SRF during early development of Xenopus laevis - NCBI

2 downloads 66 Views 2MB Size Report
Jan 16, 1991 - cDNA probe; Nick Hopwood and Jonathan Sleeman (CRC, Cambridge) ... Jackson,R.J. and Hunt,T. (1983) Methods Enzymol., 96, 50-74.
The EMBO Journal vol.10 no.4 pp.933-940, 1991

Expression of genes encoding the transcription factor SRF during early development of Xenopus laevis: identification of a CArG box-binding activity as SRF T.J.Mohun, A.E.Chambers, N.Towers and M.V.Taylor1 Laboratory of Developmental Biochemistry, National Institute for Medical Research, The Ridgeway, Mill Hill, London NW7 1AA and 'CRC Molecular Embryology Group, Department of Zoology, Downing Street, Cambridge CB2 3EJ, UK

Communicated by J.R.Tata

cDNA clones encoding the sequence-specific DNA binding protein, serum response factor (SRF), have been isolated from a Xenopus laevis neurula library and their nucleotide sequence determined. The Xenopus SRF (SRFX) gene produces multiple-sized transcripts, present at 105 copies per unfertilized egg. A similar level is detected in the embryo during early cleavage, but SRFX transcripts accumulate rapidly following gastrulation. The protein they encode is similar in sequence to human SRF in its central and carboxy-terminal regions, but possesses a divergent amino-terminal portion. We have previously described a Xenopus embryo sequence-specific binding activity that recognized the CArG motif of the cardiac actin gene promoter. Here we show that the DNA-binding characteristics of synthetic SRFX are indistinguishable from those of the embryo factor. Moreover, antiserum raised against the synthetic SRFX recognizes this factor. Together, these results establish that the same factor binds to elements required for constitutive transcription in Xenopus oocytes, musclespecific gene expression in Xenopus embryos and serum-responsive transcription in cultured amphibian cells. Key words: CArG box/myogenesis/serum response/ transcription factorIXenopus

Introduction Exposure of cultured mammalian cells to growth factors and wide variety of environmental stimuli results in the rapid and transient transcription of 'immediate early' genes (Lau and Nathans, 1985, 1987; Almendral et al., 1988), including several proto-oncogenes such as c-fos, c-myc and c-jun (reviewed in Bravo, 1990). Since several of these genes encode regulatory proteins, it is likely that their activation directly results in a change in cellular gene expression in response to the initial stimulus. The mechanisms by which extracellular signals trigger this genetic cascade and the nature of the genes involved are of considerable interest since they may identify critical steps in the control of cell proliferation. Activation of c-fos transcription in response to serum mitogens in cultured cells has provided a model for the study of these events. A conserved DNA sequence, first identified in the c-fos promoter and subsequently found in several other 'immediate early gene' promoters, mediates

a

Oxford University Press

transcriptional response of the c-fos gene and has been termed the serum response element (SRE) (see Treisman, 1990, for a review). The SRE is a binding site for a ubiquitous, nuclear phosphoprotein, the serum response factor (SRF), which participates in the activation of c-fos transcription (Treisman, 1987; Prywes et al., 1987; Schroter et al., 1987). In embryonic development, fertilization is followed by a period of cell proliferation during which the cells become restricted in their developmental fate and commence differentiation into distinct cell types. Little is known about the regulation of embryonic cell proliferation, but embryological and biochemical studies suggest that polypeptide growth factors are involved in establishing the earliest regional differences in the embryo (Smith, 1990; Ruiz-i-Altaba and Melton, 1990). The molecular basis of their action is unknown and it is therefore important to establish whether stimulation of 'immediate early' genes in cultured cells provides a model for the activity of growth factors in embryogenesis. In this study we have examined the expression of the Xenopus SRF gene during early development in the amphibian, Xenopus laevis. cDNAs encoding the Xenopus homologues of human SRF have been isolated, levels of SRF-encoding transcripts measured through early development and the sequence of the Xenopus protein compared with its human counterpart. Several lines of evidence have suggested that in addition to regulating serum-responsive transcription in fibroblasts, SRF may play a role in the terminal differentiation of muscle cells. All vertebrate sarcomeric actin genes possess several copies of a conserved, 10 nucleotide motif, the CArG box, within their promoters (Minty et al., 1986; Bergsma et al., 1986; Mohun et al., 1986, 1988; Miwa et al., 1987; Muscat et al., 1987). Mutagenesis studies have shown that the CArG box is an essential element for transcription from cardiac and skeletal actin gene promoters in muscle cells (Miwa and Kedes, 1987; Mohun et al., 1989b; Walsh, 1989). A similar sequence lies at the centre of the SRE. In vivo assays demonstrate that these two promoter elements are to some extent functionally interchangeable in embryos (Taylor et al., 1989) and in cultured cells (Walsh, 1989; Taylor et al., 1989; Tuil et al., 1990) suggesting that they might provide binding sites for the same transcriptional activator(s). In vitro binding studies are consistent with this view. SREs will compete with CArG boxes for binding to an SRF-like factor present in nuclear extracts of mammalian cells (Phan-Dinh-Tuy et al., 1988; Boxer et al., 1989; Taylor et al., 1989) and both motifs can form a specific complex with purified human SRF (Taylor et al., 1989). These findings raise the possibility that SRF may act as a transcriptional activator of musclespecific actin genes. In early Xenopus embryos, sarcomeric actin genes are activated exclusively in the somitic mesoderm of gastrulating embryos (Mohun et al., 1984) but the mechanisms responsible for both temporal and spatial restriction of actin 933

T.J.Mohun et al.

gene transcription are unknown. In previous studies, we characterized a sequence-specific DNA binding activity from Xenopus embryo extracts that recognized the CArG motif of the cardiac actin gene promoter (Mohun et al., 1989b; Taylor et al., 1989) and that also formed specific complexes with SRE sequences (Mohun et al., 1989a; Taylor et al., 1989). Such behaviour would be expected for an amphibian SRF, but identification of the Xenopus embryo factor was inconclusive since it failed to react with antiserum raised against purified human SRF. Here we compare the properties of synthetic Xenopus SRF with those of the embryo CArG box-binding factor we previously reported and find that they are indistinguishable. Our results therefore indicate that in addition to a well documented role in the transcriptional activation of 'immediate early' genes, SRF protein is also involved in regulating somite-specific expression of the cardiac actin gene in early embryos.

amino acids compared with 508 in the human protein. Like its human counterpart, the frog polypeptide is relatively rich in serine (57/448), threonine (45/448), proline (30/448) and glutamine (30/448). Alignment of the two sequences (termed SRFX and SRFh) indicates that the protein divides into three distinct regions. The amino-terminal portion (SRFX residues 1 -85; SRFh residues 1 -130) is the most divergent (49% identity) and includes a large deletion within the frog sequence. A negatively charged region encompassing a casein kinase II recognition site (Norman et al., 1988; Manak et al., 1990) is retained in the frog sequence (Figure 2). The central portion of the protein (SRFX residues 86-229; SRFh residues 131 -274) is virtually identical between the frog and human proteins and encompasses the DNA binding/protein dimerization domain (Norman et al., 1988). The remaining region, comprising almost half the molecule including the carboxy-terminus (SRFX residues 231-448; SRFh residues 275 -508), is well conserved between the

Results

two

Characterization of Xenopus SRF (SRFX) cDNAs Xenopus cDNAs encoding SRF were isolated from a neurula stage cDNA library using a human SRF cDNA as probe. Three types of clone were obtained (Figure lA) and representatives characterized by DNA sequencing. XXSRF3, a 3.5 kb cDNA, encodes a complete Xenopus SRF polypeptide (Figure 1B). The predicted mRNA comprises a long 5' untranslated region of at least 556 nucleotides and 3' untranslated region of 1617 nucleotides, containing three putative polyadenylation signals (Proudfoot et al., 1976). Partial cDNAs encoding a second SRF transcript were also isolated from the neurula library. The longest of these, XXSRF9, is 7% divergent in sequence from XXSRF3 over the protein coding region, the majority of differences (57/79) being third base changes. Its 5' untranslated region shows only a limited region of homology to XXSRF3 and extends for 1500 nucleotides (Figure IA; sequence submitted to EMBL database). XXSRF9 probably originated from aberrant priming during library construction (or from incomplete second strand cDNA synthesis) since it lacks a 3' untranslated region. In contrast, a third cDNA, XSL15, is 81 % similar in sequence to the 3' untranslated region of XXSRF3 over a shared 1160 nucleotide region, but extends a further 2450 nucleotides after its poly(A) tail (Figure lA). This 3' region of XSL15 contains five possible polyadenylation signals (sequence submitted to EMBL database). Genomic Southern blotting data indicate that XSL15 cDNA derives from the same gene as XXSRF9 (data not shown) suggesting that only two SRF genes are expressed in the neurula embryo, giving rise to transcripts cloned as XXSRF3 and XXSRF9/XSL15. In order to determine the intron-exon structure of the SRF genes, we have characterized genomic X clones encoding the XXSRF3 transcript. Sequence comparison indicates that the gene contains six introns at codons 126, 215, 290, 331, 394 and 418 and the entire transcription unit is at least 11 kb in length (A.Chambers, unpublished data). A comparison of the human and Xenopus SRF proteins shows that the polypeptide sequence of SRF is highly conserved between frogs and humans (Figure 2). XXSRF3 encodes a protein of predicted Mr 46 116, containing 448 934

species (82% identity).

Expression of SRFX during early development In order to establish the size of the SRFX transcripts, total RNA from unfertilized eggs, embryos and swimming tadpoles was fractionated on denaturing agarose gels and Northern blots hybridized with a probe specific for the XXSRF3 transcript (Figure 3). Surprisingly, multiple RNAs were identified at all stages of development, ranging in size from 3 kb to > 9 kb. Identical results were obtained using a variety of specific hybridization probes from the XXSRF3 cDNA confirming that the hybridizing mRNAs were

transcribed from the same gene. Probes specific for the other SRFX transcript, XXSRF9, produced a similar hybridization pattern (data not shown) indicating that the two genes

are coexpressed approximately equally during early development. Multiple bands may result from the detection of unspliced or partially spliced SRF transcripts (cf. Norman et al., 1988) and indeed cDNAs corresponding to incomplete

spliced transcripts were isolated from the neurula cDNA library (T.Mohun, unpublished data). In addition, several of the potential polyadenylation sites may be utilized during RNA processing to give a variety of different sized mRNAs. SRFX transcripts are inherited from the unfertilized egg and the levels of each sized species remains approximately constant until after gastrulation. In subsequent development to the swimming tadpole the most prevalent transcripts accumulate further, whilst the others remain constant or show a more transient increase during neurulation (Figure 3). A new sized transcript, larger than any others, appears in neurula stage embryos and its level remains approximately constant during development of the swimming tadpole. Similar, complex patterns of hybridization are obtained with RNA from a variety of adult tissues, suggesting that the SRF transcripts detected in these Northern blots are ubiquitous in their distribution (data not shown). To assess their prevalence, we used an RNase protection assay to quantify the steady state level of SRFX transcripts in early embryos (Figure 4). We estimate that the unfertilized egg contains -105 XXSRF3-like transcripts. This level increases to 106 by neurula stage and a further 5-fold by the time the embryo has developed into a swimming tadpole (stage 42). These results indicate that SRF transcripts are very abundant in the early embryo.

Xenopus SRF

in Xenopus embryos. SRFX cRNA was prepared and used to produce SRFX protein by in vitro translation in a reticulocyte lysate. Figure 5A shows that the major

The DNA binding characteristics of SRF' protein With the isolation of Xenopus SRF cDNAs we re-examined the identity of a CArG box-binding factor previously detected

I I,MA ~ ~ ~

,

A vXSRF3

81%

93%

XSL1 5

XXSRF9 B

I

TGTGGCCGGTCAGGGAGAGGGAGAGCGGCCAGGGCCACCGGAACAAACCAAATCCGTCAGTGACACGGAGGACAACGAAACCAGCCGAAG CATCTAGGACGGGCAGGGAGCACCGGAACAAACCAAGTTCCTGGGGTAGAGGGGAAACTGGGCTCCCGAGCCAACCACAGGCCACAGGAC GAACCACAGCTCCGGTAGAGGGGGAACCATATCTGCAAGGAGGGAGCCCAGGTGTGAGGAGCCTAGAGGGGAAGATATTTAGCAGTAGCC CCCAGGGAGCGGTACTTGCAGGATTCTATACAGGGAAGGGCCAGTTAACCGACTGTACCCAAGTAAAGGGACCAGAAAGACATTTTCCCT AAGAAAAGGCTCAGCGGGAACATATAAGTTTTTAGCACAGGGCAGATTGTAAGTAGTCATTACCCAGGAGGGGCAGGTTAACTCTTATCC TCACAGCATTATACACAAGTGGGTTTTGCCCAAGGAAGGGCAGATTAGTAAATTTTATCCCAGGAGTGGGCACAATATTGGGGCACACCC M

L

P

S

N

Q

A

G

V

A

G

S

G

R

G

L

G

L

A

R

R A

G

N

G

CCCATGCCCCCAAGGGATGCTGCCCAGTAACCAGGCCGGGGTGGCGGGGTCAGGCCGAGGGCTGGGGTTGGCCCGCAGGGCAGGAAATGG S

G

S

S

E

S

G

A

G

C

P

G

I

R

G

P

G

G

Q

Y

S

G

R

G

I

K

R

G

L

N

E

L

E

L

P

E

Q

V

S

G

A

K

P

G

K

K

T

R

G

R V

K

F

S

K

R

K

T

G

I

M

K

K

A

Y

E

L

T

G

H

V

Y

T

F

A

T

R

K

L

Q

P

M

I

T

S

S

P

D

S

P

P

R

S

D

P

S

T

D

Q

R

M

S

A

S

E

S

D

S

S

G

E

T

K

D

S

L

K

P

A

F

T

V

T

N

L

P

V

P

T

T

S

T

S

M

Q

V

S

S

G

H

S

F

P

I

T

N

Y

L

A

T

V

L

K

T

E

G

G

A

M

S

S

G

G

L

M

Q

I

P

A

G

V

A

Q

Q

V

P

V

Q

A

I

Q

V

H

S

A A

Q

A

S

T

S

S

S

G

T

V

T

L

P

A

A

I

M

T

S

S

V

P

P

H

A V

M

Y

A

P

T

Q

G

L

T

D

G

G

L

A V

L

N

V

S

H

T

Q

E

Q

G

G

V

Q

Q

V

F

L

T

A

P

P

G

T

L

H

Q

M

T

V

I

G

Q

Q

S

S

G

S

N

K

N

D

S

E

E

E

D

P

P

G

A A

G

V

V

G

Y

P

G

A

S

G

K

M

E

F

I

D

N

K

L

R

R

Y

T

T

L

T

G

T

Q V

L

L

L

V

A

S

T

G

K

A

L

I

Q

T

C

L

900

E

CTTCAGCAAGCGCAAAACTGGCATCATGAAGAAGGCCTATGAGCTTTCCACCCTGACTGGCACCCAGGTGCTGCTGTTAGTTGCCAGTGA E

810

T

AGTGTCAGGGGCCAAACCTGGGAAGAAAACTCGGGGCCGAGTGAAGATTAAGATGGAATTTATTGATAACAAATTGCGGCGCTACACGAC S

720

T

GCGGGGGATAAAGAGGGGGCTCAATGAGCTGGAGCTGCCAGAACAAGGGGGGGCAGCAGGAGTTGTTGGGTACCCAGGGGCTAGTGGAAC I

630

G

GGCTGGGTGCCCAGGTATTAGGGGCCCTGGGGGCCAGTACAGCGGTAGTTCAGGGAGCTCAGAGTCGGGGGAAGAGGAGGATCCTCCTGG G

90 180 270 360 450 540

990

N

GACCGGCCATGTGTACACATTTGCCACGCGCAAGTTGCAGCCCATGATCACCAGCGAGACGGGGAAAGCTTTGATCCAAACCTGCCTGAA 1080 T

G

F

E

E

T

D

L

T

Y

Q V

TTCCCCTGACTCTCCTCCCCGCTCGGACCCCTCTACTGACCAGCGCATGAGTGCCACCGGCTTTGAGGAGACCGACCTCACCTACCAGGT 1170 G

T

S

N

I

Q T

ATCTGAGTCTGACAGCAGCGGGGAGACCAAGGATTCGCTAAAGCCGGCATTCACCGTGACCAACCTTCCAGGCACATCCAACATCCAAAC 1260 P

V

T

S

A

N

G

AGTGCCCACCACCTCCACTTCCATGCAAGTCAGCAGCGGGCACTCCTTTCCCATCACCAACTACCTGGCACCAGTGACCAGCGCCAACGG 1350 F

T

L

M

S

A G

G

A

CACCGTGCTGAAAACTGAGGGTGGAGCTATGAGCTCAGGGGGATTAATGCAAATTCCCGCAGGCTTCACTCTTATGTCAGCAGGGGGAGC 1440 P

S

S

D

S

S

S

E

L

V

Q

TGTAGCTCAGCAGGTACCTGTCCAGGCTATTCAGGTCCATTCTGCAGCTCAGGCATCGCCGTCCAGTGACAGTAGTTCAGAGTTGGTGCA 1530 T

T

V

S

G

H

M

M

Y

P

S

AACATCCTCTAGTGGAACAGTAACACTGCCTGCAGCCATCATGACCTCCTCCGTGCCCACAACAGTGAGTGGTCATATGATGTACCCCAG 1620 A

F

S

S

A

P

Q

M

Q

TCCCCATGCAGTAATGTATGCCCCTACACAAGGACTAACCGATGGGGGCCTCGCTGTGCTCAATGCCTTCTCCTCTGCGCCCCAAATGCA 1710 V

Q

I

P

V

S

A

V

Q

AGTTTCTCACACCCAGGAACAGGGGGGTGTCCAGCAAGTTTTCCTCACTGCTCCCCCTGGCACTGTCCAGATACCTGTGTCTGCAGTACA 1800 L

T

E

L

Q

V V

N

L

D

T

S

N

S

N

GCTCCACCAGATGACAGTTATTGGGCAGCAGTCCAGCGGCAGTAACCTGACTGAGCTGCAAGTGGTTAATTTGGACACGTCGAACAGCAA 1890

CAAGAATGACTAAAGCGCAGGGTGGGTGGCCCCACACTTTATTGACTCAAAACCTATTTATTGTTGCCTTTTTCACACGTTTCTTTAATC CTCTCCGGCTGGTGCCGTGGCACAAGGACTGGTAGTTATCTGGATGGGGTCACGCTGCTTCACCCCCCCCCCCCAAGCCAAAGAGCTCTA AACCTTCGTGTCTTTCTCCACTTTACAGCAGCAGAAAGGCCAACGCCCTGTTGGAGTGCGCTCTTTGGGATCCGCAGGGGCGGCGCTGGA ACCAGGCGAGGAGGATTTGCACTGTGCCTGTGTGATTAGACTTGGGCTTCTACCTGTTCATGTTGGGCTTGGCCTGCTGTTGGCATTAGA CACTGGGTATTCTGTGTTGAAGTTTACCCGGCAAAGCATCACATATGGGCCCCTAGATATACTTGTAGAGCCCCCAACAAGCCATTGCTA GGAAAACAGAAGGAGAAGGGACCCTAAAATTCCCTAAGAGAAGACTATATTTTTCTAATAGTTGAGTCTCAGTATTTATGTATTGAGGGT TTGTGATGCCTCAGTTTTTACGGCGAGGAGATTTTACTGTCTGTATAGCCAATACATCCCTCTGATGGGAATGTATACACTCATGCACAA TATTGTATTGCTTTGCAGGGGTCTTAGTGAGGAGCACTACAGGATTTCAGTGCCCATCCAATCACAAGTACTGTAATTAGGGTACAAATG GCTGGCCACGTTCCATCACGGAGGACAGGAGCCATCTTTGTGTCATTTCTGCTGCAACCCTGTGTGTGCAAGTTGTGTGAGTCTGTGGGA AGACAATGGCACGTGACTGGGGGTGGGTTGGCAATGGGCAGAACTTATCAGACTCAGGGTGTTGGCGCAGAAGCGTGTTGGGGGTTAATA GCGTGCTTTTGGTTGAAGAGAAATATATATATATTTTCCAGTTTCATTTGTACAGTCTGTGCTGTGTTTGGTTTGCTGGTGTGCGAGTTT TTTTACTCAACTATTAACCCATTCAGCCACCCTCTGCTCTGGTTTCTGGCAGTTAATCCCTTTTGTGCAGCAGCCGCATCTCTCCTGTAG TTTCTCTGCTCGGGATGCAAGTGAATCCTTTTGCCATTGCTTAGTTTGATGGATCCTGTGAGTAAGTTTATTAGCTGTGCAGGTATTGCA GCTTCCAAAGTCAATGGGAAAGTGCCTAATCCAATTATGCCCAGCTCTGCGCAACATACATTCAGGTTGTAGTTATAGAAGGCAGAGATG AATAAATCAGTGTAGCGCCTCTCCTGGCCACATAGTAGTTACAAGACTCCCCTCCTGGCCACATCGTAGTTACCTGGCGTTCCTCCTGGC

CACTGCGCATCTATGTGGCATCTATGTGGCGCCTGGTCACCACCCCCCTTTCCGGGCCACAGCATCTGGTTGCTCCCCTTCACAGGTCAC ATATTGTAGAACTGCATTTCCGATCCAGTGTTTGTATTTAACATGCAGTCCATGGTAGGTCTGAAATGAAAAAGTATCTGTGAGTG

1980 2070 2160 2250 2340 2430 2520 2610 2700 2790 2880 2970 3060 3150 3240 3330 3420

CTAATCTTCCACAATAGAACTGGGGGTCTTTGTCCCTTATCCTTTCTGGGGTGGTAG3GAAAGGCAGAGACCCCCTAAAAATAAAATAAT3510 TTACTTCAAAAAAAAAA

Fig. 1. (A) Xenopus SRF-encoding cDNAs isolated from a neurula library. The cDNAs are aligned with respect to each other. Open boxes represent protein-coding regions. Vertical bars indicate EcoRl restriction sites. Dotted lines delimit regions of sequence similarity (shown as a percentage value). Bar represents 500 bp. (B) Complete nucleotide sequence of Xenopus SRF cDNA (XXSRF3). Possible polyadenylation signals are shown in bold.

935

1

T.J.Mohun et al. SRFH MLPT.QAGAAAALGRGSALGGSLNRTPTGRPGGGGGTRGANGGRVPGNGAG50

111. 111 SRFX MLPSNQAG .....................

l

1:1.1 1

)

VAGSGRG15 L

LGPGRLEREAAAAAATTPAPTAGALYSGSEGDSESGEEEELGAERRGLKR 100

1.1I1:.:1:.......:1:I111.1.1111111:.:

11:11

LG... LARRAGNGAGCPGIRGPGGQYSGSSGSSESGEEEDPPG..RGIKR 60 D V V V

SLSEMEIGMVVGGPEASAAATGGYGPVSGAVSGAKPGKKTRGRVKIKMEF 150

GLNELEL.....PEQGGAAGVWGYPGASGTVSGAKPGKKTRGRVKIKMEF 105 I (A)GG

V

IDNKLRRYTTFSIKKTGIMKKAYELSTLTGTQVLLLVASETGHVYTFATR 200

IDNKLRRYTTFSaKRTGIMKKAYELSTLTGTQVLLLVASETGHVYTFATR 155 KLQPMITSETGKALIQTCLNSPDSPPRSDPTTDQRMSATGFEETDLTYQV 250

KLQPMITSETGKALIQTCLNSPDSPPRSDPSTDQRMSATGFEETDLTYQV 205 C

SESDSSGETKDTLKPAFTVTNLPGTTSTIQTAPSTSTTMQVSSGPSFPIT 300 11111111111.111111111111 11.111.1.111.111111.11111 SESDSSGETKDSLKPAFTVTNLPG. TSNIQTVPTTSTSMQVSSGHSFPIT

254

NYLAPVSASVSPSAVSSANGTVLKSTGSGPVSSGGLMQLPTSFTLM. PGG 349 VTSANGTVLK.TEGGAMSSGGLMQIPAGFTLMSAGG

NYLAP ...

294

T

S

AVAQQVPVQAIQVHQAPQQASPSRDSSTDLTQTSSSGTVTLPATIMTSSV 399 AVAQQVPVQAIQVHSAA.QASPSSDSSSELVQTSSSGTVTLPAAIMTSSV 343 L

HS

PTTVGGHMMYPSPHAVMYAPTSGLGDGSLTVLNAFSQAPSTMQVSHSQVQ 449 PTTVSGHMMYPSPHAVMYAPTQGLTDGGLAVLNAFSSAP.QMQVSHT..Q 390 S

*

EPGGVPQVFLTASSGTVQIPVSAVQLHQMAVIGQQAGSSSNLTELQVVNL 499 I.

11.11111

..

11111111111

i.11111

::11111111111

EQGGVQQVFLTAPPGTVQIPVSAVQLHQMTVIGQQ.SSGSNLTELQVVNL 439

Fig. 3. Expression of multiple SRF-encoding transcripts during early Xenopus development. Northern blot hybridized with a Xenopus SRF cDNA probe. Each lane contained 10 yig of total RNA. (E) Unfertilized egg, (B) blastula (stage 6-8), (G) gastrula (stage 10-12), (N) neurula (stage 17-19), (Tb) tailbud (stage 22-26), (Tp) tadpole RNA (stage 46). The size of transcripts was estimated from the mobility of Xenopus ribosomal markers (arrows). This filter was stripped and rehybridized with a Xenopus cytoskeletal actin cDNA fragment. This probe contained 518 nucleotides of protein coding sequence and cross hybridized with sarcomeric actin mRNAs, which are smaller than cytoskeletal actin mRNA and are prominent from neurula stage onwards.

DTAHSTKSE 508 I I .:

.I. :

DTSNSNKND 448

i )eve lopi n e r i %

C

p

At¢.

Fig. 2. Comparison of Xenopus and human SRF polypeptide sequences. Sequence alignment was obtained using the GAP program (Devereux et al., 1984). A conserved casein kinase II recognition site (ESGEEEE/D) and two potential kinase A sites (RRYT and KRKT) are shown in bold. The second Xenopus SRF polypeptide (predicted from the sequence of XXSRF9) is very similar to that shown (XXSRF3); differences are indicated below the frog sequence, including the insertion of an extra alanine residue (shown in brackets) at position 75. The cDNAs encoding the second protein are incomplete and no data are available for the predicted amino acid sequence beyond residue 381 (marked by an asterisk).

translation product is a polypeptide of apparent Mr 54 000. This is significantly larger than the predicted size from sequence analysis (Mr -46 100) and probably results from modification of the protein by the cell-free extract. A similar discrepancy has been noted for the cell-free translation product of human SRF (Norman et al., 1988). Synthetic Xenopus SRF was then examined in binding studies using an electrophoretic mobility shift assay (EMSA) to establish if its behaviour resembled that found for the embryo CArG box-binding factor described in earlier studies (Taylor et al., 1989). The results (shown in Figure SB) demonstrate that synthetic SRF formed a complex with probes comprising either the CArG box 1 motif from the cardiac actin gene promoter, or the SRE of the Xenopus c-fos gene. Like the embryo factor, the synthetic protein showed a greater affinity for the SRE than for CArG box 1 (Figure SB, cf. lanes 1 and 5). In each case, binding was specific as judged by competition with unlabelled probe DNA. Again, like the embryo factor (Taylor et al., 1989), formation of the SRFX/CArG

3.1taridard

owl

-

936

F XK.SRF3

xPressi'r

Fig. 4. RNase protection assay to monitor the level of SRFX transcript, XSRF3, during early development. The position of the 375 nucleotide fragment obtained from full length protection of the probe is indicated (arrow). (P) Undigested probe, (C) tRNA control, (E) unfertilized egg, (B) blastula, (G) gastrula, (N) neurula, (Tb) tailbud, (Tp) tadpole, (A6) Xenopus A6 cultured fibroblast RNA. Sense orientation transcripts derived from the same template as the probe were used in similar hybridizations to provide standards for quantitation of SRF mRNA. These gave a longer protected fragment than the authentic SRF mRNA due to the presence of polylinker sequence complementary to that of the probe.

Xenopus SRF

B A a

2

b*f

4w

.-~~~~~~~~~~~~~~~A

3

4

. _ , ;^.---xl-

5

7

8 9 10 1'

-

+~.-.)-

Fig. 5. (A) The in vitro translated product of SRFX cRNA migrates as a 56 kd polypeptide. Synthetic RNA from the plasmid pSP64T.XSRF was translated in vitro in a rabbit reticulocyte lysate. The 35S-labelled protein was analysed by SDS-PAGE and visualized by autoradiography. Lane 2 shows the products of in vitro translation of a 5-fold greater amount of SRFX RNA than lane 1. In the autoradiograph exposure shown, no 35Slabelled product was detectable in mock translated reticulocyte lysate analysed in an adjacent lane. (B) The DNA binding sequence specificity of SRFX protein. EMSA of 4 fmol SRFX protein with 4 fmol of the following end-labelled DNA binding site probes: CArG box I (lanes 1 -4); Xenopus c-fos SRE (lanes 5 and 6); CArG box 2 (lanes 7 and 8); CArG box 3 (lanes 9 and 10); CArG box 4 (lanes 11 and 12). Competitor DNAs were included in the binding reactions at the following molar excesses: lane 2, lOx CArG box 1; lane 3, 1OX ACT.L; lane 4, lOx ACT.L*; lane 6, lOx Xenopus c-fos SRE; lane 8, 30x CArG box 2; lane 10, 30x CArG box 3; lane 12, 30x CArG box 4. (C) The SRFx/CArG box 1 complex comigrates with the Xenopus embryo factor/CArG box I complex. EMSA of CArG box I DNA probe with: lane 1, SRFX made by in vitro translation; lane 2, whole cell extract from neurula (stage 18) embryos; lane 3, nuclear extract from neurula (stage 18) embryos; lane 4, nuclear extract from HeLa cells.

box 1 complex was strongly competed by ACT.L, an SRE variant that is a high affinity site for human SRF (Treisman, 1987), but not competed at all by ACT.L*, a variant which is a very poor binding site for the human protein. Both SRFX and the embryo factor also formed a specific complex, albeit rather weakly, with two more CArG motifs (CArG boxes 3 and 4) from the cardiac actin promoter; neither bound to the fourth copy of this motif, CArG box 2 (Figure SB, cf. lanes 7 and 8 with lanes 9- 12; Mohun et al., 1989b). Taken together, these data demonstrate that the embryo factor is indistinguishable from SRF' in its DNA binding characteristics. We next compared the relative migration of the SRFX/ CArG box 1 complex in an EMSA with that obtained using Xenopus embryo extracts and the same DNA probe. Identical mobiities were seen with the synthetic protein and with both whole cell and nuclear extracts from neurula embryos, indicating that the embryo protein was of a similar size to synthetic SRFX (Figure SC, lanes 1-3). The complex migrates further than that obtained using HeLa cell nuclear extract (lane 4), which is consistent with the smaller size of the Xenopus SRF protein compared with its human counterpart. Identification of a major Xenopus embryo CArG box-binding factor as Xenopus SRF Finally, we demonstrated conclusively that the previously characterized CArG box-binding activity in early embryos was indeed SRF, using antisera raised against Xenopus SRF protein prepared from the cloned DNA in a bacterial expression system. A glutathione S-transferase - SRFX fusion polypeptide containing residues 175 -448 of SRFX was used

to raise rabbit antisera against the Xenopus protein. Immunoprecipitation by one of these antisera against SRFX protein made from synthetic SRFX RNA, either by in vitro translation in a reticulocyte lysate or after injection into Xenopus oocytes, is shown in Figure 6. SRFX protein was not immunoprecipitated by unimmunized rabbit serum. Moreover, the specific antiserum, but not the unimmunized rabbit serum, also immunoprecipitated a protein (Mr 56 000) from uninjected Xenopus oocytes that comigrated with SRFX (Figure 6A, lanes 7-9). A smaller polypeptide (Mr 43 000) present in oocyte extracts was also recognized by the specific antiserum but we do not know the relationship of this protein to SRFX. Other proteins detected after immunoprecipitation with the anti-SRFx serum were nonspecific, since they were also obtained using unimmunized rabbit serum. When anti-SRFx serum was added subsequent to complex formation in the binding reactions, the complex formed between the embryo factor and the CArG box 1 probe migrated with a greatly reduced mobility, producing a 'supershift', whilst the unimmunized rabbit serum had no effect (Figure 6B). These results indicate that the previously characterized CArG box-binding activity in both whole cell and nuclear extracts from stage 18 neurula embryos is antigenically related to Xenopus SRF. A CArG box 1 binding activity in nuclear extracts from adult frog heart (the tissue which expresses the cardiac actin gene in the adult) comigrates with the embryo activity (Figure 6B, cf. lanes 1, 4 and 7). The specific antiserum 'supershifts' this binding activity from adult heart, suggesting that this is also SRFX rather than an adult tissue-specific variant of SRF. The antiserum also 'supershifts' the CArG box-binding activity

937

T.J.Mohun et al.

neurula embryo contains 106 transcripts for one of these genes, which compares with 107 transcripts for both the muscle-specific cardiac actin gene (Mohun et al., 1984) and the ubiquitously expressed cytoskeletal actin gene (Mohun et al., 1989a). Each SRF gene produces multiple transcripts which are distributed throughout the embryo (data not shown). The polypeptides they encode are very similar to the human protein except for a striking divergence in their amino-terminal region. A 143 residue region, encoded within two exons of the Xenopus gene, is almost entirely conserved between the human and frog proteins and contains the DNA binding and dimerization domains. dl *~~~~~~d

Fig. 6. (A) Immunoprecipitation of SRFX. SRFX from in vitro translation reactions (lanes 1 -3), Xenopus oocytes injected with synthetic SRFX mRNA (lanes 4-6) and uninjected Xenopus oocytes (lanes 7-9) was immunoprecipitated with serum from a rabbit immunized with GST-SRFx, with (lanes 3, 6 and 9) and without (lanes 1, 4 and 7) a 'pre-clearing' step (see Materials and methods). The products were resolved by SDS-PAGE and compared with those obtained by immunoprecipitation with unimmunized rabbit serum (lanes 2, 5 and 8). In each case a 56 kd protein (arrowed) is detected with the anti-SRFx serum, but not with the unimmunized rabbit serum. Simriilar results were obtained with a second GST-SRFX antiserum and with affinity purified anti-SRFx antibody. (B) Anti-SRFx serum recognizes a CArG box 1 binding activity in Xenopus embryos, adult Xenopus heart and HeLa cells. EMSA of CArG box 1 DNA probe with whole cell extract from neurula (stage 18) embryos (lanes 1 -3) or nuclear extracts from neurula (stage 18) embryos (lanes 4-6), adult frog heart (lanes 7-9) and HeLa cells (lanes 10-12). Conditions were as follows: no serum addition (lanes 1, 4, 7, 10); addition of anti-SRFx serum (lanes 2, 5, 8 and 11) or unimmunized rabbit serum (lanes 3, 6, 9 and 12) after complex formation. Similar results were obtained with a second GST-SRFX antiserum and with affinity purified anti-SRFX antibody.

found in HeLa nuclear extracts, the source from which the human SRF was originally purified. In summary, we have found that the embryo CArG boxl-binding activity and synthetic SRFX protein show similar DNA sequence specificity, the complexes they form with a CArG box 1 probe comigrate in the EMSA and both are bound by anti-SRFx antibodies. Taken together, these results identify CArG box 1-binding factor as Xenopus SRF.

Discussion SRF mRNA is abundant in X. laevis embryos. Transcripts can be detected in the unfertilized egg and accumulate during

early embryonic development. These arise from two SRF genes, presumably reflecting the pseudo-tetraploid nature of this species (Bisbee et al., 1977). We estimate that the

938

SRF in early development What function does SRF play in the developing embryo? An obvious possibility is suggested by the observations that growth factor treatment of embryonic cells can mimic the effect of natural signals that mediate inductive cell-cell interactions. Perhaps SRF regulates transcription from embryonic equivalents of 'immediate early' genes and is a target for signal pathways activated by inducing factors. Another possibility is that SRF is involved in constitutive as well as inducible gene transcription. Purified human SRF stimulates transcription from a Xenopus cytoskeletal actin promoter in vitro (Norman et al., 1988) and the same promoter sequence is required for constitutive transcription in transfected mammalian fibroblasts and microinjected Xenopus oocytes (Mohun et al., 1987). Cytoskeletal actin genes are expressed in all regions of the embryo after gastrulation, but a microinjection assay has failed to detect any requirement for the SRE sequence (Brennan, 1990). A third possibility is that SRF is required for musclespecific actin gene expression. In an earlier study, we identified a CArG box-binding activity present in mammalian muscle cell lines as SRF by virtue of its DNA sequence specificity and its cross-reactivity with antiserum raised against the purified human SRF protein (Taylor et al., 1989). A similar binding activity was detected in Xenopus embryos, but this did not cross-react with the same antiserum (Mohun et al., 1989b; Taylor et al., 1989). We have now demonstrated that the embryo CArG box-binding activity is indistinguishable from Xenopus SRF by three criteria. Firstly, it shows the same DNA-binding sequence specificity as SRFX. Secondly, the electrophoretic mobility of the complex formed by either protein with a CArG box-binding site is identical. Thirdly, the CArG box-binding protein is antigenically similar to SRF since it is recognized by rabbit antiserum raised against a GST - SRFX fusion protein. Identical electrophoretic mobility and cross-reactivity with anti-SRFx antibodies is observed with a CArG box-binding activity present in nuclear extracts from adult Xenopus heart (Figure 6B). These results are consistent with the involvement of a single SRF protein in serum-responsive, constitutive and muscle-specific transcription and account for the functional interchangeability of SREs and CArG boxes in vivo. Whilst it is also possible that several antigenically similar, but functionally distinct, SRFs act as transcriptional regulators, we have no evidence for their existence in the frog embryo. By altering the conditions of the binding assay (most importantly, the non-specific DNA competitor), it has been possible to resolve other CArG box-binding activities in embryo extracts in addition to that documented

Xenopus SRF in this study (M.Taylor, in preparation). However, these differ from SRFX in their abundance, affinity for CArG box derivatives and the electrophoretic mobility of the protein DNA complexes they form. None appears to cross-react with the SRFX antiserum. Furthermore, by the neurula stage of development, the amphibian embryo comprises many different cell types showing diverse patterns of gene expression. If the multiple roles of SRF are in fact due to the expression of several genes encoding proteins indistinguishable by our criteria, we might expect to detect their transcripts in the differentiating embryo. However, screening of a neurula cDNA library has failed to identify any candidate SRFx-like transcripts, despite repeated attempts using low stringency hybridization conditions. Embryos do contain transcripts of more distantly related genes (unpublished data), but these do not recognize SREs or CArG motifs in vitro and their function is unknown (R.Pollock and R.Treisman, in preparation).

Modulation of SRF activity If a single SRF is involved in regulating serum-responsive, constitutive and muscle specific transcription it is necessary to account for the differing specificities of each. CArG box-containing sarcomeric actin promoters are not activated during serum stimulation of fibroblasts (Elder et al., 1984) and serum-responsive genes such as c-fos remain quiescent during differentiation of embryonic muscle. We presume that SRF activity is modulated differently according to promoter and cell type by interaction with other proteins. Three different types of protein -protein interaction are suggested from the limited evidence available. Firstly, regulatory proteins could interact directly with SRF in a ternary complex, as for example has been shown for the phosphoprotein, p62TCF (Shaw et al., 1989; Schroter et al., 1990). It will be of interest to determine whether a similar complex can be detected on muscle actin promoter CArG boxes and whether any of its components are muscle-cell specific. A second possible interaction could involve competition between SRF and other proteins for binding to the SRE/CArG sites. A candidate competitor for SRF is the protein p62/MAPF1 (Walsh et al., 1987; Ryan et al., 1989), which binds to one half of the SRE/CArG sequence. In the Xenopus embryo, several other CArG box-binding activities can be detected in addition to SRFX (M.Taylor, in preparation) and these may play a role in modulating SRFX activity at the CArG box site. Lastly, proteins binding to other sites within the promoter may affect the composition and activity of any SRE/CArG protein complex. In the c-fos gene promoter, there is evidence for interactions between SRF and proteins binding to an immediately adjacent AP1/ATF site (Fisch et al., 1988, 1989; Berkowitz et al., 1989). This type of interaction also appears to regulate muscle-specific expression from the sarcomeric actin gene promoters. Transcription of the human cardiac actin gene in transfected muscle cells requires Sp-1 and MyoDi binding sites, which lie between CArG box 1 and the TATA box (Sartorelli et al., 1990). These sites are not conserved in the Xenopus cardiac actin gene promoter, but a binding site for Xenopus MyoD 1 is located upstream of the CArG motifs. Mutagenesis studies indicate that sequences encompassing the MyoD 1 site are essential for expression of the Xenopus gene in early embryos (manuscript in preparation). We do not know the composition of the

SRF/CArG box complex in embryonic muscle cells, nor how its activity is affected by MyoD binding. With the isolation of Xenopus SRF cDNAs reported here, the nature of these interactions can now be studied.

Materials and methods Characterization of Xenopus SRF cDNAs and genomic clones A neurula stage cDNA library (Kintner et al., 1987) was screened under

conditions of low stringency using a 1367 nucleotide fragment of the human SRF cDNA (Norman et al., 1988), containing most of the protein-coding sequence (codons 82-508). Positively hybridizing phage were purified and their cDNA inserts subcloned into plasmid vectors using standard procedures (Sambrook et al., 1989). The DNA sequences of XXSRF3, XXSRF9 and XSL15 were determined using the dideoxy method (Sanger et al., 1977; Bankier et al., 1987) and compiled using the DB programs (Staden, 1982). Each nucleotide was sequenced an average of 3-4 times on each DNA strand and the consensus analysed using the Staden (1984) and UWGCG programs (Devereux et al., 1984). Clones containing the SRF gene were isolated from a X.laevis genomic library (Krieg et al., 1985) in the same way. Intron -exon boundaries were located by alignment of genomic and cDNA (XXSRF3) restriction maps and identified by sequencing of selected genomic DNA fragments. RNA analysis RNA was isolated from unfertilized Xenopus eggs, embryos (staged according to Nieuwkoop and Faber, 1967) and cultured A6 fibroblasts as described previously (Mohun et al., 1989b). Total RNA was fractionated on 1% formaldehyde-agarose gels and blotted onto GeneScreen (NEN). Hybridization probes comprised a variety of fragments derived from both coding and non-coding portions of XXSRF3 and XXSRF9. Probes were labelled using oligonucleotide primers or by synthesis from single-stranded templates and were hybridized separately (Hopwood et al., 1989). Filters were washed at 65°C in 0.1 x SSPE for 1 h and exposed to film. All probes gave similar results whether they comprised sequences common to both cDNAs or unique to each one. A 1300 bp Pvull fragment of the type 5 cytoskeletal actin cDNA (Mohun and Garrett, 1987) was hybridized to stripped filters to provide a control for integrity of RNA. This probe contains both coding and non-coding sequences with the result that it cross-hybridizes with sarcomeric actin mRNAs (see Figure 3 legend). RNase protection assays (Zinn et al., 1983; Mohun et al., 1989b) were performed using 10 Ag of total RNA in each reaction. Nuclease resistant fragments were fractionated on 6% polyacrylamide sequencing gels with Hinfl digested pBR322 as a size marker. An RNA probe was synthesized from plasmid pSP73.XSRF3 using SP6 polymerase. Plasmid pSP73.XSRF3 contains a 375 nucleotide BamHI-EcoRI fragment from XXSRF3, encoding amino acid residues 52-175 and was linearized with PvuII for RNA synthesis. For quantitative estimates of SRF mRNA, sense-orientation transcripts were synthesized in bulk from EcoRI linearized pSP73.XSRF3, using T7 polymerase. These were trace-labelled to permit calculation of transcript concentration. Serial dilutions were hybridized to the same probe as embryo RNA samples and electrophoresed in parallel.

Synthesis of SRF in vitro The entire XXSRF3 coding region (nucleotides 551 -2142) including five nucleotides of 5' untranslated region and 242 nucleotides of 3' untranslated region was reconstructed in the expression vector, pSP64T (Krieg et al., 1984), using BglII linkers. Trace-labelled, capped RNA was synthesized from the resulting plasmid (pSP64TXSRF) and translated in vitro using a rabbit reticulocyte lysate (Jackson et al., 1983) containing [35S]methionine. The yield of SRFX was estimated by TCA precipitation. In the EMSA -4 fmol of SRFX was used. Production of GST- SRF fusion protein A 2.4 kb EcoRI fragment encoding residues 53-448 of the SRFX protein was cloned into the bacterial expression vector, pGEX3T (Smith et al., 1988). The glutathione S-transferase -SRFX fusion protein (GST-SRFX) was isolated from bacterial lysates by affinity chromatography and used to raise antiserum from rabbits by standard procedures. Affinity chromatography columns of GST-SRFX and GST carrier protein coupled separately to CNBr-activated Sepharose CL4B were used to purify SRFX antibodies from serum.

Preparation of cell extracts Oocyte extracts, whole cell and nuclear extracts from neurula (stage 18) embryos and nuclear extracts from HeLa cells were made as described previously (Mohun et al., 1989b; Taylor etal., 1989). The adult heart nuclear

939

T.J.Mohun et al.

prepared similarly to embryo nuclear extracts and is described (M.Taylor, in preparation). Electrophoretic mobility shift assay (EMSA) The EMSA was carried out essentially as described previously (Taylor et al., 1989) using 4 fmol of end-labelled probe and a mnixture of salmon sperm DNA and restricted plasmid (pUC 18) as the non-specific DNA. The CArG box 1 probe was synthesized as complementary 32mer oligonucleotides (Mohun et al., 1989b) cloned via KpnI compatible ends into the KpnI site of pUC 18. A c-fos SRE probe was synthesized as a pair of 26mers (Mohun et al., 1989a), blunt-end cloned into the SniaI site of pUC 18. Both probes were obtained by excision with EcoRI and BamHI followed by end labelling with [a-32P]dATP using the Klenow fragment of DNA polymerase. Probes for CArG boxes 2, 3 and 4 were prepared as described previously (Mohun et al., 1989b). Specific competitor DNAs were as described previously (Taylor et al., 1989; Mohun et al., 1989b) except the Xenopus c-fos SRE which was identical to the labelled probe detailed above. Immunoprecipitation Full-grown oocytes (stage VI) injected with 40 ng of SRFX RNA and mixed stage, uninjected oocytes were each incubated in modified Barth X saline (MBS) containing [35S]methionine (I tCi/ttl) for 20 h at 20°C. Extracts were prepared from these as described. The equivalent of four fullgrown, injected oocytes, 12 mixed stage, uninjected oocytes or 5 fmol of SRFX made by in vitro translation was incubated with protein A-Sepharose beads and antiserum, affinity purified antibody or unimmunized rabbit serum by gentle mixing for 2 h at 4°C in RIPA (50 mM NaCl, 25 mM Tris-HCI, pH 8.2, 0.5% sodium deoxycholate, 0.5 % NP40, 0.1 S% SDS and 0.1 % sodium azide). The beads were collected by centrifugation and washed extensively with RIPA before elution of the bound protein by boiling in sample buffer and analysing by SDS-PAGE. As a 'pre-clearing' step, labelled protein was incubated with unimmunized rabbit IgG linked to Sepharose beads with gentle mixing for 1 h at 40C in RIPA, the beads removed by centrifugation and the supernatant added to the protein A-Sepharose as described above. extract was

elsewhere

Acknowledgements We thank Richard Treisman (ICRF, London) for providing the human SRF cDNA probe; Nick Hopwood and Jonathan Sleeman (CRC, Cambridge) for advice on the pGEX expression system and immunoprecipitation respectively; Jeremy Rashbass (CRC, Cambridge) for help with oocyte microinjection; and Betty Baker (NIMR), Janet Champion (NIMR) and Nigel Garrett (CRC, Cambridge) for assistance in the course of this work. We are grateful for the financial support of the Medical Research Council and

the Cancer Research

Campaign.

References Almendral,J.M., Sommer,D., Macdonald-Bravo,H., Burckhardt,J., Perera,J. and Bravo,R. (1988) Mol. Cell. Biol., 8, 2140-2148. Bankier,A.T., Weston,K.M. and Barrell,B.G. (1987) Methods Enzymol., 155, 51-93. Bergsma,D.J., Grichnik,J.M., Gossett,L.M.A. and Schwartz,R.J. (1986) Mol. Cell. Biol., 6, 2462 -2475. Berkowitz,L., Riabowol,K.T. and Gilman,M.Z. (1989) Mol. Cell. Biol., 9, 4272-4281. Bisbee,C.A., Baker,M.A., Wilson,A.C., Hadji-Azimi,I. and Fischberg,M. (1977) Science, 195, 785-787. Boxer,L.M., Prywes,R., Roeder,R.G. and Kedes,L. (1989) Mol. Cell. Biol., 9, 515-522. Bravo,R. (1990) In Seminars in Cancer Biology. Saunders Scientific Publications, London, Vol. 1, pp. 37-46. Brennan,S.M. (1990) Differentiation, 44, 111-121. Devereux,J., Haeberli,P. and Smithies,O. (1984) Nucleic Acids Res., 12, 387-395. Elder,P., Schmidt,L., Ono,T. and Getz,M. (1984) Proc. Natl. Acad. Sci. USA, 81, 7476-7480. Fisch,T.M., Prywes,R. and Roeder,R.G. (1988) Genes Dev., 2, 391-405. Fisch,T.M., Prywes,R. and Roeder,R.G. (1989) Mol. Cell. Biol., 9, 1327-1331. Hopwood,N.D., Pluck,A. and Gurdon,J.B. (1989) EMBO J., 8, 3409-3417.

Jackson,R.J. and Hunt,T. (1983) Methods Enzymol., 96, 50-74. Kintner,C.R. and Melton,D.A. (1987) Development, 99, 311-325. Krieg,P.A. and Melton,D.A. (1984) Nucleic Acids Res., 12, 7057-7070.

940

Krieg,P.A. and Melton,D.A. (1985) EMBO J., 4, 3463-3471. Lau,L.F. and Nathans,D. (1985) EMBO J., 4, 3145-3151. Lau,L.F. and Nathans,D. (1987) Proc. Natl. Acad. Sci. USA, 84, 1182-1186.

Manak,J.R., De Bisschop,N., Kris,R.M. and Prywes,R. (1990) Genes Dev., 4, 955-967. Minty,A. and Kedes,L. (1986) Mol. Cell. Biol., 6, 2125-2136. Miwa,T. and Kedes,L. (1987) Mol. Cell. Biol., 7, 2803-2813. Mohun,T. and Garrett,N. (1987) Development, 101, 393-402. Mohun,T., Brennan,S., Dathan,N., Fairman,S. and Gurdon,J. (1984) Nature, 311, 716-721. Mohun,T., Garrett,N. and Gurdon,J. (1986) EMBO J., 5, 3185-3193. Mohun,T., Garrett,N. and Treisman,R. (1987) EMBO J., 6, 667-673. Mohun,T., Garrett,N., Stutz,F. and Spohr,G. (1988) J. Mol. Biol., 202, 67-76.

Mohun,T., Garrett,N. and Taylor,M.V. (1989a) Development, 107, 835 -846.

Mohun,T., Taylor,M.V., Garrett,N. and Gurdon,J.B. (1989b) EMBO J., 8, 1153-1161. Muscat,G.E.O. and Kedes,L. (1987) Mol. Cell. Biol., 7, 4089-4099. Nieuwkoop,P. and Faber,J. (1956) Normal Table ofXenopus laevis (Daudin). North-Holland, Amsterdam. Norman,C., Runswick,M., Pollock,R. and Treisman,R. (1988) Cell, 55, 989-1003.

Phan-Dinh-Tuy,F., Tuil,F., Schewighoffer,C., Pinset,C., Kahn,A. and Minty,A. (1988) Eur. J. Biochem., 173, 507-515. Proudfoot,N.J. and Brownlee,G. (1976) Nature, 263, 211 -214. Prywes,R. and Roeder,R.G. (1987) Mol. Cell. Biol., 7, 3482-3489. Ruiz-i-Altaba,A. and Melton,D.A. (1990) Trends Genet., 6, 57-64. Ryan,W.J., Franza,B.J. and Gilman,M.Z. (1989) EMBO J., 8, 1785-1792. Sambrook,J., Fritsch,E.F. and Maniatis,T. (1989) Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory Press, Cold Spring

Harbor, NY. Sanger,F., Nicklen,S. and Coulson,A.R. (1977) Proc. NatI. Acad. Sci. USA, 74, 5463-5467. Sartorelli,V., Webster,K.A. and Kedes,L. (1990) Genes Dev., 4, 1811- 1822. Schroter,H., Shaw,P. and Nordheim,A. (1987) Nucleic Acids Res., 15, 10145-10158.

Schroter,H., Mueller,C.G., Meese,K. and Nordheim,A. (1990) EMBO J., 9, 1123-1130. Shaw,P.E., Schroter,H. and Nordheim,A. (1989) Cell, 56, 563-572. Smith,D.B. and Johnson,K.S. (1988) Gene, 67, 31-40. Smith,J.C. (1990) Curr. Opinion. Cell Biol., 1, 1061-1070. Staden,R. (1982) Nucleic Acids Res., 10, 4731-4751. Staden,R. (1984) Nucleic Acids Res., 12, 521-538. Taylor,M., Treisman,R., Garrett,N. and Mohun,T. (1989) Development, 106, 67-78. Treisman,R.H. (1987) EMBO J., 6, 2711-2717. Treisman,R. (1990) In Transcription Factors, Differentiation and Cancer. Saunders Scientific Publications, London. Seminars in Cancer Biology, Vol. 1.

Tuil,D., Clergue,N., Montaras,D., Pinset,C., Kahn,A. and Phan Dinh Tuy,F. (1990) J. Mol. Biol., 213, 677-686. Walsh,K. (1989) Mol. Cell. Biol., 9, 2191-2201. Walsh,K. and Schimmel,P. (1987) J. Biol. Chem., 262, 9429-9432. Zinn,K., Di Maio,D. and Maniatis,T. (1983) Cell, 34, 865-879. Received on December 3, 1990; revised on January 16, 1991