Ž .Mutation Research 374 1997 145–148

Short communication

Moloney murine leukemia reverse transcriptase suspect in theproduction of multiple misincorporations during hprt cDNA

synthesis

John Curry 1, Barry W. Glickman )

Centre for EnÕironmental Health and the Department of Biology, UniÕersity of Victoria, P.O. Box 3020, Victoria, British Columbia, V8W3N5, Canada

Received 26 September 1996; revised 15 November 1996; accepted 25 November 1996

Abstract

Our laboratory has characterized several hundred mutant hprt cDNAs produced using Moloney murine leukemia reversetranscriptase to convert mRNA to cDNA. During the characterization of these mutants we have detected six T-lymphocytemutants that demonstrate multiple G:C™A:T transitions along the hprt cDNA coding sequence. Attempts to repeat themRNA to cDNA conversion and subsequent characterization have demonstrated that the multiple transitions are likelyartifacts. We suggest that reverse transcriptase is directly responsible for these multiple base substitutions and as such, thatmultiple mutations be viewed as suspect requiring confirmation at the genomic level.

Keywords: Mo-MLV; Reverse transcriptase; hprt

1. Communication

The monitoring of in vivo human mutations isfacilitated by the use of the hypoxanthine–guanine

Ž .phosphoribosyltransferase HPRT T-cell clonal as-w xsay 1 . Not only can in vivo mutant frequencies be

determined, the mutations responsible for the ob-Ž R .served 6-thioguanine resistant TG phenotype can

also be characterized at the DNA sequence level.

) Ž . Ž .Corresponding author. Tel.: 1 250 472 4067; Fax: 1 250472 4075; E-mail: bwglick@uvic.ca

1 Ž . Ž .J.C.: Tel.: 250 472-4067; Fax: 250 472-4075; E-mail:jcurry@uvic.ca

This approach can provide useful data on the natureand origin of mutations occurring in vivo. However,the properties of the enzymes used in the molecularcharacterization of mutant hprt mRNAs require somecircumspection. Specifically, the cDNAs are pro-duced by the reverse transcription of mRNAs con-

w x w xtained in cell lysates 2 or mRNA extracts 3 .Reverse transcriptases commonly used to transcribecDNAs, such as the Moloney murine leukemia virusŽ . Ž .Mo-MLV and avian myeloblastosis virus AMV

Ž .reverse transcriptase RT , are characterized by highw xerror-rates 4–6 . Using the M13mp2 lacZa gene as

template, both RTs were estimated to commit anerror at least once every 30 000 nucleotides polymer-

w xized 4 . Indeed, the relatively high in vivo mutation

0027-5107r97r$17.00 Copyright q 1997 Elsevier Science B.V. All rights reserved.Ž .PII S0027-5107 96 00251-5

( )J. Curry, B.W. GlickmanrMutation Research 374 1997 145–148146

rates characteristic of these retroviruses are thoughtto be a direct consequence of this error proneness.

In principal, errors that occur during the reversetranscription of hprt mRNAs should go undetectedbecause cell lysates or extracts are derived from a

Ž 3 – 5.large number of cells 2=10 , each containingw xapproximately 8–10 hprt mRNA molecules 7 . Two

Ž .subsequent polymerase chain reactions PCR di-w xrected with two sets of nested primers 2 are used to

amplify the cDNA providing sufficient amounts ofw xproduct for direct double-stranded sequencing 8 . It

would thus be expected that errors produced byMo-MLV RT would be diluted out by other copiesnot having a similar error.

Nevertheless, over the past several years we haveupon occasion recovered human hprt mutant se-quences that we suspect were the consequence ofMo-MLV RT mis-incorporation errors. Using themethod briefly described above, as modified by Curry

w xet al. 9,10 , multiple G:C™A:T transitions occur-ring in a single cDNA product have been character-ized from several TGR T-lymphocyte clones. Todate we have observed six separate cDNA character-

izations where such multiple base substitutions haveŽ .occurred Table 1 . Attempts to repeat these observa-

tions by generating new cDNA from fresh extracts ofthe same mutant clones proved unsuccessful. In con-

Žtrast, when the original RT-PCR products primary. Ž .PCR were re-amplified secondary PCR and char-

acterized, the same multiple mutations were againencountered. From these and other observations, webelieve that the multiple transitions are an artifactoccurring in the RT phase and hence generated bythe MoMLV reverse transcriptase.

The majority of the multiple base substitutions, inthe cases that we have characterized, involve GA

Ž .dinucleotides 26 of 37 substitution at G:C basepairs ,while other dinucleotides are involved less fre-

Ž . Ž . Ž .quently; GG 8r37 , GT 2r37 , GC 1r37 . Usinga Chi-square test, with expected values for all four

Ž .dinucleotides being equal 9.25 , the observed valuesŽ 2 y7.are highly significant x s43.5, dfs3, p-10

and demonstrate a predilection for the mechanismresponsible here to act at GA dinucleotides. Forthose events occurring at GA dinucleotides, the pre-ceding nucleotide appears not to have any significant

Table 1Position and the sequence context of multiple base substitutions characterized from TGR T-cell clones

17-10 B9-6 17-15

89 GCTGAGG 96 TTTGGAA 306 ACTGAAG97 TTGGAAA 101 AAAGGGT 384 AAAGAAT

361 GATGATC 118 CATGGAC 432 GCAGACT452 TCAGGCA 197 TCTGTGT FB16B

469 AAAGATG 403 GAAGATA 361 GATGATC472 ATGGTCA 438 TTTGCTT 393 CTTGATT518 TTGGATA 492 GCTGGTG 412 ATTGACA553 CCAGACA 569 TAGGATA 432 GCAGACT

C580 CTTGACT 599 TCAGGGA 606 TTTGAAT628 AGTGAAA 601 TTTGAAT664 TGAGAGT WX1-9

674 GTTGAGT 272 ATAGATC17-7 285 TATGACT

A601 AGGGATT 334 ACAGGGG606 TTTGAAT 352 ATTGGTG

400 GTGGAAGB581 TTGACTA

ŽA . ŽB. ŽC .All of these mutations are G : C ™ A : T transitions except for three transversions; is a G : C ™ C : G; is a A : T ™ T : A; is aG : C ™ T : A

( )J. Curry, B.W. GlickmanrMutation Research 374 1997 145–148 147

consequence upon the misincorporations; TGAŽ 213r26, AGA 8r26, GGA 5r26 x s3.8, dfs2,

.p-0.14 .We can not explain how in an excess of template

the MoMLV RT reaction can produce products suchw xas the six we describe here. Pathak and Temin 11

report a similar phenomenon which they have termedhypermutation, while determining the in vivo for-ward mutation rate during a single round of replica-tion of the spleen necrosis virus. Characterization ofproviruses revealed 15 G™A transitions which areproposed to be the direct results of mis-incorporationduring viral reverse transcription. Two of the substi-tutions appear to involve tandem events where GGdinucleotides in the sequence GGGAA are replacedwith AA dinucleotides. Nine of the fifteen single-basepair substitutions occur at GA dinucleotide se-quences, compared with 2r15 at GT and 2r15 atGG. None of the substitutions are preceded by Anucleotides. A simple Chi-square analysis demon-strates a significant preference for GA dinucleotidesŽ 2 .x s11.9, dfs3, p-0.008 . This GA dinucleo-tide preference is the same as we observed in ourhprt study.

w xJi and Loeb 12 employing an in vitro system,report several multiple mutations produced by theHIV-1 RT using the enÕ gene as template. The

Žauthors observed nine multiple mutations with ei-.ther 2 or 3 mutations each along of the enÕ gene

templates. In addition, the HIV-1 RT enzyme wasestimated to have an error rate of approximately oneerror per 5000 RNA nucleotides copied and of the 64mutations, 76% were transitions with the most fre-quent type being G:C™A:T.

We have not been able to develop the conditionsrequired for the RT-PCR step to reproducibly gener-ate multiply misincorporated cDNAs. Such cDNAshave been observed seemingly randomly during thecourse of many different experiments, over severalyears. Since they are rare events, we have not soughtto develop a more error-free system. It is possiblethat an occasional clonal lysate contains much feweraccessible cells and thus very few mRNA moleculesfor reverse transcription. With just a small number ofmRNA molecules, RT errors may not be ‘washed’out by correct copies. This does not, however, ex-plain the multiple nature of these transitions. Perhapsa defective or inefficient RT enzyme inefficiently

and quite inaccurately copies only a few mRNAmolecules, thereby producing not only a cDNA withmultiple transitions, but one that predominates suffi-ciently to be scored at the sequencing level.

We can however, not rule out the possibility thatin the initial PCR step, it is the synthesis of the 2ndcDNA strand that is the source of the multiplesubstitutions. Using primers specific for exon 3 ofhprt, Taq polymerase was found to predominantly

w xproduce A:T™G:C transitions 13 rather than theG:C™A:T transitions. Moreover, there have beenno reports of multiple misincorporations with thisenzyme. As the majority of the transitions reportedhere involve G:C basepairs, the possibility of DNAdamage due to incubation of the template at hightemperatures also exists. Deamination of cytosine touracil is the most frequent DNA damage caused in

wthis manner, producing G:C™A:T transitions 14–x16 . Nevertheless the probability that multiple DNA

damages occur on a single template does seem veryunlikely.

An additional factor may be that Taq polymeraseexhibits a template preference, and for some reasonpreferentially amplifies the altered templates duringthe course of the amplification. This would requirethat a template altered by MoMLV RT would bepreferentially amplified by Taq during PCR amplifi-cation. Under such circumstances the mutated tem-plate would be over represented and preferentiallyvisible in the subsequent sequencing. While our in-ability to consistently reproduce these observationsprevents us from exploring their cause in detail, wefelt it important to report their recovery. Our firstcase of these multiple events arose while followingthe hprt mutant frequency in someone exposed to apotentially genotoxic work environment. These ob-servations thus led to unnecessary fears. Also, inaddition to acknowledging these rare events, somespeculation on their origin seemed worthwhile.

Acknowledgements

This work was supported by a grant to Dr. BarryGlickman from the Medical Research Council ofCanada. We thank Aparecido da Cruz and MagomedKhaidakov for isolating and characterizing the WX1-9 and FB16B mutants. The criticisms of Dr. Elliot

( )J. Curry, B.W. GlickmanrMutation Research 374 1997 145–148148

Drobetsky and the encouragement of Dr. AdonisSkandalis were greatly appreciated.

References

w x Ž .1 Albertini, R.J., K.L. Castle and W.R. Borcherding 1982T-cell cloning to detect the mutant 6-thioguanine-resistantlymphocytes present in human peripheral blood, Proc. Natl.Acad. Sci. USA., 79, 6617–6621.

w x Ž .2 Yang, J.L., V.M. Maher and J.T. McCormick 1989 Ampli-fication and direct sequencing of cDNA from the lysate oflow numbers of diploid human cells, Gene, 83, 347–354.

w x3 Steingrimsdottir, H., G. Rowley, A. Waugh, D. Beare, J.Ž .Cole and A.R. Lehmann 1993 Molecular analysis of muta-

tions in the hprt gene in circulating lymphocytes fromnormal and DNA-repair-deficient donors, Mutation Res., 294,29–41.

w x4 Roberts, J.D., B.D. Preston, L.A. Johnston, A. Soni, L.A.Ž .Loeb and T.A. Kunkel 1989 Fidelity of two reverse tran-

scriptases during DNA-dependent DNA synthesis in vitro,Molec. Cell. Biol., 9, 469–476.

w x Ž .5 Bakhanashvili, M. and A. Hizi 1992 Fidelity of the RNA-dependent DNA synthesis exhibited by the reverse transcrip-tases of human immunodeficiency virus types 1 and 2 and ofmurine leukemia virus: Mispair extension frequencies, Bio-chemistry, 31, 9393–9398.

w x Ž .6 Bakhanashvili, M. and A. Hizi 1993 The fidelity of thereverse transcriptases of human immunodeficiency virusesand murine leukemia virus, exhibited by the mispair exten-sion frequencies, is sequence dependent and enzyme related,FEBS Lett., 319, 201–205.

w x Ž .7 Steen, A., H. Luthman, D. Hellgren and B. Lambert 1990Levels of hypoxanthine phosphoribosyltransferase RNA inhuman cells, Exp. Cell Res., 186, 236–244.

w x Ž .8 Winship, P.R. 1989 An improved method for directly se-quencing PCR amplified material using dimethyl sulphoxide,NAR, 17, 1266–1266.

w x9 Curry J., A. Skandalis, J. Holcroft, J. de Boer and B.W.Ž .Glickman 1993 Coamplification of hprt cDNA and gamma

T-cell receptor sequences from 6-thioguanine resistant hu-man T-lymphocytes, Mutation Res., 288, 269–275.

w x10 Curry J., G.T. Rowley, V. Saddi, D. Beare, J. Cole and B.W.Ž .Glickman 1995 Determination of hprt mutant and mutation

frequencies and the molecular characterization of humanderived in vivo T-lymphocyte mutants, Environ. Molec. Mu-tagen., 25, 169–179.

w x Ž .11 Pathak, V.K. and H.M. Temin 1990 Broad spectrum of invivo forward mutations, hypermutations, and mutationalhotspots in a retroviral shuttle vector after a single replicationcycle: Substitutions, frameshifts, and hypermutations, Proc.Natl. Acad. Sci. U.S.A., 87, 6019–6023.

w x Ž .12 Ji, J. and L.A. Loeb 1994 Fidelity of HIV-1 reverse tran-scriptase copying a hypervariable region of the HIV-1 enÕgene, Virology, 199, 323–330.

w x Ž .13 Keohavong, P. and W.G. Thilly 1989 Fidelity of DNApolymerases in DNA amplification, Proc. Natl. Acad. Sci.USA, 86, 9253–9257.

w x Ž .14 Lindahl, T. 1979 DNA glycosylases, endonucleases forapurinicrapyrimidinic sites, and base-excision repair, Prog.Nucleic Acids Res. Mol. Biol., 22. 135–189.

w x Ž .15 Eckert, K.A. and T.A. Kunkel 1991 DNA polymerasefidelity and the polymerase chain reaction, PCR MethodsAppl., 1, 17–24.

w x Ž .16 Eckert, K.A. and T.A. Kunkel 1991 The fidelity of DNApolymerases used in the PCR, in: M.J. McPhearson, P.

Ž .Quirke, G.R. Taylor Eds. , Polymerase Chain Reaction I: APractical Approach, IRL Press at Oxford University Press,Oxford, pp. 227–246.