Homologies in Valpha and Vdelta genes of the T cell receptor: a statistical analysis.

The functions of the gamma/delta T cell receptor (TCR) are much less understood than those of alpha/gamma TCR. There is however a growing consensus that gamma/delta T lymphocytes operate in distinct manners, and at different levels than alpha/beta T cells. gamma/delta T cells appear to belong to a separate cell lineage, which tends to differentiate at earlier stages of development than apha/beta lineages. Several results suggest that gamma/delta cells do not normally interact with B cells in the characteristic helper role of alpha/beta T lymphocytes. The gamma/delta receptor does not show the strong specificity (restriction) for MHC proteins of class I or II, typical of the alpha/beta receptor. Hence, some authors believe that the main function of this receptor could be to recognize antigenic fragments attached to monomorphic, MHC like antigen presenting proteins. In man and mouse the TCR alpha and delta genes are confined together in a narrow region, bounded by the Calpha gene segment at the 3' end and by the mixed array of Valpha and Vdelta elements at the 5' end. It is therefore most likely that the alpha and delta genes were originated by duplications of the appropriate elements of a single ancestral gene. If the functional roles of the alpha/beta and gamma/delta receptors are indeed as distinct as they are thought to be, the evolutionary divergence of the alpha and deltagenes could have had pronounced effects on the repertoire of Valpha and Vdelta peptidic domains since the bulk of the TCR V-domain, according to several authors, is responsible for contacts with the antigen presenting protein. It is then possible that evidence in support of a functional divergence of the alpha/beta and alpha/delta receptors is revealed by careful analysis of the pattern of sequence similarities between the various Valpha and Vdelta elements. We have considered the nucleotide sequences of 59 Valpha/delta elements for man, distributed in 33 families, and of 101 Valpha/delta elements for mouse, covering 21 families. Among the human sequences, 8 correspond nominally to Vdelta elements but 4 of them (Vdelta 4-7) very rarely are used in mature 8 genes, being normally expressed in a chains. The normal VS human repertoire consists of the remaining elements (Vdelta 1,2,3,8). In the mouse sequences, 11 nominal Vdelta elements are represented. Of these, 7 cover the normal repertoire (Vdelta 1-7, with 7 close variants of Vdelta 6), while the others (Vdelta 8-11) are very rarely used in delta chains. For man and mouse separately, we have aligned the inferred aminoacld sequences pairwise, using the Needleman-Wunsch algorithm with standard scoring parameters. We measured the similarity Sij between sequences i and j, of respective length Ni and Nj, by the quantity Sij = Qij/max{Ni , Nj}, where Qij is the total score achieved by the alignment of Ii and j. In this way, any two identical sequences would have maximum similarity irrespective of their length, while a sequence j identical to a subsequence of i would have a similarity to i smaller than maximum. This is justified whenever we are confident that all loci of interest are represented by essentially complete sequences. A similarity measure of this type, among others, has been used in other articles. Distances Dij between any two sequences i and j were calculated from the matrix of similarities {Sij} using the standard conversion formula: Dij = sqrt(Sii + Sjj)-2Sij). The two distance matrices {Dij}, calculated for man and mouse respectively, provided the basic data for the two statistical methods by which we analyzed the pattern of sequence similarities between Valpha and Vdelta elements: (i) Principal Coordinate Analysis (PCA) and (ii) Average Linkage Cluster Analysis (CLA). PCA tries to represent each seq