Bacterial artificial chromosome derived simian varicella virus is pathogenic in vivo

Background Varicella zoster virus (VZV) is a neurotropic alphaherpesvirus that infects humans and results in chickenpox and herpes zoster. A number of VZV genes remain functionally uncharacterized and since VZV is an obligate human pathogen, rigorous evaluation of VZV mutants in vivo remains challenging. Simian varicella virus (SVV) is homologous to VZV and SVV infection of rhesus macaques (RM) closely mimics VZV infection of humans. Recently the SVV genome was cloned as a bacterial artificial chromosome (BAC) and BAC-derived SVV displayed similar replication kinetics as wild-type (WT) SVV in vitro. Methods RMs were infected with BAC-derived SVV or WT SVV at 4x105 PFU intrabronchially (N=8, 4 per group, sex and age matched). We collected whole blood (PBMC) and bronchoalveolar lavage (BAL) at various days post-infection (dpi) and sensory ganglia during latent infection (>84 dpi) at necropsy and compared disease progression, viral replication, immune response and the establishment of latency. Results Viral replication kinetics and magnitude in bronchoalveolar lavage cells and whole blood as well as rash severity and duration were similar in RMs infected with SVV BAC or WT SVV. Moreover, SVV-specific B and T cell responses were comparable between BAC and WT-infected animals. Lastly, we measured viral DNA in sensory ganglia from both cohorts of infected RMs during latent infection. Conclusions SVV BAC is as pathogenic and immunogenic as WT SVV in vivo. Thus, the SVV BAC genetic system combined with the rhesus macaque animal model can further our understanding of viral ORFs important for VZV pathogenesis and the development of second-generation vaccines.


Introduction
Varicella zoster virus (VZV) is a neurotropic alphaherpesvirus and the etiological agent of varicella (chickenpox) and herpes zoster (shingles). VZV establishes latency within the sensory ganglia, and reactivation from latency can cause significant morbidity and occasionally mortality in older and immunocompromised individuals. Currently the FDA vaccine Zostavax® reduces the incidence of shingles by 51% and the burden of disease by approximately 61% [1,2]. Thus, a significant portion of vaccine recipients still remains susceptible to VZV reactivation. To improve vaccine efficacy, we need to determine the function of the viral open reading frames (ORFs) that contribute to VZV pathogenesis and those that are important for the host immune response.
Simian varicella virus (SVV) is a homolog of VZV that causes varicella-like disease and establishes latency in sensory ganglia of rhesus macaques [3][4][5]. SVV shares significant DNA homology and genome colinearity with VZV [6][7][8][9]. VZV and SVV have the smallest genomes of the herpesvirus family. VZV encodes at least 70 unique ORFs and SVV encodes 69 distinct ORFs [8,10]. Despite the smaller genome size and homology to herpes simplex virus (HSV), a number of VZV/SVV genes remain functionally uncharacterized. Studies characterizing viral gene function utilizing in vitro tissue culture models do not always model the complex host-pathogen relationship that occurs in vivo. Recently, with the construction of an infectious SVV bacterial artificial chromosome (BAC), the production of mutations and deletions in specific SVV ORFs will allow the investigation of gene function during in vivo infection [11]. Previously, SVV BAC was shown to generate infectious virus with molecular properties and in vitro replication kinetics comparable to wild-type (WT) SVV [11]. SVV BAC was also used to generate an ORF10 deletion virus, which demonstrated that SVV ORF10 is nonessential for replication in vitro [11]. In the current study we further investigate SVV BAC in vivo by monitoring replication kinetics, immune response and establishment of latency in rhesus macaques and show that SVV BAC is as pathogenic as WT SVV.

Whole-genome analysis of SVV BAC
The BAC derived SVV viral genome was comprehensively analyzed by comparative genomic hybridization (CGH) and directly compared to wild-type (WT) SVV. Using this technique, any differences in genomic sequence between SVV BAC and WT SVV results in variations in hybridization intensities to corresponding segments represented on the array, giving an altered hybridization ratio between SVV BAC and WT SVV ( Figure 1A). CGH analysis revealed that two areas displayed variations when compared to WT SVV, indicating differences in nucleotide sequence at these locations. These regions were amplified via PCR and directly sequenced resulting in the identification of two nucleotide substitutions that produced 1 missense mutation and 1 silent mutation within the coding region of the SVV BAC genome. Specifically, we identified a point mutation at nucleotide 41990 from G to A within ORF22, producing an amino acid change from valine to isoleucine ( Figure 1B) and a transition of nucleotide 106546 from T to C producing a silent mutation within ORF62/71 ( Figure 1C). The nucleotide change in ORF62/71 was also previously shown in the sequencing of an ORF61 deletion virus that was generated from the same parental SVV cosmid system [11][12][13]. SVV ORF22 is a putative tegument protein based on the function of herpes simplex virus type-1 (HSV-1) UL36 homolog. The missense mutation in ORF22 did not render SVV BAC derived virus noninfectious or hamper replication kinetics and plaque size in vitro [11].

Disease severity and viral load
Rhesus macaques (RMs) were infected with SVV BAC or WT SVV at 4×10 5 PFU intrabronchially (n=4 per group, sex and age matched). We investigated the pathogenesis of BAC derived SVV in vivo by measuring disease progression, viral replication, immune response, and the establishment of latency compared to WT SVV. We collected bronchoalveolar lavage (BAL) cells and blood (peripheral blood mononuclear cells, PBMC) at various days post-infection (dpi) and sensory ganglia were collected at necropsy (84-86 dpi). All infected RMs displayed hallmarks of SVV infection including the development of rash, which lasted between 7 and 10 days. A representative RM infected with SVV BAC at 7 Figure 1 Comparative genomic hybridization and sequence analysis comparing SVV BAC to WT SVV. A) Schematic representation of the SVV genome highlighting the SVV ORFs (arrows) that contain sequence changes. Sequence variation results in different hybridization intensities indicated by the hybridization ratio between SVV BAC and WT SVV and signal potential nucleotide changes. B and C) The regions containing sequence variations were amplified by PCR and directly sequenced. Sequencing identified: B) within ORF 22 a transition occurs from G to A resulting in a missense mutation, C) within ORF 62/71 a transition from T to C results in a silent mutation (note the nucleotide and position number refers to the genomic position). Nucleotide substitution (bold italics). dpi is shown in Figure 2A and a representative RM infected with WT SVV at 7 dpi is shown in Figure 2B. A lesion area was biopsied at 10 dpi and viral loads were measured by quantitative real-time PCR ( Figure 2C). By 10 dpi, we were able to detect viral DNA in all RMs except RMs 28553 and 28621 infected with WT SVV. SVV viral loads were also measured by quantitative real-time PCR in BAL cells and whole blood samples. BAL cell viral loads peaked at 3 dpi in both SVV BAC and WT SVV infected RMs then decreased to levels near or below our limit of detection by 63 dpi ( Figure 2C). SVV viral loads in whole blood are significantly lower than in BAL cells, though we were able to detect SVV DNA in whole blood between 3 and 14 dpi in RMs infected with SVV BAC or WT SVV and then viral loads decreased to levels near or below our limit of detection ( Figure 2D). Therefore, the ability of SVV BAC to replicate in vivo was comparable to WT SVV.
Concentrations of pro-inflammatory cytokines IFNγ (involved in anti-viral activities and differentiation of T helper 1 (Th1) subsets [21]), TNFα (systemic inflammation [22]), IL-6 (pro-and anti-inflammatory responses [23]), IL-2 (T cell proliferation and homeostasis [24]) and IL-15 (proliferation of T cells and NK cells [25]) peaked at 7 dpi. The concentrations of IL-1 receptor antagonist (RA), which prevents IL-1 from signaling through the IL-1R [26] and IL-10 (pleiotropic activities in inflammation and immune regulation [27]) peaked at 7 dpi. G-CSF, a pleiotropic cytokine, produced by endothelium, macrophages and other immune cells, which stimulates the survival, proliferation, differentiation, and function of neutrophils also peaked at 7 dpi [28]. Levels of IL-12, important in the differentiation of naïve T cells into CD4 Th1 cells [29], peaked at 10 dpi. Lastly, the concentration of IFNα, a type I interferon that is important in anti-viral immunity, peaked at 3 dpi. Growth factors EGF (stimulates cell growth, proliferation and differentiation [30]), and FGF-basic (multifunctional protein involved in angiogenesis and wound healing [31]) concentrations in BAL fluid peaked at 7 dpi. While levels of growth factors HGF (regulates cell growth, cell motility, and morphogenesis, and acts primarily upon epithelial and endothelial cells [32]) and VEGF (stimulates vasculogenesis and angiogenesis [33]) peaked at 10 dpi.
Concentrations of IL-1β, IL-4, IL-5, IL-8, IL-17, GM-CSF and RANTES (CCL5) and in BAL fluid were below our limit of detection (data not shown). We did not detect any significant differences in concentrations of chemokines, cytokines or growth factors between RMs infected with SVV BAC or WT SVV. We also measured the concentration of the above chemokines, cytokines, and growth factors in plasma (data not shown). However, we were only able to detect changes in the levels of IFNγ, which peaked at 7 dpi and returned to baseline by 10 dpi and no significant differences were detected between cohorts.

B cell and antibody response to SVV BAC
We compared the magnitude and kinetics of the B cell response as well as the generation of SVVspecific IgG antibody titers post-infection in RMs infected with SVV BAC or WT SVV. The expansion of B cells is measured based on expression of Ki67, a nuclear protein involved in DNA replication [34] by flow cytometry. SVV infection induces the proliferation of B cells indicated by an increase in the frequency of Ki67 positive cells on days 7 to 14 compared to day 0 in marginal zone (MZ)-like (CD27 + IgD + ) and memory (CD27 + IgD − ) B cells, in BAL cells ( Figure 4A) and PBMC ( Figure 4B). In BAL cells and PBMC we measured similar proliferation of both subsets of B cells and no statistical differences in RMs infected with either SVV BAC or WT SVV except for at 10 dpi in BAL cells of WT SVV infected RMs we detected higher proliferation of MZ-like B cells (p<0.05) compared to SVV BAC. We also measured SVV-specific IgG ( Figure 4C) antibody endpoint titers in plasma using standard ELISA. The kinetics of IgG production were comparable during SVV BAC and WT SVV infection of RMs, the titers peaked around day 14 post-infection and remained stable until necropsy. BAC or WT SVV infected RMs, SVV-specific CD4 and CD8 T cells were detected 7 dpi, their frequency peaked between 14 and 21 dpi and declined to a memory set point. There were no statistically significant differences between animals infected with SVV BAC or WT SVV, and both cohorts did not produce a measurable CD4 or CD8 response in PBMCs following stimulation with overlapping viral peptide pools (data not shown).

SVV viral load in sensory ganglia
SVV DNA viral loads in sensory ganglia were measured by quantitative PCR ( Table 1). The viral loads reported   in Table 1 reflect SVV genome copy numbers in a portion of the ganglia and are therefore not representative of the entire organ. Four of the RMs went on to further studies therefore Table 1 shows the latent viral loads for two animals from each cohort. We detected SVV DNA within at least one sensory ganglia of each RM infected with SVV BAC and WT SVV indicating that the SVV BAC, like WT SVV is able to traffic to the sensory ganglia, the site of SVV latency.

Discussion
Cloning viral genomes as bacterial artificial chromosomes (BAC) is an efficient tool to manipulate the viral genome facilitating the study of viral genes in vitro and in vivo. BACs were constructed for the VZV parental Oka virus and the vaccine Oka virus [35][36][37][38]. All VZV ORFs have been deleted using BACs or cosmids and tested in cell culture providing valuable information on which VZV genes are essential for viral replication in vitro (reviewed in [39,40]). Subsequent in vivo analysis of VZV mutant viruses using the SCID-hu mouse model showed that VZV ORFs 1, 2, and 3 are dispensable for viral replication [38], whereas VZV ORF23, which was dispensable for replication in culture, was found to be required for replication in human skin xenografts [41]. Similarly, VZV deleted for ORF7 replicated in MeWo cells but in vivo ORF7 was shown to be important for neuroinvasion [40,42]. Lastly, mutating the furin recognition site of VZV ORF31 (gB), an essential viral protein, did not affect VZV replication in vitro but attenuated viral replication in vivo [43]. Studying VZV mutant viruses in the SCID-hu mouse model has provided valuable insight into VZV biology, further evaluation of the role of different VZV ORFs in pathogenesis and the immune response in vivo is hampered by the fact that VZV is an obligate human pathogen. SVV is a simian homolog of VZV that causes varicella and zoster in nonhuman primates [4][5][6].
Recently, the SVV genome was cloned as a BAC and SVV virus generated using the SVV BAC genetic system was found to be similar to WT SVV in vitro [11]. Mutagenesis of SVV ORF10 showed that SVV ORF10 is nonessential for replication in vitro [11]. Additionally, generation of a SVV ORF 63/70 mutant demonstrated impaired growth in Vero cells [44]. In the present study we further the analysis of SVV BAC by infecting rhesus macaques and investigating the pathogenesis of SVV BAC compared to WT SVV in vivo. The combination of the rhesus macaque animal model and the SVV BAC will provide a robust tool to examine viral ORFs important for pathogenesis and help to target VZV ORFs that will improve vaccine efficacy.
To compare the genome of the SVV BAC to WT SVV we utilized comparative genomic hybridization. CGH analysis was employed as a cost-effective and accurate strategy to analyze genomic DNA from multiple viruses. This technique is sensitive enough to detect single base changes in addition to insertions, deletion or rearrangements in the genome [45][46][47]. We sequenced two sites in the SVV BAC genome that displayed variations in hybridization when compared to WT SVV. In ORF22 a point mutation at nucleotide 41990 was found producing an amino acid change from valine to isoleucine. SVV and VZV ORF22 are putative tegument proteins based on homology to HSV-1 UL36 [48]. UL36 is a HSV-1 late gene and the HSV-1 virion contains 100-150 copies of UL36 [49,50]. UL36 is essential to HSV-1 replication and the phenotype of a null mutant virus showed accumulation of capsids containing cytosolic DNA that did not mature into enveloped virions [51][52][53]. Though, the evidence suggests that the amino acid change in SVV BAC does not constitute a significant change in the protein. In vitro SVV BAC displays similar plaque size and replication kinetics in Vero cell monolayers as compared to WT SVV [11]. Our data in vivo shows that SVV BAC displays similar replication kinetics, immune response and establishment of latency compared to WT SVV. Potentially the position of the amino acid change or that the change is a nonpolar side chain to a nonpolar side change allows for WT behavior.
The replication kinetics of SVV BAC in the bronchoalveolar lavage cells (the site of virus inoculation) and in the peripheral blood was statistically similar to WT SVV. Related, the spread of varicella rash was also similar between cohorts and lasted between 7 and 10 days post-infection. In Figure 2A and B we show pictures of two RMs infected with SVV BAC or WT SVV, which showed representative rash and also illustrates the variation in rash spread we see in our animals. In two animals infected with WT SVV, we did not detect viral DNA in the skin lesions but due to the nature of the skin punch biopsies as well as the timing of the rash, which varies between animals usually between 7 to 10 days post-infection, we may not have obtained a lesion spot with detectable viral DNA.
We also followed the immune response to SVV BAC during acute infection and found SVV BAC to elicit a parallel immune response in vivo. We analyzed the proliferation kinetics of antigen experienced B cells, the production of SVV-specific IgG antibodies as well as the proliferation kinetics of memory T cells and the IFNγ/ TNFα response of SVV-specific CD4 and CD8 T cells, and each parameter was analogous. We did measure a statistical difference in MZ-like B cells at 10 dpi in the BAL where RMs infected with WT SVV displayed a higher peak percentage of Ki67 positive cells. However, this difference did not translate to a higher antibody titer in the WT SVV infected RMs. This difference could be due to the small sample size (n=4) and the outbred nature of rhesus macaques.
Infection with SVV BAC also resulted in a comparable upregulation of chemokines, cytokines, and growth factors during the early stages of acute infection in the lungs compared to RMs infected with WT SVV. Peak levels of IFNα, an important antiviral cytokine, occurred at 3 dpi, which corresponds to peak viral loads in the lungs also at 3 dpi. Type I interferons are early immune effectors and are important cytokines during initial infection to limit viral replication and spread, including herpesviruses [54][55][56]. The concentrations of several T cell-recruiting chemokines (MCP-1, MIG, I-TAC) peaked 7 days prior to the observed peak in proliferating T cells in BAL samples. We also detected increased concentrations of TNFα (3 dpi) and IFNγ (7 dpi), which is indicative of Th1 immune responses and correlates with an increase in CD4 CM in BAL samples. Peak concentrations of IL-2 were detected 7 days prior (7 dpi) to the peak proliferation of CD4 and CD8 T cells (14 dpi). Levels of IL-12, which play a role in enhancing cytotoxic function of CD8 T cells, peaked 4 days before peak proliferation of CD8 T cells in BAL samples. Interestingly, we also detected an increase in growth factors in the BAL fluid, which peaked from 7 to 10 days postinfection. Many of these growth factors are involved in wound healing and might represent a response to tissue injury induced by SVV replication within the lungs (site of inoculation).
Lastly we found that both SVV BAC and WT SVV established latency within the sensory ganglia. Viral DNA was detected in at least one ganglion from each RM measured by quantitative PCR. In summary, SVV BAC is as pathogenic in vivo as WT SVV. Future studies will utilize the SVV BAC genetic system to generate knockout viruses to help characterize the role of SVV genes in acute infection, the establishment and maintenance of latency, and reactivation in vivo, a critical step in the understanding of viral factors that impact VZV pathogenesis and the immune response to VZV.

Cells and viruses
Bacterial artificial chromosome (BAC)-derived SVV was generated from self-excisable pSVV-BAC resulting in complete excision of plasmid sequences from the virus genome [11]. Wild-type (WT) simian varicella virus (SVV, Cercopithecine herpesvirus 9) and SVV BAC were propagated as previously described, briefly Vero cells maintained in Eagle's minimal essential medium (EMEM) supplemented with 5% newborn bovine serum, penicillin and streptomycin, WT SVV and SVV BAC infected Veros were harvested by scraping and frozen in Vero media supplemented with 10% dimethyl sulfoxide (DMSO) [11]. Virus stocks were titered by plaque assay on primary rhesus fibroblasts maintained in Dulbecco's modified Eagle's Medium (DMEM) supplemented with 10% fetal bovine serum and penicillin, streptomycin and L-glutamine. WT SVV cell lysate was obtained by scraping infected primary rhesus fibroblasts at the height of CPE followed by centrifugation and sonication using 7 pulses of 70-80 Watts (Sonicator 3000, Misonix Inc., Farmingdale NY) and frozen at −80°C.

Animals and sample collection
All rhesus macaques were housed at the Oregon National Primate Research Center and were handled in accordance with good animal practices as defined by the Office of Laboratory Animal Welfare. Animal work was approved by the Oregon National Primate Research Center Institutional Animal Care and Use Committee. Rhesus macaques (RM, Macaca mulatta) were SVV seronegative prior to infection measured by ELISA. RMs (n=4 per group) were infected intrabronchially with 4×10 5 PFU WT SVV or SVV BAC infected Veros. Peripheral blood mononuclear cells (PBMC) and bronchoalveolar lavage (BAL) cells were collected from rhesus macaques as previously described [5]. Animals were euthanized at 84 to 86 days post-infection. Sensory ganglia: trigeminal ganglia (TG), cervical, thoracic and lumbarsacral dorsal root ganglia (DRG-C, DRG-T, and DRG-L/S respectively) were divided, flash frozen and stored at −80°C until analysis.

DNA extraction and quantitative real-time PCR (qPCR)
DNA was extracted from heparinized whole blood (WB), BAL cells, and portions of frozen ganglia using Archive Pure DNA Cell/Tissue Kit (5 PRIME, Gaithersburg MD) according to the manufacturer's protocol. SVV DNA viral loads in WB, BAL cells and sensory ganglia were measured by qPCR using Maxima Probe/ROX qPCR Master Mix (2X) (Fermentas, Glen Burnie MD) and primers/Taqman probe specific for SVV ORF21. Following an initial 10 minute 95°C step, 40 cycles of 15 sec at 95°C and 1 minute at 60°C were completed using StepOnePlus (Life Technologies, Carlsbad CA). SVV BAC DNA was used as quantification standards [11].

Cytokine analysis
Plasma and BAL supernatant samples (stored at −80°C) were thawed and analyzed using Cytokine Monkey Magnetic 28-plex panel as per the manufacturer's instructions (Life Technologies). IFNα levels were measured using Cynomolgus/Rhesus IFNα Serum ELISA Kit according to the manufacturer's instructions (PBL Interferon Source, Piscataway NJ). Samples were run in duplicate. Values below the limit of detection were designated as ND, or not detected.

Enzyme-linked immunosorbent assay (ELISA)
ELISA plates were coated with SVV lysate overnight at 4°C, blocked with 5% milk in wash buffer (0.05% Tween in PBS) for 1 h at room-temperature (RT), washed three times with wash buffer, and incubated with heatinactivated (55°C, 30 min) plasma samples in 3-fold dilutions in duplicate for 1 h. After washing three times with wash buffer, horseradish peroxidase (HRP)-conjugated anti-rhesus IgG (Nordic Immunology, Netherlands) was added for 1 h, followed by addition of chromagen ophenylenediamine•2HCl (OPD) (Sigma, St Louis MO) substrate for 20 minutes to allow detection and quantitation of bound antibody molecules. The reaction was stopped with the addition of 1 M HCl. The optical density was measured at 490 nm using an ELISA plate reader (SpectraMax 190, Molecular Devices, Sunnyvale CA). Endpoint IgG titers were calculated using log-log transformation of the linear portion of the curve with 0.1 optical density (OD) units as the cut-off. Titers were standardized using a positive control sample included with each assay.

Intracellular cytokine staining
BAL cells and PBMC were stimulated with SVV lysate (1 μg) or SVV overlapping peptide pool containing open reading frames (ORFs) 4, 31, 61 and 63 for 1 h followed by addition of Brefeldin A (Sigma, St Louis MO) to block cytokine export for 14 h. After stimulation cells were surface stained with antibodies against CD4, and CD8β, as described above. Samples were fixed, permeabilized (BioLegend) and dual-stained using antibodies against IFNγ (eBioscience) and TNFα (eBioscience). Samples were analyzed using the LSRII instrument and FlowJo software.

Statistical analysis
Statistical analysis and graphing was conducted with GraphPad Prism software (GraphPad Software Inc., La Jolla CA). Significance values for Figures 2, 3, 4, 5, and 6 utilized repeated measures of ANOVA with the Bonferroni post-test to explore differences between groups (SVV BAC and WT SVV) at each time-point.