Disulfide bond configuration of human cytomegalovirus glycoprotein B

Abstract

Glycoprotein B (gB) is the most highly conserved of the envelope glycoproteins of human herpesviruses. The gB protein of human cytomegalovirus (CMV) serves multiple roles in the life cycle of the virus. To investigate structural properties of gB that give rise to its function, we sought to determine the disulfide bond arrangement of gB. To this end, a recombinant form of gB (gB-S) comprising the entire ectodomain of the glycoprotein (amino acids 1 to 750) was constructed and expressed in insect cells. Proteolytic fragmentation and mass spectrometry were performed using purified gB-S, and the five disulfide bonds that link 10 of the 11 highly conserved cysteine residues of gB were mapped. These bonds are C94-C550, C111-C506, C246-C250, C344-C391, and C573-C610. This configuration closely parallels the disulfide bond configuration of herpes simplex type 2 (HSV-2) gB (N. Norais, D. Tang, S. Kaur, S. H. Chamberlain, F. R. Masiarz, R. L. Burke, and F. Markus, J. Virol. 70:7379-7387, 1996). However, despite the high degree of conservation of cysteine residues between CMV gB and HSV-2 gB, the disulfide bond arrangements of the two homologs are not identical. We detected a disulfide bond between the conserved cysteine residue 246 and the nonconserved cysteine residue 250 of CMV gB. We hypothesize that this disulfide bond stabilizes a tight loop in the amino-terminal fragment of CMV gB that does not exist in HSV-2 gB. We predicted that the cysteine residue not found in a disulfide bond of CMV gB, cysteine residue 185, would play a role in dimerization, but a cysteine substitution mutant in cysteine residue 185 showed no apparent defect in the ability to form dimers. These results indicate that gB oligomerization involves additional interactions other than a single disulfide bond. This work represents the second reported disulfide bond structure for a herpesvirus gB homolog, and the discovery that the two structures are not identical underscores the importance of empirically determining structures even for highly conserved proteins.

FIG. 1.

gB-S peptides expected from CNBr digestion and initial model of disulfide bonds of CMV gB. Shown is a sequence alignment of the ectodomains of gB from CMV and HSV-2. The amino terminus of each glycoprotein is oriented at left and cysteine residues are numbered sequentially. The asterisk denotes the additional cysteine residue of CMV gB that is predicted to be unpaired. The disulfide bond configuration of HSV-2 gB is indicated by solid lines connecting appropriate cysteine pairs. Due to the high conservation of cysteine residues between the two gB homologs, our initial model for the disulfide bond configuration of CMV gB paralleled that of HSV-2 gB. Shown above the CMV gB sequence are the positions and sizes of cysteine-containing fragments of CMV gB expected from cyanogen bromide digestion. Peptide fragments that do not contain cysteine residues were omitted for clarity.

FIG. 2.

Purification of gB-S and digestion with CNBr. Fractions collected during the purification of gB-S were analyzed by SDS-PAGE and Coomassie blue staining. (A) Lane 1, predialysis supernatant; lane 2, postdialysis supernatant; lane 3, column flowthrough; lane 4, elution. (B) Digestion of gB-S with cyanogen bromide (lane 1) resulted in 16 fragments. Shown in this Coomassie blue-stained gel of CNBr-digested gB-S are numerous gB-S fragments (arrows) as well as PNGase F used to deglycosylate the peptides (*).

FIG. 3.

HPLC separation of CNBr-generated gB-S peptides. All of the peaks visible in this trace were analyzed by MALDI-TOF (MS) as indicated in Materials and Methods, and fractions labeled A through E were assigned to gB-S peptides that contain cysteine residues. All gB-S peptides that did not contain cysteine residues were identified by MALDI-TOF (MS) but are not labeled in this HPLC trace.

FIG. 4.

Disulfide bond C94-C550. Spectra of fraction A before (left) and after (right) treatment with DTT and iodoacetamide. The fragment corresponding to CNBr-C94 was not identified in the spectrum following reduction and carboxyamidation. The m/z ratio and charge state of each peak are indicated.

FIG. 5.

Disulfide bond C111-C506. Spectra of fraction B before (left) and after (right) treatment with DTT and iodoacetamide. The m/z ratio and charge state of each peak are indicated.

FIG. 6.

Disulfide bond C246-C250. Spectra of fraction C before (left) and after (right) treatment with DTT and iodoacetamide. The m/z ratio and charge state of each peak are indicated.

FIG. 7.

Disulfide bond C344-C391. Spectra of fraction D before (left) and after (right) treatment with DTT and iodoacetamide. The m/z ratio and charge state of each peak are indicated.

FIG. 8.

Disulfide bond C573-C610. Spectra of fraction E before (left) and after (right) treatment with DTT and iodoacetamide. The m/z ratio and charge state of each peak are indicated.

FIG. 9.

Cysteine substitution mutant C185A. The additional cysteine residue of CMV gB, C185, was predicted to be involved in dimerization between two gB monomers. Shown is an immunoblot of 1% CHAPS-soluble cell lysates from wild-type gB or C185A mutant gB expressing 293T cells. In contrast to predictions, the C185A mutant does not show a significant defect in dimerization.

FIG. 10.

Model of disulfide bonds of CMV gB. (A) Shown is an alignment of the ectodomains of gB from CMV (top) and HSV-2 (bottom). The amino terminus is oriented at left and cysteine residues are numbered sequentially. The disulfide bond configurations of both HSV-2 gB and CMV gB are indicated by solid lines connecting appropriate cysteine pairs. The similarity of disulfide bonding reflects the high conservation of cysteine residues between these two glycoproteins although significant differences exist in the region from C185 to C250 of CMV gB. The asterisk denotes the additional cysteine residue of CMV gB that was predicted to be unpaired but was found in a disulfide bond with C246. (B) The disulfide bond between CMV gB C246 and C250 likely maintains these residues in a tight loop that may not exist in the HSV-2 gB homolog. This model shows a wireframe view of the alpha carbon atoms of the putative C246-C250 loop. The structure was modeled using the 3D-PSSM protein fold recognition (threading) server (http://www.sbg.bio.ic.ac.uk/∼3dpssm/).