A Systematic Analysis Of Neighboring-site Effects Of Amino Acids And Amino Acid Substitutions

Neighboring-site effects mean that there are composition preferences for neighboringsites of amino acids and amino acid substitutions in protein sequences. For amino acidsubstitutions, they also mean that the substitution process is context-dependent. As animportant characteristic of amino acids and amino acid substitutions, neighboring-site effectsplay a key role in stabilization of protein structure and function. In the current study,neighboring-site effects of amino acids and amino acid substitutions in proteins werecomprehensively investigated in four parts as follows.Part1. Using the relative entropy method, the neighboring-site effects of20amino acidsin human (Homo sapiens) proteins were investigated. The results showed that:(1) For mostamino acids, the highest relative entropy appeared at the first two immediate flanking sites ofN-and C-termini, indicating that these sites have the strongest composition preferences.(2)The neighboring-site effects of C and G were exceptional. The highest relative entropy of Cappeared at the third flanking site of each side, while an evident periodic pattern was detectedat G neighboring sites.(3) Besides a few amino acids as F, W, Y, the others commonlyshowed the feature of self-preference at the neighboring sites (i.e. the same kind of aminoacid tended to show high propensity at the neighboring sites).(4) The neighboring-site effectsof amino acids varied in different secondary structures. In α-helix, the highest relativeentropy was commonly seen at the third or fourth flanking site; in coil, the highest relativeentropy was mainly observed at the nearest two neighboring sites.(5) The neighboring-siteeffects of amino acids also varied in different cellular locations.Part2. We further investigated the neighboring-site effects of amino acids in fly(Drosophila melanogaster), yeast (Saccharomyces cerevisiae) and E.coli (E.coli K12MG1655), and compared the differences in different species. The results showed that:(1)Some amino acids showed similar neighboring-site effects in different species, such as theexceptional neighboring-site effect of C which was observed in all species; while some othersdiffered greatly, such as the periodic pattern of G which only existed in human proteins.(2) Incomparison with eukaryotes, the neighboring sites of amino acids showing apparentcomposition preferenceswere fewer in prokaryotes, indicating that the neighboring-siteeffects of amino acids in prokaryotes were weaker.(3) The feature of self-preference ofamino acids was much evident in eukaryotes than in prokaryotes. Part3. By grouping the20amino acids into nonpolar (NON), polar neutral (NEU),positive (POS) and negative (NEG) amino acids, we investigated the neighboring-site effectsof amino acid substitutions in the mouse genome on the basis of overcoming the influence ofneighboring-site effects of amino acids themselves. The results showed that:(1)12out of16(4×4) substitution types showed significant neighboring-site effects.(2) The mostover-represented substitution pattern was the NEGâ†'NEG occurring in NEG_NEG context,the frequency of which was2-fold higher than that of expectation. In contrast, the mostunder-represented substitution pattern was NEUâ†'NEU occurring in NON_NON context, inwhich the number of observation was about87%of the expectation.(3) The neighboring-siteeffects of amino acid substitutions varied in different protein secondary structures. Four andnine substitution types were evidently neighbor dependent in α-helix and coil, respectively;while no substitution type with apparent neighboring-site effect was observed in β-strand.Part4. Applying relative entropy, the neighboring-site effects of380(20×19)substitution types were comprehensively investigated in the human, mouse and dog. Theresults showed that:(1) Nearly half of380substitution types showed evident neighboring-siteeffects. Among20amino acids, P was the most preferred residue at the neighboring sites ofmany substitution types. Besides, E, S, A, G, K and Q were also frequently observed at theneighboring sites.(2) When the site was substituted by P, E, S, A, G or K, the neighboringsites of it tended to show high propensity for the amino acid type as the post-substituted one.(3) In α-helix, the highest relative entropy of amino acid substitutions commonly appeared atthe third or fourth flanking site; in coil, the nearest two flanking sites of amino acidsubstitutions showed the greatest composition preference in most cases.In summary, a systematic analysis of neighboring-site effects of amino acids and aminoacid substitutions was conducted in this study. The results could improve secondary structureprediction algorithms and protein design; moreover, this study provided useful informationfor developing context-dependent amino acid substitution models in the future.