The Polymyositis-Scleroderma Autoantigen Interacts with the Helix-Loop-Helix Proteins E12 and E47*

The basic helix-loop-helix (bHLH) transcription factors E12 and E47 regulate cellular differentiation and proliferation in diverse cell types. While looking for proteins that bind to E12 and E47 by the yeast interaction trap, we isolated the rat (r) homologue of the human (h) polymyositis-scleroderma autoantigen (PM-Scl), which has been localized to the granular layer of the nucleolus and to distinct nucleocytoplasmic foci. The rPM-Scl and hPM-Scl homologues are 96% similar and 91% identical. We found that rPM-Scl mRNA expression was regulated by growth factor stimulation in cultured rat aortic smooth muscle cells. rPM-Scl bound to E12 and E47 but not to Id3, Gax, Myb, OCT-1, or Max. The C terminus of rPM-Scl (amino acids 283–353) interacted specifically with a 54-amino acid domain in E12 that is distinct from the bHLH domain. Finally, cotransfection of rPM-Scl and E47 specifically increased the promoter activity of a luciferase reporter construct containing an E box and did not affect the basal activity of the reporter construct. rPM-Scl appears to be a novel non-HLH-interacting partner of E12/E47 that regulates E2A protein transcription.

The basic helix-loop-helix (bHLH) 1 transcription factors E12 and E47 regulate cell type-specific transcription and growth by binding to the E box (CANNTG) on target genes (1). The two proteins result from alternative splicing of E2A gene exons that encode similar bHLH domains (2). Because E12 and E47 are expressed in many embryonic and adult tissues (3), they cannot regulate cell type-specific gene expression. Tissue selectivity is achieved by the formation of E12/E47 heterodimers with cell type-specific bHLH transcription factors, which promotes differentiation to individual cell types. For example, the skeletal muscle bHLH factors MyoD, Myf5, myogenin, and Myf6 regulate each step of myogenesis by forming heterodimers with E12/E47 (4 -6). Similarly, the differentiation of hematopoietic cells (7), presomitic mesoderm, and neural tissue (8,9) also depends on the interaction of E12/E47 with cell type-specific bHLH factors. The E2A proteins can also function in the absence of cell type-specific bHLH factors. Although no cell typespecific bHLH factor has been identified in fibroblasts, overexpression of E47 retards the progression of fibroblasts through the cell cycle at the G1 restriction point (10).
The function of E12 and E47 is antagonized by the proteins Id1, Id2, and Id3 (11,12). Although the Id proteins have an HLH domain, they lack a DNA-binding domain. The Id proteins use the HLH domain to sequester E12/E47 and prevent subsequent activation of the E box. Increased transgenic expression of Id1, under the influence of an immunoglobulin promoter/enhancer, prevents B cell differentiation (13). Also, the Id proteins increase cellular proliferation by inhibiting E12/E47-dependent suppression of growth (14 -16). Because of the important role of the E2A proteins in the regulation of cell growth and differentiation, proteins that bind to the E2A proteins may also have an important impact on these processes (17,18).
To identify proteins that interact with the E2A proteins, we used E12 as a bait in the yeast two-hybrid interaction trap. By this technique we cloned the rat (r) homologue of the human (h) polymyositis-scleroderma autoantigen (PM-Scl) from a rat aorta cDNA library. PM-Scl had been shown previously to reside predominantly in the granular layer of the nucleolus (as well as within distinct nucleocytoplasmic foci) and not to be affected by nuclease digestion (19 -21). We found that the C terminus of rPM-Scl interacts specifically with a 54-amino acid domain in E12/E47 that is distinct from the bHLH domain. This interaction appears to affect E2A protein function because cotransfection of rPM-Scl and E47 specifically increased E47mediated transcriptional activation of an E box-luciferase reporter construct.

MATERIALS AND METHODS
Plasmids and Strains-Plasmid DNAs and the yeast strain EGY48 (MAT␣ trp1 ura3 his3 LEU2::pLexAop6-LEU2) used in the interaction trap assays were provided by Dr. Roger Brent and colleagues (Massachusetts General Hospital) and used as described (22,23). The Escherichia coli K-12 strain KC8 (pyrF::Tn5, hsdR, leuB600, trpC9830, lacD74, strA, galK, hisB436), the gift of Dr. Kevin Struhl (Harvard Medical School, Boston, MA), was used to rescue cDNA plasmids from yeast. The E12 bait plasmid used in the interaction trap was constructed by cloning a 0.53-kilobase pair cDNA fragment encoding amino acids 477-654 of human E12 (E12-(477-654)) (24), which had been obtained by polymerase chain reaction amplification of pGEM3E12 (kindly provided by Dr. David Baltimore, Massachusetts Institute of Technology, Cambridge) into the EcoRI-and XhoI-digested yeast expression vector pEG202 (23). (The pEG202 plasmid is used to constitutively express a LexA fusion or bait protein.) The DNA sequences coding for the bait and interactant proteins were generated by polymerase chain reaction and reverse transcription polymerase chain reaction and cloned between the unique EcoRI and XhoI sites of pEG202 and pJG4-5 (23), respectively. The various deletion mutants of E47 were the gifts of Dr. Lennart Philipson (New York University Medical Center, New York, NY) (10). The bait proteins used were full-length rat Id3 (9), mouse Gax (amino acids 121-303) (25), mouse Myb (amino acids 1-185) (26), human OCT-1 (amino acids 294 -429) (27), and full-length rat Max (28). All constructs were sequenced to confirm in-frame expression of fusion proteins. For expression in eukaryotic cells, rPM-Scl and Id1 were cloned into the pCR3 vector (Invitrogen), whereas full-length E47 was expressed in the pEMSV plasmid (29). The pGEX4T-1 vector (Pharmacia Biotech Inc.) was used for the expression of glutathione S-transferase (GST) fusion proteins in E. coli host HB101. The E box reporter plasmid pGL2(E box) 3 was generated by cloning an oligonucleotide (30) containing three E box sequences into the minimal SV40 promoter in the pGL2 promoter vector (Promega).
Yeast Interaction Trap-We screened a rat aorta cDNA library for E12 interactants by the yeast two-hybrid interaction trap as described (31). Briefly, EGY48 (MAT␣ trp1 ura3 his3 LEU2::pLexop6-LEU2) was used as host yeast strain. The bait plasmid was constructed by cloning E12-(477-654) in-frame of the LexA gene contained in the pEG202 plasmid. An oligo(dT)-primed cDNA library from rat aorta (CD strain, Charles River Laboratories) was constructed with the yeast galactoseinducible expression plasmid pJG4-5.
The yeast were transformed by the method of Gietz et al. (32) with modification. A total of 4 ϫ 10 6 primary yeast transformants were selected and plated onto Ura Ϫ His Ϫ Trp Ϫ Leu Ϫ plates containing 2% galactose. After 4 days at 30°C, 241 Leu ϩ colonies appeared, of which 90 showed a galactose-dependent blue color in 5-bromo-4-chloro-3-indolyl ␤-D-galactopyranoside medium. We prepared plasmid DNAs from 42 of these colonies (33) and introduced them into KC8 cells by electroporation. One of the clones encoded residues 283-353 of rPM-Scl. The full-length rPM-Scl cDNA clone was then isolated by screening a rat aortic smooth muscle cell (RASMC) cDNA library in ZAPII (Stratagene). The longest of the three positive clones analyzed contained a cDNA insert of ϳ1.6 kilobase pairs, which was sequenced completely by using customized oligonucleotide primers and plasmid DNA as a template.
To assess the specificity of the interaction, we transformed yeast of the EGY48/pSH18-34 strain with the interactant plasmid rPM-Scl(283-353) and the bait constructs indicated in Fig. 4 and applied them to glucose Ura Ϫ His Ϫ Trp Ϫ plates. Four colonies from each bait/ interactant combination were picked and applied to Ura Ϫ His Ϫ Trp Ϫ plates containing 5-bromo-4-chloro-3-indolyl ␤-D-galactoside and either 2% glucose or 2% galactose and 1% raffinose. We checked the color of the yeast 48 h later.
Liquid ␤-galactosidase assays were performed to map the interaction domains. A single yeast colony bearing the appropriate bait and interactant plasmid was inoculated into minimal Ura Ϫ His Ϫ Trp Ϫ medium with 2% glucose and grown to saturation overnight at 30°C. The next day, cells were diluted 1:50 into medium containing 2% galactose and 1% raffinose and allowed to grow to an A 600 of 1.0 -2.0. The A 600 of the culture was measured, and the cells were harvested and permeabilized as described (34). ␤-Galactosidase units were calculated by the equation 1000(A 420 )/(time ϫ volume ϫ A 600 ), where time is expressed in min and volume is expressed in ml (35). Immunoblots were probed with rabbit anti-LexA antiserum (1:2000, gift of Barak Cohen, Massachusetts General Hospital) to detect bait proteins and with mouse monoclonal antihemagglutinin antibody 12CA5 (1:1000, Berkeley Antibody Co.) to detect interactant proteins as described (31).
Cell Culture and RNA Isolation-RASMC were isolated from the thoracic aorta of adult male Sprague-Dawley rats by enzymatic digestion and grown in a humidified incubator (37°C, 5% CO 2 ) in Dulbecco's modified Eagle's medium (RH Biosciences, Lenexa, KS) supplemented with 10% fetal calf serum (HyClone Laboratories, Logan, UT), 25 mM Hepes (pH 7.4), 600 mg of glutamine/ml, 100 units of penicillin/ml, and 100 g of streptomycin/ml (growth medium). Cells were passaged every 3-5 days, and cells from passage 4 were used for RNA isolation. RASMC were rendered quiescent by exposure for 72 h to a similar medium in which the fetal calf serum concentration was 0.4% (quiescence medium). Quiescent subconfluent RASMC were then stimulated again with growth medium. NIH 3T3 cells (American Type Culture Collection) were cultured in growth medium from which Hepes had been omitted.
Northern Analysis-Total RNA was harvested from subconfluent RASMC in growth medium and quiescent RASMC at discrete points after serum stimulation with growth medium. Total RNA was isolated by guanidinium isothiocyanate extraction and centrifugation through cesium chloride as described (36). Total RNA (10 g) was separated on a 1.3% formaldehyde-agarose gel and transferred to nitrocellulose filters. The filters were then hybridized with a [ 32 P]dCTP-labeled randomly primed rPM-Scl cDNA probe. The filters were washed in 30 mM sodium chloride, 3 mM sodium citrate, and 0.1% SDS at 48°C and then autoradiographed for 16 h on Kodak XAR film at Ϫ80°C. The filters were also exposed to a phosphor screen, and radioactivity was measured on a PhosphorImager running the ImageQuant software (Molecular Dynamics Inc., Sunnyvale, CA).
In Vitro Binding-GST fusion proteins were expressed and purified, and in vitro binding experiments were performed essentially as described (31,37). In brief, fresh cultures of E. coli (HB101) transformed with pGEX-4T or pGEX-4T-E12-(477-654) were induced with 0.4 mM isopropyl-␤-D-thiogalactopyranoside when the A 600 reached 0.8. After centrifugation the bacterial pellet was resuspended in phosphate-buffered saline and then lysed by sonication and the addition of Triton X-100 to a final concentration of 1%. The clarified lysate was incubated with glutathione-Sepharose 4B (Pharmacia). 35 S-Labeled proteins were generated by the TNT T7/T3 Coupled Reticulocyte Lysate System (Promega) with the pCite4-rPM-Scl and pEMSV-E47 expression plasmids. 35 S-Labeled proteins (4 l) were incubated with GST-E12 and GST beads in 50 mM NaCl and bovine serum albumin (1 mg/ml) for 1 h at 4°C. The beads were then washed four times with 0.1% Nonidet P-40 in phosphate-buffered saline. The proteins were eluted by boiling in SDScontaining sample buffer, fractionated by 12% SDS-PAGE, stained with Coomassie Blue, and exposed to Kodak XAR film.
DNA Transfection-NIH 3T3 cells were transfected by the calcium phosphate method. The E box reporter plasmid (2.5 g) was transfected with pSV␤gal (2.5 g) to correct for differences in transfection efficiency. We cotransfected 10 g of the empty pCR3 vector or 10 g of the pCR3 vector containing full-length rPM-Scl to determine its effect on basal reporter plasmid activity. In the same experiment we cotransfected 1 g of a pEMSV vector containing the full-length E47 insert with 9 g of pCR3 or pCR3 vector containing rPM-Scl or Id1. Luciferase activity was measured in duplicate for all samples with a luminometer (Autolumat 953, EG&G, Gaithersburg, MD) and the Promega luciferase assay system. ␤-Galactosidase activity was also assayed. The ratio of luciferase activity to ␤-galactosidase activity in each sample served as a measure of normalized luciferase activity. Transfection measurements were then reported relative to the average normalized luciferase activity of the pCR3-E box samples. Each construct was transfected in duplicate, and data for each construct are presented as the mean Ϯ S.E. from three separate experiments. For multiple treatment groups, a factorial analysis of variance was applied followed by Fisher's least significant difference test when appropriate. Statistical significance was accepted at p Ͻ 0.05.

RESULTS
Isolation of cDNA Clones Encoding E12-interacting Proteins-A human E12 cDNA spanning amino acids 477-654, containing the basic and helix-loop-helix domains, was fused in-frame with the LexA DNA-binding domain and used in a yeast interaction trap screen (23). This region is devoid of transcriptional activity but retains the ability to interact with MyoD and the Id proteins (11). Forty-two cDNAs were isolated and partially sequenced from an adventitia-stripped rat aorta cDNA library. Twenty-nine cDNAs encoded Id3 (12), and five encoded Id1 (11). As described previously, five cDNAs encoded a ubiquitin-conjugating enzyme, UbcE2A, which promotes the degradation of the E2A proteins by the ubiquitin-proteasome system (31). One of the remaining three was a 0.6-kilobase pair cDNA (clone p17) that shared 81% identity with the 75-kDa hPM-Scl cDNA (19) (Fig. 1, arrow). This partial cDNA was then used to isolate a complete 1.6-kilobase pair cDNA from a RASMC cDNA library. The predicted open reading frame it encoded was 91% identical to the hPM-Scl (Fig. 1). The conclusion that clone p17 is the rat homologue of PM-Scl was confirmed by its immunoreactivity to a rabbit antiserum (gift of Edward Chan and E. M. Tan, Scripps Clinic and Research Foundation, La Jolla, CA) raised against recombinant hPM-Scl (data not shown).
Serum Stimulation Induces rPM-Scl mRNA Expression in Quiescent RASMC-Smooth muscle cells are the dominant cell type in aortic tissue. We performed Northern analysis on total RNA isolated from cultured early passage RASMC to confirm expression of rPM-Scl in this cell type and determine the effect of growth factor stimulation on rPM-Scl expression. rPM-Scl mRNA expression was high in growing subconfluent RASMC and low in quiescent RASMC (Fig. 2, Growing versus 0 h). After the quiescent RASMC had been stimulated with serum, the rPM-Scl mRNA level increased within 6 h and reached a level comparable to that in growing nonconfluent RASMC within 18 h. rPM-Scl expression was also suppressed in quiescent as compared with growing skeletal myoblasts (C2C12), NIH 3T3 fibroblasts, and rat aortic endothelial cells (data not shown).
rPM-Scl Interacts Specifically with E12/E47-To assess the interaction between rPM-Scl and the E12 proteins, we performed in vitro binding assays with GST fusion proteins. The in vitro translation products of full-length E47 and rPM-Scl were retained selectively by GST-E12 in comparison with GST alone (Fig. 3). This result confirmed the binding interaction observed in the yeast interaction trap.
We then used the yeast interaction trap to determine the specificity of rPM-Scl binding to the E2A proteins in comparison with that of other transcription factors expressed as LexA hybrids. A ␤-galactosidase reporter construct was used to provide a measurement of rPM-Scl binding (Fig. 4). The synthesis of each LexA-bait fusion protein was confirmed by immunoblotting yeast extracts with polyclonal antiserum specific for LexA (data not shown). The specificity of the E2A-rPM-Scl interaction was confirmed by the dark hatches in the Galactose panel (Fig. 4), which indicate activation of the lacZ reporter gene.
Because expression of a C-terminal (283-353) rPM-Scl transcription activation domain (TAD) fusion protein is under the control of the galactose-inducible GAL1 promoter, ␤-galactosidase activity is observed only in the presence of galactose. The rPM-Scl (283-353) TAD bound both E12 and E47 (Fig. 4) but not five other transcriptional regulators (Id3, Gax, Myb, OCT-1, and Max).
Map of Regions That Interact with E12/E47 and rPM-Scl-Next we used the yeast interaction trap to map the respective binding domains of E12 and rPM-Scl. We generated three additional rPM-Scl TAD hybrids (Fig. 5, top) to map the region of rPM-Scl that interacts with E12. These constructs were transfected into yeast harboring a LexA-E12-(477-654) construct, which contains the basic and HLH domains, and the ␤-galactosidase reporter plasmid pSH18-34. For each construct, fusion protein biosynthesis was confirmed by Western blot analysis. ␤-Galactosidase production was directly proportional to the degree of hybrid binding (Fig. 5). The C-terminal region (283-353) of rPM-Scl was necessary and sufficient for interaction with E12. rPM-Scl binding to E12 was not altered by the addition of the rest of the molecule (rPM-Scl construct 1-353), whereas deletion of the C-terminal region abolished binding (rPM- Scl constructs 1-288, 1-192, and 1-112). These observations show that the 70 C-terminal amino acids of rPM-Scl are responsible for its interaction with E12.
rPM-Scl Enhances Transcriptional Activation of E47-To determine whether the binding of rPM-Scl to an E2A protein would affect its function, we studied the effect of rPM-Scl on E47-mediated enhancement of transcription of an E box luciferase reporter plasmid (Fig. 7). Cotransfection of full-length rPM-Scl with the E box reporter plasmid did not alter its basal activity. Cotransfection with full-length E47 did increase the activity of the E box reporter plasmid; however, this increase in activity was prevented when E47 was cotransfected with Id1, in keeping with the antagonistic effect of the Id proteins on E2A protein transcription (11). Although rPM-Scl alone had no effect on the activity of the E box reporter plasmid, rPM-Scl increased activation of the reporter plasmid by E47 by 2.4-fold (Fig. 7). To confirm the specificity of the effect of rPM-Scl on the E box reporter plasmid, we cotransfected the rPM-Scl plasmid with a pGL2 control plasmid (SV40 promoter/enhancer luciferase). rPM-Scl did not enhance luciferase activity mediated by pGL2 (data not shown). The increase in E47-mediated E box activation after cotransfection with rPM-Scl (Fig. 7) substantiates a cellular interaction between the two proteins. DISCUSSION hPM-Scl was first identified with autoantibodies from patients with the syndrome for which it is named (20,38). Immunoelectron microscopy identified prominent hPM-Scl staining in the granular layer of the nucleolus and in distinct nucleoplasmic clusters (21), and anti-hPM-Scl antibodies precipitated a multimeric complex of 11 or more peptides and phosphoproteins (39). This complex is distinct from the ribonucleoprotein, and indirect evidence suggests that it associates with DNA rather than rRNA (21,39). The corresponding cDNA was cloned by expression screening with antibodies isolated from patients suffering from the polymyositis-scleroderma overlap syndrome (19). hPM-Scl encodes a 355-amino acid protein whose highly acidic C terminus is responsible for a characteristic pattern of aberrant migration at 75 kDa on SDS-PAGE, despite a predicted size of 39.5 kDa. We have found that rPM-Scl also has an aberrant migration pattern on SDS-PAGE owing to its C terminus (Fig. 5, bottom).
The localization of hPM-Scl to the granular layer of the nucleoli and to distinct nucleocytoplasmic foci suggests that it is associated with the nuclear matrix. The nuclear matrix is a complex lattice of nonhistone proteins that is resistant to nuclease digestion and high salt extraction (40,41). The matrix affects transcription by binding to trans-acting factors that include the steroid receptors (42), ATF (43), OCT-1 (43,44), and the retinoblastoma protein (45). Retention of trans-acting factors in the nuclear matrix can increase or decrease transcriptional activity. In addition, the nuclear matrix orchestrates the formation of distinct subnuclear compartments into specific foci of transcription (46 -48).
To consider the possibility that PM-Scl interacted with different families of trans-acting factors, we tested the specificity of the rPM-Scl-E12/E47 interaction by the yeast two-hybrid assay. rPM-Scl did not interact with the HLH protein Id3 or with the trans-acting factors Max, Gax, Myb, and OCT-1 (Fig.  4). By contrast, rPM-Scl interacted well with E12 and E47. E12 and E47 differ by their HLH domains, which result from differential splicing of the E2A gene. The two bHLH exons encode similar domains that have clearly different affinities for other HLH domains (2). The absence of a difference between E12 and E47 in binding affinity for rPM-Scl implies a non-bHLH binding domain.
We again used the yeast two-hybrid assay to identify the domains through which rPM-Scl and E12/E47 bind. The E2A protein binding epitope is in the C terminus of rPM-Scl. This epitope binds to a 54-amino acid domain upstream of the E12/ E47 bHLH domain. Removal of the basic or HLH domains did not affect the binding interaction of E47 with rPM-Scl. Gen-Bank TM analysis of this 54-amino acid domain revealed conservation of amino acid sequence between the E2A proteins and E2-5/ITF-2. This domain is also conserved among the human, rat, mouse, and guinea pig E2A proteins.
To validate further the cellular interaction between E12/E47 and rPM-Scl, we performed transient transfection assays with an E box-luciferase reporter plasmid. rPM-Scl increased E47dependent luciferase activity without increasing the basal activity of an E box-luciferase or an SV40-luciferase (pGL2) reporter plasmid (Fig. 7). Although a nonspecific reduction in reporter plasmid activity is common in cotransfection assays, an increase in activity indicates a specific interaction. This observation supports the conclusion that rPM-Scl and E12/E47 are interacting partners within mammalian cells.
The regulated expression of rPM-Scl may indicate that this nucleolar protein is a novel non-HLH regulator of the E2A proteins. Like PM-Scl, B23 is a nucleolar protein that associates with a trans-acting factor, YY1. Both B23 and YY1 have been colocalized to the nuclear matrix and the nucleolar remnant (49), and transfection of B23 with YY1 relieves YY1induced repression of transcription. In contrast, our cotransfection experiments (Fig. 6) indicate that rPM-Scl enhances activation of E2A protein transcription. Still, the net effect of B23 on YY1 and rPM-Scl on E12/E47 is to increase transcription.
We have cloned and characterized UbcE2A (31), a ubiquitinconjugating enzyme that binds the same E2A protein domain as rPM-Scl. UbcE2A antagonizes the functional activity of the E2A proteins by promoting their degradation. Also, Sloan et al. (50) have found that B cell differentiation may be regulated by phosphorylation of E47 serine residues 514 and 529, which are contained within the rPM-Scl binding epitope. The convergence in one E2A protein region of these three regulatory mechanisms (degradation, phosphorylation, and possibly nuclear matrix attachment) may indicate an important non-HLH regulatory domain. Transient transfection was performed by the calcium phosphate method. NIH 3T3 cells were cotransfected with the E box-luciferase reporter plasmid and the pSV␤gal plasmid to correct for differences in transfection efficiency. For each category the corrected luciferase activity was normalized to the average luciferase activity of the pCR3 samples. Shown is the mean Ϯ S.E. from three experiments.*, p Ͻ 0.05.