Preclinical development of humanized monoclonal antibodies against CD169 as a broad antiviral therapeutic strategy

Abstract

New therapies to treat or prevent viral infections are essential, as recently observed during the COVID-19 pandemic. Here, we propose a therapeutic strategy based on monoclonal antibodies that block the specific interaction between the host receptor Siglec-1/CD169 and gangliosides embedded in the viral envelope. Antibodies are an excellent option for treating infectious diseases based on their high specificity, strong targeting affinity, and relatively low toxicity. Through a process of humanization, we optimized monoclonal antibodies to eliminate sequence liabilities and performed biophysical characterization. We demonstrated that they maintain their ability to block viral entry into myeloid cells. These molecular improvements during the discovery stage are key if we are to maximize efforts to develop new therapeutic strategies. Humanized monoclonal antibodies targeting CD169 provide new opportunities in the treatment of infections caused by ganglioside-containing enveloped viruses, which pose a constant threat to human health. In contrast with current neutralizing antibodies that bind antigens on the infectious particle, our antibodies can prevent several types of enveloped viruses interacting with host cells because they target the host CD169 protein, thus becoming a potential pan-antiviral therapy.

Keywords: Antibody therapy; Antivirals; Humanization; Methods; Monoclonal antibody.

Fig. 1.

Humanization of anti-CD169-mAbs and functional validation. A. Diagram illustrating the design of humanized anti-CD169 mAbs. Created with Biorender. B. C. D. Competition for Raji-Siglec-1 cells binding between HIV-1_Gag-eGFP VLPs and 3 different anti-CD169 mAbs for 1 h at 37°C. The antibody isotype control was used as a reference (IsoC, with grey star symbols). A comparison between original murine antibody (m-, triangle and lighter color), chimeric antibody (c-, circle), and humanized CDR-engrafted antibody (h₀, square and darker color) is shown for each antibody clone (#1F5, B; #3F1, C; #5B10, D [corresponding to a non-linear fit to a variable response curve from one representative experiment out of two]). E. IC₅₀ values of individual mAbs that achieved a total HIV-1_Gag-eGFP VLP blocking effect. Data from two independent experiments.

Fig. 2.

Optimization of humanized anti-CD169 mAbs and functional validation. A. Potential sequence liabilities found in CDR-engrafted humanized antibodies define variants indicated by suggested amino acid exchanges. B. Competition assay with Raji-Siglec-1 cell binding and HIV-1_Gag-eGFP VLPs to test anti-CD169 mAb variants to eliminate selected potential liabilities. C. Competition for Raji-Siglec-1 cell binding between HIV-1_Gag-eGFP VLPs and anti-CD169 mAbs. We tested the three different forms of #1F5 and #3F1 antibodies: murine (m-, triangle and lighter color), humanized (h₀-, square), and stabilized (h₁-, diamond). The comparison between these antibodies is shown with a non-linear fit to a variable response curve from one representative experiment out of two. D. IC₅₀ values of individual mAbs were calculated with a non-linear fit to a variable response curve. Data from two independent experiments.

Fig. 3.

Humanized anti-CD169 mAbs block DC-mediated viral entry. A. Relative viral uptake of HIV-1 by LPS-activated DCs. Cells were pre-incubated with humanized anti-CD169 mAbs and then pulsed for 4 h at 37°C with non-replicative HIV-1_{NL4–3 Gag-iGFP}. Values are normalized to cells incubated with isotype control mAb, showing a mean ± SD entry of 78.8 ± 9.7%, which is considered 100%. B. Relative HIV-1_NL4–3 trans-infection mediated by LPS-treated DCs pre-incubated with anti-CD169 mAbs. Values are normalized to cells pre-incubated with isotype control mAb, showing a mean ± SD entry of 2.83 ×106 RLU ± 2.3 ×106, which is considered 100%. C. Relative viral uptake of EBOV_eGFP-VP40 VLPs by LPS-activated DCs. Cells were pre-incubated with humanized anti-CD169 mAbs and then pulsed for 4 h at 37°C. Values are normalized to uptake by cells incubated with isotype control mAb, with a mean ± SD entry of 51.2 ± 18.7%, which is considered 100%. D. Relative internalization of EBOV_BlaM-VP40 VLPs by LPS-activated DCs pre-incubated with anti-CD169 mAbs. Values are normalized to cells incubated with isotype control mAb, with a mean ± SD entry of 46.8 ± 14.2%, which is considered 100%. E. Relative viral uptake of SARS-CoV-2 by LPS-activated DCs. Cells were pre-incubated with humanized anti-CD169 mAbs and then pulsed for 3 h at 37°C. Values are normalized to cells incubated with isotype control mAb, with a mean ± SD entry of 13.65 ± 8.2 ng/ml, which is considered 100%. F. Relative SARS-CoV-2 trans-infection mediated by LPS-activated DCs pre-incubated with anti-CD169 mAbs. Values are normalized to cells incubated with isotype control mAb, with a mean ± SD entry of 12.3 ± 2.9 ng/ml, which is considered 100%. Data information: In (A–F), data are presented as mean values and SD from two independent experiments and include cells from six donors. Statistical differences were assessed using a one-sample Wilcoxon test.

Fig. 4.

Stability and affinity analysis of humanized and stabilized anti-CD169-mAbs under oxidative stress. A. Chromatogram of anti-CD169-mAbs before and after H₂O₂-induced oxidative stress measured by SEC-MALS. B. Competition for Raji-Siglec-1 cell binding between HIV-1_Gag-eGFP VLPs and anti-CD169 mAbs. We tested two different forms of #1F5 and #3F1 antibodies, namely, humanized (h₀-, square) and stabilized (h₁-, diamond), before (filled symbols) and after (empty symbols) treatment with H₂O₂-induced oxidative stress. Non-linear fit to a variable response curve from one representative experiment out of two. C. Sensograms of anti-CD169 mAbs coupled on SPR sensor arrays at 60 nM show binding to their target CD169 receptor in solution using the IBIS MX96 platform before and after oxidative stress. The response scale (Y-axis) is adjusted to show on and off rates for all antibodies, as sensor coupling seemed to be affected by amino acid changes introduced in #1F5 mAb, compared to #3F1 mAb, which showed similar coupling efficiencies upon optimization. The graphs show that on- and off-rates for both antibodies are unaffected by amino acid exchange and oxidative stress. D. Calculation of K_D by interpolation of equilibrium fit for #3F1 mAb from Fig. 4C. CD169 binding to #1F5mAb was too weak to calculate K_D. E. Flow cytometry profiles show the level of binding of h₁-3F1 at 1 μg/ml to HEK293 cells transiently transfected with ZsGreen1-CD169 or ZsGreen1-SUSD5 recombinant proteins. Antibody binding using a secondary antibody labelled with the fluorophore AF647.