Tracing globin phylogeny using PSIBLAST searches based on groups of sequences

PSIBLAST search of a protein database with a query sequence is a widely used tool for the detection of related but evolutionarily distant sequences. Iterations carried out until convergence (i.e., until the majority of the sequences most similar to the query sequence are extracted from the database) also produces an ordered list of more distantly related (i.e., less similar sequences) and false positives belonging to other protein families. Thus, a PSIBLAST search based on one group of globin sequences will provide sequences left in the database that are more distantly related (i.e., belong to other groups of globins) ordered according to the E value or bit score. The relative order of the scores should yield a clue to the relative similarity of the query group to the other groups of globins. Histograms of E values or bit scores from PSIBLAST searches using vertebrate myoglobins, cytoglobins, alpha- and beta-globins, and neuroglobins as query groups show distributions that are congruent with the accepted phylogenetic tree. An illustration of more distant relationships is demonstrated by the results using neuroglobins as the query group, which show a striking similarity to bacterial single-domain globins and flavohemoglobins from bacteria and eukaryotes. Furthermore, it is observed that sequences belonging to the three undoubtedly ancient globin lineages form very broad distributions, while recently evolved groups such as cytoglobins have narrow distributions. Thus, the breadth of a distribution of E values or bit scores for a query group may be related to its evolutionary age.