The N Terminus of Human Cytomegalovirus Glycoprotein O Is Important for Binding to the Cellular Receptor PDGFRα

Abstract

The human cytomegalovirus (HCMV) glycoprotein complex gH/gL/gO is required for the infection of cells by cell-free virions. It was recently shown that entry into fibroblasts depends on the interaction of gO with the platelet-derived growth factor receptor alpha (PDGFRα). This interaction can be blocked with soluble PDGFRα-Fc, which binds to HCMV virions and inhibits entry. The aim of this study was to identify parts of gO that contribute to PDGFRα binding. In a systematic mutational approach, we targeted potential interaction sites by exchanging conserved clusters of charged amino acids of gO with alanines. To screen for impaired interaction with PDGFRα, virus mutants were tested for sensitivity to inhibition by soluble PDGFRα-Fc. Two mutants with mutations within the N terminus of gO (amino acids 56 to 61 and 117 to 121) were partially resistant to neutralization. To validate whether these mutations impair interaction with PDGFRα-Fc, we compared binding of PDGFRα-Fc to mutant and wild-type virions via quantitative immunofluorescence analysis. PDGFRα-Fc staining intensities were reduced by 30% to 60% with mutant virus particles compared to wild-type particles. In concordance with the reduced binding to the soluble receptor, virus penetration into fibroblasts, which relies on binding to the cellular PDGFRα, was also reduced. In contrast, PDGFRα-independent penetration into endothelial cells was unaltered, demonstrating that the phenotypes of the gO mutant viruses were specific for the interaction with PDGFRα. In conclusion, the mutational screening of gO revealed that the N terminus of gO contributes to efficient spread in fibroblasts by promoting the interaction of virions with its cellular receptor.IMPORTANCE The human cytomegalovirus is a highly prevalent pathogen that can cause severe disease in immunocompromised hosts. Currently used drugs successfully target the viral replication within the host cell, but their use is restricted due to side effects and the development of resistance. An alternative approach is the inhibition of virus entry, for which understanding the details of the initial virus-cell interaction is desirable. As binding of the viral gH/gL/gO complex to the cellular PDGFRα drives infection of fibroblasts, this is a potential target for inhibition of infection. Our mutational mapping approach suggests the N terminus as the receptor binding portion of the protein. The respective mutants were partially resistant to inhibition by PDGFRα-Fc but also attenuated for infection of fibroblasts, indicating that such mutations have little if any benefit for the virus. These findings highlight the potential of targeting the interaction of gH/gL/gO with PDGFRα for therapeutic inhibition of HCMV.

Keywords: cytomegalovirus; glycoproteins; receptors; virus entry.

FIG 1

Charge cluster-to-alanine scanning of pUL74. A charge cluster was defined as two or more charged amino acids within a group of five amino acids. A gray background indicates conservation among various HCMV strains. Purple amino acids are negatively charged, and blue amino acids are positively charged. The boxes indicate peptide sites at which charged amino acids were exchanged with alanines.

FIG 2

Spread of charge cluster-to-alanine mutants in fibroblasts. Mutated viral genomes based on TB40-BAC4-IE-GLuc were transfected into fibroblasts, and samples of the GLuc-containing supernatants were drawn every 3 days. GLuc activity was measured after the last sample was taken. Linear regression of the increase of GLuc activity over time allowed determination of the growth rate of the respective virus. (A) One example of the calculation by linear regression; (B) mean growth rates of all mutant viruses in three experiments relative to the growth rate of wild-type virus in the respective experiment. Error bars indicate the standard error of the mean. The phenotypes of the mutants were defined according to their growth rates in comparison with that of the wild type (far left) as follows: unaltered (black; difference of <2 SEM), moderately impaired (blue; reduced by 2 to 10 SEM), severely impaired (red; reduced by >10 SEM), or enhanced (purple; increased by >2 SEM).

FIG 3

Mutations of aa 56 to 61 and 117 to 121 confer resistance to inhibition by PDGFRα-Fc. Virus stocks were preincubated with various concentrations (3 to 200 ng/ml) of PDGFRα-Fc for 2 h at 37°C before infection of fibroblasts. The inhibitory effect of PDGFRα-Fc on the different virus mutants was assessed by the visualization of infected cells via indirect immunofluorescence against the viral IE antigens at 1 day postinfection. (A) Dose-response curves normalized to untreated control. Each curve represents mean values from 3 to 4 independent experiments. (B) In a total of 4 experiments, the EC₅₀ of PDGFRα for inhibition of mutants was compared to that for wild-type virus. Error bars indicate the standard error of the mean. The wild type is shown in black, ccta56–61 in turquois, and ccta117–121 in blue. *, P < 0.05.

FIG 4

PDGFRα-Fc binds ccta56–61 and ccta117–121 virus particles less efficiently. Freshly produced wild-type and mutant virus particles were preincubated with 20, 100, or 500 ng/ml PDGFRα-Fc for 2 h. To allow for easy visualization, the virus particles were attached to fibroblasts on ice for 90 min before fixation with 80% acetone. Virus particles were indirectly stained with a mouse anti-pp150 antibody and rabbit anti-mouse Cy3. Bound PDGFRα-Fc was detected using Alexa488-labeled goat anti-human-Ig antibodies. (A) Representative examples of virus particles treated with 500 ng/ml of PDGFRα-Fc. (B) Mean particle intensities of 100 virus particles per condition for the different viruses. Particles were chosen for quantification according to their pp150 signal and low background in the surrounding area. (C) To control for specificity of the observed reduction in PDGFRα-Fc binding, signal intensities of the two N-terminal ccta mutants, a mutant without a phenotype, and UL74stop were compared to those of the wild-type virus after treatment with 500 ng/ml of PDGFR. One hundred particles per condition were quantified. The signal intensities of the corresponding PDGFRα-Fc signals were measured using AxioVision software. Error bars indicate the standard error of the median. *, P < 0.05 for comparison with the wild type.

FIG 5

UL74ccta56–61 and UL74ccta117–121 virions contain normal amounts of the gH/gL/gO trimer. Gradient-purified virions were subjected to SDS-PAGE under nonreducing conditions. The gH/gL/gO complex was detected using rabbit anti-gO polyclonal antibodies. As a reference, the lower parts of the membranes were stained with a murine monoclonal antibody recognizing the viral major capsid protein (MCP). For comparison, virus mutants with severely impaired spreading capacities were included. (A) Representative example of an immunoblot as used for quantification of signals. (B) Quantification of six experiments, as described in panel A, using various volumes of two biologically independent virion preparations of wild-type, ccta56–61, and ccta117–121 virions. The ratios of gH/gL/gO signals and MCP signals were normalized to the wild type in each experiment. Error bars indicate the standard error of the mean. (C) Virion preparations were titrated on fibroblasts to determine the number of infectious units per milliliter. The number of HCMV genomes in an aliquot of the virion preparation was determined. Shown are the mean infectivity/particle ratios determined for at least three independent virion preparations. *, P < 0.05 for comparison with the wild type.

FIG 6

The mutations ccta56–61 and ccta117–121 impair PDGFRα-mediated penetration into fibroblasts. (A) Representative images of the nuclear translocation assay (as described for panel B) in fibroblasts. (B) To determine the number of particles that successfully penetrated into the cells, gradient-purified virions of the wild type, ccta56–61 (turquoise), and ccta117–121 (blue) were incubated with fibroblasts or endothelial cells for 6 h. Fixed cultures were than stained for pp150 to visualize the particle distribution. The percentage of particles that have successfully translocated to the nucleus in a total of 80 cells per condition in 2 independent experiments is shown. (C) Gradient-purified wild-type virions were pretreated with 500 ng/ml PDGFRα-Fc for 2 h before incubation with fibroblasts. After 6 h, the efficiency of nuclear translocation was determined as for panel B. About 60 cells per condition were analyzed. (D) Fibroblasts were preincubated with PDGF-AA for 1 h on ice before infection. Infection with wild-type virus (black), ccta56–61 (turquoise), and ccta117–121 (blue) was performed for 2 h at 37°C in the presence of PDGF-AA. At 1day postinfection, the cells were fixed and stained for the viral immediate early antigens. Shown is the percentage of infected cells normalized to the untreated control (no PDGF-AA). All error bars indicate standard error of the mean determined from 3 experiments. *, P < 0.05.

FIG 7

Mutation of N-terminal charge clusters also confers partial resistance in other HCMV strains. (A) Sequence alignment of three different HCMV strains, TB40, AD169, and VHL/E, shows that the putative receptor binding sites (blue boxes) are conserved. The charged amino acids that were exchanged with alanines are shown in blue (positively charged) and purple (negatively charged). Numbering is based on the TB40 sequence. (B) Stocks of TB40-BAC4, AD169-pHB5, and VHL/E-BAC19 and the respective ccta mutants were diluted to an MOI of 1 and preincubated with 3 to 200 ng/ml of PDGFRα-Fc for 2 h at 37°C before infection of fibroblasts. The inhibitory effect of PDGFRα-Fc on the viruses was assessed by the visualization of infected cells via indirect immunofluorescence against the viral IE antigens at 1 day postinfection. The infection rates were normalized to the respective untreated control. The gray lines indicate the EC₅₀. Each curve represents mean values from 3 biologically independent experiments.