Human cytomegalovirus gH/gL/gO binding to PDGFRα provides a regulatory signal activating the fusion protein gB that can be blocked by neutralizing antibodies

Abstract

Herpesviruses require membrane fusion for entry and spread, a process facilitated by the fusion glycoprotein B (gB) and the regulatory factor gH/gL. The human cytomegalovirus (HCMV) gH/gL can be modified by the accessory protein gO, or the set of proteins UL128, UL130, and UL131. While the binding of the gH/gL/gO and gH/gL/UL128-131 complexes to cellular receptors, including PDGFRα and NRP2, has been well-characterized structurally, the specific role of receptor engagements by the gH/gL/gO and gH/gL/UL128-131 in regulation of fusion has remained unclear. We describe a cell-cell fusion assay that can quantitatively measure fusion on a timescale of minutes and demonstrate that binding of gH/gL/gO to PDGFRα dramatically enhances gB-mediated cell-cell fusion. In contrast, gH/gL/pUL128-131-regulated fusion is significantly slower, and gH/gL alone cannot promote gB fusion activity within this timescale. The genetic diversity of gO influenced the observed cell-cell fusion rates, correlating with previously reported effects on HCMV infectivity. Mutations in gL that had no effect on the formation of gH/gL/gO or binding to PDGFRa dramatically reduced the cell-cell fusion rate, suggesting that gL plays a critical role in linking the gH/gL/gO-PDGFRa receptor binding to activation of gB. Several neutralizing human monoclonal antibodies were found to potently block gH/gL/gO-PDGFRa-regulated cell-cell fusion, suggesting this mechanism as a therapeutic target.

Importance: Development of vaccines and therapeutics targeting the fusion apparatus of human cytomegalovirus (HCMV) has been limited by the lack of an in vitro cell-cell fusion assay that faithfully models the receptor-dependent fusion characteristic of HCMV entry. The cell-cell fusion assay described here demonstrated that the binding of gH/gL/gO to its receptor, PDGFRα, serves to regulate the activity of the fusion protein gB, and this is specifically vulnerable to inhibition by neutralizing antibodies. Moreover, the measurement of fusion kinetics allows for mutational studies of the fusion mechanism, assessing the influence of genetic diversity among the viral glycoproteins and studying the mechanism of neutralizing antibodies.

Keywords: glycoproteins; human cytomegalovirus; membrane fusion; virus entry.

Fig 1

Syncytium formation resulting from co-expression of HCMV glycoproteins. ARPE19 cells stably expressing plasma membrane-anchored GFP (lck-GFP) were infected with Ad vectors encoding HCMV glycoproteins gB and gH/gL, gH/gL/gO, or gH/gL/pUL128-131 with or without PDGFRα. Nuclei were stained with DAPI, and syncytium formation was monitored 48–72 hours post infection (h.p.i.) by immunofluorescence. White arrows indicate representative syncytia for each condition.

Fig 2

Quantitative assessment of HCMV fusion glycoproteins in real time. A live-cell, biomolecular complementation assay (adapted from Atanasiu et al. [55]) for which target cells were added to effector cells expressing gB and gH/gL (HL), gH/gL/gO (HLO), or gH/gL/pUL128-131 (HLP) and luminescence was measured every 10 min for 20 hours. (A) Fusion traces for HCMV glycoprotein complexes with ARPE19 cells used as targets. (B) ARPE19 cells were infected with Ad encoding PDGFRα (V242K [66]) for 24 hours and then used as targets. (C) Fusion rates for all conditions were determined by linear regression over the linear phase of each luciferase activity trace. Error bars reflect the standard deviation of three experiments, and P-values reflect two-way analysis of variance and Tukey’s multiple comparisons test between PDGFRα+/−target cells (ns > 0.05, * >0.01, ** >0.001, *** >0.0001, **** <0.0001).

Fig 3

Fusion sensitivity to gH/gL/gO surface levels. The expression of gH/gL/gO was titrated by adjusting the Ad input over an eightfold range, with 1× being the conditions used in Fig. 2. (A) Surface expression was determined by CELISA using an antibody specific to gH (mAb 14–4b). (B) Fusion traces corresponding to the titrated gH/gL/gO levels. (C) Fusion rates corresponding to the titrated gH/gL/gO levels. Error bars reflect the standard deviation of three experiments, and P-values were generated using one-way analysis of variance and Tukey’s multiple comparisons test (A) or Welch and Brown–Forsythe analysis of variance with Dunnett’s T3 comparisons (C) (ns >0.05, * >0.01, ** >0.001, *** >0.0001, **** <0.0001).

Fig 4

Regulation of gH/gL/gO-dependent fusion by endogenously expressed PDGFRα. (A) Fusion traces for HCMV glycoprotein complexes with MRC-5 cells used as targets. Fusion regulation by gH/gL/gO was tested with endogenous levels (pink) and overexpressed (brown) PDGFRα. (B) Fusion rates for all HCMV glycoprotein complexes with endogenous (–) or overexpressed (+) of PDGFRα. Error bars reflect the standard deviation of three experiments, and P-values were generated using two-way analysis of variance and Tukey’s multiple comparisons test (ns >0.05, * >0.01, ** >0.001, *** >0.0001, **** <0.0001).

Fig 5

Effect of gO allele on gH/gL/gO-PDGFRα fusion regulation. (A) Real-time fusion traces for gH/gL/gO-dependent cell–cell fusion using four different gO alleles: 1b, 1c, 2a, 1a. (B) Fusion rates for different gO alleles. (C) Surface expression of gH/gL/gO with different gO alleles determined by CELISA. Error bars reflect the standard deviation of three experiments, and P-values were generated using Welch and Brown–Forsythe analysis of variance with Dunnett’s T3 comparisons (B) or one-way analysis of variance and Dunnett’s multiple comparisons test (C) (ns >0.05, * >0.01, ** >0.001, *** >0.0001).

Fig 6

Effect of gL mutagenesis on gH/gL/gO-PDGFRα regulation of fusion. Real-time cell–cell fusion was performed using gL scanning alanine mutants (previously described in Schultz et al. [60, 71]). Fusion traces (left) and rates (right) are presented. Error bars reflect the standard deviation of three experiments, and P-values were generated using Welch and Brown–Forsythe analysis of variance with Dunnett’s T3 comparisons (ns >0.05, * >0.01, ** >0.001).

Fig 7

Assessing the ability for HCMV-nAb to block gH/gL/gO-PDGFRα regulation of fusion. The susceptibility of gH/gL/gO-dependent cell–cell fusion was assessed by preincubating effector cells with 50 µg/mL of nAb for 1 hour prior to addition of target cells. Ab treatment was maintained, and fusion was monitored over 20 hours. Several anti-gH/gL nAbs were tested including MSL-109 (76), 14–4b (77), and eight novel Abs isolated from the B-cell compartment of HCMV-infected donors (75). Specific antigenic domains (75) are designated in parentheses. Fusion traces (left) and rates (right) are presented. Error bars reflect the standard deviation of three experiments, and P-values were generated using one-way analysis of variance and Dunnett’s multiple comparisons test (ns >0.05, * >0.01, ** >0.001, *** >0.0001, **** <0.0001).

Fig 8

Comparison of the apo and PDGFRα-bound gH/gL/gO structure. Cryo-electron microscopy coordinates for the apo (pdb 7LBE) and PDGFRα-bound (7LBF) structures (72) of gH/gL/gO were aligned in Chimera (UCSF [84]) and represented as ribbon structures. Individual subunits gH (blue), gL (grey), and gO (orange) are shaded to designate the apo (dark shade) and PDGFRα-bound (light shade). Residues of gL within the L139 (top right) and L201 (bottom right) regions that were mutated to alanine are displayed as sticks, with relevant interactions denoted with bond distances.