Mutational Mapping of UL130 of Human Cytomegalovirus Defines Peptide Motifs within the C-Terminal Third as Essential for Endothelial Cell Infection

Abstract

The UL130 gene is one of the major determinants of endothelial cell (EC) tropism of human cytomegalovirus (HCMV). In order to define functionally important peptides within this protein, we have performed a charge-cluster-to-alanine (CCTA) mutational scanning of UL130 in the genetic background of a bacterial artificial chromosome-cloned endotheliotropic HCMV strain. A total of 10 charge clusters were defined, and in each of them two or three charged amino acids were replaced with alanines. While the six N-terminal clusters were phenotypically irrelevant, mutation of the four C-terminal clusters each caused a reduction of EC tropism. The importance of this protein domain was further emphasized by the fact that the C-terminal pentapeptide PNLIV was essential for infection of ECs, and the cell tropism could not be rescued by a scrambled version of this sequence. We conclude that the C terminus of the UL130 protein serves an important function for infection of ECs by HCMV. This makes UL130 a promising molecular target for antiviral strategies, e.g., the development of antiviral peptides.

Human cytomegalovirus (HCMV) is a widespread betaherpesvirus that causes lifelong persistent infections with occasional reactivations. While HCMV infection is usually clinically unapparent in the immunocompetent host, it can cause severe disseminated infections under conditions of immunosuppression, with manifestations in the lung, retina, and gastrointestinal tract, among others (12). Various cell types support viral replication, including epithelial cells and endothelial cells (ECs), smooth muscle cells, fibroblasts, and cells of hematopoietic origin (13, 14, 18, 19, 25, 26, 37). Among these target cells, endothelial cells are assumed to contribute particularly to hematogenous dissemination of HCMV (24).

While recent clinical HCMV isolates are characterized by this broad cell tropism, the target cell range becomes restricted during long-term propagation on fibroblasts (28, 33). The underlying mechanism for this cell culture adaptation is a modulation within the viral genes UL128, UL130, and UL131A (8, 11). These three genes have been shown to be essential for infection of granulocytes, dendritic cells, epithelial cells, and endothelial cells but are dispensable for infection of fibroblasts (1, 9, 11, 34, 35). The encoded proteins pUL128, pUL130, and pUL131A were reported to form a complex with the viral glycoproteins gH and gL that is distinct from the glycoprotein complex gCIII (gH/gL/gO) (35). Whereas poorly endotheliotropic HCMV strains bear just the gH/gL/gO complex in their envelopes, highly endotheliotropic strains bear both gCIII variants: gH/gL/gO and gH/gL/pUL128-131A. Deletion of any of the three genes UL128-131A results in loss of EC tropism (11), most likely because only a complete complex of gH/gL and pUL128, pUL130, and pUL131A can efficiently function in endocytic entry in ECs (21). However, functional sites within the proteins (e.g., mediating binding to the viral complex partners or interaction with a putative cellular receptor) have not yet been identified. One approach to search for candidate protein-protein interaction sites is charge-cluster-to-alanine (CCTA) mutagenesis. This method is based on the assumption that clusters of charged amino acids tend to be exposed in the tertiary structure of a protein and are thus likely to be sites of interaction with other proteins. Replacement of these charged amino acids by uncharged alanines should then target protein-protein interaction sites without destroying the protein backbone (5, 7). Applying this method to HCMV pUL128, we were able to identify a central core region within pUL128 essential for EC infection as well as contributing sites in the N-terminal half and the C terminus of the protein (22). We now aimed to extend the study to the scanning of UL130 by markerless mutagenesis in the context of a highly endotheliotropic HCMV BACmid, TB40-BAC4. The resulting mutant viruses were then characterized with regard to their ability to infect ECs to identify the relevant parts of the protein.

With regard to the role of UL130 in EC infection by endocytosis, the C-terminal part of pUL130 was of special interest. A frameshift mutation that changes the last 11 amino acids (aa) of pUL130 is the most prominent difference between the poorly endotheliotropic HCMV strain Towne and the highly endotheliotropic strain HCMV-TB40-BAC4 in this region (8, 11, 27). Rhee and Davis have described a cell-penetrating pentapeptide (CPP) motif (PFVYLI) mediating internalization by endocytosis, which is clathrin and caveolin independent but may involve lipid rafts (17). Not only do the last five amino acids of pUL130 (PNLIV) bear a striking similarity to this motif, but also the entry of HCMV into ECs has been reported to occur by an endocytic pathway (20, 23). Thus, we hypothesized that the pentapeptide motif PNLIV in pUL130 might be involved in mediating endocytic uptake of HCMV in ECs, and if so, deletion of this motif should result in a nonendotheliotropic virus. A number of CPPs that are thought to be taken up by endocytosis have now been described, including VPMLK, PMLKE, VPTLK, KLPVM, and others (32). These CPPs all bear some similarity, but the exact amino acid sequence seems to be irrelevant. We thus hypothesized for UL130 that a scrambled mutant (PNLIV changed to PINVL) should still be able to mediate endocytosis of HCMV in ECs. To test these assumptions we generated a series of mutant viruses where the PNLIV motif was either deleted, scrambled (PNLIV changed to PINVL), or exchanged against a known CPP (PFVYLI [17]) and characterized them with regard to EC infectivity.

MATERIALS AND METHODS

Cells and viruses.

Primary human foreskin fibroblasts (HFFs) were cultured in minimal essential medium (MEM) (GIBCO/Invitrogen) containing 5% fetal calf serum (FCS), 2 mmol/liter l-alanyl-l-glutamine, 100 μg/ml gentamicin, and 0.5 ng/ml basic fibroblast growth factor (bFGF) (GIBCO/Invitrogen). Primary human umbilical vein endothelial cells (HUVECs) were cultured in RPMI 1640 (GIBCO/Invitrogen) with 50 μg/ml endothelial cell growth supplement (ECGS) (Becton Dickinson), 10% human serum (HCMV seronegative), 2 mmol/liter l-alanyl-l-glutamine, 5 IU/ml heparin, and 100 μg/ml gentamicin. HCMV strain TB40-BAC4 (27) and derived mutants were propagated in HFFs. An HCMV-TB40-BAC4 mutant carrying a deletion of the complete UL128-132 region was generated as previously described by Hahn et al. (11). For preparation of virus stocks, infectious supernatants from HFF cultures were harvested at days 5 to 6 postinfection. Cellular debris was removed by centrifugation at 3,220 × g for 10 min, and the supernatants were stored at −80°C until phenotypic testing on ECs.

Markerless mutagenesis of HCMV genomes.

The highly endotheliotropic HCMV-TB40/E-derived bacterial artificial chromosome (BAC) TB40-BAC4 was used for the generation of mutant strains (27). The mutant BACs were generated with the markerless mutagenesis protocol as described by Tischer et al. (31). For charge-cluster-to-alanine (CCTA) mutations, codons for charged amino acids in the sequence of interest were exchanged against codons for alanine (Table 1). Recombination fragments were generated by PCR from plasmid pEP-Kan-S (primer information is given in Table 2). The resulting recombination fragments consisted of the 18-bp I-SceI restriction site and a kanamycin resistance cassette flanked on both sides by overlapping HCMV homologies containing the sequence of interest. The respective recombination fragments for the different mutants were inserted into TB40-BAC4 by homologous recombination in Escherichia coli strain GS1783 (kindly provided by Gregory A. Smith, Northwestern University, Chicago, IL). After successful kanamycin selection, all non-HCMV sequences were removed by an intrabacterial I-SceI digest and a subsequent Red recombination. Correct mutagenesis was confirmed by sequencing. For virus reconstitution, BAC DNA was isolated using the NucleoBond Xtra Midi kit (Macherey-Nagel) and transfected into HFFs using the MBS transfection kit (Stratagene). Cells were propagated until viral plaques appeared and virus could be harvested.

TABLE 1.

Overview of TB40-BAC4-UL130ccta mutants and their respective amino acid exchanges

Amino acidsa	CCTA mutation	Mutant TB40-BAC4
KPHD	APAA	UL130ccta48-51
ECRNE	ACANA	UL130ccta82-86
RE	AA	UL130ccta94-95
ER	AA	UL130ccta101-102
KK	AA	UL130ccta108-109
RTASK	ATASA	UL130ccta127-131
EDAK	AAAA	UL130ccta142-145
KQTK	AQTA	UL130ccta154-157
DGTR	AGTA	UL130ccta165-168
HVFRD	AVFAA	UL130ccta181-185

^aAmino acids changed in the CCTA mutants are in bold.

TABLE 2.

Primers used for UL130ccta mutagenesis

Mutant BAC	Primer direction	Primer sequencea
TB40-BAC4-UL130ccta48-51	Forward	AGAATCCGTCCCCGCTATGGTCTAAACTGACGTATTCCGCACCGGCAGCAGCGGCGACGTTTTACTGTaggatgacgacgataagt
	Reverse	GGCGAGGGATAGATAAAAGGACAGTAAAACGTCGCCGCTGCTGCCGGTGCGGAATACGTCAGTTTAGAcaaccaattaaccaattctga
TB40-BAC4-UL130ccta82-86	Forward	CAATTCTCGGGGTTCCAGCGGGTATTAACGGGTCCCGCATGTGCAAACGCAACCCTGTATCTGCTGTaggatgacgacgataagt
	Reverse	GGTCTGGCCTTCCCGGTTGTACAGCAGATACAGGGTTGCGTTTGCACATGCGGGACCCGTTAATACCcaaccaattaaccaattctga
TB40-BAC4-UL130ccta94-95	Forward	TCCCGAGTGTCGCAACGAGACCCTGTATCTGCTGTACAACGCAGCAGGCCAGACCTTGGTGGAGAGaggatgacgacgataagt
	Reverse	TTTTCACCCATGTAGAGCTTCTCTCCACCAAGGTCTGGCCTGCTGCGTTGTACAGCAGATACAGGGcaaccaattaaccaattctga
TB40-BAC4-UL130ccta101-102	Forward	CCTGTATCTGCTGTACAACCGGGAAGGCCAGACCTTGGTGGCAGCAAGCTCCACCTGGGTGAAAAAaggatgacgacgataagt
	Reverse	CGCTCAGGTACCAGATCACCTTTTTCACCCAGGTGGAGCTTGCTGCCACCAAGGTCTGGCCTTCCCcaaccaattaaccaattctga
TB40-BAC4-UL130ccta108-109	Forward	GGAAGGCCAGACCTTGGTAGAGAGAAGCTCCACCTGGGTGGCAGCAGTGATCTGGTACCTGAGCGGaggatgacgacgataagt
	Reverse	GGAGGATGGTCTGGTTGCGACCGCTCAGGTACCAGATCACTGCTGCCACCCAGGTGGAGCTTCTCTcaaccaattaaccaattctga
TB40-BAC4-UL130ccta127-131	Forward	AGCGGTCGCAACCAGACCATCCTCCAACGGATGCCCGCAACAGCTTCAGCACCGAGCGACGGAAACGaggatgacgacgataagt
	Reverse	GTCTTCCACGCTGATCTGCACGTTTCCGTCGCTCGGTGCTGAAGCTGTTGCGGGCATCCGTTGGAGGcaaccaattaaccaattctga
TB40-BAC4-UL130ccta142-145	Forward	CTTCAAAACCGAGCGACGGAAACGTGCAGATCAGCGTGGCAGCAGCCGCAATTTTTGGAGCGCACATGaggatgacgacgataagt
	Reverse	AGCTTGGTCTGCTTGGGCACCATGTGCGCTCCAAAAATTGCGGCTGCTGCCACGCTGATCTGCACGTTcaaccaattaaccaattctga
TB40-BAC4-UL130ccta154-157	Forward	TGGAAGACGCCAAGATTTTTGGAGCGCACATGGTGCCCGCACAGACCGCACTGCTACGCTTTGTAGTCaggatgacgacgataagt
	Reverse	TGATAACGTGTGCCATCGTTGACTACAAAGCGTAGCAGTGCGGTCTGTGCGGGCACCATGTGCGCTCCcaaccaattaaccaattctga
TB40-BAC4-UL130ccta165-168	Forward	TGCCCAAACAGACCAAGCTGCTACGCTTCGTCGTCAACGCAGGCACAGCATATCAGATGTGTGTGATGaggatgacgacgataagt
	Reverse	TGAGCCCAGCTCTCCAGCTTCATCACACACATCTGATATGCTGTGCCTGCGTTGACGACGAAGCGTAGcaaccaattaaccaattctga
TB40-BAC4-UL130ccta181-185	Forward	TATCAGATGTGTGTGATGAAGCTGGAGAGCTGGGCTGCAGTCTTCGCAGCATACAGCGTGTCTTTTCaggatgacgacgataagt
	Reverse	GGTGAACGTCAATCGCACCTGAAAAGACACGCTGTATGCTGCGAAGACTGCAGCCCAGCTCTCCAGCcaaccaattaaccaattctga
TB40-BAC4-UL130ΔPNLIV	Forward	CACCGAGGCCAATAACCAGACTTACACCTTCTGCACACATTGAGCCCGTCGCACGAGCAGaggatgacgacgataagtaggg
	Reverse	CGCGCGGTTTTCAAAATTCCCTGCTCGTGCGACGGGCTCAATGTGTGCAGAAGGTGTAAGcaaccaattaaccaattctgattag
TB40-BAC4-UL130addPNLIV	Forward	CACCGAGGCCAATAACCAGACTTACACCTTCTGCACACATCCCAATCTCATCGTTTGAGCCCGTCGCACGAGCAGaggatgacgacgataagtaggg
	Reverse	CGCGCGGTTTTCAAAATTCCCTGCTCGTGCGACGGGCTCAAACGATGAGATTGGGATGTGTGCAGAAGGTGTAAGcaaccaattaaccaattctgat
TB40-BAC4-UL130scramPNLIV	Forward	CACCGAGGCCAATAACCAGACTTACACCTTCTGCACACATCCCATCAATGTTCTCTGAGCCCGTCGCACGAGCAGaggatgacgacgataagtaggg
	Reverse	CGCGCGGTTTTCAAAATTCCCTGCTCGTGCGACGGGCTCAGAGAACATTGATGGGATGTGTGCAGAAGGTGTAAGcaaccaattaaccaattctgat

^aUppercase, HCMV homology; lowercase, homology to template plasmid pEP-Kan-S; bold, sequence of interest.

Measurement of the endothelial cell tropism.

Infectious supernatant from fibroblast cultures was tested for cell-free infectivity in HUVECs and HFFs. A total of 2 × 10⁴ HFFs and HUVECs per well were seeded in 96-well plates coated with 0.1% gelatin 1 day prior to the experiment. The cells were preincubated for 30 min at 37°C with MEM and then infected with the respective virus suspension for 1 h. The virus suspension was replaced by fresh medium, and the cultures were incubated overnight at 37°C. Cells were then fixed with 80% acetone and stained for viral immediate-early antigen (pUL122/123) by sequential incubation with antibody E13 (Biosoft) and Cy3-conjugated goat anti-mouse IgG F(ab)₂ (Jackson ImmunoResearch). Nuclei were counterstained with DAPI (4′,6′-diamidino-2-phenylindole). The infection efficiencies in HFFs and HUVECs were quantified, and the relative EC tropisms of the mutant viruses in comparison to the wild-type HCMV-TB40-BAC4 were determined as the ratio of infection efficiency in HUVECs to infection efficiency in HFFs.

Purification of HCMV virions by glycerol-tartrate gradients.

Virion fractions were isolated from late-stage infected cell culture supernatants by use of glycerol-tartrate density gradients essentially as described previously (30). Briefly, 60 to 120 ml of infectious supernatant was centrifuged at 3,220 × g for 10 min to remove cellular debris. HCMV particles were then pelleted by ultracentrifugation at 80,000 × g for 70 min. The pellets were resuspended in 2 ml sodium phosphate buffer (8 mM NaH₂PO₄·1H₂O, 32 mM Na₂HPO₄·12H₂O, pH 7.4) and layered onto a preformed linear glycerol-tartrate gradient (15% sodium tartrate-30% glycerol to 35% sodium tartrate in phosphate buffer). The gradient was centrifuged at 80,000 × g for 45 min, resulting in separation of HCMV particles into noninfectious enveloped particles (NIEPs), virions, and dense bodies. The virion fraction was collected using a syringe and needle, resuspended in phosphate buffer, and recentrifuged at 80,000 × g for 70 min. The supernatant was discarded, and the virion pellets were stored at −80°C until used for Western blot analyses.

Immunoblot detection of proteins UL128, UL130, and UL131A.

Infected cells or virions were lysed with 0.5% NP-40 in phosphate-buffered saline (PBS) supplemented with a protease inhibitor cocktail (Roche) and standard Laemmli loading buffer. Lysates for detection of pUL128 were prepared with “total lysis” buffer (62.5 mM Tris, 2% [vol/vol] SDS, 10% [vol/vol] glycerol, 6 M urea, 0.01% [wt/vol] bromphenol blue, 0.01% [wt/vol] phenol red, and 5% [vol/vol] β-mercaptoethanol) supplemented with benzonase nuclease (Novagen). Proteins were separated on 10% or 15% acrylamide gels and transferred to polyvinylidene difluoride (PVDF) membranes. The UL128 protein was detected with monoclonal antibody 4B10 and pUL130 with monoclonal antibody 3C5 (kindly provided by Tom Shenk, Princeton University, Princeton, NJ) (35). The UL131A protein was detected with a rabbit polyclonal antiserum (kindly provided by Brent Ryckman and David Johnson, Oregon Health and Sciences University, Portland, OR) (21). Secondary detection was performed with horseradish peroxidase-conjugated goat anti-mouse Ig antibody (Dako or Santa Cruz) or horseradish peroxidase-conjugated goat anti-rabbit Ig antibody (Santa Cruz) and chemiluminescence (Super Signal; Pierce). Chemiluminescence signals were documented with Fluor-S Max (Bio-Rad). Quantity One software (Bio-Rad) was used for the densitometric evaluation of signals.

Statistical analysis.

The statistical significance of differences in cell tropism values of the various viruses was determined using two-tailed Mann-Whitney U test analyses. A P value of <0.001 was considered highly significant.

RESULTS

Identification of charge clusters in pUL130 and generation of mutants.

Charge clusters are likely to be exposed sites within the tertiary structure of a protein and may thus form interaction sites with other proteins. Replacement of these charged amino acids with uncharged alanines, rather than mere deletions, will probably destroy putative interaction sites on the protein surface while the overall structure of the protein is preserved. Charge-cluster-to-alanine (CCTA) scanning therefore provides a tool for the identification of specific protein interaction sites while avoiding unspecific effects due to destruction of the protein backbone. A charge cluster was defined as at least two charged amino acids within five successive amino acids (36). According to this definition, 11 charge clusters in TB40-BAC4 pUL130 could be identified, which were distributed over the complete protein (Fig. 1). From charge clusters 2 to 11, a sequence of up to five amino acids was chosen for CCTA mutagenesis (indicated by boxes in Fig. 1). The first charge cluster was omitted because amino acids 1 to 25 have been shown to be a signal peptide that is cleaved off (15). The peptide motifs chosen for CCTA scanning were designated according to their amino acid positions: KPHD_48-51, ECRNE_82-86, RE_94-95, ER_101-102, KK_108-109, RTASK_127-131, EDAK_142-145, KQTK_154-157, DGTR_165-168, and HVFRD_181-185. The generation of mutants was performed in E. coli by markerless replacement of the selected charged amino acids with alanines in BACs containing the genome of the highly endotheliotropic HCMV-TB40-BAC4. A complete list of the generated mutants and the respective amino acid exchanges as well as the primer sequences used for markerless mutagenesis are given in Tables 1 and 2. The mutated BAC genomes were transfected into primary human fibroblasts (HFFs) for reconstitution of virus. As UL130 is dispensable for replication of HCMV in fibroblasts, reconstitution of mutant viruses perfectly resembled reconstitution of wild-type virus HCMV-TB40-BAC4. The transfected HFFs showed viral plaques within 5 to 10 days after transfection, grew to 100% cytopathic effect (CPE) during further cell culture passages, and finally produced virus titers of >10⁶ infectious units per ml. The correctness of the recombinations was controlled in every mutant by sequencing of UL130. Stocks of each virus were then produced in HFFs and stored at −80°C for phenotypic testing in ECs.

FIG. 1.

Charge clusters in the amino acid sequence of pUL130 of TB40-BAC4. Charged amino acids are indicated by bold letters. Eleven charge clusters were identified in pUL130 and are highlighted in gray. Boxed amino acids represent the motifs chosen for CCTA mutagenesis (i.e., amino acids 48 to 51, 82 to 86, 94 and 95, 101 and 102, 108 and 109, 127 to 131, 142 to 145, 154 to 157, 165 to 168, and 181 to 185). Colors indicate the phenotypes of the CCTA mutants: green, no phenotype; yellow, intermediate phenotype; red, severe phenotype. The backslash (\) between amino acids 25 and 26 marks the end of the predicted signal peptide.

Endothelial cell tropism of HCMV-TB40-BAC4-UL130ccta mutants.

To evaluate the effects of the various mutations on replication of the respective viruses in ECs, cell-free virus preparations were tested on human umbilical vein endothelial cells (HUVECs) and HFFs in parallel. Fibroblast infection served as a control and as an indirect indicator of the virus titer, because the analyzed genes are dispensable for infection of fibroblasts. The infection rates in the two cell types were compared, thus revealing the “relative EC tropism” of each mutant.

CCTA mutation of any of the charge clusters in the N-terminal half of pUL130 (KPHD_48-51, ECRNE_82-86, RE_94-95, ER_101-102, KK_108-109, and RTASK_127-131) had no significant effect on EC tropism. The phenotypes of these mutants resembled that of the wild-type HCMV-TB40-BAC4. CCTA mutation of charge clusters in the C-terminal part of UL130, in contrast, strongly influenced the EC tropism of the respective mutants. Mutation of amino acids EDAK_142-145 or KQTK_154-157 caused an intermediate phenotype, with a reduction in EC tropism of about 50% with either mutant in comparison to the wild type. CCTA mutation of amino acids DGTR_165-168 or HVFRD_181-185 caused a strong reduction in the relative EC tropism (infection rate in HUVECs/ infection rate in HFFs), from 0.59 for wild-type HCMV- TB40-BAC4 to 0.04 for mutant HCMV-TB40-BAC4-UL130ccta165-168 and 0.07 for mutant HCMV-TB40-BAC4-UL130ccta181-185 (Fig. 2A and B).

FIG. 2.

Relative endothelial cell tropisms of UL130ccta mutants. Fibroblasts (HFFs) and endothelial cells (HUVECs) were infected with various UL130ccta mutants at an infection multiplicity of 0.9 infectious unit/cell. One day after infection, viral immediate-early antigens were detected by indirect immunofluorescence staining (Cy3, red nuclear signals). Cell nuclei were counterstained with DAPI (blue nuclear signals). (A) Infection efficiency in HUVECs and HFFs. One example for each phenotype category is shown: wild-type HCMV-TB40-BAC4, no-phenotype HCMV-TB40-BAC4-UL130ccta101-102, intermediate-phenotype HCMV-TB40-BAC4-UL130ccta142-145, and strong-phenotype HCMV-TB40-BAC4-UL130ccta165-168. (B) Relative EC tropisms of all UL130ccta mutants, determined as the ratio of infection efficiency in HUVECs and infection efficiency in HFFs. Bars represent mean values from at least three experiments (the standard error of the mean is indicated with each bar). Asterisks indicate a highly significant (P < 0.001) difference between the wild-type virus and the marked mutant. (C) Relative EC tropisms of UL130ccta revertants as described for panel B.

Revertant viruses were constructed to exclude the possibility that the change in EC tropism of the C-terminal CCTA mutants was due to unspecific second-site mutations. Again, markerless mutagenesis was applied to restore the wild-type sequence. The relative EC tropisms of all revertant viruses were indistinguishable from that of the wild-type HCMV-TB40-BAC4, thus proving that the phenotypic changes of the mutants were caused by the respective charge cluster mutations (Fig. 2C).

Taken together, the phenotypes of the CCTA mutants clearly indicate the C terminus as the EC tropism-relevant portion of UL130, either for complex formation or for endocytic uptake in ECs.

Deletion of the C-terminal pentapeptide motif PNLIV destroys EC tropism.

The C-terminal amino acids PNLIV of pUL130 bear a striking similarity to cell-penetrating pentapeptides (CPPs) described in the literature (17, 32). CPPs are thought to mediate endocytic uptake, which is also the proposed entry mechanism of HCMV in ECs (3, 20). To test the hypothesis that amino acids PNLIV of UL130 are also a cell-penetrating motif, we generated a deletion mutant on the basis of highly endotheliotropic TB40-BAC4 (TB40-BAC4-UL130ΔPNLIV) by markerless mutagenesis. As with the CCTA mutants, the phenotype of reconstituted virus HCMV-TB40-BAC4-UL130ΔPNLIV was characterized with regard to its EC tropism by comparing infection efficiencies of cell-free virus preparations in HUVECs and HFFs. HCMV-TB40-BAC4-UL130ΔPNLIV was almost unable to infect endothelial cells, whereas it replicated normally in HFF. The relative endothelial cell tropism (infection rate in HUVECs/infection rate in HFFs) was reduced from 0.6 for wild-type HCMV-TB40-BAC4 to 0.03 for mutant HCMV-TB40-BAC4-UL130ΔPNLIV (Fig. 3). A revertant virus (HCMV-TB40-BAC4-UL130addPNLIV) showed the same phenotype as the wild-type HCMV-TB40-BAC4, confirming that loss of EC tropism was specifically caused by deletion of amino acids PNLIV.

FIG. 3.

Deletion of the C-terminal pentapeptide motif PNLIV of UL130 renders HCMV-TB40-BAC4 nonendotheliotropic. (A) Infection efficiency of HCMV-TB40-BAC4-UL130ΔPNLIV and its revertant virus HCMV-TB40-BAC4-UL130addPNLIV in HUVECs and HFFs in comparison to the wild-type HCMV-TB40-BAC4. Immediate-early antigen was detected at 1 day postinfection (Cy3); nuclei were counterstained with DAPI. (B) Relative EC tropisms of all UL130PNLIV mutants, determined as the ratio of infection efficiency in HUVECs and infection efficiency in HFFs. Bars represent mean values from at least three experiments (the standard error of the mean is indicated with each bar). Asterisks indicate a highly significant (P < 0.001) difference between the wild-type virus and the marked mutant.

Since CPPs are usually not dependent on a specific order of amino acids, we hypothesized that changing the sequence of amino acids PNLIV should have no influence on EC tropism. Surprisingly, a scrambled mutant (HCMV-TB40-BAC4-UL130scramPNLIV, generated on the basis of TB40-BAC4-UL130ΔPNLIV) in which the PNLIV sequence was changed to PINVL was also as nonendotheliotropic as the deletion mutant HCMV-TB40-BAC4-UL130ΔPNLIV, with a relative EC tropism of 0.03. As this result was unexpected, an identical mutant was generated independently on the basis of TB40-BAC4, but the reconstituted virus showed the same nonendotheliotropic phenotype.

To analyze whether a known cell-penetrating peptide motif would restore the function of the last five amino acids of UL130, we exchanged PNLIV against the CPP described by Rhee and Davis (17), PFVYLI. If the last five amino acids of UL130 had only cell-penetrating function, HCMV-TB40-BAC4-UL130PFVYLI should be similar to wild-type HCMV-TB40-BAC4 with regard to uptake into and infection of ECs. The characterization of this mutant, however, showed it to be poorly endotheliotropic (relative EC tropism of 0.04), similar to the deletion mutant HCMV-TB40-BAC4-UL130ΔPNLIV and the scrambled mutant HCMV-TB40-BAC4-UL130scramPNLIV (Fig. 3B).

Poorly endotheliotropic phenotypes are caused by loss of the gH/gL/pUL128-pUL130-pUL131A complex from virions.

Previous reports regarding the function of pUL130 suggested that this protein might be critical for formation of the pentameric gCIII complex. The C-terminal frameshift naturally occurring in HCMV strain Towne has been reported to cause accelerated lysosomal degradation of pUL130 compared to the wild-type protein (15), and lack of the UL130 protein results in loss of the complete pentameric complex from the virion envelope (21). It was therefore tempting to speculate that the C-terminal UL130 mutations that reduced EC tropism in our analysis also affected incorporation of this complex into mutant virions. To test this hypothesis, we purified virion particles of the PNLIV deletion mutant and one of the C-terminal CCTA mutants with the most severe phenotype (HCMV-TB40-BAC4-UL130ccta165-168) and analyzed them for the presence of pUL128, pUL130, and pUL131A by immunoblotting. Virions from wild-type HCMV-TB40-BAC4 and a mutant lacking the complete UL128 locus (HCMV-TB40-BAC4delUL132-128) served as positive and negative controls, respectively. Cell lysates were tested for the presence of pUL130 in order to prove that the mutated protein is expressed properly (Fig. 4A). As expected, wild-type virus incorporated all three proteins of the UL128 locus into virions. In contrast, virions of the PNLIV deletion mutant and the UL130ccta165-168 mutant were almost completely lacking pUL128, pUL130, and pUL131A (Fig. 4B and C), although the mutated UL130 protein was detectable in cell lysates (Fig. 4A). Obviously, at least with the two mutants tested, alteration of the C terminus was critical for incorporation of the pentameric gCIII complex into virus particles.

FIG. 4.

Detection of pUL128, pUL130, and pUL131A in virions of UL130 mutant viruses. (A) Immunoblot detection of pUL130 in lysates of cells infected with wild-type TB40-BAC4, mutant TB40-BAC4-UL130ΔPNLIV, and mutant TB40-BAC4-UL130ccta165-168. Lysates of TB40-BAC4ΔUL132-128 served as a negative control. (B) Immunoblot detection of pUL128, pUL130, and pUL131A in lysates of purified virions of wild type TB40-BAC4, mutant TB40-BAC4-UL130ΔPNLIV, and mutant TB40-BAC4-UL130ccta165-168. Virion lysates of TB40-BAC4ΔUL132-128 served as a negative control. (C) Densitometric evaluation of immunoblot signals from three experiments as shown in panel B. Columns represent mean values relative to the wild-type control. Error bars represent standard errors of the means.

DISCUSSION

Here we report the identification of peptide motifs within pUL130 that contribute to endothelial cell tropism of HCMV. We combined a comprehensive charge-cluster-to-alanine scanning approach and a hypothesis-driven approach investigating a C-terminal candidate cell penetration peptide. The effects of mutations introduced in the respective sites ranged from almost complete destruction to about 50% reduction of EC tropism. All of the identified peptide sites were located in the C-terminal third of UL130 (aa 142 to 214), emphasizing the role of this domain in the proper function of the protein. Support for this finding comes from sequence comparisons between 12 clinical HCMV isolates from four geographic regions (see Fig. S1 in the supplemental material). While the N-terminal half of pUL130 showed interstrain variation on 11 out of 107 amino acids, the C-terminal half was remarkably conserved, with only one isolate showing a T-to-I exchange at amino acid position 167. This region is also highly conserved between HCMV and chimpanzee CMV (38), with 82% identity and 98% similarity. Particularly, charge clusters 165 to 168 and 181 to 185, mutation of which caused a severe reduction in EC tropism, are 100% conserved in CCMV. Among charge clusters whose mutation caused a 50% reduction in EC tropism, cluster 154 to 157 is completely conserved and cluster 142 to 145 carries only one neutral mutation (glutamic acid to aspartic acid). In contrast, the homologous parts of the N-terminal region (aa 25 to 139) show only 56% identity and 66% similarity between HCMV and CCMV. Not unexpectedly, charge clusters 48 to 51, 82 to 86, and 127 to 131 within this region are mutated in the CCMV genome. Remarkably, however, the series of three charged amino acid doublets between aa 94 and 109 is also conserved, showing only one neutral mutation (glutamic acid to aspartic acid). It is tempting to speculate that this site of the protein is functionally important. Even though mutation of these charge clusters did not have any effect on EC tropism in our study, a function for this striking motif might be revealed by further analysis of a combined mutant where all three amino acid doublets are changed in one virus.

Possible functions of the four phenotypically relevant charge clusters are (i) interactions with other components of the gH/gL/pUL128/pUL130/pUL131A complex or with other viral or cellular proteins during assembly and maturation of the complex or (ii) interaction with cellular receptor or coreceptor molecules during viral entry as part of the pentameric gCIII complex.

The UL130ccta165-168 mutation caused loss of all three proteins of the UL128 locus from virion particles, which proves that the C terminus is involved in complex formation. While this agrees well with a previous report showing that lack of UL130 causes lack of the whole complex within virus progeny (21), it does not formally exclude an additional function of this part of the protein in virus-cell interactions. Still, a direct involvement of pUL130 in attachment and/or fusion is only hypothetical to date. Indirect evidence comes from the fact that antibodies against pUL130 have been reported to neutralize HCMV virions with regard to EC infection in an epitope-specific way (35) and to inhibit plaque formation specifically in ECs (10). Charge clusters 142 to 145 and 154 to 157, causing an intermediate phenotype, are preferential candidates for such a function.

By choosing the charge-cluster-to-alanine approach, we focused in this study on the identification of protein-protein interaction sites, as charge clusters are assumed to be exposed on the surface of the protein and may interact with charged areas on other proteins. In principle, any of the charge clusters might also contribute to the hypothetical chemokine function of pUL130 that has been suggested (11, 29). However, this study focused on the contribution of pUL130 to EC tropism, as this function of UL130 is well established. Functional evidence for a chemokine-like function of UL130 is still lacking and was not tested here (i.e., C57 and C83 were not mutated).

As pointed out in a number of publications, extended propagation of HCMV strains on fibroblast cultures leads to a loss of EC tropism, which is regularly associated with mutations in the UL128 locus (1, 2, 4, 8, 11, 15). In HCMV strain Towne, UL130 is mutated with a frameshift replacing the C-terminal 11 amino acids by 26 amino acids (8). This mutation severely reduces EC tropism of this strain, indicating an important sequence motif within the C-terminal 11 amino acids. One distinctive feature is the C-terminal pentapeptide PNLIV, which is conserved between HCMV and CCMV and resembles a previously published cell-penetrating peptide (17). Actually, loss of this pentapeptide severely reduced EC tropism, supporting the idea that this motif is essential. Surprisingly, however, the function could not be restored by addition of a scrambled version of this pentapeptide (PINVL) or by addition of a known cell-penetrating peptide (PFVYLI) as described in the literature (17). While this emphasizes the remarkable sensitivity of the C terminus of pUL130 to even slight modifications, it also indicates that the “cell-penetrating” feature of the C terminus is alone not sufficient to restore the function. We therefore assumed that the PNLIV motif is involved in interactions other than those previously reported for the PFVYLI motif, and this assumption was supported by the finding that deletion of PNLIV resulted in loss of the complete pentameric complex from virion particles. Accelerated lysosomal degradation has been recently reported with Towne pUL130 compared to wild-type pUL130 (15), suggesting a function of the C terminus in stabilization of the protein, which is not restored by other CPP motifs. Still, in the intact protein the PNLIV motif may have the additional function to support endocytic uptake. A CPP-like function of PNLIV could be tested by fusing it to otherwise inert peptides, which may be interesting for the issue of targeted drug delivery.

The identification of functionally relevant peptide sites within pUL130 may also have implications for the development of novel antiviral strategies. The two charge clusters whose mutation caused intermediate phenotypes may represent sites for targeted modest attenuation of HCMV strains. This may become relevant in the context of live vaccine development, when reduced but not completely abolished viral fitness is desired. Such a concept is suggested by the finding that the poorly EC-tropic vaccine candidate strain Towne was inefficient in inducing neutralizing antibodies against epithelial cell infection, which may explain its failure to induce protective immunity (6, 16). Under the assumption that at least some of the charge clusters identified in this study may be involved in virion-host cell interaction during viral entry, these charge clusters may provide targets for antiviral approaches using interfering peptides containing the respective amino acid sequences.