JOURNAL OF VIROLOGY, Dec. 1994, p. 7840-7849 Vol. 68, No. 120022-538X/94/$04.00+0Copyright © 1994, American Society for Microbiology

Chinese Hamster Ovary Cells Contain Transcriptionally ActiveFull-Length Type C Proviruses

YOLANDA S. LIE,1 ELICIA M. PENUEL,lt MARI-ANNE L. LOW,1 T. P. NGUYEN,2t JOCELYN 0. MANGAHAS,1KEVIN P. ANDERSON,1§ AND CHRISTOS J. PETROPOULOSl*

Molecular Virology Laboratory,1 and DNA Sequencing Laboratory,2 Genentech, Inc.,South San Francisco, Califomia 94080

Received 22 April 1994/Accepted 7 September 1994

We have isolated a genomic locus from Chinese hamster ovary (CHO) cells that contains a full-lengthprovirus. Nucleotide sequence analysis indicates that it is a defective member of the rodent type C retrovirusfamily with an env region that is similar to those of mouse amphotropic retrovirus and subgroup B felineleukemia virus. We were able to demonstrate that this provirus is a member of a closely related family offull-length proviruses in CHO cells and Chinese hamster liver. Hybridization probes generated from thisgenomic clone were used to characterize type C retrovirus RNA expression in CHO cells. Full-length genomicRNA and subgenomic envelope mRNA were detected in CHO cell lines but not in the human-derived 293 cellline. Interestingly, we discovered that the site of retrovirus integration lies within a G repeat sequencebelonging to the short interspersed element family of retroposons.

Since their development in the 1950s, CHO cells have beenused extensively for genetic studies (see reference 13). Re-cently, CHO cells have been used to manufacture recombinantprotein therapeutics (7, 11, 18, 33). Several features make theCHO cell culture system a suitable choice for the manufactureof recombinant protein therapeutics (for discussions, see ref-erences 12, 24, and 40). Adherent CHO cell populations can bereadily adapted to growth in suspension culture to facilitatescale-up and manufacturing. In general, complex proteinsundergo more appropriate glycosylation and posttranslationalmodification when produced in CHO cells versus other eukary-otic (yeast or insect cells) or prokaryotic cell systems. Anefficient system based on cell lines that are deficient indihydrofolate reductase activity is available for generating cellclones that produce large amounts of exogenous protein(reviewed in reference 44; see reference 3 for methodology). Inaddition, CHO cells are nonpermissive for the replication ofmany human-pathogenic viruses (43).

In contrast to many rodent cell lines, CHO cells do notproduce demonstrable levels of infectious retrovirus (9, 17,22). Most CHO cell lines are resistant to infection by all of themajor classes of murine retroviruses, including amphotropicstrains, which are known to infect cells from a wide variety ofmammals (5, 14, 38). However, both intracytoplasmic type Aand budding type C retrovirus-like particles have been de-tected in CHO cells by electron microscopy (16, 17, 22, 23, 39,42). The small number of retrovirus-like particles produced byCHO cells and their lack of demonstrable infectivity suggestthat they are the product of endogenous retrovirus sequencesand not exogenous retrovirus infection. Previously, Andersonet al. (2) reported the molecular and biochemical character-

* Corresponding author. Mailing address: Genentech, Inc., 460 Pt.San Bruno Blvd., South San Francisco, CA 94080. Phone: (415) 225-1328.Fax: (415) 225-2866. Electronic mail address: petropoulos.christos@gene.com.

t Present address: Department of Molecular and Cell Biology,University of California, Berkeley, CA 94702.

t Present address: Applied Biosystems, Foster City, CA 94404.§ Present address: Department of Molecular and Cell Biology, ISIS

Pharmaceuticals, Carlsbad, CA 92008.

izations of the retrovirus-like particles concentrated fromCHO cell culture fluids. Purified particles were shown to beimmunologically related to mammalian type C retroviruses andto possess detectable levels of reverse transcriptase. It was alsoshown that the particles contained RNA sequences homolo-gous to rodent retrovirus RNA. Related sequences weredetected in CHO cell and Chinese hamster liver DNA prepa-rations.To extend the investigation of CHO retrovirus-like particles,

a recombinant X phage library was constructed from CHOgenomic DNA and phage plaques containing retrovirus-re-lated sequences were identified by using a cDNA probe thatwas generated from purified CHO retrovirus-like particleRNA. In this report, we present the first complete nucleotidesequence of a full-length type C provirus isolated from CHOcells. We show that it is a member of a family of full-lengthtype C proviruses that are found both in CHO cells and inChinese hamster liver. Some of these elements are activelytranscribed and produce both full-length genomic RNA andsubgenomic envelope mRNA. The observations we report areconsistent with an endogenous origin of type C retrovirus-likeparticles in CHO cells.

MATERUILS AND METHODS

CHO genomic DNA library. A genomic DNA library wasconstructed from a Sau3AI partial digestion ofCHO cell DNAwith the Lambda Fix vector system used in accordance with themanufacturer's (Stratagene, LaJolla, Calif.) instructions. Xphage plaques containing type C retrovirus-related insertswere identified by conventional filter hybridization methods byusing 32P-radiolabeled ML10 sequences (3). ML10 is a cDNAsequence that was generated from the RNA of CHO cellretrovirus-like particles and is homologous to the endonucle-ase region of rodent type C retroviruses (2).DNA sequence analysis. XML2G sequences were subcloned

into the pUC219 vector by using standard molecular cloningprotocols (3). DNA sequencing was performed by the dideoxy-chain termination method with single-stranded DNA tem-plates (1-3). Internal primers were synthesized on the basis ofderived DNA sequence information. Computer analyses of

7840

on August 3, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

ENDOGENOUS TYPE C RETROVIRUS EXPRESSION IN CHO CELLS

DNA sequences (GenBank homology search, homology plots)were performed with the sequence analysis package (thirdedition) generated by the Scientific Computing Group ofGenentech.DNA and RNA analysis. Genomic DNA was prepared from

CHO cells by using proteinase K digestion and phenol-chloro-form extraction (3). The products of restriction endonucleasedigestion were separated by electrophoresis and analyzed byconventional Southern transfer hybridization (1-3). Total cel-lular RNA was prepared from cell cultures with RNAzol(Tel-Test "B", Inc., Friendswood, Tex.), and the poly(A)+RNA fraction was isolated by oligo(dT) cellulose affinitychromatography (Pharmacia Biotech, Inc., Piscataway, N.J.).The poly(A)+ RNA fractions were analyzed by Northern(RNA) transfer hybridization methods (1). Hybridizationprobes were generated by preparing purified long terminalrepeat (LTR), gag,pol, and env sequences as depicted in Fig. 1.Radiolabeled probes were prepared by nick translation(Gibco/BRL, Bethesda, Md.). Filter hybridizations were per-formed overnight at 42°C in hybridization buffer (50% form-amide, 50 mM HEPES [N-2-hydroxyethylpiperazine-N'-2-eth-anesulfonic acid; pH 7.0], lx Denhardt's solution, 3x SSC [1xSSC is 0.15 M NaCl plus 0.015 M sodium citrate], 150 ,g ofdenatured salmon sperm DNA per ml). Filters were washedonce for 15 min at room temperature in 2x SSC and twice for60 min each time at 55°C in 0.1X SSC-0.1% sodium dodecylsulfate. Radioanalytic quantitation of hybridization signal in-tensities was performed by using an AMBIS 100 Detector(Ambis, San Diego, Calif.).PCR. B52-related LTR sequences were amplified by PCR

(3) from CHO cell genomic DNA by using the followingoligonucleotide primers: 5'-CGC-GTC-GAC-TGA-AAG-ACC-CCT-GTC-C`-l-TAG-CCC-3' (forward) and 5'-GGG-TCT-AGA-ATC-CCG-GAC-GAG-CCC-CCA-3' (reverse).The underlined sequences represent Sall andXbaI recognitionsites that were incorporated into the forward and reverseprimers, respectively. The PCR product was characterized bysequence analysis.CAT assays. The ML2G LTR region was inserted into the

pCAT vector series (Promega Corporation, Madison, Wis.)between the Sall and XbaI sites by using a Sall site at position1 and an XbaI site at position 1276 of the ML2G sequence.CHO Kl cells were transfected with LTR-chloramphenicolacetyltransferase (CAT) plasmids by the calcium phosphateprecipitation protocol (3), and cell lysates were prepared 48 hafter transfection. CHO cell lysates were adjusted to anequivalent protein concentration by using a modified Bradfordassay (Bio-Rad Laboratories, Hercules, Calif.), and CATactivity was measured by conventional methods (3). Levels ofCAT activity were quantitated by radioanalytic imaging (Am-bis).

Nucleotide sequence accession number. The nucleotide se-quence reported here (see Fig. 2) has been submitted toGenBank and assigned accession number U09104.

A.

F 14.6 Kb XS S S

,1 1 1

X SScr_ 1I I I

RV RVL 8Kb l

B.

LTR LTR*11 gag I pol I nv

H H

Xb Xb

provirus

probesS P

genomic RNA

env mRNA

FIG. 1. CHO cell endogenous provirus map. (A) Recombinant Xphage ML2G was isolated from a genomic library of CHO cell DNA.XML2G contains 14.6 kb of CHO cell DNA (thick bar) that can beexcised from the A vector arms (thin bars) by digestion with restrictionendonuclease SalI (S). The 5' and 3' boundaries of the mammaliantype C retrovirus sequences within the ML2G insert are defined by theprovirus LTRs (solid boxes). An EcoRV (RV) recognition site ispresent in each LTR but not elsewhere in the provirus sequence. Theprovirus contains several unique restriction endonuclease recognitionsites: ClaI (C), XhoI (X), SalI, (S) and ScaI (Sc). (B) Nucleotidesequence analysis has defined a complete type C provirus withinML2G. The positions of the gag, pol, and env genes (thin line) andflanking LTR sequences (solid boxes) are shown. Each region wassubcloned separately to create nucleic acid hybridization probes (solidbars) specific for the LTR, gag,pol, and env regions. These probes wereused to identify related sequences in CHO cell DNA and to define thestructures of RNAs (solid lines) expressed in CHO cells. PstI, P;BamHI, B; XbaI, Xb; HindIII, H; SalI, S.

RESULTS

Isolation of endogenous retrovirus sequences from CHOcells. To isolate endogenous retrovirus sequences from CHOcells, a genomic DNA library was constructed in AFIX (seeMaterials and Methods for details). Approximately 5 x 104recombinant phage, which we estimate represented 10 to 20%of the CHO cell genome, were screened by nucleic acidhybridization. The hybridization probe consisted of a cDNAsequence (named ML10) which is homologous to the endonu-clease region of murine type C retroviruses. ML10 was isolated

from a cDNA library generated from the RNA of CHO cellretrovirus-like particles. Its characterization has been reportedelsewhere (2). Hybridization screening identified 19 recombi-nant A phage-containing ML10-related sequences. On thebasis of preliminary restriction site mapping studies, cloneXML2G, containing a 14.6-kb insert, was chosen for furthercharacterization (Fig. 1).AML2G contains a full-length retrovirus sequence. The

nucleotide sequence of a continuous 9.6-kb stretch of theXML2G insert was determined by using conventional subclon-ing and dideoxy-chain termination sequence analysis protocols

7841VOL. 68, 1994

on August 3, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

7842 LIE ET AL. J. VIROL.

1 GIrGCCGA A TGATCGAT PCrGCGW AcATI'Ir CWAT= AIGAGC MAMMA= CCAGTIGATA101 GrACAr A!Ic TAcxxr CCA AAGO TAAT TinI'ILAI'G AGrATR GMAI CrAGCATCC ACCATK3G AAGCOCCA201 TiwATAG AfGlPT cocIA=C TPAcAC AI3S= GCcn;G= A3XAX:n T=EAcwx TIGAAC r301 GocTIUa= ci7m GGPDCCWAA? ATMG ATA/OM ACCCCCACC C GCRL*i:U3 5 AG: SGC401 GCAT Uri TAQA TGATT= AGCTGCI AQ= QQCCA=

501 Q CAAGGCC GC CACG GOCA MAT AAG&'AMATGAAA601 Ic=TP crC MTAAGGACCTACCTTrZC TATCAUAPGAAX CCM MCT

701 ATTACATCMAATa CTGA GlC=i!T ACGAAGA CTCTCCSWO GTr=TTATPBS SD

801 G7MMSUC GAGAC CCCAGGAC ACCCAC AOTIGGhM TAAI1T IrGt:tC1001 AACG ATC SGGGCG ACACG TCT=ArIGM GCCCM

1001 CCAG~AGA WcIGgPOM ATIGAMfCA TAGTGGO'I ACCITIUTGATCIITIITI A GAr-ACTrCC CAACC'ICTrr1101 GGPGA= C A GA TICA ATCCrca3 G3GAGAAGA AwItC GtUII CCox AA TrAMr

GAG1201 GTcAAAan ocAGcGAAA GOt'IUFtfI TM== tTnIGAar TAACTTAT CiIIrA CrAAMMAG AAATar, ACAAACPAU1301 AccAci=r WAC ACICAC TGUMAG McAGTr TGCICATA CAATCIG I ACCAAAIG A_T7CI1401 AGGGOCAMM TI'GACriM G AGAGO GTAC TrTAACt'ItC A TATATTcAGGTGA GAGA AGTA W1501 AGcma=} CATcOcAcc AWIG=A CATAGCAcr CGATA CC C ACCCA-1601 CroCIw Cri1it:r CrAct= CACCO ACCOCAAm T1UAT= CiCWrC GOCAIttW, CCmrIc1701 AcGAG=AA GAGAETA GOIAAGAGA AGAACA[ AWI AAACSIGPGMCIGAr GACIGAA CC1801 ACGCGCC GCCAC: OR GAAXXMt GGCAAGACG GTIrOU= OCACI IA GGO'IAC1901 GAS TC.AGG CACOCCI CAGA.TICGC.C1TXk_ CI.MMACM GACCGAAP CACTCAA T ATc2001 AAOIGGAAAA ATAATA1A11UCIr11UICr GCACACt' r GCTGAI CACAACWIAC CAACCM2101 GGmAwIAaw = T1WCAGI MAGWGAAGAGAAA CAC1I'IM TACrGAGA.GAA.AAT CC3AAOOMG2201 GGCcAfccPAGCIGOtcAAG AAA=AGC GGrO1C CTI WfAGC CGAATGG TITTACrACr GAAC_flGTA GGACCCTCT2301 V1UCU7IG3G77rIG CAGGAC ACCTACCAAT TACCGAACGr AATACPGG GC=GAAT COaXCrC2401 T AGACAAAA GAATAr, AAUMATA= COUCCqAGAIC CAAAGAaU AGCAGO ACCACC MIVITA T rICA2501 AGCCOCGGPC ATAAAA-CA AGcCCAAA GcIAAAAT TrACAAT ATACO'ICCA AGICITA AA=.PAGCAG AAnMUr C TATAAGAG2601 GAAW &A R AGAACI'I7G AGAAAGGAA C GAAAGAC AGAA CY=GGAA GCIr AGAGAtC=2701 AGoGAATA AAGATGAGC AGOTATI ccAGrAGr GAGICAG AGACAGAATA GGCAGAGGA TGGCAGAAG GG0C0C =ACAG

POL2801 TcAAaocr TwUGAAAG AAAAMAcA TiuGCAAAA GAATOWA AGM GGCtAAG= COAOOCCAA GACIGT CTIOAAW2901 vnNGAAG7TCAcAGAAMA GoccAGG AXI CGA AACAC'AGC G ATC3GCAC cTIW.rAGCGATACAGSG3001 CArATA= K=CGAAz CGIUC GA I MACCGAr GCAIGGI= AGAC'IC APGoaAAW CAanITC GTrAC'AAA3101 oGcAI CAitcA crAnrTA_GCAT=T TI AGa CCCAGAT'I cccc TrAWrAGGA AT GACCAAATT3201 AAG AA ATAcTI AG PGIUr AGCAAAGI ACAGOGCAG ACGAAI'CCrrArCCCTIGACAI CCATAGAAGA LATACAGA3301 CTcAAAAGGAA TrnAAcAm cAGGAAA=C TGATcAI GcTG A T MAG AACAGGRA GC C3401 CATcAAccAG GccAATA TAGrAAC AAAAGII ACAG'AI AICrAAAr AAGOXAGA AGGAAIM3501 ccAcATwm AAAGrrAC GcAAA TAAA CCTIAAA C AAGCAGAA CCAA7GAM3601 TAoGoOAGTA cAGGATcIA oGAGGIA TAAAATA GAGACATAC A c1rr CrCCAACCCT TATAATI 7GAGGAT GCOAAC3701 AT.IGG ATALMT AGAICTTAAGAAGATC'1:C CrATATTIG C=MUATI C.AGGAIGCXr3801 AccriAAT C CPACTIGGA CAGTAC 7CAAGG I AAGAAG OCUCTII MAAAGT TrAATAGG A MCAGG3901 A A AAr C'ITAGr a GA.MA 'iU:Cltt'IX AGCCAAAAC AAGGAAGAAT GCAA MAAGG CACP.AG4001 CICAZPB C TC CCAArAAGC CCAGATAlU CAGAAACAGr I AGGATACAAG ATAaAGGA=4101 GAGGGGAIG GcAnGA G A GC GOCA1 ACAT ACCCACAAA ATaXXMCCA ACIGAGAGAG TMI CRCAG=C4201 cIanoc'r U oaMrAATGOCrnr rCC1CA a GCrTFr%AA GGGAAGAGC CAAAAGAA4301 GOCt1ACCA ACAAAAA GAATCACrAG AOGACrA GCAAG= TIGAACI TATAGAMAG MAMA=4401 ATGCGAAG.AGccT cAAAAAI GC G AAGGP,C.CA GCATAL IC AAGAAAIT I GCAC4501 lIlATG ACCAC AGAAAAGAT CM= CCATACA.W I G ;GIGCAG4601 ATCAGA I AGAMC C=AmCG AATACAT T1GACCA1A IGG CAA-CGG GACTG GI4701 GCilana GcAGCAW TGC'IttCTCr<XXGGAG CCAGAGGC ATAATI= ACAGGACI GCIGG ATGGGG ACCCGAC4801 AciGAcCA C CCCGGA=AC AIIMA CGGAMGAAG CAGTITM CA=kAGGAA AACGAAGGC GMACGAACAG4901 AGAauAGrC I Cr CAcA= CTCCC C AG GCAGAA TCATrGCALT CACAGCrr CAnTIG CpGAAGGSA5001 GAGGOIUAG ACACAG Ar-GItUTAari ACICCATA TACATGAGA AATTrA CXA GS GGC IML= GOGAM5101 GArATTAAm ATAAGA AATITICr= CjITAGAr C1rI G W TAAG7A= TAcATIG CGGACAIAA A_kzMGGATr5201 CTGAAGC CAGCAA.T CGGA=GGCAG A CCAGAGGO GcoIarA CAGACACCAC CCACPAAG5301 Cormc1 ACIRMAA AAGAcc CAAA AGAAATIG AGOCACAAzG GAACAaA GGGRIAC TICTAAGA5401 AAGACIA A GMACCAAGr ACCAIC AMTAA AAGATCW.AC AGOI,AG CAACGAA,AM GX3AGO 'IO CAr

FIG. 2. Complete nucleotide sequence of a CHO type C provirus. The nucleotide sequence of a 9.6-kb region of the XML2G insert is shown.The region contains a complete 8,422-bp type C endogenous retrovirus sequence flanked by 345 bp of the upstream cellular sequence and 836 bpof the downstream cellular sequence. Sequences that make up the 5' and 3' LTR regions are underlined. Retrovirus integration has generated a4-bp duplication (ATAG) that borders the 5' and 3' ends of the provirus and defines the site of retrovirus integration. Several important andcharacteristic features of the provirus sequence are shown in boldface. These include the inverted repeats at each end of the LTR sequences, theTATA box in the 5' LTR, the polyadenylation signal sequence in the 3' LTR, and the env mRNA splice donor (SD) and splice acceptor (SA)sequences. The gag (GAG), pol (POL), and env (ENV) regions were tentatively assigned on the basis of homology to the MoMLV sequence. Thenegative-strand PBS specifying tRNAPro and the positive-strand primer consisting of the polypurine tract (PPT) are shown in italics.

on August 3, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

ENDOGENOUS TYPE C RETROVIRUS EXPRESSION IN CHO CELLS 7843

550156015701580159016001

AGAA= A mTGAGAAcImAACAA!r GCnCACcGc AG ACA11MGGlOG AGCBCG CCCGAA APGAA AG=IA GA5GiTrAA CAGATA 70GA7TIA

TrITIUiW C G? TGGX AAGCCTrc C 1PAAAC GLAACACA AGA,t7Ml=AC TAAAAAT CIGAAGAA I

TIOGTG CAtiIr I ACAGA TAA1GtC GIq T CCCAGATAAG TCAGCG GCCACCI= -lG A7GA TIGAAATr1CAM== ACrACG TACOItTGA CAGGEIcC A ACW MCAEM ACACCr TI'IGAAATIC PCTGG GCACA

CrACCISCl X TACGIT1iCZ AA)CACCV GGCkrGGI-AA TM mAAA= TGCATGATAACICSA

6101 A7IATrAAcr rici7AIA AGAIIIA G ATC=OrA r CAAG= C CACrACAATrACAACXG GATI'IIGA6201 AAca)c TcAGCIAT AAw.ACCAGA GoAwCAAC cAcakTc CATicIPcc A CACCI GZ T CPACCAAGc6301 ACCAi= AActA= GAGA3c cc C ciCArM eCGA c r

EW6401 CAaurAAzcaCGA cAcATIC GcfxA-A CAGATaXAcAACACICAcAAcIC C1'ITAA.AGr AAGACIT CI!TCA=6501 ni lr rI vI'Iuit'IG TimCC ATMr.CAC ThAPAC AGITrA AMTCACCr GAAAATAWC AAItGA cooGOAU6601 AccAA=X:MAcxc A TAmGAI AcATcATa- TICIIU= C=II TrAGrA= CIG I IIGC

6701 cAGGAAGCA TctcA cA¶ocA cAcio GAA.TAm GAAc AAATI GarIATAGA cAAciA6801 CCACrAGaG AAGCIUIOG k AAACA CAGGC TnACTGGAAA ctcT= CrTIGATIT CA7IItI6901 A acr APiUcAm Gr.1GcAGGG AMAC G GCAAAGcc T T3GGATAGT 11= m CAAGGOCCA7001 c_ccG AAAGc GcIcA= TAAGi'Ic AGAO'IGG AAAAGGAAWiXATrMUUC'rA:IC IACIScYCX CrCACM7101 ACAAA s CT.Ar IC CCIGMAC" CAAATnACC=coGccA ACAAs=IC GWAAAA CAGAATAC GACCAM7201 GocoasAC AATiccAAGr CciGACC CAAACTAIA AGwr GC= AIt C CsCCCAA ucCrA7301 ATcGAGiAA nccAGAGAc fX3rG GGAIAT mTr cAAT AA=AAGG GITIcAAG rAA.A TTCAGA AACAAGI=7401 A sAATCI GiTMccm GI'IKtG G CA A7UAAS AATAMUA CCAAGIWAG CATrkI A(GAOA ATACCTIGIT7501 GAACTrA= VCrIGCArAIIr rAcCACr IrCAACA CCACATTrGA rrI T GATAA.7601 mriGGCCAA AGrCCAAT CACA-GM AATUI cArDrM GGAAA7IAA C=ATATA GAOAr AM rA cr

7701 ATTM'rA GGAA.TAACA TCATA4C1GC AT.OAG GAA:CGMC CC'CAGG ATnACA TrAACI ACAACAACC7801 ATGAA AcA caar Ac.GAcA cG-=cctAA ATcrAc v AmGMW <'GCAAAAA CGGAM7901 ATIrATr Trn CG - GI,71GCCC CAAGAA ATGCACITA TACmAGAA GrMAGGA GCATGOCCAA8001 AcarAA A^AGc%A AGPnGAA- 7C MACa AGAc A.clW

8101 TrrrAci= k ArA AAr TATA'A.r. ACA T AACICrAATU AM'IGAGGA .TAIu8201 cATAcAAA= TrrpTlrjAA ciKAAcAAT cctAA AAGCAGATA, ATcAAA T.AGAGcG.AATGA AGAIIA= CAGrAACAG

PPT8301 AzGAAAmsw COCCCAC 3Q=AQ a= C Il 3M ICArTL=1 I C C7

8401 -M ACAr G erccc c cC GC7 AA P8501 ACG4C GC=AAA AGCA= CrIt3CT 7AAC19M G ATTGA tlrc=:. MAMA= lwuliv

8601 ULyGaMlA=AGAG= 1TACTCCAAA8701 GI CGCIMT= l l a TC C 2= ATA riur T1AG' (GG8801 A CCITG'GA TIGACATG'A AACA.ACI TAAI AAA.TrAAAA TI'IGAAAAT CAOC.AGA GAAAIItr8901 TrAAGT GOTrrICAAT GATArsc ACA=Ic AAG'ITIUG GGACACTCAA ICTI TAGOGA9001 CT TAA= A TACAAr Gl3MriA TAArroIG AATAAGATM CAMAGCIM A G=iG= TACAA

9101 GAATIUCIM CAAATAAIC CAT=AIGACCAGCnIAAMIACSBGGICAGIAAarai C CGAC T3IGAA GOMrA=9201 ciGAc GCGAAciA GCATcA TAAGAAAwr TAATG U1'ilImlI CTGAGCAIG AIrCAG cAnrATAGA ACrACTAaGA9301 GcTocAr ACATTITrA GICAawTt'I ATAT'IU Si'rwT AAcTC Cr CACTGIC CAGAAAAI A

9401 ccAGA=' cA Gc AAA Am%w ATMAAaIA .AcwArrr PTA r AUcGIAGrAcAr ATAT CCITrrAA¶C9501 AAGTCA ATI1TTCATAT AarAGT TII AAAATrA TA 'TCTCAWT GI'TATrIuIT-i iTIU

9601 Tm3

FIG. 2-Continued.

(1-3). The nucleotide sequence obtained from this analysis isshown in Fig. 2. The derived sequence was screened againstsequences compiled in the GenBank database. This compari-son identified an 8.4-kb retrovirus-related sequence containinggag, pol, and env regions flanked by an LTR sequence that isabout 450 bp long (Fig. 1). The best match over the entirelength of the ML2G retrovirus sequence (approximately 64%)was to members of the murine leukemia virus (MLV) family.The overall similarity between ML2G and the Moloney strainof MLV (MoMLV; GenBank accession numbers J02255,J02256, and J02257) is illustrated in the dot matrix homologydisplay shown in Fig. 3A. As might be expected, homology to

MoMLV was highest within the pol region (69%), followed bygag (63%) and env (55%). No similarities were detected whenthe ML2G sequence was compared to full-length mouse (25),Syrian hamster (30), and Chinese hamster (10) intracisternal Aparticle (IAP) sequences.The ML2G LTR consists of about 310 bp of the U3

sequence, followed by about 70 bp of the R sequence and

another 70 bp of the U5 sequence. Similarities between the

ML2G and MLV sequences within or adjacent to the LTRregions are restricted to small stretches containing criticalcis-acting regulatory elements. These include the tRNA prim-er-binding site (PBS), polypurine tract, inverted repeat se-

quences, TATA box, polyadenylation signal sequence, tran-

scription start (CAP) site, and env splice donor sequence.Sequences resembling or analogous to the 75-bp enhancerelement repeats within the U3 region of the MLV LTR (21)were conspicuously absent from the ML2G LTR sequence.

Recently, Zhu et al. (45) reported the isolation of a CHOcDNA with retrovirus characteristics (referred to as B52;GenBank accession number, M89808). B52 is a partial cDNAthat contains a large portion of the transmembrane (TM)region of a retroviral envelope gene along with the U3 and Rregions of a 3' LTR. Comparison of the ML2G and B52sequences indicated that the two are similar in the TM (91%homology) and R (72% homology) regions but contain quitedifferent U3 sequences (Fig. 3B). Aside from the nucleotidesequence distinctions, the B52 U3 region is also considerablylarger than its ML2G counterpart (662 versus 310 bp). A

VOL. 68, 1994

on August 3, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

7844 LIE ET AL.

2000 4000 6000 6000

ML2G

nl . |nIPOL

K0KI-

I B

200

400

600

1000

.~~\ I

\ 1 I

Ml:2GI~~~~~~~~~~~~~~~~~~

ENVENV L-AU3R

wU'I

IIF---U3 R

FIG. 3. Dot matrix similarity plot comparing the CHO provirus to MoMLV and the B52 cDNA. The nucleotide sequence of the CHO type Csequence was compared to the MoMLV sequence (A) or the CHO B52 cDNA sequence (B) by using a dot matrix display program. The matchcriterion was set at 10 matches in a window of 12 nucleotides. The diagonal line generated by this analysis identifies regions of identity betweenthe two sequences. Each sequence is schematically portrayed along the corresponding axes of each plot. The jagged edges in the B52 illustrationdenote that this is a partial cDNA sequence.

comparison of the ML2G and B52 LTR sequences revealedtwo distinct regions within the B52 U3 region that are absentfrom the ML2G LTR. Interestingly, these regions of the B52sequence contain repeated sequence elements; however, thefunctional significance of these repeats has not been demon-strated.The ML2G sequence is a mammalian type C provirus.

Several structural features distinguish the ML2G sequence asa member of the type C retrovirus family. In Table 1, thepositive-strand primer sequence, the inverted repeat sequence,and the negative-strand PBS of ML2G are compared to thecorresponding sequences of murine type C retroviruses, as wellas mouse, Syrian hamster, and Chinese hamster LAPs. Thepolypurine tract, inverted repeat, and PBS of ML2G clearlymatch the murine type C retrovirus sequences more closelythan the mouse or hamster LAP. The PBS of ML2G (ignoringa 1-bp insertion) specifies tRNAPr' as the probable primer fornegative-strand DNA synthesis during replication. The latter is

a common feature of many, but not all, rodent type Cretroviruses, yet it is not restricted exclusively to rodentretroviruses (see reference 6). In contrast, the PBSs of mouseand hamster LAPs typically specify a tRNAPhe primer (10, 25,30, 35; see reference 20).The predicted amino acid sequence in the region joining

ML2G gag andpol specifies translation ofpol sequences in thesame frame as gag by nonsense suppression of an in-frameUAG codon. Translation of the pol region by suppression of astop codon (UAG) between the gag and pol open readingframes is another characteristic feature of the murine type Cretrovirus family (for reviews, see references 15 and 19).Typically, IAPs translate their pol region by using a -1frameshift (10, 20, 25; see references 15 and 19).The sequences at the junctions between the LTRs and the

adjacent cellular sequences reveal characteristic features ofretrovirus integration. Integration has resulted in the loss of 2bp from both ends of the virus (i.e., the start of U3 in the 5'

TABLE 1. Comparison of Chinese hamster retrovirus sequences with corresponding murine type C retrovirus and mouse, Syrian hamster,and Chinese hamster LAP sequences

Sourcea Positive-strand Inverted-repeat Negative-strandprimer sequence primer

CH-ML2G...... AAGAGAAATGGGGG TGAAAGACCCCC tRNAProCH-B52...... AAGAGAAATGGGGG AATGAAAGACCCCC UnknownMoMLV...... AGAAAAAGGGGGG AATGAAAGACCCC tRNAProCH-IAP....... AAAAAAAGGGGGA TGT tRNAPheSH-LAP...... GAATAGGGGGAGA TGT tRNAPheM-IAP...... AAAAGGAAAGGGGGATGTT tRNAPhe

a Sources of sequences are as follows: Chinese hamster ML2G (CH-ML2G), Fig. 2; Chinese hamster B52 (CH-B52), Zhu et al. (45); MoMLV, Shinnick et al. (36);Chinese hamster IAP (CH-IAP), Dorner et al. (10); Syrian hamster LAP (SH-LAP), Ono et al. (30); mouse IAP (M-IAP), Mietz et al. (25).

A

2000

4000

6000

80m0

mz

CO)

I GAGRU5

J. VIROL.

IL

on August 3, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

ENDOGENOUS TYPE C RETROVIRUS EXPRESSION IN CHO CELLS 7845

GAG POL ENV

Fa121l 11l:11111 11 1116I

o 1 2 3 4 5 6 7 8 9Kilobase

FIG. 4. Locations of open reading frames. The predicted aminoacid sequence was determined for the three reading frames of theML2G provirus (frame 1, 0; frame 2, + 1; frame, 3, -1). The positionsof all termination codons were identified and are displayed as verticallines.

LTR and the end of U5 in the 3' LTR) and has produced aduplication of the cellular sequence ATAG at the site ofinsertion (Fig. 2). Four-base-pair duplications such as this arecharacteristic of murine type C retroviruses (see reference 41),and although some TAP elements may have 4-bp duplicationsat their integration sites, the majority, including a full-lengthChinese hamster type A sequence (10), are flanked by a 6bp-repeat of the cellular sequence (20). All of the aboveobservations are consistent with the placement of ML2Gwithin the mammalian type C retrovirus group.The ML2G provirus is defective. Although the ML2G

sequence contains all of the components of a full-length type Cprovirus (i.e., LTR-gag-pol-env-LTR), it possesses many de-fects that would prevent the production of infectious retrovirusparticles. The defects consist of a large number of accumulatednucleotide substitutions and small insertions and deletions thathave disrupted both the gag-pol and env reading frames (Fig.4). Despite multiple disruptions in the gag, pol, and env genes,relatively long stretches of amino acid sequences with obvioushomology to murine type C retrovirus gag-, pol-, and env-encoded proteins can be predicted from the ML2G nucleotidesequence (for example, see Fig. 5). The ML2G sequence alsocontains a defective tRNAPrO PBS sequence. The 18-bp se-quence has been disrupted by the insertion of a G residue (Fig.2).The ML2G env region is related to murine amphotropic env

sequences. Within the env region, the extent of homology toMoMLV ranged considerably: from 52% within the surface(SU) domain to 67.5% in the TM region. Further analyseshave indicated that the CHO type C env region is more closelyrelated to the env genes of the nonecotropic MLVs than it is toecotropic MLV. The greatest degree of homology (approxi-mately 60% in the SU domain and 68% in the TM region) isseen with amphotropic (4070A) MLV, an amphotropic MLVrecombinant strain (lOA1), and subgroup B feline leukemiavirus [FeLV(B)]. The CHO env SU region has only 50%homology with the sequence of gibbon ape leukemia virus, aretrovirus that recognizes cell surface receptors in some CHOcell lines (27).Workers in several laboratories have independently local-

ized determinants of nonecotropic receptor specificity to vari-able domains in the amino-terminal region of the env SUdomain (4, 28, 31, 32). We used the nucleotide sequence of theML2G provirus to predict the amino acid sequences for theseregions and compared them to the corresponding regions ofamphotropic, lOA1, xenotropic, and polytropic MLV andFeLV(B) (Fig. 5). A comparison of the first variable region(designated vr-A by Battini et al. [4]) indicates that ML2Gresembles amphotropic MLV (4070A), an amphotropic MLV

recombinant (lOA1), and FeLV(B) more closely than eitherxenotropic or polytropic MLV. The second variable region(vr-B) of the ML2G env SU domain is less conserved than vr-Abut again appears to match amphotropic MLV and FeLV(B)more than the xenotropic or polytropic strain.CHO cells contain a family of proviruses related to ML2G.

To assess what fraction of ML2G-related sequences exist asfull-length proviruses, we digested CHO cell DNA withEcoRV and analyzed the digestion products with hybridizationprobes prepared from ML2G gag, pol, and env sequences (Fig.1). EcoRV was chosen for this analysis because the onlyEcoRV recognition sites found in the ML2G sequence exist inthe LTR regions. Consequently, EcoRV digestion shouldgenerate an 8-kb unit length retrovirus fragment from full-length endogenous proviruses that are closely related toML2G. As shown in Fig. 6, an 8-kb EcoRV fragment was easilydetected in CHO cell DNA by using the gag, pol, and envprobes. No signal was detected in DNA prepared from thehuman-derived 293 cell line. With radioanalytic imaging, thesignal intensities of the 8-kb bands were found to comprise 10to 20% of the total signal present in the lanes. This observationindicates that at least 10 to 20% of the type C sequencesrecognized by the ML2G hybridization probes are containedwithin full-length proviruses.

Hybridization probes prepared from the gag, pol, and envregions were also used to test for the presence of relatedsequences in DNAs prepared from Chinese hamster (Cricetu-lus gniseus) liver, BHK (Mesocricetus auratus) cells, mouse A9cells, and human 293 cells. Abundant levels of ML2G-relatedsequences, including full-length 8-kb proviruses, were detectedin Chinese hamster liver but not in BHK, A9, or 293 cells (Fig.6). Extended film exposure times did indicate that distantlyrelated sequences exist in BHK cells (data not shown). Thedata indicate that full-length ML2G-related proviruses existnaturally as endogenous elements within the Chinese hamstergerm line. The hybridization profiles in Fig. 6 also identifyseveral prominent EcoRV fragments in CHO cells that are notpresent in Chinese hamster liver.By using PCR and oligonucleotide sequence primers de-

signed from the B52 sequence (see Materials and Methods),we were able to amplify B52 LTR-related sequences fromCHO cell DNA. Hybridization probes prepared from ampli-fied B52-related sequences or the ML2G LTR hybridizedequally well to CHO cell DNA, in each case producingheterogeneous smearing patterns (Fig. 6). Note, however, thatthe 8-kb EcoRV fragment representing the full-length proviruswas recognized only by the ML2G LTR probe and not by theB52 LTR probe. This observation indicates that the two probesrecognize different sequences and, therefore, that CHO type Cretrovirus elements may fall into at least two distinct groups:those with ML2G-like LTRs and those with B52-like LTRs.

Full-length type C proviruses are transcribed in CHO cells.To characterize type C retrovirus transcription in CHO cells,we analyzed poly(A)+ RNA prepared from CHO Kl andhuman 293 cells. Northern (RNA) transfer analysis was per-formed with gag-, pol-, and env-specific hybridization probes.The data shown in Fig. 7 demonstrate that type C retrovirusRNA is expressed in CHO cells but not in the 293 cell line.CHO cells contain two major RNA species that are approxi-mately 8 and 3.5 kb long. The 8-kb transcript was detected byusing gag, pol, and env hybridization probes, consistent with thestructure of full-length type C retrovirus genomic RNA. The3.5-kb RNA was detected only with the env-specific hybridiza-tion probe, consistent with the structure of spliced subgenomicenv mRNA. We further examined RNA expression in sixadditional CHO cell clones, and in every case, we were able to

VOL. 68, 1994

on August 3, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

7846 LIE ET AL.

Avra.ml2gvra.amphovra.lOalvra.felvvra.xenovra.poly

Bvrb.ml2gvrb.amphovrb.lOalvrb.felvvrb.xenovrb.poly

1 SDWDPSDQEPFPGYGCHHPGGRIGTRSKDFYVCLSGHRPTHDI1EE WDPSDOEPYVGYGC KY PG R OT R TFDFYVC - PG H TV KS G1EEWDPSDOEPL GYGCKYPGGRKRRRTFDFYVC-P GHTVKSGiNT NPSDOEPFPGYGCDOPWjRWO RNT FYVC-PGHANRKO1DYWD--DEPDIGDGCRTPGGRRR RLY DFYVC-PGHTVPIG1DDWD-- ETG L GCRTPGGRKRA TF DFYVC-P GHTVPTG

111

111

. PG Y AG K G P WR C GO RA C DPVMIVG G F GATPEE

. PWDT CRSKVA CG -PCY DLSKVSNS|FOGATW

OGi I IOCSIEGGWC G- PCY D KA- VHS STTGAEG-PKDO------ G- PCYDSS-VSSDI KGATG- ]RN - - - - - - G- PCY DSSAVSSD I KGAT P G

FIG. 5. Predicted amino acid sequence encoded by ML2G env regions. The ML2G sequence was used to predict the amino acid sequencescorresponding to variable regions A (vra) and B (vrb) of env (4). The predicted sequences are aligned with the variable region A and B sequencesof MLVs that exhibit different nonecotropic env receptor specificities (lOA1, amphotropic [ampho], xenotropic [xeno], and polytropic [poly]) andFeLV(B) (felv). Amino acid positions that are identical to the predicted sequence of ML2G are boxed.

detect the 8- and 3.5-kb RNAs (data not shown). When filmexposure times were extended, the gag probe also detected twoless abundant transcripts that were approximately 4.5 and 3.5kb long.The probe prepared from the ML2G LTR region did not

hybridize as efficiently to the 8- and 3.5-kb transcripts as didthe gag, pol, and env probes (Fig. 7). In addition, this probeidentified three additional transcripts (approximately 9.5, 4.5,and 2.5 kb) that were not seen with the other probes. Inter-estingly, the hybridization pattern generated from the B52LTR probe more closely resembled the pattern we observedwith the gag-, pol-, and env-specific probes (Fig. 7). This probehybridized specifically to the 8- and 3.5-kb transcripts and didnot hybridize to the three additional RNAs detected with theML2G LTR probe. The data suggest that the prominent 8- and3.5-kb RNAs expressed in CHO cells may be transcribed fromelements that are closely related to the ML2G sequence in thegag, pol, and env regions but contain LTRs more similar to theB52 sequence.To assay the transcriptional activity of the ML2G LTR

directly, we subcloned the LTR sequences into a reporterplasmid containing the bacterial CAT gene and introduced theLTR-CAT plasmid into several CHO cell lines by DNAtransfection. The results of several transient CAT expressionassays indicate that the ML2G LTR functions poorly in CHOcells in comparison with the MLV LTR or the simian virus 40promoter-enhancer region (data not shown). In contrast, theB52 LTR has been shown previously to act as a strongtranscriptional promoter in CHO cells (45).

Integration of the ML2G provirus interrupted a SINE. Toevaluate the site of ML2G integration, the cellular sequencesimmediately flanking the ML2G provirus were screenedagainst sequences in the GenBank database. This analysisidentified the sequences flanking both the 5' and 3' LTRs ashamster G repeat sequences (Fig. 8A). G repeats (26) belongto the larger short interspersed element (SINE) sequencefamily (8). On the basis of the sequence organization of theML2G locus, it appears that the ML2G retrovirus integratedinto the Chinese hamster genome within a G element member(Fig. 8B).

DISCUSSION

We have continued to investigate the origin of retrovirus-like particles in CHO cell lines by isolating and characterizinga genomic locus that contains a full-length provirus. Theoverall nucleotide sequence of this provirus indicates that it isrelated to murine type C retrovirus sequences. We were able toidentify several features commonly associated with rodent typeC retrovirus integration and replication. These include a PBS

. L. 1.

cnq=y ,q XC'CO CNo ,.t., 0 0I :3x:I n = MM :x: = oxx =SCnII ^I r) =R OOms IZOt)XE IVOOE RC O NO

kB

23.1

9.4 i6.6 X

2.3 '|2.0 J

gag po1 env LTR LTR(ML2G) (B52)

FIG. 6. Detection of full-length type C proviruses in CHO cells.Southern transfer hybridization analysis was used to test for thepresence of ML2G proviral sequences in DNAs prepared from CHOcells (CHO), Chinese hamster liver (CH liver), a Syrian hamsterkidney cell line (BHK), human 293 cells (Hu 293), and mouse A9 cells(Mu A9). DNA was digested with EcoRV, which cuts once in the 5'LTR and once in the 3' LTR, and the digestion products wereseparated by electrophoresis in agarose gels. DNA was transferred tonitrocellulose membranes. Membranes were incubated with radiola-beled hybridization probes consisting of the ML2G LTR and gag, pol,and env regions (Fig. 1) and amplified B52-related LTR sequences.The 8-kb EcoRV fragment detected in CHO cells by all four ML2Gsequence probes represents full-length proviruses belonging to theML2G family. XH3, HindIll-digested X DNA marker.

J. VIROL.

on August 3, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

ENDOGENOUS TYPE C RETROVIRUS EXPRESSION IN CHO CELLS 7847

Cen cn en I C.)

- iX

kB

8.0-

3.5 :.:

so

o oz zC.) U.

10 20 30 40 50A ml2g. flank.rc TCTCTGTGTGTGTGACCAATTAATTTTCTCAACTTTTAATTTGAATTAGA

gsinespcl910 20 30 40 50

60 70 80 90ml2g. flank.rc AACAGCATTG-TTTACATGTCAATCTCAGAGGGAGGCCCTCTTCCCCTGC

gsinespcl9 AACAATCTTGTTTTACTTATCAATCCCAG ----- TTCCCTCTCC-----C60 70 80 90

4'a 100 110 120 130 140ml2g. flank.rc CCCTCTCCAACTAACACCCTACCTATCCCATATCCTTTCTGGCTCCCCAA

**CC**O**C*C***AA*C**********T*******T-********gsinespcl9

ml2g . flank.rc

gag pol env B-actinML2G B52

LTR

FIG. 7. Detection of type C retrovirus transcription in CHO cells.Northern transfer hybridization analysis was used to test for thepresence of type C retrovirus RNA in CHO cells (CHO) and thehuman 293 cell line (293). Polyadenylated RNA fractions were sepa-

rated by electrophoresis in agarose gels, transferred to nitrocellulosemembranes, and incubated with radiolabeled hybridization probesspecific for the ML2G gag, pol, and env regions (A) and the ML2G andB52 LTRs (B). A control hybridization using a ,B-actin probe is alsoshown. Sizes of the full-length genomic RNA (8.0 kb), the spliced env

mRNA (3.5 kb), and control ,B-actin RNA (2.0 kb) are shown. RNAsizes were estimated on the basis of comparative migration of an RNAladder set (0.24 to 9.5 kb; Gibco/BRL).

that predicts tRNAPrO as a primer for negative-strand DNAsynthesis, a four-base duplication of the cellular sequence atthe site of retrovirus integration, and translation of the polregion by suppression of a stop codon within the gag-polboundary. Our sequence analyses indicate that this proviralsequence is unrelated to the CHO type A sequences previouslyreported by workers in this laboratory (1) and others (10, 35).The provirus characterized in this study contains a largenumber of nucleotide substitutions that have disrupted boththe gag-pol and env reading frames, as well as the tRNA PBS.In addition, the ML2G LTR did not exhibit significant pro-moter-enhancer activity when tested in a CAT reporter geneassay. For these reasons, we conclude that the ML2G provirusitself cannot produce type C particles. However, the isolationof ML2G demonstrates that CHO cells do contain full-lengthretrovirus elements. Thus, it is not unreasonable to speculatethat some of these proviruses are capable of producing the typeC retrovirus-like particles that have been observed buddingfrom CHO cells (22, 23, 39) and that are immunologicallyrelated to mammalian type C retroviruses (2).To characterize type C retrovirus RNA expression in CHO

cells, we constructed LTR, gag, pol, and env hybridizationprobes from cloned ML2G proviral sequences. We were ableto detect two major retroviral RNA transcripts in CHO celllines. The 8-kb RNA contains LTR, gag, pol, and env se-

quences, and therefore it is most likely full-length, unsplicedgenomic RNA. The 3.5-kb RNA was detected only with env

and LTR probes, consistent with the structure of subgenomic,spliced env mRNA. Further analyses of these two RNAs withLTR sequence probes suggest that many of the 8- and 3.5-kbtranscripts contain LTR sequences that are more homologousto the B52 cDNA sequence (isolated from CHO cells by Zhuet al. [45]) than to the ML2G LTR. In the latter study, the B52cDNA was reported to detect abundant levels of 5.2- and2.7-kb transcripts in CHO cells and not the 8- and 3.5-kb

TCCCCTGCACCCCACAGACCCCCTATCCCATCCCCTTTCT -GCTCCCCAT100 110 120 130

150 160 170 180 190GGGAAGGGTGAAGGCCTTACATAGGGGGTCTTCAGAGTCTGTCATATCCT

gsinespcl9 GG--AGGGTG-AGTCCTTCCAAGGGAGATCTTCAAAGTCTGTCATATCAT140 150 160 170 180

200 210 220 230 240ml2g. flank.rc TTCGA--TAGGCCTAGGCCCAACCCTGTGTGTCTAGCTAGGCTCAGGGAG

gsinespcl9 TTGGAGCAGGGTCTAGGCCCTTACCCTTGTGT ----CTAGGCTGAGAGAG190 200 210 220 230

250 260 270 280 290ml2g. flank. rc TATCCCCTCTATGTGGAATGGGGCTTGAAAAGTCCATTCCTATGCTAGGG

gsinespcl9 TAT-CCCTCTATGTGGGAT-GAGCTCCCAAAGTCCATTTGTATGCTAGGG240 250 260 270 280

ml2g. flank.rc

gsinespcl9

300 310 320 330 340ATAAGTACTGATCTACTACAACCAACTAATTTTAACCAGGAGCTTCTGTG

ATAAATACTGA-GTCGCCTAAAGAAGAGGGGATAGCTATTGCCTTCTGAC290 300 310 320

350 360 370 380ml2g. flank.rc TGGCTGTAGATTTGTAACTATCCACTGGAGTCTGGT

gsinespcl9 ATG--AAAGGTGAGTGACTGGTGCTTTGACTGTACC330 340 350 360

B _ ----_

rG-SINE

FIG. 8. ML2G retrovirus integration within a SINE. The sequence

immediately flanking the ML2G provirus was converted to its reverse

complement and then aligned with the sequence of the spcl9 G

element (GenBank accession number, X15420), a member of the

SINE sequence family (A). The relative position of the ML2G proviruswithin the spcl9 G element is shown schematically (B). Horizontal

arrows designate the 5 '-to-3' orientation of the provirus and G

element sequences.

transcripts that we report here. Since the B52 cDNA contains

only partial env and LTR sequences, the researchers were

unable to describe fully the sequence composition of the 5.2-

and 2.7-kb transcripts they detected.

By comparing the B52 and ML2G sequences, we demon-

strated that the two are closely related in the env TM region

and the R region of the LTR but are more divergent in the U3

regions of the LTR. By using a B52 LTR-CAT reporter

construct in transient transfection assays, Zhu et al. (45) have

shown that the B52 LTR sequence can function as a potent

transcriptional promoter-enhancer. In contrast, we did not

detect promoter-enhancer activity by using the ML2G LTR

sequence in similar experiments. On the basis of the data

available (from Northern analyses and CAT assays), we pro-

pose that the type C retrovirus RNAs expressed in CHO cells

are produced from endogenous elements that contain gag, pci,

A

kB

8.0-3.5-

2.0-

VOL. 68, 1994

on August 3, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

7848 LIE ET AL.

and env sequences related to the ML2G sequence but possessLTRs similar to the B52 sequence. To test this hypothesis andto evaluate the protein-coding potential of the 8- and 3.5-kbtranscripts, we have begun to characterize a number of cDNAclones generated from a CHO cell line which expresses type Cretroviral RNA and protein at high levels.

In a previous study, the number of type C retrovirussequences in CHO cells was estimated at approximately 100 to300 copies per cell (2). The identification of 19 positive clonesfrom a library screen, representing 10 to 20% of the CHO cellgenome, is in good agreement with this earlier finding. In thecurrent study, we were able to demonstrate that 10 to 20% ofthe CHO sequences recognized by type C hybridization probesexist in full-length provirus structures. Together, these obser-vations suggest that CHO cells contain 10 or more full-lengthproviruses related to the ML2G provirus (i.e., they hybridize toML2G gag, pol, env, and LTR probes and contain an EcoRVsite in the 5' and 3' LTRs). The results of both Southern andNorthern blot analyses presented in this study, as well as thesequence of the B52 cDNA (45), demonstrate that type Celements containing LTR sequences distinct from ML2G (i.e.,B52 like) also exist in CHO cells and are actively transcribed.We found that EcoRV digpstion of CHO cell DNA pro-

duced several subgenomic-length fragments that were notproduced by digestion of Chinese hamster liver DNA. Thesefragments suggest that type C elements with significant dele-tions may have accumulated in CHO cell lines during theirestablishment in culture. Alternatively, they may simply reflecta polymorphic distribution of EcoRV sites within Chinesehamster type C proviruses. Others have reported finding nodifferences in the Southern hybridization profiles of type Aretrovirus sequences in CHO cells and Chinese hamster liver(10).A comparative analysis of the env SU sequences of the

ML2G provirus indicates that this CHO retrovirus sequenceresembles amphotropic MLV. The evolutionary origin of themurine amphotropic env gene is unknown. In nature, exoge-nous amphotropic retrovirus infection is limited to distinctareas in southern California (14, 34). Genetic studies haveshown that murine amphotropic retroviruses are not present inthe germ line of the North American mouse (Mus musculusdomesticus) population (29). Similarly, we were unable todetect CHO env-related sequences in the murine A9 cell line.CHO env sequences also resemble the env region of FeLV(B).This finding is not surprising considering the similarity betweenamphotropic MLV and FeLV(B) env sequences noted previ-ously (29). Unlike the amphotropic env sequences of mice,endogenous FeLV(B) env sequences have been detected in thegerm line of cats (37).

During the characterization of an endogenous type C pro-virus in CHO cells, we discovered that sequences flanking theprovirus share homology with the previously described hamsterG elements. G elements (262) are a group of short (<500-bp),moderately repetitive (>10 copies per haploid genome) se-quence elements belonging to the larger SINE sequence family(8). SINEs are mobile genetic elements which replicatethrough RNA intermediates. Members of the SINE sequencefamily are ubiquitous genetic components of the eukaryoticgenome and appear to be distributed fairly randomly (oneelement every 5,000 bp, on average). The sequence organiza-tion of the ML2G locus reflects retroviral integration within aG element of the Chinese hamster genome (Fig. 8B).

ACKNOWLEDGMENTSWe acknowledge Steve Williams for constructing the genomic

library, Marshall Dinowitz and Florian Wurm for enthusiastic support

of this project, the Bioorganic Chemistry group at Genentech forsupplying primers for sequencing and PCR, Allison Bruce for com-puter graphics assistance, and Colin Watanabe and Wolfgang Hoeckfor assistance with DNA and amino acid sequence analysis.

REFERENCES1. Anderson, K. P., Y. S. Lie, M. L. Low, S. R. Williams, E. H. Fennie,

T. P. Nguyen, and F. M. Wurm. 1990. Presence and transcriptionof intracisternal A-particle-related sequences in CHO cells. J.Virol. 64:2021-2032.

2. Anderson, K. P., M. L. Low, Y. S. Lie, G.-A. Keller, and M.Dinowitz. 1991. Endogenous origin of defective retroviruslikeparticles from a recombinant Chinese hamster ovary cell line.Virology 181:305-311.

3. Ausubel, F. M., R. Brent, R. E. Kingston, D. D. Moore, J. G.Seidman, J. A. Smith, and K. Struhl. 1994. Current protocols inmolecular biology. Greene Publishing Associates and J. Wiley &Sons, Inc., New York.

4. Battini, J.-L., J. M. Heard, and 0. Danos. 1992. Receptor choicedeterminants in the envelope glycoproteins of amphotropic, xeno-tropic, and polytropic murine leukemia viruses. J. Virol. 66:1468-1475.

5. Cloyd, M. W., M. M. Thompson, and J. W. Hartley. 1985. Hostrange of mink cell focus-inducing viruses. Virology 140:239-248.

6. Coffin, J. M. 1991. Retroviridae and their replication, p. 645-708.In B. N. Fields, D. M. Knipe, et al. (ed.), Fundamental virology,2nd ed. Raven Press, New York.

7. Collen, D., J. M. Stassen, B. J. Marafino, Jr., S. Builder, F.DeCock, J. Ogez, D. Tahiri, D. Pennica, W. F. Bennet, J. Salwa,and C. F. Hoyng. 1984. Biological properties of human tissue-typeplasminogen activator obtained by expression of recombinantDNA in mammalian cells. J. Pharm. Exp. Ther. 231:146-152.

8. Deininger, P. L. 1989. SINEs: short interspersed repeated DNAelements in higher eucaryotes, p. 619-636. In D. E. Berg andM. M. Howe (ed.), Mobile DNA. American Society for Microbi-ology, Washington, D.C.

9. Dinowitz, M., Y. S. Lie, M. L. Low, R. Lazar, C. Fautz, B. Potts, J.Sernatinger, and K. Anderson. 1991. Recent studies on retrovirus-like particles in Chinese hamster ovary cells. Dev. Biol. Stand.76:201-207.

10. Dorner, A. J., F. Bonneville, R. Kriz, K. Kelleher, K. Bean, andR. J. Kaufman. 1991. Molecular cloning and characterization of acomplete Chinese hamster provirus related to intracisternal Aparticle genomes. J. Virol. 65:4713-4719.

11. Egrie, J. C., J. Browne, D. Lai, and F. K. Lin. 1985. Characteriza-tion of recombinant monkey and human erythropoietin. Prog.Clin. Biol. Res. 191:339-350.

12. Elbien, A. D. 1991. The role of N-linked oligosaccharides inglycoprotein function. Trends Biotech. 9:346-352.

13. Gottesman, M. M. 1985. Molecular cell genetics. J. Wiley & Sons,Inc., New York.

14. Hartley, J. W., and W. P. Rowe. 1976. Naturally occurring murineleukemia viruses in wild mice: characterization of a new "ampho-tropic" class. J. Virol. 19:19-25.

15. Hatfield, D. L., J. G. Levin, A. Rein, and S. Oroszlan. 1992.Translational suppression in retroviral gene expression. Adv.Virus Res. 41:193-239.

16. Heine, U. I., B. Kramarsky, E. Wendel, and G. Suskind. 1979.Enhanced proliferation of endogenous virus in Chinese hamstercells associated with microtubules and the mitotic apparatus of thehost cell. J. Gen. Virol. 44:45-55.

17. Hojman, F., R. Emanoil-Ravier, J. Lesser, and J. Peries. 1989.Biological and molecular characterization of an endogenous ret-rovirus present in CHO/HBs-A Chinese hamster cell line. Dev.Biol. Stand. 70:195-202.

18. Hubbard, R. C., N. G. McElvaney, P. Birrer, S. Shak, W. W.Robinson, C. Jolley, M. Wu, M. S. Chernick, and R. G. Crystal.1992. A preliminary study of aerosolized recombinant humandeoxyribonuclease I in the treatment of cystic fibrosis. N. Engl. J.Med. 326:812-815.

19. Jacks, T. 1990. Translational suppression in gene expression inretroviruses and retrotransposons. Curr. Top. Microbiol. Immu-nol. 157:93-124.

J. VIROL.

on August 3, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

ENDOGENOUS TYPE C RETROVIRUS EXPRESSION IN CHO CELLS 7849

20. Kuff, E. L., and K. K. Leuders. 1988. The intracisternal A-particlegene family: structural and functional aspects. Adv. Cancer Res.51:183-276.

21. Laimins, L. A., P. Gruss, R Pozzatti, and G. Khoury. 1984.Characterization of enhancer elements in the long terminal repeatof Moloney murine sarcoma virus. J. Virol. 49:183-189.

22. Leiber, M. M., R E. Benveniste, D. M. Livingston, and G. J.Todaro. 1973. Mammalian cells in culture frequently release typeC viruses. Science 182:56-59.

23. Manley, K. F., J. F. Givens, R L. Taber, and R. F. Zeigel. 1978.Characterization of virus-like particles released from the hamstercell line CHO-Kl after treatment with 5-bromodeoxyuridine. J.Gen. Virol. 39:505-517.

24. Marino, M. H. 1989. Expression systems for heterologous proteinproduction. Biopharm 2:18-33.

25. Mietz, J. A., Z. Grossman, K. K. Leuders, and E. L. Kuff. 1987.Nucleotide sequence of a complete mouse intracisternal A-parti-cle genome: relationship to known aspects of particle assemblyand function. J. Virol. 61:3020-3029.

26. Miles, C., and M. Meuth. 1989. G-repeats: a novel hamster SINEfamily. Nucleic Acids Res. 17:7221-7228.

27. Miller, A. D., J. V. Garcia, N. von Suhr, S. M. Lynch, C. Wilson,and M. V. Eiden. 1991. Construction and properties of retroviruspackaging cells based on gibbon ape leukemia virus. J. Virol.65:2220-2224.

28. Morgan, R A., 0. Nussbaum, D. D. Muenchau, L. Shu, L.Couture, and W. F. Anderson. 1993. Analysis of the functional andhost range-determining regions of the murine ecotropic andamphotropic retrovirus envelope proteins. J. Virol. 67:4712-4721.

29. O'Neill, R R, J. W. Hartley, R. Repaske, and C. A. Kozak. 1987.Amphotropic proviral envelope sequences are absent from theMus germ line. J. Virol. 61:2225-2231.

30. Ono, M., H. Toh, T. Miyata, and T. Awaya. 1985. Nucleotidesequence of the Syrian hamster intracisternal A-particle gene:close evolutionary relationship of type A particle gene to types[sic] B and D oncovirus genes. J. Virol. 55:387-394.

31. Ott, D., R Friedrich, and A. Rein. 1990. Sequence analysis ofamphotropic and 1OAl murine leukemia viruses: close relation-ship to mink cell focus-inducing viruses. J. Virol. 64:757-766.

32. Ott, D., and A. Rein. 1992. Basis for receptor specificity ofnonecotropic murine leukemia virus surface glycoprotein gp7OSu.J. Virol. 66:4632-4638.

33. Patzer, E. J., G. R. Nakamura, R D. Hershberg, T. J. Gregory, C.Crowley, A. D. Levinson, and J. W. Eichberg. 1986. Cell culture

derived recombinant HBsAg is highly immunogenic and protectschimpanzees from infection with hepatitis B virus. Bio/Technology4:630-636.

34. Rasheed, S., M. B. Garner, and E. Chan. 1976. Amphotropic hostrange of naturally occurring wild mouse leukemia viruses. J. Virol.19:13-18.

35. Servaney, M., J. J. Kupiec, L. d'Auriol, F. Galibert, J. Peries, andR Emanoil-Ravier. 1988. Nucleotide sequence of the Chinesehamster intracisternal A-particle genomic region corresponding to5'LTR-GAG. Nucleic Acids Res. 16:7725.

36. Shinnick, T. M., A. A. Lerner, and J. G. Sutcliffe. 1981. Nucleotidesequence of Moloney murine leukemia virus. Nature (London)293:543-548.

37. Stewart, M. A., M. Warnock, A. Wheeler, N. Wilkie, J. I. Mullins,D. E. Onions, and J. C. Neil. 1986. Nucleotide sequences of afeline leukemia virus subgroup A envelope gene and long terminalrepeat and evidence for the recombinational origin of subgroup Bviruses. J. Virol. 58:825-834.

38. Teich, N. 1984. Taxonomy of retroviruses, p. 25-207. In R. Weiss,N. Teich, H. Varmus, and J. Coffin (ed.), RNA tumors viruses, 2nded. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y.

39. Tihon, C., and M. Green. 1973. Cyclic AMP-amplified replicationof RNA tumour virus-like particles in Chinese hamster ovary cells.Nature (London) New Biol. 244:227-231.

40. Vapnek, D. 1990. Choosing a production system for recombinantproteins, p. 1-14. In M. D. White, S. Reuveny, and A. Shafferman(ed.), Biologicals from recombinant microorganisms and animalcells: production and recovery. VCH Publishers, New York.

41. Varmus, H. E. 1983. Retroviruses, p. 411-503. In J. A. Shapiro(ed.), Mobile genetic elements. Academic Press, Inc., New York.

42. Wheatley, D. N. 1974. Pericentriolar virus-like particles in Chinesehamster ovary cells. J. Gen. Virol. 24:395-399.

43. Wiebe, M., F. Becker, R Lazar, L. May, B. Casto, M. Semense, C.Fautz, R Garnick, C. Miller, G. Masover, D. Bergmann, and A. S.Lubiniecki. 1989. A multifaceted approach to assure that recom-binant t-PA is free of adventitious virus, p. 68-71. In R. E. Spier,J. B. Griffith, J. Stephenne, and T. J. Crooy (ed.), Advances inanimal cell biology and technology for bioprocesses. Butterworths,London.

44. Wurm, F. M. 1990. Integration, amplification and stability ofplasmid sequences in CHO cell cultures. Biologicals 18:159-164.

45. Zhu, W., M. Kriajevskaia, and W.-G. Chou. 1992. A retroviralsequence of the Chinese hamster ovary cell line. Oncogene7:2081-2083.

VOL. 68, 1994

on August 3, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from