Crystal Structure of the Human Cytomegalovirus Glycoprotein B

Abstract

Human cytomegalovirus (HCMV), a dsDNA, enveloped virus, is a ubiquitous pathogen that establishes lifelong latent infections and caused disease in persons with compromised immune systems, e.g., organ transplant recipients or AIDS patients. HCMV is also a leading cause of congenital viral infections in newborns. Entry of HCMV into cells requires the conserved glycoprotein B (gB), thought to function as a fusogen and reported to bind signaling receptors. gB also elicits a strong immune response in humans and induces the production of neutralizing antibodies although most anti-gB Abs are non-neutralizing. Here, we report the crystal structure of the HCMV gB ectodomain determined to 3.6-Å resolution, which is the first atomic-level structure of any betaherpesvirus glycoprotein. The structure of HCMV gB resembles the postfusion structures of HSV-1 and EBV homologs, establishing it as a new member of the class III viral fusogens. Despite structural similarities, each gB has a unique domain arrangement, demonstrating structural plasticity of gB that may accommodate virus-specific functional requirements. The structure illustrates how extensive glycosylation of the gB ectodomain influences antibody recognition. Antigenic sites that elicit neutralizing antibodies are more heavily glycosylated than those that elicit non-neutralizing antibodies, which suggest that HCMV gB uses glycans to shield neutralizing epitopes while exposing non-neutralizing epitopes. This glycosylation pattern may have evolved to direct the immune response towards generation of non-neutralizing antibodies thus helping HCMV to avoid clearance. HCMV gB structure provides a starting point for elucidation of its antigenic and immunogenic properties and aid in the design of recombinant vaccines and monoclonal antibody therapies.

Fig 1. HCMV gB ectodomain structure.

(A) Schematic representation of the full-length HCMV gB (top) and the crystallized construct, gB78-706-7M-E (bottom). Disulfide bonds are represented as black brackets, antigenic domains (AD-1-5) are indicated in blue brackets, and mutations are shown using red bars. Structural domains, are colored as follows: domain I = blue, II = green, III = yellow, IV = orange, V = red, as in [16, 17]. SS = signal sequence, MPR = membrane proximal region, TM = transmembrane domain, and Cyto = cytoplasmic domain. Numbers denote construct boundaries. (B) The crystal structure of the HCMV gB ectodomain is shown as a protomer and a trimer in cartoon representation as well as a trimer in surface representation. Chain B is colored by domain as in (A). (C) Side and top down view of the coiled coil in DIII with a coordinated calcium ion (CA) (blue sphere). Side chains of D508 (yellow) with carboxyl oxygens (red) and C506 (green) are also shown. Dashed lines indicate distances between the carboxyl oxygens in D508 and the calcium ion. All structure figures were made in Pymol (http://www.pymol.org).

Fig 2. Multiple sequence alignment of gB homologs.

Protein sequence alignment of HCMV (strain AD169), EBV (strain B958), and HSV-1 (strain KOS) (UniProtKB accession numbers P06473, Q777B0, P06437, respectively) was generated and analyzed using ClustalW2 [32] and ESPript 3.x [33]. The secondary structure of HCMV is displayed above the alignment and the secondary structure of HSV-1 (2GUM) [16] is displayed below the alignment. Secondary structure elements are colored by domain, as in Fig 1. Unresolved residues are denoted with dashed lines, signal sequences as orange letters and residues in fusion loops as blue letters. Cysteines participating in disulfide bonds are highlighted in green, furin cleavage sites, in pink, glycosylation sites, in cyan. First residue in the crystallized HCMV gB construct is highlighted in brown. Identical residues are shown as white text on red background, similar residues are shown as red text.

Fig 3. Structures of HCMV, HSV-1, and EBV gB ectodomains.

The structures of HSV-1 (2GUM), HCMV (5CXF), and EBV (3FVC) gB ectodomains are aligned on the central trimeric helices and are displayed side-by-side as (A) trimers in surface and (B) protomers in cartoon representations. The domains I-V of each homologue are colored as in [16, 17]. (C) Single protomers (chain B in HSV-1 and HCMV) are aligned on the core helix α4. DIV and DIII are enlarged and shown from the side and top down. Red arrows indicate differences in the domain placement of the homologues, while black arrows indicate the hinge between domains. HSV-1 gB is colored in green, HCMV gB is colored in orange, and EBV gB is colored in cyan. (D) Single protomers (chain B in HSV-1 and HCMV) are aligned on DII, demonstrating the species-specific twist of each, using the same color scheme as in C.

Fig 4. Alignment of gB homologs.

(A) Individually aligned domains I of HSV-1 (2GUM) [16] in green, HCMV (5CXF) in orange, and EBV gB (3FVC) [17] in cyan are shown side by side and as an overlay. Residues used in alignments and RMSDs are listed in S1 Table. Orientation of the β hairpin β7-β8 (HCMV gB residues 188–201) is marked with an arrow. (B) Aligned trimers of domain V of each homolog are shown side by side and as an overlay. Colors of (B) are the same as in (A).

Fig 5. Observed glycosylation and fully-glycosylated model of HCMV gB.

(A) HCMV gB is shown in surface representation in wheat. Glycans observed in the structure are shown in space-filled representation in teal. Residues shown to be important for antibody binding are colored as in Fig 6. (B) To obtain a glycosylation model representative of mammalian glycosylation, high-mannose type glycans (Man8-GlcNAc2-N-Asn) were modeled onto the structure of gB. Paucimannose and high-mannose are shown schematically.

Fig 6. Neutralizing antibody epitopes.

(A) The overall view of the epitopes on the surface of gB. (B) Side and top view of domain IV (AD-1). Residues shown to decrease binding of anti-AD-1 Abs are shown in orange. Those specific to neutralizing AD-1 Abs (G595 and F632) are colored forest green. (C) Domain II (AD-4) in complex with a human neutralizing Ab, SM5-1 (4OSN and 4OT1) (left) and domain II of gB, chain B, modeled with Fab SM5-1 (right) to illustrate how glycosylation would affect antibody binding. The heavy chain of SM5-1 is colored in magenta, the light chain is in slate, and AD-4 epitope is in lime. (D) A close-up view of domain I (AD-5). Residue involved in binding of 1G2 (N284) is colored in magenta, residues involved in binding of 2C2 (N293/D295) are in blue, and residue important for binding of both (Y280) is in purple. Residues shown not to affect binding of 1G2 or 2C2 are colored in pale blue.