Genetic variation of HIV-1 : molecular epidemiology and viral evolution

Abstract

HIV-1 displays a very high genetic variability. The evolutionary process of the virus generates a nucleotide substitution rate which is approximately one million times faster than that of higher organisms. This remarkable rate in combination with a half-life of six to eight hours, gives HIV-1 an enormous potential of genetic adaptation. This is probably an important reason for the inability of the immune system to combat the virus. Another striking example of the genetic adaptation is seen during antiviral treatment. Often when a patient is treated, drug sensitive virus is cleared only to be replaced by drug resistant escape mutants. Another important field where the genetic variability can be explored is in phylogenetic analyses. Many studies have applied such analyses to investigations of great epidemiological interest, ranging from questions on the origin of HIV to detailed studies of local transmission chains.

The aim of this thesis was to investigate the genetic variation on several inter-related levels. First, a method to estimate the population structure in a clinical sample was developed. HIV-1 in an infected individual has been shown to consist of a population rather than a single genetic variant, and has therefore suggested to follow the quasispecies model. The method is based on direct PCR amplification of DNA from uncultured lymphocytes or RNA from plasma. The amplicons are subsequently sequenced by a method that incorporates dideoxy and deoxynucleotides with similar efficiency. An accurate estimate of the population structure is obtained, where minor sequence variants (down to 10%) could be detected in polymorphic positions.

Second, phylogenetic tree building methods and nucleotide substitution models were compared and evaluated. Different gene fragments were also tested to elucidate which gene is best suited for phylogenetic analysis. For inference of the topological order, we found that the choice of method was less important than the choice of gene fragment. In fact, the best reconstruction of a true tree was derived from a combined gag-env fragment. This shows that itis better to combine all available sequence information rather than to use it separately. In our investigation the V3 fragment gave better opportunities than pl7 gag and pol RT to resolve transmission events close in time, since this region evolves faster. For estimating accurate genetic distance and branch lengths the substitution model is of utmost importance. The most realistic model was the general reversible Markov process including a gamma distribution for rate variation across sites. This model estimates 9 parameters from the data set and a given tree topology. Thereby, unbalanced nucleotide frequencies, rate variation across sites, and different nucleotide substitution rates, such as preferred G to A substitutions, are taken into account during estimation of the genetic divergence.

We have derived experience from many molecular investigations, including mother-to-child transmission, several forensic investigations, screening for genetic subtypes of HIV-1 and the detection of a new subtype. These investigations have suggested infection routes and likely epidemiological relationships. They have also shown that the transmitted virus appears to be derived from the dominant population of the donor, without any obvious selection. In addition, more than one viral form can be transmitted. In conclusion, this study has addressed and answered several important questions concerning the genetic variability of HIV-1 from both a macroscopic and microscopic perspective.