Mutation comicLast week, the Effect Measure blog1 talked about a paper that offered a new way of treating influenza.2 Briefly, the approach is to attack the virus by treating the host cell: Eliminating host functions that the virus requires, but that the host cell does not.

The authors of the paper commented that “targeting host cell determinants temporarily dispensable for the host but crucial for virus replication could prevent viral escape,” and Effect Measure observed that “It’s not obvious to me why the virus can’t as easily mutate in ways to adapt to a missing “office tool” as to a drug that affects an important viral function.” In the comments, I said:

I think the fundamental difference is that in the latter case, the virus needs just modify an already-present function; but if the tool is missing altogether, the virus would have to develop a whole new function from scratch, an altogether more difficult task.

That’s not to say that a virus could not do it — we see examples of this all the time, with viruses that have co-opted host functions and in some cases even host genes. But the process is usually much slower; we tend to recognize those events in hindsight, whereas we can often watch viral genes adapting to drugs in real time.

I still think my explanation is generally true, but here’s a counterexample, of what appears to be a virus evolving a brand-new function in just 5 weeks.

Background: Mammals have what seems to be a general defense against retroviruses like HIV. Several members of the APOBEC family of proteins are anti-retroviral;3 they force widespread mutations into the HIV genome, so many mutations that the virus can’t replicate or produce normal proteins.4 The reason HIV is able to replicate in spite of the APOBECs, is that HIV has in turn an anti-APOBEC protein, vif, that causes rapid destruction of several APOBECs. (I’ve mentioned this before, here, here, and here.)

vif structure (Zhang et al, Org Biomol Chem. 2007 Feb 21;5(4):617-26)
Predicted vif structure (Zhang et al.)

Variants of HIV that don’t have vif (either natural or artificial) can replicate pretty much normally in cells that don’t produce APOBECs; but they’re dead in cells that do have APOBECs, and their natural targets for infection do have APOBECs. So HIV is pretty much absolutely dependent on vif for its life-cycle.

There’s a lot of interest in trying to use this fact as an anti-HIV treatment. If there was a safe, effective anti-vif drug, then the APOBECs that are normally present could go ahead and destroy the virus. Now, we know from experience with other anti-HIV drugs that the virus could probably mutate vif to avoid this hypothetical drug, but let’s say it couldn’t. Let’s say the drug was completely effective in blocking vif, and there was no way for HIV to build a drug-resistant vif. Would the virus be completely helpless, or would it be able to develop a whole new anti-APOBEC function from scratch?

At least in a specific and somewhat limited set of conditions, that’s just what happened. Hache et al5 took a vif-deleted virus and tossed it into cells that have APOBEC3G. For several weeks, as you’d expect, there was almost no virus recovered (because without vif, the virus was destroyed by the APOBEC3G). But (see the figure to the left here; click for a larger version) after about 45 days, in 3 of the 48 cultures, virus abruptly started to grow again.

Hache et al, Fig. 1C
“Highlights of the long-term spreading-infection experiments for Vif-deficient viruses on vector-control- or APOBEC3G-expressing CEM-SS cell lines. Of the cultures, 45/48 showed no virus replication on APOBEC3G-expressing cells (flat lines not graphed).” 5

These new viruses still didn’t have any vif, but they were pretty much resistant to APOBEC3G — they had developed a brand-new function that conferred resistance to APOBEC3G. This new function behaves quite differently from vif. For one thing, vif protects against several different members of the APOBEC family, while the new variants were only resistant to APOBEC3G (they were still susceptible to APOBEC3F). And it took two simultaneous changes in the viruses for this to work: “Virus replication was only detectable after two mutations appeared: a noncoding A200T(C) transversion and a Vpr null mutation.

The mechanisms underlying this aren’t quite clear, and it’s really mysterious why getting rid of vpr would help make HIV resistant to APOBEC3G (vpr is a fairly mysterious protein in its own right, so it doesn’t offer a lot of handles to work it out). Anyway, although the authors did offer a number of possible explanations, that’s not really what I wanted to talk about. The point I wanted to make is that viruses can acquire new functions out of nothing, as well as modifying already-present functions.

Vif and APOBEC3G (Kao et al, 2004)
Co-expression of Vif and APOBEC3G in HeLa cells

Having said that, I think this actually does support my answer to some extent. Would this sort of resistance actually arise in vivo? Remember the question the authors were looking at originally: If we have an anti-vif drug, will resistance to it quickly arise? And if we look at the characteristics of the resistance in this artificial system, it’s actually somewhat encouraging:

  • The resistance took quite a while to pop up — several weeks, anyway 6
  • The resistance was rare even over that timescale. Only three of the 48 cultures threw out resistance variants. Each culture was infected with thousands of viruses originally,7 so you could say that the rate was lower than 3 in 400,0008

(Both of these are sort of what you’d expect from variants that require multiple mutations for resistance. Single mutations occur very frequently, but multiple mutations are exponentially9 less frequent.)

  • The resistance was partial. These viruses were only resistant to APOBEC3G. In a natural infection, these resistant viruses would probably still be killed, because they’re still susceptible to APOBEC3F

So: Yes, viruses can develop new functions, but it’s probably still fair to say they’re not as adept at this as at modifying existing functions.

  1. Which you should be reading, if you’re not already[]
  2. Karlas, A., Machuy, N., Shin, Y., Pleissner, K., Artarini, A., Heuer, D., Becker, D., Khalil, H., Ogilvie, L., Hess, S., Mäurer, A., Müller, E., Wolff, T., Rudel, T., & Meyer, T. (2010). Genome-wide RNAi screen identifies human host factors crucial for influenza virus replication Nature DOI: 10.1038/nature08760[]
  3. Sheehy, A. M., N. C. Gaddis, J. D. Choi, and M. H. Malim. 2002. Isolation of a human gene that inhibits HIV-1 infection and is suppressed by the viral Vif protein. Nature 418:646-650.[]
  4. Mangeat, B., P. Turelli, G. Caron, M. Friedli, L. Perrin, and D. Trono. 2003. Broad antiretroviral defence by human APOBEC3G through lethal editing of nascent reverse transcripts. Nature 424:99-103[]
  5. HACHE, G., SHINDO, K., ALBIN, J., & HARRIS, R. (2008). Evolution of HIV-1 Isolates that Use a Novel Vif-Independent Mechanism to Resist Restriction by Human APOBEC3G Current Biology, 18 (11), 819-824 DOI: 10.1016/j.cub.2008.04.073[][]
  6. How does this compare to the time it would take HIV to pop up resistant variants to one of the drugs in HAART therapy?  Anyone know? Bueller? Anyone?[]
  7. It’s not clear from their methods exactly how many, but 10,000 seems like a reasonable guess[]
  8. That’s not really a valid interpretation but it offers an upper limit, anyway[]
  9. Exponentially? Geometrically? Anyway, “much less”[]