Fc-engineered antibody therapeutics with improved anti-SARS-CoV-2 efficacy

Abstract

Monoclonal antibodies with neutralizing activity against SARS-CoV-2 have demonstrated clinical benefits in cases of mild-to-moderate SARS-CoV-2 infection, substantially reducing the risk for hospitalization and severe disease^1-4. Treatment generally requires the administration of high doses of these monoclonal antibodies and has limited efficacy in preventing disease complications or mortality among hospitalized patients with COVID-19⁵. Here we report the development and evaluation of anti-SARS-CoV-2 monoclonal antibodies with optimized Fc domains that show superior potency for prevention or treatment of COVID-19. Using several animal disease models of COVID-19^6,7, we demonstrate that selective engagement of activating Fcγ receptors results in improved efficacy in both preventing and treating disease-induced weight loss and mortality, significantly reducing the dose required to confer full protection against SARS-CoV-2 challenge and for treatment of pre-infected animals. Our results highlight the importance of Fcγ receptor pathways in driving antibody-mediated antiviral immunity and exclude the possibility of pathogenic or disease-enhancing effects of Fcγ receptor engagement of anti-SARS-CoV-2 antibodies upon infection. These findings have important implications for the development of Fc-engineered monoclonal antibodies with optimal Fc-effector function and improved clinical efficacy against COVID-19 disease.

Extended Data Fig. 1:. Cloning and characterization of the IgG binding activity of hamster FcγRs.

a, Syrian hamster FcγRs were cloned, and their sequences were determined. The FcγR ectodomains are underlined. b-e, The affinity of human IgG1 and Fc variants (b, e, SPR sensorgrams), as well as of mouse (c) and hamster (d) IgG subclass variants for the various classes of hamster FcγRs was determined by surface plasmon resonance (SPR), using soluble hamster FcγR ectodomains. n.d.b., no detectable binding.

Extended Data Fig. 2:. Comparison of the FcγR expression levels in the various effector leukocyte populations between young and older FcγR humanized mice.

FcγR expression was assessed by flow cytometry in peripheral blood leukocyte populations from young (6-7 weeks old; orange) and older (17 weeks old; blue) FcγR humanized mice. a, Gating strategy to identify the various leukocyte populations, b, Representative histogram overlay plots of FcγR expression in young and older FcγR humanized mice. Corresponding isotype controls are indicated in lighter shade. c, Quantitation of FcγR expression (MFI, median fluorescence intensity subtracted from the respective isotype control) in various leukocyte populations. Results are from 4 or 5 mice per group for young and older mice, respectively.

Extended Data Fig. 3:. Histopathological analysis of lung tissue from SARS-CoV-2-infected FcγR humanized mice.

Lungs from SARS-CoV-2-infected (MA10 strain, 10⁴ pfu, i.n.) FcγR humanized mice (16-22 weeks old) were harvested on day 4 post-infection and evaluated histologically to assess the pathological changes associated with SARS-CoV-2 infection. a, Uninfected mice were characterized by clear alveolar spaces and absence of inflammatory cell infiltrates (low magnification (left panel, 200x); high magnification (center and right panels, 400x). b, In contrast, SARS-CoV-2 infection was associated with perivascular and peribronchial mononuclear leukocyte infiltration (200x left; 400x center panel), as well as the presence of macrophages and neutrophils in alveoli and necrotic cellular debris in alveolar spaces (600x, right panel). c, In addition, SARS-CoV-2-infected mice exhibited perivenular mixed neutrophilic, histiocytic, and lymphocytic inflammation, reactive endothelium and extravasation of leukocytes (left panel, 400x), as well as foci of interstitial neutrophilic and macrophage inflammation with hemorrhage and single cell necrosis (center and right panels, 400x). Images are representative of one uninfected and six infected mice.

Extended Data Fig. 4:. In vitro neutralization activity of anti-SARS-CoV-2 mAbs against SARS-CoV-2 strains.

Neutralization activity of REGN and BMS/RU mAbs (individual mAbs or as a cocktail) against SARS-CoV-2 pseudotyped reporter viruses was measured by in vitro neutralization assay. (a) In vitro neutralization curves and (b) IC₅₀ values of REGN (upper panel) and BMS/RU (lower panel) against SARS-CoV-2 WT, MA10, or B.1.351. (c) Fold change of SARS-CoV-2_MA10 and SARS-CoV-2_B.1.351 IC₅₀ was calculated over WT. n= 1 experiment performed in duplicates. (d) Cryo-EM structures of REGN10933 and REGN10987 complexed with SARS-CoV-2 (PDB: 6XDG) or C135 (PDB: 7K8Z) and C144 (PDB: 7K90) complexed with the SARS-CoV-2 spike trimer. Residues within the SARS-CoV-2_WT RBD that are mutated in the SARS-CoV-2_MA10 strain (Q493K, Q498Y, P499T) are indicated in magenta.

Extended Data Fig. 5:. In vitro neutralization activity and antigenic specificity of Fc variants of anti-SARS-CoV-2 mAbs.

To confirm that changes in the Fc domain have no effect on the neutralization activity and Fab-mediated functions of anti-SARS-CoV-2 mAb, Fc domain variants were characterized in (a-c) in vitro neutralization assays using SARS-CoV-2_MA10 pseudotyped reporter viruses and (d, e) in ELISA assays using SARS-CoV-2 RBD. n= 1 experiment performed in duplicates. (a, d) REGN and (b, e) BMS/RU mAb cocktails were expressed as Fc variants and their in vitro neutralization activity (a, b, c, IC₅₀ values) and (d, e) antigenic specificity was compared among Fc variants.

Extended Data Fig. 6:. In vitro and in vivo stability of Fc variants of anti-SARS-CoV-2 mAbs.

(a) Size-exclusion chromatography (SEC) analysis of Fc variants to determine aggregate formation among Fc domain variants. The SEC elution profiles and abundance (percentage) of monomeric IgG is indicated for the different Fc variants. (b) Fc variants of the REGN mAb cocktail were administered (i.v.; 50 μg) to FcγR humanized mice and antibody serum levels were determined by ELISA at various time points after antibody administration. n= total of 3 mice per group from two independent experiments. Data are mean ± s.e.m.

Extended Data Fig. 7:. High-dose treatment of SARS-CoV-2-infected FcγR humanized mice with anti-SARS-CoV-2 mAbs Fc variants enhanced for activating FcγR binding is not associated with enhanced disease.

(a) Following the experimental strategy in Fig. 3b, SARS-CoV-2-infected (MA10, 10⁴ pfu, i.n.) FcγR humanized mice (n=3 for PBS and n=5 for mAb-treated groups) were treated (i.v.) with 40 mg/kg REGN mAb cocktail expressed as Fc variants with diminished (GRLR) or enhanced activating FcγR binding (GAALIE). (b) SARS-CoV-2-infected mice were treated with 10 mg/kg BMS/RU mAb cocktail expressed as Fc variants GRLR or GAALIE on day 2 post infection. n= total of 10 mice per group for PBS, n=9 for GAALIE, and n=6 for GRLR from two independent experiments Weight loss (mean ± s.e.m.) was compared between GRLR and GAALIE-treated groups by two-way ANOVA (Bonferroni post hoc analysis adjusted for multiple comparisons). NS, not significant. Red arrow indicates time point of mAb treatment post-infection.

Extended Data Fig. 8:. Mouse and human C1q binding of Fc domain variants of IgG1.

The capacity of IgG1 Fc domain variants to interact with mouse (a) and human (b) C1q was assessed by ELISA (n= 1 experiment performed in duplicates). The KA (K322A) variant, which has previously described as a complement-deficient mutant, was included as control.

Fig. 1:. Contribution of Fc effector function to the protective activity of neutralizing anti-SARS-CoV-2 mAbs in hamster infection models.

a, Overview of the FcγR locus organization in humans, mice, and Syrian hamsters. b, Fc variants of human IgG1 were evaluated for their affinity for hamster FcγRs. Numbers indicate the fold-change in affinity compared to wild-type human IgG1. n.d.b., no detectable binding. c, d, Wild-type and FcR null (GRLR) variants of REGN mAb cocktail (c) or S309 mAb (d) were administered i.v. (5 mg/kg) to Syrian hamsters one day before (prevention model, c) or after (therapy model, d) i.n. challenge with SARS-CoV-2 (NYC isolate, 10⁵ pfu) (n=9 hamsters per group for PBS and GRLR-treated, n=10 for wild-type from two independent experiments for c and n=6 hamsters per group from two independent experiments for d). Hamsters were monitored for weight loss (left; mean ± s.e.m.) and lung viral titers (right, analyzed on day 7 (c) or 6 (d) post-infection) were compared between treatment groups by one-way ANOVA (Bonferroni post hoc analysis adjusted for multiple comparisons). P values are indicated. e-g, SARS-CoV-2-infected hamsters (10⁵ pfu, NYC isolate) were treated on day 1 post-infection with Fc variants of the REGN mAb cocktail (5 mg/kg, i.v.) exhibiting differential hamster FcγR binding affinity and A/I ratio (calculated based on FcγRIV/FcγRIIb affinity). Weight loss (e, plotted over time (mean ± s.e.m.) or f, as max change) and lung viral titers (g, assessed on day 6 post-infection) were compared by one-way ANOVA (Bonferroni post hoc analysis adjusted for multiple comparisons). P values are indicated. n=9 hamsters per group for PBS-treated, n=7 for GRLR and GAALIE, and n=5 for V11 from two independent experiments. Red arrow indicates time point of mAb treatment post-infection. Boxes and whiskers represent the median, quartiles, and range (minimum to maximum).

Fig. 2:. Fc-FcγR interactions are required for the therapeutic activity of neutralizing anti-SARS-CoV-2 mAbs in mouse infection models.

a, b, FcγR humanized mice were infected with mouse-adapted SARS-CoV-2 (MA10 strain, 10⁴ pfu, i.n.) and weight loss (mean ± s.e.m.) was compared in (a) young (7 weeks old; n=5) and older (18 weeks old; n=4) mice, as well as in (b) mice (16–19 weeks old) challenged with the indicated inoculum dose. n=5 mice per group; n=4 for 10⁴ and 10 pfu dose groups from two independent experiments. c, d, The therapeutic activity of REGN mAb cocktail (expressed as human IgG1 and administered at 5 mg/kg one day post-infection) was evaluated in FcγR humanized and FcγR deficient (FcγR_null) mouse strains challenged with SARS-CoV-2 (MA10 strain, 10⁴ pfu i.n.). n= total of 10 mice per group for FcγR_null and n= total of 11 (PBS) and n=12 (REGN) mice per group for FcγR humanized mice from two independent experiments. e, f, SARS-CoV-2-infected FcγR humanized mice (MA10 strain, 10⁴ pfu i.n.) were treated with wild-type human IgG1 or GRLR variants of REGN mAb cocktail one day post-infection. n= total of 8 (PBS), 5 (WT), and 6 (GRLR) mice per group from two independent experiments. Weight loss (c, e; mean ± s.e.m.) and survival curves (d, f) were compared to the corresponding PBS-treated group by two-way ANOVA (Bonferroni post hoc analysis adjusted for multiple comparisons) and log-rank (Mantel–Cox) test, respectively. P values are indicated. NS, not significant. Red arrow indicates time point of mAb treatment post-infection.

Fig. 3:. Selective engagement of activating FcγRs improves the therapeutic activity of anti-SARS-CoV-2 mAbs.

a, Human IgG1 Fc variants with differential affinity for specific classes of human FcγRs were generated for anti-SARS-CoV-2 mAbs. Numbers indicate the fold-change in affinity compared to wild-type human IgG1. b-g, Following the experimental strategy in panel b, SARS-CoV-2-infected FcγR humanized mice were treated (i.v.) at the indicated dose with REGN (c-e) or BMS/RU (f-g) mAb cocktail, or the non-neutralizing mAb CR3022 (h), expressed as wild-type human IgG1 or as Fc variants with differential affinity for human FcγRs. Weight loss (mean ± s.e.m.) (h and d, g, left panels; e, curves from individual mice) and survival curves (c, f and d, g, right panels) of antibody-treated mice were compared with the corresponding PBS-treated group by two-way ANOVA (Bonferroni post hoc analysis adjusted for multiple comparisons) and log-rank (Mantel–Cox) test, respectively. P values are indicated. NS, not significant. c, n=6 mice per group from two independent experiments; d, e, n= total of 10 mice per group (for PBS, GAALIE, GA groups), n=9 for WT, and n=11 for ALIE from four independent experiments; f, n= total of 8 (PBS and 1 mg/kg) or 9 (10 mg/kg and 4 mg/kg groups) mice per group from two independent experiments; g, n= total of 7 (GAALIE) or 12 (PBS, WT) mice per group from three independent experiments. h, n= total of 8 (GAALIE) or 10 (PBS, WT) mice per group from two independent experiments. Red arrow indicates time point of mAb treatment post-infection.

Fig. 4:. Prophylactic activity of anti-SARS-CoV-2 mAbs is enhanced by selective engagement of activating FcγRs.

In a model of mAb-mediated prophylaxis of SARS-CoV-2 infection (a), the activity of wild-type and GAALIE variants of the REGN mAb cocktail was assessed. FcγR humanized mice were treated (i.v.) at the indicated dose with REGN mAb cocktail expressed as wild-type human IgG1 or as GAALIE variant one day prior to challenge with SARS-CoV-2 (MA10, 10⁴ pfu i.n.). Weight loss (mean ± s.e.m.) (c, left panel) and survival curves (b and c, right panel) of antibody-treated mice were compared with the PBS-treated group by two-way ANOVA (Bonferroni post hoc analysis adjusted for multiple comparisons) and log-rank (Mantel–Cox) test, respectively. P values are indicated. NS, not significant. b, n= total of 7 mice per group (n=6 mice/group for 2 mg/kg) from two independent experiments; c, n= total of 7 (WT) or 9 (PBS, GAALIE) mice per group from three independent experiments.