A Bispecific Antibody Composed of a Nonneutralizing Antibody to the gp41 Immunodominant Region and an Anti-CD89 Antibody Directs Broad Human Immunodeficiency Virus Destruction by Neutrophils

Abstract

In addition to the direct neutralization of virus, there is a broader potential for antibody-mediated inhibition of human immunodeficiency virus (HIV) by targeting HIV to effector cells. We demonstrate here that a bispecific antibody incorporating a broadly reactive anti-gp41 antibody, F240, and an anti-IgA receptor (CD89) antibody is effective at directing neutrophils to destroy HIV. Not only are neutrophils the predominant type of white blood cells and very efficient at mediating cell cytotoxicity, they are relatively resistant to infection with HIV. Therefore, they represent a significant weapon against infection if they can be directed and armed to destroy HIV and infected cells.

Despite a sustained antibody response to env antigens, numerous studies have clearly demonstrated the inability of the humoral immune system to develop functionally effective neutralizing antibodies during human immunodeficiency virus (HIV) infection (5, 25, 30). However, several studies have shown that specific human monoclonal antibodies protect against infection in nonhuman primate models of infection (1, 9, 10, 19-21), and the administration of a combination of 2F5, 4E10, and 2G12 antibodies induced a delay in viral rebound during structured interruption of antiretroviral therapy (36).

There remains a need to develop novel treatments for infected individuals who may no longer respond to or who have significant toxicity from antiretroviral therapy and to prevent HIV transmission. To this end, bispecific antibody constructs may be used to target HIV and infected cells to effector cells for destruction, resulting in greater control and prevention of infection. Pathogens undergo opsonization with serum antibody and phagocytosis after binding to neutrophils. Neutrophils also mediate antibody-dependent cellular cytotoxicity through Fcγ and Fcα receptors. Neutrophils have been shown to have ingested viral particles inside phagosomes (29), yet relatively little is known of the potential of neutrophil-mediated destruction of HIV. Here, we demonstrate that a bispecific antibody construct incorporating the variable regions of the anti-gp41 antibody F240 and the anti-CD89 (immunoglobulin A [IgA] receptor) antibody 14A8 promotes destruction of HIV type 1 (HIV-1) by neutrophils.

14A8 is a fully human monoclonal antibody specific for human CD89, binding outside the IgA binding site and generated in Medarex-Mouse human Ig transgenic mice (11). The nonneutralizing human monoclonal antibody F240 recognizes an extremely conserved extracellular epitope (residues 598 to 604) on gp41 within cluster I. F240 reacts with primary isolates from all clades of HIV-1 (4), similar to other cluster I antibodies (2, 27). The majority of clade A, B, and C isolates in the HIV-1 sequence database were identical to the peptide used to map F240 (amino acids 592 to 606); similar results were seen with clade D, with the exception of a consistent L602H mutation.

A bispecific construct of the 14A8 and F240 monoclonal antibodies was constructed by chemical conjugation using Sulfo-SMCC cross-linker. F(ab′)₂ fragments of each antibody were prepared using immobilized pepsin (Pierce). Anti-gp41 antibody F240 was reduced to F(ab) fragments using immobilized Tris(2-carboxyethyl)phosphine hydrochloride (Pierce) and conjugated with Sulfo-SMCC (Pierce). 14A8 F(ab′)₂ was reduced with 2-mercaptoethylamine-HCl to generate reactive sulfhydryl groups. An equimolar concentration of F240 F(ab) with SMCC cross-linker was added to an equimolar concentration of 14A8 F(ab). Final recovery of bispecific or cross-linked F(ab) fragments was 20%, with the majority of the bispecific antibody running with an apparent molecular mass of 100 kDa [50 kDa was contributed by each F(ab)] upon sodium dodecyl sulfate-polyacrylamide gel electrophoresis analysis under nonreducing conditions and, under reducing conditions, as a doublet at 25 kDa (data not shown).

The reactivity of the F240 component of the bispecific antibody was determined by enzyme-linked immunosorbent assay (ELISA) using vaccinia virus-expressed HIV gp160 (vT240; clade C 96ZM651.8). BSC-1 cells infected for 24 h with vT240 were dried onto the plates, which were blocked prior to the addition of samples. Bound bispecific antibody was detected using horseradish peroxidase-conjugated goat anti-human kappa and o-phenylenediamine substrate. The immunoreactivity of the 14A8 component of the bispecific antibody with CD89 on neutrophils was determined by flow cytometry. Neutrophils were obtained using dextran sedimentation of the red blood cell/granulocyte pellet following Ficoll-Hypaque sedimentation of peripheral blood collected from healthy donors after informed consent and as approved by the Institutional Review Board at Beth Israel Deaconess Medical Center. The neutrophils were preincubated with normal mouse serum to block nonspecific binding prior to the addition of bispecific antibody, which was detected using fluorescein isothiocyanate-conjugated goat anti-human kappa chain antibody. The reactivity of the bispecific antibody with HIV was retained by the bispecific antibody (Fig. 1A), as was immunoreactivity with neutrophils (Fig. 1B).

FIG. 1.

Immunoreactivity of bispecific antibody with HIV and neutrophils. (A) Reactivity with clade C gp160 (as measured by optical density values at 490 nm) of serial dilutions of bispecific antibody compared to that of serial dilutions of F240 antibody. (B) Bispecific antibody was reacted with neutrophils to determine the reactivity of the 14A8 component of the bispecific antibody with CD89. The relative fluorescence was compared to that of14A8 antibody alone.

While these results demonstrate that the antibody fragments are immunoreactive, it was also ascertained, using HIV-coated ELISA plates, that the ligated fragments simultaneously bound both HIV and neutrophils. After unbound bispecific antibody was removed from the vT240 plates, neutrophils (1 × 10⁵ cells/well) were added. Unbound neutrophils were removed by washing the plates, and the number of bound cells was determined by counting a minimum of 10 fields per well. There was on average one cell per field in the wells receiving F240 alone, presumably due to nonspecific FcRγ reactivity of the Fc region of F240 with neutrophils. There were significantly more neutrophils bound to HIV-1 gp160 in the presence of the bispecific antibody, with an average of 12 neutrophils per field. Neutrophils failed to bind to HIV-coated wells to which medium or the 14A8 antibody was added (data not shown).

Given its intact immunoreactivity, the bispecific antibody was then tested for the functional ability of neutrophil-mediated antibody-dependent cell-mediated virus inhibition (ADCVI) (12). Antibodies were titered in 96-well round-bottom plates in 50 μl medium containing 20% heat-inactivated fetal bovine serum. Target cells were peripheral blood mononuclear cells (PBMC) productively infected with HIV-1 4 days prior to use, as previously described (23), and 1 × 10⁵ infected cells were added per well in 50 μl. Within 10 min of the combination of antibody and infected cells, neutrophils were added to the wells at 1 × 10⁶ effector cells/well in 100 μl, resulting in an effector/T-cell ratio of 10:1. After 4 h, in order to measure the surviving infectious virus, phytohemagglutinin-stimulated PBMC were added as indicator cells (1 × 10⁵/well). These indicator PBMC were incubated for 7 days in the presence of interleukin 2, at which time p24 was quantitated by ELISA (5). Given the short half-life of neutrophils, the survival of neutrophils beyond a few hours under these conditions was negligible. Therefore, this assay measured the immediate reduction in infectious virus and cell-cell transmission that would result in the viral infection of the fresh PBMC (6).

Thus, as shown in Fig. 2, bispecific-antibody-mediated destruction of HIV and HIV-infected cells was determined by assaying the inhibition of subsequent HIV replication or p24 levels. The ability of bispecific antibody to mediate ADCVI was shown for both clade B isolates, R5 (SF162 and BaL) and R5X4 (89.6), as well as a clade C isolate (R5; 93MW960). The combination of unconjugated 14A8 F(ab) with F240 F(ab) also inhibited HIV replication, albeit not as efficiently. The activity of the unconjugated but combined antibody components more than likely reflects the nonspecific activation of neutrophils by the monovalent anti-CD89 component. It is known that the monovalent anti-CD89 components of other bispecific antibodies are effective at activating polymorphonuclear leukocytes (PMN) (7, 15-17, 33-35). There was no viral inhibition in control wells containing antibody, target cells, and indicator cells without neutrophils. Viral replication was similar for control wells containing effector cells and target cells without antibody and target cells alone (data not shown).

FIG. 2.

Antibody-dependent cell-mediated viral inhibition mediated by bispecific antibody and neutrophils. Effector neutrophils were incubated with serial dilutions of conjugated bispecific antibody (▪) or the combination of unconjugated 14A8 F(ab) and F240 F(ab) (□) antibodies, and HIV-infected target cells were added. A 1:1 ratio of F(ab) was used for the unconjugated antibody, and the concentration shown represents the final concentration. After 4 hours of incubation, uninfected indicator T cells were added. After 1 week, p24 was quantitated by ELISA, and the values were compared to those for control wells in which effector and target cells without antibody were incubated. Percent inhibition was determined by the following formula: [(p24 control − p24 test)/p24 control]·100. ADCVI activities against BaL-, 89.6-, SF162-, and 93MW960-infected cells were measured.

There have been contradictory reports on the functional activity of neutrophils in HIV-infected individuals. While it was reduced in some studies (8, 24, 26), in other studies, there were no differences in specific activities of neutrophils from HIV-seropositive and HIV-seronegative individuals (3). Neutropenia is commonly found in individuals diagnosed with AIDS or with CD4 counts of <200 (18, 31). However, immunomodulatory molecules, including cytokines and antibody for opsonization, have been shown to restore functional activity to defective neutrophils (24, 32, 37). More importantly, highly active antiretroviral therapy has been shown to significantly improve the functional activity of neutrophils (22).

It has been demonstrated that HIV can bind to resting and activated neutrophils, which then may transfer the virus to uninfected cells (13, 28). Furthermore, there remains the potential that HIV complexed with bispecific antibody may be redirected to other CD89-positive (CD89⁺) cells (such as monocytes), resulting in the transmission of infection. However, it can be argued that, given the complex envelope glycoprotein and associated host cell antigens acquired by the virus, HIV binding to neutrophils may have a larger impact in vitro than in vivo. That is, it would be expected that innate and adaptive immune functions of neutrophils, including phagocytosis and ADCVI, would diminish the ability of neutrophils to transfer virus to uninfected cells in vivo. This ADCVI activity would be further potentiated by a bispecific antibody that could bind to the effector cells (neutrophils) in the presence of plasma immunoglobulin, as well as react broadly and with high affinity to all clades of HIV. It can be further postulated that the range of inhibition (50% to 90%), even at the highest concentration tested, may limit usefulness in vitro. However, since PMN do not survive long in culture, it can be argued that in vivo there would be a continuous pool of PMN to be armed to continue to destroy virus. The arming of neutrophils to effectively destroy HIV represents a significant arsenal, given the presence of neutrophils in the circulation, on the mucosal surface, in lymphoid tissue, and at sites of inflammation (14).

In conclusion, we have demonstrated that a bispecific antibody created by conjugating the broadly reactive anti-gp41 antibody F240 and an antibody to the IgA receptor (14A8) can be used to induce neutrophil destruction of HIV. These in vitro results suggest that the use of bispecific antibody is a feasible approach for immunotherapy and that it has significant potential to be tested in animal models.