Pre- and postexposure efficacy of fully human antibodies against Spike protein in a novel humanized mouse model of MERS-CoV infection

VelocImmune mice were immunized with DNA encoding for MERS-CoV S protein and recombinant purified S protein and polyclonal titers were monitored. When sufficient responses were observed, spleens were harvested and monoclonal antibodies were isolated by both conventional hybridoma methodology and B-cell sorting technology. Supernatants containing antibodies were screened, as outlined in Fig. S1. To enrich for antibodies that could block the binding of its S protein to its receptor DPP4 and then potentially neutralize MERS-CoV, we tested more than 1,000 supernatants for binding to purified MERS-CoV-S-RBD (13). A blocking ELISA was then used to screen for antigen-positive supernatants that were able to block the interaction between MERS-CoV-S-RBD and DPP4. The ability of these antibodies to block entry of retrovirus particles pseudotyped with MERS-CoV S protein (MERSpp) was then quantified. All lead antibodies were subsequently cloned on to hIgG1 Fc regions, sequenced, expressed in CHO cells, and purified for further characterization. Using these methods, we identified approximately 30 fully human monoclonal antibodies that can very potently neutralize MERS-CoV by blocking the S protein–DPP4 interaction.

Although betacoronaviruses have lower mutation rates compared with other RNA viruses (24), they can still rapidly generate mutants and escape the pressure of a single therapeutic agent, including monoclonal antibodies (25–27). Therefore, we sought to isolate blocking antibodies that do not cross-compete for binding to MERS-CoV S protein and thus could be used together as a antibody mixture that is less susceptible to virus escape. To this end, we monitored the binding kinetics of each of the purified antibodies on MERS-CoV S protein precomplexed with a second antibody to select for pairs that do not cross-compete with each other. Based on the data presented in Fig. S2 A and B and Table S1, we conclude that REGN3048 and REGN3051 can cobind MERS-CoV RBD, suggesting that the two antibodies bind to discrete epitopes. Mutation of one epitope as a result of selective pressure by one antibody should not affect binding of the other antibody.

REGN3051 and REGN3048 were selected as the two lead antibodies that met all of the screening criteria.

To determine the binding kinetics of our monoclonal antibodies to MERS-CoV-RBD protein, surface plasomon resonance Biacore studies were performed (Fig. S2 C and D). The complete kinetic binding parameters for REGN3048 and REGN3051 are shown in Table S2. Both antibodies exhibit tight binding to the recombinant S protein with equilibrium dissociation constant (K_D) values ranging from 43–48 pM at 37 °C, suggesting that they have a high affinity for binding to MERS-CoV S protein.

The neutralization activities of REGN3051 and REGN3048 were first evaluated using MERSpp that express a luciferase reporter. These particles were incubated with decreasing concentrations of REGN3051 and REGN3048. Huh-7 cells were then transduced with the MERSpp/antibody complexes and intracellular luciferase activity was measured 3 d later. As seen in Fig. 1A, both antibodies can block cell entry of the pseudoparticles at picomolar concentrations; the 50% inhibitory concentration (IC₅₀) was 65 pM and 70 pM for REGN3051 and REGN3048, respectively. The entry of MERSpp was not affected by an isotype control antibody used at up to 10-fold higher concentrations. Importantly, the neutralization activity of these antibodies was specific for MERSpp, as neither inhibited the entry of control pseudoparticles [vesicular stomatitis virus glycoprotein (VSVg) pseudotyped particles] (Fig. 1B).

We next tested the ability of these two monoclonal antibodies to neutralize live MERS-CoV virus in vitro. Decreasing concentrations of antibodies were incubated with a defined amount of the clinical isolate MERS-CoV strain EMC/2012 (1). Vero E6 cells were then infected with the mixture and cell-viability was measured 2 d later. As shown in Fig. 1C, both REGN3051 and REGN3048 exhibited potent inhibitory activity against MERS-CoV, with IC₅₀ values of 460 pM and 180 pM, respectively.

Taken together, these data suggest that REGN3051 and REGN3048 potently block entry of MERS-CoV into susceptible cell lines and neutralize infectivity.

RNA viruses encode a low-fidelity genome replication machinery that allows them to rapidly adapt to their host environment (28). However, coronaviruses, such as MERS-CoV, encode for a nonstructural protein with a 3′-to-5′ exo-ribonuclease activity, which provides a proof-reading function during their replication and greatly increases the coding capacity of the genome without leading to mutational catastrophe (29, 30). Despite that finding, sequencing of multiple clinical isolates sampled during the first 2 y of the MERS-CoV epidemic has revealed that the MERS-CoV S protein of the virus is evolving (4, 31). We aligned 38 National Center for Biotechnology Information deposited sequences of MERS-CoV clinical isolates and compared them to the sequence of the first isolated strain, EMC/2012 (1). Based on the design of our screening assay, both REGN3051 and REGN3048 are expected to bind within the RBD of the MERS-CoV S protein, so the comparison of the sequences specifically focused on amino acids 367–606. We identified seven different amino acids that differ between the sequenced clinical isolates and the prototype EMC/2012 sequence: A431P, S457G, S460F, A482V, L506F, D509G, and V534A (Table S3). To test the ability of our antibodies to bind to conserved regions within the RBD and neutralize all MERS-CoV clinical isolates sequenced as of July 2014, we used site-directed mutagenesis to create plasmid constructs encoding for all S protein variants. We then used these to generate MERSpp pseudotyped with the modified S proteins and performed neutralization assays. REGN3051 is able to neutralize all isolates with IC₅₀ values ranging from 13–65 pM (Fig. S3). Similar values were also observed for REGN3048, with the exception of the V534A variant, which exhibited a partial resistance to the antibody. It is worth noting however, that this amino acid change was only observed in a single MERS-CoV isolate (Riyadh_2), which was isolated in 2012 and is part of a dead branch of the MERS-CoV evolutionary tree (31).

These data suggest that REGN3051 and REGN3048 bind to regions of MERS-CoV S protein that are conserved during the natural evolution of the virus.

We next compared the potency of our two antibodies with previously isolated monoclonal antibodies. Three groups have used in vitro antibody isolation methods—that is, phage display (26, 27) and yeast display (32)—to select for antibodies that bind to MERS-CoV S protein and block virus entry. For our studies, we selected a panel of three antibodies with published variable domain sequences, which were reported to potently neutralize MERS-CoV and bind to diverse epitopes. The published sequences for antibodies 3B12, MERS-4, and MERS-27 (27, 32) were cloned onto human IgG1 constant domains, then expressed and purified similarly to REGN3051 and REGN3048. We tested the three purified antibodies for binding to MERS S protein, blockade of the S protein–DPP4 interaction, and published neutralization properties (Fig. S3). We then tested the neutralization efficacy of all five antibodies using pseudoparticles generated with the prototypical EMC/2012 sequence and all clinical isolates described above (Fig. S3). REGN3051 and REGN3048 neutralize MERSpp with IC₅₀ values at least one log lower than the most potent comparator antibody, MERS-4. In addition, both 3B12 and MERS-27 appear to bind to less conserved sites on the MERS-CoV S protein, as pseudoparticles with sequences of multiple clinical isolates cannot be neutralized with these antibodies. These data suggest that antibodies isolated after in vivo selection using the VelociImmune platform are able to neutralize a broader range of MERS-CoV isolates with improved potency compared with several antibodies isolated solely based on in vitro biochemical properties.

In vivo testing of antiviral molecules requires a small animal model that is susceptible to MERS-CoV infection; however, wild-type mice are not susceptible to MERS-CoV infection (33). Sequence comparison of the sequence of mouse and human DPP4 revealed that the amino acids that have previously been identified as contact sites between MERS-CoV S protein and its receptor (13) differ between the two species. In addition, expression of human DPP4 in mouse cells allows for MERSpp entry and MERS-CoV propagation (34), indicating that entry of the virus is the limiting step in infection of mouse cells.

We hypothesized that mice expressing human DPP4 in place of mouse DPP4 would be susceptible to MERS-CoV and allow for in vivo testing of anti–MERS-CoV therapeutics. We used VelociGene technology to replace the 79 kb of the mouse Dpp4 gene with 82 kb of its human ortholog (Fig. S4A). The resulting mice express fully human DPP4 under the control of the mouse regulatory elements, to preserve the proper expression regulation and protein tissue distribution. Immunohistochemistry was performed to confirm huDPP4 expression in the lungs of huDPP4 mice. Using an antibody specific for huDPP4, we detected huDPP4 expressed in club cells, type 2 alveolar epithelial cells, and alveolar macrophages (Fig. S4B).

To test if humanized DPP4 mice (huDPP4 mice) can support infection, we intranasally inoculated mice with 2 × 10⁵ pfu of MERS-CoV. Although no mortality or clinical signs of disease was observed up to day 4, at days 2 and 4 postinfection mice were killed and their lungs were dissected. To obtain virus RNA levels, lungs were homogenized and analyzed by real-time PCR using primers specific to MERS-CoV (Fig. 2 A and B). To obtain virus titers, lungs were homogenized, clarified by centrifugation, and titered (Fig. 2C). Robust MERS-CoV replication in the lungs was evident at 2 and 4 d postinfection (dpi). RNA quantification, using primer sets specific for the MERS-CoV genome and MERS-CoV leader (designed to only amplify replicating MERS-CoV), demonstrated high levels of MERS-CoV replicating RNA in lungs collected at day 2, and these levels were maintained through day 4 postinfection. Plaque assay of lung homogenate quantified MERS-CoV levels of ∼3.63 × 10⁶ pfu/g lung at day 2 and ∼1.88 × 10⁷ pfu/g lung at 4 dpi, demonstrating active replication of MERS-CoV in the lungs of the huDPP4 mice. These data suggest that humanization of DPP4 using VelociGene technology has created a robust model of MERS-CoV in mice that can be used to assess MERS-CoV treatment in vivo.

Lungs from huDPP4 mice intranasally inoculated with either MERS-CoV or sham infected with PBS were analyzed for pathological changes (Fig. 3). At day 2 postinfection, peri-bronchiolar inflammation was evident with alterations in bronchiolar cell structure found throughout the lungs. Minimal peri-vascular inflammation or effects on alveolar structures were noted at day 2 postinfection. At 4 dpi, significant interstitial infiltration was observed with peri-vascular cuffing and extensive alveolar thickening. Bronchiolar alterations were still present as well. Importantly, this lung pathology is consistent with the radiographic findings of development of interstitial pneumonia and significant lung disease seen in humans with MERS-CoV, suggesting that this humanized DPP4 in vivo model of MERS-CoV infection recapitulates pathological sequelae that are seen in MERS-CoV infection of humans.

Unlike what is seen with other mouse models of MERS-CoV infections, virus replication and pathology in the huDPP4 mice is localized in the lungs. Histological analysis of brains section 4 dpi demonstrated no inflammation (Fig. S5). Homogenate of brain from MERS-CoV infected huDPP4 mice was tested for MERS-CoV specific mRNA at both day 2 and day 4 postinfection. No MERS-CoV specific RNA was detected.

Having established that humanized DPP4 mice were susceptible to MERS-CoV, we used this model to evaluate the activity of REGN3051 and REGN3048 in vivo. Mice were intraperitoneally injected with 200, 20, or 2 μg of either REGN3051, REGN3048, or 200 μg of hIgG1 isotype control antibody 24 h before intranasal infection with 1 × 10⁵ pfu of MERS-CoV. As seen in Fig. 4 A and B, both REGN3051 and REGN3048 were able to significantly decrease MERS-CoV specific RNA levels in the lungs by over 2 logs at the 200 μg per mouse dose, compared with the isotype control antibody. REGN3051 was more effective at reducing MERS-CoV RNA levels at the 20-μg dose compared with REGN3048 at the same dose. The 2-μg dosing of either antibody was ineffective at reducing viral RNA levels compared with isotype control-treated mice. When MERS-CoV titer was analyzed in the lungs (Fig. 4C) we found that both the 200- and 20-μg dose of REGN3051 reduces virus levels to near the level of detection in the assay (1 × 10⁵ pfu/g). REGN3048 is equally efficient at the 200-μg dose as REGN3051, whereas the 20- and 2-μg display a dose dependent inhibition of viral inhibition. These data suggest that REGN3051 and REGN3048 can effectively block MERS-CoV infection in vivo.

Histological analysis was also performed on lungs from mice treated 24 h preinfection with REGN3048, REGN3051, or hIgG1 isotype control antibody at 4 dpi (Fig. S6A). Lungs from mice pretreated with hIgG isotype control mice displayed significant lung pathology with increased interstitial inflammation, perivascular cuffing, and thickening of alveolar septa. Mice treated with 200 μg of either REGN3048 or REGN3051 had reduced inflammation with minimal foci of inflammatory cells in the interstitium and minor bronchiolar cuffing. In mice pretreated with 20 μg of REGN3048 and REGN3051, we found moderate levels of perivascular cuffing and interstitial inflammation compared with the higher antibody group. In contrast, the 2-μg antibody pretreated group had similar pathology to the hIgG1 isotype control displaying significant interstitial inflammation and predominant peri-vascular inflammation. Blinded histological scoring (Fig. S6B) reflects the reduced inflammation scores for treated mice. These data demonstrate that REGN3048 and REGN3051 confer a dose-dependent reduction in lung pathology following MERS-CoV infection corroborating viral RNA levels and virus titers determined for these mice.

Taken together, these data indicate that REGN3051 and REGN3048 can block MERS-CoV infection and disease in vivo when injected 1 d before infection. To our knowledge, REGN3051 and REGN3048 are, to our knowledge, the first fully human antibodies that have been shown to display efficacy in an in vivo model of MERS-CoV infection.

The ability to inhibit MERS-CoV replication and lung pathology after infection is a desired trait in a potential therapeutic. To assess whether REGN3051 is able have an effect therapeutically, we infected huDPP4 mice with MERS-CoV, and then 24 h later injected the infected mice intraperitoneally with either 500 μg of hIgG isotype control or REGN3051 at 500 or 200 μg per mouse. At 4 dpi, mice were killed and mouse lungs were analyzed for viral RNA, virus titer, and lung pathology. Both the 500- and 200-μg doses of REGN3051 were able to reduce viral RNA levels by about 10-fold in the lungs of mice compared with control antibody treated mice (Fig. 5 A and B). Lung titers of the same mice demonstrated significant reduction in MERS-CoV levels in the lungs with a greater than 2 log reduction at day 4 postinfection (Fig. 5C). These data demonstrate that REGN3051 can significantly inhibit viral replication even when administered 24 h following viral inoculation.

Histological analysis was performed on mice treated 24 h postinfection with hIgG control antibody, 500 μg REGN3051, or 200 μg REGN3051 (Fig. S7A). Mice treated with control antibody displayed similar pathology to the control above with significant interstitial inflammation, peri-vascular cuffing and thickening of alveolar septa. Mice treated with either 200 or 500 μg of REGN3051 had minimal interstitial inflammation and with reduced and only focal peri-vascular inflammation throughout the lungs. Blinded histological scoring demonstrates reduced inflammation scores for treated mice (Fig. S7B). These data demonstrate that therapeutic doses of REGN3051 reduce MERS-CoV induced lung pathology even when given 24 h postinfection.

Taken together, these data demonstrate that by coupling our VelocImmune and VelociGene technologies, we were able to rapidly produce candidate monoclonal antibodies and rapidly create relevant mouse models that were useful to evaluate potential therapeutics to prevent or treat infections by an emerging infectious disease threat.

The 293T cells (human embryonic kidney cell line) and Huh-7 cells (human hepatocarcinoma cell line) were grown in high glucose DMEM (Gibco) containing 10% (vol/vol) FBS (Gibco), 1% l-glutamine, and antibiotics (penicillin/streptomycin; Gibco) at 37 °C in a 5% (vol/vol) CO₂ atmosphere. Vero E6 (monkey kidney epithelial cells; ATCC, CRL-1586) were grown in minimal essential media (MEM; Corning) supplemented with 10% (vol/vol) FBS (Sigma Aldrich), 1% l-glutamine (Gibco), and 1% penicillin/streptomycin (Gemini bio-products) at 37 °C in a 5% (vol/vol) CO₂ atmosphere.

Transfections were performed using Lipofectamine LTX and Plus Reagent (Life Technologies) following the manufacturer’s instructions.

The Jordan MERS-CoV strain (GenBank accession no. KC776174.1, MERS-CoV- Hu/Jordan-N3/2012) was kindly provided by Kanta Subbarao (National Institutes of Health, Bethesda, MD) and Gabriel Defang (Naval Medical Research Unit-3, Cairo, Egypt). The EMC/2012 strain of MERS-CoV (GenBank accession no. JX869059.2) was kindly provided by Bart Haagmans (Erasmus Medical College, Rotterdam, The Netherlands) to M.B.F.

Codon-optimized cDNA sequence encoding for MERS-CoV S protein (EMC/2012 – GenBank JX869059) was synthesized by GeneArt and cloned into standard expression vectors. Plasmids encoding MERS-CoV S protein variants in the RBD were generated by mutagenesis using QuikChange II (Agilent Technologies) according to the manufacturer’s instructions. psPAX2 and pWPXLd-GFP encoding for HIV gag pol and GFP in the context of an HIV genome were obtained from the Trono laboratory (Ecole Polytechnique Fédérale de Lausanne, Switzerland). pCK-retro-luc was constructed by replacing the GFP ORF in pWPXLd-GFP with the gene encoding for firefly luciferase.

MERS-CoV S protein RBD (amino acids E367–Y606) (13) was expressed with the Fc portion of the human IgG1 molecule at the C terminus (MERS RBD-hFc). Human DPP4 (hDPP4) expressed with a 6xHistidine tag at the C terminus (hDPP4-6His) was purchased from R&D Systems. An anti-hFc polyclonal antibody was purchased from Jackson Immunochemical.

The variable regions of the previously described monoclonal antibodies against MERS-CoV S protein 3B12 (27) and MERS-4 (32) were synthesized and cloned with human IgG1 constant domains. Antibodies were produced in CHO cells and purified by Protein A affinity chromatography.

VelocImmune mice (37, 38) comprising DNA encoding human Ig heavy and κ light-chain variable regions and mouse constant regions were immunized with DNA and protein immunogens using standard techniques. When the desired immune response was achieved, splenocytes were harvested.

Some splenocytes were fused with P3 × 63Ag8.653 myeloma cells to form hybridomas and grown under hypoxanthine/aminopterin/thymidine selection. Media harvested from the antigen positive wells were used to determine the antibody affinity of binding to antigen using a solution competition ELISA. Of a total of 696 hybridomas that express antigen-specific antibodies, 182 were found to block the interaction between DPP4 and S protein and 123 blocked entry of MERSpp into target cells. Mouse constant domains were exchanged for human ones. The most potent antibodies were selected for subsequent analysis.

A separate pool of splenocytes was used to isolate MERS-CoV S reactive B cells. Cells were stained with FITC-conjugated anti-mouse Fc, incubated with biotinylated MERS RBD-hFc, and then stained with PE-conjugated streptavidin. The stained cells were analyzed by flow cytometry. Single B cells that were positive for both anti-mFc and MERS S protein were sorted and plated. RT-PCR of antibody genes from these individual B cells was performed as described previously (39). Briefly, cDNAs for each single B cell were synthesized via reverse-transcriptase reaction (Superscript III, Invitrogen). Each resulting reverse-transcriptase product was then split and transferred into two corresponding wells on two 384-well plates. One set of the resulting reverse-transcriptase products was first amplified by PCR using a 5′ degenerate primer specific for human IgG heavy-chain variable region leader sequence and a 3′ primer specific for mouse heavy-chain constant region, to form an amplicon. The amplicon was then amplified again by PCR using a 5′ degenerate primer set specific for framework 1 of human IgG heavy chain variable region sequence and a 3′ degenerate primer set specific for framework 4 of human IgG heavy chain variable region sequence. The other set of the resulting reverse-transcriptase products was first amplified by PCR using a 5′ degenerate primer specific for human κ light-chain variable region leader sequence and a 3′ primer specific for mouse κ light-chain constant region to form an amplicon. The amplicon was then amplified again by PCR using a 5′ degenerate primer set specific for framework 1 of human κ light-chain variable region sequence and a 3′ degenerate primer set specific for framework 4 of human κ light-chain variable region sequence. The heavy-chain and light-chain PCR products were cloned into expression vectors containing human IgG1 heavy-chain constant region and κ light-chain constant region, respectively. A total of 792 antigen-specific antibodies were tested in MERS-CoV pseudoparticles (MERSpp) neutralization assays. The most potent neutralizers were further analyzed.

Recombinant anti–MERS-CoV S antibodies were produced in CHO cells after stable transfection with paired expression plasmids containing heavy chain and light chain derived from the same B cell. Antibodies secreted into conditioned culture media by the transfected cells were purified by Protein A affinity chromatography.

Sequences of comparator antibodies were cloned in the same expression vectors and produced, purified, and characterized for binding to S protein and neutralization using the same techniques.

Equilibrium dissociation constants (K_D values) for anti–MERS-CoV S antibodies binding to MERS-RBD-mFc protein were determined on a real-time surface plasmon resonance biosensor platform using Biacore T-200 instrument (GE Healthcare). The CM5 Biacore sensor chip was first derivatized by coupling a monoclonal mouse anti-human Fc antibody (GE, # BR-1008-39) using the standard EDC-NHS chemistry. Filtered and degassed HBS-P buffer [10 mM Hepes, 150 mM NaCl, and 0.05% (vol/vol) surfactant P-20, pH 7.4] was used as running buffer during the immobilization of anti-human Fc and also for determining antibody binding kinetics parameters. The chip surfaces were activated by incubating with a 1:1 (by volume) mixture of 0.4 M 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride and 0.1 M N-hydroxy succinimide. This mixture was injected over the chip surface at a flow rate of 30 µL/min for 7 min. Following surface activation, monoclonal mouse anti-human Fcγ antibody (15 µg/mL in 10 mM sodium acetate buffer, pH 5.0) was injected over the activated chip surface at a flow rate of 30 µL/min for 4 min until a resonance unit (RU) signal of about 7,000 RU was obtained. Remaining active groups on the CM5 chip surface were then blocked by injecting 1 M ethanolamine over the surface at a flow rate of 30 µL/min for 7 min. The coupled chip surfaces and the reference surface were then washed and treated with 20 mM phosphoric acid to remove uncoupled residual proteins. The chip was washed with HBS-P buffer using the built-in prime procedure before performing the capture kinetic experiments.

To determine kinetics of antibody binding to MERS-RBD-mFc, antibody was captured at two surface densities through their Fc region using the monoclonal mouse anti-human Fc antibody-coupled surface. Following the capture step, different concentrations of MERS-RBD-mFc (10 nM to 78 pM, twofold dilution series) were injected over the captured antibody surface for 7 min at a flow rate of 50 µL/min, followed by the dissociation of antibody bound MERS-RBD-mFc for 30 min in HBS-P running buffer. The real-time binding of MERS-RBD-mFc to different antibody captured surfaces was monitored and the entire experiment was performed at 37 °C. Specific Biacore kinetic sensorgrams were obtained by the standard double-referencing procedure. The kinetic parameters were obtained by globally fitting the data to a 1:1 binding model using Biacore T200 Evaluation Software v1.0 curve-fitting software. The dissociation rate constant (k_d) was determined by fitting the change in the binding response during the dissociation phase, and the association rate constant (k_a) was determined by fitting MERS-RBD-mFc binding at different concentrations. The equilibrium dissociation constant (K_D) was calculated from the ratio of the k_d and k_a (K_D = k_d/k_a). The time needed (in seconds) for half of the bound ligand to dissociate from the captured antibodies (t1/2), is calculated by dividing 0.693 (natural log of 2) by the experimentally determined k_d.

Binding competition between MERS-CoV S antibodies was determined using a real time, label-free bio-layer interferometry assay on an Octet RED96 biosensor (Pall ForteBio). The entire experiment was performed at 25 °C in HBS-P Octet buffer [10 mM Hepes, 150 mM NaCl, and 0.05% (vol/vol) surfactant Tween-20, pH 7.4, 1 mg/mL BSA] with the plate shaking at the speed of 1,000 rpm. To assess whether antibodies are able to compete with one another for binding to recombinant MERS-CoV-RBD-mFc protein, a premix assay format was adopted and 50 nM of MERS-CoV-RBD-mFc was preincubated with 500 nM of different anti-MERS monoclonal antibody (subsequently referred to as mAb-2) for at least 2 h before performing the binding competition assay. A nonspecific monoclonal antibody was incubated with MERS-RBS.mFc as an Isotype control. Octet biosensors coated with an anti-hFc polyclonal antibody (Pall ForteBio, Cat# 18–5060) were first submerged into wells containing 50 µg/mL solution for 4 min to capture around 2 nm of anti-MERS antibody (subsequently referred to as mAb-1). Following the capture step, unoccupied anti-hFc polyclonal antibodies were then saturated by submerging Octet biosensors into wells containing 100 µg/mL of nonspecific fully human monoclonal antibody. Finally, the biosensors were immersed for 4 min into wells containing the premixed samples of 50 nM of MERS-RBD-mFc and 500 nM of mAb-2. The biosensors were washed in HBS-P Octet buffer in between every step of the experiment. The real-time binding response was monitored during the course of the experiment and the binding response at the end of every step was measured. The response of mAb-1 binding to the precomplex of MERS-RBD-mFc and mAb-2 was compared and competitive/noncompetitive behavior of different anti-MERS monoclonal antibodies was determined.

Retrovirus particles pseudotyped with MERS-CoV Spike protein (MERS pseudoparticles, MERSpp) were generated by cotransfecting 293T cells with a mix of plasmids expressing MERS-CoV S protein, psPAX2 and pCK-retro-luc. Supernatants were harvested 48 h posttransfection, clarified using centrifugation, concentrated using Lenti-X Concentrator (Clontech), aliquoted, and frozen at −80 °C. Control pseudoparticles were generated by substituting the plasmid expressing MERS-CoV Spike protein with a plasmid encoding for VSVg.

Mouse lung was homogenized in 1 mL of PBS with glass beads for 60 s at 6,000 rpm using a Magnalyzer (Roche). Homogenate was then centrifuged for 10 min at 10,600 × g and supernatant analyzed by plaque assay on Vero E6 cells to quantitate levels of virus remaining after treatment. Plaque assays were performed as previously described (40), except that plates were left for 3 d for plaques to appear.

RNA was extracted from mouse lung by homogenization in 1 mL of TRIzol (Life Technologies) using a Magnalyzer (Roche) according to manufacturer’s instructions. Levels of MERS-CoV RNA were assessed using the Taqman Fast virus one-step master mix (Applied Biosystems) according to the manufacturers’ instructions using a duplex of primers obtained from Life Technologies targeting a region of the genome upstream of the envelope gene (UpE); the leader sequence of the nucleocapsid messenger RNA (leader primer) and compared with mouse 18S rRNA was used as an endogenous control. Quantitative PCR reactions in Microamp fast optical reaction plates (Applied Biosystems) were read on a 7500 fast DX real-time PCR instrument (Applied Biosystems) and data were analyzed using the ΔCt method, with an uninfected control set to 1. Percent MERS-CoV RNA detected was expressed relative to levels of RNA detected in infected mice treated with isotype-matched control antibodies.

We used the VelociGene method as described previously, which enables large insertion or replacement of mouse gene in mice with the corresponding human genomic sequence (41). Briefly, a large targeting vector (LTVEC) was constructed to contain 82 kb human DPP4 genomic DNA from exons 2–26, including 3′UTR to replace 79-kb mouse Dpp4 counterparts sequence. Human BAC RP11-68L22 and mouse BAC RP23-362N15 containing the Dpp4 gene were identified using Blast and sequence confirmed using Illumina bench top sequencer MiSeq. The LTVEC was electroporated in the F1 hybrid (129S6SvEvTac/C57BL6NTac) ES cells. G418-resistant colonies were picked 10 d after electroporation and screened for correct targeting by the loss-of-allele assay. We used the VelociMouse method (42), in which targeted ES cells were injected into uncompacted eight-cell stage Swiss Webster embryos, to produce fully ES cell-derived F0 generation mice carrying the human DPP4 gene allele. Using these technologies, F0 mice were born about 5 mo after the targeting constructs were ready. All animal procedures were carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the NIH. The protocol was approved by the Regeneron Pharmaceuticals Institutional Animal Care and Use Committee.

Six- to 8-wk-old humanized DPP4 mice were injected intraperitoneally with indicated amounts of antibody or sterile saline as a control at 1 d preinfection or 1 dpi, depending on the experiment. Before intranasal inoculation, mice were anesthetized by intraperitoneal injection using a mix of xylazine (0.38 mg per mouse; Lloyd laboratories) and ketamine (1.3 mg per mouse; Henry Schein Animal Health), diluted in PBS to make a total volume of 50 μL per mouse. Once anesthetized, mice were intranasally inoculated with either PBS or 2 × 10⁵ pfu of MERS-CoV (Jordan) diluted into PBS for a total inoculum of 50 μL. During the experiment, mice were weighed before infection and every day of the experiment to assess MERS-CoV induced weight loss. Mice were killed at day 2 or day 4 postinfection using a lethal dose of isofluorane (Butler Animal Health Supply). Lungs were harvested for further analysis of MERS-CoV replication and pathology. For all experiments, 5–10 mice per group were euthanized and analyzed at each time point.

Histological slides were prepared from 5-μm sections of fixed, paraffin-embedded tissues, and stained with H&E. Fields were examined by light microscopy and analyzed. The degrees of interstitial, peribronchiolar, and perivascular inflammation were scored from 0 to 5. For each experimental group, the slides were blinded and scored from 0 to 5 and tabulated to compare strains, time points and treatments. Other histologic features, such as the presence of bronchiolar epithelial and alveolar damage, pleural changes, and the extent of peribronchovascular inflammation, were also noted for each group in the text. The overall inflammatory scores for each mouse were averaged for each group, and presented as average scores of all mice in each group and time point.

HuDPP4 mouse lungs was stained with anti-M MERS-CoV antibody. For staining, 5-μm paraffin sections of fixed lung tissue were cut by the histology core at University of Maryland, Baltimore. Slides were deparaffinized in xylenes and rehydrated in an ethanol series. Endogenous peroxidase activity was quenched by incubating slides in 0.3% hydrogen peroxide in methanol for 30 min at room temperature. Antigen retrieval was performed by incubating slides for 20 min in immunohistochemical select citrate buffer pH 6.0 (Millipore). Tissue sections were then preincubated in 10% (vol/vol) goat serum in PBS for 1 h at room temperature. Rabbit polyclonal anti–MERS-CoV M antibody (gift from Eric Snijder, Leiden University Medical Center, Leiden, Netherlands) was used at a concentration of 1:100 diluted in 10% (vol/vol) goat serum in PBS. Sections were incubated in primary antibody for 30 min at room temperature. Slides were washed before incubation with biotinylated secondary antibody from the ABC Elite Rabbit kit (Vector Labs) for 30 min at room temperature. Slides were again washed before addition of ABC reagent (Vector Labs) for 30 min at room temperature. After three changes of PBS, slides were developed using DAB (3,3-diaminobenzidine) for 2 min per slide. Development was stopped using a 5-min incubation in dH₂O. A 10-min incubation in 0.5% methyl green in 0.1 M sodium acetate pH 4.2 solution was used for counterstaining. Counterstain was cleared using two changes of PBS followed by two changes of N-butanol for 2 min each. Slides were dehydrated using two, 5-min incubations in xylenes before being mounted and imaged.

Neutralization experiments with pseudoparticles and live virus were performed in triplicates. Mice were randomly assigned to groups for each experimental cohort. Four animals per group were included for all in vivo experiments. RNA analysis was performed in triplicates and plaque assays in duplicates. In all graphs, values correspond to averages and error bars to SD. Histological analysis was performed in a blinded fashion. All statistical analyses were performed using one-way ANOVA and Tukey’s multiple comparison tests or the two-tailed Student t test. A P value < 0.05 was considered statistically significant. The data were analyzed using the GraphPad Prism online software package v5. For all figures, *P < 0.05, **P < 0.01, and ***P < 0.001.