Molecular evolution of dinoflagellate luciferases, enzymes with three catalytic domains in a single polypeptide

Abstract

Enzymes with multiple catalytic sites are rare, and their evolutionary significance remains to be established. This study of luciferases from seven dinoflagellate species examines the previously undescribed evolution of such proteins. All these enzymes have the same unique structure: three homologous domains, each with catalytic activity, preceded by an N-terminal region of unknown function. Both pairwise comparison and phylogenetic inference indicate that the similarity of the corresponding individual domains between species is greater than that between the three different domains of each polypeptide. Trees constructed from each of the three individual domains are congruent with the tree of the full-length coding sequence. Luciferase and ribosomal DNA trees both indicate that the Lingulodinium polyedrum luciferase diverged early from the other six. In all species, the amino acid sequence in the central regions of the three domains is strongly conserved, suggesting it as the catalytic site. Synonymous substitution rates also are greatly reduced in the central regions of two species but not in the other five. This lineage-specific difference in synonymous substitution rates in the central region of the domains correlates inversely with the content of GC3, which can be accounted for by the biased usage toward C-ending codons at the degenerate sites. RNA modeling of the central region of the L. polyedrum luciferase domain suggests a function of the constrained synonymous substitutions in the circadian-controlled protein synthesis.

Fig. 1.

Bar representation of the organization (Upper) and partial sequence alignments (Lower) of dinoflagellate LCFs. (Upper Top) Two copies of LCF genes arranged in tandem, separated by an intergenic region. (Upper Middle) The structure of Lp LCF, showing the N-terminal region followed by three repeated domains, D1, D2, and D3. (Upper Bottom) A diagram of the second domain of Lp subdivided into three regions: the N-terminal, the central, and the C-terminal. The central region (126–275) is more conserved than the flanking ones and is likely the catalytic core. (Lower) Alignment of sequences 30–70 of the N-terminal second domains of seven LCFs showing the four conserved histidines at positions 35, 45, 60, and 66.

Fig. 2.

Molecular phylogeny of dinoflagellate LCFs based on their nucleotide sequences. Numerals are the percentage of the bootstrap values. (A) Distance tree of the full-length coding sequences constructed by using the neighbor-joining method, separating seven LCF genes into three clades. The first includes LCFs from two Alexandrium species and Pr; the second comprises three Pyrocystis LCFs; and the third is made up of Lp LCF as an orphan. (B) Maximum likelihood tree (natural logarithmic likelihood =–11505.72613) of the catalytic domains. The corresponding domains of different LCFs group together, and the domains of Lp separate early from those of the other LCFs. (C) Parsimony tree (tree length = 694; consistency index = 0.7334 and retention index = 0.5727) of the N-terminal domains of the seven LCFs and of Lp_LBP and Pl_GST. The tree was generated using paup 4.0b. The bootstrap value in support of Lp LCF, Lp_LBP, and Pl_GST as a clade was lower (in parentheses) when the data were analyzed with maximum likelihood method. (D) Small subunit ribosomal RNA tree for the seven species studied here along with Noctiluca scintillans (Ns).

Fig. 3.

Cumulative numbers of synonymous and nonsynonymous substitutions along the domains; numbers were counted with the snap program. For intramolecular comparisons, only D1 and D2 are shown; similar results were obtained for D1/D3 and D2/D3. For intermolecular comparisons, only the first domain pairs are shown; the other two gave the same result. (A) The changes in the cumulative numbers of the nonsynonymous substitution rates between the domains are similar along the domains for all LCFs, with pronounced reductions in the central regions. (B) The cumulative numbers of synonymous substitutions for intramolecular domains, especially for the first domain as compared with one of the other two, are lower in the central region for Lp and Pr LCFs but not for others. (C and D) Similar comparisons also were carried out pairwise for corresponding domains of different LCFs. Both the cumulative numbers of nonsynonymous substitutions and those of synonymous substitutions increase at about the same rate along the entire domain; the absolute numbers are smaller for the more closely related LCFs, such as At vs. Pr.

Fig. 4.

Relationship between synonymous substitution rates and codon bias. The three intramolecular domains (D1, D2, and D3) are compared pairwise for entire ORFs (1–376) or as three regions, the N-terminal (codon 1–125), the central (codon 126–275), and the C-terminal (codon 276–376). For each LCF, the values from three pairwise comparisons were averaged, and their standard deviations were computed. Regional analysis of synonymous (K_s) and nonsynonymous (K_a) substitution rates and their ratios are shown in A, B, and C, respectively. Note that the scales on the y-axis are different for K_s and K_a. The codon bias is measured in ENC and CBI (D and E). GC₃ contents for each region are indicated in F. All values were calculated with the dnasp package. The values in each category are all arranged from left to right for the LCFs from Aa, At, Lp, Pf, Pl, Pn, Pr, and the averages.