Eur. J. Biochem. lY4, 323-330 (1990) (C] FEBS 1990

Structural characterization of the human DNA topoisomerase I gene promoter Norbert KUNZE, Michael KLEIN, Arndt RICHTER and Rolf KNIPPERS Division of Biology, Univcrsity of Konstanz, Federal Republic of Germany (Received June 27, 1990) - EJB 90 0749

We have isolated a genomic DNA fragment from HeLa cells containing the promoter region and the first two exons of the human gene encoding DNA topoisomerase I (hTOPI). Transcription of hTOPl mRNA initiates at multiple sites which are clustered 247 nucleotides and 210 nucleotides upstream of the translation-initiation site of the protein coding region. The nucleotide sequence of the region preceding the transcription-initiation sites is G/C rich and contains sequence motifs which are known binding sites of the transcription factors Octl (octameric transcription factor l), Spl and AP2 (activator protein 2). Furthermore, one CAMP-responsive element is present 50 nucleotides upstream of the transcription-initiation site nearest the 5’ end. Neither TATA nor CAAT boxes were found in the promoter region of the hTOPl gene. A 918-bp fragment containing the sequence elements described above drives the transient expression of a chloramphenicol acetyl transferase (CAT) gene sequence in transfected HeLa and 293 cells. In addition we analyzed a 10-kb fragment containing the promoter and exons 1 and 2 for regions of DNase I hypersensitivity. We detected one prominent DNase-I-hypersensitive region in the promoter close to the putative transcription-factor-binding sites and several weaker regions in intron 2. Human type I DNA topoisomerase (hTOP1) is a monomeric protein with an apparent molecular mass of 100 kDa. In vitro, the purified enzyme catalyzes the relaxation of positive and negative superhelical turns in DNA molecules by successive cycles of single-strand breakage and rejoining of the phosphodiester bonds of the DNA backbone (reviewed in [l -31). About 106TOP1 molecules/nucleus are present in mammalian cells. They are translated from a 4.2-kb poly(A)+ mRNA which is encoded by the single hTOPl gene located on chromosome 20q11.2-13.1 [4, 51. In the eukaryotic cell the double-stranded DNA is organized as a nucleoprotein complex and thereby topologically restrained. Transient changes in the torsional tension of this DNA molecule can be generated by exogeneous stimuli or induced endogeneously by the processes of transcription or DNA replication. These changes result in an alteration of twist and writhe in topologically restrained DNA molecules. Active topoisomerases recognize these topological changes and catalyze the relaxation of DNA by changing the linking number (reviewed in [6]). There is overwhelming evidence that TOP1 is one of the two major enzymes controlling the torsional tension of chromatin in eukaryotic cells. It has been shown that TOP1 (together with topoisomerase 11) acts as swivelase in DNA replication and removes the superhelical tension which would otherwise accumulate as a consequence of replication fork Correspondence to N. Kunze, Division of Biology, University of Konstanz, W-7750 Konstanz, Federal Republic of Germany Note. The novel nucleotide sequence data published here has been deposited with the EMBL sequence data banks and is available under accession number X52603. Abbreviations. AP2, activator protein 2; hTOPl, human DNA topoisomerase type I; hTOPI, human DNA topoisomerase type I gene; Octl, octamer transcription factor 1 ; PMA, 4/?-phorbol 12myristate 13-acetate. Enzymes. Restriction endonucleases BamHI, EcoRI, HindIII, KpnI, PvuII, SmaI, Sphl, SstI and XhoT (EC 3.1.21.4); avian myeloblastosis virus reverse transciptase (EC 2.7.7.49); nuclease S1 (EC 3.1.30.1); pancreatic RNase (EC 3.1.27.5); DNA polymerase (Klenow fragment) (EC 2.7.7.7); DNaseI (EC 3.1.21.1); proteinase K (EC 3.4.21.14); chloramphenicol acetyl transferase (EC 2.3.1.28).

movement (reviewed in [7]). Furthermore the enzyme is involved in the transcription of protein-coding genes and essential for transcription of rRNA genes [8, 91. It has been suggested [lo] that TOPI functions as a topological sensor and regulates the conformation of the chromosomal domain to which the enzyme is bound. Consistent with this hypothesis, the enzyme is present in the nucleus in large amounts corresponding to about one enzyme molecule/l0 nucleosomes. During the S-phase of the cell cycle, when the amount of chromatin is doubled, the TOPl protein level and the relaxation activity of the enzyme increases twofold [ l l , 121. This shows that the gene is regulated in a way that ensures a constant relationship between the amount of TOP1 and the amount of chromatin. Transcription of the TOPl gene seems to be highly regulated since the transcriptional activity of the gene as well as TOPl mRNA levels are affected by a variety of environmental and endogeneous stimuli: Duguet et al. [13] reported that the amount of TOPl mRNA increased 20-fold in regenerating rat liver and, more recently, Romig and Richter have shown a sixfold increase in mRNA levels in serum-stimulated cell culture cells [12]. Furthermore, infection of HeLa cells with adenovirus 5 results in a fivefold increase in the transcriptional activity of the hTOPI gene followed by an increase in the TOPI mRNA level [14]. Finally, exposure of human skin fibroblasts to 4p-phorbol 12-myristate 13-acetate (PMA) causes an immediate and transient increase in the level of hTOPI mRNA [15]. Reduced levels of TOPl mRNA and enzymatic activity are reported for an established murine leukemia cell line, P388PC, which is resistant to the TOPl inhibitor SKF 1048644 [16]. The fact that this reduced level of expression of TOP1 seems to be ballanced by an increase in the expression of TOP2 in this cell line, may indicate that eukaryotic topoisomerase genes are in some way autoregulated by topoisomerases. As a first attempt to investigate the regulation of the hTOP1 gene expression, we have isolated a fragment of genomic DNA containing the promoter of the hTOPl gene, the first two exons and 5 kb of the 5’ flanking region. Here we report the molecular structure of this genomic fragment.

324 We have identified the transcription-initiation sites of the hTOPl mRNA and have characterized the 5' sequences necessary for promoter activity. We have also mapped a strong DNase-I-hypersensitive region residing within the promoter of the topoisomerase I gene.

I

I

0

5

E

S I

I

/

I

\

/

Exon

I " ESp

\

Sm

7. E

Sm

1

2

3 kb

Fig. 1. Map of the genomic DNA insert in the phage clone .J7.l. A restriction map of the 5' end of the hTOPl gene is shown (upper). The first two exons are indicated by stipplcd boxes in the enlarged map of the 2.7-kb EcoRI restriction fragment (lower). The scquencing strategy is indicated by arrows below the 2.7-kb fragment. B, BarnHI: E, EcoR1; H, HindIII; K, KpnI; P, PvuII: S, SstI; Sm, S m a I ; Sp, SpkI; x, XhoI

Isolation of'u gcriornic IiTOPI clone, subcloning und scquencc mnal~.si.s

A bacteriophage iL47.1 library of human genomic lymphocyte DNA was kindly provided by B. Horsthemke. Phages were plated on E. coli host strain LE392 at 5 x lo4 phages/ 150 mm dish. Duplicate nitrocellulose filter lifts from 20 plates were probed with "P-labeled synthetic oligonucleotides corresponding to nucleotides 49 - 75 and to nucleotides 22 - 39 of the published hTOPl cDNA-sequence [18]. One positive phage clone, 37.1, was identified and purified to homogeneity. Sequences complementary to the cDNA and flanking regions were subcloned into cloning vectors pUC18 or M13mp18/19 and sequenced by the dideoxy-chain-termination method [19]. Primer-extension uncily,ris 1

In order to generate a single-stranded 32P-labeled DNA probe, the oligonucleotide described above was annealed to single-stranded M 13mpl8 DNA containing the 1 kb EcoRI -

X

I

-+-

0

Chemicals were purchased from Merck/Darmstadt, biochemicals and enzymes from Boehringer/Mannheim and radionucleotides from Amersham/Braunschweig. HeLa cells and human embryonic kidney cells transformed by the adenoviral early region (293 cells) were from Flow Laboratories. Restriction enzyme mapping, gel electrophoresis and blot hybridization were performed as described by Maniatis et al. in [17]. The oligonucleotides used in this study were prepared on a 380B DNA synthesizer (Applied Biosystems).

Nuclease-S,-protet tron u m i y

H I

---+ -+-

Materiul und ytmwd rnetliods

Total HeLa cell RNA was prepared as described [20]. The oligonucleotide 5'-AGACTCCAGAAACGGCTGAGGCTGCTA-3' corresponding to nucleotides 49 - 75 of the hTOPI cDNA [18] was used as primer. This oligonucleotide was 5'end labeled using polynucleotide kinase and [ Y - ~ ~ P ] A to T Pa specific activity of approximately 3 x 10' cpm/pg. Annealing of the 3ZP-labeledoligonucleotide (4 x lo5 cpm) to 40 pg total HeLa cell RNA was performed in 0.9 M NaCI, 0.15 M Mops acetate (pH 7.4) and 1 mM EDTA by heating to 75°C and subsequent cooling to room temperature. After ethanol precipitation, the pellet was redissolved in 24 p1 50 mM Tris HCl (pH 8.0) 5 mM MgC12, 5 mM dithiothreitol, 50 mM KCI, 50 pg/inl bovine serum albumin, 0.6 mM of all four deoxynucleoside triphosphates and 50 units RNasin and incubated at 42 "C for 30 min with 1 7 or 50 U avian myeloblastosis virus reverse transcriptase (Genofit). The reaction was stopped by addition of 1 pi 0.5 M EDTA. Subsequently, 1 pg pancreatic RNase was added and the mixture was incubated at 37°C for 30 min. After phenol/chloroform extraction and ethanol precipitation, the pellet was redissolved in gel-loading buffer and analyzed by electrophoresis in a 4% polyacrylamide/ 7.5 M urea sequencing gel.

P Sm

I

HH S H E I I I l l

\

2

1

I

MATERIALS AND METHODS

EH I I

\

/ /

15 kb

10

S K

E I

HE I1

Xhol fragment of 57.1 (see Fig. 1). The primer was extended with the Klenow fragment of DNA polymerase I and [ti32P]dCTP. The double-stranded DNA produced was cleaved at the PvuII site (see Fig. 1). The DNA was then denaturated by boiling in alkaline gel-loading buffer and the 32P-labeled single-stranded fragment was isolated from a 5 % polyacrylamide gel. Approximately 2 x lo5 cpm of the 32P-labeled probe were annealed with 40 pg total HeLa cell mRNA or, as a control, with 40 pg yeast tRNA in 20 mM Tris HCI (pH 7.4) 0.4 M NaCI, 1 mM EDTA, 0.1% SDS and 75% deionized formamide (final volume 40 p1) by heating to 75°C and subsequent slow cooling to 55°C. After cooling, 0.35 ml S1 nuclease mix [300 mM NaCl, 3 mM ZnSO,, 60 mM sodium acetate (pH 4 . 9 , 5 pg/ml denaturated salmon sperm DNA and 500 U S1 nuclease] was added. The mixture was then incubated for 1 h a t 37°C. After phenol/chloroform extraction and ethanol precipitation, the pellets were redissolved in gelloading buffer and analyzed on a sequencing gel as described for primer-extension analysis. Isolation of nuclei und detection of DNase-I-hypersensitive regions

Monolayers of HeLa cells grown to 70% confluency were harvested, pelleted by centrifugation, washed in buffer A (0.15 mM spermine, 0.5 mM sperinidine, 15 mM Tris HCI, 60 mM KCI, 15 mM NaCl, 2 mM EDTA, 0.5 mM EGTA and 1 mM phenylmethylsulfonyl fluoride, pH 7.4) and subsequently lysed in buffer A supplemented with 0.5 M sucrose and 0.5% Triton X-100 by 10 strokes with a tight-fitting pistil in a Dounce homogenizer (2 x lo7 cells/ml). The nuclear pellet was recovered by centrifugation at 1000 g , washed in buffer A supplemented with 0.35 M sucrose and again homogenized. After an additional washing step, the nuclei were resuspended in DNase I digestion buffer (0.15 mM spermine. 0.5 mM spermidine, 15 mM Tris HCI, 60 mM KCl. 15 m M NaCI, 0.2 mM EDTA, 0.2 mM EGTA and 1 mM phenylmethylsulfonyl fluoride, pH 7.4). DNase I digestion was performed using 65 520 U/ml DNase I at a concentration of lo8 nucleiiml. The reaction was started by the addition of MgCtz to a final concentration of 5 mM at 4 ° C and terminated after 15 min by addition of EDTA (final concentration 10 mM). Nuclei were collected by centrifugation, lysed by addition of proteinase K buffer (30 mM Tris HCl, 300 mM EDTA, 0.5% SDS,

325 pH 7.5) and incubated overnight with proteinase K (0.5 mg/ ml) at 55 ' C. The DNA was purified, digested with the appropriate restriction enzymes and separated in agarose gels by standard methods. The DNA was blotted to nylon membranes (Amersham) and hybridized with the appropriate 32P-labeled genomic DNA fragments (specific activity 5 x lo8 cpm/pg). The KpnI - Hind111 fragment was used in case of HindIII digested DNA, the EcoRI - SstI fragment in case of the SstIor SstIIHindIII-digested DNA and the SmaI - EcoRI frag-842 -800 -750 -700 -650 -600

-550 -500 -450 -400 -350



A P 2

A P

-250 GGGGCTGGGGTTCCGAGAAAAAGCGTCTGGAGAGGAAGGGGAGGCT-

2 -200

Construction o f p TPCATI and chloramphenicol acetyl transferuse (CAT) activity assay

The 91 8-bp SphI -XhoI fragment upstream of hTOPl exon 1 (see Fig. 1) was subcloned into the appropriate polylinker sites of pBLCAT3 [21] in order to generate pTPCATl. Transfection of cultured human cells was performed by the DEAE-dextran method described by Banerji et al. [22] using 2 pg of each plasmid. The DNA was dissolved GAATTCAGAGCTGGGCACAGAATAGATGCTCTATAAACGGTA in 300 pl buffer B (25 mM Tris HC1, 137 mM NaC1, 5 mM GTTCTATTCCCCATTTGCCCCGTCCTCCCAGCATGCCAGGGATACAGGGG CGGATAGAGACAGGTTCAGTTCCCTTTGGTCTCTTGCATTCCCCCTTCCC KC1,0.7 mM CaCI2, 0.5 mM MgC12 and 0.6 mM Na2HP04, TCAGCACAATTCCTGGGCTCATAGCGGATGCTCAATAAGTATTTGTAGAA pH 7.4), DEAE-Dextran (300 pl, 1 mg/ml in buffer B) was DR 16.1 added and mixed well at room temperature. TGAACCAAAGAAATCAGACCCTTGGTGACCCAGGGCTACATGTACTCGGC The cells (semiconfluent; 9-cm dishes) were washed twice DR 16.1 IR 11 with buffer B before the DNA samples were added. The plates GGAAAACTGGGGTTAAACCGGACCCTTGGTTGGTTGAAAAATTAAAATAA were kept at room temperature for 30 min with occasional > > -phageclone containing the first two exons of the hTOPl gene. Genomic DNA sequences in these exons exactly correspond to the published cDNA sequence except for the first nine nucleotides [18].These additional nucleotides present in the hTOPl cDNA were most likely derived from the EcoRI linker added during cDNA library construction and are therefore not part of the hTOPI transcript. As already discussed by D’Arpa et al. [IS], the reading frame coding for hTOPl extends up to the 5’ end of the cDNA clone, and the possibility was considered that the cDNA clone may not contain the translation-initiation codon. However, our analysis of the genomic sequence does not reveal an additional, more 5‘ located ATG codon in the reading frame of hTOPl . Moreover, the transcription-initiation points could be mapped to nearly identical positions by nuclease-S ,-protection assay and primer-extension analysis excluding the possibility that there may be an additional intron upstream of the published cDNA sequence. Thus, the first ATG in the cDNA most likely represents the translation-initiation codon. The minor differences between primer-extension analysis and S1 mapping present in the upstream cluster of transcription-initiation sites may be due to methodological limitations. One possibility may be, that the 5’ ends of the longest RNA . DNA hybrids are vulnerable to S1 nuclease attack because of the relative high A/T content at their extreme 5’ ends. The existence of multiple transcription-initiation sites is a characteristic feature of promoter sequences lacking a TATA box [30]. Recently, Means and Farnham [31] described their analysis of the transcription-initiation site in the dihydrofolate reductase gene, another housekeeping gene without TATA box. They identified a protein, the housekeeping initiator protein 1, that binds to the transcription-initiation site. The con-

329 sensus sequence of this binding site was shown to be composed region, which is a typical feature of housekeeping gene regulaof two sequence elements (AATTTC and GCCA), separated tory elements and therefore may serve as a transcriptionby 1 - 19 nucleotides. However, no such sequences were found factor-binding site. Recently, we have described two pseudogenes correspondin vicinity of the hTOP1 transcription-initiation sites. The hTOPl promoter shares some features with other ing to the 3' part of the hTOPl gene. These truncated prohousekeeping gene promoters. The sequence of the 500 cesssed pseudogenes are present on human chromosomes 1 nucleotides upstream of the transcription-initiation sites is and 22 [S]. In a detailed analysis [39], we found promoter-like extremely G/C rich (67%) and contains a high frequency of sequences at the 5' side of one of these truncated pseudogenes potential methylation sites (46 CpG pairs). The CpG/GpC (Y1 TOPI). These promoter like sequences, however, do not ratio, in the region 200 bp upstream of the transcription- show any similarity to the promoter sequence of the functional initiation sites is nearly 1, compared to 0.2-0.3 in bulk DNA hTOPI gene described here. [25]. Within this region, we note two GGCGGG sequence We thank Dr. B. Horsthemke for providing the genomic library, motifs, potential core elements of transcription-factor-SplDr. B. Luckow for the gift of the pBLCAT2 and pBLCAT3 constructs binding sites. 50 nucleotides upstream of the transcription-initiation site and Dr. W. Earnshaw for providing the cDNA sequence of hTOP1 nearest the 5' end we identified a CAMP-responsive element prior to publication. We are grateful to R. Mettke for excellent technical assistance and to B. Baumgartner for help with the transfection (TGACGTCG). This sequence is known to bind the cellular of eukaryotic cells. The work was supported by the Deutsche ForCAMP-responsive-element-binding protein which is involved schungsgemeinschaft through SFB 156 and a grant to N. K. in mediating the cellular response to CAMP(reviewed in [27]). The same sequence element also appears to be involved in the transcriptional activation of genes by adenoprotein E l A [26, REFERENCES 32, 331. Indeed, Romig and Richter [14] recently reported a 1. Wang, J . C. (1985) Annu. Rev. Biochem. 54, 665-697. 3 - 5-fold increase in hTOP1 mRNA levels in HeLa cells dur2. Vosberg, H.-P. (1985) Curr. Top. Microbiol. Imrnunol. 114, 19ing the early phase of infection with adenovirus 5. This in102. crease was dependent on the expression of an early adenovirus 3. Wang, J. C. (1987) Biochim. Biophys. Acta 909, 1-9. protein product encoded by the E1A region. This observation 4. Juan, C. C., Hwang, J., Lia, A. A., Wang-Peng, J., Knutsen, T., is consistent with the relatively higher level of hTOPl proHuebner, K., Croce, C. M., Zhang, H., Wang, H. C. & Liu, F. moter activity observed in 293 cells, a cell line constitutively L. (1988) Proc. Natl Acad. Sci. USA 85, 8910-8913. expressing the adenovirus E1A and E1B proteins [34]. 5. Kunze, N., Yang, G. C., Jiang, Z. Y., Hameister, H., Adolph, S., Wiedorn, K.-H., Richter, A. & Knippers, R. (1989) Hum. Genet. As recently shown by Hwong et al. [15], hTOPl mRNA 84, 6-10. levels in human skin fibroblasts are transiently elevated upon 6. Saavedra, R. A. (1990) Bio Essays 12, 125-128. exposure to PMA. This stimulatory effect of PMA seems to 7 . Richter, A. & Knippers, R. (1989) Life Sci.Adv. 8, 125- 134. be the result of increased transcriptional activity. Induction 8. Stewart, A. F., Herrera, R. E. & Nordheim, A. (1990) Cell 60, of gene transcription by PMA is probably mediated by several 141- 149. transcription factors (for a review see [35]). One of these is 9. Brill, S. J., DiNardo, S., Voelkel-Meiman, K. & Sternglanz, R. activator protein 1 which recognizes a consensus sequence (1987) Nature 326,414-416. (TGASTCA) very similar to the CAMP-responsive element, 10. Camilloni, G., DiMartino, E., DiMauro. E. & Caserta, M. (1989) Proc. Natl Acad. Sci. USA 86, 3080- 3084. which is present in the hTOP1 promoter region. Another phorbol-ester-responsive element, present in the metallo- 11. Heck, M. M. S., Hittelman, W. N. & Earnshaw, W. C. (1988) Proc. Natl Acad. Sci. USA 85, 1086- 1090. thionein IIA promoter, is recognized by transcription factor AP2. The hTOPl promoter region also contains sequence 12. Romig, H. & Richter, A. (1990) Biochinz. Biophys. Acta 1048, 274 - 280. motifs showing some similarity to AP2 consensus sites. 13. Duguet, M., Lavenot, C., Harper, F., Mirambeau, G. & Like several other promoters of constitutively expressed DeRecondo, A.-M. (1983) Nucleic Acids Res.11, 1059-1075. proteins, the hTOPl promoter also contains a perfect consen- 14. Romig, H. &Richter, A. (1990) Nucleic Acids Res. 18, 801 -808. sus sequence of the Octl-binding sites. Fletcher et al. [36] and 15. Hwong, C.-L., Chen, M.-S. & Hwang J. (1989) J . Bid. Chem. LaBella et al. [37] have demonstrated that the octamer motif 264, 14 923 - 14 926. and the binding of Octl are necessary for cell-cycle-regulated 16. Tan, K. B., Mattern, M. R., Eng, W-K., McCabe, F. L. & Johnson, R. K. (1989) J . Natl Cancer Inst. 81, 1732- 1735. expression of histone H2B and, recently, a 3 -4-fold increase 17. Maniatis, T., Fritsch, E. F. & Sambrook, J. (1982) Molecular strated [12]. cloning, a laboratory manual, Cold Spring Harbor Laboratory, Further work must demonstrate which of the putative Cold Spring Harbor, NY. transcription-factor-binding sites are actually involved in 38. D'Arpa, P., Machlin, P. S., Ratrie, I11 H., Rothfield, N. F., mediating the various effects on transcriptional regulation of Cleveland, D. W. & Earnshaw, W. C. (1988) Proc. Natl Acad. the hTOPl gene described above. Sci. USA 85,2543 - 2547. A correlation between the location of transcriptional regu- 19. Sanger, F., Nicklen, S. & Coulsen, A. R. (1977) Proc. Natl Acad. latory elements and the positions of DNase-I-hypersensitive Sci. USA 74, 5463 - 5467. regions has been documented for a number of genes (for 20. Chirgwin, J. M., Przybyla, A. E., MacDonnald, R. J. & Rutter, W. J. (1979) Biochemistry 18, 5294-5299. review see [38]). This also seems to be true for the hTOPl gene, 21. Luckow, B. & Schiitz, G. (1987) Nucleic Acids Res. 15, 5490. where we located the most prominent DNase-I-hypersensitive region at about 50 bp upstream of the 5' transcription-in- 22. Banerij, J., Olson, L. & Schaffner, W. (1983) Cell33, 729-739. 23. Gorman, C. M., Moffat, L. F. & Howard, B. H. (1982) Mol. Cell. itiation site. This site in the promoter region thus coincides Biol.2, 1044-1051. with transcription-factor-binding sites. Whether or not the 24. Russel, G. J., Walker, P. M. B., Elton, R. A. & Suhak-Sharpe, J. additional DNase-I-hypersensitive regions present in intron 2 H. (1976) J . Mol. Biol. 108, 1-23, are correlated with potential enhancer elements remains to be 25. Schreiber, E., Muller, M., Schaffner, W. & Matthias, P. (1989) shown. At least the first of the hypersensitive regions in intron Tissue specfie gene expression (Renkawitz, R., ed.) pp. 33 - 54, VCH publishers, Weinheim. 2, immediately adjacent to exon 2, is located in a G/C-rich

330 26. Rocsler, W. J., Vandenbark, G . R. & Hanson, R . W. (1988) J . Biol. Clicvn. 263, 9063 -9066. 27. Kadonaga, J. T., Jones, K. A. & Tjian, R. (1986) Trcwds Biochem. Sci.11, 20-23. 28. Maniatis, T., Goodbourn, S. & Fischer, J. A. (1987) Science 236, 1237-1245. 29. Wu. C . (1980) Nature 286, 854-860. 30. Dynan, W. S. (1986) Trends Genet. 2, 196-197. 31. Means, A . L. & Farnham, P. J. (1990) Mol. Cell. Bid. 10, 653661. 32. Lin, Y.-S. & Grccn. M . R. (1988) Proc. Nut1 Acad. Sci. U S A 85, 3396 - 3400.

33. Hardy, S. & Shenk, T. (1988) Proc. Natl Acnd Sci. U S A 85, 4171 -4175. 34. Graham. F. L., Smiley, J.. Russell, W. C. & Nairn, R. (1977) J . Gen. Virol. 36, 59 -72. 35. Karin, M. (1985) Trends Genet. 5, 65-67. 36. Fletcher, C., Heintz, N. & Roeder. R. G. (1987) Cell 51, 773781. 37. LaBella, F., Sive, H. L.. Roeder, R. C . & Heintz, N. (1988) Genes & Dev. 2, 32 - 39. 38. Gross, D. S. & Garrard, W. T. (1988) Annu. Rev. Biochem. 57, 159-197. 39. Yang, G., Kunze, N., Baumgartner. B., Jiang, Z. Y., Sapp, M. Knippers, R. & Richter, A. (1990) Gene 91, (Amst.) 247-253.