Expansion of the Actin Gene Family in Amphioxus

: Actins are a small family of ubiquitous proteins that are essential cytoskeletal components and are highly conserved during evolution. Actins are usually divided into two classes, the cytoplasmic and muscle actins, which have different functional roles. Here we systematically analyzed the actin genes in the genome of the primitive chordate amphioxus ( Branchiostoma floridae ). We found that amphioxus contains more than 30 actin genes, many of which are linked. Phylogenetic analysis suggests the amphioxus actin genes have clearly undergone extensive expansion through tandem duplications. The actin genes’ structure also varies a lot, containing 2 to 7 exons. We also cloned two muscle type of actin genes from the amphioxus （ B. belcheri ） and compared their expression patterns during early development. The slight difference in their expression suggests functional diversification of these actin genes. Our results shed light on the evolution both of actin genes themselves and their functional roles in chordate development.

cytoplasmic and muscle actin are different at about 20 amino acid positions, which serve to distinguish the two types of actins (Vandekerckhove & Weber 1979, 1984Kovilur et al, 1993). Also, the cytoplasmic and muscle actin genes have different intron-exon structures in mammals.
The actin genes in the primitive chordates, urochordates and cephalochordates, have been subjected to several studies (Bovenschulte & Weber, 1997;Kusakabe 1995Kusakabe , 1997Kusakabe et al, 1997aKusakabe et al, , b, 1999Suzuki & Satoh, 2000). The amphioxus, which is thought to be the closest living relative of the vertebrates, has been a model system to study the evolution and mechanisms of development in vertebrates (Holland et al, 2004). The Florida amphioxus (Branchiostoma floridae) cytoplasmic and muscle actins are clearly distinguishable and closely related to vertebrate cytoplasmic and muscle actins, respectively (Kusakabe et al, 1997b). The cytoplasmic actin gene BfCA1 is widely expressed while the muscle actin BfMA1 is expressed strongly in somatic muscle and weakly in smooth muscle associated with gill slits. Also, southern blot analysis of genomic DNA has suggested that the amphioxus genome contains multiple muscle and cytoplasmic actin genes that might have arisen from recent duplication and gene conversion.
Recent results indicate that vertebrate actin isoforms evolved after the separation of the vertebrates and urochordates (Kusakabe, 1995;Kusakabe et al, 1997a). It has also been suggested that the actin genes have arose independently in each of the three chordate lineages.
With the nearly complete genome sequencing of the Florida amphioxus (B. floridae, Putnam et al, 2008), we can systematically study the actin genes in this primitive chordate species. We found that the amphioxus contains more that 30 actin genes, many of which are linked.
Phylogenetic analysis suggests that the amphioxus actin genes have clearly undergone expansion through tandem and clusteral duplications. The comparative analyses of the actin genes' structure suggest a model of the evolution of these genes. We also cloned two muscle type of actin genes from the amphioxus and compared their expression patterns during early development. The slight difference in their expression suggests functional diversification of these actin genes. Our results shed light on the evolution both of actin genes themselves and their functional roles in chordate development.

Blast searches and phylogenetic analysis
The coding region of Branchiostoma floridae muscle actin gene (BfMA1, GenBank accession number D87407) was used to blast the B. floridae genome (http://genome.jgi-psf.org/Brafl1). The corresponding regions were extracted and the gene structures were analyzed using Genomescan (http://genes.mit.edu/ genomescan.html) and manual corrections. A phylogenetic tree was constructed by the neighborjoining method using the MEGA4 program (Zhao et al, 2007).

Cloning of Branchiostoma belcheri muscle actin genes and in situ hybridization
Adult amphioxus (B. belcheri) were collected during the breeding season from the South China Sea and maintained in the laboratory. Naturally fertilized eggs were collected and cultured at room temperature. The 3' partial sequences plus the 3' untranslated regions of two B. belcheri muscle type actin genes were cloned using 3' RACE. One corresponds to the published BfMA1 (Kusakabe et al, 1999), and a second one corresponds to a new muscle actin gene (corresponding to the 23-9 gene, see below) whose corresponding genomic sequence is available in B. belcheri. The gene specific primers used were: 5'-ACATCCGTAAGGACCTGTACGCCAAC-3' (BfMA1) and 5'-GATCGTCAGCAGAAGGAGATCA CCTT-3' (23-9). Total mRNA of 36-hour larvae was used as template for the reverse transcription. The PCR products were then cloned into pGEM-T vector, sequenced and used to prepare probed for in situ hybridization. In situ hybridizations were carried out as described (Zhang & Mao, 2009).

The actin genes in amphioxus genome
Taking advantage of the nearly completed actin homologous genes. We made blast searches using the published B. floridae muscle actin (Kusakabe et al, 1997b) coding region as the query. In total 11 scaffolds were found to contain actin homologous sequences. The corresponding regions were extracted and analyzed by Genomescan and manual corrections. Surprisingly, 37 genes were predicted, 27 of which were full length and the rest contained gaps in the sequences (Fig.1, Tab. 1).
Most of them (31 out of 37) are organized into 5 gene clusters (Fig.1). The genes are referred to according to their locations on the scaffolds (Fig.1, Tab. 1). We also found that scaffold 264 (16 kb) and partial sequence of scaffold 226 (8 kb, containing genes 226-1 to 226-3) show a high similarity with sequences in scaffold 231.
We can not rule out the possibility that they might represent different alleles of the same cluster, or they can result from recent cluster duplications. Taking into account also the possibility of miss-assembly of the sequences, the confirmation of the exact number of actin genes in amphioxus still awaits the improvement of the sequence data. Phylogenetic analysis (Fig.2) of the 27 predicted full length actin proteins showed that they fell into three different groups: the muscle type, the cytoplasmic type and those that showed similar homology to both muscle and cytoplasmic types. In addition to the reported cytoplasmic actin (Kusakabe et al, 1997b, corresponding to gene 68), a cluster of 4 genes (on scaffold 207) was also clustered into the cytoplasmic type. They likely represent a new type of cytoplasmic actin.

The structure and evolution of the actin genes
The amphioxus muscle and cytoplasmic actin genes have been reported to contain 4 and 5 exons respectively.
Interestingly, genes with different structures also exit.
5/31 contain 6 exons, two of them have only two exons, and one gene even contains 7 exons (Tab. 2). Most of genes locate in the same scaffold have the same structure ( Fig.1, Tab. 2), strongly suggesting that tandem  Annotated hypothetical proteins and genes in the GenBank identical or close to our predicted genes were listed.

Tab. 1 The predicted actin genes in the Branchiostoma floridae genome
1, The hypothetical proteins in the database contain less or more amino acids than our predicted version.
2, The hypothetical protein 126063 is a fusion version of 231-6 and -7.
duplication was the major means of expansion of the actin gene family in amphioxus. Since most of the intron positions (42,123,206,270,Tab. 2) are conserved across species (Kusakabe et al, 1999), even in invertebrate actin genes, we prefer the model that the ancestor actin gene to be multi-exon. The actin genes contain only two exons might arise due to intron loss during evolution. Since intron gain and loss happen heavily during evolution, it is not easy to reconstruct the evolutionary history of the amphioxus actin genes according only to their structures.

Tab. 2 Comparison of intron positions in amphioxus actin genes
23-9 at larval stages. At early larva stage (24 h

Discussion
Actins are very well conserved at protein level during evolution. Here we showed that the amphioxus actin genes have clearly undergone rapid lineage specific expansion and contain more than 30 actin genes. Some of them fall clearly into the muscle or cytoplasmic types, while the rest show similar homology to both muscle and cytoplasmic actins. We also extracted the 1kb sequences upstream of the start codon and 300 bp 3' untranslated region (UTR) sequences of the actin genes and tried to construct phylogenetic trees using these sequences.
However, the result showed that the 3' UTR sequences have become quite divergent and the tree was no longer reliable (data not shown). The 5' 1kb regions contain the 5' UTR sequences and the potential promoter regions.
Comparisons of these sequences revealed partial conservation of some of the sequences (data not shown).
Detailed analysis of the potential promoter regions of these genes might help to resolve their transcriptional regulation mechanisms. It will be of interest to check systematically the expression patterns of the different actin genes to follow their potential functional divergences. In sea urchins, the actin gene family has been also reported to be rapidly evolving whose members are expressed in distinct cell lineages in a developmentally regulated fashion ( Kissinger et al, 1997). Since they are usually clustered, the amphioxus actin gene family is also a good system to study gene evolution and the evolutionary relationships between available amphioxus species. It will be of interest to check the gene number, site of expression, and gene linkages of the actin gene family in different amphioxus species.