cDNA cloning and characterization of murine transcriptional enhancer factor-1-related protein 1, a transcription factor that binds to the M-CAT motif.

The M-CAT motif is a cis-regulatory DNA sequence that is essential for muscle-specific transcription of several genes. Previously, we had shown that both muscle-specific (A1) and ubiquitous (A2) factors bind to an essential M-CAT motif in the myosin heavy chain beta gene and that the ubiquitous factor is transcriptional enhancer factor (TEF)-1. Here we report the isolation of mouse cDNAs encoding two forms (a and b) of a TEF-1-related protein, TEFR1. The TEFR1a cDNA encodes a 427-amino acid protein. The coding region of TEFR1b is identical to 1a in both nucleotide and predicted amino acid sequence except for the absence of 43 amino acids downstream of the TEA DNA-binding domain. Three TEFR1 transcripts (approximately 7, approximately 3.5, and approximately 2 kilobase pairs) are enriched in differentiated skeletal muscle (myotubes) relative to undifferentiated skeletal muscle (myoblasts) and non-muscle cells in culture. In situ hybridization analysis indicated that TEFR1 transcripts are enriched in the skeletal muscle lineage during mouse embryogenesis. Transient expression of fusion proteins of TEFR1 and the yeast GAL4 DNA-binding domain in cell lines activated the expression of chloramphenicol acetyltransferase (CAT) reporter constructs containing GAL4 binding sites, indicating that TEFR1 contains an activation domain. An anti-TEFR1 polyclonal antibody supershifted the muscle-specific M-CAT.A1 factor complex in gel mobility shift assays, suggesting that TEFR1 is a major component of this complex. Our results suggest that TEFR1 might play a role in the embryonic development of skeletal muscle in the mouse.

The M-CAT motif is a cis-regulatory DNA sequence that is essential for muscle-specific transcription of several genes. Previously, we had shown that both musclespecific (A1) and ubiquitous (A2) factors bind to an essential M-CAT motif in the myosin heavy chain ␤ gene and that the ubiquitous factor is transcriptional enhancer factor (TEF)-1. Here we report the isolation of mouse cDNAs encoding two forms (a and b) of a TEF-1related protein, TEFR1. The TEFR1a cDNA encodes a 427-amino acid protein. The coding region of TEFR1b is identical to 1a in both nucleotide and predicted amino acid sequence except for the absence of 43 amino acids downstream of the TEA DNA-binding domain. Three TEFR1 transcripts (ϳ7, ϳ3.5, and ϳ2 kilobase pairs) are enriched in differentiated skeletal muscle (myotubes) relative to undifferentiated skeletal muscle (myoblasts) and non-muscle cells in culture. In situ hybridization analysis indicated that TEFR1 transcripts are enriched in the skeletal muscle lineage during mouse embryogenesis. Transient expression of fusion proteins of TEFR1 and the yeast GAL4 DNA-binding domain in cell lines activated the expression of chloramphenicol acetyltransferase (CAT) reporter constructs containing GAL4 binding sites, indicating that TEFR1 contains an activation domain. An anti-TEFR1 polyclonal antibody supershifted the muscle-specific M-CAT⅐A1 factor complex in gel mobility shift assays, suggesting that TEFR1 is a major component of this complex. Our results suggest that TEFR1 might play a role in the embryonic development of skeletal muscle in the mouse.
Commitment and subsequent differentiation of skeletal muscle involves a cascade of transcription factors that act through several different cis-regulatory elements. The best characterized of these elements are the E-box and A/T-rich motifs, which interact with the MyoD/bHLH (1) and MEF-2/MADS (2) families of transcription factors, respectively. Members of the MyoD family have been shown to be critical to the development of skeletal muscle (1,3). Members of the MEF-2 family are involved in both cardiac and skeletal muscle development (2, 4 -6).
In addition to elements that are involved in muscle-lineage determination, other cis-regulatory elements that have been shown to be involved in the expression of muscle-specific genes include the serum response element, SP-1 element, and M-CAT motif (6 -10). The M-CAT motif was initially identified as a cis-regulatory element in the cardiac troponin T promoter (6). The M-CAT motif has subsequently been shown to be required for muscle-specific expression from the myosin heavy chain ␤ (9 -11) and ␣-skeletal actin (7) promoters in vitro, as well as the myosin heavy chain ␤ (12) and ␣ (13) promoters in vivo.
Two mammalian M-CAT-binding proteins have been identified, transcriptional enhancer factor-1 (TEF-1) 1 (17,20,22) and embryonic TEA domain-containing factor (ETF) (23). ETF transcripts are strictly limited in their distribution during embryogenesis, appearing predominately in the hindbrain at embryonic day 10 by in situ hybridization analysis. Little is known about ETF function except that it shares binding-site specificity with TEF-1. TEF-1 is a cellular transcriptional activator that was originally identified as a viral cis-regulatory elementbinding protein that interacts with the GT-IIC and Sph motifs (14 -17). The GT-IIC motif is almost identical to the M-CAT motif, which TEF-1 has been shown to bind as well (7, 13, 18 -21). TEF-1 transcripts are present in most adult tissues, and are particularly abundant in the kidney, skeletal muscle, heart, and lung. During embryogenesis, TEF-1 appears to be expressed ubiquitously (24). TEF-1 contains a TEA DNA-binding domain (25) and an activation function that involves the cooperation of at least three regions of the protein (acidic and proline-rich and C-terminal regions) (26). TEF-1 also interacts with cell-specific co-factors, some of which are required for activation (27), others of which block activation (28). A transgenic mouse strain containing a null mutation of the TEF-1 gene shows homozygous embryonic lethality around embryonic day 11 due to a heart malformation characterized by an abnormally thin ventricle wall (24). However, the skeletal muscle lineage appears to be unaffected in these mice, which suggests that TEF-1 is not essential for the early stages of skeletal muscle development.
We have previously shown that two M-CAT (A-element) 1 The abbreviations used are: TEF-1, transcriptional enhancer factor-1; CAT, chloramphenicol acetyltransferase; aa, amino acid(s); ETF, embryonic TEA domain-containing factor; GST, glutathione S-transferase; nt, nucleotide(s); PAGE, polyacrylamide gel electrophoresis; RT, reverse transcription; PCR, polymerase chain reaction; mTEF1, mouse TEF-1; cTEF1, chicken TEF-1; TEFR1, TEF-1-related protein 1; bp, base pair(s); kb, kilobase pair(s). binding factors, A1 and A2, could be distinguished by gel mobility shift assays using nuclear extracts from differentiated skeletal muscle cells in culture (myotubes) (29). The A2 factor appears to be ubiquitous, while the A1 factor is seen only in myotubes. We cloned the mouse homolog of TEF-1 (mTEF-1) and showed that it is one component of the A2 factor (20). However, the identity of the A1 factor remained unknown. Because the A1 and A2 factors had similar DNA-binding properties (29), we concluded that these factors might be closely related. By using TEF-1 cDNA as a probe, we isolated cDNAs for TEFR1, an M-CAT-binding transcription factor that bears a close resemblance to TEF-1. We present evidence here suggesting that TEFR1 is a candidate for the muscle-specific A1 factor. We also show that TEFR1 transcripts are enriched in the skeletal myogenic lineage during mouse embryogenesis.

MATERIALS AND METHODS
Preparation of Sol8 Myotube cDNA Library-Poly(A) ϩ RNA from Sol8 myotubes was prepared using oligo(dT)-cellulose (Invitrogen). The Sol8 myotube Uni-ZAP cDNA library was prepared using the ZAP-cDNA synthesis kit (Stratagene). Bluescript SK(Ϫ) phagemids containing cDNA inserts between EcoRI and XhoI sites were excised from the ZAP clones using the ExAssist helper phage.
Cloning of TEF-1-related Transcription Factors-An ϳ800-bp cDNA clone encoding a TEF-1-related protein was isolated by screening an adult mouse cardiac ZAPII cDNA library (Stratagene) using human TEF-1 cDNA (kindly provided by Dr. Pierre Chambon) (17) as a probe under reduced stringency conditions as described previously (20). Using the TEF-1-related cDNA as a probe, we further screened the Sol8 myotube cDNA library. Prehybridization was performed at 42°C in a solution containing 50% formamide, 0.2% polyvinyl-pyrrolidone, 0.2% bovine serum albumin, 0.2% Ficoll, 0.05 M Tris-HCl, pH 7.5, 1.0 M NaCl, 0.1% sodium pyrophosphate, 1.0% SDS, 10% dextran sulfate, and 100 g/ml denatured salmon sperm DNA. Hybridization was performed at 42°C in the same solution with the denatured probe, which was labeled by the random oligo priming method using a kit (Boehringer Mannheim) according to the manufacturer's instructions. The filter was washed in 2 ϫ SSC (1 ϫ SSC is 0.15 M NaCl and 0.015 M sodium citrate) and 0.1% SDS at room temperature, followed by washing in 0.1 ϫ SSC and 0.1% SDS at 65°C.
DNA Sequencing and Analysis-Nested deletions of plasmids TEFR1a and 1b were created using an ExoIII/mung bean nuclease deletion kit (Stratagene). Double-stranded DNA sequencing was performed on both strands using the Sequenase kit (U. S. Biochemical Corp.) and/or using the Taq DyeDeoxy Terminator Cycle Sequencing kit and DNA Sequencer (Applied Biosystems). All computer analyses of nucleotide (nt) and amino acid (aa) sequences were done using the GCG package of sequence analysis tools (30).
Northern Blot Analysis-Total RNA from cells was extracted as previously described (31). Total RNA from different adult mouse tissues was extracted as described elsewhere (32). Poly(A) ϩ RNA from Sol8 myotubes and 3T3 cells was prepared using oligo(dT)-cellulose (Invitrogen). Either total RNA (30 g) or poly(A) ϩ RNA (3 g) was electrophoresed on 1% agarose-1.1 M formaldehyde gels, and transferred to nylon filters. All probes were prepared by random oligo priming using a kit (Boehringer Mannheim). Prehybridization and hybridization were performed at 42°C in the same solution as described for cDNA cloning (see above). The filters were washed in 2 ϫ SSC and 0.1% SDS at room temperature, followed by washing in 0.1 ϫ SSC and 0.1% SDS at 60°C.
In Situ Probe Preparation-A PCR product containing the first noncoding exon of the mouse ␣-cardiac actin gene, nt Ϫ26 to ϩ113 (34), was prepared from Sol8 genomic DNA and cloned into HindIII-BamHI sites of Bluescript SK(ϩ). The TEFR1 probe template corresponds to the complete coding region, nt 110-1393 of TEFR1a (Fig. 1A), generated by PCR from the original cDNA clone and ligated into the BamHI-XbaI sites of Bluescript SK(ϩ). Radioactive RNA probes (riboprobes) were prepared using either T3 or T7 RNA polymerase (Stratagene) and 35 S-UTP (DuPont NEN) as described elsewhere (35). TEFR1 probes were alkaline hydrolyzed as previously described (36) to an average length of 150 nt, which was verified by denaturing polyacrylamide gel electrophoresis (PAGE).
In Situ Hybridization-Embryos were obtained from timed-pregnant CD-1 mice (Charles River) and fixed in 4% paraformaldehyde in phosphate-buffered saline overnight at 4°C. Dehydration, embedding, and section preparation were conducted as described elsewhere (35). Prehybridization treatment of tissue sections, hybridization, and posthybridization treatment were conducted as previously described (34), with the omission of proteinase K digestion. Riboprobes were applied directly to tissue sections at an activity of approximately 75,000 dpm/l. Emulsion autoradiography using Kodak NTB-2 emulsion, counterstaining with toluidine blue, and subsequent mounting were conducted as previously described (35). Slides dipped in emulsion were developed after exposure for 10 days in all cases.
In Vitro Transcription and Translation-Plasmids mTEF-1 (20) and TEFR (1a and 1b) were linearized with NdeI and XhoI, respectively. The linearized plasmids were transcribed using either T7 or T3 RNA polymerase in the presence of m7G(5Ј)ppp(5Ј)G in addition to NTPs. One microgram of RNA product was translated with 20 l of rabbit reticulocyte lysate (Stratagene) in the presence of [ 35 S]methionine (ICN).
Preparation of Antibody-Anti-TEFR1 polyclonal antibody was produced by immunization of chickens with GST-TEFR(1-38) fusion protein. All immunizations, collections of serum and IgY purifications from eggs were performed by East Acres Biologicals, Inc. (Southbridge, MA).
Immunoprecipitation-Five microliters of in vitro transcription/ translation products of TEFR1b and mTEF-1 cDNA clones (see above) in 200 l of radioimmune precipitation buffer (150 mM NaCl, 50 mM Tris-HCl, pH 7.5, 1 mM EDTA, and 0.2% Nonidet P-40) were precleared by incubating for 1 h at 4°C with rocking in the presence of 20 g of preimmune chicken IgY, 15 g of rabbit anti-chicken IgY and 30 l of 50% protein A-Sepharose (Pharmacia) slurry. After spinning for 2 min at 500 ϫ g, the supernatants were incubated for 1 h on ice with 20 g of anti-TEFR1 antibody and 15 g of rabbit anti-chicken IgY in the absence or presence of an equimolar amount of different competitors. They were further incubated for 1 h at 4°C with rocking in the presence of 30 l of 50% protein A-Sepharose slurry. The Sepharose beads were washed four times with radioimmune precipitation buffer, resuspended in SDS sample buffer (37), and incubated in a boiling water bath for 5 min. Proteins were separated on an SDS-PAGE gel.
Gel Mobility Shift Assay-Preparation of nuclear extracts and binding reactions for gel mobility shift assays were carried out as described elsewhere (29). For supershift experiments, nuclear extracts were preincubated with either preimmune or anti-TEFR1 antibody at room temperature for 30 min in the absence or presence of an equimolar amount of different competitors. Reaction mixtures were further incubated with labeled DNA probe and poly(dI-dC) at room temperature for 15 min. Native PAGE gels were run at 150 V in a 0.5 ϫ TBE buffer (35). The sequence of the sense strand of oligo A (M-CAT) is 5Ј-CAG GCA GTG GAA TGC GAG GAG-3Ј (29). This oligo was annealed to a complementary oligo to form the double-stranded probe.
Cell Culture-Mouse skeletal muscle (Sol8 and C2C12), mouse fibroblast (Swiss 3T3), and human cervical carcinoma (HeLa) cell lines were maintained in growth medium containing Dulbecco's modified Eagle's medium (Life Technologies, Inc.) with either 20% (Sol8 and C2C12) or 10% (3T3 and HeLa) fetal bovine serum. Differentiation of Sol8 and C2C12 myoblasts was induced by exposure of confluent cultures to differentiation medium containing Dulbecco's modified Eagle's medium and 10% horse serum.
Transfection and Enzyme Assay-DNA transfection was performed by the calcium phosphate precipitation method (38) as described previously (29). CAT and ␤-galactosidase activity in the cell extracts were assayed as described previously (39,40).

RESULTS
TEFR1a and 1b Are TEF-1-related Transcription Factors-We isolated TEF-1-related cDNA clones by screening a mouse cardiac cDNA library using human TEF-1 cDNA as a probe under low stringency conditions as described previously cDNA Cloning of M-CAT-binding Factors (20). We then screened a Sol8 myotube cDNA library using one of the cardiac cDNA clones, TEFR1 (ϳ0.8 kb), as a probe under high stringency conditions. About 20 clones were isolated from an initial screening of 5 ϫ 10 5 plaques. Two different clones (TEFR1a and 1b) containing full-length coding regions were selected by restriction mapping and partial DNA sequencing for further analysis. The TEFR1a and 1b clones were ϳ1.6 and ϳ1.7 kb in length, respectively. The DNA sequences of these clones are shown in Fig. 1A.
The open reading frame of the TEFR1a cDNA encodes a 427 aa protein containing a TEA DNA-binding domain (aa 31-98; Fig. 1, A-C). The initiation codon used by TEFR1 appears to be AUU (Ile) at nt 110 (see "Discussion"). The open reading frame of the TEFR1b cDNA encodes a 384-aa protein, which is identical to TEFR1a except that 43 aa downstream of the TEA domain are absent from TEFR1b (Fig. 1, B and C). The full coding nucleotide sequence of TEFR1b is identical to that of TEFR1a, except that a 129-nt region (nt 443-571 in TEFR1a) is absent from TEFR1b (Fig. 1, A and B). The noncoding regions of the two cDNAs are divergent, except for sequences proximal to the start and stop codons, which are identical (Fig. 1, A and  B). Comparison of the TEFR1 and mTEF-1 (20) predicted amino acid sequences revealed 76% overall identity and 93% identity within the TEA domain (Fig. 1C). The TEA DNAbinding domain consists of three putative ␣-helices (25) and is highly conserved throughout evolution. Except for TEFR1, all other vertebrate TEA proteins share 100% amino acid identity within the TEA domain ( Fig. 1D) (23).
TEFR1 Transcripts Are Enriched in Embryonic Skeletal Muscle and Adult Lung-We examined the expression of TEFR1 transcripts in various cell lines and tissues by Northern blot analysis using a probe containing the full coding region of the TEFR1a cDNA clone (Fig. 2). A major ϳ7-kb transcript and minor ϳ3.5and ϳ2-kb transcripts were observed in differentiated skeletal muscle cells (Sol8 and C2C12 myotubes), but not in undifferentiated skeletal muscle cells (Sol8 and C2C12 myoblasts) or non-muscle cells (3T3 and HeLa) ( Fig. 2A). In adult tissues, TEFR1 transcripts are abundant in lung, present at a low level in kidney, skeletal muscle, and heart, and undetect-able in thymus, brain, spleen, and liver (Fig. 2B). In poly(A) ϩ RNA, TEFR1 transcripts are present at a low level in nonmuscle (3T3) cells and are abundant in differentiated skeletal muscle cells (Sol8 myotubes) (Fig. 2C). In contrast to the expression pattern of TEFR1 transcripts, a ϳ12-kb mTEF-1 transcript was found at a similar level in both Sol8 myotubes and 3T3 cells, as previously observed (Fig. 2C) (20).
RT-PCR was used to distinguish between TEFR1a and 1b transcripts (Fig. 3). Both TEFR1a (620 bp) and 1b (491 bp) PCR products were seen in Sol8 and C2C12 myotubes (Fig. 3B). Both TEFR1a and 1b were also observed in adult lung, skeletal muscle, heart, and kidney (Fig. 3C). A very small amount of TEFR1 was seen in Sol8 and C2C12 myoblasts and non-muscle cells (3T3) (Fig. 3B). TEFR1b has consistently appeared to be more abundant than 1a in all tissues and cell lines tested under the RT-PCR conditions used here, except in 3T3 cells. In 3T3 cells, TEFR1a has consistently appeared to be more abundant than 1b.
We also examined the expression of TEFR1 transcripts during mouse development by in situ hybridization using antisense riboprobes made by in vitro transcription of the full coding region of the TEFR1a cDNA. TEFR1 transcripts are enriched in the myotome at embryonic day 9 (Fig. 4, A-E), co-localized with ␣-cardiac actin (Fig. 4, C and F). At embryonic day 14.5, TEFR1 transcripts are enriched in embryonic skeletal muscle (Fig. 5B), co-localized with ␣-cardiac actin (Fig. 5A). ␣-Cardiac actin has been shown to be a marker for embryonic striated muscle (34). In situ hybridization using sense riboprobes showed background levels of hybridization over tissue sections (Fig. 5, C and D).
TEFR1 Binds to the M-CAT Motif (A Element) in Vitro-We examined whether TEFR1a and 1b could bind to the M-CAT motif in a gel mobility shift assay. In vitro transcription/translation products of mTEF-1, TEFR1a, and 1b cDNAs were analyzed by SDS-PAGE (Fig. 6A). The mTEF-1 product migrated as a major ϳ53-kDa band as described previously (20). The major TEFR1a product was slightly smaller (ϳ52 kDa) than mTEF-1, and the major TEFR1b product was smaller (ϳ50 kDa) than TEFR1a. Using these in vitro transcription/transla- cDNA Cloning of M-CAT-binding Factors tion products and end-labeled oligo A containing the M-CAT motif from the rabbit myosin heavy chain ␤ promoter as a probe (29), we performed a gel mobility shift assay (Fig. 6B). TEFR1a and mTEF-1 formed specific protein-DNA complexes with the same mobility as that of the ubiquitous M-CAT⅐A2 factor-complex, while TEFR1b formed a specific complex with the same mobility as that of the muscle-specific M-CAT⅐A1 factor complex. The specificity of TEFR1 binding was examined in competition experiments with unlabeled oligo A and MutA (29) as a competitor (Fig. 6C). Unlabeled oligo A competed for complex formation between labeled oligo A and TEFR (1a and 1b). In contrast, unlabeled oligo MutA, which contains a mutated M-CAT motif, had no effect on complex formation.
Muscle-specific A1 Factor Is Antigenically Related to TEFR1 and Distinct from mTEF-1-We generated a polyclonal antibody against a GST-fusion protein containing the amino-terminal 38 aa of TEFR1 (GST-TEFR(1-38)) (Fig. 7A). To determine the specificity of this antibody, we conducted immunoprecipitations of in vitro transcription/translation products of the TEFR1b and mTEF-1 cDNAs (Fig. 7B) in the presence of one of several GST-fusion protein competitors (Fig. 7A). In the absence of competitor (Ϫ) or the presence of GST as a competitor (GST), both TEFR1b (lanes 2 and 3) and mTEF-1 (lanes 8 and  9) were immunoprecipitated. In the presence of the immunogen RNAs from mouse skeletal muscle (Sol8 and C2C12) and fibroblast (Swiss 3T3) cell lines were subjected to RT-PCR using specific amplimers for either TEFR1 (24 cycles) or ␤ 2 -microglobulin (16 cycles). PCR products were separated by agarose gel electrophoresis and Southern blotted. The 620-and 491-bp bands correspond to TEFR1a-and 1b-specific PCR products, respectively. The negative control is in the absence of RNA. Mb, myoblasts; Mt, myotubes. C, total RNAs from adult mouse tissues were subjected to RT-PCR using TEFR1-specific (35 cycles) and ␤-actin-specific (22 cycles) amplimers. PCR products were separated by agarose gel electrophoresis and Southern blotted. Ethidium bromide staining of ␤-actin products is shown at the bottom. The negative control is in the absence of RNA.   5 and 11).
We then conducted gel mobility shift assays using nuclear extracts from either HeLa cells or Sol8 myotubes in the presence of either preimmune or anti-TEFR1 antibody (Fig. 7C). Using Sol8 myotube nuclear extract, addition of anti-TEFR1 antibody resulted in the disappearance of the A1 complex band, reduction of the intensity of the A2 complex band, and formation of a supershifted complex (lanes 6 and 8). However, using HeLa cell nuclear extract, addition of anti-TEFR1 antibody reduced the formation of the A2 complex and did not result in the formation of a supershifted complex (lanes 2 and 4). Supershifted complexes are generally thought to contain probe, DNAbinding protein, and specific antibody. To confirm whether the muscle-specific A1 complex contains TEFR1, we examined the ability of GST-fusion protein competitors (Fig. 7A) to block the activity of the anti-TEFR1 antibody (Fig. 7D). A GST-fusion protein containing only TEFR1-specific residues (GST-TEFR(1-24)) completely blocked the activity of the anti-TEFR1 antibody (lane 4), preventing the formation of the supershifted complex and returning the A1 and A2 complex band intensities to those observed in the presence of preimmune antibody (lane 1). The GST-TEFR1 immunogen (GST-TEFR(1-38)) also completely blocked the activity of the anti-TEFR1 antibody (lane 3).
In the presence of either GST alone (lane 6) or a GST-mTEF-1 competitor (GST-mTEF ; lane 5), no competition was observed with the anti-TEFR1 antibody. These results suggest that the A1 complex contains TEFR1.
TEFR1 Contains an Activation Domain-We conducted transfection experiments in which TEFR1 expression vectors (pCMV-TEFR1a and 1b) were co-transfected with TK-CAT reporter constructs containing either GT-IIC or M-CAT elements upstream of the thymidine kinase promoter (Fig. 8). The activity of the CAT reporter genes in the absence of co-transfected TEFR1 is presumably due to endogenous M-CAT-binding factors (mTEF-1 and/or TEFR1). When TEFR1 was co-transfected, this background CAT activity was repressed in a dosedependent manner in both HeLa cells and Sol8 myotubes. Based on the close structural similarity and sequence identity between TEFR1 and mTEF-1, this phenomenon might be due to "squelching" (see "Discussion"). Because the M-CAT motif has been shown to be a positive-regulatory element in multiple muscle-specific genes (6,7,9), we conducted experiments to determine whether TEFR1 contains an activation domain. We conducted transfection experiments using GAL4/TEFR1 chimeras and a GAL4-CAT reporter, an approach which has been used successfully in the study of TEF-1 activation function (26). The full coding regions of TEFR1a and 1b were connected in-frame to the DNA-binding domain of yeast activator GAL4 (aa 1-147) to form GAL4-TEFR 1 and GAL4-TEFR 2, respectively (Fig. 9). Co-transfection of GAL4-TEFR 1 with the p3xGAL4-BGCAT reporter (kindly provided by Dr. Erik K. Flemington) slightly increased basal CAT activity in Sol8 myotubes (5-fold) and HeLa cells (2-fold). GAL4-TEFR 2 showed more activation than GAL4-TEFR 1 in Sol8 myotubes (44-fold) and HeLa cells (6-fold).
There are two functional DNA-binding domains in both GAL4-TEFR 1 and GAL4-TEFR 2. It has been reported that GAL4/AP2, SP1/(CTF/NF1), and GAL4/TEF-1 chimeras, which contain more than one functional DNA-binding domain, are weak transactivators, although they contain strong activation domains (26,41,42). To determine whether or not strong activation functions in GAL4-TEFR 1 and 2 were being masked by the presence of two functional DNA-binding domains, we mutated two conserved amino acids within the first putative ␣-helix of the TEFR1 TEA DNA-binding domain, aa 48 (Leu 3 Pro) and 50 (Ile 3 Phe) (Fig. 1D). Mutations in this region result in loss of DNA-binding activity (26) (data not shown). Transfection of GAL4-TEFR 1M or GAL4-TEFR 2M which contain mutated TEFR1a and 1b, respectively, strongly activated the expression of the p3xGAL4-BGCAT reporter in both Sol8 myotubes (ϳ100-fold) and HeLa cells (ϳ150-fold) (Fig. 9). Deletion of aa 1-111 of TEFR1a (GAL4-TEFR 3), including acidic and basic regions and the entire TEA domain, dramatically increased the expression of the reporter plasmid (ϳ400 fold in Sol8; ϳ800-fold in HeLa). Strong activation was still observed when aa 1-206, including the proline-rich region, were deleted (GAL4-TEFR 5). However, deletion of aa 1-302, including one of two STY-rich regions, resulted in total loss of activation function (GAL4-TEFR 6). In addition, there was no activation when the C-terminal 28 aa (aa 400 -427) were removed from GAL4-TEFR 4 (GAL4-TEFR 7). Therefore, TEFR1 apparently contains one or more activation domains, at least one of which is located in the C-terminal half of the protein (aa 207-427). DISCUSSION TEF-1 is considered to be the transcription factor that is primarily responsible for the transcriptional activation FIG. 6. In vitro transcription/translation of TEFR1 and mTEF-1. A, SDS-PAGE of in vitro translated mTEF-1 and TEFR1. The mTEF-1 and TEFR1a and 1b RNAs were synthesized in vitro, using either T7 or T3 RNA polymerase, and then translated using rabbit reticulocyte lysate. Three microliters of the in vitro translation products were separated on an 11% SDS-PAGE gel. B, gel mobility shift assay of in vitro translated products using end-labeled oligo A. Three microliters of either unprogrammed rabbit reticulocyte lysate (Lysate), mTEF-1, or TEFR1 translation products, or 2 g of nuclear extracts from HeLa cells or Sol8 myotubes (Mt) were incubated with end-labeled oligo A and separated on an 8% native PAGE gel. Arrows indicate the positions of specific complexes (A1 and A2), nonspecific complexes (*), and free probe (F). C, competition for complex formation between in vitro translated products and oligo A. TEFR1 translation products were incubated with end-labeled oligo A in the absence (Ϫ) or presence of unlabeled competitor oligo A or MutA (M) at a 50-fold molar excess over the labeled probe.
cDNA Cloning of M-CAT-binding Factors through the M-CAT cis-regulatory element found in certain muscle-specific genes. This seems reasonable for both nonmuscle and cardiomyocyte cells, as in both cases there is a single M-CAT-binding activity seen in gel mobility shift assays, which has been attributed to TEF-1 by antigenic criteria (7,13,17,19). However, we had shown previously that in differentiated mouse skeletal muscle in culture two M-CAT-binding activities could be detected, which we designated A1 and A2 (29). The M-CAT⅐A2 factor-complex is ubiquitous and contains mTEF-1 as a major component (20). Here we report the isolation of a TEF-1-related transcription factor, TEFR1, which appears to be a major component of the muscle-specific M-CAT⅐A1 factor-complex.
We have isolated two TEFR1 cDNAs, TEFR1a and 1b, which have identical nucleotide sequences within their coding and proximal noncoding regions, but for the absence of 129 nt (43 aa) in TEFR1b. Southern blotting of CD1-mouse genomic DNA with the full-length TEFR1b cDNA clone showed hybridization to a single, ϳ10-kb EcoRI fragment, which suggests that TEFR1a and 1b are splice forms of a single gene (data not shown). In all cell types and tissues examined by RT-PCR, TEFR1b transcripts have been more abundant than 1a, except in the case of 3T3 cells, where the ratio is inverted (Fig. 3). It has been reported that the function of a transcription factor can be altered by structural changes occurring in vivo, such as alternative splicing (43,44) and phosphorylation (45). Further work will be required to determine whether TEFR1a and 1b are functionally distinct. TEFR1 appears to use AUU (Ile) at nt 110 as an initiation codon, similar to TEF-1 (17). Mutation of this codon to UGG (Trp) resulted in no translational initiation in vitro from nt 110, while mutation to AUG (Met) increased translational efficiency (data not shown). There are several AUG codons in the 5Ј portion of the TEFR1 cDNAs. However, comparison of the flanking sequences surrounding these various codons indicated that the AUU at nt 110 lies in the best Kozak sequence context (46). Furthermore, none of the AUG codons is appropriately positioned to account for the fact that TEFR1 contains a TEA DNA-binding domain, as indicated by its ability to bind to the M-CAT motif (Fig. 6). Also, only initiation at the AUU codon at nt 110 can account for both the molecular weights of the in vitro translated TEFR1 products (Fig. 6) and the existence of the amino-terminal residues which were detected by the anti-TEFR1 antibody (Fig. 7).
TEFR1 is the third member of the TEA family to be found in mammals. The expression patterns of ETF (23) and TEFR1 are tissue-restricted, in contrast to the widespread expression of mTEF-1 (20). Comparison of the predicted amino acid se- cDNA Cloning of M-CAT-binding Factors quences of mTEF-1, TEFR1, and ETF indicated that ETF is equally divergent, ϳ65% identity, from either mTEF-1 or TEFR1, while mTEF-1 and TEFR1 share 76% identity. Comparison of TEFR1 to the 172-bp EST 683 human cDNA clone 32B5 (GenBank TM accession no. T25108) (47) showed that they share 95% identity at the amino acid level. The sequence of EST 683, which was identified by systematic sequencing of a colorectal cancer cDNA library, coincides with TEFR1 aa 261-316. EST 683 might correspond to the human analog of mouse TEFR1, but this must be confirmed by cloning and sequencing of the full-length cDNA. EST 683 also shows a high degree of homology to chicken TEF-1 (cTEF-1) (21). Comparison of the cTEF-1 predicted amino acid and nucleotide sequences to those of mTEF-1 and TEFR1 revealed an unexpected result. At the amino acid level, TEFR1 has an overall identity with cTEF-1 of 87%, but only 76% with mTEF-1, while mTEF-1 and cTEF-1 share 77% identity. Comparison of the nucleotide sequences of the TEA domains revealed that TEFR1 shares 83% identity with cTEF-1, but only 76% with mTEF-1, while mTEF-1 and cTEF-1 share 78% identity. These comparisons suggest that cTEF-1 is the chicken analog of TEFR1 rather than mTEF-1. However, the structures of the activation domains of TEFR1 (Fig. 9) and cTEF-1 (21) are significantly different. One possibility is that cTEF-1, mTEF-1, and TEFR1 represent three distinct members of the TEA family. If, on the other hand, cTEF-1 and TEFR1 are homologs, then differences between the activation domains of cTEF-1 and TEFR1 might be due to differences in co-factors between avians and mammals (see below). A third alternative is that, in avians, there might be a single transcription factor (cTEF-1) responsible for transcriptional activation through the M-CAT motif in all striated muscle.
TEFR1 contains multiple putative activation domains, most of which are shared structurally with TEF-1, such as acidic (48,49), proline-rich (41,42), and STY-rich (50,51) regions. However, the acidic and proline-rich regions which are essential for TEF-1 activation function (26) are dispensible for TEFR1 activation function (Fig. 9). In transfection assays using TEFR1 FIG. 8. Effect of overexpression of TEFR1 and mTEF-1 on reporter plasmids containing multiple binding sites. Each reporter (2 g) was co-transfected with expression vector (0.2 or 1.0 g) and pCMV-lacZ (1 g) into either Sol8 myocytes (A) or HeLa cells (B). The 5GTIIC-TKCAT and 3A-TKCAT plasmids contain five copies of a GTIIC tandem repeat and three copies of the A element (21 bp), respectively, upstream of the herpes simplex virus thymidine kinase promoter in pTKCAT (20). The CAT activity was normalized using the ␤-galactosidase activity for each transfection. All CAT activities are given relative to the value obtained for the p5GTIIC-TKCAT, which is set at 100% for each cell type. Results represent the average of two to four separate transfection experiments. Variation of normalized values was less than 10% between experiments.
FIG. 9. Transactivation of the p3xGAL4-BGCAT reporter by GAL4/TEFR1-chimeric proteins. The p3xGAL4-BGCAT reporter (2 g) containing three copies of the GAL4 DNA-binding site was transfected with each expression vector for a GAL4/TEFR1 chimera (0.1 g) and pCMV-lacZ (1 g) into either Sol8 myocytes or HeLa cells. CAT activity was normalized using the ␤-galactosidase activity for each transfection. Values for fold induction are given relative to activity of the GAL4(1-147) construct. Results represent the average of two to four separate transfection experiments. Variation of normalized values was less than 10% between experiments. The structure of TEFR1a is shown at the top. The region absent from TEFR1b (aa 112-154) and the TEA domain (aa 31-98) are shown above the TEFR1a diagram. The acidic, basic, proline-rich (Pro), and serine-threonine-tyrosine-rich (STY) regions are indicated by different fill patterns. The positions of mutated aa 48 (Leu 3 Pro) and aa 50 (Ile 3 Phe) in GAL4-TEFR 1M and 2M are indicated by asterisks. expression vectors and CAT reporter constructs containing either M-CAT or GT-IIC elements, we observed a dose-dependent repression of reporter gene activity (Fig. 8). One possible explanation for this observation is that TEFR1 is an M-CATbinding repressor. However, two observations make this possibility unlikely. First, accumulation of transcripts of some contractile proteins in differentiated skeletal muscle is dependent upon the M-CAT positive cis-regulatory element. Second, the formation of the M-CAT⅐A1 factor (TEFR1) complex occurs in skeletal muscle in vitro as differentiation proceeds (29). An alternative explanation is that overexpression of TEFR1 results in "squelching" (52), which implies that TEFR1 requires a co-activator(s) to function, as has been determined biochemically for TEF-1 (27,28). The dose-dependence of the squelching effect and minimal structural requirements for activation are different between TEF-1 and TEFR1. Therefore, TEF-1 and TEFR1 might interact with overlapping but nonidentical sets of co-factors. Further work will be required to determine whether the structural differences between TEF-1 and TEFR1 indicate in vivo functional differences.
To directly address the question of whether TEFR1 is a component of the muscle-specific A1 complex, we generated a polyclonal antibody to the amino terminus of TEFR1. Gel mobility shift assays in the presence of this antibody resulted in the formation of a supershifted complex and reduction in both A1 and A2 complex band intensities (Fig. 7C). This effect could be eliminated by competition with a segment of the immunogen which was specific to TEFR1 (GST-TEFR(1-24); Fig. 7D, lane  4). These results, along with previous data indicating that components of the A1 complex are antigenically distinct from TEF-1 (20), suggest that a major component of the musclespecific A1 complex is TEFR1. It is possible that the A1 complex does not contain TEFR1, but rather a protein that contains a region that is antigenically identical to the amino terminus of TEFR1. This is unlikely, though, because the region corresponding to TEFR(1-24) is highly divergent among the three mammalian TEA-domain proteins identified so far, TEF-1 (17,20,22), TEFR1, and ETF (23). Further work will be required to determine whether the A2 complex in differentiated skeletal muscle also contains TEFR1 and to determine whether TEFR1a, 1b, or both are present in the muscle-specific A1 complex.
Skeletal muscle cell lines exhibit many of the properties of embryonic and perinatal muscle (53,54). The enrichment of TEFR1 transcripts in differentiated skeletal muscle cells in culture suggested that TEFR1 might also be enriched in embryonic skeletal muscle. In situ hybridization analysis of mouse embryos at embryonic day 9 (Fig. 4) and embryonic day 14.5 ( Fig. 5) showed that TEFR1 transcripts are relatively abundant in the skeletal myogenic lineage. Our Northern analysis of adult tissue RNA showed that TEFR1 transcripts are expressed weakly in adult skeletal muscle. Therefore, TEFR1 might be transiently required for some early stage of myofiber maturation, such as appropriate accumulation of key components of the contractile apparatus.
Northern analysis showed that both mTEF-1 (20) and TEFR1 (Fig. 2B) are present in adult heart. Our in situ hybridization results showed no enrichment of TEFR1 transcripts in embryonic heart (data not shown). Our initial isolation of TEFR1 was from an adult mouse cardiac cDNA library. However, after multiple screenings of this library, only a single, partial TEFR1 cDNA clone was isolated, while we recovered multiple mTEF-1 cDNA clones. This suggests that TEFR1 transcripts are less abundant than those of mTEF-1 in the adult heart. Also, there appears to be a single cardiac M-CATbinding activity (7,13,19). Although these observations imply a minor role for TEFR1 in the heart, we cannot rule out the possibility that TEFR1 is involved in cardiac gene regulation. In particular, we do not know whether TEFR1 transcripts are present at the time when contractile protein gene transcription is initiated in the heart, prior to embryonic day 8 (34).
In the adult mouse, TEFR1 transcripts are present at a higher level in lung than in any other tissue we tested. Mouse TEF-1 transcripts are also present at a high level in adult lung, approximately equivalent to striated muscle and kidney (20). Our in situ hybridization data indicated that TEFR1 is not enriched in the embryonic lung at either embryonic day 9 or 14.5 (data not shown). This suggests that TEFR1 might not be involved in early development of the lung. Though TEFR1 and possibly other members of the TEA family might be involved in lung-specific gene expression, especially postnatally, no lungspecific promoters have been identified yet that contain functional M-CAT motifs.
As more transcription factors have been characterized, it has become clear that the existence of networks of factors interacting at single cis-regulatory elements is common. The E-box and A/T-rich motifs have been shown to interact with distinct families of transcription factors, the MyoD/bHLH (1) and MEF-2/ MADS (2) families, respectively. The M-CAT motif appears to be another such element, interacting with members of the TEA family of transcription factors. Our results indicate that TEFR1 and TEF-1 are closely related members of this TEA family and that TEFR1 might play a role in the embryonic development of skeletal muscle.