JOURNAL OF VIROLOGY, Aug. 1989, p. 3479-34880022-538X/89/083479-10$02.00/0Copyright © 1989, American Society for Microbiology

Adeno-Associated Virus P5 Promoter Contains an Adenovirus ElA-Inducible Element and a Binding Site for the Major Late

Transcription FactorLONG-SHENG CHANG, YANG SHI, AND THOMAS SHENK*

Howvard Huighes Medic al Institute, Depar-tment of Biology, Princ eton University, Princeton, New Jersey 08544-1014

Received 22 February 1989/Accepted 10 May 1989

Activity of the adeno-associated virus P5 transcriptional control region was found to be induced byadenovirus EIA gene products. A pair of adjacent sequence elements was found to mediate both basal andElA-induced P5 activity. The first element is a binding site for the major late transcription factor (MLTF), a

factor first identified on the basis of its binding to a specific sequence within the adenovirus major latepromoter. The second element is a tandemly repeated 10-base-pair sequence whose relationship to previouslydescribed binding sites is unclear. Each element individually conferred ElA responsiveness on a heterologouspromoter, and deletion analysis demonstrated that each contributed to the level of P5 activity in the presence

of EIA products both in transfection- and infection-based assays. Although deletion of the MLTF binding siteled to reduced P5 transcriptional activity in the presence of EIA proteins, the deletion generated enhanced P5basal activity in the absence of the transcriptional activator. The negative effect of the MLTF binding site inthe absence of activator and its positive effect in the presence of activator combine to enhance the magnitudeof the response by the P5 control region to EIA gene products.

The human adeno-associated virus (AAV) type 2 is a

defective parvovirus (for a review, see reference 3). Theviral genome is a 4,681-nucleotide, single-stranded DNA(57), including terminal repeats of 145 bases which serve as

origins and primers for DNA replication (50, 53). The viralDNA sequence contains two major open reading frames.The right half of the conventional AAV map encodes threeviral structural (cap) proteins (5, 28, 48), and the left halfencodes four nonstructural (Rep) proteins (42). Three sets ofcapped and polyadenylated AAV mRNAs have been identi-fied (15, 17, 34, 39). Their 5' ends are located at map

positions 5, 19, and 40, respectively (16, 37), and their 3'ends coterminate at map position 96 (57). The presence ofTATA boxes approximately 30 bases upstream from eachRNA start site (16, 37) suggests that they are produced bythree separate promoters, designated P5, P19, and P40.Although AAV can replicate under appropriate conditions

in the absence of a helper virus (60), efficient replicationrequires coinfection with adenovirus or herpesvirus (3). Oneof the helper functions provided by adenovirus is the trans-activating activity of the early region 1A (ElA) gene prod-ucts (26, 33, 47, 58). The ElA gene is the first gene expressedduring an adenovirus infection, and its gene products cantranscriptionally activate other adenovirus early genes (forreviews, see references 2 and 43). In addition, ElA productshave been shown to activate transcription of the AAV P19promoter (58).Here we report that the AAV P5 promoter can be induced

by adenovirus ElA gene products. Deletion analysis identi-fied two sequence elements that mediate transcriptionalactivity of the control region. The first element is a bindingsite for the adenovirus major late transcription factor(MLTF). It rendered a heterologous promoter modestlyinducible by ElA gene products. This element reduced thebasal activity and enhanced the ElA-induced activity of theP5 control region. The second element is a tandemly re-

* Corresponding author.

peated 10-base-pair (bp) sequence. This sequence rendered a

heterologous promoter ElA responsive and enhanced theElA-induced activity of the P5 promoter. The two elementsappear to interact to make the P5 promoter highly responsiveto ElA gene products.

MATERIALS AND METHODS

Plasmids, viruses, and cells. A DNA segment including theAAV P5 transcriptional control region (nucleotide numbers190 to 310; XbaI-HhaI cleavage product) was subclonedfrom psub201 (49) and then fused with the chloramphenicolacetyltransferase (CAT) expression vector (12) to generatethe pAAVP5-CAT190 plasmid (see Fig. 1). Deleted deriva-tives of the P5 promoter were constructed by using Bal3lnuclease (38), and deletion endpoints were mapped by DNAsequencing (51). Several deletion derivatives were preparedin which the rabbit ,-globin cDNA from pSV2-p-globin (56)was substituted for the CAT coding region. To constructplasmids for generation of AAV/P5-CAT transducing vi-ruses, the P5-CAT-specific DNA segment was excised fromseveral pAAVP5-CAT plasmids and ligated to the smallXbaI fragment from the AAV recombinant plasmidpsiib2ol(+) (49). These recombinant virus plasmids containthe P5-CAT sequences inserted between the two cis-actingterminal repeats of AAV.The pElA-ElB plasmid containing the entire adenovirus

type 5 (AdS) El region (nucleotide numbers 1 through 5778),pElA containing only the ElA region (nucleotide numbers 1through 1767), and pElB containing only the E1B region(nucleotide numbers 1339 through 5778) have been describedpreviously (21, 46). pElA-CAT (30) contains the ElA pro-moter (sequence -499 to + 113 relative to the ElA cap site),pE2A-CAT (45) contains the E2 early promoter, and pMLP-CAT contains the major late promoter (sequence -400 to+30 relative to the major late cap site) linked to the CATcoding region. The pBC19 and pBC21 plasmids (obtainedfrom M. Labow, Princeton University) contain the BglII-BarnHI fragment carrying the simian virus 40 (SV40) pro-

3479

Vol. 63, No. 8

3480 CHANG ET AL.

moter (without enhancer and 21-bp repeats) and CAT codingregion from pAlOCAT (32) cloned in the BamniiHI site of thepBluescript KS(+) vector (Stratagene Inc., La Jolla, Calif.)in both orientations. A 22-bp synthetic DNA containing the10-bp direct repeat in the upstream region of the P5 promoter(-51 to -70 with respect to the AAV transcription start site)was inserted into the Smalu site of the pBC19 or pBC21plasmid to generate the pBC19-8, pBC19-12, and pBC21-24plasmids (see Fig. 6). Sequencing data indicated that pBC19-12 contains the repeat inserted 5' of the SV40 promoter andin the same orientation as that in the P5 promoter, pBC19-8contains the repeat inserted in the opposite orientation, andpBC21-24 contains the repeat inserted in the 3' end of theSV-CAT expression cassette. Similarly, a 16-bp syntheticDNA containing the MLTF binding site homology (-70 to-85 with respect to the AAV transcription start site) wasinserted into the StnaI site of the pBC19 or pBC21 plasmid togenerate the pBC19-6, pBC19-11, pBC21-30, and pBC21-31plasmids. Sequencing data indicated that pBC19-6 andpBC19-11 contain the MLTF binding site homology insertedin opposite orientations upstream of the TATA motif of theSV40 early promoter and pBC21-30 and pBC21-31 containthe MLTF binding site homology inserted in opposite orien-tations in the 3' end of the SV-CAT expression cassette.AAV/P5-CAT recombinant virus stocks were prepared by

cotransfecting 2 jg of pSM620 DNA (wild-type AAV [50])into HeLa cells by the DEAE-dextran method (40) in thepresence of adenovirus (c1309, a phenotypically wild-typeAd5 [29]) at a multiplicity of 10 PFU per cell. When maximalcytopathic effect occurred, the cells were harvested andfrozen-thawed three times. The adenovirus in the resultinglysate was inactivated by heating the virus stock at 56°C for1 h. The AAV/P5-CAT recombinant viruses were quanti-tated by determining in a dot blot assay the amount ofAAV/P5-CAT-specific DNA in virion preparations with a32P-labeled CAT DNA probe.HeLa cells (American Type Culture Collection) and ade-

novirus-transformed human 293 cells (14) were grown inmedium supplemented with 10% calf serum.

Transfection, infection, and CAT assays. Actively growingHeLa monolayer cultures were split 1:5 in Dulbecco modi-fied Eagle medium containing 10% fetal calf serum at 20 to 24h before transfection and/or infection. Calcium phosphate-DNA coprecipitate mixtures were prepared as previouslydescribed (13). Cells on each 100-mm-diameter dish werecotransfected with 2 Fg of each P5-CAT plasmid and 18 Fxgof adenovirus-specific plasmid or control pBR322 DNA. Theprecipitate was left on the cells in the presence of mediumfor 12 to 14 h, after which the medium was replaced withfresh medium containing 10% serum. When required, cellswere infected with adenovirus at 20 PFU per cell for 1 hbefore transfection. At 48 h after transfection, cells wereharvested and lysed in 0.25 M Tris hydrochloride (pH 7.8) byfreeze-thawing. The protein concentration of the extract wasdetermined, and the same amount of extract protein wasused in each CAT assay (12). To quantitate relative levels ofCAT activity, areas of silica gel containing acetylated andunacetylated chloramphenicol were scraped off the thin-layer chromatography plate, their radioactivities were deter-mined by liquid scintillation counting, and the percentage ofconversion was calculated.RNA preparation and analysis. To assay steady-state lev-

els of P-globin RNA encoded by pAAVP5-P3G190. totalcytoplasmic RNA was isolated from HeLa cells at 48 h aftertransfection (21) and analyzed by ribonuclease protection(41). The 32 P-labeled RNA probe contained the P5 control

region plus the 5' half of the P-globin coding region. Theprobe protected a B-globin RNA fragment of about 400nucleotides. Protected bands were quantified by densitomet-ric scanning of appropriate autoradiographic exposures pro-duced by using preflashed Kodak XAR-5 film.To measure transcription rates, nuclei were prepared fiom

HeLa cells 48 h after transfection, nuclei were incubated for15 min at 30°C in the presence of [32P]UTP, and labeled RNAwas isolated and hybridized to probe DNAs immobilized onnitrocellulose filters (18, 25). Probe DNAs were prepared bydigestion of pAAVP5-3G190 with XhbI plus BartnHI anddigestion of pLK221 (human f3-actin cDNA [19]) withBaml-HI, separation of the resulting fragments by electropho-resis, and transfer of the entire pattern to nitrocellulosestrips. After hybridization, the radioactivity in individualbands was quantified by scintillation counting.Measurement of DNA stability. Analysis of plasmid DNA

stability was conducted as described by Alwine (1). Briefly,at 48 h after transfection, low-molecular-weight DNA wasprepared (24), equivalent portions of each sample weredigested with XhbI, subjected to electrophoresis in a 1%agarose gel, and hybridized to a 32P-labeled P3-globin-specificprobe DNA (55). After autoradiography, bands were quan-tified by densitometric scanning.

Band-shift, footprint and photo-cross-link analysis. Twooligodeoxynucleotides containing MLTF binding sites wereprepared and used in these experiments. One contained thesequence of the MLTF binding site present on the Ad2chromosome (5'-GGTGTAGGCCACGTGACCGGGTGTTC-3'), and the other contained the homologous sequencepresent in the AAV P5 promoter (5'-GAGGTCACGTGAGTGT-3').MLTF activity was partially purified from a nuclear ex-

tract of HeLa cells (9) by chromatography on a TSKphenyl-5PW matrix in an LKB UltroPac HPLC column(Pharmacia LKB Biotechnology, Inc., Piscataway, N.J.).Activity was localized in the elution profile by band-shiftassay, using the 32P-labeled oligodeoxynucleotides corre-sponding to the adenovirus and P5 MLTF binding sites. Thepeak fraction was used as the source of MLTF for footprint,band-shift, and photo-cross-linking analysis.

Band-shift assays (10) were pertormed as described pre-viously (20), with minor modifications. Each reaction mix-ture contained 2 pd of partially purified MLTF, 10 fmol of32P-labeled probe DNA, and 2 Vig of poly(dI) poly(dC).Binding mixtures were incubated at 25°C for 30 min, DNA-protein complexes were separated from free DNA by elec-trophoresis in a 4(4 polyacrylamide gel (bisacrylamide-acryl-amide, 1:20), and gels were transferred to 3MM paper, dried,and then exposed to film.

Footprint analysis (11) utilized a DNA substrate compris-ing the sequence between -96 to +24 relative to the P5transcriptional start site. The substrate DNA was 32P-labeledat the end at position -96 by filling in the overhanging end ofan XhbI cleavage site with the Klenow fragment of DNApolymerase I and [a-32P]dCTP. DNA-protein complexeswere formed as described above for band-shift assays andtreated with 10 pg of DNase I per ml for 1 min at 25°C; DNAwas purified and then analyzed for protected regions byelectrophoresis in a 6% polyacrylamide gel containing 8 Murea.To photo-cross-link the MLTF to its binding site, reaction

mixtures were prepared as described above for band-shiftassays and incubated to allow binding to reach equilibrium.Reaction mixtures were pipetted onto parafilm and irradiatedfor various times at a distance of 4.5 cm from a UV lamp

J. VIROL.

AAV PS PROMOTER 3481

A. A4vp-: F K190 Coc M20 27o0 24CGGT0C TC?TAT7IGA&;Ti-C.CT.AGTTTT7T^'r-CAT-T C,C3.LTG<-C4Cr-T

-90 -s0 -70 -tj -50 40MLTF Rl R2

260 270 3SC; 290 7;0 .vuGTATTT AAT TT

-30 -20 -10 10 +0

TATAC PAARVP5-CAT 190 a: : 4-

B. pAAVP5-CATl90CAT

AAVP5 31 SPLICELPOLY A

v e o~~~~~~~~~~~~~~~~iy

I~~~~~190.S'

TABLE 1. Rate of P5 transcription was increased bypEIA-EIB gene products'

Cotransfecting plasmidProbe DNA (cpm hybridized) Induction"

pBR322 pEIA-EIB

3 globin 18 162 11.1Actin 406 330 1.0

,* w _ " HeLa cells were cotransf'ected with pAAVP5-)3G190 plus either pBR322W or pElA-EIB, and nuiclei were prepared 48 h latter. Run-on transcription was

performed in the presence of [os-32P]UTP and nuclear RNA was prepaired andhybridized to 3-globin and aictin (non-ElA-indLucible) probe DNAs whichwere immobilized on nitrocellulose filter patper. After filters were washed,radioactivity was quantified.

b Induiction ratios were calculated, normalizing the radioactivity hybridizedto actin DNA in the presence and aibsence of EIA products to reflect the flct

* 9 9 * that EIA products do not inflLtence actin transcription r-ates.

rq~ CI I,jl.e ll$^ o qo -

FIG. 1. AAV P5 5' flanking sequence and its response to aden-ovirus EIA proteins. (A) Sequence of the AAV PS 5' flankingregion. Numbers above the sequence are nucleotide positions ac-

cording to Srivastava et al. (57), and numbers below are nucleotidepositions relative to the P5 cap site at +1. The MLTF binding site,R1-R2 repeated element, and TATA box sequence are indicated. (B)Structure of pAAVP5-CAT190. AAV P5 sequences (nucleotideposition 190 to 310) are represented by a solid segment in thecircular plasmid. CAT coding sequences. SV40 splice site, and SV40poly(A) addition site are indicated. The CAT mRNA predicted to beencoded by the plasmid is designated by an arrow. (C) CAT assaysshowing the effect on pAAVP5-CAT190 expression of gene productsintroduced by cotransfection with plasmids (pElA. pBR322. pElA-E1B) or by transfection and simultaneous infection (AdS). Extractsof HeLa cells were prepared at 48 h after transfection and assayedfor CAT expression. pSV2-CAT-transfected HeLa cells were in-cluded as a control.

(310-nm filter; Fotodyne Inc., New Berlin, Wis.). Sampleswere not treated with nuclease since the substrate DNA was

labeled at its 5' ends. Cross-linked complexes were analyzedby electrophoresis in an 12.5% polyacrylamide gel contain-ing 0.1% sodium dodecyl sulfate. After electrophoresis, thegel was dried onto filter paper, and radioactive bands werevisualized by autoradiography.

RESULTS

Induction of the AAV P5 control region by adenovirus ElAproducts. pAAVP5-CAT190 was constructed to assay theeffect of adenovirus ElA gene products on the activity of theAAV P5 transcriptional control region. This plasmid con-tains the AAV P5 5'-flanking domain (Fig. 1A) appended tothe CAT reporter gene (Fig. IB). Extracts were prepared at48 h after transfection of HeLa cells with pAAVP5-CAT190DNA and assayed for CAT activity. A low basal, i.e.,uninduced, level of expression was detected (Fig. 1C, +pBR322). As shown below, at least a portion of this basalactivity is due to proper initiation at the P5 start site. Ifplasmids encoding either the adenovirus ElA gene (Fig. 1C,+ pElA) or ElA plus E1B genes (Fig. 1C, + pElA-EIB)were included in the transfection mixture with pAAVP5-CAT190, the level of CAT activity increased by a factor of 7and 23, respectively. The adenovirus ElB gene alone did notstimulate pAAVP5-CAT190 expression above basal levels(data not shown). Nevertheless, pElA-EIB was consistently

observed to be a more potent inducer of AAV P5-controlledCAT expression, presumably because the E1B coding regioncontains a (cis-acting element that enhances ElA expression(27). CAT expression was also stimulated if ElA functionwas provided by adenovirus infection of pAAVP5-CAT190-transfected HeLa cells (Fig. 1C, + AdS).

In multiple experiments, pAAVP5-CAT190 expressionwas induced by a factor of 40 + 15.1 in the presence ofpE1A-ElB. This level of induction was equal to or some-what greater than that obtained for the well-studied adeno-virus E2 early promoter (pE2A-CAT, 29 + 14-fold induction)and much greater than that obtained for the ElA controlregion (pElA-CAT, 2.9 + 1.1-fold induction).A transcriptional rate analysis was performed to directly

demonstrate that EIA gene products could stimulate tran-scriptional activity of the AAV P5 control region (Table 1).HeLa cells were cotransfected with either pAAVP5-,3G190(identical to pAAVP5-CAT190 except that the 3-globincodinQ region is substituted for the CAT sequence sinceCAT-encoding mRNAs were found to be unstable) pluspBR322 (included to control for DNA concentration in thetransfecting mixture) or pAAVP5-3G190 plus pElA-ElB. Invitro transcriptional run-on analysis demonstrated that tran-scription of P3-globin sequences was stimulated by a factor of11 When ElA gene products were present.To be certain this result reflected induction of AAV

P5-directed transcriptional initiation, two control experi-ments were performed. First, an experiment was conductedto confirm that the transcriptional induction resulted frominitiation at the AAV P5 start site (Fig. 2A). RNA wasprepared at 48 h after cotransfection of HeLa cells withpAAVP5-3G190 and pBR322 or pElA-EIB, and 3-globintranscripts initiated at the AAV P5 start site were assayed byRNase protection analysis. There was a 15-fold increase inthe level of appropriately initiated 3-globin RNA whenpAAVP5-3G190 was cotransfected with the ElA-expressingplasmid. A faster-migrating protected fragment was alsoobserved in this analysis, and its origin is uncertain. Short-ened 3-globin transcripts have been observed previously (44,56), and they may be due to initiation downstream of themajor start site. The shortened fragment was also induced byElA gene products. In the second control experiment, thestability of pAAVP5-3G190 DNA was assayed subsequentto cotransfection in the presence of pBR322 compared withpElA-EIB DNA (Fig. 2B). Low-molecular-weight DNAwas prepared from the same experiment in which steady-state P-globin RNA levels were analyzed above. pAAVP5-3G190 DNA levels were assayed by DNA blot analysis,

Vot- 63. 1989

3482 CHANG ET AL.

A.PNassePROTECTION*pAAVP;;-

c,Jc\.}wi< a::LLJCQ

co

B NCN&STABLITYpA P5- j3I19OrEIA-EIB pBR322 MKR

X! X a

-x c Od X d C O

- -400nt.. -11I

MLTF RI RZ...........~~~~~~~~~..........C....T..F...r ---. ,W-S- -- . _~~~

.i;

B;iv-;.>

..

- 190I'"I

- 21T- 2.t0-- 2 4-,R.._ 2 L 2

i,

- - + - + ;CA 13.3 0.2 0.2 0.4 78 3 b6 C. i4. . (;

~.4.3 1.0 44.5 4.3 4 7 1.0 iAi..j i19O 31.I- i D 2i' 3 243 pv

FIG. 2. Steady-state levels of RNA transcribed under control ofthe AAV P5 promoter and transfected plasmid DNA stability in thepresence and absence of ElA gene products. (A) RNase protectionanalysis of 3-globin RNA in the cytoplasm of HeLa cells at 48 h aftertransfection with pAAVP5-PG190 plus either pBR322 or pElA-ElB.The probe RNA corresponded to the sequences containing the P5control region plus the 5' region of the ,B-globin coding region. Thefaster-migrating band probably results from internal initiation within3-globin sequences. (B) DNA blot analysis of pAAVP5-,G190 DNA

isolated at 48 h after transfection of HeLa cells. 32P-labeled ,-globin-specific DNA served as probe. The relative amounts of DNAapplied to individual lanes are indicated at the top of each lane. LaneMKR, Purified pAAVP5-PG190 DNA which was not transfectedinto HeLa cells. 111, Position of linear (form 111) pAAVP5-CAT190DNA; nt, nucleotides; x, fold dilution.

using a 32P-labeled probe DNA. Although about twofoldmore pAAVP5-PG190 DNA persisted in cells that alsoreceived pElA-ElB compared with those receiving pBR322,this difference is not large enough to account for the in-creased level of 3-globin-specific transcription.

In sum, pAAVP5-,BG190 DNA is not markedly stabilizedin HeLa cells expressing pElA-EIB-coded products, andthe 3-globin RNA molecules encoded by pAAVP5-3G190are initiated at the AAV P5 start site. Therefore, we canconclude, on the basis of the enhanced rate of ,B-globintranscription in the presence of pElA-ElB, that the AAV P5control region is stimulated by ElA gene products.AAV P5 sequence elements that respond to adenovirus EIA

products. In addition to a TATA motif, the AAV P5 5'-flanking region contains two additional sequence elementsthat stand out on initial inspection (Fig. 1A). First, a se-quence homologous to the adenovirus MLTF binding site isevident between AAV sequence position 204 to 215 (-83 to-72 relative to the P5 cap site). In all, 9 of 12 bp in the AAVP5 sequence match the equivalent adenovirus sequence.Second, a 10-bp sequence which is directly repeated liesbetween AAV sequence position 217 to 236, just down-

--4 - + - pt.;-\~~~0.-LiB02 3 K 3 0 2 22r' 9 NVE P5i2Nv 1. ..I) ,... X, E 18, FLANLLJ1 N

292 2 2 271 2 7 n 3ni A AWV P5-CaT

FIG. 3. Effect of deletion mutations on AAV P5 activity. (A)Structure of deleted P5 derivatives. The top of the diagram is arepresentation of the P5 promoter on which the MLTF binding site,R1-R2 repeat, and mRNA cap site are designated. Deleted deriva-tives contain unidirectional deletions running from position 190toward the mRNA cap site. Mutant templates are designated by thenumber of the last nucleotide to be removed by their deletion. (B)CAT assays showing the effect of deletions within the P5 controlregion on its activity. Extracts were prepared at 48 h after transfec-tion, and CAT activity was determined. Assays were performedeither in the absence (-) or presence (+) of cotransfecting pElA-E1B to provide EIA function. Percent conversion was determinedby measuring the radioactivity in acetylated versus unacetylatedspots. ElA induction was calculated by dividing the percent con-version in the presence of EIA by the percent conversion in theabsence of ElA.

stream of the MLTF binding site homology. The relationshipof this repeat, termed R1-R2, to other known factor bindingsites is unclear.To evaluate the role of these and other sequence elements

in the function of the AAV P5 control region, a series ofunidirectional deletion mutations was constructed within theP5 segment of pAAVP5-CAT190 (Fig. 3A). Basal and E1A-induced activity of each mutant AAV P5 control region wasthen evaluated by CAT assay of transfected HeLa cellextracts. The experiment was repeated three times, and arepresentative trial is displayed in Fig. 3B. Basal activityremained low and constant for all mutant plasmids with theexception of pAAVP5-CAT216 (unidirectional deletion ex-tends from upstream to position -216, Fig. 3B). This variant

J. VIROL.

AAV P5 PROMOTER 3483

A. RECOMBINANT VIRUSESITR P5 CAT

CATl310 CAT

191 243

B. CAT ASSAYS

ITR

91 1AAVP5-CAT 190

.J AAVP5-CAT243

^.^._~~~~~~~~~~~~~~~~~.0 .1wstws_~~~~~~~~~~~~~1 4 0

24 48 24 48 12 240.7 0.6 0.7 0.7 2.8 74

190 243 l

36 48 12 2493 94 1.0 7.5

190 24

36 48 hr p.i.14 14 % CONVERSION

ADENOVIRUS.3 AAVP5-CAT

FIG. 4. Analysis of AAV PS function in the context of a viral infection. (A) Structure of recombinant AAV chromosomes. Invertedterminal repeat (ITR), PS promoter, and CAT coding regions are designated. The first and last nucleotides deleted in AAVP5-CAT243 aredesignated (191. 243). (B) CAT assays showing the effect of the deletion of the PS sequence between position 191 to 243 on PS activity.Extracts were prepared at various times after HeLa cells were infected (hr p.i.) with recombinant AAV viruses, and CAT activity wasdetermined. Assays were performed either in the absence (-) or presence (+) of coinfecting adenovirus to provide ElA function. Percentconversion was determined by measuring the radioactivity in acetylated versus unacetylated spots.

exhibited enhanced (about threefold) CAT expression in theabsence of ElA products. In contrast, it exhibited a reduced(about threefold) level of ElA-induced activity comparedwith that of plasmids that contained the sequence intervalbetween 190 to 216. As a result, pAAVP5-CAT216 exhibiteda substantially reduced EIA induction ratio (CAT activity inthe presence divided by activity in the absence of EIAproducts). The behavior of pAAVP5-CAT216 suggests theremay be a sequence between position 190 to 227 that nega-tively influences basal activity but positively contributes toElA-induced activity of the AAV P5 control region. TheMLTF binding site homology is located within this interval.ElA-inducible activity was completely lost when the dele-tion was extended to sequence position 243. pAAVP5-CAT243 displays identical basal and induced levels of CATactivity. Thus, the interval between position 216 to 243appears to contain a sequence element that responds posi-tively to the presence of ElA products. This interval con-tains the R1-R2 repeat.Two recombinant AAV viruses were produced to confirm

that the response of P5 elements to EIA products measuredin transfection assays properly reflected their role in AAV-infected cells. pAAVP5-CAT190 and pAAVP5-CAT243were used to generate derivatives of psub201, an infectiousclone of AAV (49). DNA segments containing the P5 tran-scriptional control region plus CAT reporter gene weresubstituted for the AAV coding region between the viralterminal repeat sequences (Fig. 4A). Recombinant viruseswere propagated by using pSM620 (recombinant clone ofwild-type AAV [50]) as a helper DNA. The same amount ofthe AAV/P5-CAT recombinant viruses was used to infectHeLa cells either alone or in the presence of adenovirus.

AAVP5-CAT190 virus generated substantial CAT activity incells coinfected with adenovirus, whereas AAVP5-CAT243generated a much reduced signal under similar conditions(Fig. 4B). The level of CAT activity observed in this exper-iment for AAVP5-CAT243 subsequent to coinfection withthe AAV variant, wild-type AAV, and adenovirus (Fig. 4B)was substantially greater than background levels of CATactivity generated after transfection with pAAVP5-CAT243plus pElA-ElB (Fig. 3B). This difference is due to the factthat AAVP5-CAT243 DNA has replicated, greatly amplify-ing the CAT template copy number.The domain of the P5 transcriptional control region lo-

cated within sequence position 190 to 243 clearly plays a keyrole in P5 activity within AAV-infected cells.MLTF binding site and R1-R2 repeat can each confer

ElA-inducibility on a heterologous promoter. The roles of theR1-R2 repeat and the MLTF binding site homology in theresponse of the AAV P5 promoter to ElA products werefurther investigated by placing them upstream of a heterol-ogous promoter that was not ElA responsive. The twoelements were separately inserted into plasmids containing aCAT reporter gene controlled by the SV40 early promoter(TATA motif plus start site) and tested for the ability torespond to EIA products in HeLa cells transfected with testplasmids plus pElA-E1B (Table 2).As expected, expression of the parental plasmids, pBC19

and pBC21, was not induced by ElA products since theycontained a TATA motif previously shown to be nonrespon-sive (54). pBC19-8 and pBC19-12 contained the R1-R2 ele-ment on a 22-bp segment (5'-GTTTTGCGACATTTTGCGACAC-3') inserted on the 5' side of the CAT sequence inopposite orientations. The inserts did not increase basal

Vot- 63, 1989

f:::.

3484 CHANG ET AL.

TABLE 2. AAV P5 R1-R2 element rendered a heterologouspromoter ElA inducible'

Test element" % Conversion'Plasmid Induc-

Sequence Loca- Orien- -E1A +E1A tion"tin tation

pBC19 None 0.27 0.27 1.0pBC19-8 R1-R2 5' - 0.27 0.84 3.1pBC19-12 R1-R2 5' + 0.29 2.33 8.0pBC19-11 ML' 5' - 0.31 0.60 1.9pBC19-6 ML 5' + 0.32 1.30 4.1pBC21 None 0.28 0.29 1.0pBC21-24 R1-R2 3' + 0.28 0.47 1.7pBC21-31 ML 3' - 0.29 0.32 1.1pBC21-30 ML 3' + 0.28 0.30 1.1pAdMLP-CAT 0.47 3.16 6.7pAAVP5-CAT190 0.55 18.78 34.1

" Extracts were prepared at 48 h after transfection of HeLa cells andassayed for CAT activity. The values reported are the averages obtained byanalysis of duplicate extracts.

b The test element was inserted into pBC19 or pBC21 in the same (+) or theopposite (-) orientation relative to the transcriptional start site as it is foundin the AAV P5 control region.

' The percent conversion of CAT (0.12%) by extracts of cells transfectedwith a plasmid that did not contain a CAT coding region was subtracted fromall experimental values.

" Induction was calculated as the percent conversion of CAT obtainedwhen indicated plasmids were cotransfected with pElA-ElB DNA divided bythe percent conversion of CAT when they were cotransfected with pBR322DNA.

" ML, MLTF binding site.

CAT expression, but did increase the level of expression inthe presence of ElA products. Expression from pBC19-8and pBC19-12 was induced by factors of 3 and 8, respec-tively. pBC21-24 contained the R1-R2 repeat at the 3' side ofthe CAT coding region, and it exhibited a marginal, 1.7-fold,induction in the presence of ElA products.The response of pBC19-8 and pBC19-12 to ElA products

was further studied by transfection of 293 cells whichconstitutively express ElA gene products (Table 3). BothR1-R2-containing plasmids generated about 20-fold moreCAT activity than the parental plasmid that lacked thesequence. The orientation of the R1-R2 insert upstream ofthe SV40 TATA motif reproducibly (three independentexperiments) influenced the magnitude of the response toElA products in the HeLa cell cotransfection assay (Table 2)but not that in the 293 cell assay (Table 3). The reason for the

TABLE 3. AAV P5 R1-R2 element enhanced expression froma heterologous promoter in 293 cells"

Test element"Plasmid Induction'

Sequence Location Orientation

pBC19 None 1.0pBC19-8 R1-R2 5' - 19.5pBC19-12 R1-R2 5' + 21.6PBC19-11 ML 5' - 2.3pBC19-6 ML 5' + 3.8

" Extracts were prepared at 48 h after transfection of 293 cells and assayedfor CAT activity. The values reported are the averages obtained by analysis ofduplicate extracts.

" The test element was inserted into pBC19 or pBC21 in the same (+) or theopposite (-) orientation relative to the transcriptional start site as it is foundin the AAV P5 control region.

' Induction was calculated as the percent conversion of CAT obtained forindicated test plasmids divided by the percent conversion of CAT obtained forpBC19.

assay-dependent difference is not clear. Possibly it reflects ahigher level of ElA protein in 293 cells than in transfectedHeLa cells. This could result in more efficient trans-activa-tion in 293 cell assays (4).pBC19-6 and pBC19-11 contain the MLTF binding site

homology on a 16-bp segment (5'-GAGGTCACGTGAGTGT-3') inserted in opposite orientations upstream of theTATA motif of the SV40 early promoter. These plasmidsdisplayed the same basal level CAT activity as the parentalplasmid but were induced by a factor of 2 to 4 in response toElA products when assayed in HeLa cells by cotransfectionwith pElA-ElB (Table 2) or in 293 cells (Table 3). The levelof their response to EIA compared favorably with that seenfor pMLP-CAT which contains the entire major late tran-scriptional control region (Table 2). Plasmids containing thissequence inserted to the 3' side of the CAT sequence,pBC21-30 and pBC21-31, were not responsive to ElA geneproducts (Table 2).

In sum, both the MLTF binding site homology and theR1-R2 sequence endowed EIA responsiveness on a heterol-ogous promoter. Both sequence elements presumably con-tributed to the overall response of the PS promoter to ElAgene products. The R1-R2 sequences displayed the greaterresponse to the adenovirus trans activator. The magnitude ofthe induction mediated by R1-R2 depended on the orienta-tion of the sequence and the assay employed.MLTF binds to the MLTF binding site homology in AAV

P5. Although the P5 control region contained a sequence thatclosely resembled a bona fide MLTF recognition site, it wasnecessary to prove that the factor did, indeed, bind to thesequence. Accordingly, MLTF activity was partially purifiedfrom a nuclear extract on a TSK phenyl-5PW matrix, and itsinteractions with the P5 homology were analyzed.

Footprint analysis (11) was performed by using the par-tially purified MLTF activity and a P5-specific DNA frag-ment as substrate (Fig. 5). A 16-nucleotide sequence wasprotected from DNase I cleavage (AAV sequence 202 to 217,inclusive, or -70 to -85 relative to the PS start site). Thisprotected region was centered on the MLTF binding sitehomology (sequence position 204 to 213, inclusive, Fig. 1A).Thus, the footprint analysis is consistent with the possibilitythat MLTF binds to the AAV PS promoter.

Next, band-shift analysis (10) was utilized to test theability of the PS and adenovirus MLTF binding sites tocross-compete for factor binding (Fig. 6). Two double-stranded oligodeoxynucleotides were chemically synthe-sized that corresponded to the bona fide adenovirus MLTFbinding site and the PS homology. When either of theoligodeoxynucleotides was 32P-labeled at its 5' ends andused as a substrate in a DNA band-shift assay employingpartially purified MLTF, two characteristic shifted bandswere evident. The bands produced by using either of the twoDNA substrates migrated identically. The generation of theshifted bands could be inhibited either by self competition, inwhich an excess of unlabeled homologous oligodeoxynucle-otide was included in the binding mixture, or by competitionwith the heterologous DNA. Although the adenovirus-spe-cific oligodeoxynucleotide appears to be a stronger cross-competitor than PS DNA in the experiment displayed in Fig.6, this result was not reproducible. The two DNAs exhibitedsimilar affinities for the factor in repeated cross-competitionexperiments (data not shown). However, it is not possible toconclude that the two binding sites have similar affinities forMLTF since the two DNAs were different in size (16 versus26 bp) and flanking sequences could influence binding af-finity. Nevertheless, the fact that the adenovirus- and AAV-

J. VIROL.

AAV P5 PROMOTER 3485

4 : !, !' rS11 ~- -R-F'- '- '- _E-

do m~

AdiAd AAV5 25 5 25

AAV PMOBEAAV Ad C0MPETITOR5 25 5 25 FOLD EXCESS

IMLTFJCOMPLEX

do

FIG. 5. DNase I footprint analysis of partially purified MLTF onthe AAV P5 control region. The leftmost lanes contain the end-laibeled P5 DNA segment subjected to chemical cleavage to producea sequence standard. Reaction mixtures received no protein (lanes0) or received partially purified MLTF (lanes 4 or 10 Vd). Theprotected sequence of the AAV minus-strand DNA is indicaited atthe right.

specific binding sites generated shifted bands of identicalmobility and exhibited cross-competition argues stronglythat the two DNAs bind the same factor.

Finally, cross-linking with UV light (8) was employed toexamine the size of the polypeptide that bound to the P5homology (Fig. 7). The 32P-labeled oligodeoxynucleotidecontaining the P5-specific sequence was mixed with partiallypurified MLTF, binding was allowed to reach equilibriuIM.the mixture was subjected to UV irradiation for varioustimes, and the cross-linked material was analyzed by elec-trophoresis. A band migrating at about 46 kilodaltons (kDa),the reported size of MLTF (43 to 48 kDa [8, 52]), wasgenerated by this procedure. The intensity of the 46-kDaband increased linearly with increasing time of irradiation,consistent with a simple interaction between the activatedDNA and a closely associated protein. Production of the46-kDa band could be inhibited by competition with eitherunlabeled homologous DNA or with the heterologous. ade-novirus-specific DNA. The generation of an appropriatelysized cross-linked product and the cross-competition bothprovided further evidence that the MLTF binds to the AAVP5 control region.

DISCUSSION

We have demonstrated that the AAV P5 promoter is EIAresponsive (Fig. 1, 2, and 4, Table 1). Two sequenceelements play a role in the induction process, a MLTF

IinEIininI __ini ]FREE DNAFIG. 6. Band-shift analysis of partially purified MLTF. using

oligodeoxynucleotide substrates that contain MLTF binding sites.Ad. DNAs that contain the Ad2 MLTF binding site; AAV, DNAsthat contain the AAV P5 homology to the binding site. Numbersabove the lanes indicate the molar excess of unlabeled, competitorDNAs. Bands corresponding to MLTF complex and free DNA arelabeled.

binding site and a direct repeat termed R1-R2 (Fig. 3, Tables2 and 3).A variety of observations support the conclusion that the

MLTF binds to the P5 control region. First, in a DNase Iprotection assay, a partially purified preparation of MLTFprotected a 16-nucleotide stretch that included the MLTFbinding site homology in P5 (Fig. 5). Second, specific shiftedcomplexes formed by using oligodeoxynucleotides contain-ing either the bona fide adenovirus MLTF binding site orthe P5 homology migrated identically, and the two DNAsexhibited cross-competition for the factor (Fig. 6). Third,photo-cross-linking and analysis of the resulting complex byelectrophoresis demonstrated that the P5-specific oligodeox-ynucleotide bound to a polypeptide of about 46 kDa, thereported size of the MLTF (8. 52), and inclusion of an excessof unlabeled oligodeoxynucleotide containing the adenovirusMLTF binding site in the binding mixture could preventcross-linking to the P5 DNA (Fig. 7).

Besides its role in adenovirus major late and AAV P5transcription, the MLTF has been shown to bind and acti-vate both the mouse metallothionein I (6) and the raty-fibrinogen transcriptional control regions (7). Inspection ofthe four binding sites indicates they share an invariant 6-bpcore sequence, CNCGTGA. The core region of the AAV P5sequence is palindromic, TCACGTGA.The second P5 sequence element that is involved in

ElA-mediated induction comprises a direct repeat, R1-R2,each copy of which is 10 bp in length. The relationship of thissequence to known factor binding sites is unclear. TheR1-R2 repeat, TTTTGCGACA, is related to elements withenhancer activity that reside upstream of the adenovirusEIA (ATTTACCACA, a 7 of 10 match [22]) and polyoma-virus early (TTTTGCAAGA, an 8 of 10 match [23J) codingregions. As yet, however, we do not know whether the samefactor binds to these related sequences.Very few discrete sequence elements, only certain TATA

box variants (TATAA [54, 59]), an 83-bp sequence thatincludes an E2F binding site (31), and a 100-bp segment that

Vol. 63, 1989

_ _I -- ..

3486 CHANG ET AL.

z

H

NO COMPETITOR- MIN. UV COMPETITOR0z 0 5 10 15 30Ad AAVNS

-97kd

-68kd

-43kd

A * _ 29 kdFIG. 7. UV cross-linking of MLTF to the P5 binding site. Par-

tially purified MLTF was mixed with the 5'-end-labeled oligodeox-ynucleotide corresponding to the P5 MLTF binding site, bindingwas allowed to reach equilibrium, reaction mixtures were irradiated.and cross-linked complexes were analyzed by electrophoresis insodium dodecyl sulfate-containing polyacrylamide gels. Lanes: Noprotein, no MLTF; No competitor, exposed to UV light for thetimes (minutes) indicated above each lane; Competitor, the productsof reaction mixtures exposed to UV light for 30 min that included a200-fold molar excess of unlabeled oligodeoxynucleotide containingthe adenovirus MLTF binding site (Ad), a 200-fold molar excess ofunlabeled oligodeoxynucleotide containing the AAV P5 MLTFbinding site (AAV), or a 380-fold molar excess of a nonspecificoligodeoxynucleotide (NS). The positions of unlabeled size markersare designated. kd, Kilodaltons.

includes a cyclic AMP response element (35) have beenshown to confer ElA responsiveness on a heterologouspromoter. Both the P5 MLTF binding site and R1-R2 ele-ments displayed this property (Tables 2 and 3). The MLTFbinding site, when appended to the SV40 early start site plusTATA motif (neither the SV40 TATTTAT nor the AAV P5TATTTA sequences would be predicted to be ElA respon-sive [54]), generated a control region that was modestlyresponsive to EIA products (two- to fourfold). Insertion ofthe R1-R2 sequence generated a more substantial 3- to20-fold increase in the same assays. It appears very likelythat the R1-R2 element binds a cellular factor whose activitycan be directly or indirectly modulated by the adenovirusElA gene products.The MLTF binding site and the R1-R2 element both

contribute to the activity of the P5 promoter in the presenceof ElA products. Deletion of the MLTF binding site reducedP5 activity by a factor of about 3 (Fig. 3B, pAAVP5-CAT216), and simultaneous deletion of the R1-R2 repeatfurther reduced P5 activity by a factor of about 25 (Fig. 3B,pAAVP5-CAT243). Deletion of both elements generated P5derivatives with very low basal activity which did notrespond to ElA products.

In addition to its effect on induced expression, deletion ofthe MLTF binding site alone reproducibly led to an approx-

imately threefold increase in the basal activity of the P5promoter (Fig. 3B, pAAVP5-CAT216). Apparently, theMLTF binding site leads to a reduction in uninduced basalactivity of P5. The MLTF is believed to enhance the basalactivity of other control regions that contain its binding site

(discussed in references 7 and 36). although it did not do sowhen inserted upstream of the SV40 TATA motif in pBC19-6and pBC19-11 (Table 2). The negative effect of the MLTFbinding site on basal P5 expression likely derives from thecontext in which it resides within the PS promoter. Perhapsthe MLTF and the factor that binds the R1-R2 elementinterfere with each other's binding or activity in the absenceof EIA products. The trans-activating proteins could thenstimulate the P5 promoter by directly or indirectly alteringone or both of the factors so that they no longer interfere butcooperate to induce transcription.

It is clear that the combination of the MLTF binding siteand R1-R2 repeats influences both basal and induced P5activity. The close apposition of the two binding sites hasapparently generated a transcriptional control region thatdisplays a greater induction in response to EIA productsthan either site alone could generate. This has been achievedby combining an inhibitory effect on basal activity withstimulatory effects achieved in the presence of adenovirusElA gene products.

ACKNOWLEDGMENTS

We thank L. Kedes, M. Labow, B. Thimmappaya. and C.Tibbetts for various plasmids; M. Flocco and J. Song-Nichols forsynthesizing DNAs: and D. Engel and U. Mueller for critical readingof the manuscript.

This work was supported by Public Health Service grant CA38965from the National Cancer Institute. L.-S. Chang was a fellowshiprecipient from the Leukemia Society of America and T. Shenk wasan American Cancei- Society Research Professor during the initialphase of this work.

LITERATURE CITED

1. Alwine, J. C. 1985. Transient gene expression control: effects oftransfected DNA stability and trans-activation by viral earlyproteins. Mol. Cell. Biol. 5:1034-10)42.

2. Berk, A. J. 1986. Adenovirus promoters and EIA transactiva-tion. Annu. Rev. Genet. 20:45-79.

3. Berns, K. I., and R. A. Bohenzky. 1987. Adeno-associatedviruses: an update. Adv. Virus Res. 32:243-306.

4. Brunet, L. J., and A. J. Berk. 1988. Concentration dependenceof transcriptional activation in inducible ElA-containing humancells. Mol. Cell. Biol. 8:4799-4807.

5. Buller, R. M. L., and J. A. Rose. 1978. Characterization ofadeno-associated virus polypeptides synthesized in viha and iniiruo. p. 394-410. In D. Ward and P. Tattersall (ed.), Replicationof mammalian parvoviruses. Cold Spring Harbor Laboratory,Cold Spring Harbor, N.Y.

6. Carthew, R. W., L. A. Chodosh, and P. A. Sharp. 1987. Themajor late transcription factor binds to and activates the mousemetallothionein I promoter. Genes Dev. 1:973-980.

7. Chodosh, L. A., R. W. Carthew, J. G. Morgan, G. R. Crabtree,and P. A. Sharp. 1987. The adenovirus major late transcriptionfactor activates the rat -y-fibrinogen promoter. Science 238:684-688.

8. Chodosh, L. A., R. W. Carthew, and P. A. Sharp. 1986. A singlepolypeptide possesses the binding and transcription activities ofthe adenovirus major late transcription factor. Mol. Cell. Biol.6:4723-4733.

9. Dignam, J. D., R. M. Lebovitz, and R. G. Roeder. 1983.Accurate transcription initiation by RNA polymerasell in asoluble extract from isolated mammalian nuclei. Nucleic AcidsRes. 11:1475-1489.

10. Fried, M., and D. M. Crothers. 1981. Equilibria and kinetics oflac repressor-operator interactions by polyacrylamide gel elec-trophoresis. Nucleic Acids Res. 9:6505-6525.

11. Galas, D. J., and A. Schmitz. 1978. DNase footprinting: a simple

J. VIROL.

AAV PS PROMOTER 3487

method for the detection of protein-DNA binding specificity.Nucleic Acids Res. 5:3157-3170.

12. Gorman, C. M., L. F. Moffat, and B. H. Howard. 1982.Recombinant genomes which express chloramphenicol acetyl-transferase in mammalian cells. Mol. Cell. Biol. 2:1044-1051.

13. Graham, F., and A. J. van der Eb. 1973. A new technique for theassay of infectivity of human adenovirus S DNA. Virology52:456-457.

14. Graham, F. L., J. Smiley, W. C. Russell, and R. Nairu. 1977.Characteristics of a human cell line transformed by DNA fromhuman adenovirus type 5. J. Gen. Virol. 36:59-72.

15. Green, M. R., and R. G. Roeder. 1980. Definition of a novelpromoter for the major adenovirus-associated virus mRNA.Cell 22:231-242.

16. Green, M. R., and R. G. Roeder. 1980. Transcripts of theadeno-associated virus genome: mapping of the major RNAs. J.Virol. 36:79-92.

17. Green, M. R., S. E. Straus, and R. G. Roeder. 1980. Transcriptsof the adeno-associated virus genome: multiple polyadenylatedRNAs including a potential primar-y transcript. J. Virol. 35:56(-565.

18. Groudine, M., M. Peretz, and H. Weintraub. 1981. Trainscrip-tional regulation of hemoglobin switching in chicken embryos.Mol. Cell. Biol. 1:281-288.

19. Gunning, P., P. Ponte, H. Okayama, J. Engel, H. Blau, and L.Kedes. 1983. Isolation and characterization of full-length cDNAclones for human (x-. [-. and -y-actin mRNAs: skeletal but notcytoplasmic actins have an amino-terminal cysteine that issubsequently removed. Mol. Cell. Biol. 3:787-795.

20. Hardy, S., and T. Shenk. 1988. Adenovirus ElA gene productsand adenosine 3'. 5'-cyclic monophosphate activate transcrip-tion through a common factor. Proc. Ncatl. Acad. Sci. USA85:4171-4175.

21. Hearing, P., and T. Shenk. 1983. The adenovirus type S EIAtranscriptional control region contains a duplicated enhancerelement. Cell 33:695-703.

22. Hen, R., E. Borelli, P. Sassone-Corsi, and P. Chambon. 1983. Anenhancer element is located 340 base pairs upstream from theadenovirus 2 ElA cap site. Nucleic Acids Res. 11:8747-8749.

23. Herbomel, P., B. Bourachot, and M. Yaniv. 1984. Two distinctenhancers with different cell specificities coexist in the regula-tory region of polyoma. Cell 39:653-662.

24. Hirt, B. 1967. Selective extraction of polyoma DNA frominfected mouse cell cultures. J. Mol. Biol. 26:365-369.

25. Hofer, E., and J. E. Darnell. 1981. The primary trianscriptionunit of the mouse [3-major globin gene. Cell 23:585-593.

26. Janik, J. E., M. M. Huston, and J. A. Rose. 1981. Locations ofadenovirus genes required for the replication of adenovirus-associated virus. Proc. Natl. Acad. Sci. USA 78:1925-1929.

27. Jochemsen, A. G., L. T. C. Peltenburg, M. F. W. tePas, C. M.deWit, J. L. Bos, and A. J. van der Eb. 1987. Activation ofadenovirus 5 ElA transcription by region E1B in transformedprimary rat cells. EMBO J. 6:3399-3405.

28. Johnson, F. B., T. A. Thompson, P. A. Taylor, and D. A. Vlazny.1977. Molecular similarities among the adenovirus-associatedvirus polypeptides and evidence for a precursor protein. Virol-ogy 82:1-13.

29. Jones, N., and T. Shenk. 1979. Isolation of adenovirus type 5host range deletion mutants defective for- transformation of ratembryo cells. Cell 17:683-689.

30. Jones, S. N., and C. Tibbetts. 1989. Upstream DNA sequencesdetermine different autoregulatory responses of the adenovirustypes 5 and 3 ElA promoters. J. Virol. 63:1833-1838.

31. Kovesdi, I., R. Reichel, and J. R. Nevins. 1987. Role ofan adenovirus E2 promoter binding faictor in ElA-medicatedcoordinate gene control. Proc. Natl. Acad. Sci. USA 84:2180-2184.

32. Laimins, L. A., G. Khoury, C. Gorman, B. Howard, and P.Gruss. 1982. Host-specific activation of transcription by tandemrepeats from simian virus 40 and Moloney murine sarcomavirus. Proc. Natl. Acad. Sci. USA 79:6453-6457.

33. Laughlin, C. A., N. Jones, and B. J. Carter. 1982. Effect ofdeletions in adenovirus early region 1 genes upon replication of

adeno-associated virus. J. Virol. 41:868-876.34. Laughlin, C. A., H. J. Westphal, and B. J. Carter. 1979. Spliced

adenovirus-associated virus RNA. Proc. Natl. Acad. Sci. USA76:5567-5571.

35. Lee, K. A. W., and M. R. Green. 1987. A cellular transcriptionfactor E4F1 interacts with an ElA-inducible enhancer andmediates constitutive enhancer function in vitro. EMBO J.6:1345-1353.

36. Leong, K., L. Brunet, and A. J. Berk. 1988. Factors responsiblefor the higher- transcriptional activity of extracts of adenovirus-infected cells fractionate with the TATA box transcriptionfactor. Mol. Cell. Biol. 8:1765-1774.

37. Lusby, E. W., and K. I. Berns. 1982. Mapping of the 5' terminiof two adeno-associated virus 2 RNAs in the left half of thegenome. J. Virol. 41:518-526.

38. Maniatis, T., E. F. Fritsch, and J. Sambrook. 1982. Molecularcloning: a laboratory manual. Cold Spring Harbor Laboratory.Cold Spring Har-bor. N.Y.

39. Marcus, C. J., C. A. Laughlin, and B. J. Carter. 1981. Adeno-associated virus RNA transcription in vivo. Eur. J. Biochem.121:147-154.

40. McCutchan, J. H., and J. S. Pagano. 1968. Enhancement of theinfectivity of simian virus 40 deoxyribonucleic acid with dieth-ylaminoethyl-dextran. J. Natl. Cancer Inst. 41:351-357.

41. Melton, D. A., P. A. Krieg, M. R. Rebagliati, T. Maniatis, K.Zinn, and M. R. Green. 1984. Efficient iti vitro synthesis ofbiologically aictive RNA and RNA hybridization probes fromplasmids containing a bacteriophage PS6 promoter. NucleicAcids Res. 12:7035-7056.

42. Mendelson, E., J. P. Trempe, and B. J. Carter. 1986. Identifi-cation of the trans-acting Rep proteins of adeno-associated virusby antibodies to a synthetic oligopeptide. J. Virol. 60:823-832.

43. Moran, E., and M. B. Mathews. 1987. Multiple functionaldomains in the adenovirus ElA gene. Cell 48:177-178.

44. NMulligan, R. C., B. H. Howard, and P. Berg. 1979. Synthesis ofrabbit [-globin in cultured monkey kidney cells following infec-tion with a SV40 [-globin recombinant genome. Nature 277:108-114.

45. Murthy, S. C. S., G. P. Bhat, and B. Thimmappaya. 1985.Adenovirus EIIA early promoter: transcriptional control ele-ments and induction by the viral pre-early ElA gene, whichappears to be sequence independent. Proc. Natl. Acad. Sci.USA 82:2230-2234.

46. Pilder, A., NI. Moore, J. Logan, and T. Shenk. 1986. Theadenovirus E1B-55K transforming polypeptide modulates trans-port or cytoplasmic stabilization of viral and host cell mRNAs.Mol. Cell. Biol. 6:470-476.

47. Richardson, XW. D., and H. Westphal. 1984. Requirement foreither early region la or early region lb adenovirus geneproducts in the helper effect for adeno-associated virus. J. Virol.51:404-410.

48. Rose, J. A., J. V. Maizel, Jr., J. K. Inman, and A. J. Shatkin.1971. Structural proteins of adenovirus-associated viruses. J.Virol. 8:766-770.

49. Samulski, R. J., L.-S. Chang, and T. Shenk. 1987. A recombi-nant plasmid from which an infectious adeno-associated virusgenome can be excised in vivo and its use to study viralreplication. J. Virol. 61:3096-3101.

50. Samulski, R. J., A. Srivastava, K. I. Berns, and N. Muzvczka.1983. Rescue of adeno-associated virus from recombinant plas-mids: gene corr-ection within the terminal repeats of AAV. Cell33:135-143.

51. Sanger, F., S. Nicklen, and A. R. Coulson. 1977. DNA sequenc-ing with chain terminating inhibitors. Proc. NatI. Acad. Sci.USA 74:5463-5467.

52. Sawadogo, M., M. W. Van Dyke, P. D. Gregor, and R. G.Roeder. 1988. Multiple forms of the human gene-specific tran-

scription factor USF. I. Complete purification and identifica-tion of USF from HeLa cell nuclei. J. Biol. Chem. 263:11985-11993.

53. Senapathy, P., J.-D. Tratschin, and B. J. Carter. 1984. Replica-tion of adeno-associated virus DNA: complementation of natu-

Vol. 63. 1989

3488 CHANG ET AL.

rally occurring rep-mutants by a wild-type genome or an ori-mutant and correction of terminal palindrome deletions. J. Mol.Biol. 179:1-20.

54. Simon, M. C., T. M. Fisch, B. J. Benecke, J. R. Nevins, and N.Heintz. 1988. Definition of multiple, functionally distinct TATAelements, one of which is a target in the lisp 70 promoter forEIA regulation. Cell 52:723-729.

55. Southern, E. M. 1975. Detection of specific sequences amongDNA fragments separated by gel electrophoresis. J. Mol. Biol.98:503-517.

56. Southern, P. J., B. H. Howard, and P. Berg. 1981. Constructionand characterization of SV40 recombinants with beta-globincDNA substituting in their early regions. J. Mol. Appl. Genet.1:177-190.

57. Srivastava, A., E. W. Lusby, and K. I. Berns. 1983. Nucleotidesequence and organization of the adeno-associated virus 2

genome. J. Virol. 45:555-564.58. Tratschin, J.-D., M. H. P. West, J. Sandbank, and B. J. Carter.

1984. A human parvovirus, adeno-associated virus, as a eucary-otic vector: transient expression and encapsidation of the pro-caryotic gene for chloramphenical acetyltransferase. Mol. Cell.Biol. 4:2072-2081.

59. Wu, L., D. S. E. Rosser, M. G. Schmidt, and A. J. Berk. 1987.A TATA box implicated in ElA transcriptional activation of a

simple adenovirus 2 promoter. Nature (London) 326:512-515.60. Yakobson, B., T. Koch, and E. Winocour. 1987. Replication of

adeno-associated virus in synchronized cells without the addi-tion of a helper virus. J. Virol. 61:972-981.

J. VIROI.