Conjugation of anti-HIV gp41 monoclonal antibody to a drug capable of targeting resting lymphocytes produces an effective cytotoxic anti-HIV immunoconjugate

Abstract

HIV-infected cells persisting in the face of suppressive antiretroviral therapy are the barrier to curing infection. Cytotoxic immunoconjugates targeted to HIV antigens on the cell surface may clear these cells. We showed efficacy in mouse and macaque models using immunotoxins, but immunogenicity blunted the effect. As an alternative, we propose antibody drug conjugates (ADCs), as used in cancer immunotherapy. In cancer, the target is a dividing cell, whereas it may not be in HIV. We screened cytotoxic drugs on human primary cells and cell lines. An anthracycline derivative, PNU-159682 (PNU), was highly cytotoxic to both proliferating and resting cells. Human anti-gp41 mAb 7B2 was conjugated to ricin A chain or PNU. The conjugates were tested in vitro for cytotoxic efficacy and anti-viral effect, and in vivo for tolerability. The specificity of killing for both conjugates was demonstrated on Env+ and Env- cells. The toxin conjugate was more potent and killed more rapidly, but 7B2-PNU was effective at levels achievable in patients. The ricin conjugate was well tolerated in mice; 7B2-PNU was toxic when administered intraperitoneally but was tolerated intravenously. We have produced an ADC with potential to target the persistent HIV reservoir in both dividing and non-dividing cells while avoiding immunogenicity. Cytotoxic anti-HIV immunoconjugates may have greatest utility as part of an "activate and purge" regimen, involving viral activation in the reservoir. This is a unique comparison of an immunotoxin and ADC targeted by the same antibody and tested in the same systems.IMPORTANCEHIV infection can be controlled with anti-retroviral therapy, but it cannot be cured. Despite years of therapy that suppresses HIV, patients again become viremic shortly after discontinuing treatment. A long-lived population of memory T cells retain the genes encoding HIV, and these cells secrete infectious HIV when no longer suppressed by therapy. This is the persistent reservoir of HIV infection. The therapies described here use anti-HIV antibodies conjugated to poisons to kill the cells in this reservoir. These poisons may be of several types, including protein toxins (immunotoxins) or anti-cancer drugs (antibody drug conjugates, ADCs). We have previously shown that an anti-HIV immunotoxin had therapeutic effects in animal models, but it elicited an anti-drug immune response. Here, we have prepared an anti-HIV ADC, which would be less likely to provoke an immune response, and show its potential for use in eliminating the persistent reservoir of HIV infection.

Keywords: HIV persistent reservoir; antibody drug conjugate; immunoconjugate; immunotoxin.

Fig 1

Cytotoxicity of ricin and potential drugs for use in ADCs. Cells were incubated for 3 days in the presence of the indicated drugs. For the final 3 h, MTS dye was added, and A₄₉₀ determined. Percent cytotoxicity is on the vertical axis, molar concentration of cytotoxic drug on the x-axis. The arrows above each graph indicate the concentration that achieves cytotoxicity that is highly significantly greater (P < 0.001) than the untreated control. (A) Drugs were tested against the Env− parent cell lines: H9 is a CD4+ lymphoma cell line, 293T is of human embryonic kidney origin. (B) C8166.R5 is a CD4+ lymphoma cell line transfected to express CCR5 and used for acute infections in vitro. PBMCs are human peripheral blood mononuclear cells (a mixture of lymphocytes and monocytes) placed into tissue culture medium without stimulation.

Fig 2

Characterization of ADCs and immunotoxins. (A) The chemical structures of the drug-conjugated linkers are shown, with maleimide at the top and the disulfide linker below. (B) Chromatographic analysis of 7B2–PNU conjugated with the maleimide linker was analyzed by FPLC, with tracings from the unconjugated antibody in blue, and the conjugate in red. Red vertical lines in the hydrophobic interaction tracing likely indicate 0, 1, 2, and 3 PNU molecules per antibody. (C) 7B2–dgA conjugation was assessed by SDS-PAGE electrophoresis under non-reducing conditions on gradient gels. Size markers are indicated in red in lane 6. Unconjugated dgA and 7B2 are shown in lanes 1 and 2, and different preparations of conjugate without purification are shown in lanes 3 and 4. Lane 5 shows the preparation post-purification on protein A-agarose. The immunotoxin consists of 0, 1, 2, and 3 dgAs per 7B2. (D) Mass spectrometry of unconjugated 7B2 (top) and 7B2–PNU (bottom). The peaks corresponding to the number of PNUs conjugated to 7B2 are indicated by the number of diamonds. The MS results also show a mixture of 7B2:PNU ratios.

Fig 3

Specificity of cytotoxicity of 7B2 immunoconjugates on Env+ and Env− cells. Cytotoxicity was assayed as in Fig. 1. Four different immunoconjugates were tested against Env+ cells (92 UG and H9/NL4-3) and their Env− parent cells (293T and H9). Concentrations of immunoconjugates used were as follows: on Env+ cells 7B2–dgA 20 ng/mL, 7B2–PNU and 7B2–calicheamicin 200 ng/mL, 7B2-amanatine 1111 ng/mL; on Env− cells all at 2,000 ng/mL. The concentrations indicated here and in further figures are based on total protein. Statistical significance, compared with untreated control, is shown by asterisks: *P < 0.05, **P < 0.01, ***P < 0.001.

Fig 4

Titration of immunoconjugates on different cell types. Each row demonstrates the cytotoxicity of immunoconjugates on Env+ cell (left) and its Env− parent (right). The immunoconjugates included two forms of the PNU-containing ADC and the dgA-based immunotoxin. The arrows above each graph indicate the concentration that achieves cytotoxicity that is highly significantly greater (P < 0.001) than the untreated control.

Fig 5

Effects over time of immunoconjugates on persistently infected H9/NL4-3 cells. (A) Viability determined by trypan blue dye exclusion and manual counting. Viability counts have been performed >10× with similar results. (B) Apoptosis as assessed by binding of fluorescent annexin-V. Cells were gated on viable cells by size and scatter. This experiment was repeated thrice. (C) Virus production by these constitutively infected cells was measured by p24 present in the cell supernatant, as quantified by antigen capture ELISA. Statistical significance of suppression of p24 production by the CICs compared with untreated control is shown by asterisks: *P < 0.05, **P < 0.01.

Fig 6

Antiviral effects of immunoconjugates during acute tissue culture infection. Infection was initiated by the addition of 10⁵ C8166.R5 cells to a 1:2,500 dilution of an inoculum consisting of a mixture of viable cells and cell debris from C8166.R5 infected with the clade B HIV isolate BaL. The immunoconjugate or Ab was added at time 0. Supernatants were sampled at multiple time points, but p24 was not detectable until 24 h post-infection. The final three time points are shown here. Note the logarithmic increase in p24 production between each time point. Statistical significance of p24 suppression by the CICs compared with untreated control is shown by asterisks: **P < 0.01, ***P < 0.001.