 (c) complexes formed by the indicated `full-length' zTCF-1 derivatives. (*) Represents a complex formed by a C-terminally degraded/truncated zTCF-1 protein whereas (# ) represents complexes formed by a protein from the rabbit reticulocyte lysate. Equal molar quantities of each protein are used in the binding reactions

Figure 4 The DNA binding specificity of zTCF-1. (a) Gel retardation analysis of zTCF1-92 binding to a panel of ets binding sites. (b) Quantification of the data from part (a) in comparison to the binding of the equivalent DNA-binding domains from Elk-1 (Elk-1-93) and SAP-1 (SAP1-92) to the same series of sites (Brown et al., 1998a). Data are quantified relative to the binding of each protein to the S27 site (taken as 100%). The central six nucleotides of the ets-motifs are boxed

Figure 5 zTCF-1 is an in vitro target for MAP kinases. (a) Schematic illustration of the domain structure of the GST - TCF fusion proteins which contain the C-terminal regions of Elk-1, zTCF-1 and SAP-1. (b) The phosphorylation of the indicated GST - TCF fusion proteins by MAP kinases was examined using either activated ERK2, JNK-2, p38 or p38. The activity of each protein kinase towards the substrate GST-Elk310 was standardized with respect to activated ERK2 (0.1U, NEB). Kinase assays were performed for 15 min at 30°C with equal molar quantities (5pmoles) of GST fusion proteins as substrates. (c) Binding and phosphorylation assays. Equal molar quantities (37.5 pmol) of the indicated GST fusion proteins were immobilized onto glutathione agarose beads and incubated with ERK2 (1U) for 4 h at 4°C. After extensive washing, the beads with the remaining bound kinase were incubated with [-³²P]ATP in kinase buffer for 2 h at 30°C. (d) Phosphorylation of the indicated GST - TCF fusions by ERK2 in the presence of a competitor peptide corresponding to the Elk-1 D-domain. Kinase assays were carried out as described in part (b) except the MAP kinases were preincubated in the absence (lane 1) or presence of competitor peptides (100 - 1000-fold excess over the indicated GST-fusion substrates); 1 nmoles (lane 2) and 10 nmoles (lanes 3). (e) Gel retardation analysis of non-phosphorylated (lanes 1 and 3) and phosphorylated (lanes 2 and 4) zTCF-1 binding to the c-fos E74 (lanes 1 - 2) or SRE (lanes 3 - 4) binding sites. Binding to the SRE was carried out in the presence of core^SRF. Labelling of the complexes is as described in Figure 3

Figure 6 zTCF-1 is an in vivo target for MAP kinase cascades. COS-7 (a and c) or CHO (b) cells were transfected with the indicated CMV promoter-driven GAL - TCF constructs and a GAL4-driven luciferase reporter plasmid. (a) Cells were serum starved for 12 h and either left unstimulated or stimulated with EGF. (b) Cells were left unstimulated or stimulated with IL-1. Expression vectors encoding the JNK pathway inhibitor protein JIP-1(DBD) were cotransfected where indicated. (c) Cells were cotransfected with a constitutively active form of MKK6 on its own or in combination with p38 or p38 expression vectors. Transfection efficiency was monitored by using the -galactosidase expression vector pCH110. The luciferase activities relative to the unstimulated cells containing pCMV5-GAL-Elk-205 (means±s.d.; n=3) are presented