An engineered antibody-lectin conjugate targeting the HIV glycan shield protects humanized mice against HIV challenge

Abstract

Enveloped viruses responsible for global health pandemics often display a glycan shield on their surface envelope glycoproteins. In HIV, the glycan shield is formed by clusters of high-mannose glycans and plays essential roles in viral fitness and immune evasion. A few mannose-binding lectins potently inactivate HIV but have not been fully exploited due to poor pharmacokinetics and short serum half-lives. To address this, we engineered an antibody-lectin conjugate comprising the anti-HIV lectin griffithsin (GRFT) to the Fc region of human IgG1, with the aim of extending its serum half-life and augmenting anti-HIV activity by inducing immune effector responses. Engineered mGRFT-Fc produced in bacteria exhibited picomolar anti-HIV activity and an extended serum half-life, and mGRFT-Fc produced in mammalian cells (mGRFT-Fc_glyc) elicited immune effector responses. In HIV-infected CD34⁺-humanized mice, both GRFT and mGRFT-Fc_glyc effectively suppressed viral loads for up to 8 weeks after a single dose. Significantly, mGRFT-Fc_glyc prevented HIV infection by neutralizing HIV and provided sustained protection from break-through infections via Fc-mediated immune effector responses, exhibiting a dual mode of protection. This study demonstrates the successful engineering of a lectin-based biologic and provides early evidence that a glycan-targeting agent alone can confer protection from viral infection in vivo.

Graphical abstract

Kumariya, Bewley, and colleagues demonstrate that an Fc-lectin conjugate that specifically targets the HIV glycan shield on surface envelope glycoproteins has improved pharmacokinetic properties and potently inhibits HIV infection in a humanized mouse model via neutralization of viral particles together with Fc-mediated elimination of HIV-infected cells through an immune effector response.

Introduction

HIV is an enveloped virus whose trimeric surface envelope glycoproteins gp120/gp41 are heavily glycosylated to form a protective “glycan shield.” Composed of high-mannose oligosaccharides (isomers of oligomannose-5 to oligomannose-9) biosynthesized by the host, the glycan shield effectively helps HIV to escape the host immune response by shielding key protein epitopes from the immune system. This immune evasion strategy has posed an enormous challenge in developing effective HIV vaccines. Moreover, the glycan positions or clusters on the envelope trimer are conserved among different variants as they play important roles in viral fitness.^1,2 While a few classes of patient-derived, broadly neutralizing anti-HIV antibodies (bnAbs) that recognize conserved glycan epitopes on the HIV trimer have been identified and characterized, these are rare.² Thus, the glycan shield offers an attractive target against which to develop novel and effective anti-HIV therapeutics.³

Lectins are ubiquitous proteins that exhibit a distinct ability to selectively identify and bind to specific carbohydrate structures.⁴ Many non-human lectins have been shown to significantly inhibit HIV entry through multivalent interactions with HIV’s glycan shield. Examples include cyanovirin-N (CV-N), banana lectin (BanLec), actinomycete lectin (actinohivin), and red algae-derived griffithsin (GRFT).⁵ Each of these lectins recognizes multivalent clusters of high-mannose glycans, including those present on the surface of HIV. In addition, they tend to be non-toxic to healthy human cells that lack dense displays of high-mannose glycans.^6,7 However, high-mannose glycans are present on HIV-infected human cells.⁸ Among the most potent antiviral lectins, GRFT has shown superior HIV pseudovirus neutralizing activity with picomolar half maximal inhibitory concentration (IC₅₀) values and minimal toxicity in in vivo safety studies,^9,10 effectively making it an anti-HIV warhead. Despite its promise, GRFT, like other lectins, exhibits poor pharmacokinetics with a half-life of only a few hours.¹¹ To overcome a similar pharmacokinetic challenge, Hamorsky et al. engineered an antibody-lectin fusion comprising human immunoglobulin (Ig)G1 Fc and avaren (an actinohivin lectin variant) and explored its potential as a biotherapeutic targeting cell-free HIV particles as well as productively HIV-infected cells.¹² The engineered avaren-Fc exhibited potent anti-HIV activity, elicited Fc-mediated killing of HIV-infected cells, and showed an extended serum half-life in rats and rhesus macaques. However, the efficacy of glycan shield-targeting anti-HIV agents has never been demonstrated in an in vivo animal model.

Animal models used to determine protection by broadly neutralizing antibodies (bnAbs) and HIV inhibitors, and to monitor disease progression include macaques and humanized mice. Although non-human primates are the model of choice, such animal studies can be prohibitively expensive and require specialized facilities. In this investigation, we utilized mice engrafted with human CD34⁺ hematopoietic stem cells for HIV-infection studies. These humanized mice develop diverse human immune cell lineages and serve as a validated platform for researching human diseases.¹³ This humanized mouse model has several advantages compared with non-human primate models; they include the small size of the animals, which permits easier handling and lower quantities of drugs or biologics needed for testing, and significantly lower costs that may allow for testing an increased number of treatment protocols and/or against a larger number of animals, the combination of which will result in data that are more statistically reliable.¹⁴

Here, to test the validity and efficacy of a glycan-targeting anti-HIV antibody-lectin therapeutic, we engineered an mGRFT-Fc conjugate by fusing monomeric GRFT with the Fc region of IgG1 translationally. We assessed the characteristics required to designate it as an effective pre-exposure prophylactic agent, including extended plasma half-life, potential to activate Fc-mediated effector function, and in vivo efficacy to protect humanized mice against HIV infection.

Results

Engineering and production of mGRFT-Fc in bacteria

A select group of mannose-specific lectins potently inhibit HIV entry and infection.⁵ The most potent antiviral lectins are capable of multivalent and high-avidity binding to clusters of closely spaced high-mannose glycans.⁹ Such displays are mostly limited to the surfaces of viral envelope glycoproteins. GRFT has been shown by multiple laboratories to inhibit HIV with picomolar concentrations.^7,15,16 GRFT is a homodimer that displays a unique 3-dimensional fold wherein a beta-hairpin of each monomer is domain swapped with its partner giving it remarkable stability.¹⁷ Moreover, each domain of GRFT contains three mannose binding sites that engage two equivalents of Man-9. Together with its homodimeric form, and with each mannose binding face oriented opposite to one another, GRFT can engage four or more high-mannose oligosaccharides.¹⁰

In this work, we sought to improve the plasma half-life of GRFT, reduce immunogenicity that may be associated with a non-human protein, and activate Fc-mediated effector function by genetically fusing the Fc region of human IgG1 and a monomeric GRFT (mGRFT) variant previously reported by Moulaei et al.¹⁸ (Figure 1A). In our design, engineered mGRFT-Fc contained only the Fc region (C_H2 and C_H3) of human IgG1, while the C_H1 and V_H domains were replaced with mGRFT. We selected mGRFT first because it would preclude the need for each GRFT subdomain to fold into a domain-swapped dimer while extended from the Fc region, a requirement that was proving challenging to produce in Escherichia coli in initial studies. Second, we reasoned that the bivalent nature of the native lectin would be restored if two monomers were displayed on neighboring arms of the Fc, thereby restoring neutralizing activity comparable to the homodimeric form in contrast to mGRFT alone that is inactive. After preliminary studies successfully demonstrated recombinant expression of the first construct in E. coli, we further optimized the sequence by introducing mutations predicted to increase stability and folding and eliminate eukaryotic sites of Asn-linked glycosylation on the GRFT domain (Figure S1). In particular we introduced the Met340Leu and Asn346Ser mutations in the Fc domain as these so-called LS mutations are known to improve the serum half-life of IgG1.²¹ In preparation for eventual expression in mammalian cells, each of the four eukaryotic glycosylation sites in the GRFT domain (consensus sequence N-X-S/T) was mutated to Gln. The mGRFT-Fc construct was produced in Escherichia coli and native GRFT was produced as described previously.¹⁰ GRFT was purified by Ni-NTA affinity chromatography and mGRFT-Fc by protein A chromatography. Both the proteins were then subjected to size-exclusion chromatography and exhibited high purity on non-reducing SDS-PAGE gels (Figure 1B). Both GRFT and mGRFT-Fc eluted as dimers from a calibrated size-exclusion column (Figure 1C) and confirm that the IgG1 Fc of mGRFT-Fc is dimeric.

Figure 1.

Engineering and characterization of mGRFT-Fc

(A) Structure of GRFT (PDB: 2GTY) and model of mGRFT-Fc.^19,20 Cartoon rendering created with PyMOL with mGRFT or GRFT shown in yellow, the Fc region of IgG1 in blue, and the linker region (GGGSGGG) in green. (B) Coomassie-stained SDS-PAGE of purified GRFT and mGRFT-Fc in non-reducing conditions. Monomeric molecular weights of GRFT and mGRFT-Fc are 14.8 kDa and 39 kDa, respectively. (C) Size-exclusion chromatography of GRFT and mGRFT-Fc on a HiLoad 16/600 Superdex 200 pg column. Theoretical dimeric mwt’s of GRFT and mGRFT-Fc are 29.6 kDa and 78 kDa, respectively. The arrows indicate the elution volumes and molecular weights (kDa) of protein standards used for calibration.

mGRFT-Fc retains the binding and neutralizing properties of GRFT

To determine whether mGRFT-Fc recognizes HIV envelope protein gp120 in a glycan-dependent manner, we used surface plasmon resonance (SPR) to determine whether engineered mGRFT-Fc binds to the BG505 DS-SOSIP.664 (gp140 trimer) with similar affinity as GRFT binding to the gp140 trimer. The association rate of GRFT with gp140 trimer was 100-fold higher than that of mGRFT-Fc (1.42 × 10⁷ vs. 1.35 × 10⁵ M⁻¹s⁻¹, respectively) as was the dissociation rate (1.36 × 10⁻² vs. 5.68 × 10⁻⁴ s⁻¹, respectively), resulting in very similar equilibrium dissociation constants (K_Ds) for GRFT (0.96 nM) and mGRFT-Fc (4.22 nM, Figures 2A and 2B).

Figure 2.

Binding and HIV neutralization properties of GRFT and mGRFT-Fc

(A and B) SPR sensorgrams showing association and dissociation curves for interactions between GRFT (A) or mGRFT-Fc (B) and gp140 trimer immobilized on a CM5 chip using covalent amine coupling. Serial dilutions of 10 different concentrations ranging from 2 μM (red) to 0.25 nM (violet) of GRFT or mGRFT-Fc were run over gp140 trimer. The black lines show the theoretical curves for all runs. (C) HIV-1 neutralization curves for GRFT and mGRFT-Fc. The curves show assays performed using HIV-1 pseudovirus strain YU-2. Single-round infectivity assays were performed as described in materials and methods, using envelope pseudotyped viral particles and TZM-bl cells as targets. Data are shown as mean ± SD of three independent experiments. (D and E) Heatmap and scatterplot showing the HIV neutralization breadth and potential of GRFT and mGRFT-Fc against multiclade pseudoviruses comprising 15 clinical HIV strains. Standard broadly neutralizing antibodies (bnAbs) VRC07-523-LS and VRC01 were used as positive controls. Other glycan-targeting antibodies viz., 10–1074, 2G12, BG18, PG16, PG9, PGT121, PGT128, and PGT135, were used to compare their neutralization potency and breadth with GRFT and mGRFT-Fc. A non-HIV strain (SIVmac251.30.SG3; shown as black dots in E) was used as a control for which standard bnAbs were not effective, but GRFT and mGRFT-Fc were as they target highly conserved mannose-rich glycans found on gp120 on both HIV and SIV. Data are shown as mean of two independent experiments each with two technical replicates. Horizontal bars (shown in E) represent the median of IC₅₀ values of neutralizable pseudoviruses, and strains with IC₅₀ values of neutralization >330 nM or <0.06 nM are shown on upper and lower black dotted line, respectively. Statistical analysis was done using two-tailed Mann-Whitney U test between mGRFT-Fc and other antibodies. ns, not significant, ^∗p < 0.05, ^∗∗∗p < 0.001, and ^∗∗∗∗p < 0.0001.

We next tested whether gp140-binding of mGRFT-Fc translates into HIV neutralization activity by performing single-round neutralization assays with mGRFT-Fc in parallel with GRFT. Both proteins potently inhibited infection with the pseudotyped HIV-1 strain YU2, with IC₅₀ values of 2.43 pM (0.072 ng/mL) and 28.6 pM (2.23 ng/mL) for GRFT and mGRFT-Fc, respectively (Figure 2C). Although the potency of mGRFT-Fc against this strain is ∼12-fold less than that of GRFT, an IC₅₀ in picomolar range is remarkable and demonstrates mGRFT-Fc’s potential as an effective anti-HIV agent. To examine the breadth of each construct, we tested each against 15 HIV clinical isolates belonging to four different clades (A, B, C, and AD), and a simian immunodeficiency virus (non-HIV) strain. The selected panel of 15 HIV clinical isolates spans multiple HIV-1 clades and exhibits varying sensitivity to the well-known bnAbs VRC07-523-LS and VRC01 (a clinical candidate). Both GRFT and mGRFT-Fc were found to be equally or more effective against nearly all the strains compared with bnAbs VRC07-523-LS and VRC01 and other glycan-targeting bnAbs including 10–1074, 2G12, BG18, PG16, PG9, PGT121, PGT128, and PGT135 (Figures 2D and 2E). Although the median IC₅₀ of HIV neutralization by mGRFT-Fc against the panel showed no significant difference compared with that by GRFT and most of the antibodies, it was significantly superior to that of VRC01, 2G12, and PG16 antibodies (Figure 2E). Also, both GRFT and mGRFT-Fc exhibited activity against all 15 HIV clinical strains in the nanomolar range, underlining their potency and HIV neutralization breadth (Figure 2D). This wide breadth can be attributed to the fact that GRFT targets high-mannose glycans present on the surface of gp120, which are highly conserved among different HIV strains. Finally, we tested whether mGRFT-Fc could block cell-to-cell transmission of SHIV_AD8-EO in rhesus peripheral blood mononuclear cells (PBMCs). As seen in Figure S2, mGRFT-Fc inhibited HIV transmission with an IC₅₀ value of 12.7 nM, showing comparable potency to the positive control antibody 10–1074 (IC₅₀: 7.3 nM).²²

mGRFT-Fc has improved plasma half-life over GRFT

Assessment of toxicity, off-target effects, and in vivo pharmacokinetics are necessary prerequisites for preclinical evaluation of any therapeutic candidate. We performed both therapeutic- and high-dose pharmacokinetic and toxicity studies in Sprague-Dawley rats by administering GRFT or mGRFT-Fc intravenously at 250 nmol/kg or 50 mg/kg, respectively. Both GRFT and mGRFT-Fc were well tolerated in rats at both doses, and neither adverse side effects nor toxicity were observed in terms of behavior and clinical observations (Table S1) and body weight and temperature at 250 nmol/kg (Figure S3) as well as at 50 mg/kg (Figure S4) during the study. The studies further established that mGRFT-Fc showed an improved pharmacokinetic profile compared with GRFT at the same dose with elimination half-lives in rat sera of 7.5 h and 30.4 h for GRFT and mGRFT-Fc, respectively (Figure 3A). Last, for both GRFT and mGRFT-Fc, serum concentrations greatly exceeding their respective IC₅₀ values (ca. 380- and 40-fold, respectively) remained through the end of the 14-day study (Figure S5).

Figure 3.

Pharmacokinetic assessment of GRFT and mGRFT-Fc

(A) Pharmacokinetic profiles of GRFT and mGRFT-Fc in rat sera. Serum concentrations of GRFT and mGRFT-Fc in rats injected intravenously with GRFT or mGRFT-Fc (250 nmol/kg) were assessed by ELISA at 1, 4, and 8h, and 1, 2, 4, 6, 10, and 14 days post-administration. Data are means ± SD of two technical replicates from two biological replicates. (B and C) SPR sensorgrams showing association and dissociation curves for interaction of mGRFT-Fc with the human FcRn receptor (immobilized on a CM5 chip via amine coupling) at pH 6.0 (B) and pH 7.4 (C). Serial dilutions of 10 different concentrations ranging from 1 μM (red) to 1.95 nM (violet) of mGRFT-Fc were run over human FcRn at pH 6 or pH 7.4. The black lines show the theoretical curves for all runs.

Neonatal Fc receptors (FcRn) bind to IgG antibodies in a pH-dependent manner, binding with high affinity at pH 6.0 (endosomal pH) and lower affinity at pH 7.4 (serum pH). This leads to antibody recycling through reticuloendothelial cells and an increased plasma half-life.²¹ To investigate mGRFT-Fc’s potential to undergo endothelial recycling we performed SPR assays to assess the interaction of mGRFT-Fc with FcRn receptors at pH 6.0 and 7.4. Our SPR data showed efficient binding to the FcRn receptor at pH 6.0 (K_D: 31.9 nM), while binding was significantly reduced at pH 7.4 (Figures 3B and 3C). The binding efficiency of mGRFT-Fc to FcRn receptor at pH 6.0 is very similar to that of the standard anti-HIV antibody VRC01 (K_D: 78.2 nM; Figure S6). Together with the FcRn receptor binding results, these data suggest that the enhanced plasma half-life of mGRFT-Fc is due to inclusion of the Fc region that allows for FcRn-mediated recycling.

Glycosylated mGRFT-Fc triggers an immune effector response

The studies described above were conducted with mGRFT-Fc produced in E. coli and therefore lacks post-translational modifications, including glycosylation of the IgG1 Fc region at Asn 297 (Kabat numbering). The presence of the complex glycan at Asn297 is essential for Fc-receptor-mediated interactions on immune cells and in turn, for immune cell activation and eliciting cellular- and complement-mediated immune responses. Thus, we next produced glycosylated mGRFT-Fc (mGRFT-Fc_glyc) in Expi293 cells, an established expression system for producing glycosylated antibodies.²³ As seen in Figures 4A and 4B, we were able to produce and purify mGRFT-Fc_glyc to high purity as a dimer. Glycosylation of mGRFT-Fc_glyc at the predicted Asn residue (Asn 209) in the Fc of this construct was confirmed by performing PNGase treatment and mass spectrometry (Figure S7) and was compared with mGRFT-Fc produced in bacteria (Figure S8). HIV neutralization and breadth of mGRFT-Fc_glyc activity were tested in the same 15 HIV clinical isolates and a simian immunodeficiency virus (non-HIV) strain used previously (Figure S9) and in parallel with mGRFT-Fc. We also performed HIV neutralization assays on GRFT, mGRFT-Fc, and mGRFT-Fc_glyc against SHIV_AD8-EO strain (Figure S10), which was used in the rhesus PBMC cell-to-cell transmission assay (Figure S2). All three proteins neutralized SHIV_AD8-EO strain with nanomolar IC₅₀ values and were more potent than standard bnAbs VRC07-523-LS and VRC01. Neutralization potencies and coverage of the panel of viruses by mGRFT-Fc_glyc was similar to that of mGRFT-Fc except for a few strains where mGRFT-Fc was 3- to 5-fold more potent than mGRFT-Fc_glyc. The differences in potency between mGRFT-Fc and mGRFT-Fc_glyc in a few strains may arise due to the glycosylation status of the Fc domain in mGRFT-Fc. Structural analysis shows that Fc glycosylation limits the dynamic nature of the C_H2 domains of IgG1 and restricts range of movement of the domains.²⁴ Being non-glycosylated, mGRFT-Fc may thus be more flexible in binding to the gp120 and hence neutralize HIV more efficiently than mGRFT-Fc_glyc in a few of the strains. This indicates that glycosylation of the Fc region had little or no impact on the HIV-neutralizing activity. Additionally, GRFT, mGRFT-Fc, and mGRFT-Fc_glyc were inactive against murine leukemia virus (MLV) in the pseudovirus neutralization assay (Figure S10), confirming their HIV specificity.

Figure 4.

Immune effector functions of mGRFT-Fc_glyc

(A) Coomassie-stained SDS-PAGE of purified mGRFT-Fc_glyc in non-reducing conditions. Monomeric molecular weight of mGRFT-Fc_glyc is 40.5 kDa. (B) Size-exclusion chromatography profile of mGRFT-Fc_glyc on a HiLoad 16/600 Superdex 200 pg column. Dimeric molecular weight of mGRFT-Fc_glyc is 81 kDa. The arrows indicate the peaks of the molecular weight of protein standards in PBS in kDa. (C and D) SPR sensograms showing association and dissociation curves for interaction of mGRFT-Fc_glyc with FcγRI (C) and FcγRIIIa-V176 (D) captured on anti-His chip (anti-His chip was prepared by immobilizing anti-His antibody on CM5 chip via amine coupling). Serial dilutions of 10 different concentrations ranging from 500 nM (red) to 0.98 nM (violet) and eight different concentrations from 2000 nM (red) to15.6 nM (violet) of mGRFT-Fc_glyc were run over FcγRI and FcγRIIIa-V176, respectively. The black lines show the theoretical curves for all runs. (E) FcγRIIIa activation by mGRFT-Fc_glyc against CHO-ZA and CHO-ZA-gp160_SC in reporter-cell-based assay using Jurkat T-CD16A cells expressing FcγRIIIa receptors and luciferase as effector cells (ECs). Data are shown as mean ± SD of two independent experiments. (F) Complement-mediated killing of CHO-ZA-gp160_SC on activation of complement cascade by mGRFT-Fc_glyc in a concentration-dependent manner. CHO-ZA cells were used as a negative control of target cells. Data are shown as mean ± SD of three independent experiments.

We investigated whether mGRFT-Fc or mGRFT-Fc_glyc led to auto- or poly-reactivity by performing anti-nuclear antibody (ANA) staining on human epithelial cells (HEp-2), and anti-cardiolipin and anti-ANA ELISAs (Figures S11A–S11C). Neither construct showed detectable autoreactivity against cardiolipin (Figure S11B), and only minimal auto- or poly-reactivity was observed and only at the highest concentration tested (500 nM) in HEp-2 cell ANA staining (Figure S11A) and anti-ANA ELISA (Figure S11C). None of the proteins were cytotoxic toward PBMCs (Figure S11D).

We performed SPR assays to assess binding of mGRFT-Fc_glyc with Fcγ receptors that are present on immune cells and necessary for their activation and eliciting effector cell responses. mGRFT-Fc_glyc binds FcγRI and FcγRIIIa receptors with respective affinities of 19 nM and 166 nM (Figures 4C and 4D), similar to its binding to a reference bnAb VRC01 (K_D: 8.5 nM; Figure S12). In contrast, mGRFT-Fc produced in E. coli (which is non-glycosylated) failed to bind either of the two Fc receptors (Figure S13). Because cell-mediated immune effector responses such as antibody-dependent cellular phagocytosis (ADCP) and antibody-dependent cellular cytotoxicity (ADCC) are mediated by interactions with the FcγRI (ADCP) and the FcγRIIIa receptors, the observed mGRFT-Fc_glyc binding profiles suggest that the glycosylated conjugate should be capable of eliciting both types of cell-mediated immune responses.

To test immune activation, we performed a cell-based ADCC assay that depends on Fcγ-receptor binding by mGRFT-Fc_glyc to elicit an immune effector response via the FcγRIIIa receptor (also known as CD16a, Figure 4E). Natural killer (NK) cells and macrophages constitutively express CD16a receptors on their surface that are responsible for eliciting cellular-mediated immune effector responses upon activation via these receptors.²⁵ For the ADCC assay, we used Jurkat T-CD16A cells (that constitutively express cell surface CD16a) as effector cells, CHO-ZA-gp160_SC (Chinese hamster ovary [CHO] cells transfected to display gp160 on their surface) as the positive target cells, and CHO-ZA cells as negative control target cells. The Jurkat T-CD16A cell line contains an integrated luciferase gene under the control of Nuclear Factor of Activated T cells Response Element (NFAT-RE). Upon CD16a receptor engagement with target-bound antibodies (in this case CHO-ZA-gp160_SC-bound mGRFT-Fc_glyc) luciferase gene expression is activated through NFAT-RE. Little to no luminescence is expected when mGRFT-Fc_glyc is incubated either with CHO-ZA cells and Jurkat T-CD16A cells and/or Jurkat T-CD16A cells alone. Shown in Figure 4E, mGRFT-Fc_glyc activated Jurkat T-CD16A cells in the presence of CHO-ZA-gp160_SC cells with a half maximal effective concentration (EC₅₀) of 1.07 nM (Figure 4E). We observed non-specific activation by mGRFT-Fc_glyc when incubated with Jurkat T-CD16A cells alone or with CHO-ZA cells and Jurkat T-CD16A cells with EC₅₀ values of 2.79 and 2.73 nM, respectively (Figure 4E). Because the EC₅₀ of activation in the presence of CHO-ZA-gp160_SC cells is ∼3-fold less than that of the negative controls, this suggests that more than half of the activation is due to the engagement of mGRFT-Fc_glyc with CHO-ZA-gp160_SC cells (Figure S14). We propose that the background, non-specific activation is due to the presence of aberrantly glycosylated surface proteins on the Jurkat T-CD16A cancer cell line^26,27 that leads to self-activation (Figure S14). We performed the same assay using mGRFT-Fc or GRFT in place of mGRFT-Fc_glyc. Both mGRFT-Fc and GRFT failed to activate effector cells in the absence or presence of CHO-ZA or CHO-ZA-gp160_SC cells (Figures S15A and S15B), proving the indispensability of Fc glycosylation in CD16a receptor activation. Further, to rule out the involvement of non-CD16a cell surface receptors in the activation of effector cells, we used Jurkat-Lucia NFAT-cells as effector cells (EC^−CD16a) that express luciferase under NFAT promoter but do not express CD16a receptors. EC^−CD16a cells were only activated in the presence of PMA and ionomycin (used as a positive controls) and not by mGRFT-Fc_glyc in the absence or presence of CHO-ZA or CHO-ZA-gp160_SC cells (Figure S15C), further supporting the fact that the ADCC response is due to the specific interaction between CD16a receptors and the glycosylated Fc domain.

Apart from Fcγ receptors, antibodies also interact with the complement proteins via their Fc region and can activate the complement cascade leading to target cell lysis. As an orthogonal validation of mGRFT-Fc_glyc function, we performed complement-dependent cytotoxicity (CDC) assays to determine if target-bound mGRFT-Fc_glyc also can activate the complement cascade. We again used CHO-ZA-gp160_SC as the target cells and CHO-ZA cells as the negative control target cells. Human complement serum was used as a source of complement proteins that on activation by target-bound antibodies form the membrane attack complex. CHO-ZA or CHO-ZA-gp160_SC cells were incubated with human complement serum at different concentrations of mGRFT-Fc_glyc and without mGRFT-Fc_glyc as a negative control. Monitoring of live and dead cells by flow cytometry revealed that CHO-ZA-gp160_SC target cells incubated with human complement serum and mGRFT-Fc_glyc underwent cell lysis in a dose-dependent manner, while there was no effect on the viability of the negative control cells (Figures 4F and S16) nor on CHO-ZA-gp160_SC cells incubated with complement serum in the absence of mGRFT-Fc_glyc. This shows that upon binding to CHO-ZA-gp160_SC cells, mGRFT-Fc_glyc activates the complement cascade and induces CDC. We also performed this assay with GRFT and mGRFT-Fc to rule out any non-specific effect exerted by GRFT domain of the conjugate. Neither GRFT nor mGRFT-Fc showed any CDC activity (Figure S16).

In vivo HIV infection in humanized mice model

A primary goal in engineering an antibody-lectin conjugate was to test its utility as an anti-HIV pre-exposure prophylactic. To this end, we performed preclinical testing of mGRFT-Fc in an in vivo mouse model for HIV. We employed CD34⁺ NSG humanized mice (hu-CD34) for this study. The hu-CD34 mouse model was generated to mimic a human immune system through adoptive transfer of CD34⁺ stem cells,¹³ and is a proven in vivo platform for assessing the efficacy of biologics that influence the human immune system. We tested the in vivo efficacy of GRFT, mGRFT-Fc, and mGRFT-Fc_glyc as pre-exposure prophylactics by administering them intravenously into hu-CD34 mice 6 h before challenging them with HIV-1 BAL strain intraperitoneally (Figure 5A). The progression of HIV-1 infection in mice was monitored by determining plasma viral load (Figures 5B–5E) and the number of CD4⁺ T cells (Figure S17) and CD8⁺ T cells (Figure S18) in the peripheral blood. In the control group, one mouse died within 2 weeks of challenge and the remaining two were infected within 3 weeks after challenge (Figure 5B). In the GRFT group, two mice were protected during the study while one was infected after 5 weeks of challenge (Figure 5C). This late onset of infection can be due to the release of new HIV-1 particles from the CD4⁺ T cells infected by HIV-1 that escaped neutralization by GRFT. Because GRFT cannot kill HIV-1 infected cells, this resulted in the multiplication and release of HIV-1 particles from the infected cells and increase in the plasma viral load in later weeks. In the mGRFT-Fc group, all mice became infected within 4 weeks of challenge showing the inability of mGRFT-Fc to provide protection (Figure 5D). This inefficiency may be attributed to two factors. First, mGRFT-Fc is less potent than GRFT (as shown in in vitro HIV neutralization data, Figures 2C and 2D). Second, since mGRFT-Fc is non-glycosylated, it cannot elicit Fc-mediated immune effector functions to clear both cell-free HIV-1 particles and HIV-1 infected cells because glycosylation of the Fc region is indispensable for these functions. In the mGRFT-Fc_glyc-treated group, one mouse escaped early neutralization while two mice remained protected through the duration of the 8-week study (Figure 5E). This suggests that mGRFT-Fc_glyc can provide long-lasting protection with a single dose as we did not observe any late onset of infection in the mGRFT-Fc_glyc group in contrast to the GRFT group.

Figure 5.

Anti-HIV activity of GRFT, mGRFT-Fc, and mGRFT-Fc_glycin vivo in a humanized mice model

(A) Schematic representation of the experimental design. CD34⁺ NSG mice were divided into four groups of control, GRFT, mGRFT-Fc, and mGRFT-Fc_glyc with three mice in each group. All mice were treated intravenously with control, GRFT, mGRFT-Fc, or mGRFT-Fc_glyc 6 h before HIV-1 BAL challenge. Blood was collected at day −1 and every week thereafter for RT-PCR and flow cytometry studies to monitor the progression of HIV-1 infection. Plasma HIV-1 RNA level in copies/mL at day −1 and every week in control (B), GRFT (C), mGRFT-Fc (D), and mGRFT-Fc_glyc (E). The dotted line represents the limit of detection, which was 400 copies/mL. For samples where HIV-1 RNA was not detected, a value of 100 copies/mL was assigned for depiction. (F) Kaplan-Meier curve showing the percentage of mice that remained protected from HIV-1 BAL infection in control, GRFT, mGRFT-Fc, and mGRFT-Fc_glyc groups. Statistical analysis was performed using Log-rank (Mantel-Cox) test.

In the control group, the number of CD4⁺ T cells decreased after challenge, while the number of CD8⁺ T cells either remained constant or showed an increasing trend, hallmarks of HIV infection (Figures S17 and S18). In GRFT-, mGRFT-Fc-, and mGRFT-Fc_glyc-treated groups, the number of CD4⁺ T cells remained constant, whereas the number of CD8⁺ T cells showed a decreasing trend (Figures S17 and S18). The stable CD4⁺ T cell levels in infected mice from the mGRFT-Fc- and mGRFT-Fc_glyc-treated groups could potentially indicate a slower progression of HIV infection in these individuals.

Overall, these results showed GRFT and mGRFT-Fc_glyc were potent in protecting two-thirds of the mice against HIV-1 infection, while mGRFT-Fc could prevent HIV-1 infection for only 4 weeks (Figure 5F). This validates the requirement for glycosylation at Asn209 of mGRFT-Fc_glyc corresponding to Asn297 of the Fc region of IgG1 in conferring sustained protection against HIV-1 infection by engaging immune effector cells. mGRFT-Fc_glyc was found to be equally potent as GRFT in vivo. However, mGRFT-Fc_glyc is superior in that it shows an ability to kill HIV-1-infected cells by inducing Fc-mediated immune effector responses that GRFT cannot.

Discussion

The HIV envelope glycoprotein trimer gp120/gp41 (Env) is the only patient-derived immunogen to which bnAbs effective against HIV have been isolated, making it the key target in HIV vaccine efforts. Underlining its importance, HIV Env is also the target of small molecule and peptide therapeutics. Numerous approaches and Env immunogens are being tested as HIV vaccines^28,29,30; however, this remains an enormous challenge due to the unique properties of HIV Env. In addition to the shielding by high-mannose glycans of underlying protein epitopes, the HIV Env proteins exist in distinct metastable states, and when fully triggered undergo massive conformational changes leading to gp120 shedding.³¹ Together these dynamics make the trimer a moving and temporal target for immune recognition.³² Lectins that bind to high-mannose glycans through multivalent, high-avidity interactions with HIV Env have proven to be durable and potent inhibitors of HIV infection because they are able to recognize the glycan shield despite the plasticity of the trimer and can exhibit extremely slow off rates that favor inhibition of dynamic targets.^10,33

Among high-mannose binding lectins, GRFT has been shown repeatedly to be potent and non-toxic in cell-based assays and PK studies in animals.^7,9,10,11,15 Moreover, its ability to retain neutralizing activity despite genetic drift and changes in Env glycosylation makes it all the more attractive as an anti-HIV candidate.³⁴ Arguably, the one shortcoming of GRFT is its short in vivo half-life¹¹ accounting for its clinical development as a topical microbicide to prevent sexual transmission of HIV^35,36,37 and as a nasal spray for inhibition of respiratory viruses,³⁸ applications that obviate the need for extended in vivo half-lives. For these reasons, we viewed GRFT as an ideal candidate for engineering and answering the question of whether an anti-HIV warhead whose only function is to target high-mannose glycan clusters can be co-opted for in vivo protection against HIV challenge. This led to the development of the mGRFT-Fc fusion proteins described here. Gratifyingly, the in vitro and in vivo data indicate that the design principles and predicted inhibitory and immune-activating function of the antibody-lectin conjugate are sound.³⁹ In addition to an extended in vivo half-life from FcRn-mediated serum-to-endosome recycling, mGRFT-Fc_glyc protected hu-CD34 mice from HIV challenge and elicited cellular immunity and extended protection beyond that of lectin alone (Figure 6). Prior to this work, only one study has reported the development of an anti-HIV antibody-lectin conjugate (AvFc) that also had an extended half-life and activated immune effector functions in cell-based assays.¹² That work left open the critical question of whether a glycan-targeting lectin-Fc alone would be effective in an in vivo HIV challenge study using animal infection models.

Figure 6.

Mechanisms of protection by lectin versus lectin-Fc in humanized mice

(A) In non-treated humanized mice (CD34⁺ mice), HIV infects CD4⁺ T cells, leading to HIV replication and release from the infected T cells. (B) In lectin-treated CD34⁺ mice, lectin neutralizes HIV, protecting CD4⁺ T cells from infection. However, HIV that evades lectin neutralization can still infect CD4⁺ T cells, resulting in HIV replication and release from the infected cells. (C) In lectin-Fc-treated CD34⁺ mice, lectin-Fc not only neutralizes HIV, protecting against infection, but also eliminates HIV-infected cells (from HIV that escaped lectin-Fc neutralization) through immune effector responses such as ADCC, ADCP, and CDC. ADCC, antibody-dependent cellular cytotoxicity; ADCP, antibody-dependent cellular phagocytosis; CDC, complement-dependent cytotoxicity.

In this study, the in vivo HIV challenge results demonstrated that GRFT and mGRFT-Fc_glyc successfully protected two-thirds of the hu-CD34 mice from HIV infection (Figures 5C and 5E). Notably, in the GRFT-treated group, one mouse became infected at week 6. Although the number of animals in this pilot study is limited, this delayed onset of infection is intriguing and could be attributed either to viral escape or to GRFT’s inability to eliminate HIV-infected cells. In contrast, in the mGRFT-Fc_glyc-treated group, a single mouse became infected at week 3, while all mice in the mGRFT-Fc-treated group showed infection at week 3 or 4, similar to the control group. This timing suggests that the infection in these groups might be due to insufficient serum concentrations of mGRFT-Fc or mGRFT-Fc_glyc, rather than viral escape. To enhance HIV prevention and therapy, strategies to mitigate viral escape should be developed.^40,41,42 Moreover, since mGRFT-Fc_glyc-treated mice did not exhibit late-onset infection, lectin-antibody conjugates may offer a viable solution in such cases.

Furthermore, mGRFT-Fc demonstrated lower efficacy than GRFT both in vitro (Figure 2C) and in vivo (Figures 5C and 5D), likely due to the incorporation of the IgG1 Fc domain and the conversion of GRFT from dimeric to monomeric form in the mGRFT-Fc conjugate. Despite its shorter half-life compared with mGRFT-Fc, GRFT outperformed it in providing in vivo protection against HIV. This can be attributed to GRFT’s sub-nanomolar neutralization potency that effectively balances its in vivo efficacy, making it ultimately more protective than the non-glycosylated mGRFT-Fc. Notably, this study demonstrates the stability of GRFT in vivo for an extended time while maintaining protective efficacy against HIV challenge. While both GRFT and mGRFT-Fc_glyc exhibit potent antiviral activity, GRFT functions solely as an HIV-neutralizing agent, whereas mGRFT-Fc_glyc offers dual functionality, neutralizing HIV and eliminating infected cells via Fc-mediated immune effector responses, as demonstrated in our in vitro and in vivo studies. The use of hu-CD34 mice (reconstituted with human hematopoietic stem cells) in this in vivo study supports this conclusion. Human hematopoietic stem cells give rise to NK cells and macrophages of human origin in hu-CD34 mice,⁴³ which in turn enable Fc-mediated immune effector responses upon encountering antibodies or Fc-fusion proteins bound to target cells. Consequently, mGRFT-Fc_glyc protects hu-CD34 mice against HIV-1 by combining HIV neutralization with Fc-mediated immune effector functions, thereby offering a dual mode of protection. Furthermore, the inability of mGRFT-Fc (which has a non-glycosylated Fc domain incapable of inducing immune effector responses) to protect against HIV-1 in the hu-CD34 mice further supports the protection mechanism conferred by mGRFT-Fc_glyc. Thus, it is reasonable to propose that unlike GRFT, mGRFT-Fc_glyc provides sterilizing protection by clearing HIV and HIV-infected cells rather than merely delaying HIV infection.

Other studies on the development of anti-HIV Fc-fusion proteins, for example, sCD4-Fc,⁴⁴ CD24-Fc,⁴⁵ and IBP-C24⁴⁶ have shown enhancements in the anti-HIV activity of the fusion proteins resulting from leveraging the Fc-mediated effector functions. Fc-mediated effector functions account for 25%–45% of the total antiviral activity of anti-HIV IgG1,^47,48 and bnAbs with enhanced CDC activity are more efficient in restricting spreading HIV infection than wild-type bnAbs or bnAbs with reduced CDC activity.⁴⁹ More broadly, antibody-lectin and Fc-lectin conjugates are showing utility and efficacy in cancer biology and medicine in systems where glycan profiles are cell-type specific and can be targeted.^50,51,52

This study has two primary limitations. First, due to prohibitive costs and limited resources, the in vivo challenge studies were conducted with only three mice per group as a pilot study. Further studies are planned with larger cohorts and extended duration to thoroughly characterize protective immune responses and their timing in relation to HIV progression. Second, we could not correlate the protective efficacy of GRFT and mGRFT-Fc with their serum concentrations in the hu-CD34 mice because we did not perform the pharmacokinetics during the in vivo infection study due to the sample limitations and restrictions on the use of HIV-infected sera. Although pharmacokinetic studies were carried out in rats, the results could not be directly extrapolated to the hu-CD34 mice. To address these limitations, the next set of in vivo infection studies will include pharmacokinetic analysis of proteins in the sera along with the viral load and CD4⁺ and CD8⁺ T cell analyses.

In closing, this study represents a primary example of an antibody-lectin conjugate preventing HIV infection and providing sustained protection, solely by targeting the HIV glycan shield. This opens avenues for future studies and development of additional antiviral antibody-lectin therapeutics and improvements in mGRFT-Fc constructs that may show further increases in potency or plasma half-lives and mucosal distribution, the latter being a key factor in HIV prevention.^53,54 Because mGRFT-Fc_glyc shows the potential to kill HIV-infected cells by engaging immune effector cells, it is worth exploring its post-exposure potential, studies that are under way in our laboratory.

Materials and methods

Study design

To expand the utility of GRFT as an efficient antiviral therapeutic, we engineered an antibody-lectin conjugate (mGRFT-Fc) comprising monomeric GRFT fused to the Fc domain of IgG1. We produced an engineered mGRFT-Fc in E. coli and mammalian expression systems and performed subsequent purification. We evaluated the binding of mGRFT-Fc to the BG505 DS-SOSIP.664 gp140 trimer and assessed its HIV neutralization potential and breadth compared with the parent GRFT. Upon confirming the anti-HIV potential of mGRFT-Fc, we conducted toxicity and pharmacokinetics studies in rats to assess the adverse effects and plasma half-life of GRFT and mGRFT-Fc. We used a mammalian expression system to produce mGRFT-Fc_glyc (glycosylated Fc domain) and assessed the ability of mGRFT-Fc_glyc to induce immune effector functions. Further, to test the in vivo protective efficacy of GRFT and engineered mGRFT-Fc and mGRFT-Fc_glyc, we performed 8-week in vivo HIV infection studies in humanized mice. Infection progression in all the groups was monitored weekly through collected sera to measure viral load (RT-qPCR) and number of CD4⁺ and CD8⁺ T cell counts (flow cytometry). The study aimed to track protection against HIV-1 infection or resurgence of infection from HIV-infected cells.

Engineering, production, and purification of GRFT, mGRFT-Fc, and mGRFT-Fc_glyc

Construct GRFT_pET14b encoding GRFT with N-terminal 6×His tag was synthesized previously.¹⁰ Constructs mGRFT-Fc_pET14b and SP-mGRFT-Fc_pcDNA3.1 encoding mGRFT-Fc (E. coli expression) and mGRFT-Fc_glyc (mammalian expression), respectively, were synthesized by GenScript (Piscataway, NJ). The following mutations were introduced in the amino acid sequence of mGRFT-Fc for the ease of expression (Figure S1): (1) GS insertion in GRFT sequence, (2) four potential glycosylation (Asn-X-Ser/Thr) sites in GRFT sequence were mutated (Asn to Gln), and (3) LS mutation was done in Fc region (at the residues corresponding to Met340Leu and Asn346Ser of IgG1). For mammalian expression in Expi293 cells (Invitrogen, Waltham, MA), mGRFT-Fc gene was cloned with mouse Ig heavy chain signal peptide (SP) sequence at its N terminus for the efficient glycosylation and secretion of mGRFT-Fc_glyc.

The GRFT, mGRFT-Fc, and mGRFT-Fc_glyc were produced in E. coli BL21(DE3) cells (NEB, Ipswich, MA), E. coli Shuffle T7 Express cells (NEB), and Expi293 cells, respectively, according to manufacturer’s instructions. GRFT was purified by Ni-NTA chromatography using HisTrap column (Cytiva, Marlborough, MA), while mGRFT-Fc and mGRFT-Fc_glyc were purified by protein A chromatography using HiTrap protein A column (Cytiva). All the proteins were further purified by size-exclusion chromatography using HiLoad 16/600 Superdex 200 pg column (Cytiva) and their purity was checked by SDS-PAGE under non-reducing conditions.

Surface plasmon resonance

SPR measurements were performed using a Biacore T100 instrument (Cytiva) at 25°C. The gp140 trimer (BG505.DS-SOSIP.664) or the FcRn receptor was covalently immobilized to a carboxymethylated dextran matrix (CM5) S series sensor chip (Cytiva) using amine coupling at pH 5.0. A separate flow cell was capped with ethanolamine using amine coupling as a control for non-specific binding. The gp140 trimer was kindly provided by Peter Kwong (Vaccine Research Center, NIAID, NIH) and FcRn receptor was purchased from Acrobiosystems (Newark, DE). All binding analyses were done in running buffer (1× PBS at pH 7.4 or pH 6.0). All samples (GRFT or mGRFT-Fc) were prepared in filtered and degassed running buffer and their 2-fold serial dilutions ranging from 2 μM to 0.25 nM (GRFT or mGRFT-Fc) and 1 μM to 1.95 nM (mGRFT-Fc) were used as analytes for gp140 trimer and FcRn binding, respectively. All sample concentrations were injected at 30 μL/min with 60 s association and 180 s (gp140 trimer binding) or 200 s (FcRn binding) dissociation. A duplicate injection of one concentration and a buffer injection (blank) were used for assessing chip surface performance and double referencing, respectively. Surface regeneration was done by flowing a solution of 2 M mannose in 1× PBS at pH 7.4 (for gp140 trimer binding) or 50 mM Tris HCl pH 8.0 (for FcRn binding) at 30 μL/min for 30 s, followed by running buffer for reconditioning.

For binding assays of mGRFT-Fc_glyc with Fcγ receptors (FcγRI and FcγRIIIa-V176), an anti-His tag monoclonal antibody (Thermo Fisher Scientific, Waltham, MA) was immobilized to all four flow cells of a CM5 chip using amine coupling at pH 5.0 to approximately 4,000 response units (RU). His-tagged FcγRI and FcγRIIIa-V176 receptors (Acrobiosystems) were captured on flow cells 2 and 4 as ligands, respectively, to approximately 50 RU. Flow cells 1 and 3 were treated similarly but without capture of any ligand and used as a blank control. The mGRFT-Fc_glyc (ranged from 500 nM to 0.98 nM [for FcγRI] and 2,000 nM–15.6 nM [for FcγRIIIa-V176] in 1× PBS plus 0.005% P20) was injected at 30 μL/min with 2-min association and 10-min dissociation. A duplicate injection of one concentration and a buffer injection (blank) were used for assessing chip surface performance and double referencing, respectively. For surface regeneration, 10 mM glycine at pH 1.5 was injected at 30 μL/min for 30 s, followed by running buffer for reconditioning.

The binding kinetics, k_on (1/Ms), k_off (1/s), and K_D (M) were calculated from global fittings using a bivalent analyte model for GRFT and mGRFT-Fc binding to gp140 trimer and using 1:1 kinetics binding model for the rest of the binding experiments on the BIAevaluation software (Cytiva).

HIV single-round neutralization assays

Single-round HIV neutralization assays were performed as described previously.⁵⁵ Briefly, for 50% inhibitory concentration (IC₅₀) analysis, 40 μL of serial 5-fold dilutions of each GRFT or mGRFT-Fc were prepared in cDMEM (complete DMEM containing 10% fetal bovine serum and 1 mg/mL penicillin/streptomycin) in individual wells of a 96-well plate. Viral particles pseudotyped with HIV strain YU2 (200 TCID₅₀) were added to each well, followed by 20 μL of TZM-bl target cells (5 × 10⁵ cells/mL). GRFT, mGRFT-Fc, or virus controls were included, where 40 μL cDMEM was added in GRFT and mGRFT-Fc controls in place of GRFT and mGRFT-Fc, respectively, and 80 μL cDMEM was added in virus control in place of virus and GRFT or mGRFT-Fc. Following 72-h incubation at 37°C, HIV infection was measured by luciferase activity using Bright-Glo luciferase assay system (Promega, Madison, WI) according to the manufacturer’s instructions. Data were normalized to controls and plotted using Prism 10 (GraphPad Software, San Diego, CA) to obtain IC₅₀ values. All experiments were performed in triplicate for IC₅₀ analysis.

For IC₅₀ analyses against 15 HIV clinical strains, neutralization was measured using single-round-of-infection HIV-1 Env-pseudoviruses and TZM-bl target cells, as described previously.⁵⁶ Neutralization curves were fit by nonlinear regression using a 5-parameter hill slope equation. The IC₅₀ was reported as the concentrations required to inhibit infection by 50%. GRFT, mGRFT-Fc, and mGRFT-Fc_glyc proteins were tested at a starting concentration of 5 μM and the bnAbs, VRC01, and VRC07-523-LS and other glycan-targeting antibodies (obtained from the Vaccine Research Center, NIAID, NIH), were tested at a starting concentration of 50 μg/mL. For comparison and data representation, protein and antibody concentrations are shown in nM in tables.

PBMC cell transmission assay

The in vitro neutralizing activity of mGRFT-Fc was assessed using a 21-day peripheral blood mononuclear cell (PBMC) replication assay⁵⁷ employing rhesus PBMCs and SIV/HIV chimeric SHIV_AD8-EO.⁵⁸ Briefly, mGRFT-Fc was serially diluted (3-fold, starting at 6,000 nM), and an aliquot of each dilution was incubated with 100 TCID₅₀ of the SHIV_AD8-EO challenge virus at room temperature for 1 h and then added to activated rhesus PBMCs (1 × 10⁵ cells per well) in quadruplicate. Infected cultures were maintained for 2 weeks, and virus replication was monitored by ³²P-RT assay.⁵⁹ Any infectious SHIV_AD8-EO replicating during the 2 weeks of incubation would be amplified to levels detectable by the assay. Neutralization antibody titers were calculated by the method of Reed and Muench.⁶⁰ The 10–1074 mAb⁶¹ and IgG purified from an HIV-1 uninfected chimpanzee (Kurt IgG)⁶² were analyzed together with mGRFT-Fc as positive and negative controls, respectively.

Pharmacokinetics

Toxicity studies in Sprague-Dawley rats were performed at BIOQUAL Inc. (Rockville, MD). The animals’ care was in accordance with institutional guidelines. Sprague-Dawley rats were divided into three groups: GRFT, mGRFT-Fc, and control. The GRFT and mGRFT-Fc groups included three male and three female rats, and the control group included two males and two females. On day 0 of the study, groups were administered intravenously with vehicle (sterile PBS supplemented with 10 mM mannose), GRFT (250 nmol/kg), or mGRFT-Fc (250 nmol/kg). Purified preparations of GRFT and mGRFT-Fc were checked for the presence of endotoxins and the levels were well below the recommended limit (5 EU/kg/h) for preclinical research in animals. Body weights were recorded pretest, on dosing day, and each day until completion of the study. Clinical observations took place every day. Animals were examined for any altered clinical signs, including gross motor and behavioral activity and observable changes in appearance. Body temperatures were measured rectally at regular intervals. In-life bleeds were taken at −21 days pre-dose, and 1-, 4-, and 8-h, 1-, 2-, 4-, 6-, and 10-days post-dose and from two animals in each group (one male and one female) at each time point to avoid excess blood withdrawal. A terminal bleed was taken at 14 days. The collected blood was centrifuged to yield sera samples to measure the plasma concentration of GRFT and mGRFT-Fc using quantitative-ELISA.

For ELISA, maxisorp 96-well plates (Thermo Fisher Scientific) were coated with gp140 trimer (100 ng per well) in coating buffer (50 mM carbonate buffer pH 9.6) overnight at 4°C. After washing twice with PBS-T (1× PBS with 0.05% Tween 20), plates were blocked with blocking buffer (5% skim milk in 1× PBS) for 2 h at room temperature, followed by PBS-T wash. Plasma from rats was diluted in PBS-T and was incubated on the plates alongside known GRFT and mGRFT-Fc concentrations as standards for 90 min at 37°C. Plates were washed with PBS-T and incubated with rabbit anti-GRFT primary antibody (1:5,000 dilution in PBS-T) for 1 h at room temperature. After washing with PBS-T, plates were incubated with goat anti-rabbit horseradish peroxidase (HRP)-conjugated secondary antibody (1:5,000 dilution in PBS; Abcam, Cambridge, MA) for 1 h at room temperature. Plates were washed and incubated with TMB substrate (Thermo Fisher Scientific) for 30 min. Reaction was stopped by adding stop solution (Thermo Fisher Scientific) and absorbance was measured at 450 nm using Synergy H1 microplate reader (BioTek, Winooski, VT). ELISA was performed two independent times for each sample, and the data reported as mean ± SD. Half-lives of GRFT and mGRFT-Fc were determined by analyzing the data in Prism 10 using exponential two-phase decay model.

Mass spectrometry for intact protein analysis

Mass spectrometry was performed on a Waters Xevo G2 qTOF mass spectrometer operating in the positive ESI ion mode with a resolution of 25K and an ESI capillary voltage of 2.8 kV. The desolvation gas was nitrogen and the gas temperature was 280°C. Protein samples were dissolved in liquid chromatography (LC)/mass spectrometry (MS)-grade pure water and using a 10-μL loop the protein sample was injected with a Rheodyne Titan MX valve into the LC solvent stream. The LC solvents were solvent A (0.1% formic acid in LC/MS water) and solvent B (0.2% formic acid and 0.1% TFA in MS grade acetonitrile). The LC column was a ProSwift RP-4H 1 × 50 mm HPLC column. The LC pump was a Waters Acquity Class I system. The initial LC setting was 100% solvent A and held for 3 min. Solvent B was then stepped to a value of 60% and held there until all protein signal was exhausted. The raw ESI signal was deconvoluted using the MaxEnt I algorithm.

Auto- or poly-reactivity assays

Both mGRFT-Fc and mGRFT-Fc_glyc were tested for auto- or poly-reactivity using three platforms: HEp-2 cell ANA antibodies staining (ZEUS Scientific, Branchburg, NJ), anti-cardiolipin ELISA (ZEUS Scientific), and anti-ANA ELISA (ZEUS Scientific) following the manufacturer’s instructions. For HEp-2 cell ANA antibodies staining, mGRFT-Fc or mGRFT-Fc_glyc was diluted to 400, 200, and 100 nM in Zorba-NS diluent. Positive and negative controls were included on each slide. For anti-cardiolipin ELISA and anti-ANA ELISA, mGRFT-Fc or mGRFT-Fc_glyc was diluted to 500 nM in sample diluent and tested in 5-fold serial dilutions (500–0.032 nM). Positive and negative controls were included on each plate.

Cytotoxicity assay

We tested GRFT, mGRFT-Fc, or mGRFT-Fc_glyc for cytotoxicity against PBMCs. Briefly, 25 μL of GRFT, mGRFT-Fc, or mGRFT-Fc_glyc (2-fold serially diluted from an initial concentration of 500 nM) was added into each well. One well was kept as a protein control, where 25 μL RPMI1640 complete medium was added in place of GRFT, mGRFT-Fc, or mGRFT-Fc_glyc. Then, 75 μL of PBMCs resuspended in RPMI1640 complete medium were added to give a final density of 1.0 × 10⁵ cells/mL. Cells were incubated at 37°C for 24 h and then cell viability was assessed using CyQUANT MTT cell viability assay kit (Thermo Fisher Scientific) following the manufacturer’s instructions.

Antibody-dependent cellular cytotoxicity assay

ADCC activity of mGRFT-Fc_glyc, mGRFT-Fc, and GRFT was assessed using an ADCC reporter bioassay kit (Promega) following the manufacturer’s instructions. The kit employs Jurkat T-NFAT-Luc2-CD16A (Jurkat T-CD16A) cells as effector cells (ECs) that express luciferase under the NFAT promoter and CD16a receptors. As a negative control of effector cells, we used Jurkat-Lucia NFAT-cells (EC^−CD16a; InvivoGen, San Diego, CA) that express luciferase under NFAT promoter but do not express CD16a receptors. We used a CHO-based cell line stably expressing HIV-1 gp160 (CHO-ZA-gp160_SC)⁶³ as target cells and CHO-ZA cells as a negative control of target cells. Target or negative control cells were plated on 96-well plates at a density of 2.5 × 10⁴ cells in 25 μL RPMI1640 complete medium per well. Fifty microliters of mGRFT-Fc_glyc, mGRFT-Fc, or GRFT (5-fold serially diluted from an initial concentration of 50 nM) was added into each well. One well was kept as a protein control, where 50 μL RPMI1640 complete medium was added in place of mGRFT-Fc_glyc, mGRFT-Fc, or GRFT. Then, ECs or EC^−CD16a were added at a density of 1.5 × 10⁵ cells in 25 μL RPMI1640 complete medium per well for a target:effector cell ratio of 1:6. The assay was developed after 6-h incubation at 37°C according to the manufacturer’s instructions. In this assay, incubation of proteins with ECs or EC^−CD16a in the absence of target cells or in the presence of negative control cells (CHO-ZA) was used to determine the non-specific activation by proteins. Ionomycin (5-fold serially diluted from an initial concentration of 2 nM) in the presence of PMA (81 pM) was used as positive control in the assays performed with EC^−CD16a cells. The fold-change in induction of luminescence was calculated based on the ratio of luminescence produced in wells with or without proteins and was plotted against the protein concentrations to determine the EC₅₀ value for elicitation of ADCC.

Complement-dependent cytotoxicity (CDC) assay

To assess the CDC activity of mGRFT-Fc_glyc, CDC assay was performed using pooled human complement serum (Innovative Research, Plymouth, MN) as a source of complement proteins. Like ADCC assay, we used CHO-ZA-gp160_SC and CHO-ZA as target cells and negative control of target cells in this assay. Target or negative control cells were seeded into clear U-bottom 96-well plates at a density of 7.5 × 10⁴ cells in 50 μL Opti-CHO medium per well. Fifty microliters of mGRFT-Fc_glyc (5-fold serially diluted from initial concentration of 500 nM) was added into each well. One well was kept as an mGRFT-Fc_glyc control, where 50 μL Opti-CHO medium was added in place of mGRFT-Fc_glyc. Then, 50 μL of pooled human complement serum (diluted 1:4 with Opti-CHO medium) was added to each well. The assay was incubated for 5 h at 37°C. Then, propidium iodide (PI) was added to all the wells to a final concentration of 1 μg/mL. Cells were then analyzed by flow cytometry. Flow cytometry data were analyzed by plotting side scatter-area (SSC-A) vs. phycoerythrin (PE) channel in which propidium iodide was detected and the cells were gated to count the number of live and dead cells to determine the percentage of dead cells. Percent dead cells (i.e., PI-positive cells) were plotted against mGRFT-Fc_glyc concentrations to determine the EC₅₀ value for exhibiting CDC activity by mGRFT-Fc_glyc.

In vivo HIV-infection study in humanized mice

Treatment and HIV challenge

In vivo HIV-infection study was performed at BIOQUAL Inc. CD34⁺ NSG humanized mice (hu-CD34) used for this study were purchased from the Jackson Laboratory (Bar Harbor, ME). The animals’ care was in accordance with institutional guidelines and the experiments were performed in accordance with all national or local guidelines and regulations. The HIV-1 BAL strain used for HIV infection in mice was obtained from the Institute of Human Virology, University of Maryland, Baltimore, and the stock was further propagated in PBMCs. The hu-CD34 mice were divided into four groups (n = 3): control, GRFT, mGRFT-Fc, and mGRFT-Fc_glyc. The GRFT, mGRFT-Fc, and mGRFT-Fc_glyc groups were intravenously administered with 250 nmol/kg of GRFT, mGRFT-Fc, and mGRFT-Fc_glyc, respectively, and the control group was intravenously administered with equal volume of 1× PBS supplemented with 10 mM mannose (Figure 5A). All administrations were done 6 h before HIV-1 BAL challenge. All groups received HIV-1 challenge intraperitoneally at the dose of 20,000 median infectious dose (ID₅₀). The progression of HIV infection and viral load in mice was monitored from HIV-1 RNA levels in plasma using qRT-PCR, and CD4⁺ and CD8⁺ T cell counts using flow cytometry. Thus, in-life blood collections were made on day −1 and every week until week 7. Terminal blood collection was completed on week 8 for all mice. Purified preparations of GRFT, mGRFT-Fc (expressed in E. coli), and mGRFT-Fc_glyc (expressed in mammalian cells) were checked for endotoxin levels. All samples were well below the recommended limit (5 EU/kg/h) for preclinical research in animals.

HIV-1 qRT-PCR assay

For measuring viral load, qRT-PCR assay was performed using mouse plasma collected at different time points. Briefly, HIV-1 RNA was isolated from mouse plasma using the Qiagen MinElute Virus Spin Kit (Qiagen, Hilden, Germany). The qRT-PCR was performed using gag-specific primers SK101 HIV Fwd (5′-GCTATGTCAGTTCCCCTTGGTTCTC-3′) and SK143 HIV Rev (5′-AGTGGGGGGACATCAAGCAGCCATGCAAAT-3′) and SK102 HIV Probe (56-FAM/GAGACCATC/ZEN/AATGAGGAAGCTGCAGAATGGGAT/3IABkFQ) to determine viral loads. The qRT-PCR was performed using Sensifast Probe Lo-ROX One-Step PCR kit (Bioline, Memphis, TN) on thermocycler (7500 real-time PCR system, Applied Biosystems, Waltham, MA). As a positive control for the amplification reaction, plasmid DNA with an insert containing the gag gene was used as a template in every qRT-PCR run. The number of copies of HIV-1 RNA was calculated by extrapolation from the standard curve, which gave a practical range of 50 to 5 × 10⁸ HIV-1 RNA copies/mL. The standard curve was diluted from plasmid DNA containing a region of the gag gene covered by the primers and probe from a range of 10⁷ copies/μL to 1 copy/μL. All samples were done in triplicate. The limit of quantification of the qRT-PCR was determined to be 400 HIV-1 RNA copies/mL and all values below this limit were assigned 100 HIV-1 RNA copies/mL for plotting graph in Prism 10.

Flow cytometry

The whole blood samples were washed then stained with the following antibodies: anti-human-CD3-APC (clone UCHT1, BioLegend, San Diego, CA), anti-human-CD4-FITC (clone OKT4, BioLegend), and anti-human-CD8-eFlour 450 (clone OKT8, Invitrogen). The samples were lysed with 1× FACS lysis buffer (BD Biosciences, Franklin Lakes, NJ) after staining with FBS stain buffer (BD Biosciences). The cells were washed, resuspended in FBS stain buffer, and used for acquisition. FACS data were acquired using Cytek Aurora full-spectrum flow cytometer (Cytek Bioscience, Fremont, CA). Cells were first gated based on their size and granularity using side scatter-area (SSC-A) and forward scatter-area (FSC-A) to isolate the lymphocyte population. To ensure single-cell events, doublets and aggregates were excluded by gating on forward scatter-height (FSC-H) vs. FSC-A. The lymphocyte population was then analyzed for CD3 expression, using the APC channel to identify CD3⁺ T cells. Subsequently, the CD3⁺ T cells were further gated to differentiate subsets based on their CD4 and CD8 expression. CD4⁺ T cells were detected using FITC channel, while CD8⁺ T cells were identified using e-Fluor 450 channel. Data analysis was done on FlowJo v10 (Ashland, OR) to count the cells. Percent CD3⁺CD4⁺ T cells or CD3⁺CD8⁺ T cells was calculated by the formula: [(CD3⁺CD4⁺ T cell or CD3⁺CD8⁺ T cell counts/CD3⁺ T cell counts) × 100] and plotted against weeks post-challenge in Prism 10.

Statistical analysis

For statistical analysis and generating graphs, GraphPad Prism software (version 10) was used. Data are shown as mean ± SD unless noted otherwise. The exact numbers of biological or technical replicates (number of mice or rats or samples) are indicated in the figure legends. Statistical analysis on HIV neutralization breadth data was done using two-tailed Mann-Whitney U test between mGRFT-Fc and other antibodies. For generation of survival curves, Kaplan-Meier estimator integrated in GraphPad Prism 10 software was used and analyzed using Log rank (Mantel-Cox) test. All flow cytometry data were analyzed using FlowJo software version 10.10.0 (BD Life Sciences). No unexpected or unusually high safety hazards were encountered during the whole study. All raw, individual data are presented in Table S2.

Data availability

Data generated from each experiment or study reported in this manuscript can be found in the main text or the supplemental information.

Acknowledgments

This work was generously supported by the Office of AIDS Research, NIH (1ZIADK075187, C.A.B.) and the NIH Intramural Research Program (NIDDK, 1ZIADK032103, C.A.B. and NIAID, 1ZIAAI005095, N.A.D.-R.). We thank P. Kwong (Vaccine Research Center, NIAID, NIH) for providing gp140 trimer, S.D. Schmidt (Vaccine Research Center, NIAID, NIH), and J. Lloyd (LBC, NIDDK) for technical assistance in HIV neutralization studies and mass spectrometry, respectively; W. Li (University of Pittsburgh) for providing CHO-ZA and CHO-ZA-gp160_SC cell lines; Y. Nishimura and O. Donau (Laboratory of Molecular Microbiology, NIAID, NIH) for the PBMC HIV cell-to-cell transmission studies; J. Reece (Advanced Light Microscopy & Image Analysis Core, NIDDK) for technical assistance in fluorescence microscopy; and V. Klein (LBC, NIDDK), B. Chawla (BIOQUAL Inc.), and S. Kar (BIOQUAL Inc.) for helpful discussions. Illustrations in Figure 6 were generated using BioRender.

Author contributions

R.K., J.S., S.L., and C.A.B. conceptualized the study. R.K., J.S., S.L., S.O., G.Z., N.A.D.-R., and C.A.B. developed the methodology. R.K., J.S., S.L., and C.A.B. led the investigation. C.A.B. acquired the funding. C.A.B. supervised the studies. R.K. wrote the original draft of the manuscript. All authors contributed to the review and editing of the manuscript.

Declaration of interests

The authors declare no competing financial interests.