Potent In Vivo NK Cell-Mediated Elimination of HIV-1-Infected Cells Mobilized by a gp120-Bispecific and Hexavalent Broadly Neutralizing Fusion Protein

ABSTRACT Antibodies bound to human immunodeficiency virus type 1 (HIV-1) envelope protein expressed by infected cells mobilize antibody-dependent cellular cytotoxicity (ADCC) to eliminate the HIV-1-infected cells and thereby suppress HIV-1 infection and delay disease progression. Studies treating HIV-1-infected individuals with latency reactivation agents to reduce their latent HIV-1 reservoirs indicated that their HIV-1-specific immune responses were insufficient to effectively eliminate the reactivated latent HIV-1-infected T cells. Mobilization of ADCC may facilitate elimination of reactivated latent HIV-1-infected cells to deplete the HIV-1 reservoir and contribute to a functional HIV-1 cure. The most effective antibodies for controlling and eradicating HIV-1 infection would likely have the dual capacities of potently neutralizing a broad range of HIV-1 isolates and effectively mobilizing HIV-1-specific ADCC to eliminate HIV-1-infected cells. For this purpose, we constructed LSEVh-LS-F, a broadly neutralizing, defucosylated hexavalent fusion protein specific for both the CD4 and coreceptor gp120-binding sites. LSEVh-LS-F potently inhibited in vivo HIV-1 and simian-human immunodeficiency virus (SHIV) infection in humanized mouse and macaque models, respectively, including in vivo neutralization of HIV-1 strains resistant to the broadly neutralizing antibodies VRC01 and 3BNC117. We developed a novel humanized mouse model to evaluate in vivo human NK cell-mediated elimination of HIV-1-infected cells by ADCC and utilized it to demonstrate that LSEVh-LS-F rapidly mobilized NK cells to eliminate >80% of HIV-1-infected cells in vivo 1 day after its administration. The capacity of LSEVh-LS-F to eliminate HIV-1-infected cells via ADCC combined with its broad neutralization activity supports its potential use as an immunotherapeutic agent to eliminate reactivated latent cells and deplete the HIV-1 reservoir. IMPORTANCE Mobilization of antibody-dependent cellular cytotoxicity (ADCC) to eliminate reactivated latent HIV-1-infected cells is a strategy which may contribute to depleting the HIV-1 reservoir and achieving a functional HIV-1 cure. To more effectively mobilize ADCC, we designed and constructed LSEVh-LS-F, a broadly neutralizing, defucosylated hexavalent fusion protein specific for both the CD4 and coreceptor gp120-binding sites. LSEVh-LS-F potently inhibited in vivo HIV-1 and SHIV infection in humanized mouse and macaque models, respectively, including in vivo neutralization of an HIV-1 strain resistant to the broadly neutralizing antibodies VRC01 and 3BNC117. Using a novel humanized mouse model, we demonstrated that LSEVh-LS-F rapidly mobilized NK cells to eliminate >80% of HIV-1-infected cells in vivo 1 day after its administration. The capacity of LSEVh-LS-F to eliminate HIV-1-infected cells via ADCC combined with its broad neutralization activity supports its potential use as an immunotherapeutic agent to eliminate reactivated latent cells and deplete the HIV-1 reservoir.

infection in humanized mouse and macaque models, respectively, including in vivo neutralization of an HIV-1 strain resistant to the broadly neutralizing antibodies VRC01 and 3BNC117. Using a novel humanized mouse model, we demonstrated that LSEVh-LS-F rapidly mobilized NK cells to eliminate Ͼ80% of HIV-1-infected cells in vivo 1 day after its administration. The capacity of LSEVh-LS-F to eliminate HIV-1infected cells via ADCC combined with its broad neutralization activity supports its potential use as an immunotherapeutic agent to eliminate reactivated latent cells and deplete the HIV-1 reservoir.
KEYWORDS NK cell, human immunodeficiency virus B inding to the human immunodeficiency virus type 1 (HIV-1) envelope and neutralizing viral infectivity by preventing viral entry and the subsequent infection of cells constitute the major mechanism utilized by broadly neutralizing antibodies (bNAb) to control HIV-1 infection (1). Antibodies can also utilize another mechanism to inhibit HIV-1 infection, antibody-dependent cellular cytotoxicity (ADCC), which is mediated by their binding to HIV-1 envelope molecules expressed on the surface of infected cells and utilization of their constant (Fc) domains to recruit Fc␥RIIIa-expressing effector cells, such as NK cells, to kill the infected cells (2). ADCC-mediated elimination of reactivated latent HIV-1-infected cells is a potential mechanism by which antibodies can markedly reduce the HIV-1 reservoir and contribute to a functional cure (3). It is likely that the most effective therapeutic antibodies for treating HIV-1-infected individuals are bNAb with the dual capacity to potently neutralize a broad range of HIV-1 isolates and effectively mobilize HIV-1-specific ADCC to eliminate HIV-1-infected cells, including reactivated latent infected cells (4). However, despite the capacity of potent bNAb, such as VRC01 and 3BNC117, to neutralize up to 90% of HIV-1 isolates tested (5), markedly suppress HIV-1 viremia after infusion into HIV-1-infected patients (6), and eliminate HIV-1-infected cells by ADCC (4), the rapid emergence and expansion of bNAb-resistant HIV-1 during bNAb monotherapy (6,7) limits the efficacy of bNAb monotherapy (8,9) and would likely also limit their capacity to mobilize ADCC. To overcome this limitation and prevent or delay the in vivo emergence of bNAb-resistant HIV-1 (5), we developed a bispecific hexavalent CD4-antibody fusion protein, 4Dm2m, composed of two engineered domains, mD1.22 and m36.4, each specific for a different neutralizing gp120 epitope. mD1. 22, an engineered mutant of the D1 extracellular domain of CD4, selectively binds to the gp120 CD4-binding site (10), while m36.4, an antibody domain, targets the highly conserved CD4-induced (CD4i) gp120 coreceptor-binding site (11). Because CD4 binding to gp120 induces full exposure of the m36.4-targetted gp120 epitope, the linkage in 4Dm2m of the soluble one-domain CD4, mD1.22, to the m36.4 domain greatly augments the binding and neutralizing activity of m36.4 (10). 4Dm2m is a bispecific hexavalent fusion protein consisting of four mD1.22 molecules and two m36.4 molecules linked to a heavy-chain constant domain 1 (CH1), a kappa light-chain constant domain (CK), and an IgG1 Fc domain (10). The potential for hexavalent binding of 4Dm2m to gp120 increases its avidity for gp120 and enables it to neutralize HIV-1 ϳ10-fold more potently than the native bNAb, VRC01 (10). Furthermore, the bispecific binding of 4Dm2m to two independent gp120 epitopes should constrain the in vivo emergence of 4Dm2m-resistant HIV-1 by requiring independent mutations at each targeted site, as reported for combination bNAb treatment (5). Finally, because mD1. 22 was designed to mirror the CD4 structure, mutations in gp120 which reduce mD1.22 binding should be paralleled by decreased CD4 binding, which would diminish HIV-1 replicative capacity and thereby inhibit the emergence of mD1.22 escape mutations. We generated a structural variant of 4Dm2m, LSEVh-LS, with a significantly increased in vivo half-life due to its improved structural stability and increased binding to the FcRn (12). We further augmented the capacity of LSEVh-LS to mobilize ADCC activity by defucosylating its Fc domain to increase its affinity for Fc␥RIIIa and thereby amplify its ability to recruit effector cells (13). In the current study, we examined the in vitro and in vivo anti-HIV-1 activities of the defucosylated LSEVh-LS, named LSEVh-LS-F, and Journal of Virology the luciferase reporter gene after incubation with CHO-gp160SC cells but not with CHO cells (Fig. 2D).

LSEVh-LS-F eliminates HIV-1-infected cells by an NK cell-mediated mechanism.
We adapted the hu-spl-PBMC-NSG mice to be used as a novel in vivo model for evaluating HIV-1-specific ADCC activity to investigate the in vivo capacity of defucosylated LSEVh-LS-F to eliminate HIV-1-infected cells. hu-spl-PBMC-NSG mice were infected by intrasplenic injection with HIV-1 JRCSF -LucR. Five days later, after in vivo infection was established, mice were either left untreated or treated with LSEVh-LS-F, VRC01, or an isotype control MAb. Because full HIV-1 replication in primary T cells takes at least 24 h (27), we posited that the reduction in LucR activity measured 1 day after treatment should indicate the elimination of HIV-1-infected cells by ADCC. One day after HIV-1 JRCSF -LucR-infected hu-spl-PBMC-NSG mice were treated with LSEVh-LS-F, the number of HIV-1-infected cells in the mouse spleens, as indicated by LucR activity, was reduced by almost 90% compared to that after treatment with the control MAb, which was 2-fold greater (P Ͻ 0.01) than the ϳ40% reduction mediated by VRC01 treatment (Fig.  4A). In vivo visualization of productively infected cells by IVIS imaging demonstrated a marked reduction in LucR activity in the spleens of hu-spl-PBMC-NSG mice treated with LSEVh-LS-F compared to the spleens of untreated hu-spl-PBMC-NSG mice (Fig. 4B). To human IgG1 hinge. The calculated molecular mass is shown in parentheses. (B) High-resolution mass spectrometry analysis of LSEVh-LS produced in 293-F cells, which was designated LSEVh-LS, and LSEVh-LS produced in GFT gene knockout CHOF6 cells, which was designated LSEVh-LS-F. Mass spectra were shown, with deconvoluted mass for the major peak indicated at the top. G0, Fc oligosaccharides without galactose and fucose; G0F, Fc oligosaccharides without galactose but with fucose. (C) Binding kinetics of LSEVh-LS and LSEVh-LS-F with recombinant human Fc␥RIIIa as measured by SPR. SPR analysis was performed on Biacore X100 by using a single-cycle approach according to the manufacturer's instructions. Analytes were tested at 1000, 500, 250, 125, and 62.5 nM concentrations. The kinetic constants shown on the right were calculated from the sensograms fitted with monovalent binding model of the BiacoreX100 evaluation software 2.0. k a , association rate constant; k d , dissociation rate constant; K D , equilibrium dissociation constant. confirm that the LSEVh-LS-F-mediated reduction in HIV-1 infection was a consequence of NK cell-mediated ADCC (28), we depleted the NK cell population from the PBMC prior to their injection into hu-spl-PBMC-NSG mice and examined whether this would abrogate LSEVh-LS-F activity. NSG mice were intrasplenically injected with either unfractionated PBMC or NK cell-depleted PBMC and intrasplenically infected with HIV-1 JRCSF -LucR. Five days later the humanized mice were either left untreated or treated with LSEVh-LS-F, and 1 day later, HIV-1 infection in the spleens was quantified by measuring LucR activity. As described above, LSEVh-LS-F treatment markedly reduced HIV-1 infection by Ͼ70% in the spleens of NSG mice constructed by intrasplenic injection of unfractionated PBMC. In contrast, LSEVh-LS-F treatment did not decrease HIV-1 infection in the spleens of NSG mice intrasplenically injected with NK celldepleted PBMC (Fig. 5A).
Activation of NK cells can increase their capacity to mediate ADCC. For example, the ADCC-mediated in vivo elimination of human B cell lymphoma cells via NK cells recruited by anti-CD20 MAb was significantly increased by NK cell activation with IL-15SA (ALT-803), a bivalent IL-15R␣-Fc fusion protein binding two mutant IL-15 molecules (29). We previously demonstrated that in vivo IL-15SA treatment of hu-spl-PBMC-NSG mice activates the human NK cells populating the mouse spleen and enables them to potently reduce intrasplenic HIV-1 infection (26). To determine whether activation of NK cells by IL-15SA increased LSEVh-LS-F-mediated depletion of HIV-1-infected cells, we infected hu-spl-PBMC-NSG mice with HIV-1 JRCSF -LucR and 5 days later either did not treat the mice or treated the mice with IL-15SA, LSEVh-LS-F, or a combination of both. For these experiments, we reduced the treatment dose of LSEVh-LS-F by half to enable us to identify the additive effect of IL-15SA treatment on the LSEVh-LS-F-mediated depletion of HIV-1-infected cells. The combination of LSEVh-LS-F and IL-15SA reduced HIV-1 JRCSF -LucR infection by 80%, 2-fold more than treatment with either LSEVh-LS-F or IL-15SA alone (Fig. 5B). Taken together, these results demonstrate that the bispecific hexavalent antibody LSEVh-LS-F potently inhibited HIV-1 infection, as well as markedly reduced established HIV-1 infection via NK celldependent ADCC, which could be further enhanced by IL-15SA activation of NK cells.

LSEVh-LS-F potently inhibits SHIV infection in macaques.
To examine the capacity of LSEVh-LS-F to inhibit acute SHIV infection of rhesus macaques, 7 days after infection with SHIV SF162P3 and prior to the onset of peak viremia, we injected them with either LSEVh-LS-F (10 mg/kg) or phosphate-buffered saline (PBS). SHIV RNA in plasma was measured at the indicated times (Fig. 6A). PBS-treated animals reached the peak of infection on average at day 9 after infection, which was suppressed by LSEVh-LS-F treatment on day 7 (Fig. 6B). A well-recognized limitation of macaque studies involves the cost and logistic factors that limit the number of animals available for inclusion in each experimental group. When the number of animals in each group is relatively small, as in the case of the present study, and variations are large, which is typical for macaques, a significant quantitative effect may not be identified by direct statistical comparison with a control group. Individual variations of viremia were seen in each macaque. Therefore, to evaluate the statistical significance of the difference between treated and untreated macaques, we normalized the virus RNA by dividing by the value of virus RNA at day 7 and subtracting 1. The administration of LSEVh-LS-F at day 7 resulted in a highly statistically significant (P ϭ 0.0001) reduction of peak viremia at day 9 compared to peak viremia in untreated macaques (Fig. 6C). While at subsequent time points there was a trend of lower virus RNA concentrations in the treated than in the untreated macaques, other than at day 18 (P ϭ 0.02), there was no statistically significant decrease in virus RNA (Fig. 6D). These in vivo results corresponded with the potent capacity of LSEVh-LS-F to neutralize in vitro SHIV SF162P3 infection (IC 50 ϭ 1.6 g/ml, IC 80 ϭ 3.9 g/ml) (Fig. 6D). To determine whether the transitory inhibitory effect of LSEVh-LS-F on SHIV viremia was due to a short serum half-life, we measured LSEVh-LS-F pharmacokinetics in the infected macaques. The LSEVh-LS-F serum concentration was high (up to 100 g/ml) at 2 days after infusion but rapidly decreased to Ͻ1 g/ml by 7 days after infusion, a concentration unlikely to be effective in suppressing in vivo SHIV replication (Fig. 6E). One month after infection, the macaques were sacrificed, and SHIV infection in the macaque tissues was evaluated for the levels of SHIV RNA. In the ileum, rectum, duodenum, and colon of LSEVh-LS-F-treated macaques, SHIV RNA levels were significantly lower than in the untreated group (Fig. 7A), but the levels of SHIV RNA in the lymph nodes (inguinal, mesenteric, and axillary), brain, and testis/ovary were comparable (Fig. 7B). Thus, while LSEVh-LS-F is highly effective in suppressing in vivo SHIV replication, the durability of in vivo inhibitory effects of a single dose against SHIV replication was limited by the relatively short serum half-life of LSEVh-LS-F.

DISCUSSION
HIV-1-specific antibodies can eliminate HIV-1-infected cells by ADCC through Fc␥mediated recruitment of innate immune effector cells, particularly NK cells, which lyse infected cells by exocytosis of granzymes and perforin (30,31) and may contribute to the prevention and control HIV-1 infection (32,33). This is supported by studies which  (37), in vivo efficacy of bNAb therapy (31,38), emergence of immune escape variants (39), and accelerated elimination of HIV-1infected cells in 3BNC117-treated humanized mice (40). Because ADCC activity correlates with total IgG bound to HIV-1-infected cells (33), the bispecific hexavalent binding of LSEVh-LS-F to the spatially adjacent CD4 and the CD4i coreceptor-binding sites (41) should favor multivalent targeting of individual gp120 trimeric molecules in the HIV-1 In Vivo Elimination of HIV-Infected Cells by ADCC Journal of Virology membrane, whose wide separation may otherwise prevent efficient bivalent binding by monospecific natural bNAb (42)(43)(44)(45), and thereby increase its capacity to mobilize ADCC-mediated elimination of HIV-1-infected cells. Using a strategy reported to enhance the in vivo ADCC activity of antibodies used for cancer therapy (17,18,46), we defucosylated LSEVh-LS-F to increase the affinity of its Fc domain for Fc␥RIIIa on innate effector cells and thereby further increase its capacity to mediate HIV-specific ADCC activity. We developed a hu-spl-PBMC-NSG humanized mouse model, which contains human NK cells with high cytotoxic function (26) The level of HIV-1 infection in hu-spl-PBMC-NSG mice was reduced by Ͼ80% at 1 day after LSEVh-LS-F treatment, which was 2-fold greater than the reduction mediated by VRC01 treatment (Fig. 4A). These results were confirmed by direct in vivo imaging of the LSEVh-LS-F-treated mice by IVIS scanning (Fig. 4B). NK cell depletion abrogated LSEVh-LS-F-mediated reduction in HIV-1 infection, indicating that elimination of HIV-1-infected cells by LSEVh-LS-F was mediated by ADCC (Fig. 5A). This was further supported by our finding that activation of NK cell cytotoxic activity by IL-15SA treatment increased LSEVh-LS-F-mediated in vivo reduction in HIV-1 infection (Fig. 5B). In addition to permitting us to determine the in vivo ADCC activity of LSEVh-LS-F, the hu-spl-PBMC-NSG humanized mouse model we developed for in vivo ADCC evaluation should provide investigators with a crucial high-throughput model to identify HIV-1-specific antibodies, including those with mutated Fc domains (47), and immunomodulators with the most potent in vivo capacity to mobilize ADCC activity to eliminate HIV-1-infected cells.  In vivo studies in humanized mice, macaques, and humans have demonstrated that treatment with a single bNAb can potently prevent and suppress HIV-1 infection and possibly reduce the HIV-1 latent reservoir by Fc domain-mediated recruitment of effector cells such as by ADCC (6,7,9,(47)(48)(49)(50)(51). However, a limitation of current bNAb is that no single bNAb neutralizes all strains of HIV-1, enabling the emergence of immune escape variants mutated in bNAb-specific epitopes, likely from preexisting mutant virus (52), and mediating the recurrence of viremia (5,53). A single amino acid mutation can enable HIV-1 to escape from natural bNAb such as VRC01, 3BNC117, and 10-1074, which is a major impediment currently limiting the efficacy of bNAb monotherapy (7,(54)(55)(56). A substantial number of HIV-1-infected individuals are populated with preexisting VRC01-resistant virus, and consequently, treatment of some of these individuals with VRC01 only minimally suppressed viremia (9). VRC01-resistant HIV-1 and 3BNC117-resistant HIV-1 also rapidly emerged during analytical antiretroviral therapy (ART) treatment interruption studies in patients after VRC01 and 3BNC117 administration, respectively (7,8). Consequently, treatment with combinations of bNAb specific for nonoverlapping epitopes will be required for ADCC-mediated elimination of cells infected with immune escape variants to prevent recurrence of viremia effectuated by resistant virus, emulating the successful strategy of combining different antiretroviral drugs (53). We previously demonstrated that a bispecific Fc domain fusion protein expressing CD4 and the CD4i-gp120 epitope-specific human antibody domain m36 displayed more potent in vitro neutralizing activity than Fc domain fusion proteins expressing either CD4 or m36 alone (57). The targeting of two independent gp120 epitopes by LSEVh-LS-F, which contributes to its potent and broad neutralizing activity, should enable it to mediate ADCC elimination while also constraining the in vivo emergence of LSEVh-LS-F-resistant HIV-1 escape mutants by requiring mutations in two nonoverlapping gp120 epitopes for immune escape. We demonstrated the in vivo inhibitory activity of LSEVh-LS-F against HIV-1 and SHIV by using a humanized mouse model and a nonhuman primate model, respectively. The breadth of LSEVh-LS-F activity was demonstrated by its in vitro neutralization of VRC01-resistant Env strains by LSEVh-LS-F (Fig. 2B) and its in vivo inhibition of HIV-1 infection by a VRC01-resistant strain, which has also shown resistance to 3BNC117 (Fig. 3D).
An alternate approach to combination therapy with two different bNAb targeting nonoverlapping gp120 epitopes is to generate a bispecific IgG molecule consisting of the antigen-binding fragments from two different bNAb linked to an Fc domain (23,42). LSEVh-LS-F, which combined a gp120-binding CD4 subunit, mD1.22, and a CD4i coreceptor-binding site-specific antibody fragment, m36.4 (10), differs from bispecific antibodies in three crucial ways. First, instead of using a bNAb domain to recognize the gp120 CD4 binding site, it utilizes a one-domain CD4, mD1.22, that was mutated to exhibit high solubility and stability while retaining binding to all HIV-1 Envs. By mimicking CD4 binding to gp120, mD1.22-resistant escape mutants should display compromised viral replication due to the reduced capacity of their mutated gp120 to bind CD4 and initiate cellular entry and subsequent infection. Second, by engaging with the CD4 binding site of gp120, the mD1.22 induces the exposure of CD4-induced antibody epitopes on gp120, a major target of antibodies circulating in infected patients which mediate ADCC, and boost their capacity to eliminate HIV-1-infected cells via ADCC (20,58). Third, the bispecific and hexavalent binding of LSEVh-LS-F may increase its avidity for gp120 beyond that of natural monospecific and bivalent antibodies capable of mediating ADCC by facilitating its binding to the limited number of gp120 molecules expressed by HIV-1 and on the surface of HIV-1-infected cells (43). The enhanced avidity of LSEVh-LS-F for gp120 expressed by infected cells was indicated by its increased binding to gp120-expressing ACH2 cells compared to that of VRC01 ( Fig. 2A).
Two days after treatment of macaques during acute SHIV infection with LSEVh-LS-F, peak viremia was suppressed by about 1.2 logs compared to that in untreated macaques (Fig. 6B). A similar approach was used to demonstrate that VRC01 alone or the combination of VRC07-523 and PGT121 suppressed acute SHIV infection and reduced the seeding of cell-associated viral reservoirs, but the experimental protocol for that study differed from ours by sustaining bNAb-mediated viral suppression by initiating ART at 11 days after bNAb treatment (59). The 1.2-log reduction in plasma virus RNA induced by 2 days of LSEVh-LS-F treatment corresponds to k ϭ 1.38 day Ϫ1 , which is higher than the k ϭ 0.36 to 0.89 day Ϫ1 calculated for macaques chronically infected with SHIV and treated with the bNAb 3BNC117 and 10-1074 (51), indicating that LSEVh-LS-F displays very potent antiretroviral activity. The marked reduction in SHIV infection of LSEVh-LS-F-treated macaque mucosal tissues may be a consequence of the significant reduction in peak viremia mediated by LSEVh-LS-F. This reduction in SHIVinfected mucosal tissues is unlikely to be due to elimination of SHIV-infected cells by LSEVh-LS-F-mediated ADCC, because species differences in the specificity and affinity of the human Fc domain for macaque Fc receptors may limit their capacity to mobilize ADCC in macaques (59,60). While LSEVh-LS-F plasma levels averaged about 100 g/ml for the first 2 days after administration, levels subsequently declined more rapidly than those of bNAb such as VRC01 and 3BNC117. As previously described, several factors may contribute to the shorter half-life of LSEVh-LS-F than of native bNAb, including the reduced ability of the CCH1-CK interaction to generate extremely stable heterodimers and the increased susceptibility to proteolytic degradation associated with the use of polypeptide linkers to link mD1.22 and m36.4 to the Fc scaffold (12). We are investigating new approaches to enhance the half-life of LSEVh-LS-F. Nevertheless, its potent capacity to neutralize HIV-1 strains resistant to other bNAb and to eliminate HIV-1infected cells should counterbalance its shorter half-life, and this supports the potential inclusion of short, intense courses of LSEVh-LS-F as a component of therapeutic regimens that combine immunostimulants, such as ALT-803, and latency-reactivating agents to reduce the HIV-1 reservoir and contribute to the final goal of HIV-1 eradication. , NIH, Bethesda, MD) and processing the cell culture supernatant as described previously (10,12). Defucosylated LSEVh-LS (LSEVh-LS-F) was produced using CHO cells with their GDP-fucose transporter (GFT) genes inactivated using the clustered regularly interspaced short palindromic repeat (CRISPR)/Cas9 system as described previously (19). Briefly, three GDP-fucose transporter gene DNA fragments (target 1, 5= CGCTGGTCGTCTCTCTCTAC 3=; target 2, 5= AGATAGGGGTATCCAGCTGC 3=; and target 3, 5= GTACTTG TTGAGGAATACCA 3=) were cloned into the pCas-Guide-GFP vector (OriGene Technologies, Rockville, MD) and were transfected into CHO cells using Polyfect (Qiagen, Frederick, MD) according to the manufacturer's protocol. Three days after transfection, the green fluorescent protein (GFP)-positive single cells were sorted into 96-well plates and expanded. Successful GDP-fucose transporter gene deletion in one CHO cell knockout mutant clone, CHO-F6, was demonstrated by mass spectral analysis of purified antibodies obtained after transient antibody expression. CHO-F6 was adapted for growth in serum-free medium and was transfected with the LSEVh-LS vector to establish stable cell lines producing LSEVh-LS-F using the standard glutamine synthetase-based selection system (61). 4Dm2m, LSEVh-LS, and LSEVh-LS-F were purified from the 293-F and CHO-F6 cell culture supernatants by protein A-Sepharose 4 Fast Flow column chromatography (GE Healthcare, Piscataway Township, NJ) as described previously (11).

Cells
High-resolution MS. The proteins were mixed with buffer (7.5 M guanidine-HCl, 0.1 M Tris-HCl, and 1 mM EDTA) in the presence of 20 mM dithiothreitol (DTT) and incubated at 70°C for 15 min. Mass spectrometry (MS) data were acquired on an Agilent 6520 Accurate-Mass Q-TOF LC/MS system (Agilent Technologies, Danbury, CT) equipped with a dual electrospray source, operated in the positive-ion mode. Separation was performed on a Zorbax 300SB-C3 Poroshell column (2.1 mm by 75 mm; particle size, 5 m). The analytes were eluted at a flow rate of 1 ml/min with a 1 to 90% organic gradient over 5 min and holding organic for 1 min. Both mobile phases, water and acetonitrile, contained 0.1% formic acid. The instrument was used in a full-scan time-of-flight (TOF) mode. MS source parameters were set with a capillary voltage of 4 kV, fragmentor voltage of 220 V, and skimmer voltage of 65 V. The gas temperature was 350°C, the drying gas flow 12 liters/min, and nebulizer pressure 55 lb/in 2 gauge. Data were acquired at high resolution (3,200 m/z) and 4 GHz. TOF-mass spectra were recorded across the range of 100 to 3,200 m/z. Mass accuracy was maintained during the run time by infusing an internal mass calibration sample continuously during the liquid chromatography (LC)-MS runs. Data acquisition was performed using a Mass Hunter work station (Agilent, version B.02.00), and data analysis and deconvolution of mass spectra were performed using Mass Hunter qualitative analysis software (Agilent, version B.03.01) with BioConfirm Workflow.
SPR. The kinetics and affinity of LSEVh-LS and LSEVh-LS-F binding to Fc␥RIIIa were quantified by surface plasmon resonance (SPR) analysis on a Biacore X100 instrument (GE Healthcare, Port Washington, NY) using a single-cycle approach according to the manufacturer's instructions. Briefly, purified proteins were diluted in sodium acetate (pH 5.0) and immobilized directly onto a CM5 sensor chip with the standard amine coupling method. The reference cell was injected with N-hydroxysuccinimide-1-ethyl-3-(3-dimethyaminopropy)carbodiimide and ethanolamine without injection of Fc␥RIIIa. Analytes were tested at 1,000, 500, 250, 125, and 62.5 nM concentrations. Kinetic constants were calculated from the sensograms fitted with the monovalent binding model of the BiacoreX100 evaluation software 2.0.
Generation of HIV-1 infectious molecular clones. Inhibition of in vitro and in vivo HIV-1 infection was determined utilizing the LucR reporter-expressing HIV-1 infectious molecular clone (IMC) system (62), in which the proviral genome has been engineered to express, in cis, the LucR reporter gene and heterologous HIV-1 env gene sequences of choice. The approach is well established in multiple contexts, including the evaluation of the in vitro and in vivo HIV-1 inhibitory capacities of antibodies, NK cells, CD8 ϩ T cells, dual-affinity retargeting proteins (DARTs), and microbicides (26,63). Proviral plasmids, collectively referred to as pNL-LucR.T2A-Env.ecto (62), carrying the env genes indicated below were transiently transfected into 293T cells to produce virus stocks of HIV-1 IMC expressing LucR (HIV-1 Env -LucR). HIV-1 Env -LucR reporter viruses are replication competent and continue to express LucR over multiple cycles of replication, which permits the highly sensitive and specific detection of active HIV-1 replication for several weeks after inoculation (62). Because LucR has a short cellular half-life of approximately 3 h (64), the measurement of LucR levels correlates with active HIV-1 replication. For these experiments we used HIV-1 Env -LucR viruses with VRC01-sensitive HIV-1 Env strains JR-CSF, Bal, and 45_01dH5 (clade B) and C.1176 (clade C) and resistant Env strains 45_01E11 (clade B) and C.Du151.2, C.Du172.17, and C.Du422.1 (clade C) (21,22). HIV-1 Env C.Du422.1is also resistant to neutralization by the bNAb 3BNC117 (23,24). The infectious titer of HIV-1 Env -LucR viruses (on average 5 ϫ 10 6 to 10 ϫ 10 6 infectious units [IU]/ml) was determined by a limiting-dilution infection assay using TZM-bl cells as described previously (62).
Assay of antibody binding to cell surface Env. Binding of LSEVh-LS-F to Env trimers expressed on the surface of infected cells and comparison to VRC01 binding was performed by flow cytometric analysis. LSEVh-LS-F, VRC01, and a control MAb recognizing Ebola virus, ADI-15878, were biotinylated using EZ-Link sulfo-NHS-LC-biotin (Life Technologies) followed by a desalting step using a Zeba spin desalting column (Life Technologies). ACH-2 cells, a latently infected T cell line used as a model of HIV-1 latency (20), were activated with phorbol myristate acetate (PMA) (2 ng/ml) for 48 h and then stained with each biotinylated antibody at the indicated concentration determined after biotinylation for 30 min at 4°C, followed by washing and incubation with streptavidin-phycoerythrin (PE) (Life Technologies) at a 1:1,000 dilution for 30 min at 4°C. Data were acquired using an LSR II flow cytometer (BD Biosciences) and analyzed with FlowJo software (FlowJo, Ashland, OR).
In vitro ADCC assay. As target cells, CHO cells or CHO cells which stably express a clade-B HIV-1 Env (CHO-gp160SC cells) suspended in R10 were plated (2.5 ϫ 10 4 cells/well) into 96-well plates (25 l/well), and the indicated concentration of LSEVh-LS-F diluted in R10 was added to each well (50 l/well). For effector cells, genetically modified Jurkat cells expressing the human Fc␥RIIIa with a luciferase reporter gene under the transcriptional control of the NFAT-RE (Promega) were added (1.5 ϫ 10 5 cells/well in 25 l R10) for a target/effector cell ratio of 1:6. After overnight incubation at 37°C, the Bio-Glo Luciferase assay reagent (Promega) was added, and luminescence quantified as relative luminescence units (RLU) was measured using a luminescence plate reader.
Pharmacokinetic measurement of serum levels of VRC01 and LSEVh-LS-F in mice and in rhesus macaques. NSG mice were intravenously injected with 0.5 mg of either VRC01 or LSEVh-LS-F on day 0. Plasma samples were collected by submandibular bleeding at the indicated times after injection.

Measurement of the in vivo capacity of LSEVh-LS-F to inhibit HIV-1 infection.
The in vivo capacity of LSEVh-LS-F or VRC01 to inhibit HIV-1 infection was determined using our previously described humanized mouse model, in which NOD-SCID-IL-2r␥ Ϫ/Ϫ (NSG) mice were intrasplenically injected with activated human PBMC (hu-spl-PBMC-NSG mice) (26). Briefly, human PBMC isolated from HIV-1-naive donors were cultured in R10 and activated with PHA (4 g/ml) and IL-2 (100 U/ml) for 24 h at 37°C. In some experiments, NK cells were removed from the PBMC after activation by immunomagnetic sorting using anti-human CD56 microbeads (Miltenyi Biotec, Cambridge, MA). Depletion (Ͼ95%) was confirmed by flow cytometric analysis after staining with anti-human CD3-FITC and anti-human CD56-PE (Bio-Legend) antibodies as described previously (26). The next day, after the activated PBMC or NK celldepleted PBMC were washed twice with sterile PBS, HIV-1-LucR was added (ϳ1 ϫ 10 6 IU/10 7 PBMC) to the activated PBMC, and the cells were centrifuged for 1 h at 25°C at 2,500 rpm, resuspended, and injected intrasplenically (ϳ10 7 cells/mouse) into NSG mice (Jackson Laboratories, Bar Harbor, ME). The mice were bred and maintained in our biosafety level 2 (BSL2)-enhanced animal facility at Albert Einstein College of Medicine as described previously (65). The capacity of LSEVh-LS-F or VRC01 to inhibit HIV-1 infection was evaluated by intravenously injecting hu-spl-PBMC-NSG mice with VRC01 (0.5 mg) or LSEVh-LS-F (0.5 mg) at 1 day after intrasplenic inoculation with the indicated HIV-1 Env -LucR. As a negative control, we used m336, a monoclonal antibody against the Middle East respiratory syndrome coronavirus (66). The level of HIV-1 infection in the engrafted human PBMC in hu-spl-PBMC-NSG mouse spleens was quantified by measuring the LucR activity in the mouse splenic lysates using the Renilla luciferase assay system (Promega), as previously described (26). To enable us to combine the results of multiple experiments for statistical analysis, the results are reported as percent neutralization of HIV-1 infection by each treatment group in reference to the untreated group using the formula y ϭ [(1 Ϫ treated group RLU)/untreated group RLU ϫ 100].
To determine the in vivo capacity of LSEVh-LS-F or VRC01 to eliminate HIV-1-infected cells, hu-spl-PBMC-NSG mice or NSG mice intrasplenically injected with NK cell-depleted PBMC (ϳ10 7 cells/mouse) were intravenously injected with either LSEVh-LS-F (0.5 mg) or VRC01 (0.5 mg) at 5 days after infection with the indicated HIV-1 Env -LucR, to provide time for the establishment of in vivo infection and sufficient HIV-1-infected targets to evaluate the effectiveness of ADCC-mediated elimination of HIV-1-infected cells.
For some experiments, the mice were intravenously injected with LSEVh-LS-F (0.25 mg) and/or subcutaneously injected with a dose of the IL-15SA ALT-803 (0.2 mg/kg), which provides the mice with a maximum concentration in serum (C max ) of ϳ25 nM (26). The day after treatment, the reduction in the number of HIV-1-infected cells in the mouse spleens was determined by quantifying the decrease in LucR activity compared to LucR levels in the untreated group using the formula y ϭ [(1 Ϫ treated group RLU)/untreated group RLU ϫ 100].
IVIS imaging. In vivo HIV-1 infection in the hu-spl-PBMC mice was visualized and quantified by bioluminescent imaging using the IVIS Spectrum imager (Caliper LifeSciences, Hopkinton, MA) after the mice were intravenously injected (5 g/mouse) with the bioluminescence substrate RediJect Coelenterazine h (Caliper Life Sciences). The images were analyzed using the Wizard bioluminescent selection tool for automatic wavelength and exposure detection, and the bioluminescent and gray-scale images were overlaid using the LivingImage 4.0 software package to create a pseudocolor image that represents bioluminescence intensity. The bioluminescent intensity of the luciferase activity in the mouse spleens was quantified using the LivingImage 4.0 software package and reported as photon counts/second.

Measurement of LSEVh-LS-F inhibitory activity in SHIV-infected macaques.
Eight healthy male and female Chinese-origin rhesus macaques were randomly assigned to two groups (four in each), injected intravenously with SHIV SF162P3 (1,000 50% tissue culture infective doses [TCID 50 ]), and 7 days later intravenously injected with LSEVh-LS or PBS. Whole blood was collected at days 0, 7, 9, 16, 18, 21, 25, and 30. All animals were sacrificed at day 30, and tissue samples were collected and stored at Ϫ80°C. In vitro neutralization of SHIV SF162P3 by LSEVh-LS-F was evaluated using the TZM-bl assay (67). Briefly, serially diluted LSEVh-LS-F and the irrelevant control antibody m336 were preincubated with SHIV SF162P3 for 1 h at 37°C before being added to the TZM-bl cells in a 96-well plate. The luciferase expression was quantified 48 h after infection upon cell lysis and the addition of luciferase substrate (Invitrogen).
To precisely measure the viral RNA from plasma and necropsy tissues, we utilized a new ultrasensitive, nested, quantitative real-time PCR (RT-qPCR) method targeting a highly conserved region in simian immunodeficiency virus (SIV) gag (68). SIV/SHIV RNA loads in plasma and tissues were quantified by extracting RNA either from plasma (500 l) using the QIAamp viral RNA minikit (Qiagen) or from tissues (ϳ100 mg) pulverized with stainless steel grinding balls and homogenized in TRIzol reagent following the manufacturer's recommendations. We used "nested" primers SIVnestF01 (GATTTGGATTAGCAGAAA GCCTGTTG) and SIVnestR01 (GTTGGTCTACTTGTTTTTGGCATAGTTTC) flanking the SIV gag target region for reverse transcription and preamplification. The RNA was reverse transcribed into cDNA using the nested primer SIVnestR01 to facilitate priming of specific target sequences, avoid generation of nonspecific sequences, and further enhance the sensitivity of detection. The cDNA was then preamplified by PCR for 12 cycles (94°C for 30s, 60°C for 1 min, and 72°C for 1 min) with the nested primers and amplified by real-time quantitative PCR for SIV DNA gag for 45 cycles (95°C for 10 s and 60°C for 1 min) using the forward primer SGAG21 (5=-GTCTGCGTCATPTGGTGCATTC-3=) and the reverse primer SGAG22 (5=-CACT AGKTGTCTCTGCACTATPTGTTTTG-3=). The amplified product was detected by hybridization to the la-beled probe pSGAG23 (5=-FAM-CTTCPTCAGTKTGTTTCACTTTCTCTTCTGCG-BHQ1-3=). Quantitative determinations for samples showing amplification in all replicates were derived directly with reference to a linearized SIVmac239 standard curve (triplicate reactions for 10 dilutions from 10 1 copies/reaction to 10 7 copies/reaction) by averaging the values determined for each reaction based on interpolation of the measured threshold cycle values onto the standard curve of input template copy number versus threshold cycle values performed with each assay.
Statistical analysis. The statistical significance of the in vivo neutralization or suppression of HIV-1 infection by the different treatment groups compared to the untreated group was determined through the one-way analysis of variance (ANOVA) multiple-comparison test or the unpaired t test. GraphPad Prism software was used for the statistical analysis, and differences were considered statistically significant when the P value was Ͻ0.05.
Study approval. All the mouse studies were approved by the Institute for Animal Studies and the Institutional Review Board at Albert Einstein College of Medicine in compliance with the human and animal experimentation guidelines of the U.S. Department of Health and Human Services and adherence to the NIH Guide for the Care and Use of Laboratory Animals. The macaque studies were conducted in the biosafety level 3 laboratory with protocols approved by the Institutional Animal Care and Use Committee of the Institute of Laboratory Animal Science, Chinese Academy of Medical Sciences.