CHAPTER 2: COMPUTATIONAL STUDIES
2.4 DISCUSSION
Table 2.5 Amino acid peptide sequences of the p2/NC cleavage from representative sequences for top and bottom ranked ligand groups
Rank Ligand name P5 P4 P3 P2 P1 P1` P2` P3` P4` P5`
Consensus Group M N T T I M M Q R G N
Consensus Subtype B S A T I M M Q R G N
Consensus Subtype C N T N I M M Q R S N
1 44 G H P N V M M Q R G N
2 281 K A I M I Q R G N
3 10 N A V M M Q K S N
4 17 G A A G I M M Q R S N
5 16 G A A G I M M Q K S N
6 217 S T K I M I Q N S N
7 39 G G H A G I M M Q R S N
8 241 S G A A A A I M M Q K S N
9 387 S N I L M Q R S N
10 78 G T N I M I Q R N N
11 202 S N I L V Q R S S
377 354 N T H I M M Q K N N
378 156 N T N I L M Q R S N
379 149 N S N I M M Q R N N
380 64 S A N I M M Q R S N
381 14 S I N I M M Q R S N
382 271 N V N I M M Q R S N
383 80 G T N I M M Q R G N
384 115 N G A I M M Q R G N
384 164 N T N I M M Q R S N
386 145 N P N I M M Q K S N
387 250 N T N I M M Q K S N
Amino acids are colour-coded based on their physio-chemical properties. Key: Yellow= small hydrophobic; green=large hydrophobic; light blue= polar hydroxyl group; dark blue= polar group negative; purple= polar group positive.
using the p2/NC CS as the substrate. No other studies (to the author’s knowledge) have used molecular docking to determine affinities between PR and the natural substrate.
Analysis of the sequence library found that the p2/NC cleavage site exhibited extensive variation, more so on the N terminal than the C terminal side of the CS. Others have also found this site to be variable (Barrie et al., 1996, Cote et al., 2001, Feher et al., 2002, De Oliveira et al., 2003, Malet et al., 2007, Ho et al., 2009, Ho et al., 2008). Polymorphisms found in the sequence data for this study included I376V and a R380K. The I376V (Malet et al., 2007, Ho et al., 2008) and the R380K (Cote et al., 2001, Myint et al., 2004, Malet et al., 2007) polymorphisms have been seen previously in viruses from both PI naïve and experienced patients.
The polymorphisms S373Q, A374P /S, and T375S have only been seen in PI experienced patients (Malet et al., 2007), but were commonly observed in this data set. Three possible explanations exist for this apparent paradox. Firstly, these sequences were incorrectly labelled as PI naïve (or the patients mistakenly identified themselves as PI naïve).
Secondly, these sequences represented transmitted PI associated mutations. Thirdly, these sequences could represent naturally occurring polymorphisms. Subtype C sequences often contain polymorphisms which are considered resistance mutations in subtype B (Velazquez-Campoy et al., 2001, Martinez-Cajas et al., 2008).
The results of the docking procedure found that ligands with better docking scores were mostly composed of small, hydrophobic amino acids. This was to be expected given that the substrate binding cavity of HIV PR is lined with hydrophobic residues (Brik and Wong, 2002, Perez et al., 2010). Therefore hydrophobic amino acids would theoretically have
more favourable interactions in that environment. These interactions (hydrophilic/hydrophobic and negative/positive charge) are taken into account during scoring. Compact amino acid side chains could have an advantage, as these residues were less likely to clash sterically with protein residues inside the substrate binding cavity.
Other studies have classified the CS as 4 amino acids on either side of the location of cleavage (Perez et al., 2010, Özen et al., 2011). For this study, 5 flanking amino acid were rather used, which has also been supported (Malet et al., 2007, Ho et al., 2009, Verheyen et al., 2009). Longer peptides may allow for the determination of the effect for 3D protein folding. However this docking software may not be to correctly identify the cleavage site if the peptide were longer. In addition, full length Gag precursor has not yet been crystallised.
In addition, the uncertainty regarding the recognition and binding mechanism of PR in is in part related to the lack of knowledge of the tertiary protein structure of Gag. A 2010 study (Perez et al.) used molecular dynamics simulations to predict free energy of several PR:
peptide complexes. As with docking, the implication of a lower free energy is a more stable enzyme-ligand complex, equating to higher affinity. That study found that several non-CS sequences had a lower free energy than well-defined CS. This contradicts the traditional active site binding recognition approach, suggesting an alternative recognition mechanism of PR. The authors of that paper suggested that the tertiary structure of Gag determines the location of PR cleavage by virtue of whether a site was recessed in the core of the protein or exposed on the surface. Among the exposed sites, recognition of cleavage sites is determined by which residues lie in an accessible conformation for PR to bind and cleave. With regard to the result of this study, hydrophobic residues such as those in the higher scoring ligands are less likely to be exposed to the external surface of
the protein, and therefore, according to this model are less likely to be experience optimal cleavage by PR.
With regard to possible limitations of this approach, the docking software has not been specifically designed for enzyme-peptide interactions, but mainly for studies using small drug molecules. Nevertheless, this method was chosen as it was able to score and rank a library of peptide ligands based on predicted affinity.
In summary, computational studies were used here as a tool for the objective selection of p2/NC cleavage site sequences for use in the fitness assay. Analysis of PI naïve subtype C sequences retrieved from the Los Alamos HIV Sequence Database has revealed the p2/NC CS as extensively polymorphic, with the N terminal portion of the cleavage site substantially more variable than the C terminal region. Results from the docking procedure predicted that sequences similar to the consensus sequence for subtype C would have a lower binding affinity than those with small hydrophobic amino acids in positions P5’ to P2’. Three sequences with predicted high binding affinity and three with predicted low affinity binding were selected for further characterisation studies.