Rapid, phenotypic HIV-1 drug sensitivity assay for protease and reverse transcriptase inhibitors

Journal of Clinical Virology 13 (1999) 71 80 Rapid, phenotypic HIV-1 drug sensitivity assay for protease and reverse transcriptase inhibitors Hauke Walter a, Barbara Schmidt a, Klaus Korn a, Anne-Mieke Vandamme b, Thomas Harrer c, Klaus U berla a,d, * a Institut für Klinische und Molekulare Virologie, Uni ersität Erlangen-Nürnberg, Erlangen, Germany b Rega Institute for Medical Research, Katholieke Uni ersiteit Leu en, Leu en, Belgium c Institut für Klinische Immunologie und Rheumatologie, Medizinische Klinik III, Uni ersität Erlangen-Nürnberg, Erlangen, Germany d Institut für Virologie, Uni ersität Leipzig, Liebigstr. 24, D-04103 Leipzig, Germany Received 13 November 1998; accepted 28 January 1999 Abstract Background: Development of drug resistance is one of the major reasons for the failure of antiretroviral therapy of HIV-1 infection. Knowing the drug sensitivity resistance profile of viruses present in a patient prior to treatment or change in treatment could help to optimize therapy. Objective: Development of a rapid standardized phenotypic HIV-1 drug sensitivity assay for protease (PR) and reverse transcriptase (RT) inhibitors. Design: The PR gene (codons 1 99) and the 5 part of the RT gene (codons 1 300) of HIV-1 is amplified from the plasma of infected individuals by RT-PCR and ligated into a proviral clone of HIV-1 containing a deletion of the PR gene and the 5 part of the RT gene. Bacteria are transformed with the ligation product and plasmid DNA is prepared from a library of transformed bacteria.the plasmid DNA is transfected into 293 T cells and recombinant virus is harvested from the supernatant of the transfected cells 2 days after transfection. The sensitivity of the recombinant virus is determined with the help of a sensitive indicator cell line. Results: Recombinant viruses were generated with high efficiency. Determination of the drug sensitivity of the recombinant viruses with an indicator cell line was highly reproducible. The recombinant viruses accurately reflected the sensitivity resistance profile of the parental viruses. The phenotypic drug sensitivity determined by this assay correlated well with the treatment history of patients. Conclusion: This assay system should allow rapid, high-throughput analyses of phenotypic HIV-1 drug sensitivity for PR and RT inhibitors. Due to the efficient generation of recombinant viruses, propagation of the recombinant viruses in cell culture is not required prior to the determination of the sensitivity of the recombinant viruses. The risk of selecting fitter non-resistant viruses due to culture conditions is minimized. 1999 Elsevier Science B.V. All rights reserved. Keywords: HIV; Antiretroviral therapy; Resistance * Corresponding author. Tel.: +49-341-9714314; fax: +49-341-9714309. E-mail address: ueberla@medizin.uni-leipzig.de (K. U berla) 1386-6532/99/$ - see front matter 1999 Elsevier Science B.V. All rights reserved. PII: S1386-6532(99)00010-4

72 H. Walter et al. / Journal of Clinical Virology 13 (1999) 71 80 1. Introduction Highly active antiretroviral therapy (HAART) can substantially suppress viral load in many HIV-infected individuals. Suboptimal suppression of virus replication can lead to the development of drug-resistant viruses and to failure of combination therapy. Therefore, it seems to be important to make sure that in each patient all compounds of a treatment regimen are active. Cross-resistance to previously used drugs or transmission of drugresistant viruses can reduce the efficacy of compounds in pre-treated or naive HIV-1-infected individuals, respectively. HIV-1 drug sensitivity assays might help to avoid the inclusion of inactive compounds in the combination therapy. The drug sensitivity of HIV-1 can either be determined genotypically or phenotypically (for review see Perrin and Telenti, 1998). Genotypic drug sensitivity assays are based on the determination of the nucleotide sequence of the patient s virus. Certain point mutations in the RT gene and PR gene confer drug resistance to RT and PR inhibitors, respectively (summarized by Schinazi et al., 1997). These assays only allow the detection of drug resistance, if resistance is due to mutations that have been identified previously. Mutations conferring resistance to a particular drug can also depend on the kind and timing of co-administered antiretroviral drugs (Balzarini et al., 1993). HIV-1 sequence diversity and the reversal of drug resistance by second site mutations (Tisdale et al., 1993) can render interpretation of the sequence data difficult. In contrast to genotypic drug sensitivity assays, phenotypic assays are more labour intensive. Originally, they were based on measuring the replication of virus isolates from patients in the presence or absence of drugs. Disadvantages are the selection of particular virus strains during virus isolation and difficulties in standardization. The recombinant virus assay described by Kellam and Larder (1994), overcomes some of these limitations. The RT gene of HIV-1 from the plasma of infected individuals is amplified by RT-PCR. Co-transfection of the PCR product with a laboratory adapted molecular clone of HIV-1 containing a deletion in the RT gene into a CD4 + T cell line leads to outgrowth of recombinants containing the RT of the patients virus in the genome of the laboratory adapted virus. The sensitivity of the recombinants is then determined in a standardized cell killing assay. A similar approach was subsequently used for PR inhibitors (Maschera et al., 1995; Boucher et al., 1996) and for both PR and RT inhibitors simultaneously (Hertogs et al., 1998). However, these assays have some disadvantages. Due to the low frequency of recombination, culture periods of about 10 days are required for the outgrowth of the recombinant virus. During this time selection against replication-impaired, drug-resistant viruses (Zhang et al., 1997; Zennou et al., 1998); might occur. In addition, it is very time consuming to control how many independent recombinants have been generated. This is important to ensure that the recombinants are representative of the patient s virus and not just represent one clone. Concerning the read-out system, a disadvantage of the cell killing assay is the strict dependance on the dose of infection, requiring the determination of the median tissue culture infectious dose (TCID 50 )ofthe recombinants prior to the drug sensitivity assay. We therefore introduced a number of substantial modifications which made the generation of the recombinant viruses more efficient and improved the read-out for the drug sensitivity assay. 2. Material and methods 2.1. Construction of NL4-3 PRT5 and recombinant irus DNA The 3 part of the RT gene was amplified with the primers H PRT5 (TGCAGGGCCCGG- GTACGTATCTAGAACTGGCAGAAAACAG- GGA) and H4570a (ATGTATTGTTTTTAC- TGGCCAT) from pnl4-3 (Genbank entry M19921). Underlined sequences mark relevant restriction sites. The PCR product was digested with ApaI and MscI and ligated into the ApaI- MscI site of pnl4-3. The resulting plasmid, designated pnl4-3 PRT5, contains a deletion from nucleotide 2009 to 3446 of pnl4-3. The deletion is flanked 5 by the ApaI restriction site, which is present in pnl4-3, and 3 by a XbaI site, which was created by mutating the first nucleotide of the

H. Walter et al. / Journal of Clinical Virology 13 (1999) 71 80 73 six base pair recognition site of XbaI. Since the mutated nucleotide is cleaved off by Xba1 digestion, this point mutation will not be present in the recombinant virus (REC). Using the primers H2001s (TGCAGGGCCCCTAGRAAAARGG- GCTGTT, with R being A or G) and H3454a (AGTGCTAGCTCTGCTTCTTYTGTTAGTG- GTA, with Y being C or T) a fragment comprising the region deleted in pnl4-3 PRT5 (PRT5) was amplified by RT-PCR from virus containing samples. Viral RNA was extracted by the QI- Aamp Viral RNA Kit (Qiagen) and reverse-transcribed using Superscript (Gibco BRL). Amplification was performed with the Expand High Fidelity PCR system (Boehringer Mannheim) for 30 cycles at 95 C for 20 s, 60 C for 30 s, and 72 C for 2 min. During the last 20 cycles, the extension time was prolonged by 20 s per cycle. In case of plasma samples, a heminested PCR was performed with primer pairs H2001s/H3532a (TTCTGCTATTAAGTCTTTT- GATGGGTCA) and H2001s/H3454a under the conditions described above. The PCR product was purified using Gene-Clean (Bio 101) and digested with ApaI and NheI. After purification with Gene-Clean, the PCR product was ligated into ApaI and XbaI digested pnl4-3 PRT5. Ultracompetent XL2-Blue cells (Stratagene) were transformed as described by the manufacturer. Plasmid DNA was prepared from a pool of transformed bacteria or single clones using the Plasmid Mini Kit (Qiagen). 2.2. Generation of recombinant irus stocks Plasmid DNA (2.5 g) containing the recombinant virus genome or DNA from the ligation reaction was co-transfected with 2.5 g calf thymus carrier DNA (Boehringer Mannheim) into 293 T cells using the Superfect transfection kit (Qiagen) as described by the manufacturer. Two days after transfection, the supernatant of the transfected cells was cleared by centrifugation at 1200 rpm for 10 min and stored in aliquots at 80 C. The TCID 50 of the supernatant was determined on CEMx174-SIV-SEAP cells (Means et al., 1997) as described (Johnson and Byington, 1990). To determine the complexity of the virus pool, the number of virus producing cells was determined by limiting dilution co-cultures of 293 T cells harvested 1 day after transfection with CEMx174-SIV-SEAP. 2.3. Measurement of drug sensiti ity To determine virus replication, CEMx174-SIV- SEAP cells (25 000/well) were infected in triplicates in 96 well plates in a final volume of 200 l. If these cells are infected with HIV-1, the viral Tat protein will transactivate the LTR leading to a strong increase in SEAP activity of the culture supernatant. Since the titer and the replication capacity of the recombinant virus stocks were variable, the amount of virus present in each virus stock was estimated by infecting CEMx174-SIV- SEAP cells with 1, 5, 10, or 20 l virus containing supernatant. The minimal volume of the virus stock resulting in at least 10 000 relative light units (RLU) in 20 l supernatant of CEMx174- SIV-SEAP cells 3 days after infection with the virus stock was determined as described (Means et al., 1997) using the Phospha-light-kit (Tropix, Bedford, MA, USA). If 20 l virus containing supernatant was not sufficient to result in 10 000 RLUs, CEMx174 cells were infected with the virus stock derived from 293 T cells and cultivated for a few days to prepare a higher titered virus stock. To determine the actual sensitivity to RT inhibitors, CEMx174-SIV-SEAP cells were infected with the 10 000 RLU dose of the recombinant virus stock in the presence of 0, 0.01, 0.1, 1, 10, and 100 M of drug and the SEAP activity was determined 3 days later. For PR inhibitors, the inoculum was reduced by a factor of five and the SEAP activity was determined 4 days (Saquinavir, Indinavir) or 5 days (Nelfinavir, Ritonavir) after infection. For drugs to which only low level resistance was observed (Zalcitabine, Didanosine, Stavudine, Saquinavir) intermediate drug concentrations of 0.03, 0.3, 3 and 30 M were also used. As a control, recombinant NL4-3 was included in every assay. Abacavir, Zidovudine, and Lamivudine were kindly provided by Glaxo Wellcome; Zalcitabine and Saquinavir by Hoffmann la Roche; and Didanosine and

74 H. Walter et al. / Journal of Clinical Virology 13 (1999) 71 80 Stavudine by Bristol-Myers Squibb. Nevirapine was obtained from Boehringer Ingelheim; Delavirdine from Pharmacia & Upjohn; Indinavir from Merck Sharp & Dohme; Ritonavir from Abbott; and Nelfinavir from Agouron Pharmaceuticals. 3. Results A schematic representation of the assay is given in Fig. 1. The PR gene (codons 1 99) and the 5 part of the RT gene (codons 1 300) were deleted from the proviral DNA of the HIV-1 molecular clone NL4-3. The first 300 amino acids of RT contain all known drug resistance mutations with the exception of a mutation at amino acid 333, which was recently found to be involved in resistance to Zidovudine in the presence of a Lamivudine associated resistance mutation (Kemp et al., 1998). In addition, the deletion comprises the PR cleavage site between NC/p1 and p1/p6, which is sometimes mutated in viruses resistant to PR inhibitors (Zhang et al., 1997; Mammano et al., 1998; Zennou et al., 1998). The deletion was introduced in a way that two unique restriction sites were generated. Using nested RT- PCR, the PR gene and the relevant part of the RT gene is amplified as one fragment (PRT5) from the plasma of HIV-infected patients. The PCR Fig. 1. Overview of the drug sensitivity assay. RT5: 5 part of the RT gene; RT3: 3 part of the RT gene; PRT5: PCR fragment spanning PR and RT5. fragment and the deletion mutant of NL4-3 are ligated and transformed into E. coli. Plasmid DNA is prepared from a pool of transformed E. coli clones. The plasmid DNA is transfected into 293 T cells. The supernatant of the transfected cells containing the recombinant viruses is harvested 2 days after transfection. A sensitive indicator cell line (CEMx174-SIV-SEAP, Means et al., 1997) is infected with these supernatants in the presence of different amounts of the drugs to be tested and virus replication is monitored as a function of drug concentration. The indicator cells contain the SEAP gene under the control of the simian immunodeficiency virus LTR (Means et al., 1997). If these cells are infected by HIV-1, the viral Tat protein will transactivate the LTR leading to a strong increase in SEAP activity in the culture supernatant, which can be monitored easily. To determine whether the recombinant virus would have a similar sensitivity resistance profile as the virus used as template for the PCR reaction, the PRT5 region of two molecular clones of HIV-1 with defined drug resistance mutations in the RT (RTMDR1, Larder et al., 1993) or PR (designated PRMDR1; originally described by Condra et al., 1995) region was amplified and ligated with NL4-3 PRT5. Recombinant virus stocks (REC-PR-MDR1, REC-RT-MDR1) were generated by transfection of the ligation reaction into 293 T cells and propagating the recombinant viruses on a CD 4+ cell line. The sensitivity of the recombinant viruses and the parental viruses to RT inhibitors or PR inhibitors was determined by infecting the indicator cells in the presence of different concentrations of these drugs (Table 1). Both recombinant viruses had similar sensitivity resistance profiles as the respective parental viruses, demonstrating that the recombinant viruses accurately reflect the sensitivity of the parental virus to PR and RT inhibitors. The sensitivity of a pair of primary isolates that were recovered before and after Ritonavir treatment (Schmit et al., 1996) was also determined. The recombinant virus of the posttreatment isolate was highly resistant to Ritonavir and resistance to other PR inhibitors was also observed (Table 2). The degree of resistance of the

H. Walter et al. / Journal of Clinical Virology 13 (1999) 71 80 75 Table 1 Comparison of drug sensitivity of parental and recombinant virus Drug IC 50 ( M) a of virus NL4-3 RT-MDR1 b Rec-RT-MDR1 PR-MDR1 c Rec-PR-MDR1 Zidovudine 0.069 9.712 6.746 0.322 d 0.056 Zalcitabine 0.117 1.635 0.697 0.211 0.192 Didanosine 0.722 7.577 2.589 1.034 0.571 Stavudine 0.551 4.945 1.889 0.948 0.684 Lamivudine 0.259 0.220 0.593 0.087 0.088 Abacavir 0.553 5.118 6.485 0.478 0.796 Nevirapine 0.063 9.926 6.627 0.052 0.063 Delavirdine 0.037 0.748 0.449 0.028 0.046 Indinavir 0.050 0.089 0.077 0.607 0.572 Saquinavir 0.008 0.026 0.016 0.019 0.026 Ritonavir 0.066 0.074 0.047 4.967 6.867 Nelfinavir 0.088 0.092 0.078 0.437 0.403 a 50% inhibitory concentration. b HIV-1 clone resistant to multiple RT inhibitors (Larder et al., 1993). c HIV-1 clone resistant to multiple PR inhibitors (Condra et al., 1995). d In repeat experiments the IC 50 of Zidovudine for PR-MDR1 was 0.077 and 0.104 M. post-treatment recombinant virus was similar to previously published data for the post-treatment isolate (Schmit et al., 1996) further supporting the accuracy of the drug sensitivity assay. To determine the sensitivity of patients viruses, a number of modifications were introduced which made the whole assay more sensitive, faster, and less labour intensive. To increase the sensitivity of the PCR reaction a hemi-nested PCR was performed, which allowed the isolation of the PRT5 Table 2 Phenotypic drug sensitivity of a pair of recombinant viruses derived from patient isolates before and after Ritonavir treatment Drug Ritonavir Indinavir Saquinavir Nelfinavir IC 50 a SD b ( M) Pre-treatment c Post-treatment c (fold increase) 0.079 0.009 7.246 0.605 (92.1) 0.069 0.003 0.676 0.03 (9.8) 0.019 0.003 0.075 0.015 (3.9) 0.068 0.002 0.686 0.079 (10.1) a 50% inhibitory concentration. b Standard deviation. c Isolates derived from patient B2 described by Schmit et al. (1996). fragments from 86% (n=97) of patients with viral loads higher than 500 copies viral RNA/ml plasma (branched DNA assay, Chiron). The PRT5 fragment could also be amplified from six out of 22 (27%) patients with viral loads lower than 500 copies/ml plasma. The failure to isolate a PCR product from some patients with viral loads above 500 copies/ml plasma might therefore be due to the presence of different viral strains in different patients. Attempts to analyze the drug sensitivity of a number of patient samples revealed that direct transfection of the ligation reaction into 293 T cells gave quite variable titers of the recombinant viruses, which often required the amplification of the recombinant viruses prior to the drug sensitivity test (data not shown). To increase the viral titers in the supernatant of the transfected 293 T cells, E. coli were transformed with the ligation product of the PRT5 fragment from a multidrugresistant patient with the pnl4-3 PRT5. More than 1000 ampicillin-resistant colonies were obtained per l ligation reaction. Restriction analysis of plasmid DNA isolated from a pool of transformed colonies revealed that more than 50% of the colonies contained a properely ligated NL4-3 PRT5 with a PRT5 insert. This was confirmed,

76 H. Walter et al. / Journal of Clinical Virology 13 (1999) 71 80 Table 3 Reproducibility of phenotypic drug resistance assay Drug IC 50 ( M) a /fold resistance of recombinant virus stock b coefficient of 1a 1b 1c 2 3 Mean variation Zidovudine 6.7/96 3.6/52 6.4/92 9.1/132 10/145 7.2/103 0.31 Zalcitabine 5.3/17 2.7/9 3.2/10 7.7/24 10/32 3.7/12 0.42 Didanosine 8/6.3 2.7/2.1 4.7/3.7 4.6/3.6 8.8/6.9 5.2/4.5 0.40 Lamivudine 10 2 / 386 10 2 / 386 10 2 / 386 10 2 / 386 10 2 / 386 N.a./n.a. N.a./n.a. Stavudine 3.2/5.8 0.67/1.2 2.2/3.9 5.4/9.8 10.1/18 4.3/7.7 0.60 Nevirapine 0.065/0.9 0.054/0.9 0.031/0.5 0.061/0.8 0.041/0.56 0.050/0.7 0.25 Ritonavir 5.6/85 5.9/90 6.1/92 7.3/111 3.3/50 5.6/86 0.23 Indinavir 0.89/18 0.95/19 0.88/18 0.95/19 0.89/18 0.91/18 0.03 Saquinavir 0.076/9.9 0.062/8.1 0.061/8 0.11/14 0.05/7 0.072/9.4 0.29 Nelfinavir 0.56/13 0.57/13 0.74/17 0.93/21 0.52/12 0.66/15 0.23 a 50% inhibitory concentration. b Three recombinant virus stocks were prepared from the same plasma sample. Drug sensitivity was determined in independent assays three times for one recombinant virus stock (1a c) and once for the other two virus stocks (2 and 3). N.A.: not applicable. by restriction analyses of plasmid DNA prepared from ten randomly selected individual clones of transformed bacteria. Eight of these ten clones contained a properly ligated recombinant virus (data not shown). Plasmid DNA from the pool of transformed clones was then transfected into 293 T cells. The virus titer in the supernatant of the transfected 293 T cells was in the range of 10 5 median tissue culture infectious doses TCID 50 /ml, which was high enough for directly performing the drug sensitivity assay. To ensure that the complexity of the plasmid pool would be maintained in the pool of the recombinant viruses, the minimal number of independently transfected cells was determined by a limiting dilution co-culture of the transfected 293 T cells with the indicator cell line 1 day after transfection. The TCID 50 /cell varied from 0.1 to 0.6, corresponding to at least 10 5 independently transfected cells. Using these experimental conditions, the reproducibility of the drug sensitivity assay was tested with a recombinant virus derived from the plasma sample of a multidrug-resistant patient. The IC 50 values and fold resistance values of three independent assays performed with the same virus stock on 3 different days differed by less than a factor of three for most drugs tested (Table 3, stock 1a c). A larger variation was only observed for Stavudine which differed by a factor of 4.8. To further test the interassay variability, two new recombinant virus stocks were prepared from the same plasma sample 4 and 6 months later and tested in independent assays (Table 3, stock 2 and 3). The IC 50 values and fold resistance values of these new stocks were in same range as observed for the first stock. In one of the five stavudine sensitivity tests the IC 50 value and the fold resistance value was 6.4 fold lower than the mean of the determinations. For all other tested drugs, single values of all five determinations varied by less than a factor of three from the mean of these five determinations (Table 3), indicating that phenotypic drug sensitivity can be reproducibly determined. A comparison of the results of the phenotypic drug sensitivity assay with the treatment history of three patients and the nucleotide sequences of the viruses present in the plasma of these patients is shown in Table 4. In the untreated patient (patient 1), full sensitivity to all drugs tested is revealed by phenotypic and genotypic drug sensitivity analyses. Patient 2 was treated with Didanosine, Stavudine, Lamivudine, Loviride and Saquinavir at the time of analysis. A viral load of 53 000 HIV-1 RNA copies/ml plasma indicated inefficient therapy. Phenotypic drug resistance was detected against all used nucleoside analogue RT

Table 4 Comparison of phenotypic drug sensitivity with genotype and treatment history Drug Patient 1 (untreated) Patient 2 (5/97) Patient 3 (12/97) -Fold resistance a Genotype b -Fold resistance a Genotype b Treatment -Fold resistance a Genotype b Treatment Zidovudine 2 Wt 23 67N, 70R, 4/94-9/94; 8/96-100 41L, 67N, 4/94-4/96; 6/97-214F, 219E 11/96 210W,211K, 214F, 11/97 215Y Zalcitabine 1 Wt 3 184V 9/94-8/95 21 184V 12/94-1/96 Didanosine 1 Wt 3 184V 11/96-9 184V 4/96-2/97; 11/97- Stavudine 1 Wt 4 Wt 11/96-6 Wt 4/96-7/96; 11/97- Lamivudine 1 Wt 497 184V, 214F 8/95-1380 184V, 211K, 214F 1/96-4/96; 7/96-2/97; 6/97-11/97 Nevirapine 1 Wt 15 181C (Loviride 8/95-) 1 Wt No Saquinavir 1 Wt 1 73S 11/96-5/97 7 10I, 54V, 82A 1/96-4/96; 8/97- Indinavir 1 Wt 1 20R, 73S 8/96-11/96 28 10I, 24I, 46L, 54V, 5/96-2/97 63P, 71V, 82A Nelfinavir 1 Wt 1 36I no 19 63P, 77I 6/97- Ritonavir 1 Wt 1 20R, 36I no 96 54V, 71V, 82A No a Ratio of IC of patient-derived recombinant virus and NL4-3. b Codons associated with drug resistance were taken from Schinazi et al., (1997). Wt: wild type codon at all positions associated with resistance to indicated drug; numbers indicate the codon mutated in RT (for RT inhibitors) or PR (for PR inhibitors) and are followed by the one letter abbreviation of the deduced amino acid at this position. H. Walter et al. / Journal of Clinical Virology 13 (1999) 71 80 77

78 H. Walter et al. / Journal of Clinical Virology 13 (1999) 71 80 inhibitors and against Nevirapine (the sensitivity to Loviride was not determined phenotypically, but cross-resistance to Nevirapine is frequently observed). Sequencing of the RT gene of the patient s virus also revealed drug resistance mutations against all used RT inhibitors except for Stavudine. Phenotypic drug sensitivity assay indicated full sensitivity to all PR inhibitors tested including Nelfinavir. Genotypic analysis revealed three drug resistance associated mutations in the PR gene. After Saquinavir was replaced by Nelfinavir, the viral load in this patient dropped to 920 copies/ml plasma within 2 months. The lack of therapy response to Saquinavir in the absence of resistance may be due to the low bioavailability of the Saquinavir formulation used at that time (IN- VIRASE). However, even after change to Nelfinavir the low viral load level could be maintained for only another 2 months, since only one of the five compounds used was still active. Patient 3 had also been administered various treatment regimens. He harboured viruses resistant to all nucleoside analogue RT inhibitors and PR inhibitors tested in phenotypic and genotypic drug sensitivity assays. In agreement with the observed multidrug resistance, changing the treatment regimen from Zidovudine, Lamivudine, Saquinavir, and Nelfinavir to Didanosine, Stavudine and Nelfinavir did not reduce the viral load levels by more than a factor of three. Full sensitivity to non-nucleoside RT inhibitors, a class of compounds not used in this patient before, suggests that inclusion of these compounds into the treatment regimen might have been beneficial. 4. Discussion A number of modifications were introduced in the recombinant virus assay for the determination of HIV-1 drug sensitivity as described by Kellam and Larder (1994), which overcome some of its limitations. By ligating the patient-derived PCR fragment with a deletion mutant of a molecular clone of HIV-1, a larger number of independent recombinants can be obtained than by homologous recombination. A plasmid library of these recombinant viruses was first amplified in bacteria and plasmid DNA prepared from a pool of transformed bacteria was then transfected into 293 T cells. A library of recombinant viruses was harvested from the supernatant of transfected cells 2 days after transfection. Since no virus replication can occur in 293 T cells due to the lack of the CD 4 receptor, there should be no selective pressure acting on the library of recombinant viruses. This might be beneficial since there is evidence that drug-resistant viruses emerging in patients can have reduced replication properties (Zhang et al., 1997; Zennou et al., 1998). In case of viruses resistant to PR inhibitors, mutations were detected primarily in the C-terminal PR cleavage sites of gag, which counteract the replication defect (Zhang et al., 1997; Mammano et al., 1998; Zennou et al., 1998). Since the patientderived PRT5 fragment spans the PR cleavage sites between NC/p1 and p1/p6, most of the C- terminal mutations are also part of the recombinant virus. In contrast to the previously described recombinant virus assays, the complexity of the plasmid library can be easily determined by plating out an aliquot of the transformed bacteria and counting the number of transformed colonies. This ensures that the recombinant viruses truly represent the viruses in the plasma sample. The only bottleneck in the complexity of the library of the recombinant virus seems to be the number of templates in the PCR reaction, particularly in patients with low viral loads. To increase the efficiency of the PCR reaction, we only amplified the 5 part of RT and not the entire RT gene since amplification of longer PCR products results in a loss of sensitivity. However, one has to bear in mind that there is one mutation outside the PRT5 fragment which has recently been shown to be involved in resistance to Zidovudine in the presence of a resistance mutation to Lamivudine (Kemp et al., 1998). Therefore, resistance to Zidovudine can not be excluded if the recombinant virus is resistant to Lamivudine and sensitive to Zidovudine. We also changed the read-out system for determination of the drug sensitivity. Due to the higher sensitivity of the indicator cell line used, results are obtained 3 5 days after infection rather than 6 days after infection. In addition, the cell killing assays used previously have a narrow

H. Walter et al. / Journal of Clinical Virology 13 (1999) 71 80 79 linear range, requiring the careful titration of the recombinant virus stocks prior to the drug sensitivity assay, while in the assay described here a 3 day infection period of the indicator cells with the recombinant virus stocks is used to determine the infectious dose. The assay was validated by a number of independent approaches. Using multidrug-resistant molecular clones, the recombinant and the parental viruses were shown to have very similar drug sensitivity profiles. The level of resistance of recombinant viruses derived from a pair of patient isolates recovered before and after PR inhibitor treatment was comparable with previously published resistance levels (Schmit et al., 1996). The assay was reproducible and IC 50 values differed by less than a factor of two from the mean for most drugs tested. If required, occasional larger variations as observed for AZT (Table 1, PR- MDR1) or Stavudine (Table 3, 1b) can be overcome by including intermediate drug concentrations in the drug sensitivity assay or by repeat experiments. The phenotypic drug sensitivity was in agreement with the reported treatment history. Phenotypic and genotypic drug sensitivity assays gave similar results in most cases analyzed so far. However, the genotype did not always correlate with phenotypic drug sensitivity, particularly in case of PR inhibitors (data not shown). A comparison of phenotypic drug sensitivity, genotype, and the outcome of treatment in a larger number of patients is needed, in order to determine the relative importance of phenotypic and genotypic drug sensitivity testing. It is important to note, that both approaches only assess the sensitivity of the dominant virus variant. Since drug-resistant viruses present as a minority species might lead to rapid therapy failure, the clinical impact of drug sensitivity assays remains to be determined. The phenotypic drug sensitivity assay described is now routinely used to monitor drug sensitivity and more than 60 samples have been analyzed so far. Clinical applications for routine phenotypic drug resistance testing are the initiation of antiretroviral therapy and any change in the treatment regimen to ensure that all compounds of the combination therapy are active against the virus of each HIV-infected individual. Acknowledgements The authors thank B. Moschik and C. Paatz for excellent technical assistance and E. De Clercq and B. Fleckenstein for helpful discussion and continuous support. The indicator cell line was kindly provided by R.Means and R.C. Desrosiers. The authors thank M. Helm, J.-C. Schmit, and B. Clotet, for providing clinical samples and information on treatment history. The following reagent was obtained through the AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH: HIV-1 L10R/ M46I/ L63P/ V82T/ I84V from E. Emini and HIV-1 RTMDR1 from B. Larder. References Balzarini J, Karlsson A, Perez-Perez MJ, Camarasa MJ, Tarpley WG, De Clercq E. Treatment of human immunodeficiency virus type 1 (HIV-1)-infected cells with combinations of HIV-1-specific inhibitors results in a different resistance pattern than does treatment with singledrug therapy. J Virol 1993;67:5353 9. Boucher CA, Keulen W, van Bommel T, Nijhuis M, de Jong DJM, Schipper P, Back NK. Human immunodeficiency virus type 1 drug susceptibility determination by using recombinant viruses generated from patient sera tested in a cell-killing assay. Antimicrob Agents Chemother 1996;40:2404 9. Condra JH, Schleif WA, Blahy OM, Gabryelski LJ, Graham DJ, Quintero JC, Rhodes A, Robbins HL, Roth E, Shivaprakash M, Titus D, Yang T, Teppler H, Squires KE, Deutssch PJ, Emini EA. In vivo emergence of HIV-1 variants resistant to multiple protease inhibitors. Nature 1995;374:569 71. Hertogs K, de Bethune MP, Miller V, Ivens T, Schel P, Van Cauwenberge A, Van Den Eynde CGV, Azijn H, Van Houtte M, Peeters F, Staszewski S, Conant M, Bloor S, Kemp S, Larder B, Pauwels R. A rapid method for simultaneous detection of phenotypic resistance to inhibitors of protease and reverse transcriptase in recombinant human immunodeficiency virus type 1 isolates from patients treated with antiretroviral drugs. Antimicrob Agents Chemother 1998;42:269 76. Johnson V, Byington RE. Quantitative assays for virus infectivity. In: Aldovini A, Walker BD, editors. Techniques in HIV research. New York: Stockton Press, 1990:71 6. Kellam P, Larder BA. Recombinant virus assay: a rapid, phenotypic assay for assessment of drug susceptibility of human immunodeficiency virus type 1 isolates. Antimicrob Agents Chemother 1994;38:23 30.

80 H. Walter et al. / Journal of Clinical Virology 13 (1999) 71 80 Kemp SD, Shi C, Bloor S, Harrigan PR, Mellors JW, Larder BA. A novel polymorphism at codon 333 of human immunodeficiency virus type 1 reverse transcriptase can facilitate dual resistance to zidovudine and L-2 3 -dideoxy-3 -thiacytidin. J Virol 1998;72:5093 8. Larder BA, Kellam P, Kemp SD. Convergent combination therapy can select viable multidrug-resistant HIV-1 in vitro. Nature 1993;365:451 3. Mammano F, Petit C, Clavel F. Resistance-associated loss of viral fitness in human immunodeficiency virus type 1: phenotypic analysis of protease and gag coevolution in protease inhibitor-treated patients. J Virol 1998;72:7632 7. Maschera B, Furfine E, Blair ED. Analysis of resistance to human immunodeficiency virus type 1 protease inhibitors by using matched bacterial expression and proviral infection vectors. J Virol 1995;69:5431 6. Means RE, Greenough T, Desrosiers RC. Neutralization sensitivity of cell culture-passaged simian immunodeficiency virus. J Virol 1997;71:7895 902. Perrin L, Telenti A. HIV treatment failure: testing for HIV resistance in clinical practice. Science 1998;280:1871 3. Schinazi RF, Larder BA, Mellors J. Mutations in retroviral genes associated with drug resistance. Int Antivir News 1997;5:129 40. Schmit JC, Ruiz L, Clotet B, Raventos A, Tor J, Leonard J, Desmyter J, De Clercq E, Vandamme AM. Resistance-related mutations in the HIV-1 protease gene of patients treated for 1 year with the protease inhibitor ritonavir (ABT-538). AIDS 1996;10:995 9. Tisdale M, Kemp SD, Parry NR, Larder BA. Rapid in vitro selection of human immunodeficiency virus type 1 resistant to 3 -thiacytidine inhibitors due to a mutation in the YMDD region of reverse transcriptase. Proc Natl Acad Sci USA 1993;90:5653 6. Zennou V, Mammano F, Paulous S, Mathez D, Clavel F. Loss of viral fitness associated with multiple Gag and Gag-Pol processing defects in human immunodeficiency virus type 1 variants selected for resistance to protease inhibitors in vivo. J Virol 1998;72:3300 6. Zhang YM, Imamichi H, Imamichi T, Lane HC, Falloon J, Vasudevachari MB, Salzman NP. Drug resistance during indinavir therapy is caused by mutations in the protease gene and in its Gag substrate cleavage sites. J Virol 1997;71:6662 70..