Synthetic Attenuated Virus Engineering (SAVE)
A Novel Strategy to Generate Viral Vaccine Candidates
Advances in synthetic biology pave the way for the creation of new vaccine candidates for a potentially wide range of viruses. The viruses are attenuated by means of recoding the sequences of portions of the genome such that they are codon pair deoptimized but still encode the wildtype amino acid sequence. The recoded sequences that contain hundreds of nucleotide changes compared to the wildtype, are generated by chemical synthesis. Thus far, results are very promising for poliovirus and influenza virus.
Vaccinations containing live attenuated viruses have provided relief from numerous fatal diseases in the past, and continue to protect countless people today. In contrast to inactivated vaccines, live attenuated vaccines have the benefit of a more efficient cellular and humoral immune response with a low, sometimes single dose. Thus far, however, live attenuated vaccines have had two major pitfalls. Firstly, attenuated phenotypes have been produced by either randomly introducing and testing mutations in the viral genomes or by serial passage of the viruses under suboptimal conditions, such as non-natural host cells or low temperatures. These methods often rely on chance and cannot be universally applied to a wide variety of virus types. Additionally, the extent of viral attenuation, an important consideration for the development of an efficient vaccine, cannot be further manipulated by this method because the molecular mechanism of the attenuation is usually not known. Secondly, the viruses are generally attenuated on the basis of only a few mutations and have a considerable chance of reverting to wildtype thereby causing the very disease in vaccine recipients that they are aimed to prevent.
Advances in synthetic biology are allowing researchers to overcome these barriers in live attenuated vaccine development. It is now possible to order large segments of synthetic DNA, assemble them into entire genomes of infectious agents, and boot them to life - as was first published by Eckard Wimmer and colleagues with poliovirus in 2002 , and has recently been applied to Mycoplasmamycoides in Craig Venter's lab . The ability to create viable viruses by chemical synthesis based solely on sequence design allows us to rapidly modify large segments of the viral genome without altering the amino acid sequence, leading to controlled and deliberate attenuation of the virus.
Our approach to synthetic attenuated virus engineering (SAVE) makes use of codon pair bias, a phenomenon first discovered in prokaryotic cells in 1989  but has since been observed in all other studied species, including humans . Codon pair bias refers to the preferential pairing of certain codons over others, a phenomenon that is independent of codon bias, in which some codons are used more frequently than others. For instance, in human genes the codon pair GCC-GAA encodes the adjacent amino acids alanine-glutamic acid significantly less often than the pair GCA-GAG, despite GCC and GAA being the most frequently used alanine and glutamic acid codons, respectively .
Our technique to generate a recoded (codon pair deoptimized) viral genome involves using a computer algorithm to generate a viral genome containing a high number of codon pairs that were found least frequently in the intended host species‘ genome, namely codon pair deoptimization, while strictly retaining all codons and all wildtype amino acid sequences. The nucleotide sequences are then chemically synthesized and assembled into a viral genome that is booted to life in cell culture. Polioviruses containing codon pair deoptimized regions exhibited decreased translational efficiency and an attenuated phenotype in cell culture and animals .
Viruses that are able to produce proteins with wild type amino acid sequences but at low efficiency due to codon pair deoptimization may not be able to replicate sufficiently enough to overcome host defenses. Humans infected with such viruses, therefore, may be able to mount an immune response robust enough to induce lasting protective immunity, paving the way for synthetically attenuated viruses to be excellent vaccine candidates. Indeed, transgenic mice infected with codon pair deoptimized polioviruses did not show signs of disease but were protected during a challenge with a lethal dose of wild type poliovirus , a vital characteristic of an effective live vaccine.
With such promising results from poliovirus, SAVE was applied to influenza virus, a virus that claims hundreds of thousands of lives yearly, despite the widespread availability of seasonal flu vaccines. Various portions of the influenza genome were codon pair deoptimized individually and in combination and were characterized in cell culture and in mouse models. The synthetic viruses exhibited only a slight (tenfold) attenuation in cell culture as compared to wildtype. However, in animal experiments, a virus containing three deoptimized segments exhibited a four-log higher lethal dose than wildtype and a low protective dose . This combination of traits in a virus is very conducive for vaccine production as it offers a wide margin of safety, that is, a small dose of the vaccine is sufficient to stimulate humoral and cellular immunity with no signs of disease, yet it protects the animals against a thousand-fold challenge with the wildtype virus.
Furthermore, the ability to deoptimize varying combinations of the influenza genome segments, with countless different sequences, paves a new path for production of live attenuated influenza vaccinations. It should be noted that we have introduced hundreds of unfavorable codon pairs in the polio and flu genomes to achieve attenuation; therefore reversion to wildtype, or even to more virulent versions of the vaccine is unlikely. Current seasonal vaccinations with live vaccine, such as "FluMist", rely on an unchanged given backbone every year, alternating only two of the eight proteins encoded. If vaccinated multiple times (over multiple "seasons") recipients may mount an adaptive immune response to the proteins encoded by the backbone genes, which, in turn, could interfere with the efficacy of the cold-adapted vaccine. SAVE provides the opportunity to attenuate the entire seasonal genome and elicit a precise immune response with little risk of acquiring flu-like symptoms.
Given the elegance of the computer algorithm used to design the codon pair deoptimized viruses, SAVE can easily be applied to a wide range of viruses in various species. The two viruses studied thus far are of opposite genomic polarity and both have yielded convincing results. Although any potential vaccine candidates would have to undergo careful scrutiny, this method is extremely promising since it provides a faster and cheaper way to arrive at a live attenuated vaccine candidates than traditional methods. Additionally, SAVE involves a deliberate and methodical design such that viral attenuation can be modulated, providing the opportunity to create viruses that are too weak to cause illness but viable enough to replicate sufficiently and stimulate a robust immune response. Finally, SAVE's biggest advantage over traditional methods to produce live vaccines is that the sequence of viral genome has been altered so drastically by chemical synthesis  (without changing any protein sequence) that there is little risk for reverting to wildtype and causing disease. Application of this method to vaccine development would overcome the major pitfalls associated with the current techniques used to create vaccines.
We thank our colleagues Steve Skiena, Bruce Futcher, Aniko Paul, Dimitris Papamichail, Charles B. Ward, J. Robert Coleman, Anjaruwee Nimnual, and Chen Yang for their contributions to the work summarized here. Supported in part by National Institutes of Health grants AI075219 and AI15122 (EW) and a TRO-FUSION Award by Stony Brook University (SM and Steve Skiena).
 Cello J. et al.: Science 297, 1016-1018 (2002)
 Gibson D.G. et al.: Science. 329(5987), 52-56 (2010)
 Gutman G. A. and Hatfield G. W.: Proc. Natl. Acad. Sci. U.S.A. 86, 3699-3703 (1989)
 Moura G.R. et al.: PLoS ONE.2(9):e847 (2007)
 Coleman J. R. et al.: Science. 320,1784-1787 (2008)
 Mueller S. et al.: Nature Biotechnology 28(7), 723-727 (2010)
 Wimmer E. et al.: Nature Biotechnology 27(12), 1163-1172 (2009)