Biochemistry. In the article “The TRAP220 component of athyroid hormone receptor-associated protein (TRAP) coac-tivator complex interacts directly with nuclear receptors in aligand-dependent fashion” by Chao-Xing Yuan, Mitsuhiro Ito,Joseph D. Fondell, Zheng-Yuan Fu, and Robert G. Roeder,which appeared in number 14, July 7, 1998, of Proc. Natl. Acad.Sci. USA (95, 7939–7944), the authors wish to acknowledge anearlier paper that had escaped their attention. In an articleentitled “Identification of RB18A, a 205 kDa new p53 regu-latory protein which shares antigenic and functional propertieswith p53” [Drané, P., Barel, M., Balbo, M. & Frade, R. (1997)Oncogene 15, 3013–3024], Drané et al. report the identificationof a human protein, RB18A, that interacts with several anti-p53 monoclonal antibodies, shows general DNA binding prop-erties, and stimulates p53 binding to DNA. The cDNA-derivedprotein sequences of RB18A and TRAP220 are nearly iden-tical, there being minor sequence variations and an extendedN terminus on TRAP220. Apart from the effect of RB18A onp53 binding to DNA, the study of Drané et al. did not reportany additional functions of RB18A and, in contrast to thefindings with TRAP220, provided no indications of its pres-ence within a larger multiprotein coactivator complex.

Biochemistry. In the article “Molecular cloning and charac-terization of a cellular phosphoprotein that interacts with aconserved C-terminal domain of adenovirus E1A involved innegative modulation of oncogenic transformation” by UteSchaeper, Janice M. Boyd, Sulekha Verma, Erik Uhlmann,T. Subramanian, and G. Chinnadurai, which appeared innumber 23, November 7, 1995 of Proc. Natl. Acad. Sci. USA(92, 10467–10471), we reported the sequences for humanCtBP. Reexamination of the cDNA sequences revealed certainerrors. These errors have been corrected in the GenBankdatabase (accession no. U37408).

Evolution. In the article “The role of robustness and change-ability on the origin and evolution of genetic codes” by TetsuyaMaeshiro and Masayuki Kumura, which appeared in number9, April 28, 1998, of Proc. Natl. Acad. Sci. USA (95, 5088–5093), the authors wish to note the following error in Table 2related to the initiation codons of nuclear mycoplasma. Theyshould read: AUG, GUG, and UUG [Dybvig, K. & Voelker,L. L. (1996) Annu. Rev. Microbiol. 50, 25–57]. Consequently,CMy (page 5091, last paragraph, right column) should bedeleted in the text. These corrections do not change the resultsof the paper. A corrected Table 2 is shown below.

Table 2. Assignments of deviant codons

Representative genetic system Code

Changes from SGC

Initiation codonsCodon Phenotype

Mitochondrial yeasts MYe UGA stopf Trp AUG 1AUA IlefMetCUN Leuf Thr

Mitochondrial platyhelminths MPl UGA stopf Trp AUG 1AAA Lysf AsnAGR Argf SerUAA stopf Tyr

Mitochondrial nematoda MNe UGA stopf Trp AUN UUG GUG 6arthropoda AGR Argf Sermollusca AUA IlefMet

Mitochondrial echinodermata MEc UGA stopf Trp AUG 1AAA Lysf AsnAGR Argf Ser

Mitochondrial tunicata MTu UGA stopf Trp AUG 1AUA IlefMetAGR Argf Gly

Mitochondrial vertebrata MVe UGA stopf Trp AUN GUG 5AUA IlefMetAGR Argf stop

Mitochondrial euascomycetes MEu UGA stopf Trp AUN NUG UUA 8Nuclear mycoplasma CMy UGA stopf Trp AUG GUG UUG 3Nuclear euplotes CEu UGA stopf Cys AUG 1Nuclear acetabularia CAc UAR stopf Gln AUG 1Nuclear blepharisma CBl UAG stopf Gln AUG 1Nuclear candida CCa CUG Leuf Ser AUG CUG 2Nuclear bacterial CBa — — AUN NUG 7

N denotes any of A, U, G, and C, and R denotes A and G. The values in the initiation codons indicate the number of known initiation codons.The codon reassignments of each deviant code are arranged from top to bottom in the estimated order of reassignments. Compiled fromhttp:yywww3.ncbi.nlm.nih.govyhtbin-postyTaxonomyywprintgc?mode5c.

14584 Corrections Proc. Natl. Acad. Sci. USA 95 (1998)

Dow

nloa

ded

by g

uest

on

June

8, 2

021

Dow

nloa

ded

by g

uest

on

June

8, 2

021

Dow

nloa

ded

by g

uest

on

June

8, 2

021

Dow

nloa

ded

by g

uest

on

June

8, 2

021

Dow

nloa

ded

by g

uest

on

June

8, 2

021

Dow

nloa

ded

by g

uest

on

June

8, 2

021

Dow

nloa

ded

by g

uest

on

June

8, 2

021

Plant Biology. In the article “Differential expression of twoisopentenyl pyrophosphate isomerases and enhanced carot-enoid accumulation in a unicellular chlorophyte” by ZairenSun, Francis X. Cunningham, Jr., and Elisabeth Gantt, whichappeared in number 19, September 15, 1998, of Proc. Natl.

Acad. Sci. USA (95, 11482–11488), the following correc-tion should be noted. In Fig. 3 the lightly shaded sequences,representing amino acid identity of four of five se-quences, was inadvertently lost in the electronic submissionprocess.

FIG. 3. Amino acid sequence alignment of IPP isomerases predicted by cDNAs of Haematococcus pluvialis (IPIHp1 and IPIHp2) andChlamydomonas reinhardtii (IPICr1) are compared with sequences from the flowering plants Clarkia brewerii (IPICb2) (29) and Arabidopsis thaliana(IPIAt1) (30). The sequences were aligned using the program CLUSTALW (http:yydot.imgen.bcm.tmc.edu:9331ymultialign) and shaded usingGENEDOC (http:yywww.concentric.nety;Ketchupygddl.htm). The dark shading with white letters indicates amino acid identity for the alignedresidue for all five proteins whereas the lightly shaded sequences represent amino acid identity for four of five sequences. The lines above thesequences indicate differences in the predicted amino acid sequences of IPIHp1 and IPIHp2. The inverted arrowhead () above IPIHp1 designatesthe location of a truncation from the N terminus. Residues required for catalytic activity (31) are marked by upright arrowheads (Œ).

Corrections Proc. Natl. Acad. Sci. USA 95 (1998) 14585

Dow

nloa

ded

by g

uest

on

June

8, 2

021

Proc. Natl. Acad. Sci. USAVol. 92, pp. 10467-10471, November 1995Biochemistry

Molecular cloning and characterization of a cellularphosphoprotein that interacts with a conserved C-terminaldomain of adenovirus ElA involved in negative modulation ofoncogenic transformation

(two-hybrid analysis/tumorigenesis/dehydrogenase)

UTE SCHAEPER, JANICE M. BOYD, SULEKHA VERMA, ERIK UHLMANN, T. SUBRAMANIAN, AND G. CHINNADURAI*Institute for Molecular Virology, St. Louis University Medical Center, 3681 Park Avenue, St. Louis, MO 63110

Communicated by William S. Sly, St. Louis University School of Medicine, St. Louis, MO, July 19, 1995 (received for review April 21, 1995)

ABSTRACT The adenovirus type 2/5 ElA proteins trans-form primary baby rat kidney (BRK) cells in cooperation withthe activated Ras (T24 ras) oncoprotein. The N-terminal halfof ElA (exon 1) is essential for this transformation activity.While the C-terminal half of ElA (exon 2) is dispensable, aregion located between residues 225 and 238 of the 243R EIAprotein negatively modulates in vitro T24 ras cooperativetransformation as well as the tumorigenic potential ofE1A/T24 ras-transformed cells. The same C-terminal domainis also required for binding of a cellular 48-kDa phosphopro-tein, C-terminal binding protein (CtBP). We have cloned thecDNA for CtBP via yeast two-hybrid interaction cloning. ThecDNA encodes a 439-amino acid (48 kDa) protein that spe-cifically interacts with exon 2 in yeast two-hybrid, in vitroprotein binding, and in vivo coimmunoprecipitation analyses.This protein requires residues 225-238 of the 243R ElAprotein for interaction. The predicted protein sequence of theisolated cDNA is identical to amino acid sequences obtainedfrom peptides prepared from biochemically purified CtBP.Fine mapping of the CtBP-binding domain revealed that a6-amino acid motif highly conserved among the ElA proteinsofvarious human and animal adenoviruses is required for thisinteraction. These results suggest that interaction of CtBPwith the ElA proteins may play a critical role in adenovirusreplication and oncogenic transformation.

certain genes. Thus, the transforming activities of exon 1appear to be linked to interactions with cellular proteins andthe resulting regulation of transcription.Although the functions of exon 2 have been studied less

intensively, it has been implicated in certain positive andnegative transcriptional regulatory activities (9-11). Exon 2 isrequired for immortalization (12, 13) and induction of Ad2/5-specific cytotoxic lymphocytes (14). In addition, exon 2influences the extent of oncogenic transformation. Deletionswithin the C-terminal 67 aa of the ElA 243R protein enhanceE1A/T24 ras cooperative transformation (12, 15), and tumor-igenesis of transformed cells in syngeneic and athymic rodentmodels (12). Importantly, exon 2 also plays a role in tumormetastasis. Expression of wild-type (wt) ElA efficiently sup-presses the metastatic potential of tumor cells (16-18). Incontrast, cells expressing ElA proteins lacking the C-terminal67 aa are highly metastatic (9, 12). Thus, exon 2 appears tonegatively modulate in vitro transformation, tumorigenesis,and metastasis. We have localized these activities of exon 2within a 14-aa region (residues 225-238) near the C terminusof the 243R protein (19). These transformation restrainingactivities of the C-terminal region of ElA correlate with theinteraction of a 48-kDa cellular phosphoprotein termed C-terminal binding protein (CtBP) (19). Here we report on themolecular cloning and biochemical characterization of CtBP.

The Ela region of adenovirus types 2 and 5 (Ad2/5) encodestwo major proteins of 289 and 243 aa (289R and 243R). Bothproteins contain two exons and are identical except for thepresence of an internal 46-aa region unique to the 289Rprotein. While the 289R protein is required for productive viralinfection, the 243R protein encodes all the functions necessaryfor immortalization of primary cells and for transformation ofthese cells in cooperation with other viral or cellular oncogenes(1). Exon 1 of the ElA proteins is essential for these trans-forming activities and controls cell proliferation and transfor-mation by modulating gene expression through interactionwith several cellular proteins. One of the functional domainsof exon 1 encompasses two regions, conserved region (CR) 1and CR2, that are conserved among different Ad strains.These regions are responsible for interactions between ElAand the cellular proteins pRb, p107, and p130, which causethese cellular proteins to release the E2F transcription factor,thus activating gene expression [reviewed by Dyson and Har-low (2), Nevins (3), Moran (4) and Bayley and Mymryk (5)].A second functional domain, encompassing the CR1 and theN terminus of ElA, interacts with a transcriptional adapterp300 (4, 6-8) implicated in transcriptional repression of

MATERIALS AND METHODSPlasmids. Plasmids encoding fusion proteins consisting of

the Gal4 DNA-binding domain (aa 1-147) and the entiresecond exon of the ElA 243R protein (aa 141-243) or a smallerC-terminal region (aa 176-243) were constructed in the yeastshuttle vector pMA424 (20) or pAS1 (21). The exon 2 mutantsd1181-193 (19) and d11133 to d11136 (8) were cloned intopMA424 in a similar fashion. Plasmids pGST-Cter, pGST-Cter(dI181-193), and pGSTdlll33 to d11136 have been described(19). Plasmids pTM1-30, pGEX-30, and pET21-30 were con-structed by cloning the protein coding sequences (amplified byPCR) of the cDNA clone pAct3O in vectors pTM1 (22),pGEX-5X3 (Pharmacia), and pET2lb (Novagen). PlasmidpCMV-T7-30 was constructed by cloning the T7-tagged cDNAfrom pET21b-30 in an expression vector pCMV (L. K. Ven-katesh and G.C., unpublished data). Plasmid pRcCMV-T7-30was constructed by subcloning a fragment (SnaBI/HindIII)containing a portion of the cytomegalovirus (CMV) promoterand T7-tagged cDNA from pCMV-T7-30 into pRcCMV (In-vitrogen). The 243R substitution mutants APL, ADL, and ASC

Abbreviations: CtBP, C-terminal binding protein; Ad, adenovirus;GST, glutathione S-transferase; GST-Cter, ElA-GST fusion proteincontaining C-terminal 68 aa of ElA; mAb, monoclonal antibody;ORF, open reading frame.*To whom reprint requests should be addressed.

10467

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement" inaccordance with 18 U.S.C. §1734 solely to indicate this fact.

10468 Biochemistry: Schaeper et al.

(see Fig. 5B) were constructed by PCR with primers designedto substitute Ala-Ser residues for the two amino acids targetedfor mutation. The coding sequences of E1A 243R, dl1135, andthe substitution mutants APL, ADL, and ASC were cloned intopTM1 (22). The coding sequence of Adl2 234R (12S) wassubcloned into pcDNA3 (Invitrogen).Two-Hybrid Screening. The yeast two-hybrid screen was

carried out essentially as described by Chien et al. (23) andDurfee et al. (21). The yeast strain GGY1::171 (24) wascotransfected with pMA-Exon 2 and a human (B cell) cDNAlibrary tagged with the Gal4 activation domain (21). Positivelyinteracting cDNA clones were further screened with a batteryof 10 different heterologous protein baits.

In Vitro and in Vivo Protein Interactions. In vitro proteininteractions using 35S-labeled CtBP or ElA proteins with theindicated glutathione S-transferase (GST) fusion proteinswere carried out as described by Boyd et al. (19). For immu-noprecipitation analysis, CtBP and ElA proteins were ex-pressed with the vaccinia virus/T7 RNA polymerase systemdescribed by Ausubel et al. (25). BSC40 cells were infected withvTF7-3 recombinant vaccinia virus that expresses T7 RNApolymerase (26) and then cotransfected with the plasmidspTM1-243R or pTM1-dll135 and pTM1-30 using Lipo-fectAMINE (GIBCO/BRL). Cells were labeled with a[35S]methionine/cysteine mixture (500 ACi per 75-cm2 flask; 1Ci = 37 GBq) and subjected to immunoprecipitation (19).

Phosphorylation. HeLa cells were LipofectAMINE-transfected with the plasmid pRcCMV-T7-30, which expressesT7 epitope-tagged CtBP. Forty-eight hours after transfection,cells were labeled for 20 min with 4 mCi of H332PO4 or with[35S]methionine/cysteine, lysed, and subjected to immunopre-cipitation using the T7 monoclonal antibody (mAb) (Novagen)or control mAb, pAb416 (Oncogene Science).

Purification and Peptide Sequence Analysis of EndogenousCtBP. Approximately 2.5 x 1011 HeLa cells were lysed in ElAlysis buffer. The lysate was clarified by centrifugation andpreincubated with 20 mg of GST immobilized on glutathioneagarose and then incubated with 1 ml of GST-Cter (ElA-GSTfusion protein containing C-terminal 68 aa of E1A) beadscontaining 5 mg of protein. The beads were washed five timeswith ElA lysis buffer. The bound protein was separated bySDS/PAGE and blotted onto nitrocellulose. Bands stainedwith 1% amido black were excised from the membrane and,after in situ tryptic digestion, peptides were separated byreverse-phase HPLC and sequenced in an ABI protein se-quencer (model 477A) (gas phase sequenator). These serviceswere provided by the Harvard Microchemistry Facility (W.Lane, Director).

RESULTS

Cloning of cDNA for Exon 2 Binding Protein. We used theyeast two-hybrid system (23, 27) to identify cDNAs for cellularproteins that interact with exon 2 of Ad2 ElA. The two-hybridscreen was carried out using yeast strain GGY1::171 (20),which contains the lacZ reporter gene under the control of theGALl promoter (24). GGY1::171 cells were transformed witha bait plasmid (pMA-exon 2) expressing Ad2 exon 2 (residues141-243 of 243R) fused to the DNA binding domain of Gal4(residues 1-147) in plasmid pMA424 (20) and a Gal4 activationdomain tagged B-cell cDNA library (21). Approximately 107transformants were screened and a single clone (pAct30) thatwas strongly positive for interaction with the exon 2 bait wasisolated. pAct3O specifically interacted only with the exon 2bait and not with any of 10 other heterologous baits (data notshown). The interaction was further confirmed by using adifferent indicator strain, Y153 (21). To identify the regionwithin exon 2 that is required for interaction with the proteinencoded by pAct3O, two-hybrid interaction studies were per-formed with plasmids expressing the Gal4 (residues 1-147)-

exon 2 fusion proteins containing various ElA mutants (11,19). The interactions were analyzed both qualitatively andquantitatively (Fig. 1). The protein encoded by pAct3O inter-acted positively with all exon 2 mutants except dl1135 (residues225-238; see Fig. 5B). We have previously reported that dll 135is also defective in interaction with CtBP (19). Thus, theprotein encoded by pAct3O interacts with the ElA 243Rprotein within the same region (residues 225-238) required forCtBP interaction.

In Vitro and in Vivo Protein Interactions. To confirm theresults obtained in the two-hybrid interaction studies, wecarried out a series of in vitro and in vivo protein interactionstudies. The coding sequence of pAct3O (lacking the Gal4activation domain) was subcloned in a plasmid vector, pET21b,under transcriptional control of the bacteriophage T7 pro-moter. In vitro transcription and translation from the pET21bplasmid containing pAct3O sequences (pET21b-30) produceda 48-kDa protein (hereafter designated 48 kDa). The interac-tion between this 48-kDa protein and exon 2 was first analyzedby an in vitro protein binding assay using GST-Cter (19).35S-labeled 48-kDa protein was assayed for binding to variousGST-Cter mutant fusion proteins immobilized on GST affin-ity matrix followed by SDS/PAGE (Fig. 2A). The 48-kDaprotein did not bind significantly to either GST or GST-dl135protein but it did interact with GST-Cter wt and various otherGST-Cter mutant proteins. In a converse experiment, bindingof 35S-labeled full-length 243R wt or d11135 protein to aGST-48-kDa (GST-30) fusion protein was also carried out(Fig. 2B). As expected, the wt 243R protein did bind to theGST-48-kDa protein, while there was no detectable binding ofthe d11135 protein. These in vitro binding studies indicate that

A 141

pMA-Ex2 G

pMA-Cter GaI4

243

Exon2176 243

C-terminus

I' II,. . 1--L '--d1181-193 dll1134 dl1136(181-193) (209-224) (239-243)

dll 133 dll135(193-208) (225-238)

B

Proc. Natl. Acad. Sci. USA 92 (1995) 10469

AIP-4 en IT tn z_._I

5- rJi C,E la mI £co

* X 4 E _4

ct ct u u: u: u

Anti #30 M58I 1--I

kn t

+ + + +o o 0 0us, ~s us, la

Cu0

co_._

5-4I-'I

u)

en

E1A[ =:

FIG. 2. In vitro interaction of 48-kDa protein with ElA 243Rmutants. (A) Binding of 48-kDa protein to GST-Cter and variousGST-Cter mutants of ElA (see Fig. 1A). (B) Binding of full-lengthElA (243R and d11135) to GST-48-kDa protein (GST-30). Proteinseluted from GST beads were analyzed on SDS/10% polyacrylamidegel.

the 48-kDa protein encoded by clone 30 interacts with the Cterminus or full-length 243R protein and not with an ElA243R mutant protein (dI1135) lacking the previously charac-terized CtBP-interacting domain [residues 229-238 (19)].The interaction between the 48-kDa protein and ElA 243R

protein was also examined in an in vivo coimmunoprecipitationstudy. The 48-kDa protein (clone 30) was coexpressed with the243R wt or d11135 protein in BSC40 cells using the vacciniavirus/T7 expression system (22). Cells were in vivo labeled with[35S]methionine/cysteine mixture, and cell lysates were sub-jected to immunoprecipitation with rabbit polyclonal anti-serum prepared against the GST-48-kDa fusion protein (anti-clone 30) or ElA mAb M58 (28). The wt 243R proteincoprecipitated with the 48-kDa protein, while the mutantdl1l35 protein did not (Fig. 3). Conversely, the M58 (ElA)antibody coimmunoprecipitated the 48-kDa protein from ex-tracts containing 48-kDa and 243R wt proteins and not fromcells containing the 48-kDa and d11135 proteins. These resultsare in very good agreement with the in vitro binding studies andindicate that the 48-kDa protein encoded by cDNA clone 30interacts specifically within a region encompassing residues225-238 in the manner previously reported for CtBP.

Phosphorylation of 48-kDa Protein. We tested the 48-kDaprotein to determine whether it, like CtBP, is phosphorylatedin vivo. HeLa cells were transfected with the plasmid pRc-CMV-T7-30, which expresses the 48-kDa protein tagged (atthe N terminus) with an 1 1-aa T7 epitope under control of theCMV promoter. Cells were then labeled with [35S]methionine/cysteine or 32p. The protein extracts were immunoprecipitatedwith a mAb directed against the T7 tag. The 32P-labeled48-kDa protein was precipitated (Fig. 4), indicating that the48-kDa protein encoded by clone 30 is a phosphoprotein likeCtBP.

#30- M.._._:

ElA __

FIG. 3. Coimmunoprecipitation of 48-kDa protein and ElA. The48-kDa protein (#30) and ElA proteins (243R and d11135) wereexpressed in BSC40 cells using the recombinant vaccinia virus expres-sion system. Proteins were immunoprecipitated with ElA-specificantibody M58 or the antiserum raised against the 48-kDa protein (Anti#30) and analyzed on SDS/10% polyacrylamide gel.

Binding Site for the 48-kDa Protein. The CtBP bindingdomain ofElA is located in a 9-aa region between residues 229and 238 (19). Comparative analysis revealed that a 6-aa motif(PLDLSC; aa 233-238) is relatively well-conserved among Adserotypes (29). To determine whether these conserved aminoacid residues constitute the binding site for the 48-kDa protein,we constructed three 2-aa substitution mutants (APL,PL--AS; ADL, DL->AS; ASC, SC-->AS; Fig. SB) and ana-lyzed them for in vitro binding activities. 35S-labeled mutantElA proteins were prepared by coupled in vitro transcriptionand translation and tested for binding to GST-48 kDa (GST-30) protein (Fig. 5A). No significant binding was observed forthe substitution mutants APL and ADL, comparable to the14-aa deletion mutant d11135. The mutant ASC did bind to the48-kDa protein but at a very reduced level compared to 243Rwt. It therefore appears that the 6-aa region analyzed(PLDLSC) is important for efficient binding of the 48-kDaprotein. The mutation ASC (SC->AS) results in retention of aserine residue, albeit in a different position, which may explainwhy this mutant still retains some in vitro binding activity.Alternatively, the PLDL region may constitute the core bind-

35S 32p

:'i.ii*-'.C.

* ..-,..',. I.

48 kD- * ] CtBP

FIG. 4. Phosphorylation of 48-kDa protein. HeLa cells were trans-fected with the plasmid pRcCMV-T7-30 (to express T7 epitope-tagged48-kDa protein) and labeled with [35S]methionine/cysteine or 32p.Immunoprecipitations were carried out with the T7 antibody orcontrol mAb. As a marker, 48-kDa protein was also purified onGST-Cter beads by affinity chromatography. Proteins were analyzedon SDS/10% polyacrylamide gel.

48 kDa

B

243R d11135

0

Biochemistry: Schaeper et al.

10470 Biochemistry: Schaeper et al.

AXin~~~~~~~~~Ct n > 3> Translation mix

( iC

UQU On Q : = z C : V

a__

B 2_2__ __ _ _ _ _ _ d11135 (Ads)PPL _DL SC

Ad2 243R ..ciedllnepgqPLISC rprpAS AS AS

AdM2 235R ...iqeeereqtvPVDLSVkrprcn23SFIG. 5. Mapping of binding site for 48-kDa protein. (A) Binding of

ElA mutants to GST-48-kDa (GST-30) protein. The 35S-labeled ElAproteins were incubated with GST or GST-48-kDa (GST-30) protein.Bound proteins were eluted and analyzed on an SDS/8% polyacryl-amide gel. 12H/234R refers to wt Adl2 ElA protein. This proteinmigrated in an anomalous fashion in SDS/PAGE. (B) C-terminalsequences of Ad2 243R and wt Adl2 235R proteins that were used inbinding experiments. Conserved sequence motif is indicated withcapital letters. 243R mutants (APL, ADL, ASC) carry 2-aa substitu-tions within the conserved region. The 14-aa deletion of d11135(residues 225-238) includes the conserved motif and is shown above.

ing motif for the 48-kDa protein with adjoining sequencesplaying an augmenting role. We tested Adl2 ElA, which hasa similar 6-aa motif near the C terminus (PVDLSV versusPLDLSC; see Fig. 5B) for its ability to interact with the 48-kDaprotein. The in vitro binding experiments indicate that Adl2235R ElA (12H/235R) interacts with the 48-kDa protein(GST-30), albeit at a somewhat reduced level compared withAd2 243R (Fig. 5A). These results suggest that the interactionof ElA with the 48-kDa protein has been conserved duringevolution.

Sequence Analysis of cDNA. The DNA sequence of pAct30was determined and the open reading frame (ORF) in relationto the Gal4 activation domain was established. These se-quences revealed an ORF at 439 aa (predicted size, 47.5 kDa)that starts from an ATG initiation codon located downstreamof the Gal4 activation domain (Fig. 6A). We were unable toobtain additional protein-coding sequences by various meth-ods. It therefore appears that pAct3O contains the entireprotein coding sequences for the cDNA for CtBP.

In parallel studies, we also purified CtBP from HeLa cellextracts using GST-Cter fusion protein immobilized on glu-tathione agarose affinity matrix. Purified CtBP was subjectedto proteolysis and microsequencing. From these studies, aminoacid sequences of two peptides were obtained: IGSGFD-NIDIK (peptide A) and QGAFLVNAAR (peptide B). Thepredicted ORF of the cDNA clone (clone 30) containedsequences identical to peptide A (Fig. 6A). There was a singleamino acid variation between the sequences of peptide B andthe corresponding sequence of the 48-kDa ORF. The pre-dicted sequence contains a tyrosine residue at position 263,while the peptide B sequence contains an alanine at thecorresponding position. It is possible that this variation may bedue to an isoform of CtBP. The presence of the two peptidesequences in the 48-kDa ORF further strengthens our con-clusion that cDNA clone 30 codes for CtBP.Homology of CtBP. Comparison of the amino acid se-

quences with known protein sequences revealed that CtBPshares significant homology with various NAD-dependent

ACtBP

1 MGSSHLLNKG LPLGVRPPIM NGPLHPRPLV ALLDGRDCTV EMPILKDVAT

51 VAFCDAQSTQ EIHEKVLNEA VGALMYHTIT LTREDLEKFK ALRIIVRIf101 EFDZNDIZSA GDLGIAVCNV PAASVEETAD STLCHILNLY RRATGCTRRC

A151 GRAHESRASS RSARWRPRCQ DPRGDLGHHR TWSRGAGSGA AGQRVGFNVL

201 FYDPYLSDGV ERALGLQRVS TLQDLLFHSD CVTLHCGLNE HNHHLINDFT

251 VKQMRg UkIETMGGLVD EKALAQALKE GRIRGAALDV HESEPFSFSQB

301 GPLKDAPNLI CTPHAAWYSE QASIEMREEA AREIRRAITG RIPDSLKNCV

351 NKDHLTAATH WASMDPAVVH PELNGAAYRY PPGWGVAPT GIPAAVEGIV

401 PSANSLSHGL PPVAHPPHAP SPGQTVKPEA DRDHASDQL

BvanH 188

CtBP 222

FDELLQNSDIVTLHVPLNTDTHYIISHEQIQRMKQGAFLINTGRGPLVDTYELVKALENG:::II .1-11d1th 1nen-1id.. tva:ro1111r111 aaa1111--11-:1lqdllfhsdcvtlhcglnehnhhlindftvkqmrggaflvntargglvdakalaqalkeg

, 247

r281

vanH 248 KLGGAALDVLEGEEEFFYSDCTQKPIDNQFLLKLQRMPNVIIT TAYYTEQA 300:: 111111 1: 1X1 :1 . 1. 11:1-11 I1:1.111

CtBP 282 rirg&a1dvhes.epfsf9qgp.......lkdapnlicti*Jawyseqa 322

FIG. 6. (A) Amino acid sequence of CtBP. Underlined sequencescorrespond to peptide sequences A and B. (B) Sequence homologybetween CtBP and VanH. Alignment of CtBP and the Enterococcusfaecium vancomicin-resistance gene [VanH (30)] shows 67% similarityand 50% identity over the indicated regions. Boxed histidine residuecorresponds to His-296 of D-lactate dehydrogenase from Lactobacillusplantarum. This residue has been implicated in the catalytic activity ofthis dehydrogenase (31). Vertical lines, identical amino acids; :, similaramino acids.

D-isomer-specific 2-hydroxy acid dehydrogenases (32). Thesequence alignment of CtBP and one of the 2-hydroxy aciddehydrogenases (30) is shown in Fig. 6B. However, we havethus far been unable to demonstrate any significant NAD+-binding or dehydrogenase activity associated with CtBP usinghighly purified protein preparations or in cells infected withrecombinant vaccinia virus expressing high levels of CtBP(data not shown). Thus, it appears that despite the sequencesimilarity between CtBP and various dehydrogenases, CtBPmay not possess a NAD+-dependent dehydrogenase activity.

DISCUSSIONWe have identified the cDNA clone for CtBP, a 48-kDaphosphoprotein that specifically interacts with a C-terminalregion (aa 229-238) of the Ad2/5 ElA 243R protein. Thisinteraction was confirmed by in vitro protein binding and invivo coimmunoprecipitation studies. Peptide sequences thatwere obtained from purified CtBP matched sequences in thepredicted ORF of the cDNA clone. The peptide sequenceswere obtained from CtBP purified from HeLa cells, while thecDNA clone was isolated from a B-cell library. However, bothproteins are indistinguishable with respect to their apparentmolecular mass, binding properties, and phosphorylation site.Furthermore, the polyclonal antibody raised against recombi-nant CtBP recognizes endogenous CtBP from HeLa cells (datanot shown). Taken together, these data indicate that the48-kDa protein encoded by the cDNA clone (clone 30) isCtBP.CtBP binds within the same region (aa 225-238) of the ElA

protein that modulates the in vitro transformation efficiencyand tumorigenesis of E1A/T24 ras transformed baby ratkidney (BRK) cells (19). ElA mutants that lack the C-terminal67 aa or a 14-aa region (aa 225-238) encompassing the CtBPbinding domain (aa 229-238) cooperate more efficiently withT24 ras in in vitro transformation assays than wt 243R. Inaddition, the mutant E1A/T24 ras transformed BRK cellsform rapidly growing metastatic tumors in nude mice, and theyeven form tumors in syngeneic rats. It therefore appears thatthe second exon of ElA has a tumor suppressor activity andthis function is linked to its ability to interact with CtBP.

Proc. Natl. Acad. Sci. USA 92 (1995)

'-d'

-4AK. .. a

Proc. Natl. Acad. Sci. USA 92 (1995) 10471

Interestingly, CtBP displays significant homology to en-zymes of the D-2-hydroxy acid dehydrogenase family (30, 32).Members of this family have so far been identified only frombacteria, plants, and lower fungi (30, 32-39). CtBP mightencode mammalian enzyme whose activity somehow affectstransformation and tumorigenesis. Tumor cells differ fromtheir untransformed counterparts not only in quantitativechanges of growth rate and cell division but also in dramaticqualitative changes of their energy metabolism [reviewed byBaggetto (40)]. Some of these changes are associated with theactivity of specific isoenzymes (40). CtBP may be such anenzyme, whose activity is sensitive to ElA. However, so far wehave been unable to detect any dehydrogenase enzyme activityor NAD-binding activity for CtBP. It is possible that thehomology between CtBP and the dehydrogenase family con-stitutes the preservation of structural rather than enzymaticfeatures. Members of this particular family of dehydrogenaseshave been shown to form homodimers (35). CtBP may be ableto enter enzyme complexes and thereby modulate their activ-ity. This complex could then be stabilized or disrupted by ElA.Although the various dehydrogenases are considered to be

enzymes primarily involved in energy metabolism, a number ofsurprising activities have been associated with various proteinswithin this family. For example, the nuclear uracil glycosylaseactivity that is deficient in Bloom syndrome is associated withglyceraldehyde-3-phosphate dehydrogenase (41). Similarly, anucleic acid helix-destabilizing activity is associated with lac-tate dehydrogenase 5 (42). More recently, certain dehydroge-nases have been implicated in mRNA transport (43-45). Inaddition, protein kinase activities have also been associatedwith certain dehydrogenases (42). Thus, it is possible that CtBPmay also exhibit interesting activities. It is intriguing to notethat a DNA binding activity is associated with the C-terminalregion of Ad3 ElA (46) that is significantly enhanced by acellular cofactor. It could be speculated that CtBP may be acandidate for such a cofactor.So far, CtBP is the only protein identified that binds to the

second exon of ElA. The extensive genetic and biochemicalanalysis of exon 1 binding proteins has helped to understandkey regulatory events in cell cycle control and oncogenictransformation. Such biochemical analysis of CtBP may help tounderstand the regulatory functions of exon 2 on tumorigen-esis and metastasis.

We thank S. Elledge, S. Fields, S. Bayley, K. Fujinaga, and E. Harlowfor various reagents and W. Lane for microsequencing. We arethankful to Grace Denniger for valuable technical help. This work wassupported by Research Grants CA-31719 and CA-33616.

1. Zerler, B., Moran, B., Maruyama, K., Moomaw, J., Grodzicker,T. & Ruley, H. E. (1986) Moi. Cell. Bio. 6, 887-899.

2. Dyson, N. & Harlow, E. (1992) Cancer Surv. 12, 161-195.3. Nevins, J. R. (1992) Science 258, 424-429.4. Moran, E. (1993) Curr. Opin. Genet. Dev. 3, 63-70.5. Bayley, S. T. & Mymryk, J. S. (1994) Int. J. Oncol. 4, 2131-2141.6. Eckner, R., Ewen, M. E., Newsome, D., Gerdes, M. DeCaprio,

J. A., Lawrence, J. B. & Livingston, D. M. (1993) Genes Dev. 8,867-884.

7. Arany, Z., Newsome, D., Oldread, E., Livingston, D. M. &Eckner, R. (1995) Nature (London) 374, 81-84.

8. Lundblad, J. R., Kwok, R. P. S., Laurance, M. E., Harter, M. L.& Goodman, R. H. (1995) Nature (London) 374, 85-88.

9. Linder, S., Popowicz, P., Svensson, C., Marshall, H., Bondesson,M. & Akusjarvi, G. (1992) Oncogene 7, 439-443.

10. Bondesson, M., Svensson, C., Linder, S. & Akusjarvi, G. (1992)EMBO J. 11, 3347-3354.

11. Mymryk, J. S. & Bayley, S. T. (1993) J. Virol. 67, 6922-6928.12. Subramanian, T., LaRegina, M. & Chinnadurai, G. (1989) On-

cogene 4, 415-520.13. Quinlan, M. P. & Douglas, J. L. (1992) J. Virol. 66, 2020-2030.14. Urbanelli, D., Sawada, Y., Raskova, J., Jones, N. C., Shenk, T. &

Raska, K. (1989) Virology 173, 607-614.15. Douglas, J. L., Gopalakrishnan, S. & Quinlan, M. P. (1991)

Oncogene 6, 2093-2103.16. Pozzatti, R., Muschel, R., Williams, J., Radmanabhan, R.,

Howard, B., Liotta, L. & Khoury, G. (1986) Science 232,223-227.17. Pozzatti, R., McCormick, M., Thompson, M. A. & Khoury, G.

(1988) Mol. Cell. Biol. 8, 2984-2988.18. Steeg, P. S., Bevilacqua, G., Pozzatti, R., Liotta, L. A. & Sobel,

M. E. (1988) Cancer Res. 48, 6550-6554.19. Boyd, J. M., Subramanian, T., Schaeper, U., La Regina, M.,

Bayley, S. T. & Chinnadurai, G. (1993) EMBO J. 12, 469-478.20. Ma, J. & Ptashne, M. (1987) Cell 51, 113-119.21. Durfee, T., Becherer, K., Chen, P. L., Yeh, S. H., Yang, Y.,

Kilburn, A. E., Lee, W. H. & Elledge, S. J. (1993) Genes Dev. 7,555-569.

22. Moss, B., Elroy-Stein, O., Mizukami, T., Alexander, W. A. &Fuerst, T. R. (1990) Nature (London) 348, 91-92.

23. Chien, C. T., Bartel, P. L., Sternglanz, R. & Fields, S. (1991) Proc.Natl. Acad. Sci. USA 88, 9578-9582.

24. Gill, G. & Ptashne, M. (1987) Cell 51, 121-126.25. Ausubel, F. M., Brent, R., Kingston, R., Moore, D., Seidman, J.,

Smith, J. & Struhl, K., eds. (1992) Current Protocols in MolecularBiology (Greene and Wiley, New York).

26. Fuerst, T. R., Niles, E. G., Studier, F. W. & Moss, B. (1986) Proc.Natl. Acad. Sci. USA 83, 8122-8126.

27. Fields, S. & Song, 0. (1989) Nature (London) 340, 245-246.28. Harlow, E., Franza, B. R., Jr., & Schley, C. (1985) J. Virol. 55,

533-546.29. van Ormondt, H., Maat, J. & Dijkema, R. (1980) Gene 12,63-76.30. Arthur, M., Molinas, C., Dutka-Malen, S. & Courvalin, P. (1991)

Gene 103, 133-134.31. Taguchi, H. & Ohta, T. (1993) J. Biol. Chem. 268, 18030-18034.32. Taguchi, H. & Ohta, T. (1991) J. Biol. Chem. 266, 12588-12594.33. Yomano, L. P., Scopes, R. K. & Ingram, L. 0. (1993)J. Bacteriol.

175, 3926-3933.34. Arps, P. J., Fulton, G. F., Minnich, E. C. & Lidstrom, M. E.

(1993) J. Bacteriol. 175, 3776-3783.35. Goldberg, J. D., Yoshida, T. & Brick, P. (1994) J. Mol. Biol. 236,

1123-1140.36. Christoserdova, L. V. & Lidstrom, M. E. (1994)J. Bacteriol. 176,

1957-1968.37. Greenler, J., Sloan, J. S., Schwartz, B. W. & Becker, W. M. (1989)

Plant Mol. Biol. 13, 139-150.38. Colas de Francs-Small, C., Ambard-Bretteville, F., Small, I. D. &

Remy, R. (1993) Plant Physiol. 102, 1171-1177.39. Chow, C. M. & RajBhandary, U. L. (1993) J. Bacteriol. 175,

3703-3709.40. Baggetto, L. G. (1992) Biochimie 74, 959-974.41. Meyer-Siegler, K., Mauro, D. J., Seal, G., Wurzer, J., deRiel, J. K.

& Sirover, M. A. (1991) Proc. Natl. Acad. Sci. USA 88, 8460-8464.

42. Williams, K. R., Reddigari, S. & Patel, G. L. (1985) Proc. Natl.Acad. Sci. USA 82, 5260-5264.

43. Singh, R. & Green, M. R. (1993) Science 259, 365-368.44. Elzinga, S. D., Bednarz, A. L., van Oosterum, K., Dekker, P. J. &

Grivell, L. A. (1993) Nucleic Acids Res. 21, 5328-5331.45. Kawamoto, R. M. & Caswell, A. H. (1986) Biochemistry 25,

657-661.46. Ko, J. L., Dalie, B. L., Goldman, E. & Harter, M. L. (1986)

EMBO J. 5, 1645-1651.

Biochemistry: Schaeper et al.