Chromosomal mapping, differential origin and evolution of the S100 gene family

S100 proteins are calcium-binding proteins, which exist only in vertebrates and which constitute a large protein family. The origin and evolution of the S100 family in vertebrate lineages remain a challenge. Here, we examined the synteny conservation of mammalian S100A genes by analysing the sequence of available vertebrate S100 genes in databases. Five S100A gene members, unknown previously, were identified by chromosome mapping analysis. Mammalian S100A genes are duplicated and clustered on a single chromosome while two S100A gene clusters are found on separate chromosomes in teleost fish, suggesting that S100A genes existed in fish before the fish-specific genome duplication took place. During speciation, tandem gene duplication events within the cluster of S100A genes of a given chromosome have probably led to the multiple members of the S100A gene family. These duplicated genes have been retained in the genome either by neofunctionalisation and/or subfunctionalisation or have evolved into non-coding sequences. However in vertebrate genomes, other S100 genes are also present i.e. S100P, S100B, S100G and S100Z, which exist as single copy genes distributed on different chromosomes, suggesting that they could have evolved from an ancestor different to that of the S100A genes.


INTRODUCTION
S100 proteins constitute the largest gene family within the EF-hand protein super-family. In 1965, Moore isolated from bovine brain the first protein members of the S100 family: S100A1 and S100B [17]. In the following years, many other members of the S100 family were identified based on sequence homology and similar structural properties. For example, the human S100 family includes 20 members, which share 22% to 57% sequence identity [13]. S100 proteins are small acidic proteins (9-14 kDa) and contain two distinct EF-hand motifs. The C-terminal EF-hand contains a classical Ca 2+ -binding motif, common to all EF-hand proteins while the N-terminal EF-hand differs from the classical EF-hand motif and constitutes a special characteristic of the S100 proteins. S100 proteins exhibit a unique pattern of tissue/cell type specific expression and exert their intracellular effects by interacting with different target proteins that modulate their activity [5,23,31]. Two well-known pairs are S100A11annexin A1 and S100A10-annexin A2 [9,20,24,25,27] and recently, interaction between S100A11 and annexin A6 has also been reported [3]. Until now, over 90 potential target proteins have been identified [23]. Many studies have observed an altered expression of various S100 proteins in a large number of diseases including cancer, depression, Down syndrome, Alzheimer disease and cystic fibrosis [1,13,14,26,28,29]. Therefore, S100 proteins could constitute important diagnostic markers as well as therapeutic targets of many diseases.
All known S100 genes are found only in vertebrates and no S100-like sequences have ever been detected in invertebrates such as insects, nematodes and protozoa based on the analysis of available genome sequence information. This suggests that the genes encoding S100 proteins belong to a ''young'' gene family i.e. that originated during vertebrate evolution. Interestingly, because of the short phylogenetic history and the conservation of the S100A gene cluster in man and mouse [21], their origin in the vertebrate lineages remains a challenge. Moreover, in non-mammalian systems such as fish species, information on the S100 gene family evolution and genomic organisation is very scarce and only a few S100 gene members have been identified [7]. In this work, we analysed S100 gene sequences of various vertebrates including mammals and fish from available databases using both comparative genomics and phylogenetic methods, and we present a model of the molecular evolution of the S100 genes, which contributes to a better understanding of the mechanisms of evolution and biological functions of the S100 gene family.

Sequences and positions on the chromosomes or assembly scaffolds
A search in the GenBank and Ensembl databases (v39) provided 118 sequences of the S100 gene family from seven mammals whose genomes have been sequenced. In addition, using human S100 gene sequences as query sequences, orthologous sequences were found for three teleost fish, Danio rerio, Takifugu rubripes (Japanese pufferfish), Tetraodon nigroviridis (freshwater pufferfish). The complete list of the S100 mammalian and fish sequences compiled in this study together with gene names and accession numbers are given in Table I.  Homo sapiens S100A1 NP_006262 Pan troglodytes S100a11 ENSPTRG00000001303 S100A2 NP_005969 S100a12 ENSPTRG00000001352 f136068 NEWSINFRUG00000136068 S100 gene evolution Mus musculus S100b ENSMUSG00000033208 S100a3 ENSMODG00000017395 S100g ENSMUSG00000040808 S100a4 ENSMODG00000017397 S100z ENSMUSG00000021679 S100a5 ENSMODG00000017400 ENSMODG00000002897 Rattus norvegicus S100a1 ENSRNOG00000012410 S100z ENSMODG00000019747 S100a3 ENSRNOG00000012008 Mus musculus S100a1 ENSMUSG00000044080 S100a4 ENSRNOG00000011821 S100a3 ENSMUSG00000001021 S100a5 ENSRNOG00000011748 S100a4 ENSMUSG00000001020 S100a6 ENSRNOG00000011647 S100a5 ENSMUSG00000001023 S100a8 ENSRNOG00000011557 S100a6 ENSMUSG00000001025 S100a9 ENSRNOG00000011483 S100a8 ENSMUSG00000056054 S100a10 ENSRNOG00000023226 ENSBTAG00000004777 S100g ENSCAFG00000012583 S100g ENSBTAG00000017020 S100z ENSBTAG00000020201 a Codes of fish genes were defined by authors. S100 gene evolution The chromosomal localisation of these genes is based on the Ensembl v39 genomic location data.

Gene prediction
In order to detect sequences that may contain unknown S100 sequences, genomic sequences were aligned with the exons of homologous human genes by Vector NTI software and those identified were assembled into putative mRNA sequences. These mRNA sequences were translated into protein sequences, which were aligned with the corresponding human proteins to test the validity of the prediction.

Sequence alignment and construction of phylogenetic trees
Multiple alignments were performed with the Vector NTI software and Neighbour-Joining phylogenetic trees were built using the Phylip program (Joseph Felsenstein, Washington University). The reliability of the trees was measured by bootstrap analysis with 1000 replicates and the trees were edited and viewed by Treeview software.

RESULTS
3.1. Mammalian S100A genes are duplicated and clustered on one chromosome The chromosomal organisation and location of the S100A genes identified in seven mammalian species i.e. man, chimpanzee, cow, dog, rat, mouse and opossum were determined using the Ensembl database. The results revealed that in each of these seven mammals the S100A genes are clustered on a single chromosome and comprise up to 16 members ( Fig. 1 and Tab. I). Although these genes are located on a single chromosome, two subgroups (SGs) were identified: SG1 in which S100A10 and S100A11 are always tightly linked and SG2 in which other members (S100A1-9 and 12-16) are generally clustered together (Fig. 1). The distance between the two SGs covers several megabases, whereas only a few kilobases separates genes within each SG. Interestingly, the relative positions of the genes on the chromosomes are conserved among these mammalian species, which indicates a high level of conserved synteny (Fig. 1). In addition, other putative S100A gene members, previously unknown, were predicted from available genome sequence data based on information of conserved synteny and protein homology. Five genes were identified, S100A3 and S100A14 454 in the cow, S100A12 in the dog and S100A2 and S100A14 in the rat (Fig. 2 and Tab. II). Multiple protein sequence alignments with the corresponding human S100A proteins showed a high level of homology (Fig. 2). Thus, these sequences are not pseudogenes and corresponding expressed sequence tags (EST) are present in the EST databases (for details see legend of Fig. 2).
Differences in the arrangement of the S100A genes were observed between the opossum and the other species examined, i.e. SG1 (S100A10 and 11) together with S100A1 is located at the 3 0 end of opossum chromosome 2 and at the 5 0 end of the corresponding chromosomes in the other species (Fig. 1). Also, in the opossum, the positions of S100A9 and S10012 are reversed comparatively to those in the other species. These discordances indicate that chromosomal rearrangements having occurred during mammalian speciation have disrupted the syntenic gene associations.

Two clusters of S100A genes in teleost fish
A phylogenetic tree was constructed to determine accurate predictions of orthology and paralogy relationships between fish and mammalian S100A genes (Fig. 3a). Fish S100A proteins are divided into two SGs as defined in Figure 1. SG1 includes S100A10 and S100A11 genes while SG2 contains all the other S100A genes. This distribution is supported by the data on gene organisation Figure 1. Conserved synteny and subgroup (SG) definition of the S100A gene cluster in mammals. The S100A genes from different mammalian species are clustered on a single chromosome and are divided into two subgroups (SG1 and SG2) based on their relative localisation on the chromosome. The gene distribution was analysed from data in the Ensembl database (http://www.ensembl.org). S100A1-16 genes are indicated as two blocks of synteny by two colour boxes. Dashed boxes indicate the predicted genes. The name of the species and chromosome numbers are shown on the left. S100 gene evolution for available fish genome assembly scaffolds and human chromosome 1 (Fig. 3b) although in some cases, gene members are only temporarily positioned on the scaffolds and their definite chromosome localisation needs to be confirmed. Seven zebrafish genes classified in the S100A category form two clusters on chromosome 16 and chromosome 19, respectively. Among the nine takifugu genes belonging to the S100A category, at least six form two clusters on scaffold 37 and scaffold 252, respectively. Furthermore, in tetraodon, a similar gene arrangement exists with four genes clustered on chromosome 21 and two other genes clustered on chromosome 8. Interestingly, in each synteny group, gene members of both SGs 1 and 2 are present. Thus overall, these results based on phylogenetic and comparative genomic analyses show the existence of two S100A gene clusters in fish genomes and only one in mammalian genomes. Five S100A predicted genes based on conserved synteny and homology. Predicted genes include bovine S100a3 (complete CDS) and S100a14 (partial CDS), rat S100a2 (partial CDS) and S100a14 (partial CDS) and dog S100a12 (complete CDS). The multiple sequence alignments with the corresponding human S100 proteins are shown in the centre to confirm the identity of predicted genes. Two EST sequences (GenBank Accession Nos. XM_001063574 and NM_001079634) are similar to rat and bovine S100a14, especially in the CDS regions. More information is necessary to confirm that the two sequences correspond to gene S100a14. Two other EST: DR104796 (canine cardiovascular system biased cDNA, a Canis familiaris cDNA similar to that of Hs S100 calcium-binding protein A12) and DV924106 (Bos taurus cDNA clone IMAGE: 8232591 5 0 , mRNA sequence) may be the relevant bovine and rat genes, S100a3 or S100a2, respectively. 456 X. Shang et al.

Presence of other single copy S100 genes scattered in vertebrate genomes
Four other S100 genes i.e. S100P, S100B, S100G and S100Z are present in the human genome and contrarily to the S100A genes clustered on chromosome 1 they are distributed on different chromosomes. A similar distribution pattern of the homologous genes is found in the genomes of the chimpanzee, cow, dog, rat, mouse and opossum. The absence of gene S100P could be due to the incomplete genome sequencing e.g. in the cow and the fish species examined here or to loss of the corresponding sequences during speciation e.g. in the mouse and rat (Fig. 4). Unlike the S100A genes, S100P, B, G and Z genes also exist as single copies in the three fish genomes according to the phylogenetic analysis.

DISCUSSION
We analysed all available information on S100 genes in seven mammalian and three fish species and we determined their phylogenetic relationship and genomic organisation based on abundant sequence resources in databases. Until now, S100 proteins have been detected only in vertebrates, suggesting that they first appeared during vertebrate evolution. In the mouse and man [21], it has been previously shown that all S100A genes are present on a single chromosome but form two SGs, which agrees with our results on their genomic organisation and chromosomal localisation in other mammalian species i.e. the cow, dog, chimpanzee, rat and opossum (Fig. 1). We identified five new previously unknown S100A genes [18]. The structure of mammalian S100A genes is also highly conserved, generally, comprising three exons separated by two introns with the first exon untranslated [6]. The clustered localisations on a single chromosome, the highly conserved synteny and the similarity in exon/intron organisation suggest that gene duplication is responsible for the major expansion of this gene family. Figure 3. Analysis of the phylogenetic relationships and chromosome mapping of S100A genes in mammals and fish. (a) Phylogenetic tree of S100A proteins. The numbers on the branches represent the bootstrap values from 1000 replicates obtained using the (N-J) method. The tree shows two major subgroups of S100A proteins as in Figure 1. (b) Localisation of S100A genes on chromosomes or assembly scaffolds. At least two clusters are observed in fish species but only one in man. Genes are in the boxes and chromosome or scaffold numbers are shown at the top of each linkage group or gene. z09878 is an S100 gene member, ictacalcin previously identified in zebrafish [7]. 458 X. Shang et al.
Furthermore, we analysed the organisation of S100A genes in three fish model species: zebrafish, takifugu and tetraodon. The phylogenetic tree shows that in these fish species the S100A genes are also subdivided into two major SGs as observed in mammalian species. However, in contrast to the existence of a single cluster in mammalian genomes, at least two clusters are present in fish genomes (Fig. 3). A comparison of the genomic architecture and arrangements between fish and mammalian S100A genes shows that they are remarkably consistent with the occurrence of the fish-specific genome duplication (FSGD or 3R) during vertebrate evolution. More and more studies propose that, during the evolution of vertebrates, two rounds (2R) of genome duplication occurred first and then later in the stem lineage of ray-finned fishes, not belonging to land vertebrates, a third genome duplication occurred (FSGD or 3R) [4,10,16]. Indeed, duplicated chromosomes and duplicated S100A genes are present in zebrafish i.e. chromosomes 16 and 19, in tetraodon i.e. chromosomes 8 and 21, and in takifugu i.e. scaffolds 37 and 252. In fact, previous studies have reported that tetraodon chromosomes 8 and 21 and zebrafish chromosomes 16 . Phylogenetic tree and distribution of other S100 proteins in vertebrates. Mammalian homologous genes were found in NCBI and Ensembl databases. Fish genes were identified by searching the paralogue of the corresponding human S100 gene. (a) Phylogenetic tree of S100B, S100G, S100P and S100Z proteins. The numbers on the branches represent the bootstrap values (%) from 1000 replicates obtained using the N-J method. Eight fish genes are classified into the S100B, S100G and S100Z subgroups. (b) Distribution of all known S100B, S100G, S100P and S100Z genes from seven mammals and three fish species. Chromosome numbers (for mammals) and chromosome/scaffold numbers with gene names are indicated in boxes (SF = scaffold, Un = unknown). S100 gene evolution and 19 originate from a common ancestral chromosome L. Furthermore, a high degree of conserved synteny between individual tetraodon chromosomes and zebrafish linkage groups has been observed and suggests a 1:1 chromosome correspondence in both species [8,30]. After the FSGD, interchromosomal rearrangement events (including chromosome fissions, fusions and translocations) probably occurred [10], which would explain our observations that duplicated S100A genes are asymmetrically distributed and that the gene positions in the two clusters are a little different.
We suggest that a single ancestral S100A gene was duplicated and led to the two gene member types defined as SG1 and SG2 during the 2R genome duplication event about 450 Myr (million years) ago. Then, fish genomes (e.g. zebrafish, tetraodon and takifugu) underwent FSGD (3R) and during fish speciation, two clusters of S100A genes appeared on two chromosomes about 350 Myr ago. In mammalian species, because of the absence of a 3R, only one cluster of S100A genes included in SGs 1 and 2 is present on a single chromosome. However, to adapt to diverse environmental conditions, mammals acquired multiple S100A genes by tandem gene duplications within the cluster on the one chromosome (Fig. 5) as, for example, the five copies of human gene S100A7 Figure 5. Model of the molecular evolution of S100A genes. The genomic architecture of fish and mammalian S100A genes is shown. The ancestral S100A gene was duplicated and formed two members defined as SG1 and SG2 during the 2R genome duplication about 450 Myr ago. Then, fish genomes (e.g. zebrafish, tetraodon and takifugu) underwent FSGD (3R), which generated two clusters of S100A genes on two different chromosomes about 350 Myr ago. Other rearrangements also took place during this process. Mammalian (e.g. human) S100A gene members have increased only by gene duplications on a single chromosome since a third round genome duplication (3R) did not occur in mammals. (S100A7a-S100A7e) present at the same locus [11,18]. These duplicated genes may have been retained in the genome by neofunctionalisation and/or subfunctionalisation mechanisms [12] or may lead to pseudogenes, such as S100A7d and S100A7e [18]. However, some genes have either not been duplicated or have been lost during speciation, for example, S100A2, A7 or A12, which are not found in the mouse or in the rat, respectively [18].
In the case of the S100P, B, G and Z genes, the situation is different to that of the S100A genes. In vertebrate genomes, these genes are scattered on different chromosomes and exist as single copies in both mammalian and fish species. This suggests that they could have evolved from an ancestral gene different to that of the S100A genes. Differences in the mode of their interaction with target proteins support this hypothesis. Data on the crystal structure and protein interactions show that the structures of the S100A10/annexin A2 [19] and S100A11/annexin A1 [20] complexes are alike. However, the S100B protein can form a complex with a peptide derived from the C-terminal regulatory domain of p53 [22], or a TRTK-12 peptide existing in CapZ [15], or a peptide derived from Ndr-kinase [2] and the comparison of the structures of these complexes reveals differences in the orientation of the three peptides and in the type of interaction patterns with S100B protein. Moreover, the structure of the S100A10/annexin A2 or S100A10/ annexin A1 complexes is different to that of all S100B/peptide complexes. These differences in structure indicate a large diversity of S100A and other S100 genes. However, Marenholz has previously reported that S100B, P and Z genes are evolutionarily related to gene S100A1, which might point to a common ancestor of the S100 gene family [13]. More information, i.e. whole genome comparisons with other fish species, is necessary to determine whether these two groups of S100 genes have evolved from different ancestors or a common one. The analysis presented here is based on the current information available for whole genome sequences in public databases. Data on whole genome sequences increase daily and contig assemblies are frequently updated. With the completion of the current genome projects and the beginning of future genome projects of other vertebrate model systems new information will be provided, which will help understand the evolution and function of the S100 gene family.