Background: Codon usage bias has been described for various organisms and is thought to contribute to the regulation of numerous biological processes including viral infections. HIV-1 codon usage has been previously shown to be different from that of other viruses and man. It is evident that the antiretroviral drugs used to restrict HIV-1 replication also select for resistance variants. We wanted to test whether codon frequencies in HIV-1 sequences from treatment-experienced patients differ from those of treatment-naive individuals due to drug pressure affecting codon usage bias.
Results: We developed a JavaScript to determine the codon frequencies of aligned nucleotide sequences. Irrespective of subtypes, using HIV-1 pol sequences from 532 treatment-naive and 52 treatment-experienced individuals, we found that pol sequences from treatment-experienced patients had significantly increased AGA (arginine; p = 0.0002***) and GGU (glycine; p = 0.0001***), and decreased AGG (arginine; p = 0.0001***) codon frequencies. The same pattern was not observed when subtypes B and C sequences were analyzed separately. Additionally, irrespective of subtypes, using HIV-1 gag sequences from 524 treatment-naive and 54 treatment-experienced individuals, gag sequences from treatment-experienced patients had significantly increased CUA (leucine; p < 0.0001***), CAG (glutamine; p = 0.0006***), AUC (isoleucine; p < 0.0001***) and UCU (serine; p = 0.0005***), and decreased AUA (isoleucine; p = 0.0003***) and CAA (glutamine; p = 0.0006***) codon frequencies.
Conclusion: Using pol and gag genes derived from the same HIV-1 genome, we show that antiretroviral therapy changed certain HIV-1 codon frequencies in a subtype specific way.
Keywords: Antiretroviral therapy; Bioinformatics; Codon usage frequency; HIV-1; Resistance.