A simple and cost-saving phenotypic drug susceptibility testing of HIV-1

It is essential to monitor the occurrence of drug-resistant strains and to provide guidance for clinically adapted antiviral treatment of HIV/AIDS. In this study, an individual patient’s HIV-1 pol gene encoding the full length of protease and part of the reverse transcriptase was packaged into a modified lentivirus carrying dual-reporters ZsGreen and luciferase. The optimal coefficient of correlation between drug concentration and luciferase activity was optimized. A clear-cut dose-dependent relationship between lentivirus production and luciferase activity was found in the phenotypic testing system. Fold changes (FC) to a wild-type control HIV-1 strain ratios were determined reflecting the phenotypic susceptibility of treatment-exposed patient’s HIV-1 strains to 12 HIV-1 inhibitors including 6 nucleoside reverse-transcriptase inhibitors (NRTIs), 4 non-nucleoside reverse transcriptase inhibitors (NNRTIs) and 2 protease inhibitors (PIs). Phenotypic susceptibility calls from 8 HIV-1 infected patients were consistent with 80–90% genotypic evaluations, while phenotypic assessments rectified 10–20% genotypic resistance calls. By a half of replacement with ZsGreen reporter, the consumption of high cost Bright-Glo Luciferase Assay is reduced, making this assay cheaper when a large number of HIV-1 infected individuals are tested. The study provides a useful tool for interpreting meaningful genotypic mutations and guiding tailored antiviral treatment of HIV/AIDS in clinical practice.

culture of clinical strains. However, these assays require fresh peripheral blood mononuclear cells (PBMCs) from donors, are labor-intensive and time-consuming 10 . Recently, several assays either based on a recombinant virus with single-loop infection [11][12][13] , or infectious particles with multiple cycles of replication were developed 14,15 . Although those assays show great potential in phenotypic drug susceptibility testing, their cost remains high due to the expensive Bright-Glo Luciferase Assay System and concerns regarding bio-safety.
In this study, we developed a novel lentivirus-based phenotypic assay with two reporters: luciferase and ZsGreen. The main advantage of dual reporters is that the viral response to HIV-1 inhibitors is easily observed either by cells stained with ZsGreen under the Inverted Fluorescence Microscope 60 h after virus infection or by quantification of luciferase activity. In addition, the HIV-1 based lentivirus being known powerful and safe 16,17 is an ideal vector for HIV-1 drug resistance testing. Using this new assay, the phenotypic susceptibility level of HIV-1 in infected patients was measured and analyzed against a panel of antiviral drugs available in China.

Results
A novel recombinant lentivirus system presenting phenotypic drug susceptibility of HIV-1.
The modified lentivirus packaging system consisted of three plasmids with envelope, packaging and transfer functions carrying a patient's HIV-1 pol gene and dual reporter genes of luciferase and ZsGreen (Supplemental Figs S1 and S2). The patient's HIV-1 pol gene within packaging plasmid psPAX2m-Pol was integrated into recombinant lentiviral genome with envelop plasmid pMD2.G and transfer plasmid pHAGE-CMV-Luc-IRES-ZsGreen in 293FT cells (Fig. 1). The patient's HIV-1 derived lentivirus could functionally present the drug susceptibility by luciferase or ZsGreen reporters in the transduced or infected 293A cells in the presence of different concentrations of antiviral drugs.

Establishment of optimized MOI, cell density and time point in phenotypic drug susceptibility test.
In order to establish optimal assay conditions, the multiplicity of infection (MOI), cell density and time point of detection were optimized with a representative HIV-1 inhibitor in phenotypic susceptibility testing. First, a titer of patient's HIV-1 derived lentivirus was determined within a range of virus dilution folds around 10% ZsGreen positive rate from lentivirus-transduced 293A cells. Secondly, the lentivirus with ZsGreen reporter was used for selecting suitable drug dilution range and pre-titration of the MOI (Fig. 2), in which the drug dilutions corresponding to 10-90% ZsGreen positive cells were primarily chosen for further titration of MOI by the lentivirus with luciferase reporter. When the concentration of Zidovudine (a representative of NRTIs and NNRTIs) decreased in the presence of lentivirus transduced 293A cells, the positive rate of ZsGreen cells increased ( Fig. 2A), in which the drug destroyed the function of HIV-1 pol and inhibited the expression of ZsGreen in the virus infection process. In a different mechanism from NRTIs or NNRTIs, when the concentration of Lopinavir (a representative of PIs) decreased in the presence of lentiviral packaging process in the cells (Fig. 2B-T), the positive rate of ZsGreen cells increased since Lopinavir prevents maturation of the internal structural proteins and of the viral enzymes in a dose-dependent manner.
Scientific RepoRts | 6:33559 | DOI: 10.1038/srep33559 but also the second highest activity of luciferase (Fig. 3C). Finally, stability and reproducibility of the phenotypic resistance testing were evaluated by determining the concentration of 12 antiviral drugs at 50% inhibition (IC 50 ) of lentivirus infection in quadruplicate from three separate tests (Table 1). The data appeared highly reproducibility and stability with coefficient of variation (CV) below 15%.

Analysis of genotypic drug resistance-associated mutations from patient's HIV-1 strains. The
HIV-1 infected patient information was indicated in Table 2, showing that all patients were previously treated with a combination of antiviral drugs. All eight patients had CD4 cell counts below 400 cells/mm 3 and five of them below 200 cells/mm 3 . Patient's HIV-1 RNA load ranged between 155 and 104000 copies/ml. To analyze the drug resistance associated with viral mutations from these patients, the HIV-1 pol genes were amplified and sequenced. Based on the genotypic resistance database (the Stanford University HIV Drug Resistance Database) available for 10 HIV-1 inhibitors involved in this study, the mutations from those HIV-1 strains were associated with genotypic drug resistance to NRTIs, NNRTIs and PIs, respectively ( Table 3). The antagonistic mutations K65R and M184V were found from two patients ID 38290075 and 38290022, respectively (Table 3). Mutation  K65R from the patient ID 38290075 was previously reported susceptible to zidovudine (AZT) and stavudine (D4T), but decreased the susceptibility of HIV-1 to tenofovir, didanosine, abacavir and lamivudine 18 . Mutation M184V from the patient ID 38290022 was described partly susceptible to tamoxifen (TAM), zidovudine, stavudine and tenofovir, but increased the sensitivity of HIV-1 to these antiviral agents 19 . When however, it comes to analyze the drug resistance of HIV-1strains which carried such mutations, they might be predicated resistant by genotypic resistance testing, while the phenotype of drug resistance were actually differed. Therefore, without massive correlative information of genotypic and phenotypic resistance testing of various viral mutants to individual drugs, the genotype testing was not sufficient to predict the drug resistance levels from the viral replication effect of complex mutations as well as new mutations of HIV-1 strains.

Detection of phenotypic drug susceptibility from patients' HIV-1 strains. Eight HIV-1 strains
were measured for drug susceptibility using the recombinant lentivirus system with dual reporters (Supplemental Fig. S3). A wildtype lentivirus whose gag and pol genes are derived from the subtype B strain served as control for calculation of the fold changes (FC). The phenotypic drug susceptibility or resistance level to 12 antiviral drugs are shown in Table 4. Comparing with genotypic resistance analysis available for 10 drugs from the Stanford University HIV Drug Resistance Database (http://hivdb. stanford.edu/), the majority of phenotypic testing calls were concordant with genotypic assessments (Table 4). Approximately 21% (17/80) discrepancy rate was observed between phenotypic and genotypic drug resistance testing from these HIV-1 strains, of which 10% (8/80) drug resistance calls clearly differed by nearly two levels among susceptibility (S), low (L), middle (M) and high (H) resistance levels between two assays (Table 4). For instance, the patient's HIV-1 strain 38290022 was susceptibility to DDI and D4T drugs and high resistance to RPV by phenotype, but middle or high resistance to DDI and D4T and low resistance to RPV by genotype, while HIV-1 strain 38290312 was highly resistant to ETR by phenotype but low resistance by genotype. The results suggested that the newly developed phenotypic susceptibility testing could rectify at least 10% of incorrect calls regarding drug resistance levels predicted by genotypic testing.

Discussion
Drug resistance level or susceptibility testing is considered an important issue for the management of newly HIV-1 infected and treatment-exposed HIV-1 infected patients 9,20 . When a patient was newly diagnosed with   HIV-1 treatment failure, drug susceptibility testing is recommended by the US Department of Health and Human Services (DHHS), International AIDS Society (IAS-USA) and European guidelines 21 . Phenotypic susceptibility assay is optimal to provide accurate and quantitative assessments of HIV-1 strains susceptible to a large number of antiviral drugs, especially for patients receiving antiretroviral for years 9,22,23 . The principle of phenotype testing is to quantify in vitro viral replication in the presence of serial dilutions of antiviral drugs. Drug resistance level (susceptibility) is estimated as FC calculated from a ratio of the IC 50 of the patient strain to the IC 50 of a wild-type control virus. Several phenotypic assays have been previously described for assessing HIV-1 drug resistance in clinical conditions, including the first-generation Antivirogram and PhenoSense assays 14,24,25 , the ExaVir TM Drug assay 26,27 , the modified assays with a single cycle system or single-loop infection [11][12][13] , with two round infection 28 or multiple cycles of replication 15 . In our study, a second-generation lentivirus vector carrying individual patient's HIV-1 pol gene was constructed, in which the phenotypic drug-resistance was represented by dual-reporters in a single cycle of genetically modified lentivirus infected cells. This phenotypic assay has the advantages of simple, cost and time-saving and easy performing for detection of drug susceptibility to HIV-1 in a large panel of antiviral drugs.
By using the established lentivirus system, the susceptibility phenotypes to 12 antiretroviral drugs of eight treatment-exposed patients' HIV-1 strains were assessed against a control wild-type HIV-1 strain. The modified lentiviral vector with single restriction sites of Apa I and Age I appeared to be universally capable of integrating a drug-targeting pol gene from individual HIV-1 strains. The phenotypic susceptibility of HIV-1 to inhibitors on lentivirus-infected cells was reflected by the rate of ZsGreen positive cells observed with an inverted fluorescence microscope or by flow cytometric analysis. In our pre-testing, serial wells covering 10% to 90% of the ZsGreen positive cells observed with an inverted fluorescence microscope were selected for further measurement of luciferase activity with a micro-well plate reader. By narrowing down the range of percentage of ZsGreen full positive or negative wells, the consumption of high cost Bright-Glo Luciferase Assay is reduced in half, making this assay cheaper when a large number of HIV-1 infected individuals are tested. Moreover, the fold-changes of the patient-derived lentiviruses in the presence of HIV-1 inhibitors were quantified in drug susceptibility phenotype, which could precisely guide the antiviral treatment by identifying a combination of susceptibility or low resistance drugs in clinical practice. Similarly, by using ZsGreen reporter alone, the density of green fluorescence of patient-derived lentivirus infected cells was measured by flow cytometry. The phenotypic drug susceptibility of HIV-1 strain could be evaluated by the FC against a wildtype HIV-1 control. However, the expensive flow cytometric machine was required for reading the signal of ZsGreen reporter, which might limit assay's utility.
In conclusion, the study developed a simple and accurate phenotypic testing assay based on the second-generation lentivirus with dual reporters, which facilitated the evaluation of HIV-1 susceptibility or resistance levels to antiviral drugs. In comparison with the conventional phenotypic assays, this testing is cheap, rapid and accurate for quantitative assessment of drug resistance levels of HIV-1 strains to antiviral inhibitors and provides a tool for interpreting the meaningful genotypic mutations and guiding the precise antiviral treatment of HIV/AIDS in clinical practice.

Materials and Methods
Patient samples and antiviral drugs. Plasma samples from HIV-1 infected patients were provided by the center of disease prevention and control (CDC), Shenzhen, China. All patients had been treated with antiviral drugs in the past years. All participating patients signed an informed consent for sample collection and testing. This study was approved by the Medical Ethics Committee of Southern Medical University (permit numbers: NFYY-2008-045), Guangzhou, China. All experiments were carried out in accordance with the approved guidelines.  , of which capital letters in bold with "+ " or "+ + " and underline indicate discrepancies between phenotypic and genotypic resistance for at least one or two levels, respectively. Phenotypic drug resistance is presented as the fold change (FC), of which FC < 3 indicates susceptibility, FC = 3-< 6 low resistance, FC = 6-< 10 middle resistance, FC ≥ 10 high resistance of drug to HIV-1 strains, respectively 6,31,36 .

Recombinant lentiviral vector system with three plasmids carrying dual reporter genes.
Packaging plasmid psPAX2m-Pol carried the patient HIV-1-derived pol gene. Plasmid psPAX2 (# 12260) was a gift from Dr Didier Trono (Lausanne, Switzerland), which contained the pol genes from a wild-type HIV-1 subtype B NL4.3 strain (GenBank access number AF324493.2) 37 . In order to insert the drug resistance gene fragment (pol) of HIV-1 strains into the packaging plasmid psPAX2 at restriction sites Age I and Apa I, the vector was genetically modified for dysfunction of two extra sites of Age I (position nt 7697) and Apa I (position nt 863) by amplification and site-directed mutagenesis (Supplemental Table S1), designated as psPAX2m. The targeted drug resistance genes (which corresponds to the consensus sequences) from HIV-1 infected patients were individually cloned into the psPAX2m at Age I and Apa I restriction sites, which were designated as psPAX2m-Pol and were used for recombinant lentivirus production (Supplemental Fig. S1).
Envelop plasmid pMD2.G. The vector (plasmid #12259) was a gift from Didier Trono (Lausanne, Switzerland). Transfer plasmid pHAGE-CMV-Luc-IRES-ZsGreen contained dual-reporter genes. The CMV promoter appeared to initiate higher transduction efficiency than EF1α promoter in 293A cells 38 . In order to get a stronger signal of luciferase or ZsGreen, the EF1α promoter in the pHAGE-EF1α -IRES-ZsGreen (a gift from Jeng-Shin Lee, Harvard Medical School, USA) was replaced by the CMV promoter pMD2.G. Briefly luciferase gene fragment from pGL3 Luciferase Reporter Vector (Promega, Beijing, China) was sub-cloned into the pHAGE-CMV-IRES-ZsGreen next to the IRES. The transfer vector with dual-reporters was designated pHAGE-CMV-Luc-IRES-ZsGreen and suitable for lentivirus production (Supplemental Fig. S2).

Recombinant lentivirus production and titration.
A total of 3 × 10 6 293FT cells were plated in 75 cm 2 cell culture flask 36 h prior to transfection. When 75% cell confluence was reached, cells were co-transfected with a total of 15 μ g of three plasmid DNAs (2.8 μ g envelop plasmid pMD2.G, 5.2 μ g packaging plasmid psPAX2m-Pol and 7 μ g transfer plasmid pHAGE-CMV-Luc-IRES-ZsGreen) and 45 μ l of X-tremeGENE HP DNA Transfection Reagent (Roche, USA) diluted in 1.5 ml of Opti-MEM medium and incubated at room temperature for 15 min. After cell culturing for 48 h at 37 °C, the lentivirus-containing supernatant was harvested after centrifugation at 3000 rpm at 4 °C for 20 min and filtration with a Millex-HV 0.45 μ m filter (Millipore, USA).
Lentivirus titration was performed as a previously described assay with modifications 39 . Briefly, a total of 5 × 10 5 293A cells were seeded in 6-well plate (Corning, USA), and cultured for 24 h. The cells were inoculated in a series of lentivirus and incubated for 12 h. The cell medium was changed completely, replaced with fresh medium, and further incubated for 48 h. Cells in each well were counted by flow cytometry in order to enumerate the rate of ZsGreen positive cells. The titer was determined by the following formula: (F × C/V) × D. F = frequency of ZsGreen-positive cells (the percentage obtained and divided by 100), C = total number of cells in the well at the time of infection, V = volume of lentivirus diluent in ml, D = lentivirus dilution fold. Following the formula and standardizing the results, lentivirus dilution fold was determined with a range of 1-30% GFP-positive cells observed after viral transduction.
Optimization of phenotypic drug susceptibility test. The lentivirus produced by co-transfection with psPAX2m, pMD2.G and pHAGE-CMV-Luc-IRES-ZsGreen served as a wild-type HIV-1 control of phenotypic resistance since the pol gene within psPAX2m was derived from a wild-type HIV-1 subtype B NL4.3 strain susceptible to HIV-1 inhibitors. To see whether the HIV-1 inhibitor affected the viral infection and protein production, Zidovudine (D4T) was used as a representative antiviral drug for optimizing the testing conditions according to a previous study 12 . The multiplicity of infection (MOI), cell density and time points were optimized to obtain the most suitable luciferase activity in the test. The luciferase activity was measured using the Bright-Glo Luciferase Assay System (Promega, USA) in a micro-well plate reader (Wallik 1420, Perkin Elmer, USA). Each data point was tested in quadruplicate. The susceptibility of HIV-1 pseudotyped lentivirus to those 12 antiviral drugs was determined at three different time points, and the reproducibility of testing was evaluated.
Scientific RepoRts | 6:33559 | DOI: 10.1038/srep33559 Phenotypic drug susceptibility testing for patient's HIV-1 strains. The amplified HIV-1 pol gene sequences were submitted to the Stanford HIV Drug Resistance Database. The genotypic drug resistance of HIV-1 from the treated patients was analyzed with the HIVdb program available from the Stanford University HIV Drug Resistance Database 29 . Based on the aforementioned results, phenotypic drug susceptibility testing for patient's HIV-1 samples was performed using the newly developed lentivirus-based assay. A dose of 10 × MOI of recombinant lentivirus with an individual targeting HIV-1 pol gene in the presence of different concentrations of NRTIs or NNRTIs were applied to infect 1.5 × 10 4 293A cells per well within 96 wells of cell culture plate in the test, and then the activity of luciferase was determined 60 h post-infection. All antiviral drug testing were performed in quadruplicate each test result being derived from three representative experiments.
The antiviral effect of protease inhibitors (PIs) was measured during viral production. Briefly, 4 × 10 4 293FT cells/100 μ l were seeded in each well of 96-well plate and incubated for 36 h. The cells were transfected with 15 μl of mixture with three plasmid DNAs in Opti-MEM and X-tremeGENE HP DNA Transfection Reagent (Roche, USA), and then 8 h later the cell culture was changed with 100 μ l of fresh medium containing a 3-fold serial concentrations of PIs. After 48 h of incubation of the transfected cells with PIs, cells were spun down by centrifugation at 4000 rpm for 20 min at 4 °C. Fifty μ l of virus-containing supernatant were added to the 24 h pre-seeded 293A cells for measuring the activity of luciferase after 60 h incubation from lentivirus transduced cells, respectively. Data analysis. The percentage of inhibition was calculated using the following formula 12 : Viral inhibition rate (%) = (1-RLU in the drug treated group/RLU in the control) × 100% (RLU, relative luciferase activity unit). The drug concentration producing 50% inhibition (IC 50 ) on an individual HIV-1 strain was calculated using nonlinear regression analysis. The phenotypic resistance level of a detected HIV-1 strain to an individual antiviral inhibitor was expressed as a fold-change (FC), the ratio of the IC 50 of the tested strain to the IC 50 of a wild-type strain control 34 . The adjusted criteria for classifying the phenotypic resistance levels (susceptibility) of antiviral drugs to HIV-1 strains were defined as a ratio of FC < 3 indicative of drug susceptibility (S), FC = 3-< 6 indicative of low drug resistance (L), while FC = 6-< 10 and FC ≥ 10 indicative of intermediate or middle (M) and high drug resistance (H), respectively 6,31,36 . All analyses were done with the software SPSS version 21.0.