The Evolution & Medicine Review

Influenza A viruses continue to be of enormous interest to biomedical researchers and clinicians alike.  In addition to the annual influenza epidemics, which have been inferred to cause substantial excess mortality, there is the ever-present threat of a global pandemic due to several features of influenza virus biology.  A high mutation rate associated with a segmented negative-sense RNA genome that facilitates recombination confers on these orthomyxoviruses a prodigious ability to evolve in ways that confound and evade the human immune system.  The ability of influenza viruses to infect domesticated mammals, such as pigs, as well as birds, both wild and domesticated, provides additional opportunities for the virus to try out new genetic combinations and to disseminate around the globe by means both dependent and independent of human travel.

A recent paper by Jagger et al. (Science 2012) reveals that influenza A viruses are also, not surprisingly, capable of exploiting rather subtle and ingenious genetic “tricks” to maximize the value of every base pair in a total of a mere 13.5 kilobases of genome.  These authors have discovered a new influenza gene product, PA-X, that represents a fusion protein incorporating 191 amino-terminal amino acids of the well-known RNA-dependent RNA polymerase (PA) protein and another carboxy-terminal 61 amino acids of a protein from a reading frame shifted by one nucleotide downstream.  They present evidence from multiple experiments suggesting that the frameshift is related to a highly conserved codon that is rarely employed in mammalian and avian genomes and therefore interacts with a cognate tRNA of relatively low concentration in the cytosol.  The longer-than-average wait for this tRNA to be recruited to the ribosome makes a frameshift mutation more probable.

This polypeptide chain arising from two reading frames has an N-terminal endonuclease activity that, based on experiments with reporter genes, appears to degrade host messenger RNAs.  Although the PA protein (without the 61 amino acids encoded by the +1 reading frame) also has the N-terminal endonuclease domain, it did not mediate the same inhibition of plasmid-mediated gene expression.

It would be quite reasonable to suppose that this activity qualifies PA-X as an influenza A virus virulence factor.  In support of such a notion, portions of the nucleotide sequence encoding the PA-X fusion are highly conserved among influenza A viruses.  Furthermore, PA-X is dispensable for virus replication, qualifying it as an accessory protein.

Thus, it comes as an especially intriguing result to find out what happens when PA-X is deleted from the 1918 influenza A virus strain that caused the most devastating known worldwide influenza pandemic.  Loss of PA-X expression increases pathogenicity in a mouse model of influenza infection as assessed by weight loss, survival, and aspects of the immune response.  The peak differential in weight loss between virus strains expressing and not expressing PA-X is at day 5 after infection, which is roughly coincident with the appearance of virus-specific cytotoxic T lymphocytes.  With respect to survival post-infection, there were no obvious differences among mice infected with PA-X-expressing and PA-X-non-expressing strains at infectious inocula of 10 or 1,000 plaque-forming units (pfu), but there were highly statistically significant differences in survival when the initial infection was performed with 100 pfu of virus.

In other words, it appears that PA-X reduces virulence, at least as defined by the endpoints assessed by these authors.  PA-X appears to do so in part by altering the kinetics and magnitude of host gene expression and resulting immune responses.  It remains to be seen what effect PA-X has on transmissibility.  Presumably, the genetic material encoding PA-X is not maintained in the influenza A virus genome because it reduces viral fitness.

These results suggest that substantial caution is advised in drawing conclusions about the net effect or “function” of a given gene product on the overall infectious process.  The preceding is true even if one puts aside, as seems likely, that the net effects of a pathogen-encoded gene product could vary with host genotype, within or across species (Yewdell and Ince, 2012).  One corollary is that determining what gene products should be regarded as virulence factors may depend on how one defines “virulence” and “virulence factor.”  A second corollary is that virulence can require rather precise modulation and optimization in order to maximize fitness.  Therefore, asking whether deletion of the gene encoding a putative virulence factor decreases virulence may be too crude a criterion for identifying the gene product as a contributor to virulence.


Jagger BW, Wise HM, Kash JC, Walters KA, Wills NM, Xiao YL, Dunfee RL, Schwartzman LM, Ozinsky A, Bell GL, Dalton RM, Lo A, Efstathiou S, Atkins JF, Firth AE, Taubenberger JK, Digard P. An overlapping protein-coding region in influenza A virus segment 3 modulates the host response. Science. 2012 Jul 13;337(6091):199-204. Epub 2012 Jun 28. PubMed PMID: 22745253.

Yewdell JW, Ince WL. Virology: frameshifting to PA-X influenza. Science. 2012 Jul 13;337(6091):164-165.  PubMed PMID: 22798590.