Development of a rapid cell-fusion-based phenotypic HIV-1 tropism assay

Introduction A dual split reporter protein system (DSP), recombining Renilla luciferase (RL) and green fluorescent protein (GFP) split into two different constructs (DSP1–7 and DSP8–11), was adapted to create a novel rapid phenotypic tropism assay (PTA) for HIV-1 infection (DSP-Pheno). Methods DSP1–7 was stably expressed in the glioma-derived NP-2 cell lines, which expressed CD4/CXCR4 (N4X4) or CD4/CCR5 (N4R5), respectively. An expression vector with DSP8–11 (pRE11) was constructed. The HIV-1 envelope genes were subcloned in pRE11 (pRE11-env) and transfected into 293FT cells. Transfected 293FT cells were incubated with the indicator cell lines independently. In developing the assay, we selected the DSP1–7-positive clones that showed the highest GFP activity after complementation with DSP8–11. These cell lines, designated N4R5-DSP1–7, N4X4-DSP1–7 were used for subsequent assays. Results The env gene from the reference strains (BaL for R5 virus, NL4-3 for X4 virus, SF2 for dual tropic virus) subcloned in pRE11 and tested, was concordant with the expected co-receptor usage. Assay results were available in two ways (RL or GFP). The assay sensitivity by RL activity was comparable with those of the published phenotypic assays using pseudovirus. The shortest turnaround time was 5 days after obtaining the patient's plasma. All clinical samples gave positive RL signals on R5 indicator cells in the fusion assay. Median RLU value of the low CD4 group was significantly higher on X4 indicator cells and suggested the presence of more dual or X4 tropic viruses in this group of patients. Comparison of representative samples with Geno2Pheno [co-receptor] assay was concordant. Conclusions A new cell-fusion-based, high-throughput PTA for HIV-1, which would be suitable for in-house studies, was developed. Equipped with two-way reporter system, RL and GFP, DSP-Pheno is a sensitive test with short turnaround time. Although maintenance of cell lines and laboratory equipment is necessary, it provides a safe assay system without infectious viruses. With further validation against other conventional analyses, DSP-Pheno may prove to be a useful laboratory tool. The assay may be useful especially for the research on non-B subtype HIV-1 whose co-receptor usage has not been studied much.


Introduction
A new class of drugs to combat HIV-1 infection emerged in 2007 with the marketing approval of maraviroc, a small molecule that binds specifically to the CCR5 co-receptor to block viral attachment and entry [1]. While entry inhibitors are a welcome addition to the antiretroviral arsenal, one problem with this new class of drugs is that treatment is effective only against viruses with the specified co-receptor usage. HIV-1 tropism is defined by the chemokine coreceptors used for viral attachment: R5-tropic viruses use CD4/CCR5, X4-tropic viruses use CD4/CXCR4 and R5X4or dual-tropic viruses use both CD4/CCR5 and CD4/CXCR4 [2Á4]. In clinical treatment with maraviroc, the presence of X4-or dual-tropic viruses is associated with treatment failure [5,6], and a tropism assay is mandatory before treatment initiation.
HIV-1 tropism may be examined genotypically or phenotypically. Genotypic tropism assay (GTA) is based on DNA amplification and sequencing of the third variable (V3) region of the envelope glycoprotein gp120, shown by genetic mapping to be the major determinant of HIV-1 tropism [ 7Á 10]. GTA has advantages of platform portability, low cost and rapid turnaround time [11]; however, the interpretation of the sequences is complicated because of the high variability [12]. The assay used in association with maraviroc treatment is the phenotypic tropism assay (PTA) Trofile TM (Monogram Biosciences Inc., CA, USA), a CD4 cell culture assay using replication-defective pseudoviruses [13]. Although Trofile TM is considered the gold standard, a simpler and effective PTA would be useful.
Here, we describe a novel, cell-fusion-based PTA that uses a dual split reporter protein system (DSP) [14,15] to measure HIV-1 tropism by both Renilla luciferase (RL) activity and green fluorescent protein (GFP) activity. We validated the DSP-Pheno assay using HIV-1 reference strains and applied the assay to test clinical samples from patients with HIV-1 infection.

Methods
Approval of the study and recombinant DNA experiments Plasma samples from HIV-1-positive patients attending the hospital affiliated with the Institute of Medical Science, the University of Tokyo (IMSUT) were collected and kept frozen until use. Patients provided written informed consent, and the study was approved by the Institutional Review Board of the University of Tokyo (approval number 20-31). Recombinant DNA experiments used in this work were approved by the Institutional Review Board (approval number 08-30), and by the review board in the Ministry of Education, Culture, Sports, Science and Technology (MEXT; approval number .

Construction of DSP expression plasmids
The DSP system utilizes a pair of chimeric reporter proteins, DSP 1Á7 and DSP 8Á11 , each of which is a fusion of split green fluorescent protein (spGFP) and split Renilla luciferase (spRL) [15]. DSP 1Á7 fuses the N-terminal region of RL (amino acids 1Á229) to the N-terminal region of GFP (amino acids 1Á157), with a linker sequence separating the two regions. DSP 8Á11 has the complementary structure, with the C-terminal region of GFP (amino acids 158Á231), fused to the C-terminal region of RL (amino acids 230Á311), also separated by a linker sequence. When both reporter proteins are present in the same cell, they each recover full activity.
An expression vector, pRE11 (Figure 1), was constructed for the co-expression of DSP 8Á11 and HIV-1 env by multiple rounds of PCR and subcloning. Source plasmids were pIRES2-AcGFP1 (Clontech), pmOrange (Clontech), pDSP 8Á11 [15] and pmirGLO (Promega). pRE11 incorporated multiple cloning sites under the PGK promoter for the insertion of HIV-1 env (Shown as 5?-XbaI-XhoI-3? in Figure 1b). Necessary restriction enzyme cleavage sites used for construction, including multiple cloning sites (XbaI-MluI-SwaI-AgeI-XhoI), were created using synthetic oligonucleotides and PCR. A CMV promoter drives pDSP 8Á11 directly. The same CMV promoter expresses mOrange with a nuclear localization signal that serves as a marker for successful transfection via internal ribosomal entry site (IRES). All PCR fragments were confirmed by sequencing.

NP-2-derived fusion indicator cell lines
We used the ViraPower Packaging Mix with Lipofectamine 2000 (Invitrogen) to transfect 293FT cells with pLenti-DSP 1Á7 and create pseudoviruses containing the DSP 1Á7 expression cassette (Lenti-DSP 1Á7 ). We next infected cell lines NP-2/CD4 (N4), CD4/CXCR4 (N4X4) and CD4/CCR5 (N4R5) with pseudoviruses containing LentiDSP 1Á7 for 2 hours. Cells were distributed in 96-well tissue culture plates at a density of 75 cells/plate (0.8 cell/well) and grown in the presence of 4 mg/ ml blasticidin. Approximately 50 candidate clones from each cell line were randomly selected and tested for FITC intensity using FACS Calibur (BD Biosciences, Franklin Lakes, NJ, USA) 48 hours after transfection of pDSP 8Á11 . FACS data were analyzed by Flow Jo version 8.7.1 (Tree Star Inc., OR, USA). Clones with the highest median FITC intensity were expanded in M10' supplemented with 4 mg/ml of blasticidin (M10 ' 4) for further assays.
Generation of pRE11-env strains Full-length HIV-1 env was prepared by PCR amplification from clinical plasma samples as described [23]. Viral RNA was extracted from 140 ml of patient's plasma by QIAamp Viral RNA Mini kit according to the manufacturer's recommendation (QIAGEN, Hilden, Germany). One-step RT-PCR using Super-Script III and High Fidelity Platinum † Taq DNA polymerase (Invitrogen) was carried out in five separate 15-ml reactions to minimize the bias created by PCR. The reaction mixture contained 2 ml of RNA template, 7.5 ml of 2) reaction buffer, 0.3 ml of 5 mM MgSO 4 , 0.3 ml of each 10 mM of forward primer (Env-1F, 25-mer, 5?-TAGAGCCCTG GAAGCATCCAGGAAG-3?) and reverse primer (Env-3Rmix, equimolar mixture of 30-mer, 5? -TGCTGTATTGCTACTTGTGATTGCTCCATA-3? and 30mer, 5? -TGCTGTATTGCTA CTTGTGATTGCTCCATG-3?), 0.6 ml of Schematic representation of pRE11, an expression vector for HIV-1 env and DSP 8 Á 11 . pRE11 encodes also mOrange with a nuclear localization signal as an indicator of transfection. (c) NP-2-derived clones stably expressing DSP 1 Á 7 (N4-DSP 1 Á 7 , N4X4-DSP 1 Á 7 and N4R5-DSP 1 Á 7 ) were selected by the high GFP expression after direct transfection of pDSP 8 Á 11 . The expression of CD4/co-receptors was reconfirmed by appropriate monoclonal antibodies and FACS analysis. SuperScript III and High Fidelity Platinum † Taq DNA polymerase, 0.25 ml of RNAse OUT and 3.75 ml of nuclease-free water with the final volume of 15 ml/reaction. The one-step RT-PCR condition was 558C for 30 minutes, 948C for 2 minutes followed by 30 cycles of 948C for 20 seconds, 558C for 30 seconds, 688C for 4 minutes, then extension at 688C for 5 minutes. The fragment by the first-round amplification extended from NL4-3 reference position of 5853Á8936. Products from five independent reactions were combined. Four microliters of the mixed first-round PCR products were used as the template for each of five independent secondround PCR reactions employing EnvB-2F-Xba (41-mer, 5?-TAGCTCTAGAACGCGTCTTAGGCATCTCCTATGGCAG GAAG-3?) and EnvB-4R-Xho (41-mer, 5? -TAGCCTCGAGACCGGT TACTTT TTGACCACTTGCCACCCAT-3?) as the forward and reverse primers, respectively. The second PCR was carried out according to the standard 50-ml PCR protocol of the Platinum † PCR SuperMix High Fidelity as described above. The fragments amplified by the second PCR extended from NL4-3 reference position of 5957Á8817. After digestion with Xba I and Xho I, about 3-kb PCR products were purified by 1.2% agarose gel and QIAquick gel extraction kit (Qiagen). The purified products were inserted into pRE11 at XbaI and XhoI sites, resulting in HIV-1env expression plasmid (pRE11envbulk) from each patient. pRE11-envbulk, representing a quasispecies of env population from each patient, was prepared by transfecting into E. coli JM109. For bulk analysis, transfected JM109 was expanded to 25 ml, followed by QIAGEN Plasmid Midi Kit (Qiagen) for DNA extraction.

Cell-fusion assay
On the day before transfection, 500 ml aliquots of 293FT cells in DMEM supplemented with 10% FBS (D10) were seeded in 24-well tissue culture plates at a density of 2.8 )10 5 cells/ well and incubated overnight to 70Á80% confluency. The cells were then transfected with pRE11-envstrain or pRE11envbulk according to the manufacturer's protocol (Roche). On the same day, 100 ml aliquots of N4-DSP 1Á7 , N4X4-DSP 1Á7 and N4R5-DSP 1Á7 cells in MEM supplemented with 10% FBS (M10) were seeded in a 96-well tissue culture, optical bottom plate (NUNC, Thermo Fisher Scientific Inc., NY, USA) at a density of 1 )10 4 cells/well and incubated at 378C. Fortyeight hours after transfection, the medium of transfected 293FT cells was removed by aspiration and replaced with 1 ml of PBS (Sigma) at RT. Transfected 293FT cells were resuspended by gentle pipetting.
To start the cell-fusion assay, 150 ml/well of transfected cells were overlaid onto N4-DSP 1Á7 , N4X4-DSP 1Á7 and N4R5-DSP 1Á7 cells. The cells were incubated for fusion at 378C in a humidified 5% CO 2 incubator for 6 hours, and then analyzed by automatic image capture using an In Cell Analyzer 1000 (GE Healthcare). Four fields/well of image were captured through red, green and bright field channels, and fused cells were identified by the presence of two or more red nuclei surrounded by a green area (cytoplasm). Immediately after image capturing, EnduRen TM Live Cell Substrate (Promega) was added to each well, and luciferase activity was measured three times using a Glomax 96 microplate luminometer (Promega), according to the manufacturer's instructions. The mean luciferase activity, recorded as relative light unit (RLU), was the average of three measurements per well. The experiments were conducted in triplicates and repeated independently at least three times.
To test the co-receptor specificity, 2 mM/well of the appropriate inhibitor was added to the cells 90 minutes prior to the cell-fusion assay (CXCR4 inhibitor AMD3100 (Sigma) to N4X4-DSP 1Á7 cells and CCR5 inhibitor maraviroc (Sigma) to N4R5-DSP 1Á7 cells).
Using this approach, we obtained N4-and N4R5-cells expressing high levels of DSP 8Á11 , but were unable to obtain a stable N4X4 cell line expressing DSP 8Á11 (data not shown). To circumvent this problem, we decided to generate 293FT cells transiently expressing both DSP 8Á11 and the HIV-1 env protein and develop a cell-fusion assay system using those cells together with the NP-2-derived cells stably expressing DSP 1Á7 (Figure 1a). Thus, we constructed the expression vector pRE11, containing the DSP 8Á11 expression cassette and cloning sites for insertion of HIV-1 env sequences under the control of the PGK promoter (Figure 1b).
Validation of the cell-fusion assay using the env gene from laboratory HIV-1 strains We validated the DSP assay system (DSP-pheno) using pRE11 constructs engineered to contain env sequences from reference strains with known co-receptor usage. The env reference constructs, which also contained the DSP 8Á11 expression cassette, were the following: pRE11-HXB2, pRE11-LAI, and pRE11-NL4-3 (X4 strains); pRE11-BaL (R5 strain); and pRE11-SF2 (dual strain). The cell-fusion assays were performed with N4X4-DSP 1Á7 or N4R5-DSP 1Á7 cells in combination with 293FT cells transiently expressing one of the pRE11-env constructs. In all the assays, both RL and GFP activities were restored only when cells expressing the appropriate env and co-receptor combinations were co-cultured (Figure 2a and b). Co-culture of N4X4-DSP 1Á7 or N4R5-DSP 1Á7 in combination with the 293FT cells transiently expressing the pRE11-env constructs of discordant tropism served as a negative control for expression of RL activities Assay detection thresholds and sensitivity for minor populations To evaluate assay sensitivity in identifying minor variants within a single sample, we mixed pRE11-NL4-3 (X4) and pRE-BaL (R5) in varying ratios and measured RL activities and GFP signals (Figure 3a and b). Both methods of detection identified X4 viruses more readily than R5 viruses. Based on luciferase activity, the presence of approximately 0.3% X4 viruses gave values significantly higher than background (0% X4), while R5 viruses had to comprise approximately 5% of the mixture for the signal to be detectable over background (Figure 3a). Similarly, based on GFP signals, X4 viruses comprising as little as 0.1% of the mixture could be detected, while detection of R5 viruses had a minimum threshold of approximately 1% (Figure 3b).
Validation of the chemokine receptor specificity using the CXCR4 inhibitor AMD3100 and CCR5 inhibitor maraviroc 293FT cells expressing env from reference strains NL4-3 (X4) or BaL (R5) were co-cultured with N4X4-DSP 1Á7 or N4R5-DSP 1Á7 cells in the absence or presence of AMD3100 or maraviroc (Figure 4a and b). In the absence of inhibitors, RL activities of the matched co-culture were high (Figure 4a). In the presence of AMD3100, the RL activity of the co-culture of 293FT cells expressing NL4-3-derived env with N4X4-DSP 1Á7 cells was reduced by 83%. The RL activity of the co-culture of 293FT cells expressing BaL-derived env with N4X4-DSP 1Á7 cells was low in the absence of AMD3100 and was not affected significantly by its presence. The RL activity of the co-culture of 293FT cells expressing BaL-derived env with N4R5-DSP 1Á7 was reduced by 81% in the presence of maraviroc. The RL activity of the coculture of 293FT cells expressing NL4-3-derived env with N4R5-DSP 1Á7 was low regardless of the presence or absence of maraviroc. The results indicated that DSP-Pheno could be used as an assay for entry inhibitors.

Cell-fusion assay of clinical samples
To evaluate assay performance using clinical samples, we selected plasma samples from 101 treatment-naïve, HIV-1positive patients, whose infection with clade B viruses had been confirmed (data not shown). The patient population was classified into two groups based on CD4 T cell count. The low CD4 group consisted of 57 patients with CD4 T cell counts B350 cells/ml; median 228 (range 2Á350) cells/ml, and median viral load was 4.77 (range 2.97Á6.62) log 10 copies/ml (Figure 5a and b). The high CD4 group consisted of 44 patients with CD4 cell counts 350 cells/ml; median 442 (range 351Á843) cells/ml, and median viral load was 4.04 (range 1.60Á5.41) log 10 copies/ml. The viral load differences between the two groups were statistically significant by the MannÁWhitney U test (p B0.001). Aliquots of viral envelope DNA from each plasma sample were used to construct pRE11-envbulk for transfection into 293FT cells. The plasma viral load necessary for the assay was roughly 3.00 log 10 copies/ml for subtype B viruses, although we could amplify the env gene in a patient with 1.60 log 10 copies/ml.
We used the laboratory strain, BaL as the R5 control and NL4-3 as the X4 control to define the cut-off values. We examined BaL on N4X4-DSP 1Á7 cells and NL4-3 on N4R5-DSP 1Á7 cells. We defined the cut-off value tentatively as the mean value'2SD based on 3 determinations in 12 independent experiments for each combination of negative control and indicator cell (red dashed line in Figure 5c). As expected, both combinations showed stably low RL activities, with cutoff values of 876 for N4X4-DSP 1Á7 cells and 397 RLU for N4R5-DSP 1Á7 cells.
Samples from all patients gave positive RL signals on R5 indicator cells (N4R5-DSP 1Á7 ) in the fusion assay, which suggested that the bulk of virus in each patient was able to use CCR5 as the co-receptor (Figure 5c, lanes 5 and 6). Median RLU value of the low CD4 group was significantly higher than that of the high CD4 group on R5 indicator cells (pB0.0001). Median RLU value of the low CD4 group was also higher significantly on X4 indicator cells (p00.0097) and 26/57 (46%) of low CD4 cases versus 15/44 (34%) of high CD4 cases gave positive RL signals (Figure 5c, lanes 1 and 2). Higher fusion activities on both indicator cells are compatible with higher viral loads in patients with lower CD4 T cell counts and may suggest more dual or X4 tropic (dual/X4) viruses in this group of patients.

Discussion
We developed a quick, safe and sensitive HIV-1 PTA utilizing double split proteins (DSP-Pheno) and validated the specificity of the assay using laboratory strains with known co-receptor usage. We recognize several limitations of this preliminary study, but the results nevertheless are promising. We assayed bulk envelope genes amplified from plasma from HIV-1infected patients, rather than cloned envelope genes, and our sample only included subtype B HIV-1. Future studies are necessary to demonstrate the usefulness of the DSP-Pheno.
One caveat of the DSP-Pheno assay is that it is a cell-fusion system, and cellÁcell fusion may differ in significant details from virusÁcell fusion. For example, recent studies have shown that HIV-1 virions carry fewer surface glycoproteins than previously assumed [24]. The DSP-Pheno assay uses neuroglyoma cell-derived NP-2 cell lines with overexpressed CD4 and co-receptors. Although these NP-2-derived cell lines have been characterized extensively [16,17], some unknown cell surface molecules may be involved in the fusion process. The DSP-Pheno assay is a gag-free system and requires only the assembly of reporter proteins pre-formed in the fusion partner, but infection by a retrovirus requires that the entire gag particle pass through the fusion pore. Careful comparison between DSP-Pheno and in-house pseudoviral assay or GTA using clonal clinical isolates is under way.
GFP portion is necessary as a module of DSP to compensate weak self-association of split RL [15]. Although RL would be more suitable for quantitative assay, GFP may prove single clear positive fusion in the sample with very low RL readout. This feature of DSP-Pheno incorporating two different assays may be useful for certain scientific purposes.
Although several issues remain to be clarified, DSP-Pheno has multiple advantages over the conventional pseudoviral PTA: (i) the turnaround time for DSP-Pheno is short, with results available in as few as 5 days, starting from patients' plasma; (ii) DSP-Pheno is a virus-free assay that does not require a special biosafety facility, making it particularly appealing for in-house use; and (iii) the RL assay in DSP-Pheno has high sensitivity and specificity and compares favourably with the best pseudoviral PTA published in the detection of minor X4 populations using laboratory strains. Trofile TM (Monogram Biosciences Inc., CA, USA) is currently the only commercially available PTA approved for clinical use, and the latest version, ''Enhanced Trofile TM ,'' detects X4 minor populations present in concentrations as low as 0.3% [25]. A pseudoviral PTA described by Soda and colleagues had 1% detection threshold for X4 viruses [16]. Although the RL assay in DSP-Pheno could detect X4 laboratory strains present in concentrations as low as 0.3%, further studies are needed to apply the assay for the clinical use. DSP-Pheno may also be useful for the comparison of with GTA to improve the algorithm for the co-receptor usage of non-B subtypes.