Dense Bodies of a gH/gL/UL128/UL130/UL131 Pentamer-Repaired Towne Strain of Human Cytomegalovirus Induce an Enhanced Neutralizing Antibody Response

HCMV envelope proteins are expressed on the surface of laboratory strain DBs in a conformation suitable to mediate entry of the particles into human fibroblasts by membrane fusion (25, 26). The laboratory strain DBs, however, display little efficiency in penetrating epithelial cells (C. Lehmann, unpublished data), likely related to the absence of the PC on their surface, resembling observations on HCMV laboratory strains that fail to infect epithelial cells (35). In a first set of experiments, we investigated the level of neutralization against a PC-positive viral strain by sera from animals that had been immunized with PC-negative DBs. New Zealand White rabbits were immunized with laboratory strain DBs (DB-HB5) according to the schedule displayed in Fig. 1A. Sera were collected at the indicated time points and tested in microneutralization assays. As expected, the sera neutralized the PC-negative homologous strain HB5 on fibroblasts already at a low antibody concentration (Fig. 1B). The sera also neutralized the PC-positive strain TB40/E-BAC_KL7 (here TB40/E) on HFFs (Fig. 1C), although a higher antibody concentration was necessary. Similar 50% neutralizing titers (NT₅₀) were detected when ARPE-19 cells were used as targets, indicating that the sera were equally effective blocking TB40/E infection in HFF and epithelial cells (Fig. 1D). Taken together, these results showed that rabbit sera induced by HCMV laboratory strain DBs contain neutralizing antibodies that impair viral infection of TB40/E on both HFF and ARPE-19 cells. They also indicated that the level of neutralization of TB40/E was comparable for the two cell types, despite the lack of the PC in the material used for immunization.

The initial experiments and the results from others (29) indicated that the fully assembled envelope protein complexes of laboratory strain DBs lacking the PC already induced a distinct neutralizing antibody response against infection by a PC-positive strain. The PC had, however, been identified as a dominant target of the NT-abs response against HCMV. We thus designed experiments to analyze the impact of antibodies against the PC using side-by-side immunization with PC-positive and PC-negative DBs. To be able to directly compare the results, we decided to use a pair of viruses for DB production that only differed in PC expression. As a basis for this, we selected the Towne strain, which is a high-level DB producer. The Towne strain carries a double adenine insertion in the open reading frame (ORF) of the UL130 gene, leading to a frameshift (16, 36). The parental Towne strain is thus PC negative. To generate a Towne variant that expresses the PC, the mutated UL130 open reading frame in the BACmid pTowne (37) was replaced by its functional homolog from the TB40/E strain, using the galK negative-positive selection procedure (38). The resulting BACmid was denominated pTowne-UL130rep (Fig. 2A). We chose TB40/E UL130, as the expression of the PC by TB40/E appears to be conserved upon serial passaging in fibroblast cell cultures (C. Sinzger, unpublished data) (39, 40). Following reconstitution on HFF, virions and DBs of the novel strain Towne-UL130rep were purified by gradient ultracentrifugation. These particles were subjected to immunoblot analysis and probed with antibodies against pUL130, pUL128, and glycoprotein B (loading control, Fig. 2B). Particles from the laboratory strains HB5, the parental strain Towne, and from TB40/E were carried along as a control. pUL130 and pUL128 were both detectable in virions and DBs of Towne-UL130rep. According to current knowledge, both proteins are inserted into the envelope of extracellular viral particles only in complex with gH/gL and UL131. This suggested that the PC was successfully restored in Towne-UL130rep. To formally prove the functionality of the complex, ARPE-19 cells were infected with Towne-UL130rep. Infection efficiency was compared to parallel cultures infected with the parental Towne strain or with TB40/E and was measured by counting immediate early protein 1 (IE1)-positive cells (Fig. 2C). The median number of IE1-positive cells in the TB40/E-infected cultures was set to 100%. The infection rate of Towne-UL130rep was comparable to that of TB40/E. The parental Towne strain showed less than 20% IE1-positive cells, compared to the rate of positive cells, following TB40/E infection in that culture. Since the PC is considered to be important for the infection of epithelial cells, these results provided further evidence for successful reconstitution of the PC in Towne-UL130rep.

To further address this issue, ARPE-19 cells were tested for their capacity to support Towne-UL130rep DNA replication. Cells were infected with Towne or Towne-UL130rep, using identical input genome copy numbers. HFFs were infected in parallel (Fig. 2D). Genome replication was assessed by quantitative PCR analysis. Towne-UL130rep genome replication was detectable in epithelial cells. No replication of the parental strain Towne was seen. Replication was detectable to high levels in HFF, with little difference between Towne and Towne-UL130rep. To address whether virus was released from Towne-UL130rep-infected epithelial cells and HFFs, culture supernatants were probed for viral genomes, again using quantitative PCR analysis (Fig. 2E). Supernatants from Towne-UL130rep-infected epithelial cells contained increasing amounts of viral genomes over time, indicating that infectious virus was being released. In contrast, a decrease in the genome copy numbers was seen in Towne-infected ARPE-19 cultures. HFFs infected with either strain released equal genome copy numbers. Taken together, these data showed that the Towne strain had been functionally repaired in its capacity to replicate in epithelial cells and to release progeny from these cultures.

To analyze if the phenotype of Towne-UL130rep resembled that of other PC-positive HCMV strains, we investigated the cytopathogenic changes that were induced by that virus on HFFs, using immunofluorescence analyses. Restoration of the PC in the laboratory strain AD169 is associated with the appearance of multinucleated giant cells (35). It is also a characteristic of the PC-positive strain TB40-E (C. Sinzger, unpublished data). To verify that Towne-UL130rep shared that phenotype, HFFs were infected for 8 days. Cells were then tested for the induction of cell-cell fusion by indirect immunofluorescence staining, using labeling with a pp65-specific monoclonal antibody for detection (Fig. 3). Towne-UL130rep-infected HFFs indeed formed large multinucleated cells which were absent in Towne-infected cells. Substantial cell-cell fusion was also seen in cultures that were infected with TB40/E for a control. These results confirmed that the repair of UL130 in the HCMV strain Towne created a virus that was phenotypically indistinguishable from other PC-positive viruses.

Mass spectrometry of purified DBs of strains Towne and Towne-UL130rep confirmed that both particles differed only with respect to the presence of the UL128-131 (not shown). DB preparations from the two viruses were thus perfectly suitable to investigate the difference in antibody responses mediated by the presence of a functionally active PC.

In a first set of experiments, differences in the induction of neutralizing antibodies following immunization with PC-positive (PC-pos) versus PC-negative (PC-neg) DBs were addressed on fibroblasts. In a first round of immunizations, rabbits were immunized with 25 μg DB-Towne-UL130rep or DB-Towne according to the schedule shown in Fig. 4A. The sera from these animals, collected at days 66 and 86, were analyzed for their virus neutralization capacity using TB40/E for infection and immunofluorescence staining of viral IE antigen as readout. Sera from animals that received DB-Towne-UL130rep neutralized TB40/E significantly more efficiently on fibroblasts than did sera following application of DB-Towne (Fig. 4B and C). To corroborate these findings, another immunization experiment was performed in rabbits according to a shorter schedule using another DB batch (Fig. 4D). For analysis of these sera, a neutralization (NT) assay, based on the detection of Gaussia luciferase expressed by the reporter virus TB40-BAC4-IE-GLuc, was applied (41). Again, DB-Towne-UL130rep induced a significantly more efficient neutralizing response, especially at day 28 after the first immunization, additionally reflecting an enhanced boosting capacity. Taken together, these results suggest that PC-positive DBs are superior in the induction of neutralizing antibodies against fibroblast infection by a strain that contains the complete set of envelope proteins (TB40/E).

Both epithelial cells and endothelial cells are key targets of HCMV infection in vivo (42). An important issue thus was to evaluate whether the application of PC-pos DBs increased the neutralization capacity against infection of these cells. In a first experiment, mice were immunized twice intraperitoneally in a 2-week interval with DB-Towne or DB-Towne-UL130rep. Samples were collected at the end of week 4. The serum samples of 5 mice in each group were pooled and analyzed in microneutralization assays on ARPE-19 cells. Sera from mice that had been immunized with DB-Towne-UL130rep showed higher neutralization capacity on ARPE-19 cells than did the sera from animals that had been immunized with DB-Towne (Fig. 5). The results showed indeed that PC-positive DBs induced a more potent NT-abs response in mice which is protective against the infection of epithelial cells with a PC-pos HCMV strain.

To confirm these results in another animal species and on different cells, the sera from rabbits that had been used for the analysis on HFFs (see Fig. 4) were also used for neutralization assays on HEC-LTT endothelial cells (43, 44). The sera obtained after a four-dose application schedule (Fig. 6A) of DB-Towne-UL130rep (Towne-UL130rep) and sampling at day 66 or 86 showed enhanced neutralization capacity against strain TB40/E infection on endothelial cells, compared to sera from animals immunized with DB-Towne (Fig. 6B and C, Towne). To confirm these results, the sera from the independent immunization experiment (see Fig. 4) were tested accordingly. Again, sera obtained at two different times after application of DB-Towne-UL130rep showed enhanced neutralization of strain TB40/E on endothelial cells compared to DB-Towne application. Taken together, the experiments demonstrate the potential for the PC to enhance and broaden DB immunogenicity against HCMV.

The BAC vector, originally used for the cloning of Towne BAC, expresses the green fluorescent protein (GFP). This protein is fortuitously packaged into DBs of Towne BAC and of Towne-UL130rep (proteomic analysis, not shown). To avoid the potential risk of adverse effects of GFP following DB application to humans, we deleted the GFP gene from Towne-UL130rep in order to establish a seed virus strain for the production of DBs for clinical studies. The galK gene was used to replace the GFP-gene in Towne-UL130rep, again using BAC technologies (Fig. 7A). The galK gene is driven by a prokaryotic promoter and is thus not expressed in eukaryotic cells. The resulting viral strain was denominated Towne-UL130rep-ΔGFP. The expression of pUL130, representative of PC expression, on purified DBs and virions was tested by immunoblotting (Fig. 7B). Abrogation of GFP expression was tested both by direct immunofluorescence (not shown) and by immunoblotting using a GFP-specific antibody (Fig. 7B). The functionality of the PC was verified by infecting both HFF and ARPE-19 cells with Towne-UL130rep-ΔGFP and staining for IE1 expression, using indirect immunofluorescence analyses (Fig. 7C). Taken together, these results showed that the newly established seed virus strain Towne-UL130rep-ΔGFP lacked expression of the green fluorescent protein but retained PC expression. These data show that DBs of strain Towne-UL130rep-ΔGFP are suitable for the production of a DB vaccine candidate for clinical studies.

The PC has been identified as a major target of the NT-abs response during natural HCMV infection (17). Dominant neutralizing epitopes have been described on the PC (45). A recombinant PC expressed in CHO cells induced antibodies in mice with higher neutralization capacities than with antibodies induced by gB immunization (19). A multiantigenic PC-incorporating modified vaccinia Ankara (MVA) vector vaccine candidate is able to induce neutralizing antibodies as well as mouse and human major histocompatibility complex (MHC)-restricted polyfunctional T cell responses (46). Detection of PC-specific antibodies in human serum may correlate with virus control in vivo and with decreased incidence of HCMV transmission in pregnancy (47, 48). Still, the issue of whether the PC is required for a successful vaccine formulation in general remains unsettled. Vaccines like gB/MF59 provided some levels of protection against infection, fostering efforts to solely use gB for further development (12, 14). In the light of the above-mentioned findings, most currently pursued approaches, however, favor the addition of the PC to an HCMV-targeting vaccine (4, 6, 46, 49–53). The data obtained herein appear to support this notion. Although PC-negative DBs induced moderate NT-abs responses against the infection of epithelial and endothelial cells, the responses to PC-positive DBs were superior in both animal species tested. This suggests that the PC is required on DBs to induce a broad NT-abs response. The data clearly support efforts to evaluate the potential of PC-positive DBs in clinical studies in humans.

Besides epithelial and endothelial cells, fibroblasts have also been identified as target cells of HCMV in vivo (54). Infection of these cells is considered to be mediated by membrane fusion at the cell surface, although work described in a recent publication would argue in favor of macropinocytosis for this entry pathway (55). In any case, PC-negative laboratory strains and PC-positive strains of HCMV enter fibroblasts with comparable efficiency. Thus, an exclusively PC-independent pathway of infection of these cells is commonly accepted. Surprisingly, rabbit sera obtained after the application of PC-positive DBs showed higher neutralization capacity against infection of HFFs with the PC-positive TB40/E strain than did PC-negative sera (see Fig. 4). The fact that no difference was found when HFFs were infected with the PC-negative HCMV strain Towne indicates that the antibodies mediating this effect are directed against the PC. These results are remarkable, as they raise the issue of whether the PC also contributes to entry of cell-free HCMV into fibroblasts. While, in principle, steric hindrance of virus-fibroblast interactions by PC-bound antibodies or complement-mediated disruption of the virion could explain the observed findings, there is increasing evidence favoring a contribution of the pentamer to infection of fibroblasts. Overexpression of the gH/gL/gO trimer in HFFs could not completely block HCMV infection by interference, whereas overexpression of the PC in ARPE-19 cells completely blocked HCMV infection (56). Moreover, a recent study found that a combination of antibodies against the trimer and the pentamer inhibited the PC-positive strain VR1814 more effectively than did anti-trimer antibodies alone, and the authors provided data suggesting that a pentamer-mediated activation of the ErbB pathway might be involved (57). Finally, NRP-2 was recently identified as a cellular receptor of the pentamer, and this receptor is also expressed on fibroblasts (58). Both soluble NRP-2 and antibodies against NRP-2 caused a slight reduction in HCMV infection on MRC5 cells, indicating a minor contribution of the pentamer also to infection of fibroblasts. Taken together, these reports are consistent with our finding concerning a previously unnoticed role of the pentamer for entry of PC-positive HCMV strains into fibroblasts.

Human fibroblasts are the only culture system described so far that supports DB production. Clinical isolates initially express the PC when passaged on fibroblast cultures, yet most of these strains lose PC expression due to a mutation in one of the genes of the UL128-131 locus (39, 40). They, in addition, synthesize very limited amounts of DBs. We thus chose to repair the laboratory strain Towne by inserting the UL130 gene of TB40/E into Towne BAC, thereby replacing the mutated Towne gene. The expression level of the PC in TB40/E-infected fibroblasts appears to be moderate, thus resulting in its retention upon multiple passages (39) (C. Sinzger, unpublished data). In accordance with this, we did not observe a loss of PC expression following passaging of Towne-UL130rep on fibroblasts. Thus, Towne-UL130rep is suitably stable for DB production purposes. The deletion of the gene encoding GFP, however, appeared to be mandatory to exclude possible adverse effects by the application of DBs containing this protein. Consequently, the establishment of Towne-UL130rep-ΔGFP now provides a seed strain appropriate for the establishment of a downstream production process for a DB-based vaccine to be tested in clinical studies.

In conclusion, this work shows that DBs containing the PC are suitable to induce a distinct neutralizing antibody response against HCMV infection. Antibody responses against these particles may, on the other hand, be instructive in future studies to more closely define the mechanisms of entry of HCMV into different cell types.

Primary human foreskin fibroblasts (HFFs) were cultured as described before (59). Retinal pigment epithelial cells (ARPE-19) were obtained from the ATCC and were cultured in Dulbecco’s modified Eagle’s medium (DMEM) plus Ham’s F12 medium (1:1), supplemented with 10% (vol/vol) fetal calf serum. Human umbilical vein endothelial cells (HUVECs), conditionally immortalized with tetracycline-dependent expression of the SV40 large-T antigen and human telomerase reverse transcriptase (hTERT) (HEC-LTT), were cultured as described previously (43, 44). For growth, HEC-LTTs were cultured in endothelial cell growth medium (EGM BulletKit; Lonza) supplemented with 2 mg/ml doxycycline (AppliChem); for neutralization assays, doxycycline was omitted. All HCMV strains used in this analysis were derived from bacterial artificial chromosome (BAC) clones. HB5, a derivative of the laboratory strain AD169, was obtained from Borst et al. (60). TB40/E-BAC_KL7 (here denoted TB40/E) is a derivative of the endotheliotropic strain TB40/E (61) that carries an intact UL US-gene region (62). TB40-BAC4-IE-GLuc is a derivative of TB40/E expressing the Gaussia luciferase under the HCMV IE promoter at an ectopic position (41). The Towne strain was reconstituted from Towne BAC (37, 63), kindly provided by Edward Mocarski. The Towne BAC was originally cloned from a plaque-purified derivative of the Towne strain (ATCC; VR-977). The nucleotide sequence of Towne BAC was established by Murphy et al. (64).

The generation of the UL130 repaired version of the Towne BAC strain (Towne-UL130rep) was performed by the galK positive/negative selection method as described by Warming and colleagues (38) and detailed for HCMV BAC mutagenesis by Mersseman and colleagues (33). In a first step, the galK gene was amplified from pgalK (38). The primers used for this contained 50 nucleotides homologous to the UL130 locus of HCMV, thus allowing replacement of the mutated UL130 gene in Towne BAC by galK. Clones obtained by this were verified by colony PCR analysis. In a second step, the galK gene was replaced by the wild-type (wt) UL130 gene, amplified from TB40/E-BAC_KL7. The UL130-specific primers used for this were flanked by the same 50 nucleotides that had been used for galK insertion into Towne BAC. This allowed for seamless insertion of the wt U130, thereby replacing galK. After isolation of positive colonies, following negative selection for galK expression using desoxygalactose, the correct replacement was again verified by colony PCR. The BAC DNA of individual clones was then further characterized by restriction endonuclease digestion and agarose gel electrophoresis, as well as by DNA sequence analysis.

Virus reconstitution was achieved by transfecting of BAC DNA into HFFs. BACmid DNA for transfection was obtained from Escherichia coli using the Plasmid purification kit (Macherey & Nagel, Düren, Germany) according to the manufacturer’s instructions. Transfections into HFFs were performed using the SuperFect transfection reagent (Qiagen, Hilden, Germany). HFFs were seeded on 6-well plates at a density of 1 × 10⁵ cells/well using different BAC DNA concentrations. Cells were subsequently passaged until plaques became visible. The infectious supernatant was then transferred to uninfected cells for passaging of the virus. All HCMV strains were propagated on HFFs. Viral stocks were obtained by collecting the culture supernatants from infected HFFs, followed by low-speed centrifugation to remove cell debris. Supernatants were frozen at −80°C until further use.

Virus titers were determined by staining for the expression of the immediate early 1 protein (IE1) ppUL123, using monoclonal antibody p63-27 (65), kindly provided by William Britt. For this, 5 × 10³ HFF or ARPE-19 cells were seeded in each well of a 96-well plate. The following day, virus stocks were diluted to 10⁻³ and 10⁻⁴ in culture medium and were added to the cells in octuplet replicates. Cells were fixed after 48 h for 20 min using 96% ethanol. The primary antibody p63-27 was added for 1 h in a humidified chamber at 37°C. Detection was performed by adding an anti-mouse IgG, coupled to horseradish peroxidase (HRP) (rabbit-anti-mouse immunoglobulin HRP; Dako, Hamburg, Germany) at a dilution of 1:500 for 1 h and by subsequent staining with 3-amino-9-ethyl-carbazol (AEC)–H₂O₂ for another hour. IE1-positive cells were counted, and titers were determined as the means of octuplet values.

Indirect immunofluorescence and immunoblot analyses were performed as previously published (66). Expression of pUL128 and pUL130 on viral particles was analyzed by immunoblotting, using gradient purified material. Three micrograms of virions and DBs was loaded per slot on a 4 to 12% Bis-Tris-polyacrylamide gel. After transfer to a polyvinylidene difluoride (PVDF) membrane (Immobilon-FL; Millipore, Billerica, MA, USA), proteins were detected using specific antibodies. The following antibodies were used for detection: for pUL128, 4B10 (undiluted), and for pUL130, 3C5 (1:250 dilution), both kindly provided by Thomas Shenk (15); and gB:27.287 (undiluted), kindly provided by William Britt.

Viral DNA replication and release of viral DNA into the culture supernatant were performed as described before (59), using an Applied Biosystems 7500 real-time PCR system (Life Technologies GmbH, Darmstadt, Germany). Extraction of viral DNA from infected cells or culture supernatant was performed using the High Pure viral nucleic acid kits (Roche, Mannheim, Germany).

Immunizations of rabbits were performed by Eurogentec according to all Federation for Laboratory Animal Science Associations (FELASA) recommendations (https://secure.eurogentec.com/product/research-animal-care.html). Mouse immunizations were performed at the Translational Animal Research Center of the Johannes Gutenberg-University Mainz according to national and international regulations for animal welfare. Approval for the mouse immunization experiments was obtained from the local regulatory authority (Landesuntersuchungsamt Rheinland-Pfalz, Koblenz, Germany).

A microneutralization assay, adapted from a publication by Andreoni and colleagues (67), was used for the experiments shown in Fig. 2 and 5. Serum samples were heated at 56°C for 30 min for complement inactivation. Twofold serial dilutions were prepared and added to an equal volume of virus. Virus and serum were incubated for 4 h at 37°C in a 5% CO₂ atmosphere. HFFs were seeded on 96-well flat-bottom culture plates (BD Falcon, Heidelberg, Germany) at a density of 1.5 × 10⁴ cells/well in minimal essential medium (MEM) containing 5% fetal calf serum (FCS) per well for 15 min prior to the addition of the virus-antibody mixture. ARPE-19 cells were seeded at the same density in 96-well plates that had been pretreated with 0.1% gelatin, using DMEM-Ham’s F12 medium and 10% FCS. In this case, the virus-antibody mixture was added 3 h after seeding of the cells. Incubation with the cells was for 24 h at 37°C and 5% CO₂ for both HFF and ARPE-19 cells. After that, the supernatants were removed and the cells stained for IE1 expression, using either the antibody p63-27 and horseradish peroxidase (HRP)-coupled secondary antibodies (Plachter laboratory) or E13 and Cy3-coupled secondary antibodies (65) (Sinzger laboratory). The number of positive cells was counted in quadruple replicates. Each of the four values was divided by the mean value calculated from quadruple replicates of the positive control (no serum added) and defined as residual infectivity. The mean of these four residual infectivity values was determined for each serum dilution. The 50% neutralization titer (NT₅₀) was defined as the serum dilution that resulted in an infectivity of 50% or less (i.e., neutralization of 50% or more of infectious virus).

The reporter virus TB40-BAC4-IE-GLuc containing supernatant was purified as previously published (41). Two days prior to the experiment, cells were seeded on a 96 well-plate at a density of 1.5 × 10⁴ cells per well (HFFs) or 1.2 × 10⁴ cells/well (HEC-LTTs). For HEC-LTTs, plates were coated with 0.1% gelatin (Sigma-Aldrich). HFFs were seeded in MEM supplemented with 5% fetal bovine serum (MEM5; PAN-Biotech), GlutaMAX (Life Technologies), and 100 μg/ml gentamicin (Sigma-Aldrich). HEC-LTTs were seeded in endothelial growth medium (EGM) without doxycycline. For the neutralization assay, rabbit serum samples were 2-fold serially diluted in MEM5, mixed with the reporter virus to achieve an MOI of 1 (as calculated for HFFs), and incubated for 2 h at 37°C to allow for neutralization. Thirty minutes prior to infection, the medium of the prepared cell cultures was exchanged with fresh MEM5, and after that, preincubation cell cultures were infected with the serum-virus mixture for an additional 2 h at 37°C. The mixture was then removed and replaced by the appropriate growth medium for the respective cell type. Infected cultures were incubated overnight to allow expression and release of Gaussia luciferase, and luciferase activity in the supernatants was measured the following day as previously published (41), yielding values of relative light units (RLU). All RLU values were normalized to the mean RLU value of cells treated only with virus. Normalized RLU values for the various serum dilutions thus reflected relative infection levels compared to the untreated control. To determine the neutralization capacity of the rabbit serum samples, concentrations were plotted against the relative inhibition (1 − normalized RLU). The serum dilution that reflected a half-maximal reduction in Gaussia luciferase activity was determined by nonlinear regression (SigmaPlot) and given as the NT₅₀.