The human cytomegalovirus (HCMV) is well-recognized as a clinically important pathogen. Transmission of the virus during pregnancy and the resulting congenital HCMV infection (cCMV) are frequently associated with severe sequelae [1
]. The development of a vaccine against cCMV has thus been defined as a top-priority medical goal [4
]. Additionally, HCMV reactivation is a severe complication of both solid organ and hematopoietic stem cell transplantation [6
]. The establishment of a vaccine for the prevention of HCMV-related complications in these settings is highly desirable [8
Several vaccine candidates are currently being tested in pre-clinical or clinical studies (reviewed in [9
]). However, there is still an ongoing debate with regard to the goals and the appropriate formulations of a vaccine (reviewed in [9
]). The tegument protein pp65 (pUL83) and the immediate-early protein 1 (IE1, pUL123) have gained broad endorsement as being major T lymphocyte antigens to be included in a vaccine. Lesser consensus has been reached regarding the viral proteins that may be necessary to induce protective humoral immune responses following vaccination. The glycoproteins gB (gpUL55) and gH (gpUL75) have been identified as prominent targets of neutralizing antibodies (nabs) [15
]. However, clinical studies have demonstrated only limited protective effects afforded by a gB subunit vaccine [18
]. This suggests that additional antigens might be needed to induce sufficient antibody levels for protection against infection. The pentameric protein complex (PC) of HCMV envelope proteins, consisting of gH, gL, and pUL128-131, has been identified as a crucial component of the HCMV virion that mediates viral entry into a broad spectrum of host cells, including epithelial cells, endothelial cells (EC), and dendritic cells [20
]. The PC has also been found to be a major target of the humoral response, as a large proportion of the nabs capacity in convalescent human sera has been found to be directed against this complex. These findings support the concept of including the PC as a component of a future HCMV vaccine [23
One vaccine candidate that has been studied in our laboratory and by others is based on subviral particles of HCMV, known as dense bodies (DB) [25
] (Table 1
). DB are synthesized in infected fibroblast cell cultures and are released from these cells at late stages of HCMV replication, concomitant with the release of virions [36
] (Figure 1
). DB are devoid of viral capsids and DNA and are therefore non-infectious [38
]. The internal structure of DB mainly consists of pp65 and other tegument proteins [27
]. This electron-dense core is enclosed by a phospholipid bilayer which includes the major viral envelope protein complexes. These complexes are likely inserted into the DB-membrane in a fusion-competent conformation, as they mediate swift entry into cells [40
]. Consequently, antibodies induced by DB application will likely also be suitable to target envelope complexes of infectious virions in their pre-fusion conformation, thereby preventing viral entry into cells.
DB induce both CD4- and CD8-T lymphocyte responses when applied to animals [29
]. This is a remarkable feature considering the fact that DB do not replicate following application. In addition, these responses were achieved without the addition of an adjuvant. DB application also induces distinct nabs responses against HCMV infection [29
]. The impressive immunological properties are likely related to the capacity of DB to induce both activation and maturation of immature dendritic cells (DC; Figure 2
]. The potential of DB as an HCMV vaccine has however been challenged by the fact that most studies regarding the immunogenicity of these particles have been performed with laboratory strains of the virus. DB from these strains lack the PC. As mentioned above, the inclusion of the PC in a prospective vaccine is likely important for its efficacy. DB from laboratory strains do not express the PC due to mutations in the genes encoding UL128-131. We have recently genetically modified a derivative of the laboratory strain Towne to restore expression of the PC [35
]. DB purified from the culture supernatant of primary human foreskin fibroblasts (HFF) infected with this strain contain the functional PC, as these particles enter both fibroblasts and endothelial cells. Immunization experiments in mice and rabbits with these PC-positive particles have shown that the sera of the animals had higher neutralization capacities against HCMV infection in EC and fibroblasts compared to the response following immunization with PC-negative DB [35
]. Consequently, the modified Towne strain, denominated Towne-UL130repΔGFP, provides an excellent basis to develop a production process for a DB-based vaccine. In this communication, we will discuss some of the aspects for the production of a safe DB vaccine for initial clinical studies.
2. Materials and Methods
2.1. Cells, Bacterial Artificial Chromosome (BAC)-Cloning, and Viruses
HFF were cultured as described previously [41
]. All HCMV strains used in this analysis were derived from BAC clones. For downstream cloning within this study, the recently established parental strain Towne-UL130repΔGFP [35
] (hereafter referred to as Towne-repΔGFP) was used. Human umbilical vein endothelial cells (HUVECs) conditionally immortalized with tetracycline-dependent expression of the SV40 large-T antigen and hTERT (HEC-LTT) were cultured as described previously [42
]. For growth, HEC-LTTs were cultured in endothelial cell growth medium (EGM BulletKit, Lonza, Basel, Switzerland) supplemented with 2 mg/mL doxycycline (Applichem, Darmstadt, Germany). Cloning procedures were performed based on the bacterial galactokinase (GalK) positive/negative selection as described by Warming et al. [44
]. Strains HCMV-UL51-FKBP and HB5 were kindly provided by Eva Borst and Martin Messerle [45
For the generation of Towne-UL130rep-GalK-KO (hereafter referred to as Towne-rep-GalK-KO), the GalK-gene, which initially was inserted for the depletion of GFP in the BAC cassette, was seamlessly deleted by recombination with a synthetic DNA fragment consisting of the HCMV-derived homologies flanking GalK. This resulted in the depletion of GalK without insertion of additional sequences.
For the generation of Towne-UL130rep-dUL25 (hereafter referred to as Towne-rep-ΔUL25), the UL25-gene of the parental strain Towne-rep-GalK-KO was replaced by a GalK-cassette which was amplified from the plasmid pGalK [44
]. We used primers comprising 50 base pairs of sequences homologous to the genomic region flanking the UL25 gene (Towne_UL25-GalK_fwd: ACCGGCGCCGCCAAGAAACCGAGCGAAAAGAAACGATCGTCGTCGCGTCGCCTGTTGACAATTAATCATCGGCA, Towne_UL25-GalK_rev: CCTGTGACTTTTTATCATAAACCGTTCCGC CCTGCTGCTTCGTTCCACCATCAGCACTGTCCTGCTCCTT).
Virus reconstitution from BAC-clones was achieved by transfecting column-purified BAC-DNA (Plasmid Purification Kit; Machery&Nagel, Düren, Germany) into HFF with Superfect transfection reagent (Qiagen, Hilden, Germany) as described previously [25
]. Viral stocks were generated by passaging transfected HFF until all cells showed a typical cytopathic effect. The supernatants were then collected and used as seed stocks. Supernatants were frozen at −80 °C until further use. Viral stocks of HCMV-UL51-FKBP were generated in the presence of Shield-1 (1 µM, supplemented every 48 h, Aobious, Köln, Germany).
2.2. Production and Purification of Virions and DB
Virions and DB of HCMV were prepared as previously described [25
]. HFF were infected with culture supernatants containing the virus of interest. For the letermovir experiments, 50 nM or 300 nM of the substance were added to the cell culture during infection and 3 days after initial infection. Culture supernatants from infected HFF were collected 1 week after infection, and, after removal of cellular debris, pelleted via ultracentrifugation. After resuspension, the different components of the resulting pellet were fractionated via glycerol-tartrate density gradient ultracentrifugation [37
]. Subsequently, virions and DB were isolated, concentrated, and stored at −80 °C until further use. For the production of HCMV-UL51-FKBP-derived DB, HFF were infected with HCMV-UL51-FKBP in the initial presence of Shield-1 (1 µM) to allow viral spread through the complete cell culture. After 3.5 days, the Shield-1-containing medium was replaced with Shield-1-free medium to inhibit synthesis of infectious virus while retaining DB production. Supernatants of the cells were harvested 1 week after initial infection and processed as described above.
2.3. SDS-PAGE, Silver/Instant Blue Staining, and Immunoblotting
The protein composition of purified virions and DB was analyzed by SDS-PAGE, followed by either silver staining, instant blue staining, or by immunoblotting, respectively. For silver staining, 2 µg of virions or DB per lane were loaded on a 10% tris-glycine-polyacrylamide gel. For instant blue staining, 20 µg of DB were loaded on a 4–12% bis-tris-polyacrylamide gel (Thermo Fisher Scientific, Darmstadt, Germany). For immunoblotting, 30 µg of virions or DB per lane were loaded on a 10% bis-tris-polyacrylamide gel.
For silver staining, SDS-Gels were fixed and processed with the Roti®-Black P silver staining kit for proteins (Carl Roth, Karlsruhe, Germany). For instant blue staining, the SDS-Gel was incubated in 20 mL of the staining solution according to the manufacturer’s protocol (Expedeon via BIOZOL Diagnostica, Eching, Germany). For immunoblotting, proteins were transferred to a PVDF membrane (Immobilon-FL, Millipore, Billerica, MA, USA). Expression of the pentamer-complex proteins (UL128-131, gH, gL) on viral particles and DB was analyzed using a polyclonal PC-specific antibody raised in sheep (The Native Antigen Company, Kidlington, UK) using an anti-sheep HRP-coupled secondary antibody (The Native Antigen Company, Kidlington, UK).
For indirect immunofluorescence analysis, HFF or EC (HEC-LTT) (2 × 105
per well) were grown on coverslips in 6-well plates. The next day, HFF were incubated with 2 µg and EC with 10 µg of DB derived from strains Towne-BAC or Towne-repΔGFP, respectively. After 24 h, cells were handled as previously described [26
]. For detection of viral pp65, the specific mouse monoclonal antibody 65-33 (kindly provided by William Britt, University of Birmingham, Birmingham, AL, USA) was used. Nuclei were stained with 4′,6-Diamidin-2-phenylindol (DAPI) and analyses were performed with a Leica DM IRB microscope.
2.5. Analysis of Infectious Virus by IE1-Staining
The determination of residual infectivity within DB preparations was performed by staining with monoclonal antibody 63-27, directed against IE1 [46
], provided by William Britt. For this, 5 × 103
HFF per well were infected with tenfold serial dilutions of DB preparations (1 µg/mL). Forty-eight hours after infection, cells were washed in PBS and fixed with 96% ethanol for 20 min at room temperature. After a further washing step, cells were incubated for 1 h at 37 °C with 50 µL hybridoma supernatant of monoclonal antibody p63-27. Binding of the IE1-specific antibody was detected with a horse-radish peroxidase (HRP)-coupled polyclonal rabbit anti-mouse secondary antibody (Agilent, Waldbronn, Germany). Antibody binding was visualized by incubation with a 3-amino-9-ethylcarbazole (AEC) solution. After another washing step, the numbers of the IE1-positive nuclei were counted in the microscope. The mean of 8 technical replicates was taken as the relative measure of infectivity.
2.6. IFN-β Treatment and Analysis of Genome Replication Kinetics
Subconfluent HFF were treated with IFN-β (100 U/mL diluted in 0.1% bovine serum albumin–double-distilled H2O; specific activity, according to the manufacturer’s information, 5 × 108 U/mg; catalog number 300-02BC; PeproTech, Hamburg, Germany) or left untreated as a control. After 12 h of incubation, the cells were infected with 50 genome copies/cell of Towne-repΔGFP or Towne-repΔUL25. To measure intracellular and extracellular viral genomes, DNA from 1 × 105 infected cells or 200 µL of cell culture supernatants was isolated, respectively, using the High Pure viral nucleic acid kit (Roche, Mannheim, Gemany). The amount of genome copies was determined by HCMV-specific TaqMan PCR analysis using an ABI 7500 Fast real-time PCR detection system measuring triplicate technical replicates with the probe 5′-6-carboxyfluorescein-CCACTTTGCCGATGTAACGTTTCTTGCAT-tetramethyl-rhodamine (fwd primer: TCATCTACGG GGACACGGAC; rev primer: TCATCTACGGGGACACGG AC).
2.7. Statistical Analyses
Statistical analyses were performed using GraphPad Prism version 4.2.4. (GraphPad Software Inc., San Diego, CA, USA).
The medical need to develop an HCMV vaccine was identified many years ago [4
]. Several efforts have been made to establish a vaccine for both the prevention of cCMV and to attenuate the consequences of HCMV reactivation in immunosuppressed individuals since then (reviewed in [9
]). Testing of some of these candidate vaccines in clinical studies has met with limited success. This may be related to findings from recent studies from communities with high HCMV seroprevalence which have indicated that approaches mimicking natural immunity may not suffice to afford complete protection against cCMV [57
DB are a rewarding candidate to induce an immune response that is different from natural immunity and DB may thus meet the requirements for an effective vaccine. These particles are non-infectious. Consequently, they do not induce the manifold immune evasion mechanisms that are activated following natural HCMV infection or following the application of a live HCMV vaccine [60
]. They contain large amounts of the viral antigens that are considered to be important for the induction of both humoral and cellular immunity [27
]. In particular, they contain viral envelope proteins in their fusion-competent conformation which may be favorable for the induction of virus-specific protective humoral responses [67
]. The exceptional antigenic potential of DB has been shown by both our laboratory and others [29
] and may depend on their impact on dendritic cells, which are activated by DB exposure [29
] (Figure 2
The PC, consisting of gH/gL/UL128-131, is required for the infection of key target cells of HCMV, such as epithelial, endothelial, or dendritic cells [20
]. This protein complex has received considerable attention as a target of neutralizing antibodies during natural infection, as these antibodies may bear the potential to limit infection [48
]. We recently repaired the UL130 open reading frame in the laboratory strain Towne, enabling the reconstitution of the PC in that virus [35
]. DB of Towne-rep have regained the ability for PC-dependent cell entry (Figure 4
). Side-by-side immunization experiments have shown the superior potential of Towne-rep DB for the induction of neutralizing antibody responses [35
]. Since the Towne strain, as opposed to clinical isolates, is a high-level DB producer, its repaired derivative, deleted for the expression of GFP (Towne-repΔGFP, Figure 4
), is an attractive basis for a downstream production process of a DB-based vaccine. Remarkably, also in contrast to clinical isolates, the expression of the PC has proven to be stable during multiple passages in fibroblasts, thus enabling the establishment of a seed virus stock for vaccine production.
DB are produced on fibroblast cultures infected with a suitable seed virus strain. A production process for DB on MRC-5 fibroblasts has recently been established which includes UV irradiation to remove contaminating virus from the DB fractions prior to gradient ultracentrifugation (Figure 3
). This strategy safely eliminated infectious virus from the final DB product. As the removal of pathogenic virus from a DB vaccine is, however, a fundamental requirement, we designed several additional safety strategies for DB production to provide an unimpeachably safe product for application to humans.
The attenuation of the seed virus by deletion of UL25 is one of these strategies. Removal of UL25 did not affect the efficiency of DB synthesis [25
] (Figure 5
b). However, the respective virus was remarkably sensitive to IFN-β [25
] (Figure 5
e,f). Thus, the capacity of the UL25-negative viruses to replicate in vivo will be severely attenuated. A similar strategy that will be used in parallel is the conditional expression of an essential HCMV protein in the production process. Fusion of the destabilizing FKBP-domain to the UL51 open reading frame will enable the replication of the seed virus solely in the presence of Shield-1. As DB synthesis does not require pUL51, the particles can be synthesized after initial infection in the absence of Shield-1 (see Figure 7
). By contrast, viral DNA packaging and the synthesis of infectious virus will be blocked under these conditions. Thus, the expression of pUL51 under Shield-1-control will generate a safety vector for DB production that will be completely replication-incompetent in the absence of Shield-1. More work is required to evaluate if the yield of DB in the absence of Shield-1 will be sufficient for upscaling.
One additional strategy to enhance the safety of the final DB product is the addition of substances that inhibit viral replication in cell culture without hampering DB production. Inhibitors of the viral terminase complex have been identified as effective antiviral substances without grossly impairing DB synthesis [69
]. Letermovir is a terminase inhibitor that has recently been licensed for prophylaxis of HCMV reactivation in hematopoietic stem cell recipients [53
]. In this work we have shown that the application of letermovir reduced the viral contamination of purified DB by more than two orders of magnitude (Figure 6
c). Very low concentrations of the drug (50 nM) were required to suppress virus release. Applying such low concentrations in a production process will prevent contamination of a vaccine by the drug in significant amounts. Consequently, the application of letermovir for the production of a DB vaccine is an effective strategy to reduce virus contamination.