Functional analysis of a monoclonal antibody reactive against anti-cluster A epitope obtained from a patient infected with HIV-1 CRF02_AG

Hasan Md Zahid Kumamoto University Takeo Kuwata Kumamoto University https://orcid.org/0000-0001-7793-6585 Shokichi Takahama National Institutes of Biomedical Innovation, Health and Nutrition Kaku Yu Kumamoto University Shaswata Biswas Kumamoto University Kaho Matsumoto Kumamoto University Hirokazu Tamamura Tokyo Medical and Dental University Shuzo Matsushita (  shuzo@kumamoto-u.ac.jp ) Kumamoto University

CD4mc and anti-CoRBS antibodies (Abs) bind sequentially to Env trimer, opening its conformation and allowing recognition by anti-cluster A antibodies whose epitopes are located in the gp120 inner domain and remain occluded in the native trimer (10,(14)(15)(16)(17). It is not only Fab fragments that interact with the same Env trimer, but Fc fragments of these two families of Abs also bind synergistically with FcγRIIIa (18). Fc-dependent mechanisms can impact on the viral load (19,20) and cause viral escape that can result in control of HIV-1 (21) by slowing disease progression (22).
Recent reports have described the superior nature of the IgG3 class of antibodies over the IgG1 class, not only for Fc-receptor-mediated functions, such as ADCC and antibody-dependent cellular phagocytosis (ADCP) (21,(23)(24)(25)(26)(27), but also for their neutralization capability (26). The level of IgG3 correlated with anti-HIV-1 function in the RV-144 trial (28, 29) and was involved in the control of disease in different cohorts (21,30). The hinge region of IgG3, which links the Fab and the Fc regions, is two to four times longer than that of IgG1. This increased hinge length may have an effect on the exibility of the antibody and the recognition of antigen, which would ultimately result into differences in protection (31,32). Although IgG3 has higher functional potential, it has not been advanced clinically partly because of its shorter half-life compared with IgG1 (33,34).
CRF02_AG comprises 46% of the circulating strains in West Africa and is the fourth most abundant subtype in the world (35). It is also considered as the most frequent non-B subtype spreading among European natives (36). In this study we isolated a monoclonal antibody (mAb) against the cluster A epitope, 1E5, from a CRF02_AG-infected patient. IgG1 and IgG3 forms of 1E5 were constructed and examined for their functional characteristics. Previous studies on ADCC have focused on anti-cluster A antibodies from subtype B-infected patients, such as A32 and C11 (11,12,37). We observed the ADCC activity of 1E5 against HIV-1 Env from subtype CRF02_AG and subtype A with no neutralizing activity.
Moreover, enhancement in the binding and ADCC activities of A32 were observed in the presence of 1E5.
The combination of anti-cluster A antibodies induced by different subtypes may have implications for vaccine development against HIV-1.

Results
Isolation of monoclonal antibody 1E5 from a donor infected with CRF02_AG B cells from a donor infected with CRF02_AG were transformed by Epstein-Barr virus (EBV) and the supernatants were screened for reactivity to the Env of HIV-1 93TH966.8 (CRF01_AE) strain. Recombinant mAb, 1E5, was isolated from the single cell-sorted Env-reactive culture by RT-PCR of the immunoglobulin heavy and light chain genes. Genetic analysis revealed that 1E5 used IGHV1-69*09 and IGKV3-20*01 as germline genes of the heavy and light chain genes, respectively (Additional le 1: Fig. S1). The binding activity of 1E5 to Env proteins from various HIV-1 strains was examined by ow cytometry (Fig. 1a). This mAb bound well to the Env protein of subtype A (92UG037.8), CRF01_AE (93TH976.17 and 93TH966.8), subtype C (ZM233M.PB6) and subtype B (WITO4160.33 and RHPA4259.7). The results suggested that 1E5 cross-reacts to HIV-1 strains belonging to various subtypes, although the Env proteins of subtype C (ZM109F.PB4) and subtype B (REJO4541.67 and JR-FL) were not recognized by 1E5. The binding activity of 1E5 to intra-subtype strains was tested using a panel of CRF02_AG Env proteins (Fig. 1b). The binding of 1E5 was observed in most CRF02_AG strains (13 out of 15 strains), suggesting that the 1E5 epitope is conserved among most CRF02_AG strains. However, the reactivity of 1E5 was moderate for AG-250, AG-258, AG-278 and AG-263, and was not detected for AG-235 and AG-928.
IgG3 form of 1E5 showed better reactivity than the IgG1 form Recent reports described the superior nature of the IgG3 class over IgG1 not only for Fc-receptor-mediated functions, such as ADCC and ADCP, but also for neutralization capability (21,(23)(24)(25)(26)(27). To obtain 1E5 with superior activities, we constructed an IgG3 form of 1E5 in addition to the IgG1 form. Analysis of crossreactivity using a global panel of HIV-1 (38) showed that both IgG1 and IgG3 forms of 1E5 reacted to 8 out of 11 strains (Fig. 2). The reactivity of 1E5 appeared high for subtype A and AE strains, such as p398F1 and pCNE8, as well as subtype C and BC strains, such as p25710, pCE1176 and pCH119. The 1E5 reacted moderately to pCNE55, pTRO11, pX1632 and pCH119, but showed no detectable reactivity to pX2278, pCE0217 and pBJOX2000. The reactivity of the IgG3 form of 1E5 was signi cantly higher than that of the IgG1 form (Additional le 2: Fig. S2). Taken together, the binding data revealed that 1E5 reacted to 27 out of 35 strains tested (77%).

Determination of the epitope recognized by 1E5
To determine the 1E5 epitope, a panel of Env mutants, which were affected in their reactivity of potent anti-HIV-1 antibodies (39)(40)(41), were used to examine the reactivity to 1E5. However, the results revealed that point mutations in V2 (N160K, I165A, L165A, K169E, L175P and L179P), the CD4 binding site (CD4bs, D368R) and the CD4-induced epitope (CD4i, I420R) did not change the reactivity of 1E5 (Additional le 3: Fig. S3). Furthermore, the addition of sCD4 or CD4mc, YIR-821 (42), did not change the reactivity of 1E5 (Additional le 4: Fig. S4). These results suggested that the V2, CD4bs, CD4i and V3 epitopes are not the target for 1E5. Next, we compared the binding activity of 1E5 to chimeric Env proteins from 93TH976.17, a 1E5-reactive strain, and REJO4541.67, a 1E5-non-reactive strain (Fig. 3). Chimera A, which possesses gp120 from 93TH976.17 and gp41 from REJO4541.67, retained reactivity to 1E5. Chimeric Env containing the C1-C2 domains from 93TH976.17 (Chimera B) showed reactivity, but that containing the V3-C5 domains (Chimera C) resulted in no reactivity, indicating that the 1E5 epitope is in the C1-C2 domain of gp120. Strong binding was observed when the C1 and C2 domains originated from 93TH976.17 (Chimera F), although all of the chimeric Env constructs possessing the C1-C2 domain of 93TH976.17 (Chimeras D, E, G, H and I) showed marginal reactivity to 1E5. This suggested that the C1 and C2 domains constitute the 1E5 epitope, but that the V1 and V2 domains also affect the binding of 1E5. Consistent with this, chimera J, which contains the V1-V2 domain from REJO4541.67 in a 93TH976.17 backbone, showed good reactivity to 1E5. The data also suggested that the epitope recognized by 1E5, consisting of C1 and C2, may be different from that of A32 because the W69G mutation had no effect on binding (11,43).
The conformational epitope consisting of the C1 and C2 domains of gp120 is known as the cluster A region (10,44). The binding of antibodies to cluster A, such as A32 and C11, was reported to be enhanced by a combination of CD4mc and anti-CoRBS antibodies (8, 11,37). To investigate the possibility of 1E5 as an anti-cluster A antibody, the binding activity of 1E5 was analyzed in the presence of CD4mc and anti-CoRBS antibodies (Fig. 4). The addition of CD4mc (YIR-821) and an anti-CoRBS antibody (17b or 4E9C) (45, 46) markedly enhanced the binding activity of 1E5. This enhancement required both CD4mc and anti-CoRBS antibody, and the addition of either CD4mc or anti-CoRBS antibody alone did not affect 1E5 binding signi cantly. This enhancement effect of CD4mc and anti-CoRBS antibody on 1E5 binding was even more apparent than that on A32 binding.
To further investigate whether 1E5 binds to an epitope that overlaps with the epitope for A32, we performed a binding inhibition assay using Env from CRF02_AG-257-transfected cells as the target and biotinylated 1E5 or A32 as the probe (Additional le 5: Fig. S5). The ndings revealed that 1E5 did not compete with A32 for binding (Additional le 5: Fig. S5a), but signi cant enhancement of A32 binding was observed in the presence of 1E5 (Additional le 5: Fig. S5b). These data suggested that 1E5 binds to a part of the cluster A region that does not overlap with the A32 epitope, and further that binding of 1E5 can enhance the binding of A32.

Neutralization and ADCC activities of 1E5
The neutralization activity of 1E5-IgG1 was tested by a standard single-round neutralization assay for HIV-1 strains belonging to subtype A, CRF01_AE and CRF02_AG (Additional le 6: Fig. S6a). The neutralization activity of 1E5-IgG3 was examined by 1E5 alone and 1E5 with anti-CoRBS antibody and CD4mc against CRF02_AG-257 virus (Additional le 6: Fig. S6b). Neither the IgG1 nor the IgG3 forms of 1E5 showed any neutralization activity, similar to the other anti-cluster A antibodies (9,11,37).
The ADCC activity of 1E5 was examined against nine CRF02_AG strains that showed strong binding of 1E5 (Fig. 1b) by the detection of FcgRIIIa signaling (Fig. 5a). Both IgG1 and IgG3 forms of 1E5 showed low ADCC activity against most of the strains, although 1E5-IgG3 showed higher ADCC activity than 1E5-IgG1 against several strains, such as AG-242, AG-257 and AG-280. The combination of 1E5, both IgG1 and IgG3 forms, with 4E9C and YIR-821 increased ADCC activity against all of the strains tested except for AG-252. This lack of ADCC enhancement against AG-252 was consistent with the lack of enhancement of binding activity to AG-252 by 4E9C and YIR-821 (Additional le 7: Fig. S7). The combination effect of ADCC activity was statistically signi cant, and the combination of 1E5-IgG3 with 4E9C and YIR-821 showed a signi cantly higher level of ADCC activity than the other combinations ( Fig.  5b). This was consistent with the enhancement of binding activity of 1E5 with CD4mc and anti-CoRBS antibodies ( Fig. 4) and previous reports describing the enhancement of ADCC activity by the combination of cluster A antibody, CD4mc and anti-CoRBS antibodies (8, 11, 37). However, additional analysis of the combination effect revealed that CD4mc and sCD4 were not required for ADCC enhancement (Fig. 6a).
The addition of 4E9C alone increased ADCC activity more than the combination with CD4mc or sCD4. The lack of an effect with CD4mc and sCD4 may be due to the CD4 molecules expressed on the surface of effector cells, which possibly change the Env conformation accessible to anti-CoRBS antibodies. The level of dose-dependent ADCC activity of 1E5 in the presence of 4E9C was the same in both the presence and absence of YIR-821 (Additional le 8: Fig. S8). As reported previously (10,11), anti-CoRBS antibody alone did not mediate ADCC despite its strong recognition of the target cells (Additional le 9: Fig. S9). These results suggested that 1E5 mediates ADCC in combination with CD4 and anti-CoRBS antibody.
Enhancement of ADCC by the dual and triple combination of anti-cluster A antibodies and an anti-CoRBS antibody Although 1E5 recognized part of the cluster A region of gp120, the epitope recognized by 1E5 did not overlap with that of A32, the representative anti-cluster A antibody (Additional le 5: Fig. S5a). The binding of A32 was even enhanced in the presence of 1E5 (Additional le 5: Fig. S5b), and enhancement of ADCC activity was examined using three antibodies, 1E5, A32 and 4E9C (Fig. 6b). These antibodies mediated a two-to three-fold change in ADCC when used alone, but mediated a six to eight-fold change when used in combination. Moreover, a triple combination of antibodies showed the highest ADCC activity. Taken together, these data suggested that not only the combination of anti-cluster A antibody and anti-CoRBS antibody, but also two anti-cluster A antibodies coordinately, can mediate strong ADCC activity, the phenomenon that has not been reported previously.

Discussion
We isolated a monoclonal antibody 1E5 belonging to the anti-cluster A antibody family, which targets the cluster A region of HIV-1 gp120, from a patient infected with CRF02_AG. Genetic analysis of the immunoglobulin heavy (VH) and light (VL) chain variable domain gene segments revealed that VH was derived from VH1-69 and VL from the VK3-20 germline. A recent report of germline VH1-69-derived antibodies demonstrated the de ning features of VH1-69-utilizing antibodies against gp120, namely: the hydrophobic nature of the complementarity determining region-2 (CDRH2) regions with grand average hydropathy (GRAVY) scores ranging from 0.34 to 2.7, shorter complementarity-determining region-3 (CDRH3) with a median CDRH3 length of 14 amino acids and a higher isoelectric point (pI) with a median value of 6.15 (47). The characteristics of these antibodies also apply to 1E5, which has a high GRAVY score of 1.85 for the CDRH2 region, a short CDRH3 region involving 12 amino acids and a high pI value of 8. 59. It has been reported that the interactions between VH1-69 CDRH2 and the cavities within HIV-1 gp120 are hydrophobic (48). High CDRH2 hydrophobicity was detected as a unique and universal feature of the VH1-69-utilizing antibodies (47).
Binding analysis with chimeric envelope constructs indicated that 1E5 recognizes a conformational epitope involving the C1 and C2 regions of gp120 (Fig. 3). A previous detailed study mapped three unique clusters (A, B and C) of CD4i antibodies based on a cross competition assay (10). Usually occluded Cluster A epitopes can be exposed by conformational changes mediated by cellular CD4 binding to Env trimer during the viral entry process or co-expression of CD4 and the viral envelope on the same cell surface (14,15). A32 and C11 are the major examples of anti-cluster A antibodies possessing nonoverlapping epitopes involving the C1 and C2 domains (10,44). A32-like antibodies were found to be associated with the majority of ADCC activity in chronically-infected patients (9). A ow cytometry-based inhibition assay demonstrated that the epitope of 1E5 did not overlap with that of A32, and that 1E5 even enhanced A32 binding (Additional le 5: Fig. S5). Taken together, these data suggested that 1E5 binds to a part of the cluster A region that does not overlap with the A32 epitope, but that binding of 1E5 can enhance the accessibility of A32 binding.
Despite having high polymorphism (26) and a shorter half-life than IgG1 (34), IgG3 is the most polyfunctional IgG subclass, having the most potent Fc effector function covering the widest range (23). In the RV144 HIV vaccine trial, IgG3-mediated Fc effector functions, such as ADCC, ADCP and complement deposition, correlated with protection (24,25). Considering these facts, the IgG3 form of 1E5 was constructed and used in different assays in parallel with IgG1. When comparing the binding activity to Env proteins from a global panel of HIV-1, the IgG3 form showed signi cantly stronger binding than IgG1 (Additional le 2: Fig. S2). As shown in Figs. 5 and 6, the IgG3 form of 1E5 exhibited signi cantly higher ADCC activity than IgG1 in any combination with CD4mc and/or anti-CoRBS antibody. Factors other than the epitope, such as the angle of binding, can in uence the Fc function of an antibody (47). Increased hinge length can allow more exibility and therefore may increase the Fc-mediated effector functions of IgG3 (49). Most importantly, IgG3 has the highest a nity to FcγRIIIa (27,50). Stronger binding of FcγRIIIa at an appropriate angle can favor the IgG3 form to mediate better ADCC than IgG1.
The binding of biotinylated 1E5 with CRF02_AG Env-expressing target cells was markedly increased in the presence of CD4mc (YIR-821) and anti-CoRBS antibody (4E9C). CD4mc and anti-CoRBS antibody alone could not mediate noticeable binding enhancement (Fig. 4). A similar pattern of binding was observed with A32. This indicates that 1E5 reactivity was the same as that of A32, requiring CD4mc and anti-CoRBS antibodies for enhanced binding and the stabilization of state 2A in the presence of CRF02_AG-257 Env. Most of the previous studies analyzing anti-cluster A antibody binding used Env belonging to subtype B viruses (1,37,51). Here, we observed the same phenomena for CRF02_AG Env by means of anti-cluster A antibodies 1E5 and A32.
The 1E5 did not demonstrate any neutralization activity (Additional le 6: Fig. S6) when analyzing its ability to reduce the infectivity of the subtype-A, CRF01_AE and CRF02_AG Env pseudotype viruses. Being derived from germline VH1-69, this observation indicates the ADCC potential of 1E5, as described by previous research (47). A model was described for the sequential opening of trimeric Env that required anti-CoRBS antibodies to reveal the occluded epitope recognized by anti-cluster A antibodies (37). Engagement of CD4mc with the Phe43 cavity of the CD4 binding site causes a partial opening of trimeric Env, which enable anti-CoRBS antibodies to bind to Env but does not expose cluster A epitopes. Binding of anti-CoRBS antibodies with two gp120 subunits possibly exposes epitopes recognized by anti-cluster A antibodies resulting in state 2A stabilization (1). This recognition translated into e cient ADCC by anticluster A antibodies (37) and may be involved in the ADCC exhibited by anti-cluster A antibodies in HIV + sera (10,11). Figures 5 and 6a indicate that the highest level of ADCC was exhibited by the combination of IgG3 form 1E5 and anti-CoRBS antibody 4E9C, and that CD4mc (YIR-821) did not contribute to the enhancement of ADCC activity. The lack of requirement for CD4mc is explained by the expression of CD4 on the surface of effector cells used for the ADCC assay.
Several studies have suggested that ADCC may play a role in controlling HIV-1 infection (22,52). In the RV144 trial, ADCC was mainly found to be responsible for conferring protection (6). The anti-cluster A region is immunodominant in the case of both natural infection and vaccination. The majority of the ALVAC-HIV/AIDSVAX B/E vaccine recipients developed ADCC-mediating antibodies with C1, C2 region speci c A32-like antibodies comprising the signi cant portion (7). This study demonstrated that the binding of A32 increased in the presence of 1E5 (Additional le 5: Fig. S5b). This observation raised the possibility of enhancement of ADCC using a combination of two anti-cluster A antibodies. When used in an ADCC assay with CRF02_AG-257 Env, the combination of 1E5-IgG3 and A32 showed the highest level of fold change, even higher than their individual combinations with anti-CoRBS IgG (Fig. 6b). As anticluster A antibodies are the mediators of ADCC activity exhibited by HIV-1 + sera (9,11,53), and these antibodies can be elicited by vaccination (7), targeting this combination of anti-cluster A antibodies may be a major tool for the protection against HIV-1. Moreover, a recent study on the elicitation of anti-cluster A and anti-CoRBS antibodies observed higher and more e cient induction of anti-cluster A antibodies in immunized guinea pigs (54). Our results suggested that some of the anti-cluster A antibodies, such as 1E5, can stabilize the Env conformation at state 2a in the presence of CD4mc resulting in the enhancement of binding of the other anti-cluster A antibodies, such as A32, to exert higher ADCC activities. This nding may have implications in terms of vaccine strategies to induce appropriate combinations of antibodies for improved outcomes.

Conclusion
Our ndings indicate that the IgG3 form of anti-cluster A antibody 1E5 isolated from a CRF02_AG-infected individual can mediate higher ADCC than the IgG1 form. The combination of two anti-cluster A antibodies, together with an anti-CoRBS antibody, mediated the highest level of ADCC.
Isolation of IgG-producing single B cells by uorescence activated cell sorting A blood sample was obtained from patient KMCB2 of Kyushu Medical Center, who was infected with the CRF02_AG subtype of HIV-1. B cells were transformed by EBV and cultured at a concentration of 10 3 cells/well for 10 days, as previously reported (45). Single cells were sorted from the wells of an EBVtransformed B cell culture that scored positive for binding to Env (HIV-1 93TH966.8)-expressing cells using FACSAria II (BD Biosciences, San Jose, CA, USA). The cells were stained with anti-human IgG-BV421 and anti-human IgM-APC/Cy7 (BioLegend, San Diego, CA, USA), and IgG + IgM cells were sorted at single cell density into 4 μl/well of ice-cold 0.5× phosphate-buffered saline (PBS) containing 10 mM DTT, 8 U RNAsin® (Promega, WI, USA), 0.4 U 5′-3′ Prime RNAse Inhibitor™ (Eppendorf) as previously described (59).
Cloning and analysis of 1E5 immunoglobulin variable genes cDNA was synthesized as previously described (59)  For cloning of 1E5 immunoglobulin variable genes, the rst round of nested PCR was performed according to the methods described by Tiller et al. (59) using the same primer pairs, while second-round primers were modi ed to have a 15 base overlap at the 5 end with the speci c vectors. The second PCR primer sequences are listed in Additional le 10: Table S1.
The IgG heavy and light chain expression plasmids were constructed by recombination of the designated second PCR product with pIgGH and pKVA2, respectively (45), using the GeneArt Seamless Cloning and Assembly kit (Invitrogen). The nucleotide sequences of the immunoglobulin variable regions were aligned and compared to avoid possible PCR error. The sequences were analyzed for germline gene verification, framework and CDR mapping, quantification of percent identity to germline, CDR amino acid length and pI using IMGT vquest (http://imgt.org/IMGT_vquest/vquest). CDRH2 grand average of hydropathy (GRAVY) scores were calculated using an online tool (http://www.gravy-calculator.de/).

Construction of IgG3 heavy chain-expressing plasmid
The region from CH1 to CH3 of IgG1 heavy chain-expressing vector pIgGH was exchanged with the corresponding region of IgG3, and IgG3 heavy chain-expressing vector pIgG3H was constructed. Brie y, the CH1-CH3 region of IgG3 was ampli ed using primers, CHApa-F (AGC CTC CAC CAA GGG CCC ATC GG), IgG3-R (TCA CCA AGT GGG GTT TTG AGC TCA), CHPme-R (CTG ATC AGC GGG TTT AAA CTA TCA TTT ACC CGG AGA CAG GG) and IgG3-F (ACA AGA GAG TTG AGC TCA AAA CCC C) from cDNA, which was synthesized from the RNA of healthy donor peripheral blood mononuclear cells. The CH1-CH3 region of pIgGH was excluded by digestion with ApaI and PmeI, and the IgG3 fragments were inserted into the vector using the GeneArt Seamless Cloning and Assembly kit (Invitrogen). The variable region of 1E5 was inserted into pIgG3H to obtain 1E5-IgG3.

Production and puri cation of recombinant IgG
Recombinant IgG was produced and puri ed as previously described (45). Brie y, heavy and light chain plasmids were transfected into 293A cells using TransIT®-LT1 Transfection Reagent (Mirus Bio LLC, WI, USA), and the cells stably expressing IgG were selected with G418 (1000 μg/ml) and hygromycin (150 μg/ml). IgG1 and IgG3 proteins were puri ed using a HiTrap™ rProtein A FF Column and a HiTrap™ Protein G HP column, respectively (GE Healthcare).

Analysis of the binding activity of antibodies by ow cytometry
The binding activity of antibodies was analyzed as previously described (60). Brie y, 293T cells were transfected with a plasmid expressing both HIV-1 Env and enhanced green uorescent protein (EGFP). After 48 h of transfection, the cells were stained with primary antibody for 15 min at room temperature (RT). The cells were washed twice with PBS containing 0.2% BSA, and incubated with allophycocyaninconjugated A niPure F(ab')2 Fragment Goat Anti-Human IgG (H + L) (Jackson ImmunoResearch, West Grove, PA, USA) for 15 min at RT. Cells were xed with PBS containing 10 % formalin and analyzed using the FACSCanto II (BD Biosciences, San Jose, CA, USA). The reactivity of the antibodies was analyzed after gating the EGFP+ cells using FlowJo (TreeStar, San Carlos, CA, USA).

Neutralization assay using pseudovirus
The neutralization activity of antibodies was determined as previously described (45,61). In brief, 293T cells were transfected with pSG3ΔEnv and Env expression vector, and the supernatant after 48 h of transfection was stored at 80 °C. The median tissue culture infectious dose (TCID 50 ) of each pseudovirus was determined using TZM-bl cells. Serially diluted antibody and virus (400 TCID 50 ) were incubated for 1 h, and TZM-bl cells were added. After incubation for 48 h, the galactosidase activity was measured using galactosidase substrate (Tropix Gal-Screen substrate, Applied Biosystems) and an EnSpire Multimode Plate Reader (PerkinElmer, MA, USA). The relative light units (RLU) were compared to calculate the reduction in infectivity and 50% of the maximal inhibitory concentration (IC 50 ) was calculated using nonlinear regression.

ADCC bioassay to detect FcgRIIIa-mediated signaling
The detection of FcgRIIIa-mediated signaling was performed using a Jurkat NFAT-luc FCgRIIIa cell line (BPS Bioscience, CA, USA), as described previously (62). The target cells were 293T cells expressing Env, which were transfected with Env-expressing plasmid 48 h before the ADCC assay. The target cells were washed with PBS, treated with 0.05% trypsin, and resuspended in RPMI-1640 (Thermo Fisher Scienti c) with 4% FBS at a concentration of 3 ´ 10 6 cells/ml. Then, 25 µl of the target cells were incubated with antibodies for 15 min, after which 25 µl of effector Jurkat cells were added at a ratio of 1:1 and were cocultured for 6 h. The cells were lysed and the re y luciferase activity was determined with a luciferase assay kit (Promega) and EnSpire ® Multimode Plate Reader. The co-culture in the absence of antibody provided background (antibody-independent) luciferase activity. The RLU obtained in the presence of antibody were divided by the background level to calculate the fold change.

Consent for publication
Not applicable.

Availability of data and materials
Not applicable.
Additional le 8: Fig. S8. IgG3 form of 1E5 shows stronger ADCC in the presence of anti-CoRBS antibody. The effect of anti-CoRBS antibody and CD4mc on ADCC activity, which was measured by signaling through FcgRIIIa, was analyzed using IgG3 forms of 1E5. HEK 293T cells transfected with plasmid expressing HIV-1 Env (nine binding assay-positive Env proteins from a CRF02_AG panel and one subtype A) were incubated with the indicated antibodies and co-cultivated with the ADCC indicator cell line   93TH976.17, which is strongly bound by 1E5. The regions from REJO4541.67 and 93TH976.17 are shown in light-blue and red, respectively. The W69G mutation is shown in yellow. (b) The reactivity of Env recombinants to NHG, VRC01 and 1E5 was determined by ow cytometry analysis using cells expressing each Env recombinant. The reactivity was detected by APC-conjugated anti-human IgG secondary Ab, and the mean uorescence intensity (MFI) of APC is shown. Figure 4 Anti-CoRBS antibody and CD4mc increased binding of 1E5 to CRF02_AG Env. HEK 293T cells were transfected with plasmid expressing both EGFP and CRF02_AG-257 Env. At 48 h post-transfection, cells were stained with biotinylated test IgG (1E5-IgG1 and 1E5-IgG3, 10 µg/ml) alone or with anti-CoRBS antibody (17B or 4E9C, 5 µg/ml) in the presence or absence of CD4mc YIR-821 (20 µM). Biotinylated A32 was used as a control. Then, cells were uorescently labeled with APC-conjugated streptavidin.

Figure 5
1E5-IgG3 induces higher ADCC than 1E5-IgG1 in the presence of anti-CoRBS antibody and CD4mc. The ability of IgG1 and IgG3 forms of 1E5 to mediate signaling through FcýRIIIa when bound to cells expressing Env is shown. HEK 293T cells transfected with HIV-1 Env (nine binding assay positive Env proteins from a CRF02_AG panel) were incubated with the indicated antibodies and the ADCC indicator cell line expressing FcýRIIIa. Simultaneous binding of antigen and FcýRIIIa results in activation of the NFAT transcription factor, which induces luciferase in indicator cells. IE5 was used (10 µg/ml) alone or in combination with 4E9C (5 µg/ml) and CD4mc YIR-821 (20 µM). The fold change was calculated by dividing the luminescence units in the presence of Ab with those in the absence of Ab. NHG was used as a control. (a) ADCC activity against each target cell is shown. Experiments were performed in triplicate, and the means standard errors of the means are shown. (b) ADCC activity against target cells expressing Env from nine CRF02_AG strains was plotted, and statistically analyzed. The means ± standard errors of the means are shown. Statistical significance was tested using a paired t test (*, P <0.05; **, P <0.01; ***, P <0.001; ****, P <0.0001; ns, non-significant).

Figure 6
Combination of anti-cluster A antibodies and anti-CoRBS antibody enhanced ADCC activity. The ability of 1E5 to mediate ADCC activity was analyzed by measuring the signal through FcýRIIIa. HEK 293T cells expressing AG-257 Env were incubated with the indicated antibodies, YIR-821 and sCD4, and cocultivated with the ADCC indicator cell line expressing FcýRIIIa. Fold change was calculated by dividing the luminescence units in the presence of Ab by those in the absence of Ab. NHG was used as a control.