HIV evolves very quickly permitting it to escape the immune system of an infected individual, limiting the effectiveness of drug treatment and making vaccine design extremely difficult. For these reasons identifying whether there are constraints on HIV evolution is of primary importance. In this thesis, I examined the relationships between sequence diversity and a number of factors including protein structure, co-evolution and RNA secondary structures, all of which contribute to evolutionary constraints. Firstly, I demonstrate that while there is an increase in sequence diversity over time, this variation has a tendency to be limited to specific structural regions. Relating the sequence variability of individual amino acid residues with three- dimensional protein structure, I find a significant difference between evolutionary rates in regions buried in the core of the protein as compared with those exposed on the surface. This result indicates that missense mutation is affected by structural constraints. Secondly, by relating recombinant breakpoint positions with the potential numbers of losses in amino acid interactions within the structure, I propose that as well as missense mutation, recombination is also affected by structural constraints, due to the need to maintain intra-molecular interactions. Thirdly, I demonstrate a relationship between conservation of RNA secondary structure and limited sequence variation in protein level, indicating RNA secondary structures are additional evolutionary constraint. This link between sequence and protein and RNA structures not only demonstrates the limits of recent HIV-1 evolution but also highlights the origins of evolutionary constraint on viral change. Lastly, I examined whether evolutionary constraints and co- evolution patterns of HIV genomes are applied in novel mosaic vaccine strategy and suggest that polyvalent mosaic vaccine sequence may generate proteins with stable structures and co-evolving units. Detailed understanding of the constraints that restrict HIV’s possible evolution will inform analysis of drug and immune escape.

Selection of HIV-1 variants resistant to antiretroviral therapy is well documented. However, the selection in vivo of HIV-1 mutant species that can escape host immune system HLA class I restricted cytotoxic T-lymphocyte responses has, to date, only been documented in a few individuals and its clinical importance is not well understood. This thesis analyses the observed diversity of the HIV-1 reverse transcriptase protein in a well characterised, stable, HLA-diverse cohort of HIV-1 infected patients with over two thousand patient-years of observation. The results show that HIV-1 polymorphism is selected within functional constraints and is associated with specific HLA class I alleles. Furthermore, these associations significantly cluster along the sequence and tend to occur within known corresponding HLA-restricted epitopes. Absence of polymorphism is also HLA-specific and more often seen with common HLA alleles. Knowledge of HLA specific viral polymorphisms can be used to model an individual’s viral load from their HLA type and viral sequence. These results suggest that cytotoxic T-lymphocyte escape mutation in HIV-1 is critical to the host at an individual and population level as well as to short and long term viral evolution. This work provides new insights into viral-host interactions and has clinical implications for individualisation of HIV-1 therapy and vaccine design.