Journal of Virological Methods 75 (1998) 93–104

Interlaboratory concordance of DNA sequence analysis todetect reverse transcriptase mutations in HIV-1 proviral DNA

Lisa M. Demeter a,*, Richard D’Aquila b, Owen Weislow c, Eric Lorenzo d,Alejo Erice e, Joseph Fitzgibbon f, Robert Shafer g, Douglas Richman h,

Thomas M. Howard i, Yuqi Zhao j, Eva Fisher d, Diana Huang k, Douglas Mayers l,Shelly Sylvester b, Max Arens m, Kim Sannerud e, Suraiya Rasheed i,

Victoria Johnson n, Daniel Kuritzkes o, Patricia Reichelderfer p, Anthony Japour q, forthe ACTG Sequencing Working Group

a Uni6ersity of Rochester School of Medicine and Dentistry, Rochester, NY, USAb Massachusetts General Hospital, Har6ard Medical School, Boston, MA, USA

c SRA Technologies, Rock6ille, MD, USAd Uni6ersity of Miami, Miami, FL, USA

e Uni6ersity of Minnesota, Minneapolis, MN, USAf Uni6ersity of Medicine and Dentistry of New Jersey, Robert Wood Johnson Medical School, New Brunswick, NJ, USA

g Stanford Uni6ersity, Palo Alto, CA, USAh Uni6ersity of California, San Diego, CA, USA

i Uni6ersity of Southern California, Los Angeles, CA, USAj Northwestern Uni6ersity, Chicago, IL, USA

k Rush-Presbyterian–St. Lukes Hospital, Chicago, IL, USAl Walter Reed Army Institute, Rock6ille, MD, USA

m Washington Uni6ersity Medical School, St. Louis, MO, USAn Uni6ersity of Alabama, Birmingham, AL, USA

o Uni6ersity of Colorado, Den6er, CO, USAp Di6ision of AIDS, NIAID, NIH, Bethesda, MD, USA

q Beth Israel Hospital, Har6ard Medical School, Boston, MA, USA

Accepted 25 July 1998

Abstract

Thirteen laboratories evaluated the reproducibility of sequencing methods to detect drug resistance mutations inHIV-1 reverse transcriptase (RT). Blinded, cultured peripheral blood mononuclear cell pellets were distributed to eachlaboratory. Each laboratory used its preferred method for sequencing proviral DNA. Differences in protocolsincluded: DNA purification; number of PCR amplifications; PCR product purification; sequence/location of PCR/se-

* Corresponding author. University of Rochester Infectious Diseases Unit, 601 Elmwood Avenue, Box 689, Rochester, NY 14642,USA. Tel.: +1-716-2754764; Fax: +1-716-4429328.; E-mail: lisa–demeter@urmc.rochester.edu

0166-0934/98/$ - see front matter © 1998 Elsevier Science B.V.

PII: S 0166 -0934 (98 )00100 -1

L. M. Demeter et al. / Journal of Virological Methods 75 (1998) 93–10494

quencing primers; sequencing template; sequencing reaction label; sequencing polymerase; and use of manual versusautomated methods to resolve sequencing reaction products. Five unknowns were evaluated. Thirteen laboratoriessubmitted 39 043 nucleotide assignments spanning codons 10–256 of HIV-1 RT. A consensus nucleotide assignment(defined as agreement among ]75% of laboratories) could be made in over 99% of nucleotide positions, and wasmore frequent in the three laboratory isolates. The overall rate of discrepant nucleotide assignments was 0.29%. Aconsensus nucleotide assignment could not be made at RT codon 41 in the clinical isolate tested. Clonal analysisrevealed that this was due to the presence of a mixture of wild-type and mutant genotypes. These observations suggestthat sequencing methodologies currently in use in ACTG laboratories to sequence HIV-1 RT yield highly concordantresults for laboratory strains; however, more discrepancies among laboratories may occur when clinical isolates aretested. © 1998 Elsevier Science B.V.

Keywords: HIV-1; Sequence analysis; Drug resistance

1. Introduction

Resistance of HIV-1 to reverse transcriptaseinhibitors occurs frequently in patients receivingthese drugs, and it has been associated with a lossof viral suppression (Saag et al., 1993; Richman etal., 1994; D’Aquila et al., 1995; Japour et al., 1995;Welles et al., 1996). In addition, phenotypic resis-tance of HIV-1 to zidovudine has been associatedwith a poor clinical outcome (D’Aquila et al.,1995). The use of phenotypic assays to detect HIV-1drug resistance in clinical trials is limited by expenseand the need for a viral isolate. Selective PCRhybridization has been used to detect the threonineto tyrosine mutation at reverse transcriptase codon215 (T215Y) (Shafer et al., 1996), which is the mostcommon mutation associated with AZT resistance.The development of the T215Y mutation is alsoassociated with a poor clinical outcome (Japour etal., 1995). The use of selective PCR hybridizationto detect HIV-1 drug resistance requires knowledgeof the predominant mutation that develops duringtherapy, and may not be useful in detecting multi-ple or unexpected resistance mutations (Richmanet al., 1994; Shafer et al., 1994; Shirasaka et al.,1995; Staszewski et al., 1995).

Sequence analysis may be advantageous for thedetection of novel resistance mutations and thedetermination of genotype at multiple codons.Several different methods in DNA sequence analy-sis are available. The accuracy and reproducibilityof these different methods in detecting antiretrovi-ral resistance mutations in clinical isolates of HIV-1is not known. We report the results of a comparison

of the interlaboratory reproducibility of differentsequencing methods used by laboratories partici-pating in the AIDS Clinical Trials Group (ACTG).

2. Materials and methods

2.1. HIV-1 isolates and cell lines

The following virus isolates and cell lines wereobtained through the AIDS Reference and ReagentProgram, AIDS Program, NIAID NIH: AO18H112-2 pre-AZT and AO18 G910-6 post-AZTfrom Dr Douglas Richman, and 8E5/LAV from DrThomas Folks.

2.2. HIV-1 culture

AO18 pre-AZT and AO18 post-AZT were prop-agated in phytohemagglutinin (PHA)- and inter-leukin-2 (IL2)-stimulated HIV-negative PBMCs.HIV-1 isolation from patient PBMCs was carriedout using co-cultivation with PHA and IL-2 stim-ulated HIV-negative donor PBMCs according topreviously published procedures (Hollinger et al.,1992).

2.3. Sequencing methods

The preferred method of each laboratory forsequencing the region of the HIV-1 genome encod-ing reverse transcriptase is summarized in Tables 1and 2. All laboratories used the polymerase chainreaction (PCR) to amplify a region of the pol gene

L. M. Demeter et al. / Journal of Virological Methods 75 (1998) 93–104 95

Table 1Nucleic acid and PCR product purification procedures

Gel purification of PCR product Other purification of PCR productLab code Unknowns DNA purification

Yes GeneCleanA SEQ 1–5 Crude cell lysateCrude cell lysate YesB SEQ 1–2a GeneClean

NoCrude cell lysate NoSEQ 1–5CMillipore 30 000 NMWL filterNoD SEQ 1–5 Purified

Crude cell lysate NoE SEQ 1–5 NoYesPurified GeneCleanSEQ 1–5FNo PEG precipitationG SEQ 1–5 Crude cell lysate

Wizard PCR prep (Promega)YesCrude cell lysateSEQ 1–5HPurified NoI SEQ 1–5 QiaQuickPurified YesJ SEQ 1–5 Magic PCR prep (Promega)

GeneCleanYesCrude cell lysateSEQ 1–5KNoNoL SEQ 1–5 Crude cell lysate

Crude cell lysate NoL SEQ 3–5 QiagenYesCrude cell lysate DEAE membraneSEQ 1–2aMNo QiaQuickM SEQ 3–5 Crude cell lysate

a Laboratories B and M sequenced a single clone of amplified PCR product.

from proviral DNA. Almost all laboratories se-quenced directly the resultant PCR products. Twolaboratories (B and M) cloned PCR products intobacterial plasmids and reported sequence dataobtained from a single clone. Purification meth-ods, type and number of primers, DNA sequenc-ing polymerase, labelling method, and sequencingmethod used varied among the laboratories (Ta-bles 1 and 2). The majority of laboratories am-plified proviral DNA from a crude cell lysate.Approximately half of the laboratories gel-purified the resulting PCR product (Table 1).Eight laboratories used manual sequencing. Twoof the laboratories utilized solid phase separationto prepare single-stranded DNA template, threeused alkaline denaturation, and the remaindercarried out cycle sequencing on double strandedPCR product (Table 2). Seven laboratories usedautomated sequencing (two of these laboratoriesalso undertook manual sequencing); one of theselaboratories used Pharmacia and the remainderused Applied Biosystems technology (Table 2).Most automated sequencing laboratories useddye-labelled terminators, with the exception oflabs C and H, which used dye-labelled primers(Table 2).

Sequences and locations of PCR and sequenc-ing primers used by the different laboratories aresummarized in Figs. 1 and 2, and Appendix 1.

2.4. Sequence comparisons

The Virology Quality Assurance Laboratory(VQA) at Rush Presbyterian/St. Luke’s dis-tributed five blinded non-viable PBMC pellets(1–2×106 cells each) on dry ice to the participat-ing laboratories (SEQ 1 and 2 were distributed inJanuary 1995, and SEQ 3, 4, and 5 were dis-tributed in May 1995). The identities of the un-knowns, which are listed below, were not revealedto the participants until the analysis of sequenceswas completed. With the exception of SEQ 4, allPBMC pellets were harvested at termination ofculture.

AO18 G910-6 (post-AZT)SEQ 1SEQ 2 AO18 H112-2 (pre-AZT)SEQ 3 low passage clinical isolate

8E5/LAV cell lineSEQ 4SEQ 5 duplicate of SEQ 3

2.5. Sequence alignment and analysis

Sequence data were transmitted by each labora-tory as a text file either on diskette or via theInternet. Text files were imported into MacVectorDNA analysis program. Sequence alignments

L. M. Demeter et al. / Journal of Virological Methods 75 (1998) 93–10496

Table 2Sequencing methodologies used by participating laboratoriesa

Sequencing label Sequencing methodSequenceLab Unknowns Sequence Template preparationtemplate polymerase

Heat denat. Fluorescent ddNTPs TaqA SEQ 1–5 ABIdsmanualSequenase[35S]dATPB Alkaline denat.SEQ 1–2 ds

Alf (T7) PharmaciaC SEQ 1–5 ss Solid-phase separ. Fluorescent primer32P-primer TaqD SEQ 1–5 ds Heat denat. manual

Sequenase[35S]dATP manualE Alkaline denat.SEQ 1–5 ds32P-primer TaqF SEQ 1–5 ds Heat denat. manual

manualTaq32P-primerG Heat denat.SEQ 1–5 dsHeat denat. Fluorescent primer Taq ABIH SEQ 1–5 ds

Sequenase manual[35S]dATPI Solid-phase separ.SEQ 1–5 ssHeat denat. Fluorescent ddNTPs TaqJ SEQ 1–5 ABIds

Fluorescent ddNTPs TaqK SEQ 1–5 ds Heat denat. ABISequenase[33P]dATP manualL Solid-phase separ.SEQ 1–5 ss

Heat denat. Fluorescent ddNTPs TaqL ABISEQ 3–5 dsmanualSequenaseM SEQ 1–2 ds Alkaline denat. [35S]dATP

Heat denat. Fluorescent ddNTPsM TaqSEQ 3–5 ABIds

a Denat., denaturation; separ., separation; ds, double stranded DNA; ss, single stranded DNA.

were generated using AssemblyLign software.Gaps were inserted manually to maximize align-ment of sequences.

A consensus nucleotide assignment was made ata given position if ]75% of the laboratoriesprovided the same nucleotide assignment at thatposition. A discrepant nucleotide assignment wasdefined as a nucleotide assignment by an individ-ual laboratory that differed from the consensusnucleotide assignment and which occurred at aconserved nucleotide position (Meyers et al.,1995). If a laboratory reported a nucleotide mix-ture at a conserved position and the group didnot, this was classified as a discrepant nucleotideassignment by that laboratory.

2.6. Sequence analysis of cloned PCR products

In an attempt to determine whether geneticheterogeneity was present in the clinical isolatestested, two laboratories (laboratories E and G)carried out sequence analysis of individual clonedPCR products amplified from SEQ 3 and 5. Lab-oratory E amplified reverse transcriptase se-quences using a nested PCR reaction with primersZ2660/Z2661 and P1/P2 (inner primers P1 and P2contained EcoRI sites to facilitate subsequent

cloning). PCR product from the second PCRreaction was digested with EcoRI, ligated withEcoRI-digested pUC-19, and transformed intoDH5-a. Ten consecutive clones with DNA insertswere sequenced using Sequenase version 2.0(USB). Sequencing primers used were M13 for-ward and reverse primers (USB). Nucleotide se-quence spanning reverse transcriptase codons19–82 were reported with this method. Labora-tory G amplified reverse transcriptase sequencesusing primers A and NE1 modified to containuracil-containing tails at each 5% end. PCR prod-ucts were cloned into the pAMP vector (GIBCO-BRL), according to the manufacturer’sinstructions. Twenty-six consecutive clones weresequenced, using primers 74-2B, 74-7, 215-A2,and 215-D. Nucleotide sequences spanning re-verse transcriptase codons 31–130 were reportedby Laboratory G.

3. Results

3.1. Amino acid assignments at codons associatedwith AZT resistance

Table 3 summarizes the amino acid codon as-

L. M. Demeter et al. / Journal of Virological Methods 75 (1998) 93–104 97

Fig. 1. PCR primers used by participating laboratories. Position and name of primers used by each individual laboratory arediagrammed (see Appendix 1 for sequence and exact location of each primer). Sense primers are represented by arrows pointing tothe right, and antisense primers by arrows pointing to the left. If nested PCR was carried out, both inner and outer primer sets arediagrammed. Positions of primers outside protease and reverse transcriptase coding regions are not drawn to scale.

signments for the five reverse transcriptase codons(41, 67, 70, 215, 219) most commonly associatedwith AZT resistance. Ten laboratories providedsome sequence data for SEQ 1; 13 provided datafor SEQ 2; and 12 provided data for SEQ 3, 4,and 5 (Laboratory L provided sequence data us-ing both manual and automated sequencing meth-ods, leading to a total of 13 sequences from 12laboratories). Almost all of the laboratories wereable to provide sequence data for all five AZTresistance codons (Table 3). There was completeagreement in codon assignment among those lab-oratories reporting data for codons 67, 70, and215. Three laboratories reported discrepant codonassignments at position 219 for SEQ 2, 3, 4, or 5.Each of these laboratories used automated se-quencing methods.

There was complete agreement within the group

on amino acid assignment for codon 41 in thethree laboratory isolates tested (SEQ 1, 2, and 4).However, there was significant disagreementamong laboratories with regard to the amino acidassignment at codon 41 for the one clinical isolatetested in duplicate (SEQ 3 and 5). Eight of 11laboratories reported the presence of the AZTresistance mutation leucine at codon 41 (M41L)for SEQ 3, and three laboratories reported amixture of wild type and mutant sequences. ForSEQ 5, seven of 11 laboratories reported the AZTresistant mutant sequence at codon 41, two labo-ratories reported a mixture of wild type and mu-tant sequences, and two laboratories reportedwild type sequence only. In the case of SEQ 5, thisdisagreement among laboratories lead to a lack ofconsensus as to whether the unknown was wildtype or mutant at codon 41.

L. M. Demeter et al. / Journal of Virological Methods 75 (1998) 93–10498

Fig. 2. Sequencing primers used by participating laboratories. Position and name of primers used by each individual laboratory arediagrammed as in Fig. 1 (see Appendix 1 for sequence and exact location of each primer).

3.2. Nucleotide sequence comparisons

Table 4 summarizes the nucleotide sequencecomparisons among laboratories for each of thefive unknowns. The number of different nucleotidepositions analyzed for each unknown (nucleotidepositions at which four or more laboratories madea nucleotide assignment) ranged from 713 to 737nucleotides. These nucleotide positions spanned

codons 10–256 of the HIV-1 reverse transcriptase.The percent of nonconsensus nucleotide positionsfor the three laboratory isolates (SEQ 1, 2, and 4)ranged from 0 to 0.22% (Table 4). In contrast, thenonconsensus nucleotide positions in the two un-knowns from the clinical isolate tested were 0.95and 1.23% of analyzed nucleotide positions. Asnoted above, one of these nonconsensus positionswas at codon 41, which confers AZT resistance.

L. M. Demeter et al. / Journal of Virological Methods 75 (1998) 93–104 99

Table 3Amino acid assignments at AZT resistance codons in the HIV-1 reverse transcriptase

WildtypeCodon

SEQ 5b,cSEQ 4cSEQ 2 SEQ 3b,cAmino acid SEQ 1a

Met (10/10) Leu (8/11)d Met (11/11) Leu (7/11)e41 Met Leu (9/9)Asp (13/13)Asp (13/13)Asp (13/13)67 Asp (11/11)Asp Asn (10/10)

Lys (11/11) Lys (13/13) Lys (13/13) Lys (13/13)70 Lys Arg (10/10)Thr (13/13)Thr (13/13)Thr (13/13)215 Thr (13/13)Thr Tyr (10/10)Lys (11/13)gLys (11/12)h219 Lys Gln (10/10) Lys (12/13)f Lys (11/13)g

a Amino acid changes associated with AZT resistance are bolded. Amino acids are identified by their three-letter code.b All laboratories also identified a valine to isoleucine mutation at codon 75 that has been reported to occur during nucleoside

therapy.c Twelve laboratories provided data for SEQ 3–5. One laboratory reported two sequences for each unknown, using automated

and manual methods, therefore the total number of reported sequences is 13.d Three laboratories reported a mixture of leucine and methionine.e Two laboratories reported a mixture of leucine and methionine; two laboratories reported methionine only.f One laboratory reported histidine.g One laboratory reported asparagine, one reported glutamine.h One laboratory reported glutamine.

All of the non-consensus nucleotide positions inthe five unknowns were base substitutions, withthe exception of the non-consensus nucleotide inSEQ 4 (8E5/LAV). Seven of 13 reported se-quences (54%) contained an insertion of oneadenine in a stretch of six adenines that spanscodons 219 and 220. This frameshift mutationlead to premature termination of the readingframe after codon 232, and is compatible withHIV RT sequence from this cell line reportedpreviously (Gendelman et al., 1987). Five of theseven sequences reporting the frameshift mutationwere generated using manual sequencing methods,whereas all of the five labs that did not detect theframeshift mutation used automated sequencingmethods (data not shown). The presence of theframeshift mutation was verified by sequencingboth strands of the PCR product in laboratoriesG and L.

A discrepant nucleotide assignment was definedas a nucleotide assignment reported by an individ-ual laboratory that disagreed with the consensusassignment made by the group as a whole. Thepercent of discrepant nucleotide assignments weresimilar among both laboratory and clinical iso-lates, and ranged from 0.25 to 0.32% (Table 4). Ofthe 112 discrepant nucleotide assignments, all but

six were base substitutions. The six frame shiftmutations that were discrepant nucleotide assign-ments all occurred in sequences reported by labo-ratories F and M.

3.3. Intralaboratory 6ariability

Because the one clinical isolate tested was sentin duplicate, a preliminary estimate of intralabo-ratory variability could be made. The percentnucleotide divergence in SEQ 3 versus SEQ 5sequences reported by an individual laboratoryranged from 0 to 1.48% with a median of 0.56%(data not shown).

3.4. Clonal analysis of PCR products amplifiedfrom SEQ 3 and 5

In order to assess whether the lack of concor-dance at codon 41 in SEQ 5 was due to hetero-geneity in the original starting material, PCRproducts were amplified from SEQ 3 and 5,cloned into bacterial plasmids, and individualclones were sequenced. This analysis was carriedout by two separate laboratories (laboratories Eand G). Of 36 clones analyzed, 15 had the wild-type sequence ATG (methionine), 17 had the mu-

L. M. Demeter et al. / Journal of Virological Methods 75 (1998) 93–104100

Table 4Summary of nucleotide sequence comparisons

SEQ 3 SEQ 4SEQ 1 SEQ 2 SEQ 5

12 12No. labs providing data 11 1213724 729737732No. nucleotide (NT) positions analyzed 713

8363 8439Total number of NTs (no. NT per lab×no. labs) 6573 7221 84507/737 1/724Non-consensus NT/NT positions analyzed (%) 0/713 2/732 9/729

(0.14%)(0.95%) (1.23%)(0.22%)(0%)

23/845021/843927/836323/7221Discrepant NT assignments/total NT 18/6573(0.32%) (0.32%) (0.25%) (0.27%)(0.27%)

Types of discrepancies9 10Transitions 7 7 11

15 10Transversions 11 16 1013 20Insertions/deletions 0

tant codon TTG (leucine), two had the mutantcodon CTG (leucine), one had the mutant codonTTA (leucine), and one had the sequence ACG(threonine). This degree of heterogeneity was notseen at any of the other nucleotide positionsanalyzed (data not shown).

4. Discussion

We compared the reproducibility of differentsequencing methods currently used by ACTG lab-oratories to determine the nucleotide sequence ofHIV-1 reverse transcriptase in cultured proviralDNA. It was found that, for the majority ofpositions tested, concordant assignments could bemade at the five AZT resistance codons, usingsequencing techniques that differed significantly inDNA purification steps, primer choice, and se-quencing method. The two exceptions to the con-cordance were disagreement among laboratoriesas to whether a frameshift mutation existed inSEQ 4, and whether SEQ 5, the clinical isolate,was wild-type or mutant at codon 41.

The reported frameshift in SEQ 4 occurred in astretch of six adenines that comprises codons219–220. It is not surprising that a technicalartifact could occur in such a region, either due toshifting by the sequencing polymerase duringprimer extension or to base calling algorithm am-biguities in automated sequencing. SEQ 4 wasderived from the cell line 8E5/LAV, which con-

tains an integrated copy of HIV-1 and which isdefective for viral production and reverse tran-scriptase (Folks et al., 1986; Gendelman et al.,1987). A reported sequence of 8E5/LAV describedpreviously contained the identical frameshift mu-tation reported here (Gendelman et al., 1987). Thelack of consensus in identifying this frameshiftmay reflect a bias against reporting the presenceof frameshift mutations in sequences that arepresumed to represent viable viral isolates. Alter-natively, automated sequencing methods may bemore prone to artifact in long homopolymericstretches of nucleotides, given the observationthat laboratories using automated sequencing ac-counted for all of those reporting discrepantcodon assignments for codon 219 in SEQ 3 and 5(Table 3).

It was demonstrated that genetic heterogeneityin the clinical isolate led to a lack of consensus atcodon 41. This lack of consensus was probablydue to differing sensitivities of the methods usedto detect genotypic mixtures, although this hy-pothesis was not tested formally with known mix-tures of mutant and wildtype sequences. Inaddition, because of the fact that only a singleclinical isolate was tested, we were unable toascertain how frequently this lack of consensuswould occur, and whether any of the methodstested would be more sensitive to the presence ofmixtures. Most methods used by our group wererelatively insensitive to the presence of a 50%mixture of genotypes, with a bias toward report-

L. M. Demeter et al. / Journal of Virological Methods 75 (1998) 93–104 101

ing the mutant sequence at this codon. Because ofthe small number of samples tested in this com-parison, we do not have sufficient data to estab-lish whether this bias reflects the influence oftechnical factors on selectivity of nucleotide incor-poration or preferential reporting of drug resis-tance mutations. It should be noted that all buttwo of the automated sequencing laboratoriesused Taq polymerase and dye-labeled terminatorsin the sequencing reactions. It has been demon-strated that the net rate of dideoxynucleotideincorporation by Taq DNA polymerase is depen-dent on the local sequence environment, particu-larly when dye-labeled terminators are used (Leeet al., 1992; Parker et al., 1995), resulting ininconsistent identification of genotypic mixtureswhen dye-labeled terminators are used with Taqpolymerase.

When reviewing nucleotide assignments, inde-pendent of whether these were at codons associ-ated with drug resistance, it was found that aconsensus nucleotide assignment (defined asagreement between 75% or more laboratories)could be made in over 99% of nucleotide posi-tions. There did appear to be an increase in thepercent of nonconsensus nucleotide assignmentsin the clinical isolate compared to the laboratorystrains. A determination of how frequently geneticmixtures could pose a problem in reaching aninterlaboratory consensus on reverse transcriptasegenotype would require analysis of greater num-bers of clinical isolates. In addition, our study waslimited to the analysis of proviral DNA fromcultured isolates, and did not address whether thisproblem would occur more or less frequently ifsequencing was carried out on PCR productsamplified from viral RNA.

Discrepant nucleotide assignments occurred anaverage of 0.29% of reported nucleotides, butvaried widely among the different laboratories.These discrepant assignments did not appear toresult from errors in transcribing sequence data(data not shown). There also appeared to be aclustering of reported frame shift mutations thatwere discrepant assignments, with all of theseframeshifts being reported by two laboratories.This may relate to a lack of experience with thespecific sequencing technique or submission ofsequence data without manual editing. Intralabo-ratory variation in sequencing the clinical isolatewas relatively low and ranged from 0 to 1.48%.

In conclusion, the observations made in thisinterlaboratory comparison suggest that sequenc-ing methods used by ACTG laboratories at thetime of this study yield highly concordant resultsfor sequencing reverse transcriptase from labora-tory strains of HIV-1. However, these methodsmay give discordant results when clinical isolatescontaining mixtures of genotypes are analyzed.Studies are underway to develop a consensus se-quencing assay to detect drug resistance muta-tions in HIV protease and reverse transcriptase,using automated sequencing methods that canreliably identify heterozygous positions in humangenes (Chadwick et al., 1996).

Acknowledgements

This work was supported in part by grantsfrom the National Institutes of Health (5UO1-AI-27658, AI-29193, AI-27659, UO1-AI-27675,2UO1-AI-27559, RR-00044-34S2) and theChicago Pediatric Faculty Foundation.

Appendix A. Primer sequences and locations

Sequence (5%�3%)a,b,c SensedPrimer Locatione

GTTATCTATCAATACATGG + RT1861735ATTAGATATCAGTACAATGTGC + RT150215-ACCACAGGGATGGAAAGGATCA + RT157215-A1

+ RT162215-A2 TGGAAAGGATCACCAGCAATATTCCATCTGTATGTCATTGACAGTCCAGC − RT250215-D

L. M. Demeter et al. / Journal of Virological Methods 75 (1998) 93–104102

Appendix A. (Continued)

Sequence (5%�3%)a,b,c SensedPrimer Locatione

GCTATTAAGTCTTTTGATGGGTCA RT318−215-ERT200+215p TTAGAAATAGGGCAGCATAGRT270−3CA UNI2 ATAATTGCYTTACTTTAATCCCTG

− RT1413F150-144 TGGAAGCACATTGTACTGATA− RT2553F262-256 TCCCACTAACTTCTGTATGTC

ACCCATCCAAAGGAATGGAGGTTCTTTC −3RT RT221CTGCCAGTTCTAGCTCTGCTTCTT −3RT 3461 RT296

RT163−3RT3057 GGCTCTAAGATTTTTGTCATRT272−3% 3388 ACATAATTGCCTTACTTTAATC

ACTTTAAATTTTCCMATTAGTCCTATTGA +5CA NL2 RT6TACACCTGTCAACATAATTGGAAG +5CA 2488 PRO88

RT135+5F127-134 ATACTGCATTTACCATACCTAG+ RT265F19-26 CAAAAGTTAAACAATGGCCAT

GTTAAACAATGGCCATTGAC +5RT2609 RT28+ RT4TTGCACTTTAAATTTTCCATTAG5% Larder

CACTTTAAATTTTCCCAT +5% 2536 RT3+ RT205%RT GCCAGGAATGGATGGCCC

CAGTAAAATTAAAGCCAGG +74-1A RT16RT30GTTAAACAATGGCCATTGACAGAAGA74-2B +

− RT13274-4C CTGGTGTCTCATTGTTTATACTAGRT183−74-5 CCTACATACAAATCATCCAT

CCCTATTTCTAAGTCAGATCCTACATACAAATCATC −74-7 RT184TTGGTTGCACTTTAAATTTTCCCATTAGTCCTATT +A% RT6

RT158+B GGATGGAAAGGATCACC− RT152BR GGTGATCCTTTCCATCC

GGAAGAAATCTGTTGACTCAGATTGGT +CI-Pol1 PRO95+ RT166CCAGCAATATTCCAAAGTAGCATGACACI5-156RT

TGTACAGAAATGGAAAAGGAA +MART2 RT45− RT231MART3 GTACTGTCCATTTATCAGGAT

CCTACTAACTTCTGTATGTCATTGACAGTCCAGCT −NE1% RT250AATTAAAGCCAGGAATGGATG +P1 RT19GGGATGGAAAGGATCACCAGC + RT159P11b

RT259−P2 GCCGAATTCAATTTCCCCACTAA− RT251P245 TGTATGTCATGGACAGTCCA− RT225P3 CCCATCCAAAGGAATGG

TTTCTGTATGTCATTGACAG −P34 RT252RT271TTACTTTAATCCCTGP35 −

CCTACATACAAATCATCCATGTATTG − RT181P4AAATTAAAGCCAGGAATGGATGGCCCAAAAG +P5B RT22

+ PR91ACCTACACCTGTCAACATAATTGGAAGAAATCTGTTP6upTTCTGTTAGTGCTTTGGTTCCTCTAAGGAGTTTAC − RT279P72

− RT128P8 ATACTAGGTATGGTAAATGCTGTAAAACGACGGCCAGTTAAACAATGGCCATTGACAG + RT29POL1

L. M. Demeter et al. / Journal of Virological Methods 75 (1998) 93–104 103

Appendix A. (Continued)

SensedSequence (5%�3%)a,b,cPrimer Locatione

CAGGAAACAGCTATGACCACAGTCCAGCTGTCTTTTTCTG − RT246POL2PRO87Pol2483 GCTGCAGCAGGACCTACACCTGTCAACATAATTGG +

+Pol2703 CTGAAAATCCATACAATACTC RT60Pol3109 RT179CAAATCATCCATGTATTGATAG −Pol3321 RT249−CCAAGCTTTGTATGTCATTGACAGTCCAGCTGTC

− INTCCTTGACTTTGGGGATTGTAGGGPol4674PR3ACAGGAGCAGATGATACAGT +RT1

− RT318RT2 TTAAGTCTTTTGATGGGTCAATAATCCACCTATCCCAGTAGGAGAAAT GAG+SK38AGACTTCAGGAAGTATACTGCSmp136 + RT130

−Smp141 GTCATTGACAGTCCAGCTGTC RT249TATCTAAGGGAACTGAAAAATSYB121 − RT114CCCTGGTGTTTCATTGTTTAT RT134SYB141 −GTCTCAGTTCCTCTATTTTTG RT199−SYB200

+ RT48AAATGGAAAAGGAAGGGAAAASYB40GAAAATCCATACAATACTCCASYB53 + RT60ACTGGAGTATTGTATGGATTTSYB60 − RT52

+TJ-1 RT20GCCAGGAATGGATGGCCTJ-2 RT115+TACTGGATGTGGGCGATG

− RT91CAGGATGTGGTATTCCTAATUAB 16RT249WEIR2 TCATTGACAGTCCAGCTGTC −RT54−Z2446 GGCAAATACTGGAGTATTGTATGG

+Z2447 RT83GTAGATTTCAGAGAACTTAATA+Z2660 PR79GCTATAGGTACAGTATTAGTAGG− RT263GTAAATCTGACTTGCCCAAATTCZ2661

a Sequence is actual sequence of primer, written in a 5%�3% direction.b Underlined sequences are not present in the HIV-1 genome. Pol 1 and 2 contain the M13 forward and

reverse sequences, respectively. Pol 2483 and Pol 3321 contain a PstI and HindIII site, respectively.c IUB codes refer to 50–50 mixtures, e.g. M is a 50–50 mixture of A and C, Y is a 50–50 mixture of

T and C.d + Sense primer; − antisense primer.e RT, reverse transcriptase codon number; PR, protease codon number; GAG, present in gag region

(upstream of protease); INT, integrase (downstream of RT). Number refers to the first full codon adjacentto the 3% end of the primer.

References

Chadwick, R.B., Conrad, M.P., McGinnis, M.D., Johnston-Dow, L., Spurgeon, S.L., Kronick, M.N., 1996. Het-erozygote and mutation detection by direct automatedfluorescent DNA sequencing using a mutant Taq DNApolymerase. BioTechniques 20, 676–683.

D’Aquila, R.T., Johnson, V.A., Welles, S.L., Japour,A.J., Kuri-tzkes, D.R., DeGruttola, V., Reichelderfer, P.S., Coombs,R.W., Crumpacker, C.S., Kahn, J.O., Richman, D.D., forthe AIDS Clinical Trials Group Protocol 116B/117 Team,Virology Committee Resistance Working Group, 1995.Zidovudine resistance and HIV-1 disease progression duringantiretroviral therapy. Ann. Intern. Med. 122, 401–408.

L. M. Demeter et al. / Journal of Virological Methods 75 (1998) 93–104104

Folks, T.M., Powell, D., Lightfoote, M., Koenig, S., Fauci,A.S., Benn, S., Rabson, A., Daugherty, D., Gendelman,H.E., Hoggan, M.D., Venkatesan, S., Martin, M., 1986.Biological and biochemical characterization of a clonedLeu3-cell surviving infection with the acquired im-munodeficiency syndrome retrovirus. J. Exp. Med. 164,280–290.

Gendelman, H.E., Theodore, T.S., Willey, R., McCoy, J.,Adachi, A., Mervis, R.J., Venkatesan, S., Martin, M.,1987. Molecular characterization of a polymerase mutanthuman immunodeficiency virus. Virology (160) 323–9 (Er-ratum with corrected sequence published in Virology 191,(1992), 529).

Hollinger, F.B., Bremer, J.W., Myers, L.E., Gold, J.W.M.,McQuay, L., NIH/NIAID/DAIDS/ACTG Virology Labo-ratories, 1992. Standardization of sensitive human im-munodeficiency virus coculture procedures andestablishment of a multicenter quality assurance programfor the AIDS Clinical Trials Group. J. Clin. Microbiol. 30,1787–1794.

Japour, A.J., Welles, S., D’Aquila, R.T., Johnson, V.A., Rich-man, D.D., Coombs, R.W., Reichelderfer, P.S., Kahn,J.O., Crumpacker, C.S., Kuritzkes, D.R., for the AIDSClinical Trials Group 116B/117 Study Team and the Virol-ogy Committee Resistance Working Group, 1995. Preva-lence and clinical significance of zidovudine resistancemutations in human immunodeficiency virus isolated frompatients after long-term zidovudine treatment. J. Infect.Dis. 171, 11729.

Lee, L.G., Connell, C.R., Woo, S.L., Cheng, R.D., McArdle,B.F., Fuller, C.W., Halloran, N.D., Wilson, R.K., 1992.DNA sequencing with dye-labeled terminators and T7DNA polymerase: effect of dyes and dNTPs on incorpora-tion of dye terminators and probability analysis of termi-nation fragments. Nucleic Acids Res. 20, 2471–2483.

Meyers, G., Korber, B., Hahn, B.H., 1995. Human Retro-viruses and AIDS. A compilation and analysis of nucleicacid and amino acid sequences. Theoretical Biology andBiophysics, Los Alamos, NM.

Parker, L.T., Deng, Q., Zakeri, H., Carlson, C., Nickerson,D.A., Kwok, P.Y., 1995. Peak height variations in auto-mated sequencing of PCR products using Taq dye-termina-tor chemistry. BioTechniques 19, 116–121.

Richman, D., Havlir, D., Corbeil, J., Looney, D., Ignacio, C.,Spector, S.A., Sullivan, J., Cheeseman, S., Barringer, K.,Pauletti, D., Shih, C-K., Myers, M., Griffin, J., 1994.Nevirapine resistance mutations of human immunodefi-ciency virus type 1 selected during therapy. J. Virol. 68,1660–1666.

Saag, M.S., Emini, E.A., Laskin, O.L., Douglas,J. Lapidus,W.I. Shlief, W.A. Whitley, R.J. Hildebrand, C. Byrnes,V.W. Kappes, J.C. Anderson, K.W. Massari, F.E. Shaw,G.M., L697,661 Working Group, 1993. A short-term clini-cal evaluation of L-697,661, a non-nucleoside inhibitor ofHIV-1 reverse transcriptase. New Engl. J. Med. 329, 1065–1072.

Shafer, R.W., Kozal, M.J., Winters, M.A., Iversen, A.K.N.,Katzenstein, D.A., Ragni, M.V., Meyer, W.A., Gupta, P.,Rasheed, S., Coombs, R., Katzman, M., Fiscus, S., Meri-gan, T.C., 1994. Combination therapy with zidovudine anddidanosine selects for drug-resistant human immunodefi-ciency virus type 1 strains with unique patterns of pol genemutations. J. Infect. Dis. 169, 722–729.

Shafer, R.W., Winters, M.A., Mayers, D.L., Japour, A.J.,Kuritzkes, D.R., Weislow, O.S., White, F., Erice, A., San-nerud, K.J., Iversen, A., Pena, F., Dimitrov, D., Frenkel,L.M., Reichelderfer, P.S., 1996. Interlaboratory compari-son of sequence-specific PCR and ligase detection reactionto detect a human immunodeficiency virus type 1 drugresistance mutation. J. Clin. Microbiol. 34, 1849–1853.

Shirasaka, T., Kavlick, M.F., Ueno, T., Gao, W. Y., Kojima,E., Alcaide, M. L., Chokekijchai, S., Roy, B. M., Arnold,E., Yarchoan, R., 1995. Emergence of human immunodefi-ciency virus type 1 variants with resistance to multipledideoxynucleosides in patients receiving therapy withdideoxynucleosides. Proc. Natl. Acad. Sci. USA 92, 2398–2402.

Staszewski, S., Massari, F.E., Kober, A., Gohler, R., Durr, S.,Anderson, K.W., Schneider, C.L., Waterbury, J.A., Bak-shi, K.K., Taylor, V.I., Hildebrand, C.S., Kreisl, C., Hoff-stedt, B., Schleif, W.A., van Briesen, H.,Bubsamen-Waigmann, H., Calandra, G.B., Ryan, J.L.,Stille, W., Emini, E.A., Byrnes, V.W., 1995. Combinationtherapy with zidovudine prevents selection of human im-munodeficiency virus type 1 variants expressing high-levelresistance to L-697,661, a nonnucleoside reverse transcrip-tase inhibitor. J. Infect. Dis. 171, 1159–1165.

Welles, S.L., Jackson, J.B., Yen-Lieberman, B., Demeter, L.,Japour, A.J., Smeaton, L.M., Johnson, V.A., Kuritzkes,D.R., D’Aquila, R.T., Reichelderfer, P.A., Richman, D.D.,Reichman, R., Fischl, M., Dolin, R., Coombs, R.W.,Kahn, J.O., McLaren, C., Todd, J., Kwok, S.,Crumpacker, C.S., for the AIDS Clinical Trials GroupProtocol 116A/116B/117 Team, 1996. Prognostic value ofplasma HIV-1 RNA levels in patients with advanced HIV-1 disease and with little or no zidovudine therapy. J. Infect.Dis. 174, 696–703.

..