Reactivation of the methylation-inactivated late E2A promoter of adenovirus type 2 by E1A (13 S) functions

J. Mol. Biol. (1988) 202, 255-270

Reactivation of the Methylation-inactivated Late E2A Promoter of Adenovirus Type 2 by ElA (13 S) Functions Bernd Weisshaar, Klaus-Dieter Langner, Ruth Jiittermann, Ulrich Miiller Christiane Zock, Thomas Klimkait and Walter Doerfler Institute of Genetics University of Cologne Cologne, West Germany (Received 13 July 1987, and in revised form 18 February 1988) The inactivating effect of sequence-specific promoter methylations was extensively studied by using the late E2A promoter of adenovirus type 2 (Ad2) DNA. The modification of the three 5’ CCGG 3’ sequences at nucleotides +24, +6 and -215, relative to the cap site in this promoter, sufficed to silence the gene in transient expression either in Xenopus laevis oocytes or in mammalian cells, and after the fixation of the E2A promoterchloramphenicol-acetyltransferase (CAT) gene construct in the genome of hamster cells. It will now be demonstrated that the inactivation of the late promoter of Ad2 DNA can be reversed by transactivating functions that are encoded in the 13 S messenger RNA of the EIA region of Ad2 DNA. The reactivation of a methylation-inactivated eukaryotic promoter by transactivating functions has general significance in that the value of a regulatory signal can be fully realized only by its controlled reversibility. It was demonstrated in transient expression experiments that the 5’ CCGG 3’-methylated late E2A promoter was at least partly reactivated in cell lines constitutively expressing the El region of Ad2 or of adenovirus type 5 (Ad5) DNA. The reactivation led to transcriptional initiation at the authentic cap sites of the late E2A promoter and was not associated with promoter demethylation, at least not in both DNA complements. Reactivation of the methylation-inactivated E2A promoter could also be demontrated in two BHK21 cell lines (me14 and mc20), which carried the late E2A promoter-CAT gene assembly in an integrated form. In these cell lines the late E2A promoter was methylated and the CAT gene was not expressed. By transfection of cell lines mc14 and mc20, the reactivating functions were shown to reside in the pAd2ElA-13 S cDNA clone of Ad2 DNA. The pAd2ElA-12 S cDNA clone or the pAd2ElB clone showed no reactivating function. These findings implicated the ElB 289 amino acid residue protein of Ad2, a well-known transactivator, as the reactivating function of the endogenous, previously dormant, late E2A promoter-CAT gene assemblv. The methylated promoter was not demethylated, at least not in both complements, and”it was shown that reactivation of the methylated promoter entailed transcriptional initiation at the authentic late E2A cap site. Since ElA and ElB jointly had a more pronounced effect, it was conceivable that genes in both regions acted together in the abrogation of the inhibitory effect of promoter methylations in the late E2A promoter.

short-term switch-off of genes. In this situation other regulatory principles seem to play a decisive role. At the core of promoter-controlling functions lie interactions of specific proteins with specific nucleotide sequences in the promoter and upstream regions of eukaryotic genes (for reviews, see Dynan & Tjian, 1985; McKnight & Tjian, 1986). It appears reasonable to assume that in the inactivation of genes by promoter methylations, the methyl group might attain its signal value either by preventing

1. Introduction In the inactivation of viral and non-viral eukaryotio genes, sequence-specific promoter methylations have been shown to play an important role (for surveys, see Razin & Riggs, 1980; Sutter & Doerfler, 19’79, 1980; Doerfler, 1981, 1983, 1984; Doerfler et al., 1985; Murray & Grosveld, 1985). It is less likely that promoter methylation is used as a regulatory signal in the 0022-2836/88/14025<~16

$03.00/O

255

0

1988 Academic

Press Limited

256

B. Weisshaar

the binding of regulatory factors or by the recognition of specific proteins. Proteins that bind more avidly to methylated sequences have been isolated from human placenta (Huang et al., 1984; Zhang et al., 1986); however, their function has not been elucidated. In previous studies, we reported on the inactivation of the late E2A promoter, when this promoter was methylated in, vitro at its three 5’ CCGG; 3’ sequences, i.e. at nucleotides +24, +6 and -215 relat,ive to the cap site of the promoter. This inactivation was documented by testing its function in the natural sequence context after microinjection into nuclei of Xenopus laevis oocytes (Langner et al., 1984). Tn other experiments, the El A promoter of adenovirus type 12 (Ad1 2t) DNA or the late E2A promoter of Ad2 DNA was linked to the prokaryotic gene for chloramphenicol acetyltransferase (CAT) as a reporter gene, and its inactivation by methylation was shown after transfection into mammalian cells (Kruczek & Doerfler, 1983; Langner et al., 1986). In the late E2A promoter of Ad2 DNA, at least six different sequences have been identified that bind specifically to host factors. Striking differences in host factor binding at these sites between the non-methylated and t,he 5’ CCGG 3’-methylated late E2A promoter have not been observed (Hoeveler & Doerfler, 1987). signal, which is utilized in the A regulatory

inactivation of a promoter, should be reversible. It is difficult to visualize the instantaneous removal of methyl groups from an inactivated promoter, although mechanisms of demethylation have been proposed (Gjerset & Martin, 1982; Razin et al., 1986). From earlier work on the activity of the 5’ CCGG 3’-methylated late E2A promoter after transfection into mammalian cells, which constitutively ,expressed El functions of adenoviruses, the possibility arose that this unexpected activity was caused by the El transactivation of the methylation-inactivated late E2A promoter (Langner et aZ., 1986).

We have now investigated further how a fully methylated promoter could be, at least transiently, reactivated by transactivating functions. A transactivator par excellence is the 289 amino acid residue gene product of the adenovirus EIA region (Herk et al., 1979; Jones & Shenk, 1979; Nevins, 1982; Kovesdi et al., 1987: Reichel et aZ.; 1987). In

this paper, we demonstrate

that the 5’ CCGG 3’.

methylated late promoter of the Ad2 E2A region is reactivated in cells that constitutively express adenovirus E 1 functions. Transcription is initiated at the authentic cap sites of the late E2A promoter. This reactivation is not caused by demethylation of the transfected DNA in both strands. The reactivation of the methylated late E2A promoter can be

7 Abbreviations used: Ad12 etc., adenovirus type 12 etc: CAT, chloramphenicol acetyltransferase; CAM, chloramphenicol: bp, base-pairs; kb, lo3 bp.

et al.

elicited in cells that express the ElA genes and, even better, in cells that express the ElA and ElB genes, but not in cells that produce exclusively ElB products. Moreover, it will be shown that in and BHK21 cell lines, which carry the methylated, thus inactive, late E2A promoter in front of the CAT gene, the endogenous late E2A promoter can be reactivated by transfecting the left terminal 7.78% (Hind111 G fragment) of the Ad2 genome into these cells, or by a cDNA of the 13 S but not of the 12 S ElA mRNA of Ad2 DNA (Leff et al., 1984). Hence, even a genomically fixed, dormant and methylated viral promoter can be reactivated by the 289 amino acid residue protein that is encoded by the 13 S mRNA in the ElA region of Ad2 DNA.

2. Materials and Methods (a) Cell lines and plasmid constructs

All cell lines were propagated in Dulbecco’s medium supplemented with 10% fetal calf serum. The following cell lines were used in this study: 293 cells, an Ad5 transformed human cell line (Graham et al., 1977; Aiello et al., 1979), BHK21 hamster cells (ATCC CCLIO), the Ad2transformed hamster cell line HE7 (Cook & Lewis, 1979; Klimkait & Doerfler, 1985), the Ad5transformed hamster cell line BHK297-Cl31 (Visser et al., 1979), the ElA expressing cell line BHK21-AdBHindIII G, the El B expressing line BHK21-Ad2ElB, the EIA and ElB expressing line BHK21-Ad2ElA-B, the BHK21 cell lines mc14 and mc20, which carried the 5’ CCGG 3’-methylated late E2A promoter-CAT gene assembly, and cell line uc2. which contained the same assembly in the nonmethylated form (Miiller & Doerfler, 1987). Cell line BHK21-Ad2HindIII G was generated by transfecting BHK21 cells with the p78Ad2-neo plasmid (see Fig. 5(a), below). Subsequently, G418 (analog of neomycin)-resistant colonies were selected (Southern & Berg, 1982) by adding .to the medium 800 pg of commercial G418 (Pharmacia) per ml. Among the resistant colonies, those were selected that contained the Hind111 G fragment of Ad2 DNA in an integrated form. Similarly, BHK21 cells were cotransfected with pSV2-neo DNA and the HpaI (4.3%) to Hind111 (17.1%) fragment of Ad2 DNA, which was cloned in pBR322 DNA. G418resistant colonies were selected, as described (Southern & Berg, 1982). The plasmid preparations pBR322, pAd2E2AL-CAT (Langner et al., 1986), which contained the late E2A promoter of Ad2 DNA in the Hind111 site of the vector pSVO-CAT (Gorman et al., 1982), pSVP-CAT, which contained the early SV40 promoter in the HindIII site of pSVO-CAT, pAd2HindIII G, which carried the left terminal 7,78% of Ad2 DNA (the El A and the left part of the ElB region) in the HindIIT site of pBR322 (provided by Thomas Rosahl), and pAd2XacI H (con taining the left terminal 1771 nucleotides of Ad2 DNA in the pUC18 vector) were propagated and purified by standard procedures (Clewell & Helinski, 1972). The plasmid pAd2ElB, which contained the HpaI (4.3% of the Ad2 genome) to Hind111 (17.3% of the Ad2 genome) fragment of Ad2 DNA was constructed as described in the legend to Fig. 5(b), below. The cDNA clones of the 12 S (pAd2ElA-12 S) or of the 13 S (pAd2ElA-13 S) ElA

El A Reactivation

of Methylation-inactivated

mRNA of Ad2 DXA have been described (Leff et al., 1984). These cDX.4 clones were kind gifts from Claude KBdinger. Strasbourg, France.

Promoter

RNAs that were isolated from the cytoplasm of cell line BHK21 -Ad2HindITI G or BHK2 1-Ad2E 1B and were selected for polyadenylated sequences by standard methods (Aviv & Leder, 1972).

(b) In viqro methylation of the pAd2E2AL-CAT plasmid DNA The HyaIT DEA methyltransferase (Biolabs, Beverly, MA) was used as described (Vardimon et al., 1982; Kruczek & Doerfler, 1983). Complete methylation of the pAd2E2AL-CAT construct was ascertained by cleavage with HpaII or MspI (Waalwijk & Flavell, 1978) and Southern blot hybridization analyses. as detailed earlier (Vardimon et al.. 1982) (Fig. 3). (~2)Plasmid transfections and cotransfections The calcium phosphate precipitation method was used (Graham & van der Eb, 1973). Per 6 cm diameter dish, 15 pg of the non-methylated or the 5’ CCGG 3’methylated pAd2E2AL-CAT construct DEA were transfected in a mixture with 5pg of pBR322 DNA or pAd2HindIIT G plasmid DNA. In other experiments, 10 pg or 20 fig of the non-methylated or the methylated pAd2E2AL-CAT DNA, or of non-methylated pBR322 DPU’A, of pAd2HindIIIG DKA, or of pAdBElAS cDXA or pAd2ElA-13 6 cDNA were used. At 4 h after the addition of DNA to the cells, the cultures were treated with 15 y0 glycerol in Hepes-buffered saline for 2 min. Subsequently, the cells were washed, then fresh medium was added, and incubation at 37°C was continued for 44 to 48 h. (d) Assay for chloramphenicol acetyltransferase activity in extracts of transfected cells Published methods (Gorman et al., 1982; Gorman, 1985) were used with minor modifications as described (Kruczek & Doerfler, 1983). 14C-labeled chloramphenicol (CAM) and its acetylation products were separated chromatographically. The labeled compounds were excised from the thin-layer plate, and the relative amounts of acetylated forms of CAM were measured by scintillation counting in order to assess CAT gene activity quantitatively. (e) Isolation of cytoplasmic RNA from transfected cells and S, nuclease protection analyses RNA was isolated from 293 cells transfected with the unmethylated or the methylated pAd2E2AL-CAT construct by standard procedures (Scott et al., 1983). Similarly, cell lines mc14 and mc20 were transfected with pAd2ElA-12 S or pAd2ElA-13 S or plasmids pAd2HindIII G or pBR322, and cytoplasmic RPU’A was isolated 48 h after transfection. S1 nuclease (Vogt, 1973) protection analyses using specific subfragments of the late E2A promoter construct and RNA from transfected cells were carried out as described (Berk & Sharp, 1977; Liibbert & Doerfler, 1984; Langner et al., 1986). Details of these experiments are described in the legends to Figs 2 and 8. (f) Analyses of RNAs by gel electrophoreses and by hybridization to speci$c DNA fragments after Northern transfer Published methods were used (Alwine et al., 1977; Schirm & Doerfler, 1981). These methods were applied to

257

3. Results (a) Reactivation of the 5’ CCGC: 3’.methylated late EZA promoter in Ad& or Adz-transformed cells that constitutively express El functions The non-methylated and the 5’ CCGG 3’-methylated late E2A promoter constructs, which con-

tained

the CAT

(pAd2E2AL-CAT),

constitutively

gene as an activity were

transcribed

indicator

in cells

expressed the El region of Ad2

that or

of

Ad5 DNA, although the methylated promoter construct to a lesser extent (Table 1). In these experiments, the CAT gene could have been under the control of the methylated late E2A promoter or of the previously described El -transactivated prokaryotic promoter-like sequences in the plasmid part of the construct (Langner et al., 1986). In the

Ad5-transformed human cell line 293, the left terminus of the viral genome, which comprises the El A and El B regions of Ad5 DNA, is integrated and constitutively expressed (Graham et al., 1977; Aiello et al., 1979). Cell line 293 was transfected with plasmid constructs pSVO-CAT, pSV2-CAT, or with the construct pAd2E2AL-CAT in the nonmethylated or in the 5’ CCGG 3’-methylated form. The CAT activity expressed under the control of different

promoters

was

taken

as a measure

of

promoter activities. The data (Table 1) demonstrated high activity levels for the pSV2-CAT construct and activity also for the pSVO-CAT construct. The activity of the pSVO-CAT construct was due to the function of an El-transactivated, plasmid-located surrogate promoter element, as described (Langner et al., 1986). The pAd2E2ALCAT construct was also active in 293 cells. Upon methylation, its activity was reduced by about 50% of the control value. This partial rea.ctivation of the 5’ CCGG 3’-methylated construct was documented by measuring CAT reaction kinetics (Fig. 1). Tests done at all time intervals during incubation showed that the 5’ CCGG 3’-methylated pAd2E2AL-CAT construct had lower levels of activity than the non-methylated promoter-gene assembly, even at late time points, when both constructs had reached saturation levels. Similar results were observed in transfection experiments using the AdS-transformed hamster cell line HE7, or the Ad5-transformed hamster cell line BHK297-Cl31 (Table 1). In these experiments, the activity of the methylated pAd2E2AL-CAT construct was reduced to about 50% of the activity reached with unmethylated construct. However, activity was not abolished. In contrast, in HeLa or BHK21 cells that were devoid of adenoviral functions, the methylated pAd2E2AL-CAT construct was inactivated (Tables 1 and 2). It has not previously been proven that the El functions of

B. Weisshaar et al.

258

Table 1 Expression of the non-methylated or the 5’ CCGG d’-methylated pAdZEZAL-CAT construct, the pS I/O-CAT and pSV2-CAT plasmids in cells expressing the El region of Ad2 DNA (WI line?

(‘“nstruct

Experiment:

HrLa

pS\‘O-CAT pSV2XAT pAd2EQALXAT pAd2E2AIJ--C4T,

methylated Experiment:

293

pS\‘O-(‘A’l pSVP<‘AT pAdZE2ALA’AT pAdlE2ALCAT,

methylated Experiment:

HE7

pSVO-CAT pSV2-CAT pAd2EIALC’AT pAd2E2ALCAT.

methylated Experiment:

UHK297-C!131

pSVO-CAT pGV2-CAT pAd2E2AIXAT pAdPEtLAL-C”AT,

methylated

O(, (‘onversion 1 x.7 90 8.5 0.9 I 20 90 3I 16 I 28 90 76 48 I 25 8X 71 33

of C’AM to acetylated 2 0, I x7 ,ii.1 2 r 7; 20 *5 2 37 92 74 44 2 22 86 69 35

3 3.9 9x.7 4.X 2-l 3 15 x7 Ii 13 3 41 90 79 34

forms1

4 17 75 27 19 4 3% 89 84 46

5 15 84 40 1X

5’ CCGG Smethylated (HpaII DNA-methyltransferase) construct, or pSV2-CAT DNA or pSVOThe non-methylated or the in-vitro (‘AT DNA were transfected into cell lines as indicated. At 48 h after transfection, extracts were prepared and the activity of chloramphenicol acetyltransferase was determined, as described (Gorman rt al., 1982; German, 1985). Extracts were incubated with ‘4(:-labeled C!AM for 90 min. Reaction products were separated by chromatography on silica gel thin-layer plates and located by autoradiography. Autoradiographically determined spots were excised and the 14C:activity in each spot was measured by scintillation counting. Percentages were calculated as acetylated CAM per total input of CAM. t (:ell line 293 (Graham rt al., 1977; Aiello et al., 1979) contained the left terminus of Ad5 DNA and a small fragment from the right end of the Ad5 genome. Cell line HE7 (Cook & Lewis, 1979; Klimkait & Doerfler, 1985) contained about 40% of the left part of the Ad2 genome, and cell line UHK297-Cl31 (Visser et al., 1979; Klimkait & Doerfler, 1985) l&7:/,. from the left part of the Ad5 genome. Cell lines HE7 and SHK297-Cl31 carried additional segments of the viral genome. All cell lines expressed the El region of Ad2 or Ad5 DNA constitutively. $ The results of several independent experiments are reported.

Ad2 or Ad5 had the capacity to override, at least to some extent, the inactivating effect of sequencespecific late E2A promoter methylations. In some of the experiments presented in Tables 1 and 2, the pSVO-CAT level was elevated. This finding was probably due to the activity of unidentified cellular transactivators recognizing promoter-like sequences in the plasmid part of the construct (Langner et al., 1986). (b) Initiation of transcription at the authentic late E2A cap sites even in the 5’ CCGG 3’-methylated late E2A promoter The results presented in Figure 2 demonstrated directly that in 293 cells transfected with the nonmethylated or the 5’ CCGG 3’-methylated pAd2E2AL-CAT construct, transcription was initiated at the authentic E2A cap sites in the pAd2E2AL-CAT construct, regardless of whether the promoter was methylated or not. In these experiments, the 1726 bp BglI-KpnI fragment encompassing the entire HindIII-flanked late E2A promoter and the adjacent pBR322 DNA sequences up to the BgZI site (map in Fig. 2(b), * DNA probe)

was used for the S, protection analyses of RNAs isolated 48 h after transfection of 293 cells. The same 69/71 bp fragments were protected in either analysis, using RNA from cells transfected with the methylated or the non-methylated construct. These data indicated that the authentic E2A cap sites were used (Fig. 2, lanes 1 and 2). However, as expected, the DNA signals elicited by DNA fragments, which were protected by RNAs from cells transfected with the methylated construct, were markedly weaker than those from DNA fragments protected with RNA from transfection with the non-methylated promoter construct (Fig. 2, lane 2). The protected 69/71 bp DNA fragments corresponded to the authentic cap sites of the E2A late promoter (Baker & Ziff, 1981), which were not seen when RNA from untransfected 293 cells was used in S1 protection experiments (Fig. 2, lane 3). The 69/71 bp signals also arose when cytoplasmic RNA from Ad2-infected human cells was used for protection in control experiments (Langner et aE., 1986; cf. Fig. 8). The possibility existed that part of the CAT activity elicited by the methylated pAd2E2ALCAT construct in 293 cells was due to RNAs

El A Reactivation

of Methylation-inactivated

259

Promoter

293 cells

pAd2E2AL-CAT

(non-methylated)

:;w-;

_

pAd2E2AL-CAT

5 1015

25

(methylated)

40

90 Time (min)

Figure 1. Activity of t’he non-methylated or the 5’ CCGG 3’-methylated pAd2E2ALCAT construct in the Elexpressing cell line 293. Kinetics of the CAT reaction. Monolayer cultures of 293 cells were transfected with 20 pg (per 6 cm diamet’er dish) of the non-methylated or the 5’ CCGG %methylated pAd2E2AL-CAT construct. At 48 h after transfertion. extracts were prepared, and CAT activities were determined by incubating “C-labeled CAM and acetylCoA with the extracts for time intervals as indicated. The percentage of [14C]CAM converted to acetylated forms was determined by counting autoradiographically localized spots for each time point. Since an enzyme test was used to determine promoter activities, values should not be compared at saturation levels, but in the linear parts of the curves. This experiment was performed 4 times with similar results.

Table 2 Expression of the non-methylated and of the 5’ CCGG 3’-methylated pAd2E2AGCAT constructs, the pSVOCAT and pSV2-CAT plasmids in hamster cells constitutively expressing the ElA or ElB region of Ad2 DNA (!ell line

construct

‘+c Conversion of CAM to acetylated Experiment:

IiHKPl

pHVO-CAT pSV2-CAT pAd2E2ALCBT pAd2EPAL-CAT,

methylated Experiment:

I%HK%I-AdIElA-B

pH\TO-C:AT pS\:I-CAT pAd2E2AGCAT pA4d2E2AL-CAT,

methylated

1 5.5 87.1 2.6 0.3 1 84.1 98.2 76.4 37.3

Experiment: UHKYI-AdZHind (EIA)

G pSWWAT pSV2XAT pAd2E2AL-CAT pAd2E2AIXAT,

methylated Experiment:

ISHKZI-AdSElB

pHV&CAT p8V2-CAT pAd2E2ALCAT pAd2E2AI&AT,

methylsted

2.2 59.2 5.9 1.2 1 1.4 85.4 4.8 0.2

2 2.2 37.3 1.5 0.6 2 32.9 68.4 41.8 51.4 2 2.1 60.4 22.8 12.1 2 0.9 76.1 14.3 5.0

forms 3 0.01 67.5 5.1 0.3 3 31% 83.5 59.3 18.2 3 51.6 56.2 7.6 5.3 3 4.0 33.5 3.7 0.8

‘4 70.7 93.8 68.2 53.8

The derivation of the cell lines used in the described transfection experiments is outlined in the text, and the characteristics of some of these cell lines are documented in Fig. 5. The conversion values for ‘%labeled CAM to acetylated CAM were determined by standard procedures (see Materials and Methods) and after a SO-min incubation period of ‘%labeled CAM with individual cell extracts. As will be apparent from the data. each experiment was repeated several times.

260

B. Weisslumr et al.

I 726. Cryptic

fra qments

l/69

(bl ----

__/-

pBR322

EPA-Late

V-CAT

------

+

,8 --# ,........-

-_--_--_--~~~--_---

------------

Fig. 2.

-+

DNAProtectd Cryptic

Probe fragments fragments

(71/69 (-73Of610

ntl nil

ElA

Reactivation of Methylution-inactivated

initiated in prokaryotic sequences in the plasmid part of the construct (Langner et al., 1986). The BglI-KpnI fragment used for S, protection analyses (map in Fig. 2(b)) also included the previously identified (Langner et al., 1986) cryptic prokaryotic sites for transcription initiation (broken line arrows). However, additional protected fragments commensurate in length with these previously mapped surrogate prokaryotic initiation sites were apparent as very weak signals only after prolonged exposure (arrow in Fig. 2(a)). Attempts to eliminate these surrogate prokaryotic promoter sequences by devising alternative vectors did not alleviate this minor problem. It was concluded that El functions of Ad5 or Ad2 DNA could reactivate, directly or indirectly via cellular functions, the 5’ CCGG S’-methylationinactivated late E2A promoter of Ad2 DNA. Transcription from both the methylated and the non-methylated promoter was initiated at the authentic cap sites in the late E2A promoter. (c) The 5’ CCGG 3’-methylated pAd2EZAL-CAT construct was not demethylated in AdS-transformed 293 cells It was conceivable that the 5’ CCGG 3’pAd2E2AL-CAT construct methylated was somehow demethylated after transfection into the Ad&transformed 293 human cells and might therefore have become reactivated. In the present study, 293 cells were transfected with the methylated or the non-methylated pAd2E2AL-CAT construct. At 48 h after transfection, the total nuclear DNA was analyzed by HpaII or lMsp1 cleavage and Southern blot hybridization using the 32P-labeled EcoRI-Hind111 fragment of the late E2A promoter as probe. The data presented in Figure 3 demonstrate that the nuclear pAd2E2ALCAT plasmid DNA remained completely insensitive to HpaII cleavage, when methylated DNA had been transfected, but sensitive to HpaII, when nonmethylated DNA had been transfected. In either experiment, the transfected plasmid DNA was

Promoter

261

sensitive to MspI cleavage. These findings showed directly that the patterns of in-vitro methylation had remained unaltered during the 48 h transfection period (lanes c to j in Fig. 3). As expected, non-transfected 293 cells did not exhibit E2A signals (lanes a and b in Fig. 3). It was concluded that the partial reactivation of the in-vitro methylated pAd2E2AL-CAT construct in 293 cells was not due to active demethylation of the plasmid construct, during the 4%hour period of the experiment. Reactivation was probably caused by genuine transactivation. It could not be rigorously ruled out that the transfected DNA could have been converted to hemimethylated DNA due to selective demethylation in one strand. (d) Transactivation of the 5’ CCGG Zmethylated pAdZEZAL-CAT construct in BHK21 hamster cells cotransfected with the cloned HindlIZ Gfragment of Ad2 DNA thus transcriptionally The methylated, inactivated, late E2A promoter in the pAd2E2ALCAT construct could also be transactivated in BHK21 hamster cells by cotransfection with the pAd2HindIII G clone that contained the left terminal 7.78% of Ad2 DNA, i.e. the ElA and the left part of the El B regions. The CAT gene served as reporter gene in these transient expression assays. After transfection into BHK21 cells, the non-methylated pAd2E2AL-CAT construct elicited the conversion of 8% of the 14C-labeled CAM to acetylated forms in extracts of these cells (Fig. 4). Cotransfection with the pAd2HindIII G plasmid led to a striking increase (32% conversion) in the activity of the non-methylated pAd2E2AL-CAT construct in hamster cells. This stimulation was presumably due to the transactivating functions of El A and/or ElB gene products. As expected from previous results (Langner et al., 1986), the methylated pAd2E2AL-CAT construct was inactivated (0.4% conversion) in hamster cells when cotransfected with the control plasmid pBR322 (Fig. 4). This cotransfection control experiment served to

Figure 2. Mapping of the site of transcription initiation in the non-methylated or in the 5’ CCGG 3’-methylated pAd2E2ALCAT construct by S, nuclease protection. (a) Cells of line 293 were transfected with the non-methylated (lane 1) or the 5’ CCGG 3’-methylated (lane 2) pAd2E2ALCAT construct or were not transfected (lane 3). The 1726 bp KpnI-BgEI fragment (see map in (b)) of the late E2A promoter construct was 5’-labeled ([y-‘*P]ATP. polynucleotide kinase) at the KpnI site and an amount of DNA corresponding to 15,006 cts/min was annealed at 53°C to total cytoplasmic RNA, that had been extracted (Scott et al., 1983) at 48 h after transfection. The annealing buffer contained 80% formamide, 40 mw-Pipes (pH 6.4), 0.4 M-NaCl, 1 mM-EDTA. The hybrids were treated with 150 units of nuclease S, (Pharmacia) under standard conditions (Vogt, 1973) at 20°C for 66 min, and were subsequently heated to 95°C. The diagnostic [32P]DKA fragments were resolved on a 35 cm long 7% polyacrylamide gel in 8 I-Urt?& (Maxam & Gilbert, 1980). The gels were autoradiographed on Kodak XAR-5 film. As size markers (M) the 32P-labeled HpaII fragments of pUC18 DNA were coelectrophoresed. Fragment lengths are indicated in nucleotides (nt). RKAs used for Si protection experiments were derived from the following sources. (1) 293 cells transfected with the non-methylated pAd2E2AL-CAT construct; (2) 293 cells transfected with the 5’ CCGG 3’-methylated construct; (3) non-transfected 293 cells (10 ng each). The position of the 1726 bp BgZI-KpnI fragment was marked. The arrows point to positions where presumptive surrogate initiation sites could have elicited signals. (b) The scheme presented the map of the 1726 BgZI-KpnI DNA fragment used in S, protection experiments. 4, Authentic cap site; C-+, “cryptic” initiation sites; 7, locations of 5’ CCGG 3’ sequences. The pBR322, late E2A promoter and CAT gene parts of the construct are shown. The asterisks indicate the labeled termini of the DNA fragments that were used in protection experiments.

262

B. Weisshaar

h

et al.

ahhough not to the full activity level (320/, conversion), by cotransfection with the pAdSHind G plasmid (Fig. 4). Thus, the met’hylated late E2A promoter, which was inactive in hamster or human cells devoid of adenovirus sequences, could be transactivated by products of the El region of Ad2 DNA. These products could be supplied either by constitutively expressed integrated viral genes, e.g. in cell lines 293, BHK297Cl31 or HE7 (Table 1; Fig. l), or by cotransfection of the El A and part of the El B regions of Ad2 DNA (Fig. 4). The mechanism of either transactivation remained unknown. The data demon-

i

&ate

that

methylation

inhibition

of the late E2A

promoter can be abrogated also by cotransfection with a plasmid carrying the transactivating gene. S, nuclease protection analyses of the RNAs synthesized in cotransfected cells proved difficult, as under these conditions RNA was produced only in exceedingly small amounts. (e) Functions transactivating the smethylationinactivated, late E2A promoter reside in the ElA region qf Ad2 DNA

Figure 3. The state of methylation of the pAd2E2ALCAT construct remained unaltered for 48 h after transfection into 293 cells. As described in the text. 293 cells were transfected with the non-methylated or with the 5’ CCGG 3’-methylated pAd2E2ALCAT construct. At 48 h after transfection, activity levels of CAT were determined in cytoplasmic extracts (cf. Table 1, experiment 5 with 293 cells). At the same time after transfection, the nuclear DNA was extracted and cleaved with MqI (lanes a, c, e, g and i) or HpaII (lanes b, d, f, h and j). The fragments were separated by electrophoresis on a 1.5% agarose gel, blotted (Southern, 1975) and hybridized (Wahl et al., 1979) to the “P-labeled (Feinberg & Vogelstein, 1983) ,%&-Hind111 fragment of the late E2A promoter of Ad2 DNA. The autoradiogram shown was a 4 h exposure. The DPIjA preparations analyzed were derived from the following sources: lanes a and b, non-transfected 293 cells (2pg). as a negative control. These cells did not contain the E2A region of Ad5 DNA and hence exhibited no hybridization signal. Lanes c and d, the 5’ CCGG 3’.methylated pAd2E2ALCAT construct and lanes e and f the nonmethylated construct prior to transfection (1 ng of either plasmid plus 2 pg of salmon sperm DNA). Lanes g and h, DKA from 293 cells transfected with the 5’ CCGG 3’methylated, and lanes i and j DKA from 293 cells transfected with the non-methylated pAd2E2AL-CAT construct (2 pg each). The 3 MqnI fragments correspond to the 3 5’ CCGG 3’ sites in the HindIIT-KpnI E2A fragment. The smallest (18 bp) fragment cannot be seen.

rule out a non-specific stimulation of plasmid expression by cotransfection of DNA other than that of the construct to be tested for activity. In contrast, the methylated late E2A promoter of Ad2 DNA

was

transactivated

(19 o/o

conversion),

As the first step towards the elucidation of this transactivation mechanism, we tried to determine which of the El functions facilitated the reactivation of t’he methylated late E2A promoter. By transfection of BHK21 cells with the p708Ad2-neo plasmid (cf. Fig. 5(a)) or by cotransfection with the pSV2-neo (Southern & Berg, 1982) and the pAd2ElB plasmids (Fig. 5(b)) (see Materials and Methods) followed by G418 selection, BHK21 cell lines were generated that expressed constitutively either the ElA or the ElB region of Ad2 DNA. Some of the characteristics of these cell lines were summarized in Figure 5(a) and (b). It could be demonstrated by the Southern (1975) blot hybridization technique that the selected cell lines, BHKBl-AdSHind G and BHK21-Ad2ElB, contained the DNA fragments used in transfection, the p7.8Ad2-neo fragment and the HpaI (4.3%) to Hind111 (17.3%) fragment of Ad2 DNA, respectively, in an integrated form (Fig. 5(a) and (b), Southern blots, left-hand panels). Moreover, the two cell lines constitutively transcribed the ElA or the ElB region of Ad2 DNA: respectively, as could be demonstrated by RNA transfer and DNA-RNA hybridization experiments (Fig. 5(a) and (b), Northern blots, right-hand panels). The RNA isolated from cell line BHKSl-AdSHind G contained RNA molecules of about 1900, 900 and 500

nucleotides

(13 S,

12 S

and

9 S),

which

hybridized with the Hind111 G fragment of Ad2 DNA (Fig. 5(a), Northern blot, right-hand panel). Thus, this cell line appeared to express preferentially the ElA genes of Ad2 DNA. Similarly, Ad2specific RNAs, which were expressed in the BHK21-Ad2ElB cell line, were characterized by agarose gel electrophoresis, RNA transfer (Fig. 5(b), Northern blot, right-hand panel), and hybridization to the 32P-labeled HpaI (4.3%) to Hind111 (17.3%)

ElA

?lc, acatylated

Reactivation of Methylation-inactivated

263

Promoter

BHK21 cells

CAM

pAd2E2AL-CAT

IfpolK-methylated

pAd2E2AL-CAT

non-methylated

pAd2E2AL-CAT

non-methylated

pAd2EZAL-CAT

f/poll-methylated

+p%R322

tpAd2HindlII

G

I

Figure 4. Transactivation in hamster cells of the methylation-inactivated late E2A promoter of Ad2 DNA by cotransfection with the EIA and part of the ElB regions of Ad2 DNA. BHK21 cells grown in monolayers to half confluency were transfected with 15 pg of 5’ CCGG 3’.methylated or non-methylated pAd2E2AL-CAT DPU’Aplus 5 pg of pBR322 or pAd2HindTIT G plasmid DNA per 6 cm diameter culture dish. At 48 h after transfection. extra,cts of cells CAM to acetylated forms, as indicated, was were prepared and tested for CAT activity. The conversion of 14C-labeled , tnonitored by thin-layer chromatography on silica plates, followed by autoradiography. CAT activity was used as an indicator for the functionality of the late E2A promoter. The results were also quantified by determining the 14C artivity in each spot by scintillation counting. The percentages refer to the fraction of acetylated CAM relative to input CAM. fragment of Ad2 DNA. RNA size classes of about 2300 and 1000 nucleotides (22 S, 13 S) and additional weaker size classes were identified that corresponded to RNAs encoded in the ElB region of Ad2 DNA (Fig. 5(b)). Very similar RNAs could be identified in passage 4 (~4) and p30 of the BHKBl-Ad2ElB

cell

line.

Perhaps,

it

was

sur-

prising that the ElB region of Ad2 DNA was expressed at all in cells devoid of ElA functions. Apparently, cellular functions substituted for ElA functions in the activation of the ElB region. These cellular functions may also be responsible for the low activity levels of the pSVO-CAT construct occasionally observed in HeLa or BHK21 cells (cf. Tables 1 and 2). Tn a similar way, a cell line was generated that carried in an integrated form and constitutively expressed the ElA and the ElB regions of Ad2 DNA. The cell line, termed BHK21-Ad2ElA-B, was produced by cotransfecting the pBR322-cloned Hind111 G +C fragment.s (cf. scheme in Fig. 5(b)) of Ad2 DNA and the pSV2-neo plasmid. This cell line resembled line BHK297-C131, but the results of its detailed analysis will not be described here. The activities of constructs pSVO-CAT, pSV2CAT and pAd2E2AL-CAT in the non-methylated and in the 5’ CCGG 3’-methylated forms were determined in cell lines BHK21, BHKZl-Ad2ElAB, BHK21 -Ad2HindITI G (carrying the integrated construct p7+8Ad2-neo), or BHK21-Ad2ElB (Table 2). It was important to assess the capacities of individual cell lines for the reactivation of the methylation-inactivated late E2A promoter. The data demonstrate that cell lines BHKSl-AdBElA-B and BHK21 -Ad2HindIII G (expressing predominantly the ElA region) were capable of reactivating the methylation-inactivated pAd2E2AL-CAT plasmid. Cell lines BHK21 and BHK21-Ad2ElB lacked this transactivating capability. Thus, an important function of the reactivating capacity towards the 5’ CCGG 3’ methylation-inactivated late E2A promoter resided

in tjhe ElA

region of Ad2 DNA

(Table

2). We could

not rigorously exclude the possibility that cell line BHKBl-Ad2HindIII G expressed part of the ElB region or that its gene product was involved in the reactivation of the methylation-inactivated late E2A promoter. Cell line BHKBl-AdXEl A-B was more effective in transactivation than cell line BHKSl-AdSHind G. Therefore, for optimal transactivation both ElA and ElB functions might be required. (f) The pAdZElA-13 S cDNA mediates the reactivation of the methylation-inactivated late EZA promoter We wished to determine which of the ElA gene products was responsible for the abrogation of the inactivating effect of sequence-specific methylations in the late E2A promoter of Ad2 DNA. The availability of the pAd2ElA-13s and pAd2ElA12 S cDNA clones (Leff et al., 1984) facilitated this investigation. Cell lines mcl4 and mc20, which carry the late E2A promoter-CAT gene assembly in the methylated, hence inactive, form (Miiller & Doerfler, 1987), were transfected with either of these cDNA clones. At 48 hours after t,ransfeetion, cell extracts were prepared and CAT assays performed. The data in Figure 6(a) and (b) demonstrate that transfection with the pAdBElA13 S, but not with the pAd2ElA-12 S, cl>NA clone alleviates the methylation-inhibition of the late E2A promoter in these cell lines. It is also apparent that transfection of the mc20 cell line w&h an ElBcarrying clone (pAd2El B) (either by itself or with a pBR322 plasmid control) does not lead to the reactivation of the methylated late E2A promoter (Fig. 6(a)). The results of a number of control experiments, which are included in Figure 6(a), confirm that transfections with the pAd2tiindIII G (also in Fig. 6(b)) or the pAd2SacI H (the left terminal 1771 Ad2 nucleotides cloned in the vector pUCl8) constructs also reactivate t,he met8hglated

264

B. Weisshaar et al.

late E2A promoter, whereas extracts from untreated cells or from cells transfected with the empty pBR322 vector DNA cannot elicit this effect. Cotransfection of cells with the pAd2ElB and the pAd2SacI H plasmids still permits reactivation of the methylated late E2A promoter. Hence transfect,ion with the pAd2ElB construct does not inhibit t’ransactivation. Analogous control experiments have been carried out using the BHK21 cell line uc2 in which the nonmethylated late E2A promoter-CAT gene assembly has been fixed by integration (Miiller & Doerfler, 1987). In this cell line, the CAT gene is expressed at the basal level (Fig. 6(c); see also Miiller & Doerfler, 1987). As expected for the ElA pAd2ElA-13 S cDNA clone

transactivator, and other

the clones

carrying the ElA region, i.e. clones pAd2HindIII G or pAdSSac H, but not the pAd2ElA-12 S cDNA clone or the pAd2ElB clone, enhance the activity of the integrated late E2A promoter in CAT gene expression (Fig. 6(c)). The data presented in this section demonstrate that the ElA-13 S gene product, i.e. the 289 amino acid transactivating protein, is capable of reactivating the genomically fixed, methylated late E2A

promoter. The ElA-12 S or the ElB functions by themselves cannot reactivate the methylationinhibited late E2A promoter (cf. also data in Table 2). Tn cotransfection

experiments

of HeLa

cells with

the 5’ CCGG 3’-methylated pAd2E2ALCAT construct and the pAd2ElA-13 S or the pAd2ElA-12 S cDNA clone, it was again the pAdSElA-13 S cDNA clone that caused reactivation of the methylationinactivated

late E2A promoter

(data

not shown).

These data show that the 5’ CCGG 3’-methylated late E2A promoter could be reactivated by transfection with the pAdBElAS cDNA clone, even when the late E2A promoter-CAT gene construct was permanently fixed by integration in the BHK21 cell genome. Transactivation, therefore, was not restricted to transiently expressed promoter-gene assemblies. In a similar

described with

way to the reactivation

experiments

cell line 293, a number of control

experiments were also performed with cell lines mc14 and mc20. Firstly, it was shown that in cell

line mc20 the integrated,

methylated

late E2A

promoter was not demethylated reactivation after transfecting

cDNA

construct,

in the course of its the pAd2ElA-13 S as demonstrated by HpaTI or

Figure 5. Characterization of cell lines BHK21-AdZHind G and BHK21-Ad2ElB. (a) The BHKSl-Ad2HindIII G cell line. The p78Ad2-neo construct was produced in the following way: as shown schematically, the pSV2-neo construct (Southern & Berg, 1982) and the plasmid pAdSHind G, which carried the left terminal 7.78% (bp 1 to 2798) of the Ad2 genome (El A and part of ElB regions), were linearized by cleavage with EcoRI. The pSV2-neo DNA was treated with calf intestinal phosphatase, and the two linearized molecules were ligated together. The new construct was termed p78Ad2-neo. BHKLl cells that had grown to confluency in monolayer cultures were split 1 : 10 and were transfected on the following day wit,h 20 pg of the p7.8Ad2-neo construct per 75 cm’ flask using the calcium phosphate precipitation technique (Graham & van der Eb, 1973). At about 30 h after transfection, the cells were subcultured, 1 : 15. Colonies resistant to the aminoglycoside G418 were selected essentially as described (Southern & Berg, 1982) with the selective medium containing 800 pg of G418 per ml of medium. Drug-resistant colonies were grown on 75 cm2 monolayer cultures, and the DNA was extracted for analyses by Southern (1975) blot hybridizations. In some experiments the cytoplasmic RNA of these cells was prepared according to the method of Scott et a2. (1983) with minor modifications (omission of trypsinization). The nuclear DNA was extracted, cut with HindIII, and the fragments were separated by electrophoresis on a 0.8% agarose gel and transferred to a nitrocellulose filter (Southern, 1975). DNA fragments were hybridized to the 32P-labeled, left terminal Sac1 H fragment (1771 bp) of Ad2 DNA (left-hand panel). The marker track in the left-hand panel represented the Hind111 fragments of Ad2 DNA. As will be apparent from the schemes above, the 2.8 kb Hind111 G fragment of Ad2 DNA persisted in this cell line. The polyadenylated RNA from the cytoplasm of cell line BHK21-Ad2HindIII G was prepared, separated by electrophoresis on a 1.0% agarose gel and transferred to a nitrocellulose filter. The following 32P-labeled probes (Rigby et al., 1977) were used to visualize autoradiographically specific RNA sequences: (i) pSV2-neo DNA, (ii) Ad2 DNA and (iii) pBR322 DNA (right-hand panel). A mixture of pBR322 DNA fragments generated by EcoRI, BgZI or TapI, which was alkali-denatured, was coelectrophoresed as size markers. Activities for CAT in extracts of cells as indicated were determined and quantified by published methods (Gorman et al., 1982; Kruczek & Doerfler, 1983). The autoradiogram of a typical silica gel was presented. The quantitative results of this experiment, are included in Table 2 (experiment 2 with cell line BHKZl-AdZHind G). (b) The BHK21-Ad2ElB cell line. The schematic drawing demonstrates, how clone pAd2ElB was constructed. The Hind111 G and Hind111 C fragments of Ad2 DNA had been cloned in pBR322 DNA (Klimkait & Doerfler, 1985). The Hind111 G fragment was excised from the plasmid and reisolated by preparative gel electrophoresis. Of the Hind111 C clone of Ad2 DNA a partial Hind111 cleavage product with a single Hind111 cut was prepared electrophoretically. The ends of this linearized molecule were treated with calf intestinal phosphatase. The thus prepared Hind111 fragments were religated, and a clone containing the authentic Hind111 G and C fragments in the right orientation was selected and isolated. Subsequently, the entire ElA region was excised by cleavage with EcoRI and HpaI. The ElB fragment remaining linked to pBR322 DNA was gel purified, the HpuI and EcoRI ends were filled up with Klenow polymerase (Klenow et aE., 1971) and the DNA was recircularized by ligation. The pAd2ElB and pSV2-neo plasmids were cotransfected into BHK21 cells and G418-resistant colonies were selected as described (Materials and Methods; also see (a)). The integration of the pAd2ElB DNA cell line BHK21-Ad2ElB was demonstrated by Southern blot hybridization using the 32P-labeled Hpal (4.3%) to XhoI (16-1o/0) fragment of Ad2 DNA as probe (left-hand panel). Similarly, Ad2ElBspecific RNAs were identified in the cytoplasmic, polyadenylated RNA from cell line BHK21-Ad2ElB by Northern blot hybridization as described in the text (right-hand panel).

EC0 RI

Ligasa

BHKZt-Ad2Hi/tdm

”

Fig.5.

methylotad

G cells

B. Weissham et al. Left I bl

End of AdZ-Cenome H&h71

Hindill

4.3% Eli!i yHindlU-GyLy----Hindm-C! -y --. Hind=

Hindm

11.3%

7.8%

17.3%

: 3’

I -.._

/ ‘-._

Hin dk

: ‘HindLU

Hindm

fial digestion e lreatrd

kb

A 0.05

0 5.32

J I .32K l.OO-

Fig. 5, contd.

ElA

Reactivation of Methylation-inactivated

Promoter -

267 mc20-

not tronsfected pBR322 pAd2

HindllI

pAd2

Sac1 H

pAd2

EIA-12s

G

pAd2 EIA-13s pAd2

EIB

pAd2 EIB+pBR

322

pAd2 EIB+Ad2

SocI

CAT-Enzyme

98-6

H

control

not tronsfected

- o-3 - o-3 - o-3

Figure 7. In cell line me20 the methylated late E2A promoter is not demethylated in both strands during reactivation. At 48 h after the transfection of me20 cells with the pAd2ElA-13 S cDKA clone, the nuclear DKA was extracted from the cells, and cleaved with HpaII or MspI, as described in the legend to Fig. 3. The fragments were separated by electrophoresis on a 1.5?; agarose gel, transferred to a nitrocellulose filter (Southern, 1975), and the late E2A promoter-specific fragments were visualized by hybridization to the 32P-labeled EcoRIIHindIII late EZA promoter fragment followed by autoradiography. The DKA from non-transfected mc20 cells was treated in the same way as control.

- 34.7 - 37.0

- 2-l

not tronsfected

- 3-2

pBR322

- 63-3

pAd2 HindllI

- 43.4

pAd2

SocI

- 2-2

pAd2

EIA-12s

- 52.9

pAd2

EIA-13s

- 4.2

pAd2

EIB

- 4*2

pAd2

EIB+pBR322

- 45.4

pAd2 EIB+Ad2

G H

SoCI

H

Figure 6. The pAd2ElA-13 S cDNA clone mediates abrogation of methyl&ion inhibition in the late E2A promoter of Ad2 DNA. (a) and (b) The BHK21 cell line mc20 (a) or mc14 (b), which contained in an integrated form of the methylated late E2A promoter-CAT gene construct (Miiller & Doerller, 1987), was transfected with plasmid constructs as indicated. At 48 h after transfertion, extracts were prepared and CAT activity assays were performed and quantified as described (Gorman et al.. 1982; Langner et al., 1984; Gorman, 1985). Percentage values of [14C]CAM converted to acetylated forms are indicated for direct comparison. (c) Similar experiments

MspI cleavage and Southern blot hybridization (Fig. 7). Secondly, S, nuclease protection analyses on cytoplasmic RNA isolated at 48 hours after transfection of mc14 cells with the pAd2ElA-13 S cDNA clone revealed that the authentic late E2A though promoter, methylated, served for transcription initiation, since 75/77-nucleotide DNA fragments were protected by the late E2A promoter-initiated RNA (Fig. 8). The same DNA fragments were protected with RNA from cells that had been transfected with the Hind111 G fragment of Ad2 DNA. In non-transfected mc14 cells or in me14 cells transfected with pBR322 DNA or with the pAd2ElA-12 S cDNA construct in

were performed with cell line uc2, which carried in an integrated form the non-methylated late E2A promoterCAT gene assembly (Miiller & Doerfler, 1987). In all experiments 10 pg of DNA per dish was used.

B. Weisshum et al. I

2

34M567

control experiments, this RNA was not detectable, hence the methylated late E2A promoter was not reactivated (Fig. 8). Thus, the integrated and reactivated E2A promoter was not demethylated, at least not in both complements, and initiation of transcription in the methylated late E2A promoter was effected at the authentic cap sites by transfection of the pAd2ElA-13 S cDNA construct.

M

- start

-

363-

116-

501/4 69 .404 .353

-

242

-

190

-

147

-

II0

-

69

-

67

4. Discussion

7?/75-

pBR322

__f_

E2A-Lot.

_t_

CAT

______X_ -----jc

DNA-

prob.

363nt

ll6nt +

75/77nt

Protbctod frogmmts I

Figure 8. Reinitiation of transcription in the reactivated methylated late E2A promoter by transfection with the ElA 13 6 cDNA clone. At 48 h after transfection of cell line mc14 with plasmid preparations as indicated below, cytoplasmic RNA was prepared and used in S1 protection experiments with the HindIIIPvuII E2A promoter fragment as shown in the map of the late E2A promoter-CAT

construct

in (b). Reaction

conditions were the same as for Fig. 2. Der&tions

of

cytoplasmic RNA preparations: lane 1, KB cells, 12 h after the infection with Ad2 (positive control); lane 2, transfer RNA (negative control); lane 3, cell line mc14 transfected with pAd2HindIII (20 pg); lane 4, with the pAd2ElA-13 S cDNA clone (2Opg); lane 5, with the pAd2ElA-12 S cDNA clone (20 pg); lane 6, with pBR322 DNA (20 pg); lane 7, mc14 cells that were not transfected. The marker lanes (M) carried HpaII-cut pUC18 DNA that was 32P-labeled with polynucleotide kinase and denatured. The sizes of the marker DNA fragments are indicated in nucleotides (nt). (a) Autoradiogram of protected 32P-labeled DNA fragments that were electro-

phoretically separated on a 7% polyacrylamide gel containing 8 M-urea. (b) Map of the late E2A promoterCAT construct. The HindIII-PvuTI DNA fragment used in S, protection experiments was 32P-labeled by polynucleotide kinase at the terminus designated by *.

The signal of sequence-specific methyl&ion can cause the long-term inactivation of eukaryotic promoters, unless mechanisms for transient reactivation existed and conditionally reversed the function of this signal. Such mechanisms have now been recognized in the transactivating functions residing in the El (notably in the EIA) region of Ad2 or of Ad5 DNA. The results of transfection experiments indicate that the reactivation function towards the methylation-inhibited late E2A promoter resides in the 13 S but not in the 12 S cDNA clone of ElA-derived messenger RNAs. Thus the transactivating 289 amino acid residue protein in the ElA region of Ad2 DNA is implicated also as reactivator of a methylated promoter. We have shown (Langner et al., 1986), that, in transient expression experiments, the 5’ CCGG 3’-methylated late E2A promoter of Ad2 DNA, which controls the chloramphenicol acetyltransferase gene as an activity indicator, is at least partly reactivated in human or hamster cells that constitutively express the El region of Ad2 or Ad5 DNA. In a similar line of evidence, it has been demonstrated that the ElA promoter of Ad12 DNA, which can also be inactivated by 5’ CCGG 3’ methylation at two HpaII sites (Kruczek & Doerfler, 1983), can be transactivated in BHK21 hamster cells by proteins encoded in the genome of the iridovirus FV3 (frog virus 3; Thompson et al., 1986). The existence of these viral transactivators suggests that similar cellular functions might also occur. The finding of the reactivation of a methylation-inactivated promoter has general significance and might, at least in part, account for occasional reports (GerberHuber et al., 1983) that even methylated promoters exhibit partial activity. At the present time, it is not understood how transactivating proteins are regulated, particularly in their reactivating funcsequence-specifically methylated tion towards

promoters.

There

is

evidence

that

the

ElA

transactivator of adenoviruses might exert its effects via cellular functions (Kovesdi et al., 1987; Reichel et al., 1987). In many of our previous studies on the geneinactivating function of sequence-specific promoter methylations, we have utilized the late promoter of

the E2A region of Ad2 DNA (Vardimon et al., 1980, 1982; Langner et al., 1984; 1986). Upon the in-vitro methylation of the three 5’ CCGG 3’ sequences in the late E2A promoter, significant differences between proteins binding specifically at six different sites have not been detected to date (Hoeveler &

El A Reactivation of Methyl&on-inactivated Doerfler, 1987). The failure to distinguish differences in the binding of proteins between the nonmethylated and the methylated late E2A promoter of Ad2 DNA do not preclude the possibility that the functionality of promoter-bound proteins is seriously interfered with by the presence of methyl groups at specific sites. In recent experiments using synthetic oligodeoxynucleotide fragments from the E2AL promoter, differences in protein binding between the non-methylated and the 5’ CCGG 3’oligodeoxynucleotides have been methylated observed (R. Hermann, A. Hoeveler t W. Doerfler, unpublished results). In fact, in a recent report evidence has been presented that methylation of a 5’ GCGC 3’ (HhaI) sequence in the E2F binding site of the El enhancer affects E2F binding and also impairs the function of this promoter element (Kovesdi et al., 1987). Transactivation of the methylation-inactivated late E2A promoter has been documented both in transient activity tests with transfected late E2A promoter constructs (Tables 1, 2; Figs 1, 2 and 4) and with the late E2A promoter as part of an endogenously fixed promoter gene assembly that was integrated in the host genome (Figs 6 to 8). The transactivating El functions can be provided by cells that are constitutively expressing the El region of adenoviruses (Tables 1 and 2; Figs 1 and 2) or by El functions that have been introduced into cells by transfecting the cloned ElA gene or the cloned pAdSElA-13 S cDNA (Figs 4 and 6). The cloned pAd2ElA-12 S cDNA of Ad2 DNA or the pAd2ElB clone cannot function as reactivator. The endogenously expressed El functions in 293, HE7, BHK297-Cl31 or BHK21-Ad2ElA-B cells also suffice to transactivate the methylationinactivated late E2A promoter that has been introduced by transfection (Tables 1 and 2). Similarly, cell lines expressing predominantly the ElA region (BHK21 -Ad2HindIII G) are capable of methylationtransactivating the transfected inactivated late E2A promoter, whereas cell line BHKSl-Ad2ElB is not (Table 2). As directly demonstrated in cell line 293 (Fig. 2) and in cell line mc14 (Fig. 8), transcription in the reactivated methylated late E2A promoter is initiated at the authentic cap sites. The mechanisms associated with the reactivation of methylation-inactivated promoters are not understood. Apparently, demethylation of both DNA strands is not the cause of transactivation, although the selective removal of methyl groups from one DNA strand, though unlikely, cannot rigorously be ruled out. More detailed investigations into the mechanism(s) of reactivation of methylation-inactivated promoters might provide clues to the ways in which the methylation of a promoter can lead to its inactivation. The finding that the transactivation of the 5’ CCGG 3’-methylated pAd2E2ALCAT construct leads to the utilization of the authentic late E2A cap sites and does not reactivate surrogate pBR322 promoters deserves to be commented upon. In the

269

Promoter

pSVO-CAT construct, the El functions of adenoviruses activate such promoter-like sequences (Langner et al., 1986). Once the Hind111 site in the pSVO-CAT construct is occupied by the late E2A promoter, that promoter is activated and surrogate plasmid sequences are no longer efficiently utilized, perhaps because a more attractive eukaryotic promoter has now become available. It should also be mentioned that the late E2A promoter silences surrogate plasmid sequences when this promoter is inserted at the Hind111 site of the pSVO-CAT construct in the reverse orientation. We thank C. Kedinger, Strasbourg, for providing the cDNA clones of the 12 S and 13 S El A mRNAs of Ad2 DNA. We are indebted to Petra BGhm for excellent editorial work and to Hanna Mansi-Wothke for the preparation of media. This research was supported by the Deutsche Forschungsgemeinschaft through SFB74-Cl and by donations of the Hoechst Company, Frankfurt, Germany and the Fonds der Chemischen Industrie.

References Aiello, L., Guilfoyle, R., Huebner, K. & Weinmann, R. (1979). Virology, 94, 46Cb469. Alwine, J. C., Kemp, D. J. 8t Stark, G. R. (1977). Proc. Nat. Acad. Sci., U.S.A. 74, 5350-5354. Aviv, H. & Leder, P. (1972). Proc. Nat. Acud. Sci., 7J.S.A. 69, 14081412. Baker, C. C. & Ziff, E. B. (1981). J. Mol. Riol. 149, 189221. Berk, A. J. & Sharp, P. A. (1977). Cell, 12, 721-732. Berk, A. J., Lee, F., Harrison, T., Williams, J. & Sharp, P. A. (1979). Cell, 17, 935944.

Clewell, D. B. & Helinski, D. R. (1972). J. Bacterial.

110,

1135-l 146. Cook, J. L. & Lewis, A. M., Jr (1979). Cancer Res. 39, 1455-1461. Doerfler. W. (1981). J. Gen. Viral. 57, l-20. Doerfler, W. (1983). Annu. Rev. Biochem. 52, 93-124. Doerfler, W. (1984). Advan. Viral Oncol. 4, 217-247. Doerfler, W., Langner, K.-D., Knebel. D., Weyer, U., Dobnanski, P. & Knust-Kron, B. (1985). In Biochemistry

and

Biology

of

DNA

Methylation

(Cantoni, G. L. & Razin, A., eds), pp. 133-155, Alan R. Liss, New York. Dynan, W. 8. & Tjian, R. (1985). Nature (London), 316, 774-778. Feinberg, A. P. & Vogel&in, B. (1983). Anal. Biochem. 132, e-13. Gerber-Huber, S., May, F. E. B., Westley, B. R., Felber, B. K., Hosbach, H. A., Andres, A. C. & Ryffel, G. U. (1983). CeZE,33, 43-51. Gjerset, R. A. & Martin, D. W., Jr (1982). J. Biol. Chem. 257. 8581-8583. Gorman, C. (1985). In DNA Cloning (Glover, D. M., ed.), vol. 2, pp. 143-190, IRL Press Oxford and Washington. Gorman, C. M., Moffat, L. F. & Howard, B. H. (1982). Mol. Cell. Biol. 2, 1044-1051. Graham, F. L. & van der Eb, A. J. (1973). Virology, 54, 536-539.

Graham, F. L., Smiley, J., Russell, W. C. & Nairn, R. (1977). J. Gen. Viral. 36, 59-72. Hoeveler, A. & Doerfler, W. (1987). DNA, 6, 449460. Huang, L.-H., Wang, R., Gama-Sosa, M. A., Shenoy, S. & Ehrlich, M. (1984). Nature (London), 308. 293-295.

B. Weisshaar Jones. N. & Shenk. T. (1979). Cell, 17, 683-689. Klenow. H.. Overgaard-Hansen, K. & Patkar. S. A. (1971). Eur. J. Biochem. 22, 371-381. Klimkait. T. & Doerfler, W. (1985). J. Viral. 55, 466-474. Kovesdi. I.. Reichel. R. & Kevins. J. R. (1987). Proc. ,%t. Acad. Sci.. U.S.A. 84, 218cT2184. Kruczek, I. & Doertler, W. (1983). Proc. Nat. Acad. Sci., U.S.A.

80, 7586-7590.

Langner, K.-D., Vardimon, I,., Renz, D. & Doerfler. W. (1984). Proc. Nat. Acad. Sci., U.S.A. 81, 2950-2954. Langner, K.-D., Weyer, U. & Doerfler, W. (1986). Proc. Nat. Acad. Sci., (J.S.A. 83, 1598-1602. Leff. T., Elkaim, R., Goding, C. R., ,Jalinot, I’., SassoneCorsi, P., Perricaudet, M., Kedinger, C. & Chambon, 1’. (1984). Proc. Nat. Acad. Sci.. U.S.A. 81, 4381L 4385. Liibbert. H. & Doerfler, W. (1984). J. Viral. 52, 255-265. Maxam, A. M. & Gilbert, W. (1980). Methods Enzymol. 65, 499-560.

McKnight, S. & Tjian, R. (1986). Cell, 46, 795-805. Muller. U. & Doerfler, W. (1987). J. Viral. 61, 3710-3720. Murray, E. & Grosveld, F. (1985). In Biochemistry and Biology of DNA Methylation (Cantoni G. L. & Razin, A., eds), pp. 157-176, Alan R. Liss, Inc.. Kew York. xevins. J. R. (1982). CeEZ,29, 913-919. Razin. A. & Riggs, A. D. (1980). Science, 210, 604-610. Razin, A., Szyf, M., Kafri, T., Roll, M., Giloh, H., Scarpa, 8.. Carotti, D. & Cantoni, G. L. (1986). Proc. Nat. Acad. Ski., U.S.A. 83, 2827-2831. Reichel, R., Kovesdi, I. & Nevins, J. R. (1987). Cell. 48,

et al.

Rigby, P. W. J., Dieckmann, M., Rhodes. C. & Herg, P. (1977). ,J. Mol. Biol. 113, 237%251. Srhirm. S. & Doerfler. W. (1981). J. Viral. 39, 694-702. Scott. M. R. D., Westphal, K.-H. & Rigby. I’. W. J. (1983). Cell, 34, 5577567. Southern, E. M. (1975). J. Mol. BioE. 98, 503-517. Southern, I’. J. & Rerg, I’. (1982). J. Mol. Appl. Genet. 1. 327-34 I. D. & Doerfler, W. (1979). Cold Spring Harbor Symp. Quant. Riol. 44, 565568. Sutt,er. I). & Doerfler, W. (1980). Proc. Nat. Acad. Sci., C’.S.A. 77. 253-256.

Sutter.

Thompson, ,I. P., Granoff, A. & Willis, D. R. (1986). Pm:. Nat. Acad. Sci., U.S.A.

3231-3236.

Wahl, G. M., Stern, M. & Stark, G. R. (1979). t’roc. Nat. Acad.

Sci., U.S.A.

76, 3683-3687.

Zhang, X.-Y., Ehrlich, K. C., Wang, R. Y.-H. & Ehrlich. M. (1986). Nucl. Acids Res. 14, 8387-8397.

501-506.

Edited

83, 7688-7692.

Vardimon, L., Neumann, R., Kuhlmann. I., Sutter, D. & Doerfler, W. (1980). Nucl. Acids Res. 8, 2461-2473. Vardimon, L., Kressmann, A., Cedar, H.. Maechler. M. 8r Doerfler. W. (1982). Proc. Nat. Acad. Sci., F.S. A. 79: 1073.--1077. Visser. I,., van Maarschalkerweerd, M. W., RozlJn, T. H., Wassenaar, A. D. C., Reemst, A. M. C. B. & Sussenbach, J. S. (1979). Cold Spring Harbor Symp. Quant. Biol. 44, 541-550. Vogt. V. M. (1973). Eur. J. Biochem. 33, 192-200. Waalwijk, C. & Flavell, R. A. (1978). Nucl. Acids Res. 5,

by P. Chambon