Cell–cell and virus–cell fusion assay–based analyses of alanine insertion mutants in the distal α9 portion of the JRFL gp41 subunit from HIV-1

Membrane fusion is the first essential step in HIV-1 replication. The gp41 subunit of HIV-1 envelope protein (Env), a class I fusion protein, achieves membrane fusion by forming a structure called a six-helix bundle composed of N- and C-terminal heptad repeats. We have recently shown that the distal portion of the α9 helix in the C-terminal heptad repeat of X4-tropic HXB2 Env plays a critical role in the late-stage membrane fusion and viral infection. Here, we used R5-tropic JRFL Env and constructed six alanine insertion mutants, 641+A to 646+A, in the further distal portion of α9 where several glutamine residues are conserved (the number corresponds to the position of the inserted alanine in JRFL Env). 644+A showed the most severe defect in syncytia formation. Decreased fusion pore formation activity, revealed by a dual split protein assay, was observed in all mutants except 641+A. Sequence analysis and substitution of inserted 644A with Gln revealed that the glutamine residue at position 644 that forms complex hydrogen-bond networks with other polar residues on the surface of the six-helix bundle is critical for cell–cell fusion. We also developed a split NanoLuc® (Nluc) reporter-based assay specific to the virus–cell membrane fusion step to analyze several of the mutants. Interestingly syncytia-competent mutants failed to display Nluc activities. In addition to defective fusion activity, a reduction of Env incorporation into virions may further contribute to differences in cell–cell and virus–cell fusions.

Membrane fusion is the first essential step in HIV-1 replication. The gp41 subunit of HIV-1 envelope protein (Env), a class I fusion protein, achieves membrane fusion by forming a structure called a six-helix bundle composed of N-and C-terminal heptad repeats. We have recently shown that the distal portion of the ␣9 helix in the C-terminal heptad repeat of X4-tropic HXB2 Env plays a critical role in the late-stage membrane fusion and viral infection. Here, we used R5-tropic JRFL Env and constructed six alanine insertion mutants, 641؉A to 646؉A, in the further distal portion of ␣9 where several glutamine residues are conserved (the number corresponds to the position of the inserted alanine in JRFL Env). 644؉A showed the most severe defect in syncytia formation. Decreased fusion pore formation activity, revealed by a dual split protein assay, was observed in all mutants except 641؉A. Sequence analysis and substitution of inserted 644A with Gln revealed that the glutamine residue at position 644 that forms complex hydrogen-bond networks with other polar residues on the surface of the six-helix bundle is critical for cell-cell fusion. We also developed a split NanoLuc (Nluc) reporter-based assay specific to the virus-cell membrane fusion step to analyze several of the mutants. Interestingly syncytia-competent mutants failed to display Nluc activities. In addition to defective fusion activity, a reduction of Env incorpo-ration into virions may further contribute to differences in cellcell and virus-cell fusions.
HIV-1 is a member of the enveloped viruses, and it initiates its life cycle with membrane fusion between the virus and cellular membranes (1,2). This essential step begins with the interaction between the gp120 subunit of the viral envelope protein (Env) and the cellular receptor/co-receptor complex (CD4 and chemokine receptors). The step of the actual membrane fusion is mediated by the gp41 subunit of Env, a member of the class I fusion proteins.
The gp41 subunit undergoes drastic conformational changes during membrane fusion to form a six-helix bundle (6HB) 3 structure (3) (Fig. 1A). The 6HB is generated by the anti-parallel packing of two helices of gp41 termed as N-and C-terminal heptad repeats (NHRs and CHRs, respectively) (4,5). The core of 6HB is composed of a parallel coiled coil of three NHRs, and CHRs bind to the grooves generated on the trimeric NHR bundle (Fig. 1B). The formation of 6HB is tightly linked to the fusion pore formation, its enlargement, and eventual merging of the virus and cell membrane (6,7). Because the inhibition of 6HB formation results in the abortion of membrane fusion (8,9), extensive studies of 6HB such as alanine-scanning mutagenesis have been conducted in the past (10 -13); however, the exact physical location and the mechanism of the potential association of 6HB with the fusing lipid bilayers are unknown, and the exact mechanism of pore enlargement during infection remains elusive.
In a previous study, we reported alanine insertion mutants in the HXB2 gp41 subunit (14). We found that the mutants within ␣9 following its exit from the bottom of the four-helix collar formed by ␣6, ␣7, ␣8, and ␣9 (3) did not affect Env biosynthesis but showed impairment in membrane fusion. In a cell-cell fusion assay, a certain insertion mutant in the proximal region of ␣9 was able to form fusion pores as measured by the split reporter protein called dual split protein (DSP) (15,16); however, it failed to support viral infection in a conventional viral infection assay relying on the expression of the packaged reporter gene. Viral infectivity was recovered when insertion sites were shifted toward the distal portion of ␣9.
Our data are consistent with the preceding studies indicating that 6HB formation may start from the proximal portion of CHR and that fusion pores made before the completion of 6HB formation may be stabilized and enlarged along with the progress of 6HB formation toward the distal portion of CHR (6, 7). The conventional virus-cell fusion assay used in our previous study has a technical caveat: the reporter expression requires multiple steps after the actual membrane fusion step such as reverse transcription, integration, transcription, and translation of the reporter gene.
In this study, to measure the virus-cell fusion more specifically independent of these downstream steps, we developed a new split NanoLuc luciferase (Nluc)-based (17,18) virus-cell fusion assay in which a small peptide called HiBIT derived from Nluc was targeted into virions via fusion with the viral protein, Vpr (19,20). These engineered virus-like particles (VLPs) were used to infect target cells expressing the rest of Nluc called LgBiT. We named this new assay R-BiT (Vpr-HiBiT) assay. R-BiT assay can be completed in an hour after infection.
Here, we used JRFL Env and extended our alanine insertion mutagenesis of gp41 to further downstream of ␣9 of CHR where several glutamine residues are conserved between HXB2 and JRFL. We found that a conserved glutamine residue at position 644 (based on JRFL Env numbering) is critical for cell-cell membrane fusion especially in syncytia formation. With our new R-BiT assay, we found that the mutants that were relatively efficient in cell-cell fusion activities showed severe defects in virus-cell membrane fusion and that the reduction of Env incorporation into virions in addition to the defective fusion activity may contribute to the defective virus-cell fusion of the mutants.

Effect of alanine insertion in the distal portion of ␣9 in JRFL CHR on cell-cell fusion activity
Extending our previous study of HXB2 Env, we generated alanine insertion mutants in the distal portion of ␣9 using R5-tropic JRFL Env. The primary structures of HXB2 and JRFL gp41 including NHR and CHR are shown in Fig. 2A. Similar to an authentic homotrimeric coiled coil (21), the positions a and d of gp41 NHR are often occupied by hydrophobic residues such as Ile and Leu, but the presence of polar residues such as Gln is also noted. Intriguingly, there were Gln triplets around residue 541 of JRFL gp41 and residue 550 of HXB2 Env (note, the HXB2 number can be obtained by adding 9 to that of JRFL Env) ( Fig. 2A).
Although it does not form a homotrimeric coiled coil like NHR, CHR is also given an arbitrary (abcdefg) n representation; the CHR residues in positions a and d are expected to interact with the e and g residues of NHR (Fig. 1B, right panel) and contain some hydrophobic residues. The frequency of hydrophobic residues is, however, less in the latter half of ␣9, in which JRFL and HXB2 have identical sequences ( Fig. 2A). In contrast to this conservation, the proximal half of ␣9 of JRFL and HXB2 shows several differences, although conservative in its nature.
We chose the region at residues 641-646 of ␣9 for our alanine insertion mutagenesis (indicated by the arrowheads in Fig.  2A) based on two reasons. First, this region contains multiple polar Gln residues and forms complex hydrogen-bond networks in 6HB by interacting with the NHR region, which is also rich in Gln residues (4, 5) (Fig. 1B, left and middle panels). Second, as shown in our previous study (14), mutations in this region of ␣9 are less likely to affect the synthesis and processing of gp160 to compromise functional analysis because it is far from the cleavage site between gp120 and gp41 in the native For clarity, only one gp120/ gp41 monomer is shown. The NHR and CHR portions of gp41 are shown in blue and yellow ribbons, respectively. In a prefusion structure, the gp120 subunit is shown in a surface-filled model. The postfusion structure only depicts the gp41 subunit. The portion of gp41, which undergoes a dramatic loop-tohelix transition, is indicated in green. The positions of the glutamine residues (Gln-552 in NHR and Gln-653 in HXB2 CHR, respectively) are shown in red. The numbering of the residues is based on HXB2 Env (corresponding number in JRFL can be obtained by subtracting 9 from the numbering of HXB2; therefore, Gln in JRFL Env will correspond to Gln-543 and Gln-644, respectively). The image was created by Chimera (University of California, San Francisco) and was based on Protein Data Bank codes 5FUU and 1ENV. B, the structure and organization of a 6HB. Side view and top view of a trimeric 6HB (Protein Data Bank codes 1ENV) are shown in the left and middle panels, respectively. The color coordination is the same as that in A. The conserved glutamine residues are shown as in A. The helical diagram of the portion of 6HB containing three NHRs (NHR1, 2, and 3) and one CHR (CHR1) is shown (right) to show the interacting residues in 6HB. The residues in a heptad repeat are assigned with a conventional (abcdefg) designation.

Alanine insertion mutants in JRFL gp41 ␣9
structure (3). The mutants were named based on the position of the inserted alanine residue in the JRFL Env. For example, for the 641ϩA mutant, the alanine residue is located at position 641.
We used a codon-optimized synthetic JRFL Env gene (22) to express envelope proteins. We first examined the syncytium formation activities of the mutants by transfecting expression vectors into CD4 ϩ CCR5 ϩ cells. Mutant 644ϩA showed a very severe defect in this assay (Fig. 3, A and B). Although attenuated in some mutants, syncytia formation activity was retained in other mutants. Although we reasoned that mutations in this region would not perturb the protein biosynthesis and processing of Env as discussed above, we next analyzed the protein profile of 644ϩA and other mutants. The immunoblotting result with an anti-gp120 antibody failed to reveal severe defects in expression or processing in 644ϩA (Fig. 3C), suggesting that 644ϩA has a defect in the membrane fusion step itself.
To gain further insight into the potential step(s) accounting for the defect in the cell-cell membrane fusion of 644ϩA, we used the DSP assay to test whether 644ϩA has a defect in fusion pore formation. The DSP assay detects the communication (pore formation) between effector cells and target cells by measuring the recovery of the Renilla luciferase activities of split DSPs (23). The result is shown in Fig. 3D. 644ϩA showed the lowest DSP activity corresponding to ϳ25% of that of the WT. This suggests that mutant 644ϩA has defects in pore formation, as well as the subsequent steps such as pore enlargement, that eventually lead to syncytia formation. The other mutants showed decreased DSP activities, but 641ϩA showed a similar DSP activity to that of the WT. Overall, there is a good correlation between syncytia formation and DSP assays.

The glutamine residue at position 644 plays a critical role in membrane fusion
An alanine insertion can have more detrimental effects than a simple alanine substitution by inducing a shift of amino acid residues after the insertion point. However, previous studies suggested that such a mutation can be accommodated even in an ␣-helix (24,25). Because mutants 641ϩA, 642ϩA, and 643ϩA showed a relatively mild defect in syncytia formation, it can be suggested that the ␣9 region after residue 641 is rather tolerant to shifting by insertion. Consistent with this idea,

Alanine insertion mutants in JRFL gp41 ␣9
mutants 645ϩA and 646ϩA retained cell-cell fusion activities. This led to the question why only mutant 644ϩA showed such a severe defect in syncytia formation.
When we aligned the mutant amino acid sequences, it became clear that only mutant 644ϩA differed from the others by having an alanine residue at position 644 (Fig. 2B). This alteration of Gln at position 644 did not occur in mutants 641ϩA, 642ϩA, and 643ϩA because of the presence of the Gln residue at position 643; that is, Gln-643 was shifted to position 644 by alanine insertion in these mutants. Similarly, mutants A and B, the syncytia formation assay of the mutants. The expression vector for each mutant was transfected into CD4ϩCCR5ϩ cells, and the syncytia formation was evaluated by microscopy. A representative field for each mutant is shown (upper panel, phase contrast; lower panel, Hoechst stain; bar ϭ 100 m) in A. The relative degree of syncytia formation was shown by dividing the number of nuclei in the syncytia by the total number of nuclei in the observed fields. The WT value was set as 1 (B). C, the immunoblotting of the mutants. The expression and processing of gp160 were examined by immunoblotting using an anti-gp120 antibody (upper panel). The ratios of gp120/gp160 (processing of Env) or gp160 or gp120/GADPH (expression level of Env) were determined by densitometry (lower panel). D, the DSP assay to detect fusion pore formation between effector and target cells. The raw DSP values and normalized values by the surface level of Env are shown. **, statistically significant (p Ͻ 0.01). Error bars represent S.D. The representative results of two independent experiments are shown.

Alanine insertion mutants in JRFL gp41 ␣9
645ϩA and 646ϩA did not show alteration of Gln-644. This suggests that the loss of the Gln residue at position 644 may be responsible for the observed severe defect in the membrane fusion of 644ϩA. To test this hypothesis, we mutated the inserted alanine of 644ϩA to Gln; this has been shown as 644ϩQ in Fig. 2B because it is equivalent to the insertion of Gln at position 644. We analyzed the phenotype of 644ϩQ. As shown in Fig. 3, improved syncytia formation and DSP activity were observed, suggesting that the presence of a Gln residue at position 644 is critical for Env function in mediating cell-cell membrane fusion.

Analysis of alanine insertion mutants in virus-cell fusion assay: development of a virus-cell fusion assay (R-BiT assay) by employing the split Nluc system
Next, we intended to examine these mutants in a virus-cell fusion assay because we and others (10,14,26) have observed some discrepancies between the cell-cell fusion assay and virus-cell fusion assay. For this purpose, we tried to develop a more fusion-specific virus-cell fusion assay using split Nluc as a reporter. Split Nluc recovers its activity via self-association of the Nluc-derived small peptide (HiBiT) and the remaining domain of Nluc (LgBiT). Our approach is similar to the BlaM assay (27). HiBiT was targeted into HIV-1 virions via the virion-associated viral protein, Vpr (19,20). Accordingly, we named this assay R-BiT (Vpr-HiBiT) assay. Nluc is more sensitive than Renilla luciferase, and its signal detection is simple and does not require image analysis like the BlaM assay. In our pseudotype-based assay, we also added the packageable reporter gene (firefly luciferase) used in our previous study (14) for comparison (Fig. S1). The 293 cells stably expressing CD4, CCR5, and LgBiT were used as target cells (see "Experimental procedures").
In our initial attempt of the JRFL Env-based assay, we encountered a relatively high Nluc background signals in the presence of a CCR5 antagonist, maraviroc (MVC) (Fig. 4A), whereas the Fluc reporter signal was efficiently blocked (Fig.  4C). The addition of the membrane-impermeable DrkBiT peptide (about DrkBiT; see "Experimental procedures"), a competitive peptide against HiBiT, that binds to LgBiT without recovering the Nluc activity reduced these background signals (Fig.  4A), and the signal to background ratio was dramatically improved (Fig. 4B). DrkBiT did not affect Fluc reporter signals (Fig. 4C). These data suggested that reassociation of the leaked

Alanine insertion mutants in JRFL gp41 ␣9
Vpr-HiBiT and/or LgBiT was responsible for the observed residual background.

Modification of LgBiT to reduce the background signal
To reduce the background, we tried to minimize the spontaneous leakage of LgBiT from the target cells because the sedimentation of virions by ultracentrifugation did not eliminate the background completely (data not shown). We engineered LgBiT by fusing ␤-gal (BetaGal-LgBiT), a phosphatidylinositolbinding pleckstrin homology (PH) domain (PH-Halo-LgBiT) (28), and a farnesylation motif (HCAAX) (29) (Halo-LgBiT-HCAAX) to LgBiT to reduce the leakage of LgBiT. The results shown in Fig. 4D showed that PH-Halo-LgBiT, betaGal-LgBiT, and Halo-LgBiT-HCAAX decreased the MVC-insensitive background signals compared with the unmodified Halo-LgBit. Comparison of Nluc activities between DMSO and MVC showed that the addition of the PH domain yielded the best reduction of the background signal (Fig. 4E). The modification of LgBiT did not affect Fluc reporter signals (Fig. 4F). The degree of leakage estimated by the ratio of the extracellular and intracellular amounts of LgBiT was high for ␤-gal and HCAAX. This ratio was better for PH than for ␤-gal and HCAAX (Fig.  4G). Accordingly, we used the target cells expressing PH-Halo-LgBiT in the subsequent experiments.

Specificity of the R-BiT assay for membrane fusion
The R-BiT assay was validated for its specificity for membrane fusion by neutralizing antibodies (2G12 and PGT151) (30,31) or a reverse transcriptase inhibitor (azidethymidine (AZT)). The addition of neutralizing antibodies significantly reduced the Nluc signals (Fig. 5A). As expected, AZT did not affect Nluc signals (Fig. 5A). However, after 24 h of infection with no detectable residual signals of Nluc (Fig. 5B), the Fluc activities expressed from the integrated provirus DNA were inhibited by all inhibitors (MVC, 2G12, PGT151, and AZT) (Fig. 5B). These results confirmed that our R-BiT system could faithfully monitor the membrane fusion step of viral infection. As a by-product of the incorporation of Vpr-HiBiT into virions, we found that the amount of HIV-1 virions can be estimated by titrating the virion-associated Vpr-HiBiT. Nluc signals have a wider dynamic range than the p24 assay, and the determination can be completed in 1 h (Fig. S2).

Application of the R-BiT assay to the analysis of gp41 mutants
We tested mutants 641ϩA, 644ϩA, and 644ϩQ with our new R-BiT assay. We added the dose-adjusted VLPs to the target cells (293CD4R5-PH-Halo-LgBiT cells), and the Nluc activity was monitored (Fig. 6A). The WT showed a time-dependent large increase in Nluc activity, but all the tested gp41 mutants showed almost no Nluc activity. This suggests that these mutants have a severe defect in the virus-cell membrane fusion step. The loss of infectivity for mutant 644ϩA was expected because it failed to develop syncytia in our cell-cell fusion assays. The results of 641ϩA and 644ϩQ were, however, rather unexpected especially because the syncytia formation and DSP activities of the 641ϩA and 644ϩQ were almost similar to those of the WT (Fig. 3). This consistent result was obtained with the conventional virus infection assay relying on the expression of the packaged reporter gene (Fig. 6B). Our data revealed that there are prominent differences in fusion phenotypes between cell-cell and virus-cell membrane fusions. All the other mutants also showed no Fluc activities (Fig. 6B).
To determine the reason for the observed differences in the cell-cell and virus-cell fusion assays, we analyzed the protein profiles of the virions (Fig. 6C). When the gp41/p24 ratio was calculated, the mutants exhibited ϳ30% reduction of the incorporation of processed Env compared with the WT. Furthermore, a decreased gp120/p24 ratio was observed for all the mutants, indicating a gp120 shedding for the mutants (Fig. 6C). This gp120 shedding may further contribute to the observed differences in cell-cell and virus-cell fusions. In addition to these differences, we also noted increased incorporation of gp160 into mutant virions when the bands recognized by anti-gp41 were used to calculate the gp160/p24 ratio. This tendency was less prominent when the bands recognized by anti-gp120 were used. We cannot rule out the possibility that this could arise from the differences in the recognition of the antigens by the different antibodies. The presence of gp160 on the virions may interfere with the receptor recognition by the processed gp120. In addition to the defective fusion activity, these changes in viral protein profiles of the mutants may contribute to their severe defects in virus-cell fusion.

Discussion
In this study, we performed alanine insertion mutagenesis in the distal region of ␣9 in JRFL gp41. Most of our alanine inser-

Alanine insertion mutants in JRFL gp41 ␣9
tion mutants retained syncytia formation activity except for 644ϩA. The 644ϩA mutant showed attenuated activities in fusion pore formation, as revealed by the DSP assay. Therefore, mutant 644ϩA seems to have more severe defects in membrane fusion from pore formation to the progress of membrane fusion after fusion pore formation.
Our data suggest that the presence of a Gln residue at position 644 is critical for cell-cell membrane fusion. The results of a previous alanine substitution (not an insertion) study of the Gln-644 residue in JRFL (10) showed ϳ50% reduction in cellcell and virus-cell fusion and support the importance of Gln-644 in membrane fusion. We observed more severe defects in both assays, and this is probably due to the more destructive nature of alanine insertion than alanine substitution.
Because an alanine insertion causes deregistration of amino acid residues downstream of the insertion point, it is expected that an insertion will have a stronger negative effect in the 6HB structure where multihelix interactions are involved. In contrast to this expectation, although attenuated, most of the mutants retained syncytia formation activity except for 644ϩA. We speculate that the tolerance to insertion in this region could be due to the presence of multiple polar residues, such as Gln, near the insertion sites. Polar residues such as Gln are known to destabilize the coiled coil (32), and this may provide some flexibility or plasticity in these ␣-helix regions. An insertion of Ala or Gln could be accommodated by local underwinding of the helix with associated changes in superhelix to accommodate underwinding of the involved helices (33). Even with this underwinding effect, there was a striking difference between 644ϩA and 644ϩQ in cell-cell membrane fusion.
Johnson et al. (34) have shown that the contribution of the latter one-third of 6HB, in which residue 644 resides for the stability of 6HB is modest, and this agrees with the data of Suntoke and Chan (35), who did not observe any severe impact of Q653A mutation (corresponds to Gln-644 in JRFL) on the stability of 6HB in HXB2 Env. Although we cannot tell whether the result of HXB2 is immediately applicable to JRFL, and the stability of 6HB in our 644ϩA or Q needs to be determined in the future, our data suggest that the chemical nature of the amino acid residues at position 644, and its relationship with the polar residues in the exposed surface of 6HB may play some roles during membrane fusion in addition to the 6HB stability.
Regarding the residue at position 644 in JRFL 6HB, Gln-644 (Gln-653 in HXB2) located at position e of the helix of CHR interacts with Asn-544 (Asn-553 in HXB2) at position b of NHR of the same gp41 molecule (Fig. 1B, right panel) (4,5). In this layer of 6HB, Gln-644 is involved in complex hydrogen bonding networks formed among several Gln and Asn residues of both NHRs and CHRs (5,35). Interestingly, there are five or six Gln residues clustering on the surface of 6HB. The substitution of Gln-644 with Ala affects the hydrogen bonding and clustering of Gln residues. Even though the exact mechanism remains to

Alanine insertion mutants in JRFL gp41 ␣9
be elucidated, the involvement of Gln-rich proteins in protein aggregation (36,37) may suggest that clustering of multiple Gln residues on the surface of 6HB may affect the manner of interaction of 6HB with other viral or cellular (membrane) protein(s) during membrane fusion. This may affect the fate of membrane fusion. Further mutagenesis of Gln-644 or neighboring polar residues with other amino acid residues may provide an insight into the underlying molecular mechanism.
We have developed a new Nluc-based virus-cell fusion assay and observed a discrepancy in membrane fusion efficiency between the cell-cell and virus-cell fusion assays; cell-cell fusion was more tolerant to mutations. In our previous study of HXB2 Env, we observed a similar discrepancy and raised the possibility that the insufficient fusion pore size generated by the mutant is responsible (14). This insufficient size of fusion pores may be able to explain the severe defect of mutant 644ϩA, which failed to form syncytia. However, our reporter for the R-BiT assay was a small virion-associated protein, Vpr-HiBiT, which could be more easily released from virions upon membrane fusion and subsequent disassembly of viral cores. Because we failed to observe virus-cell fusion with the R-BiT assay for the mutant 644ϩQ, which recovered the cell-cell fusion activity, the insufficient size of fusion pores alone may not be able to account for the observed differences in the other mutants.
In addition to the defective fusion activities of mutant Env, we cannot rule out the possibility that the difference in the number of available functional Envs on the cell or virus surface in a given time period may affect the outcome. Although a virus has no additional supply of new Envs during membrane fusion, newly synthesized Envs can be recruited in cell-cell fusion during the period of membrane fusion because protein synthesis is continuous during cell-cell fusion. We observed a reduction in the incorporation of the processed Env into virions accompanied with gp120 shedding. Furthermore, as described for JRFL Env previously (10), indiscriminate incorporation of unprocessed gp160 into virions was noted. These may contribute to the observed inefficient virus-cell fusion. It seems that mutations in the distal portion of ␣9 can still affect the incorporation of Env into virions in JRFL Env.
Another possibility is the involvement of cellular proteins in the process of membrane fusion. The contribution of adhesion molecules such as LFA-1 and intercellular adhesion molecules to HIV-1 Env-mediated cell-cell fusion has been described (38,39). Although these proteins are absent in our 293-derived effector and target cells, the involvement of some other cellular proteins may facilitate cell-cell fusion. The alterations of 6HB surface by our mutation may affect the cell-cell fusion by altering the interactions between 6HB and cellular proteins. Lastly, our new Nluc-based virus-cell fusion assay, R-BiT, is easier to perform and is faster for monitoring virus-cell fusion in a fusion-specific manner independent of downstream replication cycles such as reverse transcription/integration and thus will be a useful tool for investigating these potential differences between cell-cell and virus-cell fusions.

Cell lines, chemicals, antibodies, and transfection
293-CD4-CCR5 (293CD4R5), LgBiT-expressing 293CD4R5, and 293T-HiBiT cell lines, generated and used in this study (see below), were maintained in Dulbecco's modified Eagle's medium (Wako, Japan) containing 10% fetal bovine serum. For the establishment of stable cell lines expressing CCR5, HiBiT, and each LgBiT protein, recombinant pseudotyped retroviruses were produced using plat-E cells (kindly provided by Prof. Toshio Kitamura, Institute of Medical Science, University of Tokyo, Tokyo, Japan) (40) with a vesicular stomatitis virus G-expressing plasmid (pCMV-VSV-G). The cells infected with pseudotyped viruses were selected with 1 g/ml puromycin and/or 10 g/ml blastcidin for at least 1 week. These bulk selected cells were used to perform fusion assays. Maraviroc

Construction of expression vectors
A synthetic DNA corresponding to the vpr gene of HIV-1 (42) was fused with HiBiT sequence (5Ј-GTG AGC GGC TGG CGG CTG TTC AAG AAG ATT AGC TAA-3Ј) at the C terminus by PCR and cloned into the pCMV vector. For the establishment of stable cell lines expressing CCR5, HiBiT, and each LgBiT protein, CCR5 was cloned into a pMXs-IRES-blastcidin retroviral vector, and Halo-HiBiT, LgBiT with ␤-gal at the N terminus, and Halo-LgBiT with a PH domain at the N terminus were cloned into the pMXs-puromycin vector, respectively. For the establishment of Halo-LgBiT with H-Ras CAAX (HCAAX) at the C terminus, the following sequences corresponding to the CAAX domain were fused at the C terminus by PCR using KOD-Plus-Neo (Takara Bio, Otsu, Japan) followed by cloning into the pMXs-Puromycin vector: HCAAX, 5Ј-GAG TCC GGC CCC GGC TGC ATG TCC TGC AAG TGC GTG CTG TCC TAA-3Ј; KCAAX, 5Ј-AAG AAG AAG AAG AAG AAG TCC AAG ACC AAG TGC GTG ATC ATG TAA-3Ј; and QCAAX, 5Ј-CAG CAG CAG CAG CAG CAG TCC AAG ACC AAG TGC GTG ATC ATG TAA-3Ј. The gp41 CHR mutants were generated using an Env expression vector containing a previously described codon-optimized JRFL sequence for mammalian cells (22). Complementary oligonucleotide pairs containing an insertion codon for the alanine residue, GCA, were used for mutagenesis using the QuikChange site-directed mutagenesis kit (Stratagene).

Virus preparation
For the preparation of Vpr-HiBiT-containing lentiviruses, a plasmid mixture (2.5 g of pCMV-JRFLenv (JRFL-Env expression vector), 2.5 g of psPAX2 (Gag-pol, Rev, and Tat expression vector), 5 g of pLenti-CMV-Luc2 (packaging reporter to express f-luc), and 1 g of pCMV-Vpr-HiBiT (Vpr-HiBiT expression vector)) was transfected into 293T cells by calcium phosphate transfection. At 48 h after transfection, the supernatant was harvested and purified by ultra-centrifugation (1 ϫ 10 5 g for 35 min using a TLA100.3 rotor). The pellet was then resuspended using Hanks' buffer. Viral lysates were subjected to SDS-PAGE and Western blotting (see below).

R-BiT assay in a 384-well format
A day before the R-BiT assay, the target cells expressing CD4, CCR5, and LgBiT were seeded in 12-well cell culture plates (2 ϫ 10 5 cells/ml). At 2 h before the assay, the cells were treated with EnduRen (6 M EnduRen; Promega), a substrate for NLuc, to activate EnduRen. MVC (1 l) in DMSO, AZT, and neutralizing antibodies in PBS(Ϫ) were added to 384-well plates (Greiner Bioscience, Frickenhausen, Germany). Target cells were harvested by pipetting, spun down, and resuspended in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum and 6 M EnduRen (8 ϫ 10 5 cells/ml). Next, 50 l of single cell suspension and Vpr-HiBiT containing a dose-adjusted pseudotyped virus (70 ng of p24) with 10 g/ml Polybrene and 6 M EnduRen were mixed in the wells. After incubation at 37°C for 15, 30, and 60 min, light signals derived from HiBiT/LgBiT association were measured using CentroXS3 (Berthold Technology, Bad Wildbad, Germany). After incubation at 37°C for 24 h, the supernatant was removed, and Steady-Glo luciferase assay reagent (Promega) was added to dissolve cells to measure Fluc activities.

Cell-cell fusion assay
Syncytium formation assay-For the syncytia formation assay, expression vectors for Env were transfected into 293CD4R5 cells using TurboFect (Thermo Fisher Scientific) and then incubated at 37°C for 12-18 h. To visualize the nuclei of the cells, Hoechst 33,342 (5 mg/ml; Invitrogen) was applied at 37°C for 15 min. After labeling, images were captured using a Fluorescent Inverted microscope (Olympus IX71). The number of nuclei in syncytia was divided by the total number of nuclei in the field to estimate the degree of syncytia formation. Two randomly chosen fields containing ϳ500 nuclei were examined.
Dual split protein assay-A cell-cell fusion assay using a split reporter protein termed DSP was performed as described previously (14). Briefly, the 293FT/DSP 1-7 cells seeded in a 12-well plate (BD Falcon, San Jose, CA) at a density of 3.0 ϫ 10 5 cells/ well were transfected with Env expression vectors of interest using FuGENE HD (FuGENE HD (l):DNA (g):Opt MEM (l) ϭ 2.5:1:50). 24 h after transfection, the effector cells were mixed with the target cells in a 96-well View Plate (PerkinElmer) in triplicate. The target cells (293CD4R5/DSP 8 -11 ) were plated in a 6-cm dish (BD Falcon) at a density of 1.5 ϫ 10 6 /dish 2 days prior to the assay and were preincubated for 1 h with fresh medium containing 60 M EnduRen (Promega). Mem-brane fusion was initiated by spinning down the mixed cells in a 96-well plate (130 ϫ g for min in LX-131; TOMY SEIKO CO., Ltd.) at room temperature. The light signals derived from the reassociated DSP were measured using a GloMax-Multi Plus detection system (Promega). A portion of the effector cells used for the DSP assay was used for CELISA to determine the surface level of Env as described previously (14). The cells were spun down in a collagen-coated 96-well plate (Corning) (130 ϫ g for 3 min at room temperature). The cells were washed with PBS and fixed with 4% paraformaldehyde in PBS for 5 min. After washing with PBS, endogenous peroxidases were inactivated with 3% hydrogen peroxide in PBS for 10 min. After blocking with a blocking reagent (5% ECL Prime blocking reagent (GE Healthcare) in PBS) for 30 min, the cells were incubated with a saturating amount of 2G12 (13.1 g/ml, 1:1000 dilution in 0.5% ECL Prime blocking reagent) for 1 h at room temperature and then incubated with anti-human IgG-HRP (Santa Cruz Biotechnology) (1:5000 dilution at room temperature for 1 h) after washing with PBS. Finally, the luminescence signals were measured using ECL Prime Western blotting Detection Reagent (GE Healthcare) in a GloMax 96 microplate luminometer (Promega).

Western blotting
Whole-cell lysates of 293FT cells transfected with the respective Env expression vectors were separated by SDS-PAGE and transferred to polyvinylidene difluoride membranes (Immobilon P; Millipore, Milford, MA). The membranes were then incubated with a sheep anti-HIV gp120 antibody (1:1000; Aalto Bio Reagents LTD, Dublin, Ireland) or a mouse anti-glyceraldehyde-3-phosphate dehydrogenase mAb (1:1000; Beyotime Biotechnology, Shanghai, China). Immunoreactive proteins were visualized with anti-goat or anti-mouse IgG conjugated to HRP (GE Healthcare), followed by processing with an ECL detection system (GE Healthcare).
The blots for viral lysates were probed with goat anti-gp120 polyclonal antibodies (1:2000; Abcam, Cambridge, MA), goat anti-gp41 polyclonal antibodies (1:2000; Bio-Rad), and an anti-p24 mAb (kind gift from Dr. Y Tanaka) were used in addition to anti-gp120. Donkey anti-goat IgG-HRP (Santa Cruz Biotechnology) or goat anti-mouse IgG-HRP (Santa Cruz Biotechnology) was used as the secondary antibody. The blot was further treated with an ECL Western blotting kit (CWBIO, Beijing, China). Images were obtained using a LAS3000 system (Fujifilm, Tokyo, Japan).

HiBiT lytic assay and p24 ELISA
For the detection of Vpr-HiBiT or LgBiT, 25 l of viruses, culture supernatants, or cell lysates were added into 384-well plates. Next, 25 l of LgBiT protein or HiBiT peptide-containing Nano Glo HiBiT lytic detection reagent (Promega) was added to the wells. After incubation at 37°C for 1 h, luminescence signals were measured using a CentroXS3. For analysis of p24 concentration, HIV-1 p24 Antigen ELISA 2.0 (Zeptometrix, Buffalo, NY) was used according to the manufacturer's instructions.

Infection assay in a 96-well plate
The target cells expressing CD4 and CCR5 were seeded in 96-well cell culture plates (2 ϫ 10 4 cells/well). A day after seeding, the cells were infected with a dose-adjusted pseudotyped virus with 10 g/ml Polybrene. A day after infection, the supernatant was removed, and Steady-Glo luciferase assay reagent (Promega) was added to dissolve cells to measure f-Luc activities.

Statistical analysis
Statistically significant differences between the mean values were determined using a two-tailed Student's t test. All data are representative of at least two independent experiments. The values represent the means of triplicate samples Ϯ S.D. p values less than 0.05 were considered statistically significant. The p value for each analysis was indicated in each figure.