Antigenic Drift Defines a New D4 Subgenotype of Measles Virus

During our studies on antigenic variation between MeV genotypes, we noted that two MeV strains isolated in Spain and defined as belonging to genotype D4 showed distinct sensitivities to antibody-mediated neutralization. To ascertain a role of substitutions in MeV-H, we generated enhanced green fluorescent protein (EGFP)-expressing recombinant MeV in which the MeV-H gene was replaced with that of two Spanish genotype D4 isolates, Barcelona.SPA/26.08 and Madrid.SPA/10.10/1. We again tested their neutralization sensitivities with a panel of anti-MeV-H antibodies against known major antigenic sites in MeV-H (Fig. 1A). The results presented in Fig. 1B show that the BH026, BH141, and BH101 MAbs completely inhibited infection by both genotype D4 viruses and the vaccine strain (genotype A), used as a control for neutralization. In contrast, discrimination between the MeV vaccine strain and the genotype D4 viruses was observed when the BH125 and BH097 MAbs were tested instead. Surprisingly, the two genotype D4 viruses showed markedly different neutralization profiles when the NE-targeting BH129 MAb was used. Although the BH129 MAb completely blocked infection by the Barcelona.SPA/26.08 virus, it was ineffective against the genotype D4 Madrid.SPA/10.10/1 virus.

Having shown that the two genotype D4 Spanish isolates showed differing neutralization patterns by anti-MeV-H MAbs, we sought to determine the broader significance of this finding for other members of genotype D4. A new MeV genotype is typically defined by 2.5% and 2.0% nucleotide divergences in the MeV-N and MeV-H genes, respectively (85). We therefore retrieved both MeV-N and MeV-H gene sequences available for MeV isolates classified as belonging to genotype D4 from GenBank (n = 41) (Table 1). After removing redundant sequences, we reconstructed phylogenetic trees for both genes and rooted them on the reference strain for genotype A (MVi/Maryland.USA/0.54). Phylogenetic analysis showed that the two Spanish genotype D4 viruses were indeed representative of two definable subgenotypes, which were supported by high bootstrap values (Fig. 2). The average number of base substitutions per site for the MeV-N gene between the 2 subgenotypes was 4.0% (±0.8%), ranging from 2.5% to 6.5%. The nucleotide sequence divergence for the MeV-H gene was 2.3% (±0.2%) (range, 1.5% to 3.0%). These results show high intragenotypic variation for the sequences assigned to genotype D4 but do not support the designation of 2 different genotypes. Here, we distinguish subgenotype 1 (D4.1) and subgenotype 2 (D4.2).

MeV-neutralizing antibodies are directed mainly against the MeV-H protein (16). Because of our identification of two subgenotypes in MeV genotype D4, we further investigated the differences observed in the MeV-H gene at the level of the resulting protein.

The predicted amino acid differences between the MeV-H proteins of the two subgenotypes were 2.0% (±0.2%), ranging from 1.2% to 4.0%. Of note, both subgenotypes were characterized by 2 well-defined consensus sequences with variations at positions 212, 240, 249, 303, 316, 359, 412, and 481 (Fig. 3). These residues, together with residues 305 and 562, were also identified as directionally evolving sites through convergent evolution under a directional evolution of protein sequences test (Bayes factor, >10⁵) (86). Likewise, 5 pairs of interactions were detected (P < 0.05): N238T←→P276L, R34K←→T307A, F476L←→S590T, M333L←→I564L, and S318N←→F571S. All sites except position 571 were exposed on the protein surface (not shown). However, none of them showed close proximity in the two crystallographic forms resolved so far for MeV-H (form I and form II [not shown]) (39).

These results prompted us to analyze selection pressures on the MeV-H gene among genotype D4 viruses. To estimate positively and negatively selected sites, we used the following four methods: single-likelihood ancestor counting (SLAC), fixed-effects likelihood (FEL), internal fixed-effects likelihood (IFEL), and mixed-effect model of evolution (MEME). Four positively selected sites were estimated by using the MEME method (Table 2), which detected residues under both pervasive positive selection and episodic selection (40): E85Q, L172M, V562F/A, and Q574K. In addition, 15 sites were under negative selection, 5 of which were common to all four methods (Table 2): 3P, 113N, 256E, 342D, and 347D. Except for sites at positions 172 and 256, they were located on the protein surface (not shown). We also calculated the combined synonymous and nonsynonymous rate to be 0.22 (95% confidence interval [CI], 0.17 to 0.29) with the SLAC method. Overall, these results establish a basis for the distinction of D4 subgenotypes.

Apart from the BH129 MAb, the BH047 and BH059 MAbs have also been shown to bind a sequential epitope encompassing amino acids 240 through 250 of MeV-H (41). To confirm that antigenic drift of the NE site defines D4 subgenotypes, we studied the neutralization capacity of the other NE-targeting MAbs. We additionally tested different recombinant MeVs encoding the genotype-specific MeV-H genes of genotypes B3.1, C2, D4, D6, D7, D8, D9, G3, and H1 (Fig. 4A). As expected, none of the NE-targeting MAbs (BH129, BH047, and BH059) were able to inhibit infection by subgenotype D4.2 viruses, whereas all of them inhibited viruses of subgenotype D4.1. Surprisingly, these antibodies did not inhibit infection by recombinant MeV possessing the MeV-H gene of genotype D6, although all other genotypes were neutralized.

To understand in greater detail the neutralization selectivity of these NE-targeting antibodies, we aligned and compared the MeV-H amino acid sequences for all the viruses that we used (Fig. 4B). Whereas the NE antigenic site is one of the most variable regions of MeV-H, NE-targeting antibodies discriminated only the D4.2 and D6 genotypes. Subgenotype D4.2 viruses showed the single mutation L249P, whereas genotype D6 viruses also showed an S247P mutation. To further elucidate the role of these amino acid changes found within the NE antigenic site, we introduced each of the two single point mutations into MeV-H genotype A and rescued the corresponding viruses. Figure 4C illustrates that an L249P amino acid substitution completely changed the neutralization sensitivity of antibodies targeting the NE, yet the sensitivity of genotype A virus to antibodies remained unchanged with and without the single point mutation S247P. These observations demonstrate that a single point mutation, L249P, but not S247P, allows MeV to avoid neutralizing antibodies targeting the NE.

Virus neutralization requires a certain level of electrostatic and shape complementarity between the viral antigen and neutralizing antibody (42). Since 2 different mutations were found in the NE antigenic site, we further investigated their effect on virus-antibody interactions. To this end, we first produced a soluble form of the MeV-H protein that displays a C-terminal FLAG epitope. This tag enables the control of the amount of protein coating the surface of the plastic well. As expected, a FLAG antibody equally recognized all FLAG-tagged MeV-H proteins (Fig. 4D). However, NE-targeting antibodies showed defects in binding to those MeV-H proteins whose viruses escaped neutralization. Overall, these results show that an L249P substitution, but not S247P, impairs the binding of neutralizing antibodies that recognize the NE antigenic site, thereby allowing virus neutralization escape.

The data presented in Fig. 1 show that a naturally occurring MeV strain can escape neutralization by MAbs targeting 3 of 6 antigenic sites of the MeV-H protein. Because 90% of the antibodies of MeV patients are directed against the MeV-H protein (17), we next assessed whether the mutation of the NE antigenic site in the subgenotype D4.2 MeV-H protein translates into a different serum neutralization pattern than those for subgenotype D4.1 and vaccine genotype A viruses.

To obtain a representative sample for population immunity against measles, we used pooled human serum collected from approximately 60 to 80 blood type AB donors in North America. To rule out potential interference by MeV-F-specific antibodies, we first absorbed MeV-F-specific antibodies from pooled human sera using MeV-F of genotype A, the F protein expressed by all of our recombinant MeVs. This depletion was performed by incubation with MeL-JuSo cells expressing the MeV-F protein. Following this depletion procedure, the remaining measles neutralization activity of the depleted serum is mediated exclusively by anti-MeV-H antibodies. Figure 5A shows that incubation with untransfected MeL-JuSo cells results in no depletion of MeV-H/F-specific antibodies, whereas incubation with MeL-JuSo/MeV-F cells results in the elimination of MeV-F-specific antibodies while maintaining the MeV-H-specific antibodies. As an additional control for the presence of residual MeV-F-specific antibodies due to the limit of detection of the fluorescence-activated cell sorter (FACS) analysis, we used an envelope-chimeric MeV/canine distemper virus (CDV) (M. A. Muñoz-Alía and Stephen J. Russell, unpublished data). This envelope-chimeric virus expresses the MeV-F protein in combination with the Onderstepoort strain CDV-H protein possessing a Y537D point mutation (43). This envelope-chimeric MeV-F/CDV-H showed complete resistance to neutralization by MeV-F-depleted pooled human sera (Fig. 5B). Neutralization assays using recombinant MeVs expressing the two D4 subgenotype-specific MeV-H proteins showed a diminished sensitivity for subgenotype D4.2 (50% inhibitory concentration [IC₅₀] = 33.00 ± 3.50 milli-international units [mIU]/ml) compared to subgenotype D4.1 (IC₅₀ = 30.93 ± 4.31) or the genotype A vaccine strain (IC₅₀ = 30.05 ± 3.35) (P = 0.21, according to a paired t test). Therefore, the absence of the NE in subgenotype D4.2 MeV-H was associated with a definite reduction (albeit nonsignificant) in neutralization sensitivity compared to those of subgenotype D4.1 and vaccine genotype A viruses.

B95a cells (an adherent marmoset B-cell line) were maintained in Roswell Park Memorial Institute medium (RPMI 1640; Corning) supplemented with 10% (vol/vol) heat-inactivated (56°C for 30 min) fetal bovine serum (FBS) (Gibco). African green monkey kidney (Vero) cells stably expressing human SLAM (65) were maintained in Dulbecco's modified minimal essential medium (DMEM) (HyClone; GE Healthcare Life Science) containing 5% FBS and 0.5 mg of Geneticin ml⁻¹ (G418; Corning). The human melanoma cell line Mel-JuSo/wt, transfected or not with the Edmonston MeV-H (Mel-JuSo/MeV-H) or MeV-F (Mel-JuSo/MeV-F) protein, was described previously (16, 66) and cultured in RPMI 1640. In antibody depletion assays, the medium was replaced with human serum diluted 1:10 in RPMI medium, and cells were cultured for 4 days at 37°C in a humidified atmosphere of 5% carbon dioxide (17). All cell culture media were additionally supplemented with 100 IU penicillin ml⁻¹, 100 μg streptomycin ml⁻¹, and 10 mM HEPES (Gibco). All cells routinely tested negative for mycoplasma contamination.

Growth of the Edmonston strain (67) and the MeV primary isolates was performed as described previously (60). The wild-type MeVs used in this work were MVi/Madrid.SPA/15.06/3 (subgenotype B3.1) (23), FV (genotype C2) (60), MVi/Barcelona.SPA/26.08 (genotype D4), MVi/Madrid.SPA.10.10/1 (genotype D4), BCL (genotype D6) (60), MVi/Madrid.SPA/25.03 (genotype D7), MVi/Alicante.SPA/22/03 (genotype D8), MVi/Granada.SPA/32.08 (genotype D9), 65045397 (genotype G3) (68), and MVi/Madrid.SPA/50.10 (genotype H1). Viral stocks were prepared by repeated cycles of freezing and thawing and were frozen in aliquots at −80°C. The MeV titer was determined as PFU by using Vero/hSLAM cells.

Monoclonal antibodies and human sera were heat inactivated and serially diluted in Opti-MEM (Thermo Fisher Scientific). Equal volumes of the respective virus at 30 PFU per well were mixed with the respective antibodies at various 2-fold serial dilutions in 96-well plates (Costar Corp.), incubated at 37°C for 1 h, and inoculated onto 80% to 90% confluent Vero/hSLAM cells. The number of EGFP-expressing syncytia per well was counted under a fluorescence microscope after 2 days of culture and converted to a percentage in relation to infection in the absence of antibodies (100%). Each sample was run in 4 wells, and each experiment was repeated at least twice on different days.

The following anti-MeV-H protein MAbs used for neutralization were previously described: BH129, BH047, BH059 (69), BH101, BH026 (70), BH141 (71), and BH097 (72). Protein G affinity-purified MAbs were used, except for BH141 and BH059. BH141 and BH059 antibody concentrations were determined with a mouse isotype-specific immunoglobulin G (IgG) enzyme-linked immunosorbent assay (ELISA) quantitation set (Bethyl Laboratories).

Human serum used in this study was commercially purchased (product no. HP1022, lot no. C80553; Valley Biomedical). This serum represents a pool of 60 to 80 blood type AB healthy donors (communication with the supplier). The antibody concentration that causes a 50% reduction in plaque numbers (IC₅₀) was determined by nonlinear regression using software (Prism version 7.00 for Mac; GraphPad Software). The titers were transformed to international units per milliliter by using National Institute for Biological Standards and Control (NIBSC) reference serum (WHO/BS/06.2031), according to procedures described previously (73).

RNA was extracted from infected B95a cells by using an RNeasy minikit (Qiagen) according to the manufacturer's instructions. A hypervariable region of the MeV-N gene and the entire region of the MeV-H gene were reverse transcribed by using 200 U of reverse transcriptase (SuperScript II; Invitrogen) and primers (Integrated DNA Technologies) MVN(+)958 (TTGGACTGCATGAATTTGCTGG) and V181 (TTAATTAAAACTTAGGGTGCAAGATCATCGATAATG), respectively. The resulting cDNA was amplified by PCR (HotStarTaq DNA polymerase; Qiagen) with primer pairs V181/H1931 (CACTAGTGGGTATGCCTGATGT) and MVN(+)958/MVP(−)3355 (GGTTGGCAGGTAAGTTGAGCT). Amplified DNA fragments were analyzed by electrophoresis in a 0.7% agarose gel and purified from the gel for sequencing (ABI Prism 3730xl DNA analyzer; Perkin-Elmer Applied Biosystems) by using a gel extraction kit (QIAquick; Qiagen). The entire MeV-H gene flanked with SpeI/PacI restriction sites (underlined in the primer sequences) was cloned into the equally restricted pCG vector (New England BioLabs Inc.) (74). At least 4 clones were sequenced, and the direct sequences of the PCR products were used to define the consensus sequences. Only those clones with the consensus sequence were transferred into an infectious MeV genome expressing EGFP upstream of N [MVvac2(GFP)N], using PacI and SpeI restriction sites. The MVvac2(GFP)N coding capacity is identical to those of the Moraten/Schwartz vaccine strains (75). Recombinant viruses were rescued by using standard protocols with modifications (76, 77). The integrity of the insert was verified by sequencing, and the corresponding virus was renamed according to the genotype of the MeV-H gene. The sequence of the MeV-N gene was determined directly from the PCR products.

The soluble MeV-H protein was produced by amplification of the MeV-H ectodomain (residues 61 to 617) preceded by the murine IgG κ-chain leader and a Strep-tag II epitope. This construct was cloned in frame with the C-terminal FLAG tag present into the MluI- and EcoRV-restricted pCMV6-AC-IRES-GFP vector by using an InFusion HD kit (Clontech Laboratories Inc.). Expi293F cells (Gibco) were transiently transfected and grown in suspension for 6 days. The culture supernatant was harvested every 3 days, filtered through a 0.45-μm-pore-size filter, concentrated by using a centrifugal filter device (Amicon Ultra 50K; EMD Millipore Corp.), and buffer exchanged with HEPES buffer (10 mM HEPES, 150 mM sodium chloride [pH 7.5]).

MeV-H- and MeV-F-specific IgG antibody levels were determined as described previously (16, 17, 66), with the following modifications: RPMI 1600 medium supplemented with 2% FBS and 2-mM EDTA (catalogue no. 155575; Thermo Fisher Scientific) was used as a FACS buffer, and cells were stained with human IgG(H+L) Alexa Fluor-488.

Coated microplates (Strep-Tactin XT; IBA GmbH) were incubated at 37°C for 1 h with 0.1 ml of soluble MeV-H protein, previously diluted to give a final optical density of 1 under the conditions described above. The plates with the MeV-H protein absorbed through its amino-terminal Strep-tag II sequence were washed 3 times with phosphate-buffered saline (PBS) containing 0.05% Tween 20 (PBS-Tween), and consequently, 0.1 ml of antibody in PBS (5 μg/ml) was added. In parallel, mouse anti-FLAG M2 antibody (Sigma-Aldrich) was included as a control for the absorbed protein. The plates were incubated for 1 h at 37°C and washed with PBS-Tween as described above. The mixtures were then allowed to react with 0.1 ml of goat anti-mouse IgG(H+L)-horseradish peroxidase (HRP) (catalogue no. 62-6520; Thermo Fisher Scientific) in PBS-Tween at a dilution of 1:2,000 to 1:4,000 for 1 h at room temperature. The peroxidase bound to the wells was detected by incubation with 1-Step Ultra TMB (3,3′,5,5′-tetramentylbenzidine; Thermo Fisher Scientific) according to the manufacturer's protocol. The Epstein-Barr virus VCA IgG titer was assessed by using a commercial ELISA as recommended by the manufacturer (catalogue no. 57351; IBL International GmbH).

Sequence data available for MeV-N and MeV-H of genotype D4 MeV strains were retrieved from GenBank and analyzed by using BioEdit version 7.0.9.0. Multisequence alignments were performed with ClustalX (http://www.genome.jp/tools/clustalw/), and identical sequences were removed from the data set. Phylogenetic reconstructions were inferred by the maximum likelihood method on the basis of the general time-reversible model available in MEGA version 6.06 (78). To estimate the robustness of the phylogenetic inference, we performed 1,000 bootstrap analyses. Estimation of the evolutionary divergence among sequences was conducted by using the Kimura 2-parameter model in MEGA version 6. The codon positions included were the first plus second plus third plus noncoding positions.

The selective pressure acting on the MeV-H gene was investigated by using 4 different methods: SLAC, FEL, IFEL, and MEME (all available in the HYPHY package on a Web-based interface [DataMonkey]) (79).

An MeV-H model was generated from the crystallographic structure of the purified protein alone (PDB accession no. 2ZB5) and in complex with SLAM (PDB accession no. 3ALZ). A complex-type N-glycan model was constructed manually by using a complex penta-antennary glycan whose coordinates were obtained online (see http://www.glycosciences.de/). Crystallographic structures were analyzed and modeled by using PyMOL software (http://pymol.org/).

The nucleotide sequences determined in this work were deposited in the GenBank database under accession no. KY524301to KY524304. A list of nucleotide sequences used in the course of this work can be found in Table 1.