A nuclear protein derived from brain cells stimulates transcription of the human neurotropic virus promoter, JCVE, in vitro.

The 98-base pair enhancer/promoter sequence is critical for cell type-specific transcription of the human neurotropic viral promoter, JCVE, in glial cells. Transcriptionally active extracts were prepared from glial cells and used to identify cis- and trans-acting regulatory elements that are involved in the glial-specific activation of the JCVE promoter. Results indicate that multiple regulatory sequences within the 98-base pair repeat specifically bind to nuclear proteins present in glial cells and positively regulate viral early RNA synthesis. The central region of the repeat, designated domain-B, interacts with a 45-kDa nuclear protein present in brain cells. This brain-specific protein was purified by conventional and DNA affinity chromatography. Complementation of the highly purified protein with HeLa extract significantly increased JCVE promoter activity. Thus, association of the novel glial-origin transcription factor with its target sequence increases transcription of the JCVE promoter in a non-glial context.

Thus, association of the novel glial-origin transcription factor with its target sequence increases transcription of the JCVE promoter in a non-glial context.
Demyelination in brain of patients with progressive multifocal leukoencephalopathy is caused by the destruction of oligodendrocytes, the myelin producing cells of the central nervous system (CNS)' (l-5). The human papovavirus, JCV, has been repeatedly isolated from brain lesions of patients with progressive multifocal leukoencephalopathy and is thought to be the etiologic agent of this fatal demyelinating disease (3,(5)(6)(7)(8). This virus preferentially infects oligodendroglial cells of the CNS and only propagates in glial cells in tissue culture (2,7,9). We (10) and others (9,ll) have shown previously that the highly restricted host range/tissue specificity of JCV to glial cells may rest in the expression of early viral genes that encode tumor (T) antigen (12). Using transgenie mice, Small et al. (13) have recently shown specific expression of the CNS early protein in the brain and this expression results in a severe hypomyelination of the CNS. These observations suggest that the JCV control region con-* This work was supported by Grant CA47996 awarded by the National Cancer Institute and Grant 922-2-6RG from the AmFAR (to K. K.). 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 USC. Section 1734 solely to indicate this fact.
This work is dedicated in fond memory of George Khoury. I Present address: Dept. of Medicine, M-013M University of California at San Diego, La Jolla, CA 92093. 7 To whom correspondence should be addressed. 'The abbreviations used are: CNS, central nervous system; bp, base pair, DTT, dithiothreitol; Hepes, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis. tains a regulatory sequence(s) that is recognized predominantly by glial cells in the CNS.
The regulatory region of the JCV gene contains two 98base pair (bp) tandem repeats, located on the late side of the origin of DNA replication (14). There is a 15-nucleotide ATrich segment within each of the repeats and one eight nucleotide segment, GTGGAAAG, homologous to the SV40 enhancer core sequence (15), located 5' to the tandem repeats distal from the origin of replication.
Previous studies have indicated that the 98-bp repeat sequence contains enhancer/ promoter activity in primary human fetal glial cells (9,10).
Studies on a number of viral and cellular genes have indicated that the temporal and cell-specific activation of eucaryotic genes by enhancer/promoter elements requires association of &-acting DNA with tram-acting cellular proteins that recognize such DNA elements (16)(17)(18)(19)(20)(21)(22)(23)(24)(25). Recently, gel-shift and UV-cross-linking techniques were used to identify trans-acting factors in glial cells that interact with the JCVE control sequence (26). Results of those experiments indicated that the JCV 98-bp sequence contains several protein-binding domains which are recognized by nuclear proteins present in human brain and HeLa cells. Interestingly, the proteins that interact with the central region of the 98-bp repeat are distinct in HeLa and brain extracts. Thus, it is likely that the association of distinct glial-specific nuclear proteins with the 98-bp sequence mediates the cell-specific transcription of the viral genome. To explore this possibility, we have used in vitro transcription extracts from glial cells to examine the role of each protein-binding domain in transcription of the JCV early (JCVE) promoter, and to identify the glial-specific transregulatory protein(s) that determines viral promoter specificity. We demonstrate that the JCV control region is composed of multiple regulatory elements arrayed in a 98-bp tandem repeat, each of which interacts specifically with nuclear proteins present in glial extract. Among these nuclear proteins, we have identified and purified a 45-kDa brain-specific nuclear protein that, by binding to a central region of the 98-bp sequence, enhances viral RNA synthesis in non-glial extract. Furthermore, in vitro complementation studies suggest that the unique promoter specificity of JCVE may be determined, at least in part, by this novel 45-kDa transcription factor present in brain.

EXPERIMENTAL PROCEDURES
Plasmids-The plasmid pBJC was constructed by inserting the BamHI-digested JCV DNA into pBR322. The pMBP plasmid is a derivative of Pex-I (obtained from A. Roach, Cornell Medical Center, New York, NY) containing 1.3 kilobases of MBP upstream regulatory region. The c-myc plasmid was obtained from D. Hall (Thomas Jefferson University). Plasmid pBJC-CAT contains the JCVE regulatory region in front of the CAT gene (9). Oligonucleotides and Labeled Fragments-Oligonucleotides were synthesized by the phosphoramidite method on an automated syn-13899 thesizer (Applied Biosystems, Inc. shift assay for binding activity.
Transcription of JCV, Promoter in Glial Cell-free Extract- Earlier studies by us (10) and others (11)   level of in vitro transcription in the glial extract, however, was consistently more than (5-8-fold) in HeLa extract, as measured by laser densitometry of the bands (data not shown) (Fig. lA, compare lanes 1 and 3). The in uitro synthesized transcripts were sensitive to 1 pg/ml of a-amanitin, indicating that they were transcribed by RNA polymerase II (Fig. lA,  lanes 2 and 4). Transcription efficiency of the control c-myc promoter revealed higher levels of transcription in the HeLa extract than in the glial extract (Fig. lA, bottom panels).
Whether the reduced level of c-myc promoter activity is the property of HJC glial extract or it is fairly common to all glial extracts and cells remains to be elucidated. It is perhaps noteworthy that the level of c-myc RNA in brain is extremely low (estimated 5-10 molecules/cell).' To further define the optimum protein concentration for JCVn transcription, 500 ng of template DNAs were separately incubated with l-9+1 extracts (4.5-40 pg protein/assay). In general, transcription of the JCV promoter in glial extract occurred in a wider range of protein concentration (13.5-40 pg protein/assay). In the HeLa extract, low levels of fulllength transcript were detected when the template DNA was incubated with 13.5 pg of protein (Fig. 1B). Virtually no transcription off the c-myc promoter was detected with various concentrations of DNA and/or protein in the glial extract (Fig. lB, bottom panels).
Results obtained from these experiments indicated that the glial extract. efficiently transcribes the neurotropic viral promoter in vitro. The differences in JCVn promoter activity between the glial and non-glial extract suggest the presence of a positively acting factor(s) in glial cells and their absence in non-glial cells. Alternatively, a repressor(s) might exist in non-glial cells to exert negative regulations.

Multiple &-acting
Transcriptional Elements Participate in Expression of the JCV, Promoter-In previous studies we demonstrated that the JCV 98-bp regulatory sequence contains multiple protein-binding domains that may influence viral early promoter activity (26). Recently, we performed in vivo transient transfection assays where each binding domain was fused separately to the heterologous SV4On reporter promoter upstream of the CAT gene. The results from those experiments indicated that the three binding regions, A4, B, and C, independently and positively contribute to CAT gene expression in glial cells (10). To examine the role of these protein-binding domains in transcription of the JCVn promoter in glial extract, we performed competition experiments in which the specific DNA-binding proteins were sequestered from the extract by preincubation with an excess amount of oligonucleotides containing the target sequence for binding of the proteins (Fig. 2A). The pretreated extract (12 pg) was subsequently programmed to transcribe the JCVx promoter by addition of 500 ng of the linearized template DNA. Addition of 1 Kg of oligo(A1) showed only a 40% inhibition of JCVE transcription (Fig. 2B, lane 2), whereas preincubation with 1 pg of oligo(A2), oligo(A3), or oligo(B) abolished viral early RNA synthesis (Fig. 2B), lanes 3-5, respectively. Addition of oligo(C) and oligo(A4) reduced the rate of JCVe transcription 8-lo-fold (Fig. 2B, lanes 6 and 7, respectively). Neither of these competitor DNAs showed any effect on the transcription of a heterologous c-myc promoter in the HeLa extract, nor on the activity of glial specific promoter (myelin basic protein) in the glial extract." These results, together with our earlier observations (lo), indicate that the protein-binding sequences, positioned within the JCV control region, differ in their ability to stimulate transcription of the JCVs promoter in glial cells.
To determine whether the JCV &-acting elements exert their effect on the in vitro transcription of JCVz promoter by direct binding to nuclear proteins present in the glial extract, band-shift assay was performed. The "P-labeled oligonucleotides representing these elements (oligo(A), and its derivatives, oligo(B) and oligo(C)) were incubated with the glial extract, and protein-DNA complexes fractionated by gel electrophoresis (30,36,37). With the exception of oligo(A1) (Fig.  3, panel A), which showed no detectable binding activity, the DNA fragments formed multiple complexes by binding to protein components present in glial extract, and these complexes migrated with slower mobilities (Fig. 3, lane 2). When the glial extract was preincubated with 1 pg of the homologous oligonucleotide and used as a source of binding protein, the specific DNA-protein complex was abolished (Fig. 3, compare  lanes 2 and 3). Thus, preincubation of glial extract with 1 pg of competitor DNA fragments effectively depletes specific DNA-binding protein(s) and may result in observable decreased levels of JCVE promoter activity. Purification of a Brain-specific Transcriptional Factor That Stimulates JCV, Promoter Actiuity-The dissimilarity in the electrophoretic mobilities of glial and nonglial nuclear proteins that bind to the central region of the 98bp repeat (26), and the effect of this binding sequence in transcription of the JCVE promoter in uiuo (10) and in vitro (Fig. 2B, lane 5) suggest that the DNA sequences spanning this region, i.e. domain B, play a key role in the strong, glial-specific transcription of the JCVs promoter. Thus, our further investigations have focused on this domain. Using a combination of conventional and DNA affinity chromatography, we partially purified this nuclear protein with binding activity for this domain. Calf brain nuclear extract was found to contain high levels of this DNA-binding protein and was, therefore, used as a source for protein purification. Frozen calf brain stored at -80 "C was slowly thawed at 4 "C and total protein extract was prepared (see "Experimental Procedures"). Fig. 4A shows the purification scheme. The amount of extract from 1 kg of calf brain (approximately 1200 mg of protein) was applied to a phosphocellulose (PII) column washed with buffer "A" containing 0.1, 0.3, 0.5, and 1 M KCl. The active fractions were pooled, and the proteins were precipitated with ammonium sulfate (Cf = 50%). The precipitated proteins were applied to a 3-ml DNA affinity column coupled with oligo(B) (32). The active fractions from the DNA affinity column (first run) were eluted with a buffer containing 0.1, 0.5, and 1 M KCl. Fig. 4B illustrates the binding activities obtained from the phosphocellulose and the affinity column fractionations. The complex formation was abolished when the preheated extract (90 "C) or the proteinase K-treated extract was used. Use of the oligonucleotide-affinity column markedly increased the purity and specificity of the DNA binding activity (See Fig. 4, legend).

Characterization
of the JCV, Brain-specific Enhancer/Promoter-binding Protein-Fractions containing DNA binding activity from the phosphocellulose and the DNA affinity columns were subjected to SDS-polyacrylamide gel electrophoresis and silver staining. The active fractions eluted from the DNA affinity column contained two major species with molecular sizes of 45 and 53 kDa (Fig. 5A, lane 5). We previously showed by UV-cross-linking analysis (26) that two nuclear proteins 45 and 82 kDa from human brain cells bind specifically to the central region of the 98bp sequence. Thus, the 45-kDa species in the silver-stained gel appears to represent the same protein that was identified previously by the indirect approach.
To more precisely characterize this binding protein, a preparative SDS-polyacrylamide gel was obtained, sliced, and the proteins were eluted (Fig. 5A, lane 5). Eluted proteins were subsequently denatured/renatured and examined for binding activity by band-shift assay. Fraction D from the gel slice containing the 45-kDa protein species produced a DNA-protein complex that comigrated with the complex formed with the 0.5 M KC1 fraction from the affinity column (Fig. 5B,  compare lanes 2 and 6). Thus, the JCVE promoter/enhancerbinding protein in calf brain, like that in human brain, has a molecular size of 45 kDa. We have recently identified a similar binding activity in monkey and mouse brain nuclear extracts.3

Purified
Brain-specific Factor Stimulates in Vitro Transcription of the JCVE Promoter in Non-glial HeLa Extract- Binding of the protein components from the crude calf brain protein extract, fractions eluted from the phosphocellulose and the affinity columns are shown. In the last two lanes, the probe was mixed with the preheated (90 "C for 5 min), or proteinase K-(50 gg/ml) treated extract.
As a first step toward analyzing the function of this brainspecific DNA-binding protein, we tested the ability of this protein to stimulate transcription of the JCVE promoter in non-glial HeLa extract. In this experiment, 1 and 3 ~1 of the purified protein was preincubated with the DNA template pBJC. After 15 min, nucleotide triphosphates and crude HeLa extract were added. The reaction mixture was incubated for an additional 60 min at 30 "C. Mixing of 1 and 3 ~1 of the purified protein (approximately 0.1-0.3 pg of protein, respectively) to the HeLa extract resulted in a modest stimulation (2-3-fold) of JCVE transcription (Fig. 6A, compare lanes 1  and 3). Increasing the amount of the purified protein resulted in no further stimulation of the JCVn basal transcriptional activity (not shown). Supplementation of the HeLa extract with purified protein did not increase the transcription of the c-myc or glial-specific promoter such as that for myelin basic protein (MB) (Fig. 6A). Fig. 6B illustrates the quantitative effect of the purified protein on transcription of the JCVn, c- myc, and myelin basic protein promoters in uitro. This experiment suggests that the partially purified 45kDa protein derived from brain represents a novel regulatory factor that activates transcription of the JCVn promoter by interacting with a specific DNA sequence. Because the observed activation of the JCVE promoter upon addition of the purified proteins to the HeLa extract might be obscured by endogenous non-glial factors that recognize the same region of the JCV template DNA (26), the endogenous binding proteins were sequestered by preincubation of the extract with oligonucleotide B. The template DNA and the purified protein were subsequently added to the reaction mixtures. To determine the optimum concentration of the specific oligonucleotide to effectively bind to all endogenous factors, transcription of JCVs promoter in the HeLa extract was carried out in the presence of increasing amounts of oligo(B) DNA. Incubation of 0.01 and 0.1 pg of oligo(B) in the HeLa extract significantly reduced the rate of JCVs RNA synthesis in vitro; with 0.5 pg, transcription of the viral promoter was virtually blocked (Fig. 7A, lane 2). Based on this result, we used 0. Thus, the 45-kDa brain-specific DNA-binding protein, in concert with other transcriptional factors present in HeLa cells, stimulates in vitro transcription of the viral promoter in these extracts.
In the above experiment, the purified factor might have increased transcription of the JCV promoter indirectly by releasing the bound endogenous protein to the competitor oligonucleotide through exchange reactions. In particular, the binding interchanged might occur when these proteins have distinct binding affinities for the cognate DNA sequence. The factor present in the HeLa extract is distinct from its coun- terpart in brain extract and generates discernible DNA-protein complexes (Fig. 8A, compare lanes 2 and 3). Therefore, band shift experiments were performed to determine whether the HeLa and the brain proteins are exchanged in the DNAprotein complex during the incubation period. Limiting amounts of the oligo(B) probe were preincubated with the HeLa extract, and the purified factor was added at 0, 5, and 30 min postincubation. Simultaneous incubation of the probe with the HeLa and brain proteins (0 min) generated two groups of brain and HeLa complexes (Fig. 8B, lane 1). When the purified factor was added at 5 or 30 min, the HeLa complexes were predominant (Fig. 8A, lanes 2 and 3, respectively). These results suggest that after formation of the Hela/ oligo(B) complexes no appreciable exchange occurs upon addition of the purified brain factor. Furthermore, these experiments suggest that the binding affinity of the HeLa factors to the oligo(B) sequence is greater than that of the 45-kDa brain factor. DISCUSSION Progressive multifocal leukoencephalopathy is characterized by a degenerative demyelination in the CNS of immunocompromised patients (1,2). Neuropathological studies of the CNS of expired patients indicate the presence of abnormally giant glial cells and focal lesions with demyelination (1,  3, 6). JCV has consistently been isolated from these focal lesions and shown to be expressed in the glial cells around these lesions (7). It is established that the restriction of JCV propagation to glial cells is due to the expression of the JCV T-antigen which, in turn, is regulated by the 9%bp enhancer/ promoter repeat. Each repeat contains multiple binding domains that interact specifically with nuclear proteins present in brain and HeLa extracts (26). The electrophoretic mobilities of these proteins, which recognize the 5'-and 3'-terminal regions of the repeat (designated by A and C), are similar in brain and HeLa extracts. The proteins which recognize the central region of the 9%bp enhancer/promoter, i.e. domain B, are distinct (26). Fusion of these binding elements, in particular the sequence containing domain B to the SV40 enhancerless promoter showed a profound stimulation of the SV40 early promoter in glial cells (10). These results suggest the presence of an intricate network of &-responsive elements within the JCV 9%bp sequence which are required for the specificity of early gene transcription. In accord with the in uiuo experiments, detailed analysis of the viral regulatory region using band-shift and run-off transcription assays established that the JCV 9%bp sequence encompasses multiple protein binding/transcriptional domains. Each domain distinctly influences transcription of the viral early promoter in glial extracts. Addition of synthetic oligonucleotides to the extract results in a concomitant depletion of a specific DNAbinding protein and reduction of JCV early promoter transcription in glial extract.
Which domain(s) plays a central role in the cell typespecific activation of the viral promoter in glial cells has yet to be identified. Domain B, as previously shown by UV-crosslinking, interacts with the 45-and 88-kDa protein components present in brain nuclear extracts. Interestingly, this domain contains the NF-l-binding sequence that may function as a target site for binding of the ubiquitous transcriptional factor (23). Whether the 45-and/or 88-kDa nuclear proteins are members of the NF-1 family present in brain cells remains to be determined. However, in earlier studies we have demonstrated that the 45-kDa protein specifically recognizes the tetranucleotide repeat, GCCA, within this domain.
The domain B-binding protein has been purified from calf brain by affinity chromatography on the oligo(B) column. Binding activity was localized to a 45-kDa protein. The affinity purified protein was shown to activate transcription of the JCVE promoter in HeLa extract and is a good candidate for the cell type-specific factor that mediates transcription of the JCVE promoter in glial cells. Our results, however, do not exclude the possibility that other binding domains are also involved in the cell type specificity of JCVE transcription. There are also two copies of CAAT box-like sequences in the middle of domain A, designated oligo(A2), that by binding to the CTF/NF-1 may increase transcription of the JCVz promoter. Note that UV-cross-linking experiments demonstrated that the protein interacting with oligo(A2) has a molecular size of 82 kDa (in glial and non-glial cells), which is different from the size of the CTF/NF-1 protein (23,38). Domain C also functions as a transcriptional element in uiuo and in vitro (shown in Fig. 2). This domain contains a TATA sequence and a tract of adenosines (A). The A/T-rich sequence may constitute a functional TATA box for a viral early promoter. It has recently been shown that dA-dT tracts present in the promoter region can stimulate transcription mediated by polymerase II (39). The poly(dA-dT) tracts bind to a 248-residue protein called "datin" found in yeast (40). Whether this novel regulatory domain plays a functional role in transcription of the brain-specific JCVz promoter remains unknown.
Using Ba131 deletion mutagenesis, we have recently identified a silencer sequence positioned near the binding site of the 45-kDa protein within the 98-bp repeat. This silencer sequence interacts with a 56-kDa nuclear protein present in glial and non-glial cells. 4 We speculate that binding of the 45-kDa glial-specific activator protein to JCV DNA may preclude the association of the repressor with the silencer, resulting in active transcription of the viral genome in the infected cells.
Results presented here demonstrate that transcription of the JCV early promoter is under a complex control mechanism that involves several cis-and truns-regulating signals. These elements include a 45kDa protein present in brain that binds to its target site in the JCV enhancer/promoter sequence and increases transcription of the viral promoter in a non-glial extract. Whether this novel brain regulatory factor in concert with other ubiquitous regulatory proteins, stimulates the viral gene expression in lytically infected cells remains to be determined. Purification of nuclear proteins that bind to the other regulatory domains of JCVx will help in better understanding the molecular mechanisms through which the JCV early promoter is actively expressed in glial cells.