A Virulent Bioluminescent and Fluorescent Dual-Reporter Marek's Disease Virus Unveils an Alternative Spreading Pathway in Addition to Cell-to-Cell Contact

Abstract

Marek's disease virus (MDV) is a growing threat for the poultry industry. Unfortunately, despite successful vaccination against the disease, MDV remains in circulation within vaccinated flocks, leading to the selection of increasingly virulent pathotypes. Detailed knowledge of the virus biology and the host-virus interaction is required to improve the vaccine efficiency. In the present study, I engineered an original, dual-reporter MDV to track and quantify virus replication in vitro and in vivo.

TEXT

Marek's disease virus (MDV) is a fatal avian oncogenic alphaherpesvirus that causes acute T-cell lymphomas in the viscera and other organs (1). The extensive use of vaccination has been successful in controlling Marek's disease (2). Unfortunately, vaccination is inefficient against wild-type virus (re)infection, and MDV remains in circulation within vaccinated flocks, fueling the emergence of increasingly virulent pathotypes (3). To achieve sustainable MDV vaccination, a better understanding of MDV biology and MDV vaccine protection mechanisms is urgently needed (4). Several molecular aspects of MDV biology, including replication, dissemination, latency, reactivation, morphogenesis, and virus-induced transformation, remain to be clarified, in contrast with the available knowledge regarding other alphaherpesviruses.

MDV is highly transmissible in vivo, whereas the in vitro production of MDV is in the range of 3 orders of magnitude lower than for other alphaherpesvirus, and this represents a barrier for MDV research (5–11). Indeed, with the most permissive cell culture system for MDV, total MDV production in vitro during 4 days of infection is, at best, 10⁵ PFU/10⁶ cells (6, 12). This suggests that an adequately permissive and physiopathologically relevant MDV in vitro production system remains to be identified. This lack of an appropriate cell model could also impair the transposition of results from other alphaherpesviruses. For example, the inability to obtain a null mutant of a particular MDV gene, in contrast to null mutants of other alphaherpesviruses, may be due to the suboptimal MDV in vitro production system rather than a genuine vital function for the gene in MDV growth (5–9).

According to the current paradigm, MDV belongs to a unique class of viruses that rely exclusively on cell-to-cell transmission (12–15). In contrast, it has been reported that infectious, cell-free MDVs are produced in feather follicle epithelial cells, and these may directly participate in viral transmission in chickens (16, 17). This is reminiscent of human T lymphotropic virus 1 (HTLV-1), which was once assumed to rely exclusively on cell-to-cell transmission, until it was produced as cell-free infectious virus in fully permissive dendritic cells (18). These observations highlight the need for the development of a relevant tool to optimize investigation of MDV and address some of the main issues associated with this model.

In the present study, I reasoned that engineering a recombinant, fully virulent, bioluminescent and fluorescent MDV reporter virus could provide a valuable replication biosensor to monitor the spread of MDV infection both in cultured cells and throughout the host organism.

Construction of a one-step shuttle plasmid to rapidly insert genes into the MDV genome.

In previous studies, fluorescently tagged MDVs proved very useful in cell sorting of MDV-infected cells and in documenting differential tissue expression of tagged tegument protein (12, 16). However, a bioluminescent MDV reporter virus expressing the highly sensitive luciferase enzyme has not yet been reported (19). In addition, tagging of viral proteins, either at the C or N terminus, is likely to be detrimental for virus replication (12, 16). To overcome this issue, I used the cis-acting hydrolase element (CHYSEL) (20–23). This strategy is based on the insertion of the 2A ribosomal skipping site, which provides an efficient way to express comparable levels and physically separate proteins from a single promoter.

I chose the strong major tegument VP22 promoter, so that processing of the foreign gene(s)-VP22 protein could be achieved by the in-frame incorporation of the porcine teschovirus 1 2A CHYSEL upstream of the UL49/VP22 coding sequence (Fig. 1A).

FIG 1.

Generation of recombinant dual-reporter MDV constructs. The Pst-SpeI fragment derived from the plasmid p(UL48-50 eGFPUL49) (12) was subcloned in order to generate unique BglII and StuI sites, respectively, upstream and downstream of the eGFP open reading frame. The coding sequence of eGFP was replaced with the corresponding fragment from either PCR-amplified StuI-eGFP-BamHI-2A-BglII or StuI-RnLuc-BglII cassettes lacking the 2A sequence. Then, the kanamycin selection cassette with the required recombination homology arms was inserted at the StuI site, leading to the shuttle plasmids p(eGFP2A-VP22) and p(RnLuc-VP22), respectively. To obtain the p(eGFP-GLuc2A-VP22) plasmid, I inserted, in frame at the AccIII and BamHI sites of the p(eGFP2A-VP22) plasmid, the corresponding sequence of naturally secreted GLuc. In order to facilitate future use of the shuttle plasmid p(eGFP-GLuc2A-VP22), this plasmid was modified in order to disrupt the StuI site downstream of the KanR cassette as follows: the plasmid was digested with NheI, treated with the T4 DNA polymerase, and after heat inactivation the plasmid was digested with PstI and ligated to a StuI-PstI fragment prepared from the same plasmid. Thus, I obtained the final one-step shuttle plasmid p(eGFP-GLuc2A-VP22). With this one-step transfer plasmid, it is possible to rapidly clone any given gene in frame with the 2A UL49 sequence by using the StuI or AccIII sites at one end and the BamHI site or a compatible site at the other end. The p(eGFP-RnLuc2A-VP22) plasmid was obtained using the AccIII and BamHI sites to replace the sequences of the Gaussia luciferase with the sequence of RnLuc. All the plasmid constructs were sequenced to confirm their identities. All the oligonucleotides used in this work are available upon request.

The recovery of highly replicative reporter viruses requires the presence of the 2A CHYSEL.

I generated three distinct recombinant viruses that contained the 2A sequence upstream of VP22 (eGFP2A-VP22, eGFP-GLuc2A-VP22, and eGFP-RnLuc2A-VP22; eGFP is enhanced green fluorescent protein, GLuc is Gaussia princeps luciferase, and RnLuc is Renilla luciferase). All these viruses showed growth characteristics similar to the parental RB-1B virus (Fig. 2). In contrast, the construct RnLuc-VP22, which lacks the 2A sequence between the luciferase and the VP22 sequences, led to a virus that was dramatically impaired in its spreading ability, consistent with previous work showing that tagging VP22 either at its C or N terminus is detrimental for MDV replication (12, 16). Recombinant viruses eGFP-RnLuc2A-VP22 and eGFP-GLuc2A-VP22 and the parental RB-1B virus were propagated during three passages on fresh chicken embryonic skin cells (CESC) in culture before establishing the virus stocks, which were made of frozen infected cells. The virus stocks were titrated twice and in parallel on fresh CESC. The three viruses showed a similar average viral titer of 10⁴ PFU/10⁶ frozen infected cells (Fig. 2A). In order to confirm that the 2A element introduced as a linker between the reporter protein and the viral VP22 protein allowed autonomous intraribosomal self-processing of the polyprotein, I performed an immunofluorescence colocalization assay of VP22 and the eGFP-GLuc2A fluorescent protein (Fig. 2B). This representative picture illustrates the robust expression of both VP22 and eGFP-GLuc2A and demonstrates that the 2A processed proteins do not colocalize. Also, Western blot analysis (Fig. 2C) confirmed the 2A self-cleavage activity. In addition, the bioluminescence assay of the dual-reporter viruses demonstrated that the time-saving and cost-effective luciferase assay is a simple quantitative assay for monitoring MDV growth properties, in comparison to the widely used plaque size assay (Fig. 2A and D) (12, 16). Thus, in cultured cells, the level of luciferase obtained after 4 days of MDV infection was shown to be proportional to the PFU titer of the input viral inoculum (Fig. 2D). Importantly and in accordance with previous work (24), activity of the naturally secreted Gaussia luciferase was easily monitored directly in cell-free supernatant without any further treatment, whereas the activity of the Renilla luciferase was mainly found in cells and in cell debris following cell lysis treatment (Fig. 2D).

FIG 2.

(A) In vitro growth property assessment of the recovered recombinant viruses. CESC were obtained from specific-pathogen-free embryos by following standard protocols as previously described (10). Recombinant viruses were recovered by transfecting the corresponding bacterial artificial chromosomal DNA into CESC cultures by using the Amaxa nucleofector kit and protocol. Recombinant viruses were propagated during three passages on fresh CESC cultures before establishing the corresponding viral stock. The viral stocks were made of several vials of CESC-infected cells frozen in liquid nitrogen (10% dimethyl sulfoxide, 90% fetal calf serum). For virus stock titrations, 10⁶ CESC in six-well plates were infected with 3-fold serial dilutions, starting with 10,000 infected cells for each virus stock, and fixed 4 days postinfection with 4% paraformaldehyde (PFA) at 4°C for 30 min. Fixed cells were stained with a mixture of three mouse monoclonal antibodies (F19 anti-VP5, K11 anti-gB, and E21 anti-ICP4) to visualize viral plaques (a). To minimize the observed variation (about 2-fold) in MDV viral titers from the same virus stock, I repeated the stock titration twice, in triplicate, and took the average titers. (b) The total numbers of separated plaques were counted and are reported as PFU/10⁶ infected cell viral stock. (c) To determine virus growth properties of the recovered recombinant viruses, the standard MDV techniques for plaque size assays were used (12, 16). Briefly, after viral stock titration, images of at least 40 plaques were taken with a charge-coupled-device (CCD) Axiocam MRm camera mounted on an inverted epifluorescence Axiovert 200 M microscope (Zeiss). The plaque areas were manually determined by using the Axiovision (Zeiss) and Image J software (NIH), and means were determined for each virus. (B) Immunofluorescence analysis of 2A cis-acting hydrolase activity. CESC were infected for 4 days with MDV (eGFP-GLucA2-VP22). Cells were fixed and stained as described earlier. Specific VP22 monoclonal antibody was used to compare the immunofluorescence localization of VP22 (red) with the localization of eGFP-GLuc2A fluorescent protein (green). The image shows the noncolocalization of the eGFP-GLuc2A fusion protein and the viral VP22 protein, indicating that the insertion of a ribosomal skipping site between eGFP-GLuc and VP22 allows this key viral protein to be released from the upstream eGFP-GLuc2A fusion protein. (C) Western blot analysis of 2A cis-acting hydrolase activity. Protein extracts from control CESC cultures or cultures infected with the indicated recombinant virus were separated by SDS-PAGE. Monoclonal anti-GFP (Clontech-TaKaRa) was used as the primary antibody. Lanes 1, 2, and 5 show cell lysate infected with recombinant MDV expressing eGFP-RnLuc2A-Vp22, eGFP-GLuc2A-Vp22, or eGFP2A-Vp22, respectively. Lane 3 is cell lysate from mock-infected cells. Lane 4 shows lysate of CESC infected with rMDV RB-1B parental virus. Lane 6 indicates the respective positions of protein molecular mass marker (in kDa); arrows in the left indicate the theoretical molecular mass (in kDa) of the expressed eGFP fusion proteins. (D) Correlation between the plaque titration results and the bioluminescence assay for quantitative monitoring of MDV dual-reporter virus replication. CESC in 12-well plates were infected with 3-fold serial dilutions of the previously titrated MDV viral stock. (a) At 4 days postinfection, a bioluminescence assay was performed directly in cell culture plates to evaluate the expression of the cell-associated Renilla luciferase. Briefly after the elimination of cell culture supernatant and the addition of the luciferase substrate, the total photon count and the numbers and the sizes of visualized plaque in each well were determined with the CCD camera for each serial dilution. The image shows bioluminescent plaques corresponding to cells infected with the recombinant MDV (eGFP-RnLuc2A-VP22). (b) The luciferase activity in cell-free supernatant (SN) or in cell lysate was also measured by using a plate luminometer (GloMax-Multi⁺ microplate reader; Promega). The luminometer was set to automatically inject 50 μl of ready-to-use coelenterazine (Renilla-Glo luciferase assay; Promega), and photon counts were acquired for 10 s. The total luciferase count produced 4 days postinfection from 10⁶ cells was plotted against the indicated PFU viral input. (c) Comparative percentage localization of Gaussia and Renilla luciferase activities in the specified compartment of cells infected with MDV (eGFP-GLuc2A-VP22) or MDV (eGFP-RnLuc2A-VP22).

The dual-reporter MDV is as virulent as the parental MDV virus in vivo.

In order to reduce the number of animals for the set of experiments assessing the MDV in vivo virulence, I examined only one of the two dual-reporter viruses (eGFP-Gluc2A-VP22), in parallel with the parental RB-1B virus. Injected birds were housed for 10 weeks with 10 contact naive chicks in glove box isolators. The results showed that both the RB-1B parental virus and the dual-reporter virus were 100% efficient in inducing full-pattern Marek's disease in inoculated chickens (Fig. 3A). The reporter virus was found to be transmitted to contact animals in a timeframe similar to that for the parental rRB-1B virus. Importantly, fluorescence microscopy allowed us to visualize, and count, actively infected tumor cells. I confirmed a previous report (25) that showed that, in the absence of reactivation treatment, the ratio of productively infected cells in tumors is low (1 to 3%). I counted about 2% of eGFP-positive cells in cultured tumor cells, including possible reactivation as a result of cell handling (Fig. 3B). Also, and before the onset of any Marek's disease symptoms, I directly detected at day 7 and at day 12 postinoculation the luciferase activity in 20-μl blood samples from MDV-eGFP-Gluc2A-inoculated animals (Fig. 3C, graph a). In MDV-eGFP-Gluc2A-inoculated cohoused animals, I detected the GLuc activity only at day 12. However, compared to the level observed in cell culture, the GLuc blood level was low. This was probably the result of inhibition of the GLuc activity in the presence of blood. Indeed, I found that GLuc activity was quenched in the presence of blood from noninfected birds (Fig. 3C, graph b). Taken together, my results support the previously reported short window of the MDV active cytolytic phase (26). The use of whole-animal in vivo bioluminescence imaging during this critical phase will allow tracking of natural MDV dynamics in MDV-infected contact animals.

FIG 3.

(A) In vivo assessment of MDV dual-reporter virus (eGFP-GLuc2A-VP22) virulence. One-week-old, specific-pathogen-free (SPF) B13/B13 white leghorn chickens were obtained from the INRA-PFIE production unit and were maintained under biosafety level 2 (BSL-2) conditions in the experimental unit. The chickens were divided into two groups of 20 chicks each; group 1 received rRB-1B parental virus, and group 2 recei ed rRB-1B (eGFP-GLuc2A-VP22) dual-reporter virus. Ten chickens of each group were injected intramuscularly in the left pectoral region with 0.2 ml of a cell suspension containing 2,000 PFU of each virus stock. The injected birds were housed with 10 contact naive chicks in glove box isolators. The chickens were evaluated twice a day for clinical symptoms of Marek's disease. Birds presenting MD clinical signs were euthanized and examined for evidence of MD. The percentage of Marek's disease-free chickens was recorded during 10 weeks in both inoculated and contact animals. (B) eGFP expression in a cell tumor removed from a necropsied MDV (eGFP-GLuc2A-VP22)-infected chicken. Tumors from visceral organs were collected at necropsy and cultured using standard protocols (15). Pictures in the upper panel show photographs of the induced solid tumors, especially in the kidneys of both parental- and eGFP-GLuc2A-VP22-infected chickens. In the lower panel, the images of the second passage of nonadherent cells were taken in the indicated channel with a charge-coupled-device Axiocam MRm camera mounted on an inverted epifluorescence Axiovert 200 M microscope (Zeiss). eGFP fluorescents cells were counted to determine the percentage of actively infected cells. (C) Monitoring of GLuc activity in blood of MDV-infected animals. (a) Blood samples of 0.2 ml were taken from infected animals and from contact animals at 0, 4, 7, 12, 25, 35, and 63 days postinoculation in order to monitor the secreted Gaussia luciferase activity in the blood samples. GLuc activity was measured in 20-μl blood samples (in duplicates) by using a plate luminometer (GloMax-Multi+ microplate reader; Promega). The luminometer was set to automatically inject 50 μl of ready-to-use coelenterazine (Renilla-Glo luciferase assay; Promega), and photon counts were acquired for 10 s. All the samples used for the Gluc assay contained 20% (vol/vol) citrate solution. (b) Compared to the level observed in cell culture, the GLuc level in blood was low. I examined if GLuc activity was inhibited in the presence of blood. Using the same assay as above, I assayed in triplicate a predefined value of GLuc activity (5-μl aliquots of conditioned medium from eGFP-GLuc2A-VP22-infected CESC) either in the presence of blood from a noninfected chicken or in the presence of the same volume ratio of PBS. In the presence of blood from noninfected birds at a volume ratio of 1/5 or 1/2, the GLuc activity was shown to be inhibited compared to the GLuc activity in the presence of PBS.

The dual-reporter virus revealed an alternative spreading pathway for MDV.

Interestingly, by using live-cell fluorescence microscopy I was able to observe in both early and heavily infected cells a striking cell-blebbing phenomenon (27, 28) (Fig. 4A). Moreover, I also observed engulfment of fluorescent cell blebs by neighboring cells. It was then tempting to hypothesize that the blebs could be a potential vehicle for MDV infection. As a first argument, I showed that low-speed centrifugation concentration of cell-free conditioned medium from MDV-infected cells was able to transmit MDV efficiently to naive cells. In order to ascertain that the MDV transmission was not the result of contamination with intact infected cells, I further subjected the inoculum to both fluorescence microscopy (Fig. 4B) and transmission electron microscopy (Fig. 4C). Convincingly, when the initial viral inoculum was normalized to the Renilla luciferase count as a direct measure of viral production, the infectivity associated with MDV-infected cells was similar to the infectivity associated with cell-free supernatant (Fig. 5). In addition, electron microscopy of the concentrated cell-free conditioned medium confirmed the presence of MDV particles and apoptotic bodies (Fig. 4C). Taken together, these results represent the first evidence that MDV does not rely exclusively on cell-to-cell contact for spreading. Indeed, the blebbing phenomenon was very reminiscent of the apoptotic mimicry “Trojan horse” scenario, where the correlation between virus-induced cell blebbing and the cellular apoptotic corps clearance mechanism has been shown to play a major role in virus dissemination (28–30).

FIG 4.

(A) Fluorescence microscopy assessment of MDV cytopathic activity. Cytopathic activity of MDV was assessed in MDV (eGFP2A-VP22)-infected cells by live-cell fluorescence microscopy. (a) The merged image of green (eGFP) and blue (staining of nuclei) shows morphological changes and formation of apoptotic bodies that contained nuclear or cytoplasmic material. (b) Magnified details of an MDV-infected cell with typical morphological criteria of type I cell death or apoptosis, defined by chromatin condensation and fragmentation, cell blebbing of the plasma membrane, and cell shrinkage. (B) Fluorescence microscopy visualization of concentrated cell-free supernatant. Low-speed centrifugation (at 1,400 × g of relative centrifugal force [RCF]) of cell culture supernatant from day 4 MDV (eGFP-RnLuc2A-VP22)-infected cells was performed in order to collect cell fragments and apoptotic bodies. After addition of 4′,6-diamidino-2-phenylindole for chromatin staining, the pellet was examined by fluorescence microscopy in the indicated channel to confirm the presence of cell fragments and apoptotic bodies. (C) Electron microscopy analysis of concentrated cell-free supernatant. The pellet of low-speed centrifugation (RCF, 1,400 × g) of 4-day MDV (eGFP-RnLuc2A-VP22) cell culture supernatant was fixed for 24 h in 4% paraformaldehyde and 1% glutaraldehyde. After centrifugation to remove the fixative agent, ultrathin sections were cut, placed on electron microscopy grids, and stained with 5% uranyl acetate–5% lead citrate. The negatively stained preparations were then examined for the presence of all virion assembly stages, from naked capsids and empty particles to enveloped virions. Transmission electron micrographs confirmed the presence of MDV particles (with and without an electron-dense capsid) in apoptotic bodies (a) and within cell debris and cytoplasmic organelles (b).

FIG 5.

Comparison between the infectivity associated with MDV-infected cells (Cell MDV) and the infectivity associated with cell-free supernatant (SN MDV). Four-day MDV (eGFP-RnLuc2A-VP22)-infected cells or corresponding pellet of cell fragments and apoptotic bodies, prepared as indicated for Fig. 4, were used with four serial 3-fold dilutions (1, 1/3, 1/9, and 1/27) as viral inoculum to infect fresh CESC. After 75 min of contact with the cells, the initial viral inoculum was washed away and viral production was assessed 4 days later via a luciferase count. The two different inocula transmitted corresponding MDV infection in a dose-dependent manner. Interestingly, when the initial viral input was normalized and set to 10⁶ RLU, the MDV infectivity associated with infected cells and the infectivity associated with cell-free supernatant produced, 4 days later, resulted in similar values for the luciferase count.

Moreover, the induction of apoptosis has been shown to enhance both the reactivation and replication of the prototypical alphaherpesvirus, herpes simplex virus 1, and also more recently beta- and gammaherpesviruses, which implies that the viral subversion of the cell apoptotic defense process is a common feature of herpesvirus biology (31, 32). For MDV, it was demonstrated previously that immunosuppressive treatments of chickens with cyclosporine or betamethasone resulted in widespread MDV infection in target organs and enhanced reactivation of the virus (26). As those drugs have been shown to be proapoptotic, further investigation of the direct role of apoptosis and apoptotic mimicry in MDV replication and spreading is needed (28–30).

Ethics statement.

Animal studies were performed according to protocol number AH/2012/01-A-MDV Bio-Tracer, approved by the Val de Loire ethics committee CEEA Vdl. Agreement numbers A37-190 and C37-175-3 were delivered by the French authority, respectively, to the Principal Investigator of this study and to the head of the INRA-PFIE experimental unit. Animal studies adhered strictly to E.U. Animal Welfare directive 2010/63/EU, which has been transposed to French law, under decree number 2013-118.

Conclusions.

In the present work, I confirmed that 2A CHYSEL activity (20–23) is efficient in chicken cells and provides a useful tool to allow the coexpression of multiple proteins from a single promoter. This result highlights the potential of MDV as versatile avian multigene delivery vector. Importantly, this work opens the way for in vivo bioluminescence and fluorescence studies in order to track MDV dynamics during the early hours of virus infection and the ensuing spread in vaccinated versus naive chickens. Such comparative studies will allow identification of the cells involved in virus uptake, despite vaccination, and could also present an opportunity to identify novel MDV-permissive cells that could be used to develop an improved MDV in vitro production system. Ultimately, this knowledge will assist in the design of an improved MDV vaccine, with the aim to tackle the main issue of the inefficient protection that allows uptake and shedding of virulent MDV in vaccinated animals.