Identification and annotation of FABP genes

In this study, most of 34 invertebrates have annotated genomes and FABP genes were directly retrieved from the databases. For the species without an annotated genome including Echinococcus multilocularis, Echinococcus granulosus, Heterorhabditis bacteriophora，Trichinella spiralis，Strongyloides ratti，Rhodnius prolixus, Haemonchus contortus and Ciona savignyi, we searched the databases using the following strategies. Candidate FABP genes were identified using TBlastN, with experimentally or putatively identified FABP gene(s) from a closely related species as a query sequence, to search various genome databases with a cut-off e-value of 1-e10 (Table 1). Otherwise, Schistosoma mansoni FABPs (Smp_095360 and Smp_046800) or C. elegans FABPs (NP_505016, NP_508558, NP_508557, NP_491928, NP_506440, NP_491926, NP_001041249, NP_506444 and NP_001033511) were used as queries. This strategy was used because FABP genes share 20 to 70% similarity at the amino acid level. We then applied two criteria to resulting “hits” to identify FABP genes. First, considering that most known FABPs are ~130 amino acids (aa) in length, we arbitrarily set the size range of FABPs from 80 to 180aa (130 ± 50aa). In addition, the sequences within the size limit were used for secondary structure prediction and those with the putatively typical structural elements were considered to be FABP genes.

Two sequential approaches were utilized for determination of the exons and exon boundaries. Firstly the exons and their boundaries were determined from TBlastN outcomes as highly-scored segment pairs or gaps within the segment pairs as described previously [11]. FABP gene structural models were then verified and finely modified using transcriptome data or expression sequence tags (ESTs). Segment pairs that dispersed over two or more supercontigs were not considered to build gene models in this study. The intron-exon boundaries were manually checked based on consensus splicing acceptor and donor sites and they conformed to the GT/AG rule.

FABP genes, identified using the approach above, were used as query sequences to search transcriptome and EST databases for the relevant species. This provided a means of validating the findings from genomic data alone (Table 1).

Sequence alignments and secondary structure prediction

The FABP protein sequences were aligned using Clustal W algorithm (MEGA 4.0) with default parameters [12] and then manually checked (Figure S1). The secondary structures of FABPs were predicted using Psipred [13].

Construction of a phylogenetic tree

Besides all the FABP amino acid sequences identified in this study, ten human FABP sequences were also included for phylogenetic analysis. Prior to tree construction, a best model was selected using TOPALi v2.5 [14]. A Bayesian tree was built using the following settings: WAG model plus gamma, 2 runs, 500,000 generations, 10 of sample frequency and 25% burn in. To confirm the topology of the tree, a ML tree was also built using the following settings: LG model plus gamma with 100 bootstraps.

Identification and annotation of FABP genes across invertebrates

During sequence searching we obtained high-scoring hits that encoded more than 180aa or fewer than 80aa, but all of which were excluded from further analyses in this study. For instance, a Branchiostoma floridae hypothetical protein (987aa, XP_002589099) contained a region at the C terminal that shared 96% identity with Branchiostoma belcheri FABP (136aa, ADD10136).

In total, 107 sequences falling within the specified size range and exhibiting appropriate secondary structure were collected from 32 invertebrate species including one placozoan, two annelids, one mollusc, five platyhelminths, seven nematodes, twelve arthropods, one echinoderm and three chordates (Table 1 and Supplementary text file). The identity of these putative FABP amino acid sequences ranged from 29.0% to 99.3% and they were predicted to have the typical tertiary structure (Figure S2). No homologues of FABPs were identified in two Cnidaria species, Hydra magnipapillata and Nematostella vectensis. Notably, four Haemonchus contortus FABP genes were identified by TBlastN searches against transcriptome (NCBI) but none of them was found in the genome, possibly due to incomplete genomic data (Sanger). One putative FABP transcript was derived from transcriptome or EST data, but its locus was not found in the genome of each of the following species: Helobdella robusta, Lottia gigantea, Schistosoma japonicum, Heterorhabditis bacteriophora and Saccoglossus kowalevskii. With the exception of the body louse, Pediculus humanus corporis, some or all FABP genes found in genomes were validated by EST or transcriptomic data.

Numbers of genomic loci for FABPs ranged from one (several arthropods and S. japonicum) to fifteen (the chordate, B. floridae) in invertebrate genomes (Table 1). Echinococcus multilocularis, Anopheles gambiae and B. floridae each had two distinct loci that encoded identical FABPs at the amino acid level. The introns of the two E. multilocularis FABPs were identical, whilst those of the A. gambiae and B. floridae FABPs were different in size and sequence. But there is not enough evidence to support that these FABP genes are transcribed into the same mRNAs.

Phylogenetic analysis of FABPs

As shown in the Bayesian tree (Figure 1), nematode FABPs formed two distant clades and with an exception of T. spiralis, each clade was comprised of all the nematode species, suggesting that the FABP genes in nematodes may have evolved from different origins. Except S. mediterranea, the phylogenetic relationship within Platyhelminth species was clearly resolved. The subclades comprised of E. multilocularis and E. granulosus demonstrate that both parasites have a similar gene set for FABP, possibly descendent from their common ancestor.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. A Bayesian tree of FABPs.

Bayesian probabilities more than 0.8 were shown at nodes. Tadh: Trichoplax adhaerens; Ctel: Capitella teleta; Hrob: Helobdella robusta; Lgig: Lottia gigantean; Smed: Schmidtea mediterranea; Sman: Schistosoma mansoni; Sjap: Schistosoma japonicum; Egra: Echinococcus granulosus; Emul: Echinococcus multilocularis; Cele: Caenorhabditis elegans; Ppac: Pristionchus pacificus; Hbac: Heterorhabditis bacteriophora; Tspi: Trichinella spiralis; Srat: Strongyloides ratti; Bmal: Brugia malayi; Dpul: Daphnia pulex; Phum: Pediculus humanus corporis; Bmor: Bombyx mori; Tcas: Tribolium castaneum; Nvit: Nasonia vitripennis; Apis: Acyrthosiphon pisum; Amel: Apis mellifera; Dmel: Drosophila melanogaster; Agam: Anopheles gambiae; Aaeg: Aedes aegypti; Cpip: Culex pipiens quinquefasciatus; Rpro: Rhodnius prolixus; Spur: Strongylocentrotus purpuratus; Bflo: Branchiostoma floridae; Csav: Ciona savignyi; Skow: Saccoglossus kowalevskii; Has: Homo sapiens Note: to make it simpler, ‘FABP’ was omitted in every branch name. For example: Tadh1 refers to Tadh_FABP1, Tadh2 to Tadh_FABP2 and so forth.

https://doi.org/10.1371/journal.pone.0077636.g001

Extraordinary gene expansion was observed in amphioxus, B. floridae, via gene duplications. Moreover, the phylogenetic analysis revealed that the current gene set might have resulted from multiple rounds of duplications and divergence during evolution, especially Bflo11 paralogues, and that these duplication events might have occurred recently (Figure 1). Essentially, a ML tree showed a similar topology to the Bayesian tree (Figure S3).

Diversity of FABP gene structures across invertebrates

Although intronless FABP pseudogenes have been described in several species including humans [15–17], all functional mammalian FABP genes exhibit similar genomic organization, containing four exons and three introns [1]. An analysis of FABP gene organization revealed diversity in invertebrates, especially in Platyhelminthes and Nematoda, although the canonical organization (four exons) predominated. A six-exon five-intron structure for FABP was only found in the early-branching invertebrate Trichoplax adhaerens. FABP genes comprised of five exons and four introns were found in placozoans, molluscs, platyhelminths and nematodes (Table 2).

Intronless FABP genes

Single-exon FABP genes were found in the following species: Echinococcus granulosus, E. multilocularis, Strongyloides ratti, P. humanus corporis, Rhodnius prolixus, Strongylocentrotus purpuratus and S. kowalevskii (Table 2). Expression of most of these intronless genes was confirmed either by transcriptome analysis or analysis of ESTs. With the exception of R. prolixus, the species encoding intronless FABPs also encoded other FABP genes with two or more exons. The intronless FABP gene architecture dominated in three other metazoans Strongyloides ratti, P. humanus corporis and Saccoglossus kowalevskii.

Alignment of the intronless FABPs has revealed that the R. prolixus FABP contained intact key residues which are important in defining fatty acid binding [10], while absence or alterations in these sites occurred in the others (Figure S4). This suggests that R. prolixus FABP has capacity to bind to lipids, but the others may no longer be able to do so. Alternatively, they may have different binding spectra for fatty acids in comparison with those that have been characterised.

Alternative splicing in FABP genes

Alternative splicing is an important post-transcriptional modification in eukaryotic pre-mRNAs, accounting for the complexity and variety of proteomes. FABP genes underwent alternative splicing to generate isoforms in several invertebrates including S. mansoni, C. elegans, T. castaneum, Acyrthosiphon pisum, Apis mellifera and Drosophila melanogaster. Furthermore, all the transcripts derived from alternative splicing were confirmed by transcriptomic data.

Comparative analysis of FABP gene structure and transcripts revealed that FABP pre-mRNAs were alternatively spliced in different patterns (Figure 2). In S. mansoni, four FABP variants were produced via exon skipping. Moreover, these variants were differentially expressed at different developmental stages (http://www.genedb.org/Homepage/Smansoni), suggesting they have distinct roles. The arthropods Acyrthosiphon pisum (3 FABP genes) and Apis mellifera (2 FABP genes) utilized the same approach to yield four and three different FABP transcripts, respectively. In contrast to the exon exclusion mechanism seen in S. mansoni, spliced leader trans-splicing (SL trans-splicing), a type of alternative splicing whereby a spliced leader serves as a mini-exon to be added onto 5’ pre-mRNA ends [18], was used in C. elegans FABP mRNA precursors (Figure 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Alternative splicing in invertebrate FABP genes.

Typical alternative splicing patterns in C. elegans, T. castaneum and D. melanogaster are represented. In C. elegans, LBP-9 pre-mRNA is spliced to generate two variants 9a and 9b by addition of a short spliced leader sequence (SL1: 5’-GGTTTAATTACCCAAGTTTGAG-3’) at the 5’ end. Blank or filled boxes and straight lines represent exons and introns, respectively, and a poly (A) stretch present in each FABP cDNA clone or EST sequence is directly shown. In each group, an annotated FABP gene is placed above the variants that are indicated by a, b, c or/and d. Numbers above the boxes and under the lines show the sizes of corresponding exons and introns, respectively. The sizes of exons, where these differ, are indicated above the corresponding exons in the spliced transcripts. The length of variants is also shown after each transcript and the number of ESTs is shown in the brackets.

https://doi.org/10.1371/journal.pone.0077636.g002

Both T. castaneum and D. melanogaster genomes contained only one FABP gene locus but variants were found in transcriptome or EST datasets. Mapping of these transcripts revealed that another alternative splicing mechanism, intron retention, was involved in the post-transcriptional splicing of FABP genes together with exon skipping (Figure 2). Noticeably, the last exons of the FABP genes were retained during alternative splicing in all the invertebrates studied except D. melanogaster that also used the partial sequence of the second intron as the last exon to generate FABP isoforms.