In order to undertake a comparative analysis of the plant Syt-like genes, I collected and compared full-length homologues from an evolutionary range of plants. In order to perform an unbiased search for as many homologues of these relatively unknown genes as possible, I looked at all of the primary nucleotide sequence data in the NCBI sequence databases 19. This information is fragmentary, little of it being in the form of complete sequences, either of transcripts or genomes. By far the most abundant source of new plant sequences are ESTs, but these represent particularly small fragments and their sequences are not determined to high accuracy. I therefore needed to gather sets of overlapping ESTs to find full-length gene sequences. In order to focus the search to the detection of genuinely homologous sequences, I used nucleotide sequence probes of plant sequences already identified. Only those database sequences closely related to the probe sequence would be identified in a given search. These matching sequences were added to the collection and joined to any overlapping sequences already present in the collection. Reiterated searches served to expand the collection and extend the length of gene fragments. Had I used amino acid sequence probes to search for homologues of these genes, I would have detected a wider range of fragments with amino acid similarity, but these would not necessarily be homologous. Overlapping nucleotide sequences would be required in any case, to piece together whole genes from the identified EST fragments, so the simplest strategy to gather full-length relatives of these genes was to use nucleotide probes. I avoided gathering processed sequences in the sequence databases: these include genes predicted from genome annotation pipelines, as well as the vast majority of amino acid sequences which are predicted from nucleotide sequences. These sequences may not be accurate and could mislead subsequent analyses if used without verification.

So I carried out reiterated rounds of blastn searching of nucleotide sequences at NCBI 19. In the first few rounds, I used probes representing the plant gene coding sequences I had already identified (genes 85 to 117) 4. After each round, I collected all of the statistically significant hits with high scoring segments longer than 30 nucleotides and assembled these sequences into a gap4 database 20. Repeated searching with different probes, followed by gap4 assembly of only previously uncollected hits, allowed me to gradually but efficiently build a comprehensive collection. Each probe detected a unique spectrum of homologous plant sequences. Probes from a given species could be used to find similar sequences from related species. Probes covering more conserved regions could be used to find sequences from a wider range of relatives. Sequences from closely related species could be used to bridge non-overlapping contigs from a single species. In the later stages of the collection process, I carefully separated the contigs so that in most cases, each represents a set of overlapping sequences from one species only. As a final step, to ensure that the collection was as comprehensive as it could be at this time, I searched the nucleotide sequences at NCBI using tblastn with amino acid sequence probes and confirmed that the top scoring hits had already been collected.

As well as examining transcript sequences, I also collected genomic sequences where available. I particularly wanted to examine the genome of Physcomitrella patens which is currently being sequenced 21. I had previously identified Syt-like genes in the genome sequences of Arabidopsis thaliana and Oryza sativa but both of these represent relatively recently evolved angiosperms whereas the moss genome represents an ancient bryophyte. I used the trace archive at NCBI 22 as well as resources at PHYSCObase 23 where transcript sequences are also available. I confirmed the genomic and transcript sequences from several Physcomitrella patens gene loci and deposited these sequences in the public databases [EMBL:AM140045, EMBL: AM140046, EMBL: AM140047, EMBL: AM140048, EMBL: AM140049, EMBL: AM140050]. In contrast to animal Syt genes, which appear to increase in number along with organism complexity 4, I found that the haploid genome of Physcomitrella patens has even more of these plant genes (19 or more) than either Oryza sativa (13) or Arabidopsis thaliana (11). Additional file 1 lists full details of each gene identified. Additional file 2 lists alphabetically, in rough phylogenetic order, all of the plant species in which genes in this collection have been identified. Genes were identified in a wide evolutionary range of land plants, from bryophytes to rosids.

Database searching identified six distinct groups of plant genes. Since all of the genes encode relatively long proteins, most of the collection comprises gene fragments which cannot yet be extended to full-length. Only where a large number of overlapping sequences were available was it possible to derive full-length gene sequences from EST contigs. Consequently, the full-length sequences represent the relatively abundantly transcribed, or the shorter genes. Genomic sequences were useful for identifying full-length sequences, irrespective of transcript abundance, as well as for providing the intron-exon structure of the gene. Full-length amino acid sequences were compared using Multalin 24. The previously used nomenclature (SytA, SytB, SytC etc.) following 14 is somewhat arbitrary and is inadequate for a consistent and meaningful description of these plant genes. I propose the following naming convention for these plant N-terminal-TM-C2 domain genes: NTMC2Type1.1, NTMC2Type1.2, NTMC2Type6 and so on. Multiple alignments of full-length sequences from each group are presented in figures 1, 2, 3, 4, 5, 6

Figures to 1, 2, 3, 4, 5, 6 show the overall domain pattern common to all of the genes: the N-terminal region, TM region, linker, C2 domain region and C-terminal region. Strongly conserved intron patterns, as well as distinctive patterns of sequence conservation, distinguish the six types of NTMC2 genes. The six groups are not entirely homogeneous. Physcomitrella patens NTMC2Type2.3 for example, while sharing its bulk with the other members of the NTMC2Type2 group, has different N and C termini and lacks the second C2 domain. The NTMC2Type2 group is also notable in that some of its members are RNA edited (see figure 2 and full details in additional file 1). In some members of this group, the genomic sequence of the second coding exon lacks one nucleotide at its 3' terminus, resulting in a faulty, frameshifted gene. However, these genes are still able to produce functional transcripts with the missing guanosine restored. Transcripts for both Arabidopsis thaliana NTMC2Type2 genes, and the Oryza sativa NTMC2Type2.2 gene are edited in this way. Transcript sequences have not yet been deposited in the sequence databases for the Physcomitrella patens NTMC2Type2.1 and the Medicago truncatula NTMC2Type2.2 genes, but I have assumed that they are similarly edited. The genomic loci of the Physcomitrella patens NTMC2Type2.2 and NTMC2Type2.3 genes and the Oryza sativa NTMC2Type2.1 gene, do not lack the equivalent nucleotide, and are not frameshifted. The genomic locus of the Medicago truncatula NTMC2Type2.1 gene lacks the equivalent exon-intron boundary altogether and is not frameshifted. The first coding exon of the Medicago truncatula NTMC2Type2.1 gene is equivalent to a fusion of the first three coding exons of the NTMC2Type2 genes mentioned above, with the corresponding two introns missing. The frameshift error thus appears to be associated with a particular intron. Other examples of divergent members of a group are Physcomitrella patens NTMC2Type4.3, which diverges at its C terminus and Physcomitrella patens NTMC2Type5.2 and NTMC2Type5.3, which have a different intron pattern. Group 6, as a whole, is not well conserved C-terminal of the C2 domain.

I had previously identified genes in metazoans and non-metazoans which encode N-terminal-TM-C2 domain proteins sharing similarity with those of plants 4. In the meantime, with the annotation of the human genome, the three members of this gene family in Homo sapiens have been named FAM62A, FAM62B and FAM62C 25. I sought to identify homologues of these genes in other organisms by tblastn searching genomic sequences, thereby identifying full-length genes and their intron-exon structures. In contrast to the current status of primary nucleotide sequences from plants, many more animal genomic sequences are available to search. One reason for this is that animal genomes are relatively small in comparison to plant genomes and are therefore relatively less expensive to sequence. After identifying FAM62 gene homologues in genomic sequences, I searched transcript sequences using blastn with nucleotide probes, to confirm the predicted gene structures. I identified FAM62 homologues in a range of metazoan genomes. Details of each gene are listed in additional file 3.

Full-length amino acid sequences were compared using Multalin 24. Figure 7 shows a multiple alignment of the three FAM62 gene products from Homo sapiens. All three share a common gene structure, but while FAM62B and FAM62C each encode three C2 domains, FAM62A contains a repeat of the portion of the gene encoding the first two C2 domains, resulting in a total of five C2 domains. Figure 8 shows a multiple alignment of four FAM62 gene products from Danio rerio. The genome of Danio rerio encodes at least four FAM62 genes. All four share a common gene structure, but while FAM62B and FAM62C each encode three C2 domains, the FAM62A homologues contain additional repeats of the module which encodes the first two C2 domains. This results in a total of five C2 domains in FAM62A1 and nine C2 domains in FAM62A2.

Figure 9 shows a multiple alignment of FAM62 gene products from a range of metazoans. The genomes of Drosophila melanogaster, Anopheles gambiae, Apis mellifera, Strongylocentrotus purpuratus, Ciona intestinalis and Caenorhabditis elegans each appear to encode one FAM62. The genome of Tribolium castaneum has an unusual and compact FAM62 locus. It is approximately 12 kilobases long and contains three closely spaced FAM62 copies in tandem. Only the first copy retains the intron pattern common to other FAM62 genes and is shown in figure 9. The other two copies have diverged from the first and from each other, both in terms of amino acid sequence and intron position (see figure 10 and further details in additional file 3). A duplicated, alternative exon in the region of the first C2 domain of the insect FAM62 genes is shown boxed in figures 9 and 10. This is reminiscent of alternative readings of the C2B region of certain Syt1 genes 32627. In the Syt1 cases, the insect Syt1 genes employ RNA editing to alter this region, while Caenorhabditis elegans and Aplysia californica encode duplicate alternative exons like the insect FAM62 genes here. The duplicated alternative exons are absent from the second and third FAM62 copies in Tribolium castaneum (figure 10). Vertebrate genomes encode at least three FAM62 genes. The FAM62B genes of vertebrates appear to be most similar to the single FAM62 genes of other organisms and are therefore included in figure 9. Alternatively spliced regions of Homo sapiens and Mus musculus FAM62B are shown boxed as well as some alternatively spliced regions of Ciona intestinalis FAM62.

Collection and analysis of the plant NTMC2 genes and animal FAM62 genes revealed intron patterns which are highly conserved within the different groups, implying a long evolutionary history for the whole length of each gene. I have previously looked at the intron patterns of Syt genes and found strong conservation of particular intron positions 34. To make clear the differences between the plant and animal N-terminal-TM-C2 domain genes and Syt genes which are also N-terminal-TM-C2 domain genes, I analyzed the intron positions within the coding regions of Syt genes from a wide a range of metazoans. Details of Syt genes shown here but not previously reported 4 are in additional file 4.

Figure 11 shows an overview of the intron patterns in Syt genes. Intron positions and their phases are shown relative to TM, C2A and C2B domains. The conserved introns between the C2A and C2B domains stand out clearly. I have included Syt17 (also known as B/K 28) homologues here. Although Syt17 homologues lack the N-terminal TM domain and were therefore excluded from my previous analysis 4 their intron structure is indeed characteristic of Syt genes and different from other Syt-like genes, such as those encoding Doc2 and Rabphilin proteins (figure 12, details in additional file 4). The HUGO gene nomenclature committee 25 have agreed to name the Homo sapiens gene locus SYT17 so I follow this nomenclature here. The finding of a Syt9 homologue in Strongylocentrotus purpuratus expands beyond vertebrates a group of Syt genes (Syt3, Syt6, Syt9 and Syt10) previously seen only in vertebrates. I have identified additional Syt genes in genomes examined previously. The Ciona intestinalis Sytα (following the nomenclature of 9) is a previously unidentified member of a group present in Caenorhabditis elegans, Drosophila melanogaster, Anopheles gambiae and Ciona intestinalis but not present in Strongylocentrotus purpuratus, Danio rerio or Homo sapiens. The Danio rerio genome sequence is still being completed and has yielded substantially more information since my last analysis 4.

In figure 11, I have arranged the Syt genes into groups of likely orthologues and paralogues. Genes from different species, which are more similar to each other than to other genes from the same species, can be classed as orthologues, and thus defined, are taken to be related by vertical descent from a common ancestor 29. The functional implications of such a relationship are that orthologues may fulfil similar, perhaps equivalent, roles in different species. As mentioned in the Background section of this paper, this may be broadly true for Syt1 genes which appear to be present in all animals. The intron pattern distinctive of Syt1 genes, is highly similar to the intron patterns of the Syt2, Syt5 and Syt8 genes. These genes appear only in the evolutionarily more modern vertebrate lineages, so it is likely that they have arisen via Syt1 duplication during the evolution of vertebrate lineages and could therefore be classed as paralogues, relative to Syt1. The functional implications of such a relationship are that paralogues may fulfil a subset of the roles of the parent orthologue through a process of subfunctionalization, or acquire new roles through a process of neofunctionalization 29. The Syt11 genes appear similarly related to the Syt4 group and the Syt14 genes similarly related to the Syt16 group. The Syt6, Syt10 and Syt3 genes also appear similarly related to the Syt9 group. Until a more complete picture emerges from the accurate identification of complete genome complements of Syt genes and Syt-like genes from many more eukaryotic lineages, it will not be possible to classify these genes more accurately as orthologues and paralogues.

Physcomitrella patens genomic DNA was a gift from Didier Schaefer. I used this as a template for PCR reactions. I amplified genomic regions using Pfu turbo polymerase with phosphorylated primers and cloned the products into Sma digested pBSIIKS-. After sequencing, overlapping clones were selected and digested with restriction enzymes in such a way as to ligate the genomic locus into one piece. The sequence of each genomic clone was deposited in the public sequence databases [EMBL:AM410046, EMBL:AM4100449, EMBL:AM410050]. cDNA clones, also gifts from Didier Schaefer, were obtained from the M. Hasebe collection 44 at PHYSCObase 23 and sequenced completely. These sequences were deposited in the public sequence databases [EMBL:AM410045, EMBL:AM410047, EMBL:AM410048].

A full-length cDNA clone of Arabidopsis thaliana NTMC2Type2.2 was a gift from Boris Voigt. I confirmed the coding sequence and deposited this in the public sequence databases [EMBL:AM410051].