New paper in PLOS Genetics. CpG sites are costly for HIV

We are publishing a new paper today in PLOS Genetics!

Title: Within-patient mutation frequencies reveal fitness costs of CpG dinucleotides and drastic amino acid changes in HIV.

“We” here means 6 coauthors of 5 different nationalities (which shows why travel bans are bad for science)! This is the first published result of our collaboration with Adi Stern in Tel Aviv.

 

Result 1: CpG sites are costly for HIV.

The most surprising result of our paper (to me at least) is that mutations that create CpG sites are significantly more costly than mutations that do not create CpG sites. We know that they are more costly because they segregate at lower frequencies. Clearly, CpG sites are not good for HIV!

In this figure (which shows synonymous mutations only) you can see that the light blue dots are at lower frequencies than the green dots – each blue dot is the frequency of a synonymous mutation that creates a CpG site and each green dot is the frequency of a synonymous mutation that does not.

Why are CpGs bad for HIV? This paper suggests that CpG sites are recognized by our immune system because Zinc-finger Antiviral Protein “binds directly and selectively to RNA sequences containing CG dinucleotides”.

Result 2: It works!

A more technical result is that we show that we can actually use within-patient mutation frequencies to estimate fitness costs of mutations. This means that we can study costs as they occur *now* (as opposed to phylogenetic approaches) and *in vivo* (as opposed to cell-culture based approaches).

This figure shows the single-site frequency spectra for three sites. Mutation frequencies are observed in 160 different patients. The second row shows simulated mutation frequencies using inferred cost estimates from the data. They look very similar to the real site frequency spectra!

Here we show that average in vivo mutation frequencies are lowest for nonsense mutations (black), higher for non-synonymous mutations (pink) and highest for synonymous mutations (yellow). This is exactly what mutation-selection equilibrium predicts.