Codon pair bias deoptimization (CPBD) has enabled highly efficient and rapid attenuation of RNA viruses. The technique relies on recoding of viral genes by increasing the number of codon pairs that are statistically underrepresented in protein coding genes of the viral host without changing the amino acid sequence of the encoded proteins. Utilization of naturally underrepresented codon pairs reduces protein production of recoded genes and directly causes virus attenuation. As a result, the mutant virus is antigenically identical with the parental virus, but virulence is reduced or absent. Our goal was to determine if a virus with a large double-stranded DNA genome, highly oncogenic Marek's disease virus (MDV), can be attenuated by CPBD. We recoded UL30 that encodes the catalytic subunit of the viral DNA polymerase to minimize (deoptimization), maximize (optimization), or preserve (randomization) the level of overrepresented codon pairs of the MDV host, the chicken. A fully codon pair-deoptimized UL30 mutant could not be recovered in cell culture. The sequence of UL30 was divided into three segments of equal length and we generated a series of mutants with different segments of the UL30 recoded. The codon pair-deoptimized genes, in which two segments of UL30 had been recoded, showed reduced rates of protein production. In cultured cells, the corresponding viruses formed smaller plaques and grew to lower titers compared with parental virus. In contrast, codon pair-optimized and -randomized viruses replicated in vitro with kinetics that were similar to those of the parental virus. Animals that were infected with the partially codon pair-deoptimized virus showed delayed progression of disease and lower mortality rates than codon pair-optimized and parental viruses. These results demonstrate that CPBD of a herpesvirus gene causes attenuation of the recoded virus and that CPBD may be an applicable strategy for attenuation of other large DNA viruses.
