Internal and external paralogy in the evolution of tropomyosin genes in metazoans

Abstract

Nature contains a tremendous diversity of forms both at the organismal and genomic levels. This diversity motivates the twin central questions of molecular evolution: what are the molecular mechanisms of adaptation, and what are the functional consequences of genomic diversity. We report a 22-species comparative analysis of tropomyosin (PPM) genes, which exist in a variety of forms and are implicated in the emergence of a wealth of cellular functions, including the novel muscle functions integral to the functional diversification of bilateral animals. TPM genes encode either or both of long-form [284 amino acid (aa)] and short-form (approximately 248 aa) proteins. Consistent with a role of TPM diversification in the origins and radiation of bilaterians, we find evidence that the muscle-specific long-form protein arose in proximal bilaterian ancestors (the bilaterian 'stem'). Duplication of the 5' end of the gene led to alternative promoters encoding long- and short-form transcripts with distinct functions. This dual-function gene then underwent strikingly parallel evolution in different bilaterian lineages. In each case, recurrent tandem exon duplication and mutually exclusive alternative splicing of the duplicates, with further association between these alternatively spliced exons along the gene, led to long- and short-form-specific exons, allowing for gradual emergence of alternative "internal paralogs" within the same gene. We term these Mutually exclusively Alternatively spliced Tandemly duplicated Exon sets "MATEs". This emergence of internal paralogs in various bilaterians has employed every single TPM exon in at least one lineage and reaches striking levels of divergence with up to 77% of long- and short-form transcripts being transcribed from different genomic regions. Interestingly, in some lineages, these internal alternatively spliced paralogs have subsequently been "externalized" by full gene duplication and reciprocal retention/loss of the two transcript isoforms, a particularly clear case of evolution by subfunctionalization. This parallel evolution of TPM genes in diverse metazoans attests to common selective forces driving divergence of different gene transcripts and represents a striking case of emergence of evolutionary novelty by alternative splicing.

FIG. 1.

TPM gene structures in metazoans. (A) General gene structures for bilaterian TPM genes. Bilaterian genes contain two promoters. Short-form transcripts are transcribed from the downstream promoter. Long-form transcripts are transcribed from an upstream promoter, contain two additional exons (1a and 2), and do not include the first exon from short-form transcripts (exon 1b). Nonbilaterian genes encode only short-form transcripts. (B) TPM gene structures from nine bilaterians and two nonbilaterians. TPM genes in most bilaterians contain copies of various exons (boxes with the same color), in contrast to the simpler genes of nonbilaterians; these tandemly duplicated exon sets (MATEs) are alternatively spliced in a mutually exclusive manner. The bilaterians Ciona intestinalis and Hellobdella robusta represent exceptions to this pattern: TPM genes in these species lack duplicated exons, and encode either long- or short-form transcripts, but not both. Boxes/lines indicate exons/introns. Homologous exons (either orthologous or paralogous) are indicated by the same color. Number after the species name, when present, correspond to the paralog represented in the figure (i.e., Homo sapiens 1 represents human TPM1). Full species names are given in the Methods.

FIG. 2.

Production of long and short forms by coordinated processing of alternative transcript regions. (A) General scenario. For three two-exon MATEs, there are eight possible combinations (at right). However, typically only two reciprocal combinations dominate available transcript sequences, with other forms never or almost never observed. (B) The TPM1 gene of D. melanogaster contains five alternative regions: alternative promoters, three two-exon MATEs, and a pair of alternative terminal exons (top line: gDNA). Among the 32 (= 2⁵) possible combinations, only two are observed, with each alternative region found in just one transcript. Exon duplicates range in amino acid identity from 11% to 60%. (C) The more complex structure of the TPM gene of L. gigantea. The gene contains alternative promoters, two two-exon MATE, two three-exon mates, and two terminal exons. Of the 144 possible structures, only three are observed.

FIG. 3.

AS and polyadenylation of terminal exons. (A) In some species such as C. elegans, AS leads to different numbers of copies of the terminal exon, with different exons encoding the protein terminus (gray) with downstream copies being entirely untranslated regions (white). (B) On other species such as D. melanogaster, alternative polyadenylation coupled to AS leads to differential usage of 3′ exons and polyadenylation, but less variation in UTR length. In each case, the top line represents genomic DNA, and subsequent lines represent observed transcripts.

FIG. 4.

Externalization of internal paralogs by subfunctionalization. (A) General scenario. A single ancestral gene encodes both long (exons above the line) and short (below) forms of the gene. Gene duplication and reciprocal loss of the two forms leads to two descendent genes each encoding just one form. (B) Expected phylogenetic signals under “simple” gene duplication and subfunctionalization. Under both scenarios, ancestrally constitutive exons in the duplicate genes should form a clade. In the absence of externalization/subfunctionalization, the case should be the same for exons that are MATEs in the ancestral gene. However, in the presence of subfunctionalization, for ancestral MATE exons, the gene duplicates should group with the MATE exons of like type: the descendent long-/short-form gene should more closely resemble the ancestral long-/short-form transcript. (C) Phylogenetic trees for gene duplicates in the Hellobdella robusta genome and alternative promoter/AS genes from other Lophotrochozoans. Long-form genes from Hellobdella robusta group with long-form transcripts from other species, whereas short-form genes group with short-form transcripts from other species. Numbers at the nodes correspond to posterior probabilities.

FIG. 5.

Model for the origin of the long TPM form in bilaterians. (A) Alignment of translated sequences for exons 2 and 3 from different bilaterian species and exon 2/3 from the nonbilaterian N. vectensis. Grey boxes indicate residues that are shared between the ancestral exon/exon 3 and the upstream exon 2. (B) Nucleotide sequence of the human TPM3 exons 2 and 3 and adjoining intronic sequences. A single nucleotide change at the exon 2 could have created the new splice site boundary. (C) Proposed two-step scenario for the origin of the long TPM isoform by tandem exon duplication of exon 1B and 3 (top) and frameshifting sliding in the newly formed exon 2 (bottom). Nve, Nematostella vectensis; Hsa3, Homo sapiens TPM3; Bfl, B. floridae; Lgi, Lottia gigantea.