Molecular characterization of the rat insulin enhancer-binding complex 3b2. Cloning of a binding factor with putative helicase motifs.

Cell-specific expression of the rat insulin II gene is in part mediated through an element located in the 5′-flanking region. The element, termed RIPE3b (−126 to −101), confers β-cell-specific expression in conjunction with an adjacent element RIPE3a (−110 to −86). Here we report the characterization of one of the RIPE3b-binding complexes, 3b2. UV cross-linking analysis demonstrated that it is composed of at least three polypeptides: p58, p62, and p110. Furthermore, a cDNA was isolated via expression screening for binding to RIPE3b. Sequence analysis reveals that the encoded protein, designated Rip-1, possessed putative helicase motifs and a potential transcription activation domain. Overexpression of Rip-1 in cells greatly enhances the 3b2 binding complex, suggesting that Rip-1 is involved in the binding of 3b2.

The insulin gene is expressed exclusively in the ␤-cells of pancreatic islets. The tissue specificity results in part from cell type-specific transcription directed by the 5Ј-flanking region of the insulin gene (1)(2)(3). Unlike humans, who have only one insulin gene, rodents (rats and mice) have two nonallelic insulin genes, I and II (rIns I and rIns II) (for review see Refs. 4 and 5). The rIns I and rIns II genes share homologous sequences not only in the coding region but also in their promoters, suggesting that they are controlled by similar regulatory mechanisms. In fact, similar cis-elements have been defined as a result of systematic mutagenesis and in vitro binding analysis within the promoter sequence (6 -9). These elements interact with ubiquitous and/or ␤-cell-specific factors to confer cell type-specific expression (6, 10 -12).
Previous studies performed in our laboratory have identified several important cis-elements, termed the rat insulin promoter elements (RIPEs), 1 in the rIns II gene promoter. One element, RIPE3, located between Ϫ126 and Ϫ86, confers ␤-cellspecific expression when linked to a heterologous minimal promoter in either orientation (13); thus it behaves as a cell typespecific enhancer element. RIPE3 can be divided into two subelements, a and b. Mutation in RIPE3a or RIPE3b in the context of the whole promoter (Ϫ448) drastically reduced the promoter activity by 25-fold. Element a and element b cooperate with each other and give rise to full RIPE3 activity, whereas each element alone has only marginal activity (13). Therefore, it is very likely that multiple protein factors are involved in cell type-specific control. Indeed, it was later demonstrated by in vitro binding assays that cell-specific as well as ubiquitous factors bind RIPE3a and RIPE3b (11).
The RIPE3a element contains an E-box sequence, CANNTG. The E-boxes are recognized by a protein family carrying a basic helix-loop-helix (bHLH) DNA-binding domain (14). The E-boxes in the rIns I (Nir and Far box) and rIns II genes (RIPE3a) have been shown to bind islet-specific bHLH protein complexes (11,(15)(16)(17). Several laboratories have cloned related bHLH proteins that bind to RIPE3a and the corresponding region in the rIns I gene (18 -22). Most notably, our laboratory has recently cloned BETA2 (neuroD), the tissue-specific component of bHLH-binding complex (23,24). On the other hand, two specific 3b protein-DNA complexes were observed by in vitro binding assays (11). The RIPE3b1 complex is islet-specific, whereas the RIPE3b2 complex is expressed in all cell lines examined. A linker substitution mutation in the RIPE3b element (mRIPE3b) not only destroyed the ability of the element to form 3b1 and 3b2 complexes but also abolished the RIPE3b activity in transfection assays (11,13). The sequence of the 3b element shows no obvious homology with any known consensus binding motif and thus is possibly recognized by novel transcription factors.
Here we report the characterization of the RIPE3b2 complex and the cloning of the first RIPE3b-binding protein. Our results show that RIPE3b2 complex is composed of at least three polypeptides: p58, p62, and p110. By screening a HIT cell (hamster insulinoma) cDNA library, we have isolated a clone that binds specifically to RIPE3b. The encoded protein, designated Rip-1, is the hamster homologue of the human and mouse Smbp-2 (25,26), which contains putative helicase motifs and a transcription activation domain. Our data also raise the possibility that Rip-1 might be a component of the multi-subunit RIPE3b2-binding complex. (27), a hamster insulinoma cell line, and HeLa cells were cultured as described previously (11). BHK-21, a hamster kidney fibroblast cell line, was grown in Dulbecco's modified Eagle's medium supplemented with 15% horse serum, 2.5% fetal bovine serum, 100 units/ml penicillin, and 100 mg/ml streptomycin. Nuclear extracts were prepared as described previously (11). ␣TC (28) and ␤TC (29) cells are transgenic mouse ␣and ␤-cell lines, respectively. Nuclear extracts of both cell lines are generous gifts from Dr. Roland Stein (Vanderbilt University).

Nuclear Extract Preparation and Gel Mobility Shift Assays-HIT-
For in vivo expression of Rip-1, 2.5 ϫ 10 5 COS cells were seeded to 60-mm dishes the day before transfection. 5 g of either pSG5 or * This work was supported by National Institutes of Health Grant HD17379. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The  1 The abbreviations used are: RIPE, rat insulin promoter element; pSGRip were introduced into cells by the calcium phosphate method. Nuclear extract was then prepared from the dishes 40 h after transfection using the method described by Attardi and Tjian (30). Gel mobility shift assays were performed as described previously (11).
In Situ UV Cross-linking-UV cross-linking experiments were conducted essentially as described by Wu et al. (31). To prepare the probe, partially complementary oligonucleotides corresponding to sequences from Ϫ111 to Ϫ101 and Ϫ126 to Ϫ101 of the rIns II gene promoter were synthesized. The partially double-stranded oligonucleotide was then filled-in with Klenow enzyme (Promega) in the presence of [␣-32 P]dCTP (ICN), bromodeoxyuridine (Sigma), and other cold deoxynucleotides (Pharmacia Biotech Inc.). The binding reaction was scaled up 5-fold and performed as described before. After electrophoresis, the gel was covered with plastic wrap and irradiated with 254 nm UV light at a distance of 3 cm in the cold room for 1 h. The specific complexes were excised from the gel following autoradiography. The gel pieces were minced, mixed with 20 ml of 2 ϫ SDS gel loading buffer, boiled, and loaded onto an SDS-polyacrylamide gel. After electrophoresis, the gel was dried and subjected to autoradiography.
Library Screening-The HIT cDNA library was a gift from Dr. Larry G. Moss (Department of Medicine, Tufts University). A double-stranded oligonucleotide (RIPE3b) containing sequences from Ϫ126 to Ϫ101 relative to the transcription start site of the rat insulin II gene was concatenated, 32 P-labeled using a nick translation kit (Boehringer Mannheim), and used to screen the library as described previously by Vinson et al. (32) and Singh et al. (33) with minor modifications. Briefly, the nitrocellulose filters were soaked in 10 mM isopropyl-1-thio-␤-Dgalactopyranoside and air dried. LB plates containing phage that had been grown for 4 h at 42°C were overlaid with isopropyl-1-thio-␤-Dgalactopyranoside filters and subsequently incubated at 37°C for 12 h. The filters were then lifted and air-dried for 15 min at room temperature. The filters were subjected to denaturation and renaturation as described in 1 ϫ binding buffer (20 mM Hepes, pH 7.9, 60 mM KCl, and 2 mM MgCl 2 ) containing 6 M guanidine HCl. The filters were then blocked in 5% milk and incubated with 32 P-labeled, multimerized RIPE3b oligonucleotide in binding buffer containing 0.25% milk and 50 g/ml sheared, denatured calf thymus DNA at a concentration of 1 ϫ 10 6 cpm/ml. After incubation at 4°C for at least 4 h, the filters were washed three times in the same buffer for 5 min each, then air-dried, and exposed to x-ray film overnight.
Plasmid Constructions-For nucleotide sequencing, all Rip-1 cDNAs (full-length and truncated forms) were released from the gt11 vectors by EcoRI digestion and were subsequently cloned into the EcoRI site of the pGEM7fz(ϩ) vector (Promega). For in vivo expression of Rip-1 protein, the longest cDNA (l3B6) was cloned into the EcoRI site of pSG5 (Stratagene). The resulting plasmid produced nearly full-length protein in cells (except the three N-terminal amino acids).
Northern Analysis and RNase Protection Assays-Total RNAs from hamster tissues and cell lines were prepared with RNazol B (Biotecx, Houston, TX) or as described by Sambrook et al. (34) using the guanidinium thiocyanate extraction method. Poly(A) ϩ RNA was isolated from total RNA using standard conditions (34). Northern blots were prepared as described (35) using 12.5 g of poly(A) ϩ RNA and probing with a 32 P-labeled, N-terminal EcoRI-EcoRV fragment of the Rip-1 cDNA. RNase protection assays were performed as described by Wu et al. (36) using an in vitro synthesized, 32 P-labeled antisense RNA probe corresponding to a C-terminal NheI-EcoR I fragment of the Rip-1 cDNA. The [ 32 P]UTP-labeled RNA probe (1 ϫ 10 6 cpm/reaction) was incubated with 40 g of total RNA isolated from various hamster tissues overnight at 45°C in 80% formamide buffer (80% formamide, 40 mM Pipes, pH 6.7, 0.4 M NaCl, and 1 mM EDTA). The hybridized mixture was digested with RNase A (40 g/ml) and RNase T1 (75 units/ml, Boehringer Mannheim) at 30°C for 1 h and then was treated with proteinase K (150 g/ml). After phenol/chloroform (1:1) extraction, the reaction was precipitated with ethanol and run on a 6% polyacrylamide sequencing gel.
Generation and Purification of Antibody-To generate antibodies against the cloned factor, the cDNA from one of the initial clones, 3B17, which encodes amino acids 535-857, was cloned into the vector pGEX-3X in frame with the glutathione S-transferase coding sequence and expressed as a fusion protein in Escherichia coli essentially as described (37). Following purification, the fusion protein was run on an SDS/10% polyacrylamide gel. The gel piece containing the fusion protein was excised, mashed, mixed with adjuvant, and injected into rabbits. The sera (immune and preimmune) obtained from the rabbits were either clarified through a protein G-Sepharose column (Pharmacia Biotech Inc.) or affinity-purified as described below. To affinity purify the antibody, the antiserum was first passed through an E. coli protein-Sepharose column made with crude E. coli extracts coupled to CNBr-activated Sepharose 4B (Pharmacia Biotech Inc.). The flow-through was collected and affinity-purified through a second column that contained purified glutathione S-transferase Rip coupled to Sepharose beads.
Immunoblot Analysis-Protein extracts were first separated on an SDS-polyacrylamide gel and then transferred to the Immobilon-P membrane (Millipore) using the Bio-Rad electroblotting system. The membrane was blocked in 1% milk in TBST (20 mM Tris⅐HCl, pH 7.5, 136 mM NaCl, and 0.05% Tween-20) for 30 min at room temperature. The affinity-purified antibody was diluted 250-fold and incubated with the membrane for 1 h at room temperature. The specific signal was detected by ECL chemiluminescence (Amersham Corp.) following the manufacturer's instructions.

RESULTS
Two Specific Complexes Interact with RIPE3b-We have shown previously that at least two specific protein complexes interact with the RIPE3b element: 3b1 and 3b2 (Fig. 1A); 3b1 is islet-specific, whereas 3b2 is ubiquitously expressed (11). The binding site for the 3b2 complex was further analyzed here by methylation interference assay: it contacted the RIPE3b element at Ϫ107G of the bottom strand (Fig. 1B), which overlapped but was distinct from those recognized by 3b1 (Gs at Ϫ107, Ϫ108, Ϫ111, and Ϫ114). Mutation in the region from Ϫ118 to Ϫ111 affected not only 3b1 but also 3b2 binding (Fig.  1A, lane 4), suggesting that nucleotides immediately upstream of Ϫ107 are required for binding 3b2. No interference was seen with methylations in the top strand. The experiment was also repeated with nonislet extracts HeLa and BHK. Each extract showed an identical pattern of interaction (Fig. 1B).
The RIPE3b2 Complex Is Composed of at Least Three Polypeptides-To further characterize the RIPE3b2 complex, we performed in situ UV cross-linking analysis. A gel mobility shift assay was carried out with the HIT nuclear extract and a radiolabeled, bromouridine-substituted RIPE3b oligonucleotide. The gel was then irradiated with UV light, and the RIPE3b2 complex was excised. The protein components of the complex were analyzed by SDS-polyacrylamide gel electro- RIPE3b2 is a specific insulin enhancer-binding complex. A, gel mobility shift assay with the HIT cell nuclear extract and RIPE3b oligonucleotides. The probe (RIPE3b) was labeled as described before (11). Sequences of the wild type and mutant RIPE3b oligonucleotides are shown in Fig. 6A (RIPE3b and m1, respectively). RIPE7 is an oligonucleotide containing the sequence from Ϫ305 to Ϫ281 of the rat insulin II gene promoter. All competitors were used at 50-fold excess. B, methylation interference analysis of the 3b2 complex. Methylation of the probe and isolation of the binding complex was performed as described (11). The source of extracts is indicated at the top of each lane. Only the 3b2 complex was assessed here. Free probe (F) was analyzed in parallel as a control. phoresis and subsequently visualized by autoradiography (Fig.  2). The analysis revealed that the RIPE3b2 complex contained at least three proteins that differed in their sizes: p58 (58 kDa), p62 (62 kDa), and p110 (110 kDa) ( Fig. 2A). The same components were detected in the 3b2 complex isolated from BHK, HeLa, ␣TC, and ␤TC cells (Fig. 2B), which was in agreement with the ubiquitous nature of the 3b2 complex.
Isolation of cDNA Clones That Bind RIPE3b-Previous studies using linker-scanning mutagenesis (8), transient transfection experiments (13), and in vitro protein binding assays (11) demonstrated that the RIPE3b element, spanning from Ϫ126 to Ϫ101 relative to the transcription start site, is important for expression of the rat insulin II gene. To study the molecular mechanism underlying the transcriptional regulation, we set out to clone gene(s) coding for the RIPE3b binding protein(s). A HIT cDNA expression library was screened by using a concatenated, double-stranded RIPE3b oligonucleotide as a probe. Two clones, 3B17 and 3B22, were isolated in the initial screening. Both recombinant phages showed a stronger binding preference to the wild type RIPE3b than to a mutated binding site. Partial sequencing and restriction mapping of the two cDNAs demonstrated that they were overlapping clones. A fragment derived from the 5Ј-end of the 3B17 clone was then used as a probe to rescreen the library. As a result, a 3.3-kb cDNA clone (3B6) was isolated. Sequence analysis of the cDNAs revealed that the clone we had, designated Rip-1, was the hamster homologue of the human and mouse Smbp-2, a protein that binds to the immunoglobulin m chain switch (Sm) region (25,26). Fig. 3A shows the alignment of deduced amino acid sequences from the three species. The open reading frame of hamster Rip-1 encodes a protein with a calculated molecular mass of 108 kDa and is 86.6 and 77.4% homologous to its mouse and human counterparts, respectively. It was also once cloned as the human glial factor-1 (GF-1), a transcription factor that binds and activates promoters of the human neurotropic virus JCV in glial cells (38). As pointed out by Fukita et al. (26), human GF-1 is identical to and is an incomplete version of human Smbp-2.
The deduced amino acid sequence of Rip-1 reveals several interesting features (Fig. 3B). As indicated by Mizuta et al. (25), the polypeptide has putative helicase motifs in the Nterminal half, including a consensus P-loop for ATP or GTPbinding sequence, GPPGTGKT, between amino acids 213 and 220 (39). Also in the N terminus is a region rich in leucine residues. Leucine-rich regions have been implicated to be protein-protein interaction domains in a few other proteins (40,41). Finally, the hamster sequence also reveals a proline-and glutamine-rich region in its C terminus, which might be a potential activation domain as demonstrated in other transcriptional regulatory proteins (42,43). Other features not shown in Fig. 3B include a nuclear localization signal KKKKK (amino acids 860 -864) and several putative protein kinase A and C phosphorylation sites.
Overexpression of Rip-1 Protein Enhances RIPE3b2 Binding-In the process of proving the DNA binding specificity of Rip-1, we encountered the same problem that Fukita et al. (26) FIG. 2. Protein composition of the RIPE3b2 complex. UV crosslinking analyses were performed as described under "Materials and Methods." Radioactive 3b2 complexes were excised from the gel and analyzed on an SDS/8% polyacrylamide gel. Free probe was also excised and analyzed in parallel (A and B, lane 1). Complexes formed with different nuclear extracts were also compared (B). Complexes formed with ␣TC and HeLa cell nuclear extracts were reassayed using a different UV light source (transilluminator, model TM-3, UVP Inc.) with a longer exposure time (C). Arrows indicate positions of three polypeptides with sizes of 110, 62, and 58 kDa, respectively.

FIG. 3. Rip-1 is a hamster homologue of mouse and human Smbp-2.
A, comparison of the hamster Rip-1 with the human and mouse Smbp-2. Deduced amino acid sequences are aligned such that residues identical with and different from hamster Rip-1 are indicated by dashed lines and uppercase letters, respectively. Gaps were created to maximize the alignment and are delineated by dots and lowercase letters. B, schematic representation of the Rip-1 protein product. The cDNA encodes a polypeptide of 989 amino acids with a calculated molecular mass of 108 kDa and several interesting features as described under "Results." P and Q stand for the proline-and glutaminerich region. The GenBank accession number for the hamster Rip-1 is L15625.
had experienced, that is, the overexpressed protein has very weak binding affinity in electrophoretic mobility shift assays. As a result, it is impossible to perform binding assays with materials prepared from in vitro translation or bacterial expression system. Because several DNA-binding proteins had been shown to require accessory factors to bind stably to DNA in vitro, an alternative approach was employed to test the possibility. A plasmid that expressed Rip-1 (without the first three amino acids) under the control of SV40 enhancer was introduced into COS cells by transient transfection. Nuclear extracts were then prepared and examined by in vitro gel mobility shift assay.
To prove proper expression of the transfected plasmid, an antibody was generated against the recombinant Rip-1. As shown in Fig. 4, the affinity-purified antibody recognized an endogenous protein of 110 kDa in either the nuclear (lane 1) or the cytosolic extract (lane 2) prepared from HIT cells. The signal was specific, because it was not detected by the preimmune serum (data not shown), and the signal can be blocked by preincubating the Rip-1 antibody with the recombinant glutathione S-transferase-Rip-1 fusion protein (Fig. 4, lanes 3  and 4).
Upon introducing the Rip-1 plasmid into COS cells, which have lower levels of the endogenous protein, expression of a 110-kDa protein, as detected by Rip-1-specific antibody, increased dramatically (Fig. 5B). As expression of Rip-1 increased, binding of the RIPE3b2 complex was greatly enhanced (Fig. 5A, compare lanes 1 and 2), whereas binding of other complexes was not affected. The increased binding activity can be further competed by unlabeled probe (Fig. 5, lane 3). The results suggest that Rip-1 either is in the 3b2 complex or is related to a component of the 3b2 complex.
GF-1-binding Element Competes with RIPE3b for Forming the RIPE3b2 but Not the RIPE3b1 Complex-It has been shown previously that using 32 P-labeled RIPE3b oligonucleotide as a probe and HIT cell nuclear extract as the protein source, two specific binding complexes, 3b1 and 3b2, are observed. Because overexpression of Rip-1 in vivo greatly enhanced 3b2 binding, we were curious as to whether sequences recognized by the human homologue, GF-1, could compete for 3b2 binding. To this end, excess amount of the GF-1-binding oligonucleotide (GFE) was used as a competitor in the gel shift assay. The sequence recognized by GF-1 in the JCV promoter shows little similarity when compared with the RIPE3b element, except that both have two CAGCC motifs in either a direct repeat or a palindrome (Fig. 6A). As shown in Fig. 6B, wild type RIPE3b oligonucleotide competed with both 3b1 and 3b2 complexes (lane 2), whereas GFE only competed for the binding of 3b2 complex but had little effect on the 3b1 complex (lane 3). Both complexes cannot be competed by m2, a RIPE3b mutant oligonucleotide (Fig. 6B, lane 4). The same results were observed with the BHK nuclear extract (Fig. 6B, lanes 5-8). Taken together, the results clearly indicated that the RIPE3b2 complex consists of a DNA binding activity that recognizes both RIPE3b and GFE sequences and that it is distinct from the RIPE3b1 complex. This again demonstrates the specific relationship between Rip-1 and the RIPE3b2 complex.
Expression of Rip-1 in Insulin-producing and Non-insulinproducing Cells and Tissues-Although it was shown that the human and mouse counterparts of Rip-1 were widely expressed, we were curious about the expression level related to insulin production. To assess the size of the Rip-1 message we probed Northern blots containing BHK poly (A) ϩ RNA with a 0.8-kb fragment corresponding to the 5Ј-end of the cDNA and detected a single band of 3.7 kb (Fig. 7A). RNase protection analysis was used to compare the expression level of Rip-1 in insulin-producing HIT cells and non-insulin-producing BHK cells (Fig. 7B). 32 P-labeled antisense RNA corresponding to the 201 C-terminal nucleotides plus 80 nucleotides from the cloning vector was synthesized in vitro and used as a probe for hybridization. The results demonstrated similar levels of Rip-1 expression in these cell lines. To determine the tissue distribution of Rip-1 RNA, total RNA was prepared from various hamster tissues and was also examined by RNase protection assays. The result revealed the highest amount of expression in brain and testis, moderate expression in heart, spleen, and kidney, and low level of expression in other tissues. No signal was detected using the negative control yeast RNA. DISCUSSION We have characterized an insulin enhancer-binding complex, RIPE3b2, and demonstrated that it is composed of at least three subunits: p58, p62, and p110. We have also isolated a RIPE3b-binding protein, designated Rip-1, which turns out to FIG. 4. Immunoblot analysis of nuclear and cytosolic extracts with the Rip-1 antibody. The analysis was conducted with 110 g of the nuclear extract (lanes 1 and 3) and 130 g of the cytosolic extract (lanes 2 and 4) prepared from HIT cells. The specific signal (110 kDa) is indicated by an arrow. Lanes 1 and 2, blot incubated with the affinitypurified Rip-1 antibody. Lanes 3 and 4, same as lanes 1 and 2 except that the antibody was preblocked with the glutathione S-transferase-Rip-1 fusion protein for 30 min at room temperature. be the hamster homologue of the putative helicase Smbp-2 and the transcription factor GF-1. A complex involving multiple subunits is not without precedents. The same phenomenon has been observed for factors binding to the interferon-stimulated response element and several others. The complex residing on the interferon-stimulated response element is actually composed of four polypeptides; interaction between the four subunits greatly enhanced binding of the complex to the interferon-stimulated response element (44). Several lines of evidence suggest that Rip-1 is involved in the 3b2 complex. First, Rip-1 was cloned by binding to the RIPE3b element. Second, overexpression of Rip-1 in cells greatly enhanced the 3b2 complex (Fig. 5A). Finally, the formation of the endogenous 3b2 complex was competed by RIPE3b and the GF-1-binding sequences that share little sequence similarity (Fig. 6). Because the Rip-1 antibody, although it functioned well in immunoblot analysis, was unable to recognize the native protein and supershift or block the 3b2 complex, we are not able to make a definitive conclusion at the present time. It is possible that the identified clone, Rip-1, is either a component of or related to the 3b2 binding complex.
It is noteworthy that the deduced amino acid sequences of Rip-1 seems to possess both features of helicase and transcription factor. The N-terminal half of Rip-1/Smbp-2 contains putative helicase motifs (25) as well as multiple leucine-rich domains. Sequence comparison shows a high degree of conservation (ϳ85%) in these regions, suggesting that they carry important functions. Contrary to the N terminus, the C terminus is more diverse (Fig. 3A). It contains a single-stranded DNA-binding domain mapped by Fukita et al. (26) in the human Smbp-2 (corresponding to a region in Rip-1 of amino acids 637-783) and a proline-and glutamine-rich domain, which could be the putative activation domain in GF-1. The low degree of homology in the single-stranded DNA-binding domain may explain why there is no apparent sequence similarity except a string of Gs among the elements recognized by Rip-1/ Smbp-2/GF-1. Although there are CAGCC motifs in the elements recognized by Rip-1 and GF-1, they are not present in the Sm region. It is more likely that the protein recognizes the structure rather than the sequence of DNA. Alternatively, there is an additional DNA-binding domain that has yet to be identified. Additionally, differences in the context of response elements may mediate diverse functions such as recombination (in the case of Smbp-2) and transcriptional activation (in the case of GF-1). It is possible that by interacting with different factors the same protein can recognize DNA elements with slightly different sequences and thus results in different cellular responses (transcriptional activation versus recombination). The presence of multiple leucine-rich regions, the putative protein-protein interaction domains (Refs. 40 and 41 and references therein), seems to lend support to the hypothesis. Whether the CAGCC motif is required specifically for transcriptional activation (as in the case of Rip-1/GF-1) remained to be tested.
What is the function of Rip-1/GF-1/Smbp-1 in the insulin gene regulation? Because Rip-1 is expressed about equally in HIT and BHK cells, it is not likely that the protein plays a The signal detected is about 3.7 kb in size. B, RNase protection assays of total RNA prepared from various hamster tissues. The RNA probe was prepared by in vitro transcription as described under "Materials and Methods." Marker (M) was pBR322 cut with HpaII. nt, nucleotides; yRNA, yeast RNA. crucial role in determining tissue specificity. Nevertheless, one cannot exclude the possibility that through some kind of posttranslational mechanism, the protein can only function in certain cell types or in response to specific stimuli. Evidence that suggests such a possibility may be seen in a gel mobility shift assay (Fig. 6B), in which a complex closely related to 3b2 (indicated by a dashed arrow) was observed in insulin-producing HIT cells but not in the fibroblast BHK. Interestingly, the activator GF-1, which has only the C-terminal portion of human Smbp-2, activated transcription from JCV promoters 10fold in glial cells (38), whereas the full-length hamster Rip-1 had at most 2-fold activation (data not shown) when tested on the RIPE3b element in HIT cells. The same result was obtained with a construct that expressed only the C-terminal half of Rip-1. It seems that cellular environment and DNA context play important roles in transcriptional activation.
Recently, a family of transcription factors with putative DNA helicase and ATPase motifs has emerged: to name a few, the yeast protein SNF2/SWI2 (45), which is required for the activation of many yeast genes; the protein hBRG1 (46), which is necessary for normal mitotic growth and transcription; and the basic transcription factor TFIIH (47). The cloned factor Rip-1/ Smbp-2/GF-1 represents another member of this family that combines two functions in one protein: DNA helicase and transcriptional activation. The two functional domains may perform cooperatively, as Laurent and Carlson (45) have suggested for the yeast activator SNF2/SWI2. That is, the protein may facilitate transcription by altering chromatin structure, helping contact between activators and the transcription apparatus, thereby enhancing transcription. Alternatively, two domains can function independently on separate occasions, depending on the composition of associated proteins. Whether the activity of Rip-1/Smbp-2/GF-1 is mediated via the first or second mechanism awaits detailed structure-function analysis.