Jonathan S. Aeschliman
BIOL 509 – Molecular Biology and Applications
November 22, 2010
Steffen Mueller, J Robert Coleman, Dimitris Papamichail, Charles B Ward, Anjaruwee Nimnual,
Bruce Futcher, Steven Skiena & Eckard Wimmer. Live attenuated influenza virus vaccines by
computer-aided rational design. Nature Biotechnology 28, 723 - 726 (2010). Published
online: 13 June 2010 | doi:10.1038/nbt.1636
Objective of the Paper
The purpose of this research was to develop and test a new method of attenuation of influenza
viruses. Mueller et al. utilize a strategy known as SAVE (synthetic attenuated virus engineering) which
takes advantage of the codon-pair bias in organisms.
Codon-pair bias
Codon-pair bias is not the same as codon bias. This is best explained in terms of an amino acid
pair, for example, Arg-Glu. There are six (6) codons which code for arginine and two (2) for glutamic
acid; together they represent twelve (12) possible combinations of codons (to form the same two amino
acids). Knowing the frequency by which each of these codons appear (codon bias), one can determine
the expected number of codon-pairs. One can then compare the expected number of instances of this
codon-pair to the actual number found in the organism. Researchers can then use the known codon-pair
bias to design organisms which are deoptimized for that organism.
By developing synthetic viruses which have deoptimized codons, the growth and virulence of
the influenza virus is significantly reduced as compared to its wild type form. More importantly, the
mutated form can provide protection against the wild type influenza virus in future exposures. Overall,
this research shows that influenza viruses, which are attenuated through the SAVE method, have a
strong potential for safe, reliable and effective vaccines.
Experiments and Results
The wild type organism chosen for attenuation was influenza virus A/PR/8/34 (PR8). Using
computer software, researchers were able to redesign several different gene sections of its genome.
There were three different regions which were chosen for deoptimization: polymerase subunit B1 (PB1),
nucleoprotein (NP) and hemagluttanin (HA). Through an eight-plasmid system, researchers were able to
insert the deoptimized forms of each gene segment into host cells. A single gene region could be
inserted as well as combinations of the three. NPMin/ HAMin/PB1Min is a strain which contains all three
deoptimized gene segments, named PR83F.
The growth patterns of each different mutated group were then characterized by growth in
MDCK cells (in vitro). The plaques generated by each individual strain were not significantly different
than the wild type strain and the virus titer levels were ten-fold lower. Western blot analysis followed
and it showed that protein synthesis of the specific gene sequences was reduced in the deoptimized
form of that gene as compared to wild type viruses. It was also shown that the WT gene segments in the
attenuated viruses were unaffected.
In addition to these in vitro studies, mouse models were used to test the attenuated viruses and
compare them to the wild type. The BALB/c mice were infected intranasally with 104 PFU (plaque
forming units) of either the WT PR8 or the PR83F (triple-mutant) viral strain. Those mice infected with
the WT experienced severe symptoms and extreme weight loss, while those infected with PR83F showed
no symptoms and no weight loss. The replicative potential of these two strains was then compared by
testing the viral load within each mouse’s lungs. Those mice infected with the WT PR8 strain showed a
3000 times higher viral load as compared to the PR83F strain. Additionally, the WT PR8 mice had died by
Day 6, while the PR83F mice had an undetectable viral load after 9 days.
In order to design an effective vaccine, the dosage must be such that it is sufficient to provide
against future infections, but not so great that it causes lethality from vaccination. Safety studies were
then done to establish both the LD50 and PD50 for different strains. PD50 is the dosage which is required
to provide immunity to future infections in half the animals. The PD50 of the PR8 (WT) virus was 1 PFU,
while its LD50 was 61 PFU; this provides a safety margin of about 60 (LD50/PD50). The PD50 and LD50 of the
PR83F virus was 13 PFU and 790,000 PFU; giving a safety margin of about 60,000.
Protection studies were then completed to assess the potential for future vaccines. Mice were
infected intranasally with 104 PFU of PR83F and then after 28 days were challenged with 1000 times the
LD50 of WT PR8. After three more days, viral titers were measured in the mice lungs. In 80% of the mice,
the levels of WT PR8 were undetectable while mock protected mice had titers ~107 PFU.
Additional tests were done to determine the level of antibodies produced in mice models. The
levels of serum IgG anti-influenza antibodies were then measured after infection with 0.01 X LD50 for
both PR8 and PR83F. The titer levels for PR83F were 312,500, while PR8 (WT) had a level of 27,540. Even
at lower concentrations (0.001 X LD50), the titer level for PR83F was still very high (230,000).
Conclusion
The results from this series of experiments shows that attenuated influenza viruses, generated
by the SAVE method, have great potential for future vaccines. The SAVE method is interesting because it
does not rely on traditional methods of attenuation. Instead of attempting to modify/delete virulence
genes, it relies on the organism’s own codon-pair bias constraints. Since this form of attenuation takes
the form of many mutations, it is unlikely that natural mutation would cause a reversion into a more
virulent form. Additionally, this method generates strains which retain the immunogenicity of the
original WT virus.
Through other experiments, this method has also been used to generate attenuated forms of
poliovirus. Since the mechanisms for replication in influenza and poliovirus are very different, it shows
that this method has the possibility to be used in a wide range of organisms. However, the SAVE method
still requires a significant amount of work before it can be implemented for human use.
In these specific experiments, it has been shown that the triple-deoptimized PR8 virus (PR83F)
can function as an appropriate vaccine against future infections with WT PR8 in mouse models. The use
of this Influenza A strain with the SAVE method shows that attenuated viruses can be reliably generated
and that this method could possibly be extended to other organisms.