THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1993 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 268, No. 31, Issue of November 5, pp. 23093-23097,1993 Printed in U. S. A.

Retinoic Acid Biphasically Regulates the Gene Expression of Hepatitis B Virus Surface Antigen in Human Hepatoma Hep3B Cells*

(Received for publication, May 7, 1993, and in revised form, July 1, 1993)

Shih-Lan HsuSj, Yu-Fen LinS, and Chen-Kung ChouSTll From the SGradunte Institute of Microbiology and Immunology, National Yaw-Ming Medical College and the TDepartment of Medical Research. Veterans General Hospital, Taipei and the $Institute of Biomedical Science, Academia Sinica Taipei 112, Taiwan Republic of China

We used human hepatoma Hep3B/C16 cells as a model to examine the effect of all-trans retinoic acid on the gene expression of hepatitis B surface antigen (HBsAg). Hep3B/Cl6 is a clonal derivative of human hepatoma Hep3B cell which was stably transfected with HBsAg DNA sequences and can produce hepatitis B virus surface antigen. We analyzed the HBsAg prod- uct and mRNA in Hep3B/C16 cells which were exposed to retinoic acid for different periods of time. The level of HBsAg started to increase after 24 h and reached maximum at 48 h of retinoic acid treatment. However, the level of HBsAg expression was severely suppressed compared to the control cells after long term (120 h) retinoic acid treatment. Such biphasic regulation of HBsAg production by retinoic acid was paralleled by the changes of HBsAg mRNA. Nuclear run-on assays also demonstrated that the retinoic acid-mediated reg- ulation was determined at least in part at the transcrip- tional level. Furthermore, an exposure of the cells to retinoic acid for only 8 h was sufficient to show that up- and down-regulation of HBsAg gene occurred 2 and 5 days later. Using a transient expression system, we demonstrated that the retinoic acid response ele- ment is located within the 5”flanking region of the HBsAg gene.

The human hepatitis B virus (HBV)’ is a small DNA virus which causes acute and chronic liver disease and a propensity to develop hepatocellular carcinoma (1). The regulation of the HBV gene expression in humans is still poorly understood. In order to study the host control of HBV gene expression, we used Hep3B/C16 (which is a clonal derivative of the Hep3B cell line) as an experimental system. The Hep3B cell line was established from a human hepatoma biopsy. The morphological characteristics and epithelial cell shape of these cells are compatible with those of liver parenchymal cells. Histology of the Hep3B cells revealed well differentiated hepatocellular carcinoma with a trabecular pattern. Clonally

*This work was supported by National Science Council Grant NSC 81-0419-B075-6 and Department of Health Grant DOH82-HR- COS, Taiwan, Republic of China. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accord- ance with 18 U.S.C. Section 1734 solely to indicate this fact.

11 To whom correspondence should be addressed Dept. of Medical Research, Veterans General Hospital-Taipei, 201, Section 2, Shih- Pai Road, Shih-Pai, Taipei, 11217 Taiwan, Republic of China.

The abbreviations used are: HBV, hepatitis B virus; RAR, retinoic acid receptor; RXR, retinoid X receptor; RARE, retinoic acid re- sponse element; HBsAg, hepatitis B virus surface antigen; EIA, enzyme-linked immunosorbent assay; kb, kilobase(s); DMEM, Dul- becco’s modified Eagle’s medium.

integrated hepatitis B virus DNA sequences are also found in Hep3B cells, which can produce the hepatitis B virus surface antigen (HBsAg) (2, 3). This cell line thus provides an excel- lent model for investigation of the regulation of human plasma protein synthesis and the hepatitis B virus-host cell relation- ship.

We previously demonstrated that retinoic acid can regulate the expression of some secretedplasma proteins; both albumin and transferrin genes were regulated by retinoic acid in a dose-dependent and protein synthesis-dependent manner in Hep3B cells (4). Retinoic acid is a natural acidic derivative of vitamin A, acts as a morphogen or a teratogen, can affect the proliferation and differentiation of various cell types (5), suppresses carcinogenesis in uiu0(6), may be a natural mor- phogen to form a gradient for establishing the anterior-pos- terior pattern formation during chicken limb development (7, 8), and regulates the expression of many genes in various cells (4,9-11). In this report we describe that retinoic acid biphas- ically regulates the expression of HBsAg gene in a dose- response manner, as measured in both a stable transfectant and a transient transfection assay.

MATERIALS AND METHODS

Reagents-All-trans retinoic acid and other chemicals were pur- chased from Sigma. Retinoic acid was dissolved in dimethyl sulfoxide and stored in liquid nitrogen.

Cell Cultures-The hepatoma cell line Hep3B/C16 was a clonal derivative of human hepatoma cell line Hep3B which was stably transfected with cloned 2.4-kb DNA fragment corresponding to the intact transcriptional unit of HBsAg with map position 2839-1896 of HBV (ayw subtype, with the EcoRI site number 0). The Hep3B/C16 cells continuously synthesize and secrete HBsAg into culture medium. The Hep3B/C16 and another human hepatoma Huh7 cells (which have no HBV genomes integrated in its chromosome) were cultured in Dulbecco’s modified Eagle’s medium (DMEM, Flow Laboratories), supplemented with 5% fetal calf serum and antibiotics in a humidified atmosphere of 5% COz, at 37 “C.

HBsAg EIA Analysis-HBsAg was measured by enzyme-linked immunosorbent assays (EIA) using reagents provided by Evernew Biotech., Inc. Retinoic acid did not interfere the HBsAg assay.

DNA Constructions-Fig. 1 shows the recombinant plasmid con- S t N C t S used in this study. The HBV genome used in our recombinant vectors was an ayw subtype, and its cloning was described as follows. For construction of pC21, the genome-length HBV DNA was cleaved at its BglII site and the 2.4-kb BglII (2839-1986) restriction fragment inserted at the BamHI site of pBluescript I1 SK’. Plasmid pC16 was constructed by cleaving both pC21 and pSV2neo-L at the KpnI and Sac1 sites (Fig. l), including the entire HBsAg gene plus part of flanking sequences and a neomycin-resistant gene. pSV2neo-L is a derivative of pSV2neo(l2) which was constructed by cut pSV2neo at the BamHI site and ligated with a synthetic SacI/KpnI linker. pBsm was constructed by cleaving both pC17 and pSV2neo-L at KpnI and Sac1 site. pC17 was constructed by cleaving pC21 at the SmaI and BsmI site, then 3’-end filled in, and self-ligated.

RNA Isolation and Northern Blot Analysis-Total cellular RNA was isolated by lysing cells with guanidinium isothiocyanate (4). The

23093

23094 Retinoic Acid Regulates Expression of HB V Surface Antigen

PC21 PC16 pBsm

FIG. 1. Construction of the recombinant plasmids pC21, pC16, and pBsm. The HBV genome used in our recombinant vectors was an ayw subtype, and its cloning was described as follows: for construction of pC21, the genome-length HBV DNA was cleaved at its BgnI site, and the 2.4-kb BgnI (+2839 to +1986) restriction fragment was inserted at the BamHI site of pBluescript I1 SK'. Plasmid pC16 was constructed by cleaving both pC21 and pSV2neo-L at KpnI and SmI sites, including the entire HBsAg gene plus part of flanking sequences and a neomycin-resistant gene. pSV2neo-L is a derivative of pSV2neo which was constructed by cut pSV2neo at the BarnHI site and ligated with a synthetic SmI/KpnI linker. pBsm was constructed by cleaving both pc17 and pSVZneo-L at the KpnI and Sac1 site. pC17 was constructed by cleaving pC21 at the SmaI and BsmI site, then 3'-end filled in, and self-ligated. Abbreviations for enzymes: B, BgZII; Ba, BamHI; Bs, BsmI; E, EcoRI; K , KpnI; S, SacI; Sm, SmaI; H, HindIII; P, PuuII.

RNA was pelleted through a CsCl cushion and precipitated with ethanol. The total RNA was electrophoresed on a 1.0% denaturing formaldehyde-agarose gel, and then transferred on to a nitrocellulose paper. The bound RNA on nitrocellulose paper was treated as follows for hybridization with the [a-32P]dCTP-labeled probe at 42 "C. After hybridization, the filters were washed in 0.5 x SSC, 0.5% SDS at 55 'C, and autoradiographed at -70 "C. The autoradiographs were quantified by a laser densitometer.

Calcium Phosphate Transfection-Huh7 cells were cultured in Dul- becco's modified Eagle's medium supplemented with 5% fetal calf serum. 10 wg of the respective plasmid DNA and 5 pg of B-galactosid- ase plasmid (internal control for transfection efficiency) were cotrans- fected by a modified calcium phosphate coprecipitation procedure (13). Cells were exposed to the precipitate for 16 h, then refed with fresh DMEM medium supplemented with 5% fetal calf serum in the presence or absence of lo-' M retinoic acid for an additional 2 and 5 days. At the end of the experiments, medium were collected for EIA, and cells were harvested for RNA isolation. The @-galactosidase activity was measured as previously described (14).

Nuclei Isolation and Run-on Assay-Nuclei were isolated from the Hep3B/C16 cells according to the method described by Brown et al. (15). Cells were homogenized in a Dounce homogenizer with buffer A (10 mM Tris-HC1, pH 8.0, 5 mM dithiothreitol, 0.3 M sucrose, and 0.1% Triton X-100). Nuclei pellets were obtained by centrifugation at 160,000 X g at 1 "C for 90 min. The pellets were then washed with buffer C (containing 50 mM Tris-HC1, pH 8.0, 0.1 mM EDTA, 5 mM MgCl,, and 0.5 mM dithiothreitol), and then resuspended in the storage buffer (40% glycerol, 50 mM Tris-HC1, pH 8.0, 5 mM MgC12, and 0.1 mM EDTA), and stored at -70 "C before use. Transcription assays were performed as previously described by Matrisian et al. (16). Nuclei (5 X IO') were incubated in the reaction buffer for 45 min at 26 "C (4). The labeled RNA products were purified using DNase, proteinase K, and ethanol-ammonium acetate precipitation. The labeled RNA species were then hybridized for 3 days on a nitrocellulose filter immobilized with various DNAs, including pBR322, actin, and HBsAg DNA fragments. After 3 days, the filters were washed, air-dried, and autoradiographed at -70 "C.

RESULTS

Regulation of HBsAg Production in Hep3BIC16 Celk by Retinoic Acd"Hep3B/C16 cells were cultured with or with- out retinoic acid (10 /IM) for up to 5 days. The growth rate of Hep3B/C16 cells was only slightly inhibited by the addition of retinoic acid. However, the production of HBsAg was biphasically regulated by retinoic acid treatment, as shown in

1600 0 0 control

400 n cy 2 CD

m I 0

0 1 2 3 4 5 30

Time (day)

FIG. 2. Effects of retinoic acid (RA) on the secretion of HBsAg protein in the Hep3BIC16 cells. Cells were incubated in the absence or presence of retinoic acid M), cell numbers were counted, and the amounts of the HBsAg in the cultured medium were measured at various indicated times. Each point represents pooled means + S.D. of 9 wells from three independent experiments.

Fig. 2. The production of HBsAg by Hep3B/C16 cells was induced up to 2-3-fold during the first 48 h of exposure to retinoic acid, but then gradually declined after prolong treat- ment of retinoic acid. About 70% of HBsAg production was suppressed by retinoic acid treatment relative to the control cells at day 5. This retinoic acid-mediated suppression of HBsAg production was also observed at longer times of reti- noic acid treatment (data not shown).

Effect of Retinoic Acid on the mRNA and Transcription of the HBsAg-To examine the change of HBsAg mRNA during the retinoic acid treatment, the steady-state level of HBsAg mRNA was assayed by Northern blot analysis using a cloned HBsAg DNA as the probe. One major 2.1-kb mRNA of HBsAg which was transcribed from stably transfected HBsAg genome was detected. The parallel increase and decrease of HBsAg mRNA during the retinoic acid treatment suggests that both up- and down-regulation of HBsAg gene expression by reti- noic acid were mainly on the mRNA level (Fig. 3a). When

Retinoic Acid Regulates Expression of HB V Surface Antigen 23095

A 1 2 3 4 5 Time [day]

- + - + - + - + - i- R A [ 1 0 - 5 M ] "I"

._. .. - HBsAg 2.1 kb

- Actin

A 11.2

0 " " ' ~ ' ~ ' ~ ' ' ' ' l 1

0 8 16 2 4 32 40 4 8

4 0.8

10.6

c 0

E C 0 ln ln

Q. Kl I) ln w. 7J m ln

.-

E

2 day 5 day " -5 -6 -7 -8 -9 0 -5 -6 -7 -8 -9 0 R A [log M ]

~ -Act in

FIG. 3. Effect of retinoic acid (RA) on steady-state HBsAg mRNA in stable transfectant Hep3B/C16 cells. Cells were in- cubated in the absence (-) or presence (+) of lo-' M retinoic acid, a t various indicated times (a) , or incubated with various concentrations of retinoic acid for 2 or 5 days ( b ) . After treatment the total RNA was extracted and analyzed using Northern blot analysis, as described under "Materials and Methods."

2 5 Time (day)

- + - + R A ( I O - ~ M )

- - - 0 - -a-Fetoprotein

" 0 +Albumin

+ pBR 322 - - +Actin

o 0 0 -HBsAg - - "Transferrin

FIG. 4. Nuclear run-on analysis of HBsAg gene transcrip- tion rate in Hep3B/C16 cells by retinoic acid (RA) treatment. Cells were cultured in the absence (-) or presence (+) of retinoic acid

M) for 2 and 5 days as described. The nuclei were then isolated from control and retinoic acid-treated cells, and transcription was performed in the presence of [(u-~'P]UTP as described under "Mate- rials and Methods." The labeled RNAs were hybridized on to nitro- cellulose filters containing HBsAg, actin (positive control), pBR322 (negative control), albumin, transferrin, and a-fetoprotein DNA frag- ments, as described under "Materials and Methods."

the cells were treated with different concentrations of retinoic acid, both induction and suppression of HBsAg gene expres- sion at day 2 and day 5 of retinoic acid treatment followed a similar dose-dependent response (Fig. 3b).

To determine whether the biphasic regulation of HBsAg gene expression by retinoic acid treatment were either at the transcriptional or post-transcriptional levels, the rate of tran-

Time of R A treatment

B " 2 5 Time [day]

- + - + RA [ 1 0 - 5 M ]

FIG. 5. Effect of retinoic acid (RA) on the expression of HBsAg at various times of treatment in Hep3B/C16 cells. Cells were cultured without or with retinoic acid (lo-' M). After 2,4,8, 16, 24, or 48 h, cells were washed with fresh medium, and continuously cultured in 5% fetal calf serum-DMEM medium for 2 or 5 days. The cell numbers were counted, and ( a ) the compared ratio of HBsAg protein in cultured medium were calculated at 2 days (0) and 5 days (W), and ( b ) the Northern blot analysis of 8-h retinoic acid treatment was detected.

scription of HBsAg gene was examined by a nuclear run-on assay. As shown in Fig. 4, retinoic acid caused a 2-fold increase in the HBsAg transcriptional rate in acute retinoic acid- treated nuclei and an -60% decrease in chronic retinoic acid- treated nuclei. We also found that retinoic acid can up- and down-regulate transferrin and albumin transcription in Hep3B/C16 cells, which is consistent with our previous ob- servations (4). This result indicates that the regulation of HBsAg gene expression by retinoic acid was primarily at transcriptional level.

The Irreversible Effect of Retinoic Acid on the HBsAg Expression-We next investigated whether continuous expo- sure of retinoic acid to the cells is essential for exerting its biphasic regulation on HBsAg gene expression. We treated the Hep3B/C16 cells with M retinoic acid for different periods of time, from 2 to 48 h. After retinoic acid treatment, cells were washed thoroughly and cultured in medium without retinoic acid for 2 or 5 days. As shown in Fig. 5, even an 8-h exposure to retinoic acid was sufficient to establish the bi- phasic regulation of both HBsAg protein (Fig. 5a) and its mRNA (Fig. 5b) at 2 and 5 days later.

Effect of Retinoic Acid on Transient HBsAg Expression-

23096 Retinoic Acid Regulates Expression of HB V Surface Antigen

pC16 PC21 pBsm Plasmid - 2

- - 5 2 5 2 5 Time (day)

- + - + - + - + - + - + R A ( 1 0 - 5 ~ )

- HBsAg 2.1 kb

-Actin

FIG. 6. Effect of retinoic acid (RA) on the level of HBsAg steady-state mRNA in transient transfection system in Huh7 cells. Cells were transfected with 10 pg of tested plasmid (pC16, pC21, or pBsm), and 5 pg of @-galactosidase plasmid (internal control for transfection efficiency) were cotransfected by a modified calcium phosphate coprecipitation procedure. Cells were exposed to the precipitate for 16 h, then refed with fresh 5% fetal calf serum-DMEM medium in the presence (+) or absence (-) of M retinoic acid for an additional 2 or 5 days. At the end of the exDeriments, cells were harvested for total RNA isolation. Northern blot analyses were performed as described under “Materials and Methods.’;

We further examined whether there is a cis-element in HBsAg gene which was responsible for the retinoic acid regulation. We transfected HBV negative human hepatoma Huh7 cells with plasmids containing the intact HBsAg coding sequence with different lengths of promoter region of HBsAg gene. As shown in Fig. 6, one major 2.1-kb transcript of HBsAg was detected and the biphasic regulation of HBsAg mRNA by retinoic acid was also observed in such a transient expression system. However, when the promoter of HBsAg mRNA was deleted from -343 to -182 bp, we still observed the basal level of HBsAg mRNA, but the biphasic response of HBsAg to retinoic acid was completely abolished.

DISCUSSION

The most striking feature in this study is the observation that retinoic acid biphasically regulated the gene expression of HBsAg in human liver cells. The expression of HBsAg gene in human hepatoma cell has been reported to be up-regulated by 2 days of glucocorticoid treatment (17). The glucocorticoid response element in HBV genome also has been identified and localized within the coding sequence of the HBsAg gene (18). Since RU38486, a specific antagonist of glucocorticoid receptor which can block glucocorticoid, induced HBsAg expression but cannot abolish retinoic acid-mediated induc- tion of HBsAg expression (data not shown), the induction of HBsAg by retinoic acid probably is not mediated through the same pathway as glucocorticoid. We also observed that insulin suppresses the HBsAg expression in Hep3B cells, but the suppression of HBsAg expression by insulin was evidenced only after 3 h of treatment. It is unlikely that insulin and retinoic acid share the same pathway to down-regulate HBsAg expression. Since the expression of HBsAg in the control cells increased with the cell density as reported before (19), we cannot tell whether the long term retinoic acid treatment really suppresses or just abolishes the cell density-induced HBsAg expression.

The up-regulation of HBsAg expression by short term retinoic acid treatment in human liver cells is consistent with the recent observation that retinoid X receptor can bind and trans-activate the enhancer 1 of HBV genome (20). However, whether the RARE in the enhancer 1 of HBV genome is responsible for the induction of HBsAg by short term retinoic acid treatment is still unclear. A more vigorous approach, such as site-directed mutagenesis of the RARE, is necessary to elucidate the physiological role of the RARE in HBsAg gene expression in vivo.

The minimum exposure of the cell to retinoic acid in order

to establish the suppression of HBsAg expression is only 8 h. It is conceivable that retinoic acid may trigger a slow dedif- ferentiation process which causes the cell to lose its respon- siveness to retinoic acid and to repress the cell density- induced HBsAg expression. The observation that albumin synthesis was completely suppressed after 5 days of retinoic acid treatment (data not shown) is consistent with this hy- pothesis. Similar suppression of HBsAg expression was also observed in the transient transfection assay in another HBV negative human hepatoma Huh7 cell. We believed that this novel biphasic regulation of HBsAg expression by retinoic acid may represent a general response of human liver cell to retinoic acid.

From the relatively slow kinetics of the retinoic acid-me- diated effects on HBsAg gene expression, it is tempting to speculate that the response may be a late retinoic acid- mediated response and controlled by indirect action. The effects of retinoic acid on genes expression mediated through the interaction of some nuclear receptors are well known. For example, RXR-a can interact with RARs, thyroid receptors, vitamin D3 receptors, or COUP-TF to form a homo-or het- erodimer and regulates the transcriptions of the target genes (21-26). Therefore, in human hepatoma cells, retinoic acid might induce some regulatory molecules to repress the acti- vating function of RXR-a and abolish its stimulatory effect toward HBsAg expression. Furthermore, the promoter dele- tion experiment suggests that the sequence mediating the effects of retinoic acid is located within the 5’-end flanking sequences of HBsAg gene (between -343 and -182). This promoter region contains no consensus sequence reported for RARE. Therefore, either a novel RAR is involved or retinoic acid could regulate HBsAg expression indirectly through changing activities of some other transcriptional factors.

REFERENCES 1. Beasley, R. P. (1988) Cancer 61,1942-1956 2. Aden, D. P., Fogel, A,, Plotkin, S., Damjanov, I., and Knowles, B. B. (1979)

3. Knowles, B. B., Howe, C. C., and Aden, D. P. (1980) Science 209,497-499 4. Hsu, S. L., Lin, Y. F., and Chou, C. K. (1992) Biochern. J . 283,611-615 5. Roberts, A. B., and Sporn, M. B. (1984) in The Retinoids (Sporn, M. B.,

Roberts, A. B., and Goodman, D. S., eds) Vol. 2, pp. 210-286, Academic Press, New York

Nature 282,615-616

6. Sporn, M. B., and Roberts, A. B. (1983) Cancer Res. 43,3034-3040 7. Eichele, G. (1989) Trends Genet. 5 , 246-249 8. Wanek, N., Gardiner, D. M., Muneoka, K., and Bryant, S. V. (1991) Nature

9. Simeone, A,, Acampora, D., Arcioni, L., Andrews, P. W., Boncinelli, E., and Mavilio, F. (1990) Nature 346,763-766

10. Schule, R., Rangarajan, P., Yang, N., Kliewer, S., Ransone, L. J., Bolado, J., Verma, I. M., and Evans, R. M. (1991) Proc. Natl. Acad. Sci. U. S. A.

11. Okazawa, H., Okamoto, K., Ishino, F., Ishino-Kaneko, T., Takeda, S.,

350,81-83

88,6092-6096

12.

13.

14.

15.

16.

17.

18.

Retinoic Acid Regulates Exprc Toyoda, Y., Muramatsu, M., and Hamada, H. (1991) EMBO J. 10,2997-

Byvoet, P., Barber, M., Amidei, K., Lowell, N., and Trudeau, W. (1986) 3005

Schule, R., Muller, M., Otsuka-Murakami, H., and Renkawitz, R. (1988) Biochirn. Bwphy. Acto 867,152-162

Neilsen, D. A., Chou, J., Mackrell, A. J., Casabadan, M. J., and Shiner, D. Nature 332,87-90

Brown, A. M. C., Jeltsch, J. M., Roberts, M., and Chambon, P. (1984) Proc. F. (1983) Pmc. Natl. Acad. Sci. U. S. A. 80,5198-5202

Matrisian, L. M., Glaichenhaus, N., Gesnel, M. C., and Breathnach, R. Natl. Acad. Sci. U. S. A. 81,6344-6348

Chou, C. K., Wang, L. H., Lin, H. M., and Chi, C. W. (1992) Heputobgy

Tur-Kaspa R., Burk, R. D., Shaul, Y., and Shafritz, D. A. (1986) Proc. Natl.

(1985) EMBO J. 4,1435-1440 16,13-18

Acad. Sci. U. S. A . 83,1627-1631

ssion of HBV Surface Antigen 23097 19. Aden, D. P., Fogel, A., Plotkin, S., Damjanov, I., and Knowles, B. B. (1979)

20. H:an2B., and Siddigui, A. (1992) Proc. Natl. Acud. Sci. U. S. A . 89,9059- Nature 282,615-616

21. Vzi's" G Made S Gold J. D Leid M. Lutz Y. Gaub, M. P., Chambon, P., &dGudas,'Ly(199lj EiBO J.'lO: 1149-1158

22. Marks, M. S., Hallenbeck, P. L., Nagata, T., Se am, J. H., A pella, E., Nlkodem, V. M., and Ozato, K. (1992) EMBO 211,1419-14f5

23. Zhang X. K., Hoffmann, B., Tran, P. B. V., Graupner, G and Pfahl, M. (1962) Nature 365,441-446

24. Kliewer, S. A., Umesono, K., Mangelsdorf, D. J., and Evans, R. M. (1992) Nature 366,446-449

25. Yu, V. C., Delsert, C., Ande,men, B., Holloway, J. M., Devary, 0. V., Naar, A. M., Kim, S. Y., Boutm, J. M., Glass, C. K., and Rosenfeld, M. G.

26. Tran, P., Zhang, X. K., Salbec, G., Hermann, T., Lehmann, J. M., and (1991) Cell 67,1251-1266

Pfahl, M. (1992) Mol. Cell. Bml. 12,4666-4676