Evaluation of Serologic and Antigenic Relationships Between Middle Eastern Respiratory Syndrome Coronavirus and Other Coronaviruses to Develop Vaccine Platforms for the Rapid Response to Emerging Coronaviruses

Viruses, Cells, and Plaque Assays

MERS-CoV Hu/England-N1/2012 and MERS-CoV Hu/SA-N1/2012 were cultured on Vero 81 cells and grown in Opti-Mem (Gibco, Carlsbad, CA) with 5% fetal clone serum (Hyclone, South Logan, UT) and gentamicin/kanamycin (Gibco). Viral growth assays in Vero and Calu-3 cells were performed as previously described [16].

Generation of Polyclonal Mouse Antisera, Neutralization Assays, and Western and Northern Blot Analysis

Genes encoding the indicated S and N proteins were synthesized from Bio Basic (Ontario, Canada) and packaged into Venezuelan equine encephalitis virus replicon particles (VRPs). Following vaccination, mouse polyclonal sera were generated from BALB/c mice, and neutralization assays involving MERS-CoV strains and SARS-CoV were as described previously [17]. For Western blots, VRP or virus-infected cell lysates and controls were prepared as described before in detail [8], and these blots were probed using the indicated mouse polyclonal sera. Vero cells inoculated with MERS-CoV isolates were harvested 12 hours after infection by means of Trizol reagent (Invitrogen) and were used to perform Northern blots [18].

Enzyme-Linked Immunosorbent Assay (ELISA) and Blocking Assay

An ELISA using indicated virus-infected cell lysates or antigens expressed from VRPs was performed as described previously [19], and the reactivity of mouse or human serum was determined using a chemiluminescent substrate. Blocking ELISA was performed by sequentially reacting plate-bound MERS-CoV lysate antigen with convalescent-phase serum obtained from a patient with MERS, followed by mouse polyclonal serum raised against VRP-packaged MERS-CoV N or S proteins. Blocking was expressed as the percentage reduction in the reactivity of the mouse serum alone.

Molecular Characterization of MERS-CoV Hu/England-N1/2012

MERS-CoV Hu/England-N1/2012 (MERS Eng 1) was isolated from a 49-year-old patient with severe respiratory illness and was transferred to London for treatment [7]. Twenty-nine mutations in MERS Eng 1 at the amino acid level were identified and compared to the published sequence of MERS-CoV Hu/SA-N1/2012 (MERS SA 1; GenBank JX869059.2; Supplementary Figure S2). To identify whether these mutations altered virus growth, we analyzed the replication kinetics of the 2 isolates in Vero cells and a continuous epithelial cell line, Calu-3 (Figure 1A and 1B). Although the replication kinetics were slightly different between 2 isolates in Vero cells, peak viral titers were equivalent. In contrast, virus growth was markedly distinct in human Calu-3 cells and could have represented differences in strain-specific in vitro adaptation phenotypes or resulted from functional differences in the sensitivity to innate immune responses [20].

CoV replication and transcription involves production of a nested set of subgenomic messenger RNAs (mRNAs) [2], and previous reports predicted 10 open reading frames in MERS SA 1 and MERS Eng 1 [9]. Northern blot analysis identified 8 subgenomic mRNAs after infection in both viruses (Figure 1C). The observed nested set of subgenomic mRNA expression is consistent with observations for other CoVs [2, 9]. The open reading frames encoded by each mRNA in these isolates are detailed in Supplementary Table 1.

Serologic Relationships Among MERS-CoV Strains

VRPs function as efficient expression and vaccine platforms for a variety of antigens [13, 21]. We generated VRPs expressing MERS SA 1 S and N proteins and then immunized mice. N protein–specific antiserum recognized a discrete 50-kDa band at the predicted molecular weight in lysates from Vero cells (Supplementary Figure 1A) and Calu-3 cells (Figure 1D) infected with VRP-N or with the 2 different MERS-CoV isolates. For the most part, similar expression patterns were evident between VRPs and viruses; however, the N protein of MERS Eng 1 had a slightly lower molecular weight, which was consistent with amino acid deletions at positions S391 and I392 (Supplementary Figure 2).

Mouse anti-S serum identified an approximately 180-kDa S protein in VRP-S– or MERS-CoV–infected Vero cells (Supplementary Figure 1A). The observed molecular weights were consistent with the sizes of the S proteins of other CoVs [2, 22]. We noted similar results in Calu-3 cells (Figure 1D), and interestingly, an increased amount of a higher-molecular-weight form of S protein (glycosylated dimer) was noted in MERS Eng 1 in both cell lines. Antiserum against MERS SA 1 S and N proteins also recognized MERS Eng 1 in ELISAs (data not shown), and N and S proteins recognized the MERS-CoV Jordan isolate (Supplementary Figure 1B).

Cross-neutralization Patterns Across Strains

Plaque reduction neutralization tests (PRNT₅₀) indicated complete neutralization of both MERS-CoV isolates (PRNT₅₀ titer, approximately 1:1400 for each; Figure 2A) and the MERS-CoV Jordan isolate (Supplementary Figure1C) by VRP-S antiserum, whereas no neutralization was observed with N antiserum. Similar findings have been reported with SARS-CoV, as well as with other known human and animal CoVs [13, 23]. Interestingly, serum from aged mice vaccinated with VRP S showed a 6-fold reduction in PRNT₅₀ titers (approximately 1:200), indicating that immunosenescence attenuates vaccine responses to MERS-CoV antigens, as was noted with SARS-CoV vaccines [13, 17]. Using serum from NA 01 patient, ELISA demonstrated high reactivity of the patient's serum to N and S antigens of MERS SA 1 expressed from VRPs (Figure 2B and 2C). Titers of antibody against N protein in patient serum peaked 3–5 weeks after onset of illness (which occurred on 3 September 2012) and waned thereafter, but the antibodies were still detected up to 5 months after illness onset. Titers of antibody against S protein were consistent from 3 weeks to 5 weeks after illness onset, after which they remained detectable. Most importantly, patient serum collected on 16 November 2012 (which contained high titers against S protein) outcompeted the binding of mouse S antiserum to intact virus in a blocking assay (Figure 2D). These data suggest that different/overlapping epitopes are recognized by human and mouse antisera following virus or VRP-S infection.

Cross-reactive and Cross-neutralizing Antibody Responses Within and Across Alphacoronaviruses and Betacoronaviruses

MERS SA 1 and MERS Eng 1 are closely related to BtCoV HKU 5 and BtCoV HKU 4 (Figure 2E) [6, 9, 24]. To evaluate antigenic relationships with the subgroup 2c betacoronaviruses, VRPs expressing S and N proteins of BtCoV HKU 4.2 and BtCoV HKU 5.5 were inoculated into mice. Antisera against both HKU 4.2 and 5.5 N proteins recognized the N proteins of both MERS-CoV isolates, whereas MERS SA 1 N antisera also detected the VRP-expressed HKU 4.2 and 5.5 N proteins, as revealed by Western blot (Figure 3A and 3B). We obtained similar results using ELISA and immunofluorescence assays (data not shown). In contrast, there was little if any observable cross-reactivity observed between MERS SA 1 S antisera with the VRP-expressed S proteins of HKU 4.2 and HKU 5.5, whereas antisera to HKU 5.5 S protein but not HKU 4.2 S protein recognized the S proteins of both MERS-CoV isolates (Figure 3C and 3D). We also measured serologic relationships using ELISA, which captures cross-reactivity to conformational epitopes, and confirmed these antigenic relationships (Figure 6C). Consistent with results of serologic tests, antisera against HKU 4.2 and HKU 5.5 S proteins did not cross-neutralize the MERS-CoV isolates. These data indicate that the N protein but not the S glycoprotein are antigenically conserved within the subgroup 2c betacoronaviruses evaluated in this panel.

We then extended our analysis to the highly pathogenic SARS-CoV and related subgroup 2b betacoronaviruses. Polyclonal mouse sera to SARS-CoV or MERS SA 1 N or S proteins exhibited no cross-reactivity to the reciprocal strains (Figure 4A and 4B). We observed very low levels of cross-neutralization of MERS SA 1 by mouse antisera to SARS-CoV, using very high but not low concentrations of serum (Figure 4C), a finding that is consistent with a recent report [24]. Interestingly, ELISA results also showed very minimal cross-reactivity of the NA 01 patient sera obtained on 23 September 2012 to SARS-CoV S antigen (Figure 4D). Consistent with this observation, binding of mouse SARS-CoV S antiserum to SARS-CoV was not inhibited by NA01 patient sera in blocking assays (Figure 2D), indicating the absence of antibodies to SARS-CoV in the patient serum.

Consonant with these findings, no cross-reactivity was observed with antisera against the VRP-expressed N or S glycoproteins of BtCoV HKU 3 and 279 and the MERS-CoV isolates (Figure 5A–D). Furthermore, no cross-neutralization of MERS-CoV isolates by HKU 3S antiserum was observed, although this serum has previously been shown to neutralize a synthetically resurrected HKU 3 variant encoding the SARS S glycoprotein receptor binding domain [23]. Interestingly, we observed very low levels of cross-neutralization of SARS-CoV by BtCoV 279 S antiserum (Figure 4C).

To further elucidate the antigenic relationships between the S glycoproteins of alphacoronaviruses and the MERS-CoV isolates, we expressed and generated mouse antisera to BtCoV 1A and BtCoV HKU 2 (group 1b alphacoronaviruses), using the VRP platforms. Despite efficient recombinant S glycoprotein expression (Figure 7A and 7B), none of the recombinant S glycoproteins were recognized by MERS SA 1 S antisera. Antisera against BtCoV1A and HKU 2 S glycoproteins had little if any cross-reactivity with and did not neutralize MERS-CoV (Figure 7A–C).

Antigenic Relationships Among the HCoVs

We next analyzed the antigenic relationships between VRP-derived mouse serum with the following representative HCoVs from each subgroup, using ELISA (Figure 6C): MERS Eng 1, from subgroup 2c; SARS-CoV, from subgroup 2b; HCoV-NL63, from subgroup 1b; and HCoV-OC43, from subgroup 2a. MERS Eng 1 was recognized by antisera targeting the N but not the S glycoprotein of viruses within the subgroup 2c betacoronaviruses. Likewise, SARS-CoV was only recognized by antisera to N but not S glycoproteins of viruses with the subgroup 2b betacoronaviruses. None of the antiserum screened reacted with HCoV-NL63 (subgroup 1b) or HCoV-OC43 (subgroup 2a). Although BtCoV HKU 2 is genetically close to HCoV-NL63, we did not observe any cross-reactivity within the S glycoprotein.

Serum from patients infected with SARS-CoV, HCoV-NL63, or HCoV-OC43 was screened against the N proteins from representative subgroup 2c and 2d betacoronaviruses. Consistent with our previous findings, human serum to SARS-CoV recognized BtCoV HKU 3N, BtCoV 279N, and SARS-CoV N (subgroup 2b) but did not recognize N proteins from other subgroups (Figure 6B). Similarly, there was no cross-reactivity of the human antisera from HCoV-NL63 (subgroup 1b) and HCoV-OC43 (subgroup 2a) infections with any of the viral antigens within the panel. Serum collected from the patient infected with MERS-CoV NA01 showed cross-reactive binding only to BtCoV HKU 5.5 N (subgroup 2c), and little if any cross-reactivity was noted outside the subgroup (Figure 6A), apart from very low, transient cross-detection of BtCoV 279 N and from cross-reactivity to SARS-CoV S and N recombinant proteins on a single day (23 September 2012).