I’ve talked about quasispecies several times, and emphasized that RNA viruses, with their high replication error rates, are most prone to forming quasispecies.

I’ve also pointed out, though, that actually measuring quasispecies is technically difficult, and measuring it for the larger DNA viruses would be even harder. You’d need to run sequences on many viral genomes, to see how much variation develops over time; and it’s only recently that sequencing tech has approached the point where it’s even thinkable, let alone affordable, to do that:

While large DNA viruses are thought to have low mutation rates, only a small fraction of their genomes have been analyzed at the single-nucleotide level.1

So maybe DNA viruses might actually form quasispecies, and we don’t know it?

In fact, even for DNA viruses, mutants appear fairly quickly, given the appropriate selection pressure. In principle, these might not even be new mutations, but simply expansion of a particular part of a quasispecies that was already pre-existing. (It’s probably a fairly obvious point, but it’s important to remember that the introduction of new mutants, and their selection and expansion, are completely different processes. Some viruses throw out incredible numbers of mutants, but almost all of them are dead ends that are actively selected against, or at the least not selected for. Other viruses may make far fewer mutants, but given strong enough selection pressure some of these might rapidly take over the population. It’s very tricky to use observed mutations as a measure of mutation frequency, because observation often depends on selection to build up the numbers of the mutant before you can see it.)

At any rate, it’s a fair enough question,2 but recently there has been some evidence that supports the concept of DNA virus genome stability. Wayne Yokoyama’s lab has actually sequenced multiple genomes of mouse cytomegalovirus (a large DNA virus — a member of the herpesvirus family) to look at quasispecies.1

One of the things they did find was a significant number of variations in their stock, compared to the stock they had got it from years before:

… our laboratory’s Smith strain MCMV differed from the previously published Smith strain … There were 452 differences, including 50 insertion/deletions (indels) and 402 single-bp substitutions. … this high number of differences suggested that MCMV mutated in vivo, as we had previously maintained our MCMV stock by in vivo passages. 1

In other words, the standard methods of maintaining a virus — repeatedly growing new stocks in cells, and using those new stocks to make yet more — allow the accumulation of variations in the genome — which is already well known, of course, but often neglected in lab experiments.  Again, as I point out above, the number of observed mutations we see here doesn’t tell us much about the actual mutation frequency.

How often do mutations arise? By running the virus through cells repeatedly (in vitro, that is) and then seeing how individual clones differed, they determined that there are very, very few mutations per replication. What’s more, and even more impressively, very few mutations appeared after passages through mice (in vivo):

… the remaining 9 mutations allowed us to estimate the mutation rate of MCMV as 1.0 x 10–7 mutations per bp per day after in vivo passage, very similar to the mutation rate calculated for in vitro passage.1

(My emphasis) For comparison, the MCMV genome is not quite 250,000 bp long, so we’re looking at around one mutation per 40-50 genomes per day (if I’m dividing right). That’s hundreds of times more stable than most RNA viruses (see the table here3 for some RNA virus error rates).  Still, there’s plenty of room for natural selection in there, because of course there are hundreds or thousands of new MCMV genomes being made per day even in the most conservative estimate (and maybe more like millions or hundreds of millions), so dozens to thousands of them are mutants.

So, not surprisingly, Yokoyama’s group was able to detect a cluster of mutations that were almost certainly selected in the mouse; without going into detail, these mutations were in a viral gene (m157) that’s known to be recognized by the (laboratory) mouse immune system, so it wasn’t surprising that mutants were selected. And such mutants did not appear in mice without the appropriate immune component, demonstrating the role of natural selection in this cluster.

They offer a number of cautions, including one that’s raised in almost all such studies:

One caveat to our mutation analysis is that lethal mutations were probably underrepresented in the final DNA pool since, by definition, they did not propagate. Nonetheless, this limitation is intrinsic to all mutation analysis.1

Still, the results are solid and reassuring, supporting a basic concept in viral evolution.


  1. Cheng, T., Valentine, M., Gao, J., Pingel, J., & Yokoyama, W. (2009). Stability of Murine Cytomegalovirus Genome after In Vitro and In Vivo Passage Journal of Virology, 84 (5), 2623-2628 DOI: 10.1128/JVI.02142-09[][][][][]
  2. Well asked, invisible non-existent person![]
  3. CASTRO, C., ARNOLD, J., & CAMERON, C. (2005). Incorporation fidelity of the viral RNA-dependent RNA polymerase: a kinetic, thermodynamic and structural perspective. Virus Research, 107 (2), 141-149 DOI:10.1016/j.virusres.2004.11.004[]