Tl;DR there are multiple double- and triple-mutant combinations in the Vgsc gene which I think we should be more concerned about. Also, they are evolving in different ways.

We recently published a new analysis of insecticide resistance using data from the Anopheles gambiae 1000 Genomes Project (Ag1000G):

I wanted to highlight here just one point from the paper, about double- and triple-mutants. To help make sense of it, let me first rewind and give a little context.

Pyrethroid target-site resistance in Anopheles gambiae

Malaria control relies heavily on insecticide-treated bednets (ITNs). All ITNs use a pyrethroid insecticide. Pyrethroids work by binding to a mosquito nervous system protein called the voltage-gated sodium channel. Malaria mosquitoes can become resistant to pyrethroid insecticides via genetic changes in the Vgsc gene which encodes this protein. This is known as pyrethroid target-site resistance.

The first pyrethroid target-site resistance mutation found in Anopheles gambiae was a leucine to phenylalanine substitution at position 995 in the protein, L995F (Martinez-Torres et al. 1998). Shortly after, a different substitution was found at the same position in the protein, L995S (Ranson et al. 2000). A decade later, a third mutation – N1570Y – was found (Jones et al. 2012). N1570Y was never found by itself, only in combination with L995F, as a double-mutant.

(Side note: many papers number these mutations according to their position in the house fly protein sequence. In the house fly, these mutations are numbered L1014F, L1014S and N1575Y.)

Wang et al. (2015) then showed experimentally that both L995F and L995S cause pyrethroid resistance, and N1570Y by itself does not cause any resistance at all, but the double-mutant L995F+N1570Y is much more potent. For example, compared to the normal protein, L995F is 8-fold more resistant to permethrin, but L995F+N1570Y is 80-fold more resistant. How exactly this works at the molecular level is something of a mystery, as N1570Y is within an intracellular loop of the protein and cannot affect pyrethroid binding directly. But the VGSC protein is an intricate and complex molecular machine with many moving parts. Wang et al. suggested that N1570Y could induce a small but important change to the shape (conformation) of a segment of the protein, enhancing the effect of L995F.

The important point here is that the double mutant L995F+N1570Y is highly resistant to pyrethroids. The degree of resistance is potentially important because a pyrethroid-treated ITN might still kill a mosquito with mild resistance, but a highly resistant mosquito might survive and tolerate the insecticide, enabling it to feed and spread malaria. Exactly how resistant a mosquito needs to be before ITNs stop being effective is a matter for some debate. But nevertheless, highly resistant mosquitoes are cause for concern.

Vgsc double- and triple-mutants

Until our study, these were the only three resistance mutations studied in Anopheles gambiae mosquitoes. But pyrethroid resistance is a problem in lots of other insect species, including other mosquitoes that transmit disease, as well as a slew of agricultural pests. The Vgsc gene is common to all insects, and a whole carnival of other resistance mutations had been found in this gene in these other pest species, reviewed in Dong et al. (2014). We wondered if there might be other mutations in malaria mosquitoes that had previously been missed, because most previous studies did not sequence the full length of the gene.

We used data from whole-genome sequencing of mosquitoes from 13 African countries to survey the full Vgsc gene for nucleotide variants that could alter the protein sequence. As expected, L995F and L995S were at high frequency in many populations, confirming a central role for these two mutations. The double-mutant L995F+N1570Y was also present in several mosquito populations. But we also found another 20 protein-altering mutations at appreciable frequency in one or more populations. What is more, these mutations occurred in distinct combinations:

  • L995S
  • L995F
  • L995F+R254K
  • L995F+D466H+I1940T
  • L995F+T791M+A1746S
  • L995F+V1507I
  • L995F+N1570Y
  • L995F+E1597G
  • L995F+K1603T
  • L995F+V1853I
  • L995F+I1868T
  • L995F+P1874S
  • L995F+P1874L
  • L995F+A1934V
  • V402L+I1527T

I think all of the above double- and triple-mutants are cause for concern, because they potentially confer a high degree of pyrethroid resistance. There is no direct experimental evidence for this yet in Anopheles gambiae mosquitoes, other than what is known about the L995F+N1570Y, but consider the following.

Firstly, V402L+I1527T is concerning because V402L by itself is known to cause pyrethroid resistance in other insects, and I1527T occurs immediately adjacent to a pyrethroid binding site. We found this combination at up to 15% frequency in West African An. coluzzii populations, but our sampling of these countries was all done in 2012, and we don’t know what has happened since. We need more recent samples to determine whether this variant is increasing or decreasing in frequency. If it is increasing, then that would be further cause for concern, because that would mean it is out-competing L995F combinations which are also found in those populations.

Secondly, Vgsc is a highly conserved gene, under strong functional constraint. Aside from the V402L+I1527T combination, and the mutations found in combination with L995F, there are hardly any other mutations at appreciable frequency. In other words, Vgsc is not the kind of gene where mutations get to high frequency by chance. The fact that so many mutations are found exclusively in combination with L995F is a strong hint that, like N1570Y, they are somehow enhancing pyrethroid resistance, and thus under selection.

We need to be monitoring all of these mutations. And we need to know more about their resistance phenotype.

Evolutionary trajectories

One final thought. There are three distinct evolutionary trajectories here. I find that interesting.

Consider L995S. By itself it confers pyrethroid resistance, and is under strong positive selection. But there are essentially no double- or triple-mutant combinations involving this mutation. In other words, it’s an evolutionary dead end. Why?

Now consider L995F. By itself it confers pyrethroid resistance, and we also find it together with a jamboree of other substitutions. For some reason, L995F (plus selection for pyrethroid resistance) unlocks a rich new landscape of evolutionary possibilities. Again, why?

Finally, consider V402L+I1527T. We essentially only ever find these mutations together, hardly ever by themselves. They really need each other. Each by itself may be somewhat deleterious, but they really work well together. Here, there is a trough in the evolutionary landscape that needs to be crossed or leaped somehow. We know from other systems (e.g., COVID-19) that such evolutionary leaps can happen in general, but what are the specific circumstances in malaria vectors that make that possible?

Much yet here to be unravelled.