Antibody potency, effector function, and combinations in protection and therapy for SARS-CoV-2 infection in vivo

The effect of mouse adapting spike mutations on antibody neutralization in vitro

To determine if the two amino acid changes (Q498T/P499Y) in the mouse-adapted SARS-CoV-2 S RBD would affect antibody neutralization, we used HIV-1 virus pseudotyped with the SARS-CoV-2-MA S protein (S-MA). The recombinant pseudovirus was produced by cotransfection of S-MA with a replication-incompetent proviral genome (_NL4-3ΔEnv-NanoLuc) that lacks a functional env gene and encodes a nanoluciferase reporter in the place of the nef gene (Fig. 1 A; Robbiani et al., 2020; Schmidt et al., 2020). To maximize S-incorporation, we truncated the C-terminus of S-MA by 19 amino acids. Using nanoluciferase reporter expression as a measure of infection, we compared the ability of pseudotyped virus bearing either WT SARS-CoV-2 S (wtS) or S-MA to infect HT1080 cells stably expressing either human or murine ACE2. Consistent with previous reports (Zhou et al., 2020; Dinnon et al., 2020), wtS did not support pseudovirus infection of murine ACE2–expressing cells (Fig. 1 B), as relative light units (RLUs) remain at background levels observed with viruses lacking spike (Fig. 1 B). In contrast, the SARS-CoV-2 MA supported robust infection of both mouse and human ACE2–expressing cells (Fig. 1 B).

The S-MA pseudotyped virus was used to measure the neutralizing activity of eight different IgG1 hu-mAbs with variable potencies against SARS-CoV-2 ranging from inhibitory concentration (IC)₅₀/IC₉₀ (half-maximal/90% IC) of 4.4/18 to 26/140 ng/ml (Table 1; Robbiani et al., 2020). C002, C104, C105, C119, C121, and C144 all target the hACE2 interaction surface of the RBD albeit at different angles of approach, and C135 and C110 target a separate nonoverlapping epitope within the RBD (Robbiani et al., 2020; Barnes et al., 2020b). With the exception of C119, the antibody neutralization titers were similar when assayed using S-MA or wtS pseudotyped viruses (R = 0.81, P = 0.02; Table 1 and Fig. 1 C and D). C119 was inactive against S-MA pseudotyped virus because the epitope targeted by this antibody overlaps the mouse-adapting mutations (Fig. 1 C; Dinnon et al., 2020; Barnes et al., 2020a Preprint). Overall, however, mouse-adapting mutations in the S protein do not significantly affect neutralization by most RBD-targeting antibodies tested in vitro.

In vitro antibody neutralization does not uniformly correlate with its in vivo efficacy

To determine if there is a correlation between in vitro neutralization and in vivo activity, we performed prophylactic efficacy studies in aged BALB/c mice infected with SARS-CoV-2 MA. mAbs (8 mg/kg) were administered by intraperitoneal injection 12 h before intranasal infection with 10⁵ PFU of SARS-CoV-2 MA (Fig. 2 A). Viral lung titers were measured by plaque assay 2 d after infection, which is the kinetic peak of viral replication in this model (Dinnon et al., 2020). Since this is primarily a virus replication model, infected mice did not display overt disease. Mice injected with the isotype control antibody (anti-Zika antibody 3633; Robbiani et al., 2017) had mean viral lung titers of 10⁶ PFU (Fig. 2 A and Table 2). In agreement with the in vitro neutralization data, C119 failed to protect against SARS-CoV-2 MA in vivo (Fig. 1 C, Fig. 2 A, and Table 2). In contrast, the other anti–SARS-CoV-2 antibodies tested protected against infection to varying degrees (Fig. 2 A). C104 (IC₉₀ 223 ng/ml) reduced viral loads in the lungs of all mice to below the limit of detection (i.e., 50 PFU). Other antibodies that were more potent than C104 against SARS-CoV-2 MA pseudoviruses in vitro lowered viral loads by 2–4 orders of magnitude (Fig. 2 A; C002, C110, C121-LS, C135-LS, and C144-LS). The LS mutation increases antibody half-life but does not alter potency or effector function (Zalevsky et al., 2010). Interestingly, C105 only reduced the viral loads in vivo by 1–2 orders of magnitude, yet its neutralizing activity against SARS-CoV-2 MA pseudotyped virus in vitro was similar to that of C104 (Fig. 1, Fig. 2, and Table 1). Comparison of the mean viral lung titer and respective IC₉₀ revealed that the in vitro neutralizing activity in pseudovirus assays did not uniformly correlate with in vivo efficacy (Fig. 2 B).

Antibody Fc-effector function contributes to neutralization of SARS-CoV-2 MA in vivo

Virus neutralization in vitro is independent of antibody Fc-effector functions that impact in vivo efficacy against other viral infections (Lu et al., 2016; Halper-Stromberg et al., 2014; DiLillo et al., 2014; Bournazos et al., 2019; DiLillo et al., 2016; Bournazos et al., 2014). To examine the role of Fc-effector function on the neutralization of SARS-CoV-2 MA in vivo, we introduced the G236R/L328R (GRLR) mutation that abrogates antibody Fc receptor interaction in C002, C104, and C110 (Fig. 3; Bournazos et al., 2014). C002 and C104 target epitopes on the ACE2-binding interface of RBD, while C110 targets RBD but does not directly overlap the ACE2 interaction surface (Robbiani et al., 2020; Barnes et al., 2020a Preprint). As expected, the three antibody variants (C002_GRLR, C104_GRLR, and C110_GRLR) had IC₉₀ values that were not significantly different from those of WT hu-mAbs in SARS-CoV-2 MA pseudovirus assays in vitro (Fig. 3, A and B). Elimination of Fc-effector function did not significantly affect in vivo protection by C002, the least potent of the three antibodies (Fig. 3 C and Table 3; C002 = 9.9 × 10² PFU vs. C002_GRLR = 1.6. × 10³ PFU; P = 0.72). In contrast, loss of Fc-effector function significantly decreased the potency of both C104_GRLR and C110_GRLR (Fig. 3 C). Fc null C104_GRLR was 14-fold and C110_GRLR sixfold less potent than their Fc effector–sufficient counterparts (P = 0.0001 and P = 0.004, respectively; Fig. 3 C and Table 3). Variable Fc-effector requirements were also observed for anti-influenza antibodies, suggesting that the nature of the antibody–pathogen interaction can influence the ability of the Fc to engage its receptor (DiLillo et al., 2016).

To determine which Fc receptors are responsible for enhanced protective activity, we grafted the variable domains of C104 onto the mouse IgG1 or IgG2a Fc or the mouse IgG1 Fc variant IgG_D265A, which is a null-binding mutant for all mouse Fc receptors (Clynes et al., 2000). Mouse subclasses display differential affinity for the activating (FcRI, III, and IV) and the inhibitory (FcRIIB) receptors (Nimmerjahn and Ravetch, 2005). Since activating and inhibitory receptors are coexpressed on most effector cells, the in vivo activity of an IgG Fc is the result of the differential affinity of Fc binding to these receptors. Mouse IgG1 binds preferentially to FcRIIB, an inhibitory receptor, while IgG2a binds primarily FcRIV receptor, an activating receptor. FcRIV is expressed on monocytes/macrophages, neutrophils, and dendritic cells but not on murine natural killer cells (Bournazos et al., 2017). As expected, the three mouse antibody variants (i.e., C104-IgG1, IgG2, or D265A) had neutralization potencies that were not significantly different from each other’s or that of their human IgG1 counterpart in SARS-CoV-2 S-MA pseudovirus assays in vitro (Fig. 3, D and E; and Table 3). Whereas mouse C104-IgG2a provided a similar level of protection as WT human C104 (human IgG1, a human subclass that engages mouse FcRIV), mouse C104-IgG1 and the C104-IgG_D265A variant were significantly less active (Fig. 3 F and Table 3). We conclude that activating Fc receptors contributes to optimal protection against SARS-CoV-2 MA in vivo.

Antibody combinations potently neutralize SARS-CoV-2 in vivo

Combinations of antibodies targeting nonoverlapping epitopes have the potential to increase antiviral potency and can prevent the emergence of antibody escape mutations (Hansen et al., 2020; Weisblum et al., 2020 Preprint). To test whether there might be a benefit of antibody combinations for in vivo protection, we tested mixtures of antibodies that target nonoverlapping epitopes on the RBD (C135-LS/C121-LS and C135-LS/C144-LS; Robbiani et al., 2020). The antibody mixtures were tested at combined total doses of 16, or 5.3, or 1.8 mg/kg compared with 8 mg/kg of each antibody alone or 16 mg/kg of isotype control. As expected, the isotype control antibody failed to reduce virus replication (Fig. 4). While 16 mg/kg of C135/C121 provided sterilizing protection (i.e., viral titers below the limit of detection) from SARS-CoV-2 MA replication in all mice tested (n = 15), C135/C144 performed nearly as well with only 1 of 15 mice tested having a low but measurable virus titer (Fig. 4 and Table 4). Whereas C135/C144 was sterilizing at 5.3 mg/kg, half of the C135/C121 mice had low but detectable viral lung titers. Notably, the 1.8-mg/kg dose of either combination significantly reduced titers compared with isotype control antibody but was not as effectivel as 5.3 or 16 mg/kg (Fig. 4 and Table 4). In addition, the antibody combination of C135/C144 at 5.3 mg/kg improved the levels of protection achieved from C135 alone at 8 mg/kg (Fig. 4).

To determine whether the C135/C144 antibody combination is effective against authentic SARS-CoV-2 in vivo, we performed prophylactic and therapeutic efficacy studies in Syrian hamsters. The antibody combination was administered 24 h before infection (i.e., prophylactic) or 12 h after infection (i.e., therapeutic). As a negative control, one group received antibody directed against HIV (10–1074) prophylactically. Viral loads in the lungs were determined 3 d after infection, the peak of replication in this model. In hamsters receiving control antibody, high levels of SARS-CoV-2 replication were observed in the lung tissue (9 × 10⁵ PFU/100 g tissue; Fig. 5). Prophylactic administration of the C135/C144 antibody combination prevented or significantly diminished SARS–CoV-2 replication at all antibody doses tested, including 2 mg/kg (Fig. 5). Similarly, early therapeutic administration of C135/C144 reduced lung viral loads below the limit of detection at the 40- and 12-mg/kg dose levels and decreased viral lung loads >10,000-fold with the lowest dose tested (4 mg/kg; Fig. 5). We conclude that relatively low doses of C135/C144 are effective in neutralizing SARS-CoV-2 both prophylactically and therapeutically.

Cells and viruses

Human Ace2-expressing HT1080 cells (HT1080_Ace2cl.14) were described previously (Schmidt et al., 2020). For constitutive expression of murine Ace2 (muAce2) in HT1080 cells, a cDNA-encoding muAce2 was inserted into a lentiviral vector CSIB 3′ to the SFFV promoter. HT1080_muAce2 cells were generated by transduction with CSIB-based virus followed by selection with 5 µg/ml blasticidin.

Cells were cultured in DMEM supplemented with 10% FCS at 37°C and 5% CO₂. Medium for Ace2-overexpressing cell lines contained 5 µg/ml blasticidin. All cell lines had tested negative for contamination with mycoplasma, and parental cell lines were obtained from the American Type Culture Collection. Recombinant SARS-CoV-2 MA virus was generated as described previously (GenBank accession no. MT844088; Dinnon et al., 2020). For virus titration by plaque assay, the caudal lobe of the right lung was homogenized in PBS, and the resulting homogenate was serial-diluted and inoculated onto confluent monolayers of Vero E6 cells, followed by agarose overlay. Plaques were visualized with overlay of neutral red dye on day 2 after infection (Dinnon et al., 2020).

Antibody production

All antibodies were produced as previously described (Mouquet et al., 2011). Briefly, 293-6E cells that were maintained in Freestyle 293 Expression Medium (Thermo Fisher Scientific) were transiently transfected by using equal amounts of Ig heavy and light chain expression vectors using branched polyethylenimine (Sigma). 7 d after transfection, cells were spun down at 4,200 g for 40 min at 4°C, and supernatants were filtered through a 0.22-µM filter (Millipore). Antibodies were purified from filtered supernatants using Protein G Sepharose 4 Fast Flow (Cytiva). Antibodies were buffer exchanged and concentrated into PBS using an Amicon Ultra centrifugal filter (Millipore).

SARS-CoV-2/SARS-CoV-2 MA pseudotyped reporter virus

SARS-CoV-2 pseudotyped particles were produced by cotransfection of pNL4-3ΔEnv-nanoluc and pSARS-CoV-2-MA-S_trunc in 293T cells (Schmidt et al., 2020; Robbiani et al., 2020). For generation of SARS-CoV-2 MA pseudotyped particles, a plasmid expressing the mouse-adapted SARS-CoV-2 S (pSARS-CoV-2-MA-S_trunc) was generated by introducing the Q498Y/P499T mutation into pSARS-CoV2-S_trunc and was used for cotransfection with pNL4-3ΔEnv-nanoluc.

SARS-CoV-2 MA pseudotyped virus neutralization assay

A 40-µg/ml initial dilution of mAbs was fourfold serially diluted over 11 dilutions. 55 µl of antibody dilutions was incubated with a 55-µl aliquot of SARS-CoV-2 MA pseudotyped virus (containing ∼10³ infectious units) for 1 h at 37°C. 100 µl of the mixture was subsequently incubated with HT1080_muAce2 cells for 48 h. Consequently, the resulting antibody starting dilution was 10 µg/ml, and each well received 50 µl virus. Following incubation, cells were washed with PBS and lysed with Luciferase Cell Culture 5× reagent (Promega), and nanoluc luciferase activity in cell lysates was measured using the Nano-Glo Luciferase Assay System (Promega). RLUs obtained were normalized to those derived from cells infected with SARS-CoV-2 MA pseudovirus in the absence of mAbs (equivalent to 0% neutralization). The half-maximal and 90% ICs (IC₅₀ and IC₉₀) for mAbs were determined using four-parameter nonlinear regression (GraphPad Prism). In detail, we applied the least squares regression method without weighting, with top and bottom values constrained to 0 (% neutralization) and 100 (% neutralization), respectively, while allowing for a variable HillSlope.

Mouse studies and in vivo infections

All mouse studies were performed at the University of North Carolina (Animal Welfare Assurance #A3410-01) using protocols (19–168, 20–114) approved by the University of North Carolina Institutional Animal Care and Use Committee. All animal work was approved by the Institutional Animal Care and Use Committee at University of North Carolina at Chapel Hill according to guidelines outlined by the Association for the Assessment and Accreditation of Laboratory Animal Care and the US Department of Agriculture. All work was performed with approved standard operating procedures and safety conditions for SARS-CoV-2. Our institutional BSL3 facilities are designed to conform to the safety requirements recommended by Biosafety in Microbiological and Biomedical Laboratories, the US Department of Health and Human Services, the Public Health Service, the Centers for Disease Control and Prevention, and the National Institutes of Health. Laboratory safety plans have been submitted, and the facility has been approved for use by the University of North Carolina Department of Environmental Health and Safety and the Centers for Disease Control and Prevention. 12-mo-old female BALB/c mice (Envigo; #047) were inoculated with the indicated concentration of antibody intraperitoneally 12 h before infection. For infection, mice were anesthetized with a mixture of ketamine/xylazine and infected with 10⁵ PFU of SARS-CoV-2 MA in 50 µl PBS intranasally. Mice were monitored daily for body weight changes. At 2 d after infection, mice were euthanized, and lung tissue was harvested for viral titer analysis. Samples were stored at −80°C until homogenized and titered by plaque assay as described above.

Hamster studies and in vivo infections

8-wk-old Syrian hamsters obtained from Envigo were housed four per cage under ABSL3 containment. Groups of hamsters were treated with varying doses of a combination of two antibodies to SARS-CoV-2 RBD, an irrelevant (10–1074; anti-HIV) antibody, or PBS. Antibodies were administered by the intraperitoneal route either 24 h before or 12 h after challenge. Virus challenge was conducted under ketamine-xylazine anesthesia by intranasal instillation of 100 μl of SARS-CoV-2 that had been passaged once in Vero and once in Vero E6 cells; the dose determined by back-titration of the inoculum was 2.6e4 PFU. Beginning 2 d before challenge and extending until euthanasia, hamsters were evaluated clinically and weighed and their temperature recorded once daily. An equal number of hamsters from each treatment group was euthanized and necropsied 3 d after challenge. For animals euthanized on day 3, a sample of ∼100 mg of right cranial and right caudal lung lobes was excised, immersed in 0.9 ml of BA1/FBS, and homogenized using a mixer mill with stainless steel balls. Tissue homogenates were frozen to −80°C until assay. Virus titrations were performed using a double-overlay plaque assay on Vero E6 cells in 6-well plates. Briefly, serial 10-fold dilutions of tissue homogenate samples were inoculated onto drained monolayers and incubated 45 min, and 2 ml of a first overlay (0.5% agarose in MEM) without neutral red was added to each well. 1 d later, a second 2-ml overlay containing 0.06 mg/ml neutral red was added to each well, and plaques were counted 1 and 2 d later.