Development of a skin- and neuro-attenuated live vaccine for varicella

Varicella, caused by primary infection with varicella-zoster virus (VZV), is common in childhood and is associated with significant global morbidity and mortality. Although usually self-limiting, severe complications occur in both immunocompetent and immunocompromised individuals. Following primary infection, VZV establishes latency in sensory ganglia and can reactivate as herpes zoster (HZ), sometimes complicated by long-lasting post-herpetic neuralgia. Live-attenuated Oka (vOka) varicella vaccines have been widely adopted and are effective; however, vOka retains wild-type neurovirulence, establishes latency, and can reactivate, raising safety concerns including vaccine-strain HZ and neurological complications. vOka vaccines are heterogeneous mixtures of haplotypes with incompletely defined attenuation mechanisms, and some wild-type SNPs have been linked to vaccine-associated varicella and HZ. There is also concern about potential long-term effects of herpesvirus latency/reactivation on cognitive function. A safer next-generation varicella vaccine with reduced skin and neurotropism is therefore desirable. Advances in VZV molecular genetics have identified genes tied to virulence and tissue tropism. Notably, ORF7 is required for virulence in both human skin and neuronal cells but is non-essential for replication in many cultured cells, suggesting that an ORF7-deficient virus could be attenuated in target tissues while remaining immunogenic. This study reports the construction and preclinical evaluation of a pOka-derived, ORF7-deficient live vaccine candidate, v7D.

Historical use of the live-attenuated Oka strain (vOka) since the 1970s led to broad reductions in varicella burden and is recommended by WHO where disease burden is significant. Nonetheless, vOka remains neurovirulent, establishes latency, and can reactivate, with reported vaccine-associated rashes and rare neurological complications including meningitis. Deep sequencing indicates that vOka preparations are mixtures of haplotypes with attenuating and wild-type SNPs; less attenuated variants may contribute to post-vaccination disease. Although observational data suggest lower HZ incidence in vaccinated children than unvaccinated peers over the short term, long-term population impact remains uncertain. Molecular studies have identified multiple VZV genes affecting tropism (e.g., ORF47/ORF66 kinases, gI, ORF10). ORF7 uniquely is necessary for virulence in both human skin and neurons yet dispensable for replication in standard fibroblast cultures. Prior SCID-hu models demonstrated that wild-type and vOka establish lytic infection in DRG xenografts followed by reduced gene expression consistent with latent/abortive infection. Recent work identified the VZV latency-associated transcript (VLT) antisense to ORF61 and VLT-ORF63 fusions as putative latency markers.

The authors engineered an infectious BAC clone from a wild-type pOka genome (broka-GFP), then introduced a three-frame stop-codon cassette within 11 bp downstream of the ORF7 start codon to abrogate ORF7 expression (b7D-GFP). The loxP-flanked BAC vector/GFP cassette was excised via Cre recombinase in MRC-5 cells to reconstitute v7D; an ORF7-rescued virus (7R) and wild-type rOka were generated as controls. Whole-genome NGS (passages 12, 15, 16, 25) assessed genetic stability, showing only the intended ORF7 stop mutation in coding regions. In vitro characterization included growth kinetics at MOI 0.01 (MRC-5, differentiated SH-SY5Y neurons, primary HDFs; HEKs at MOI 0.2), plaque assays on MRC-5, and Western blots for viral structural proteins (capsid, tegument, glycoproteins) and cellular damage markers (cleaved PARP, cleaved caspase-3). Cell viability over 7 days was measured by CCK-8. PBMC infections (healthy donors, MOI 0.01) were analyzed at 3 dpi by flow cytometry for gE, immunofluorescence for pORF62/pORF7 localization, qPCR for genome copies (ORF31, ORF62) and RT-qPCR for transcripts. In vivo attenuation was evaluated in guinea pig and cotton rat neurotropism models via intramuscular inoculation along the spine with cell-associated virus (3×10^5 PFU), followed by DRG qPCR/RT-qPCR at 1 month. Human tissue tropism was assessed using SCID-hu mice bearing human fetal skin or DRG xenografts; xenografts were directly inoculated with cell-free virus (skin: 1×10^4 PFU, DRG: 1×10^3 PFU). Outcomes included lesion development, infectious virus recovery (plaque assay), IHC for gE, and DRG genome/transcript levels (ORF31, ORF62) at 14, 28, 56 dpi. Human dendritic cell (DC) immunogenicity assays used CD14+ monocyte-derived iDCs exposed to v7D or vOka (MOI 0.01). DC maturation markers (CD40, CD80, CD83, CD86), cytolysis (LDH), and cytokines/chemokines (e.g., TNF-α, IL-6, IP-10, MIP-1β, MCP-1, RANTES) were measured. Downstream T-cell responses were evaluated by co-culturing antigen-pulsed DCs with autologous CD4+ or CD8+ T cells, assessing proliferation (BrdU ELISA, CFSE dilution) and IFN-γ ELISPOT, plus cytokine/chemokine profiling. Small animal immunogenicity studies in mice, rats, guinea pigs, and rabbits used subcutaneous immunization at weeks 0, 3, 6 with v7D or vOka (species-specific PFU doses) and followed sera for 42 weeks for VZV gp-ELISA IgG titers, PRNT neutralization (NT50), and IgG subclass distribution; cellular responses were profiled by Luminex/ELISA after antigen stimulation (Th1/Th2/Th17 cytokines and chemokines). Nonhuman primate (NHP) safety studies included: (1) rhesus macaques receiving intrathalamic v7D (3×10^5 PFU) or saline, with clinical monitoring for 3 weeks, histopathology across brain regions, and IHC for VZV antigens; (2) cynomolgus macaques receiving repeated subcutaneous v7D (1×10^4 or 5×10^4 PFU) or saline at 3-week intervals (three doses), with comprehensive clinical pathology (weights, temperatures, ECG, hematology, biochemistry, coagulation), histopathology of major organs, qPCR for tissue VZV DNA, and immunogenicity readouts (anti-VZV IgG, PRNT, PBMC IFN-γ upon VZV antigen stimulation). Statistical analyses used one- or two-way ANOVA with Tukey’s post hoc test (P<0.05).

• Genetic engineering yielded a stable ORF7-deficient VZV (v7D) with no unintended coding mutations by NGS across multiple passages. • In vitro replication: v7D replicated comparably to rOka, 7R, and vOka in MRC-5 fibroblasts (MOI 0.01) and produced abundant viral structural proteins. In differentiated SH-SY5Y neurons, HDFs, and HEKs, v7D showed markedly impaired infectious progeny production and diminished viral protein translation, with minimal induction of cell damage markers and preserved cell viability relative to controls. • PBMC tropism: At 3 dpi (MOI 0.01), v7D-infected PBMCs displayed gE surface expression and viral genome copies/transcripts (ORF31, ORF62) comparable to rOka and vOka; pORF7 staining was absent only in v7D as expected. • Neurotropism in small mammals: In guinea pigs and cotton rats, VZV DNA/RNA was detectable in DRG from some rOka/vOka-inoculated animals but undetectable in all v7D-inoculated animals, suggesting impaired neuronal persistence. • SCID-hu human xenografts: In skin, rOka and 7R produced large lesions; vOka produced smaller lesions; v7D produced no lesions and yielded no recoverable infectious virus at 10 or 21 dpi. In DRG, infectious virus was recovered at 14 dpi from rOka (5/5), 7R (3/5), and vOka (3/5) xenografts, with high genome and transcript levels at 14 dpi declining thereafter; v7D produced no infectious virus and had undetectable genome/transcripts at all time points (14, 28, 56 dpi). • Human DC assays: Both v7D and vOka induced DC maturation (CD40, CD80, CD83, CD86), sustained secretion of pro-inflammatory cytokines/chemokines (e.g., TNF-α, IL-6, IP-10, MIP-1β, MCP-1, RANTES) without cytolysis, and drove comparable CD4+/CD8+ T-cell proliferation and IFN-γ ELISPOT responses. • Small animal immunogenicity: Across mice, rats, guinea pigs, and rabbits, v7D elicited anti-VZV IgG and neutralizing antibody kinetics comparable to vOka over 42 weeks; IgG subclass profiles indicated mixed Th1/Th2 polarization (Th2-dominant in mice; IgG2a/IgG2b prominent in rats). Virus-specific cellular responses (Th1/Th2/Th17 cytokines and chemokines) were similarly upregulated in v7D and vOka groups and sustained to week 42. • NHP neurovirulence: Intrathalamic v7D (3×10^5 PFU) caused no fever, weight loss, neurological signs, or spread of infection; histology showed only needle-track astrogliosis/microgliosis with no VZV gE or pORF62 antigen staining in brain regions. • NHP repeated-dose toxicity and immunogenicity: Subcutaneous v7D at 1×10^4 or 5×10^4 PFU (three doses) was well-tolerated with no significant differences across clinical, hematologic, and biochemical parameters versus controls; histopathology showed no systemic toxicity. VZV DNA was detectable only at or near injection sites. Immunogenicity was robust: peak anti-VZV IgG GMTs were 6,400 (low dose) and 25,600 (high dose); neutralization NT50 titers were 128 in both groups; PBMC IFN-γ responses increased and were sustained through day 84.

The study addressed the need for a safer varicella vaccine by rationally ablating ORF7, a determinant of VZV virulence in human skin and neurons while dispensable for fibroblast replication. v7D preserved replication competence in MRC-5 cells and full lymphotropism in human PBMCs and DCs, supporting efficient antigen presentation and adaptive immune priming, yet was profoundly attenuated in human skin and neuronal models in vitro and in vivo (SCID-hu xenografts). This dual attenuation reduces risks associated with vOka, including vaccine-strain rashes and neurotropic complications or reactivation. Comparable immunogenicity to vOka across multiple small animal species and in nonhuman primates indicates that attenuation did not compromise the capacity to induce humoral and cellular responses. NHP studies demonstrated a favorable safety profile, with no neurovirulence upon direct intrathalamic inoculation and no systemic toxicity after repeated subcutaneous dosing, while maintaining strong antibody and IFN-γ responses. Mechanistically, loss of ORF7 likely impairs VZV cell-to-cell spread and syncytia-mediated dissemination crucial for skin and neuronal infection, while leaving lymphoid replication largely intact. These findings support v7D as a promising candidate to mitigate vaccine-related adverse events and inform herpesvirus vaccine design targeting tissue-tropism genes.

An ORF7-deficient, genetically defined VZV vaccine candidate (v7D) was constructed and shown to be highly stable, severely attenuated in human skin and neuronal tissues, and non-neurovirulent, while retaining replication in fibroblasts and lymphotropism necessary for robust immune priming. v7D elicited humoral and cellular immune responses comparable to vOka in vitro, in multiple small animal models, and in nonhuman primates, with an excellent preclinical safety profile. These data support advancement into human clinical trials and suggest potential clinical benefits over current vOka vaccines, including reduced varicella-like rashes and lower risk of vaccine-strain HZ. Future research should investigate the molecular interactions of ORF7 in VZV pathogenesis, assess latency markers (e.g., VLT and ORF63 transcripts) in latency models, and conduct long-term clinical follow-up to evaluate durability of protection, potential reactivation, and genetic stability.

VZV is highly human-specific; therefore, available animal models and in vitro systems cannot fully recapitulate human infection or latency/reactivation dynamics, limiting direct efficacy inference. Whether v7D can establish neuronal latency remains unresolved; latency-associated transcripts (e.g., VLT and VLT-ORF63 fusions) were not assessed here. Long-term safety in humans, including risks of reversion, recombination, latency, and reactivation, will require years to decades of observation. Some preclinical findings (e.g., small animal immunogenicity) may not translate quantitatively to humans.