Fig 1.
HCMV gB is composed of five structural and antigenic domains.
(A) Domain map of full-length HCMV gB. Division of domains indicated by color and residue number. Unlabeled white boxes indicate unresolved regions. The membrane-proximal region (MPR) is shown in magenta. The transmembrane region (TM) is shown in brown. The cytoplasmic domain (CTD) is shown in light purple. Fusion loop sites are shown as light blue stars in dI and labeled accordingly. The furin cleavage site is represented by a dark purple line between dII and dIII and is labeled accordingly. (B) Ribbon model of prefusion HCMV gB with one monomer colored by the domain (RCSB 7KDP) with structural and antigenic domains (AD) indicated. Fusion loop locations are marked with light blue stars. Light purple lines represent the CTD. (C) Ribbon model of postfusion HCMV gB with one monomer colored by domain (RCSB 5CXF). MPR and TM are represented by magenta and brown cylinders, respectively. Fusion loop locations are marked with light blue stars. Light purple lines represent the CTD. Each structural domain with its corresponding antigenic domain (AD) is indicated.
Fig 2.
Design iterations to stabilize prefusion gB resulted in constructs representing three known gB conformations.
(A) Domain map of construct pEW2 with mutation positions indicated. 7M mutations are indicated in cyan. (B) Negative stain electron microscopy (NS-EM) 2D class of conformations present in pEW2 sample. Conformation ratio as calculated by 2D classification particle counts with 12,489 particles total. (C) 3D reconstruction of pEW2 (1623 particles) overlaid with the structure of the known postfusion HCMV gB ectodomain (RCSB: 5CXF). (D) Domain map of construct pEW21 with mutation positions indicated. GTS mutations are indicated in gray. (E) Negative stain electron microscopy (NS-EM) 2D classes of conformations present in the pEW21 sample. (right) Conformation ratio as calculated by 3D classification particle counts. 5071 particles were recognizably postfusion, and 11,026 particles were deemed intermediate. (F) 3D reconstructions of pEW21 overlaid with the structure of the known postfusion HCMV gB ectodomain (RCSB: 5CXF) (5071 particles) or as an empty map (2929 particles). Additional intermediate 3D classes can be found in S1 Fig. (G) Domain map of construct pEW62 with mutation positions indicated. eDDR2 mutations are indicated in violet and disulfides in plum. (H) Negative stain electron microscopy (NS-EM) 2D classes (top) of conformations present in pEW62 sample. (Bottom) Conformation ratio as calculated by 2D classification particle counts. pEW62 contained 32,447 prefusion particles and 6,452 postfusion particles. (I) 3D reconstruction of conformations in pEW62 sample (prefusion = 8936 particles, postfusion = 2358 particles). Reconstructions are overlaid with known structures of prefusion HCMV gB ectodomain (RCSB 7KDP) and postfusion HCMV gB ectodomain (RCSB 5CXF). Locations of pEW62 disulfides are indicated with plum stars. Additional mutants made in attempts to stabilize prefusion gB can be found in S1 Table. Additional 2D classes of pEW2, pEW21, and pEW62, as well as additional 3D classes for pEW21 can be found in S1 Fig.
Fig 3.
Immunization with gB variants does not elicit neutralizing antibodies in mice.
(A) HCMV gB immunization scheme and timeline. Bleeds were collected retro-orbitally at the indicated time points. (B) Neutralization of sera from all immunized mice against AD169GFP. Sera from mice, n = 7 (pEW62) or n = 5 (pEW21, pEW2), were tested individually in technical duplicate. Sera from naïve mice were used as a negative control. Sera from naïve mice with known nAb 14-4b were used as a positive control. 14-4b was added to the sera at a starting concentration of 100 μg/mL and serially diluted with the sera as indicated. Symbols represent the mean, and error bars represent the standard deviation of each group, n = 7 (pEW62), n = 5 (pEW21, pEW2), n = 3 (negative control, positive control) in B-D. (C-E) Half-maximal effective concentrations (EC50) of serum antibody binding endpoint titers to (C) pEW62, (D) pEW21, and (E) pEW2 as measured by ELISA. Samples were grouped by immunization group and are colored in accordance with the color scheme defined in (A). Sera from naïve mice were used as controls. ** indicates P < 0.01, and *, P < 0.05, as determined by Mann-Whitney tests. Stars are colored in accordance with the groups being compared, as shown in (D). Fig 3a. Immunization Scheme. Created in BioRender. McClave, M. (2025) https://BioRender.com/yipgib3.
Fig 4.
mAb isolation and binding assessment to all conformations of gB.
(A) mAb isolation strategy. Enriched human B cells were stained with fluorescently labeled pEW62. The antibody variable heavy and light chain transcripts were recovered by RT-PCR, cloned into expression vectors, and produced as recombinant mAbs. For full gating strategy, see S3 Fig. (B) Binding avidity of the indicated mAbs to pEW62, pEW21, or pEW2 at a single concentration was measured by biolayer interferometry using isolated antibodies immobilized on biosensors and indicated gB construct as the ligand. Each data point represents the maximum binding response (nm shift) response of each mAb to a 40 µg/mL solution of each of the three constructs after 150 seconds. Data are representative of one independent experiment. (C) Percent gB detected on cells. Cells were transfected with AD169 gB, permeabilized, and tested for % gB staining by the indicated mAb. Each point represents an independent transfection. Bar heights represent the mean, and the error bars represent the standard error of the mean (SEM). For full gating strategy, see S4 Fig. Fig 4a. mAb isolation strategy. Created in BioRender. McClave, M. (2025) https://BioRender.com/bzak0d0.
Fig 5.
Seven antibodies neutralize at least one HCMV strain.
HCMV with GFP transgene (AD169GFP MOI = 30, TRGFP & TB40/eGFP MOI = 5) was incubated alone or with the antibody of interest for two hours before being added to confluent HFFs. Cells were then incubated with the mAb/virus mixture for 18 hours (AD169GFP) or 48 hours (TRGFP and TB40/eGFP), then washed and fixed with 4% PFA. The plate was read in a microplate reader for bulk GFP intensity. Half-maximal inhibitory concentration (IC50) values for antibodies identified as neutralizing for one or more strains (AD169GFP orange, TRGFP green, TB40/eGFP purple). nAbs were titrated against each indicated strain of HCMV, and IC50 was calculated from the resulting linear regression (see S6 Fig). Each data point is an average of three technical replicates. Each experiment was done in triplicate. Gray bars were used to denote that an IC50 for an indicated strain/mAb was above the limit of detection of our assay (82 μg/mL).
Table 1.
IC50 values for the nAbs tested in this study.
Fig 6.
Antibodies show diverse preferences for different gB constructs.
(A) Monovalent affinities of the indicated Fabs to pEW62, pEW21, and pEW2 were measured by BLI. KDs were calculated from multiple measurements of serially diluted Fab. For full kinetic curves, see S2 Fig. Data for 1G2 and SM5-1 are the same as in S1G Fig. MLCB12 did not bind any of the constructs in Fab form and thus was excluded from analysis. (B) Fold-change differences in Fab affinity across gB constructs. To visualize relative shifts in binding strength, for each Fab, the KD values in (A) were transformed into fold-change differences by normalizing the KD values for pEW62 and pEW21 to that of pEW2. Relative affinity for pEW62 is shown in purple, relative affinity for pEW21 in gray, corresponding to coloring of constructs in panel (A). Fold-change was calculated as (KDpEW2)/(KDpEW62 or KDpEW21). Values >1 indicate increased affinity relative to pEW2, values <1 indicate reduced affinity relative to pEW2. For KD values, see S2 Table.
Fig 7.
Neutralizing antibodies bind to HCMV gB AD-4 and AD-5 and form five distinct clusters.
(A) BLI competition assay. Biotinylated mAbs were immobilized on streptavidin biosensors (y-axis) and then immersed in buffer containing pEW62 with or without the indicated non-biotinylated mAb (x-axis). The percent binding inhibition is shown as a heatmap. (B-F) Models of the prefusion gB ectodomain bound to the corresponding Fab were generated and manually fitted into the NS-EM reconstructions of pEW62/Fab complexes as described in the Materials and Methods. For representative particle images, see S9 Fig. Complexes were subdivided into clusters based on competition assay data in (A). The pEW62/MLCB10-Fab complex in (D) could not be conclusively assigned to AD-4 or AD-5, due to map symmetry and lack of competitors. For detailed information on map and model fitting, see Materials and Methods.
Fig 8.
Not all neutralizing antibodies can access their epitopes on the cell surface.
(A) Percent gB detected on the surface. Cells were transfected with AD169 gB and tested for percent gB staining by the indicated mAb. Each point represents an independent transfection. Bar heights represent the mean, and the error bars represent SEM. See S11 Fig for data on individual replicates. MLCB12 could not be calculated due to a lack of binding. (B) Percent gB detected on the surface of cells infected with AD169GFP. Cells were infected with AD169GFP at an MOI of 3 and stained with the indicated mAb. Results indicate infected cells (GFP+) that were positive for each mAb. Each point represents an independent infection. Bar heights represent the mean, and the error bars represent SEM. For full gating strategy, see S12 Fig. Antibodies are colored according to their binding sites in Fig 7.
Fig 9.
Putative mechanisms of nAbs described in this study.
Schematic illustrations representing prefusion, early intermediate, extended intermediate, and postfusion HCMV gB are shown to illustrate possible binding locations of nAbs. Domains of gB are colored in accordance with the color scheme defined in Fig 1. Fusion loops are represented by gray pentagons on the tips of dI. Domains bound by nAbs of a particular class are additionally labeled for clarity. Members of each class were determined by combining data from competition assay, cell surface staining, NS-EM reconstructions, and affinity measurements. (A) Class α consists of MLCB1, shown in orange square on the left. (B) Class β consists of the majority of the nAbs isolated here, MLCB5, MLCB6, MLCB7, and MLCB11, shown in berry-colored box on the left. Red prohibition circle on extended intermediate form represents the inhibition of membrane binding or the refolding of the intermediate into the postfusion form. (C) Class γ consists of MLCB10, SM5-1, 1G2, along with MLCB9. These are shown in a purple box on the left. Blue and green antibody borders represent dI and dII binders, respectively. Created in BioRender. McClave, M. (2025) https://BioRender.com/xydn9a1.