CORDIS - EU research results

Intra-patient evolution of HIV

Final Report Summary - HIVEVO (Intra-patient evolution of HIV)

Many important human pathogens like HIV or influenza virus are RNA viruses that evolve rapidly to evade the human immune response or develop resistance against antiviral treatment. Due to this fast evolution multiple drugs against HIV have to be taken simultaneously to avoid treatment failure and the seasonal influenza vaccine has to be updated frequently. The aim of this project was to study such rapidly evolving organisms and develop a quantitative understanding of the process. In addition to being medically important, coevolution of RNA viruses and their hosts is a paradigmatic example of general host-pathogen coevolution which occurs in all forms of life. Lessons learned from HIV and influenza inform our understanding for other system where evolution cannot be observed directly.

Virus adaptation is driven by the interplay of mutation, selection for rapid virus replication, and avoidance of immune selection. To shed light on this complex dynamics, we had proposed to monitor virus evolution within HIV positive individuals by sequencing virus from plasma samples at many time points during the infection. Using this and publicly available data, we had proposed to develop a quantitative model of HIV evolution and estimate the landscape of fitness costs along the HIV genome. In addition, we proposed to develop population genetic theory that can rationalize the observed dynamics and generalize it to other examples.

Using state of the art sequencing technology, we have deep-sequenced full genomes of viruses in about 100 samples from 11 patients and investigated the dynamics of mutations within these populations. We showed that HIV has a strong tendency to mutate back to the consensus type. About one third of all mutations observed are such reversion, indicating that with-in hosts, HIV is constantly undoing mutations that it acquired in previous hosts. We further quantified this preference for an optimal genotype by mapping the fitness cost of mutations at every site in the genome. This landscape of fitness costs rationalizes macro-evolutionary patterns of conservation, identifies sites of viral vulnerability, and highlights sites of unexplained conservation for follow-up studies. Using publicly available data from early infection, we have developed a method to estimate the contribution of individual cytotoxic T-cell (CTL) clones to viral control and shown that the effect of CTLs is often much larger than previously estimated.

In addition to the empirical work, we have developed a theory of rapid adaptation that -- in contrast to the standard neutral models -- is consistent with the type of dynamics observed in rapidly adapting pathogens such as HIV and influenza. We initially developed this theory for asexual organisms and then generalized it to describe sexual (with recombination) life cycles. The theory has identified powerful statistics that can be used to estimate parameters from data and uncovered an unexpected link between the population genetics of rapid adaptation and an active field of mathematical research (Lambda coalescents).

The theoretical work as let to another unexpected application. We have shown that the patterns of genetic diversity that arise in rapidly adapting population can be used to predict successful genotypes among the circulating diversity. We have developed this insight into a method to predict the composition of future population of seasonal influenza virus population with the aim to improve the choice of virus strains used in vaccine production.

In addition these results, we have been exploring new ways to share research and data with the public and the scientific community. All data on intra-patient HIV evolution are available in a web application at which allows the interactive exploration of the dynamics within each of the individuals we studied. The data are preprocessed for facilitate reuse and exploration. In collaboration with Trevor Bedford, we have also developed a real-time tracking platform for influenza virus evolution that visualizes our predictions always using the latest available data.