We have previously described the isolation and sequencing of the EMS1 cDNA 2849 (DDBJ/EMBL/GenBank Accession No. M98343) coding for the human cortactin protein. To evaluate the genomic structure, we determined the exon/intron-boundaries. Nucleotide sequence comparisons with human EMS1 cDNA sequence revealed homology with two human genomic clones (DDBJ/EMBL/GenBank Accession No. AP000487 and AP000405) (Table 1). The genomic structure of the EMS1/cortactin gene was determined by performing BLASTn comparisons of EMS1 cDNA against the genomic clones (Figure 1A). By amplifying the intron sequences (smaller than 2 Kb) using primers on adjacent exons followed by end-sequencing of these products, we confirmed the intron/exon boundaries of the human EMS1/cortactin gene. The EMS1 gene contains 18 exons spanning over about 38 Kb of genomic DNA. The length of the individual exons ranges from 55 to 178 bp, except the last exon (1564 bp). The splice donor and acceptor sequences, the sizes of the introns and exons of the human EMS1/ cortactin gene are provided in the supplementary materials [see Additional file 1]. The ATG is at position 169, at the first nucleotide of exon 3, indicating that the first two exons encode the 5' untranslated region (UTR). The F-actin-binding repeat domain is encoded by exon 5 to exon 12 with 5 exons of 111 nucleotides in length (exons 6, 8, 9, 10 and 11) (Figure 1A and [see Additional file 1]). The sequence encoding the DDW Arp2/3 binding site is located within exon 3 and the SH3 domain is encoded by exon 17 and 18. The 3' UTR is 1420 nucleotides in length with a polyadenylation signal AATAAA at position 3225.

Other cortactin homologues have been reported in mouse 3, rat 19, chicken 50, fruit fly (Drosophila melanogaster ) 51, and frog (Xenopus laevis ) 20. We searched in numerous databases for all known cortactin genes in other species (listed in Table 1). The identification is based on overall amino acid sequence and overall structural homology with human cortactin. Cortactin homologues exist in mammalians (human, chimpanzee, cattle, pig, mouse, rat), birds (chicken), amphibians (frog), fishes (zebrafish, pufferfish), urochordates (sea squirt), invertebrates (sea urchin), insects (fruit fly, mosquito), shrimps, worms and sponges. To date, there is no evidence for the existence of cortactin in unicellular species, nor in plants. Thus, cortactin seems to be restricted to metazoans.

For several species, both cDNA and genomic sequences (total or partial) are available and therefore we were able to reveal their genomic organization using BLASTn. The exon/intron-boundaries were determined and compared to human cortactin [see Additional file 1]. As schematically presented in Figure 1, the genomic organization and the lengths of the exons as well as the locations of the exon/intron boundaries are highly conserved from urochordates to mammalians. Pufferfishes have the shortest known genome of all vertebrate species due to much shorter introns, nevertheless most exon/intron boundaries were conserved and similar to mammalian cortactin. Intriguingly, the number of repeats in the actin-binding domain differs between species (Figure 1A–G). The number of exons and the location of the intron/exon borders of insect cortactin (Drosophila and mosquito) differ considerably with mammalian cortactin, despite the proteins sequences are very similar. Drosophila and mosquito carry 4 repeats in the actin-binding domain. In both species, repeat 1-to-3 and 4 are on separate exons with in mosquito the 4^th repeat of the actin binding domain to be encoded by a single 111 bp large exon 2 (Figure 1F,G). Both, sponge (the lowest metazoan) and sea squirt (urochordate) cortactin protein carry 5 repeats. During evolution, after creation of sponges and worms, the coelomata divided into insects and urochordates (that evolved later into vertebrates). The genomic organization of ancestors of the coelomata should reveal the roots of cortactin evolution. However, complete cDNA and/or genomic DNA of cortactin homologues in these species are not yet available.

Both nucleotide and amino acid sequence comparisons with cortactin revealed the highest similarity with the hematopoietic lineage cell-specific protein 1 (HS1). So far, HS1 homologues have been reported in human 27, mouse 33, rat and chimpanzee (NCBI database), suggesting that HS1 exists in mammalians only. We determined the intron/exon boundaries of mammalian HS1 genes by aligning the cDNA with the genomic DNA using BLASTn (Figure 1H and [see Additional file 2]). The number and lengths of the exons and the locations of the exon/intron boundaries were very similar to cortactin, especially in the exons that encode the actin-binding domain (compare [see Additional file 1] and [see Additional file 2]). The exons 10–13 of HS1 encoding the centre region between the actin-binding domain and the SH3 domain are longer (633 bp versus 489bp in cortactin) and more divergent compared to corresponding exons of cortactin.

In addition to a single cortactin homologue in all other species, nucleotide sequences comparisons using the mammalian HS1 mRNA and genomic DNA sequences revealed (incomplete) genomic sequences in chicken, pufferfish, zebrafish and frog (Table 1 and Figure 1I–M) that were more related to the HS1 protein (Figure 3 and [see Additional file 3]). Because no HS1 homologues for these species were present in the mRNA/dbEST database (except for X. laevis HS1), the cDNA (and corresponding protein) sequences were deduced from the genomic DNA with BLASTn or were predicted by Ensemble program. In these lower species, two cortactin related proteins exist. To distinguish between cortactin and HS1 variants, only the most conserved N-terminal part of cortactin and HS1 protein variants, including repeat 3 (corresponding to amino acid 1–190 of human cortactin) was used in BLASTp analysis. In each species, one protein variant turned out to be more homologous to human cortactin, and was called cortactin, whereas the other protein variant appeared to be more related to HS1 and was called HS1. This analysis unveiled HS1 proteins with more than 3 repeats in chicken and pufferfish Tetraodon nigroviridis (containing 4 1/2 repeats), pufferfish Takifugi rubripes and Xenopus laevis HS1 (5 1/2 repeats) and zebrafish HS1 (6 1/2 repeats) (Figure 1 I-M).

Moreover, alignments of the exon/intron boundaries of these HS1 genes to the mammalian HS1 genes [see Additional file 2] revealed that exon 7 (repeat 3) of HS1 was most similar to exon 10 (repeat 5) of cortactin suggesting that in mammalians exon 8 and 9 (repeat 3 and 4) of HS1 were lost during evolution. This is supported by the presence of at least one sequence of 111 nucleotides in the 5670 bp intron 6 of human HS1 (location 3271–3381) that is predicted by the program HMMER when performing alignments using a consensus sequence of the 37 amino acid repeats. However, this sequence is not functional because it does not represent an exon based on the consensus sequence of exon-intron junctions ('gt ... ag' rule of intron sequences) and no human transcripts or ESTs of HS1 including this sequence are present in the NCBI databases. In summary, HS1 is not restricted to mammalians only, but exist also in fishes, amphibians and birds and its genomic structure is very similar to that of cortactin.

Cortactin is widely expressed in most cell types suggesting to be important for vital functions, while HS1 expression is restricted to hematopoietic cells suggesting to be tailored later in evolution to serve a specific function in these cells. In concordance with their tissue-specific expression pattern, we suppose that their expression might be differently regulated. Therefore, we compared the upstream promoter regions of several cortactin and HS1 genes (Figure 2). The mammalian cortactin gene is very GC rich and contains putative SP-1 transcriptional factor binding sites that are common to many TATA-less promoters and typical for promoter regions in 'widely-expressed housekeeping genes'. Ets family transcription factors, found in the HS1 promoters, are specific for hematopoietic cells and involved in controlling the expression of many B cell- and macrophage-specific genes 52 and are critical for development of lymphoid and myeloid cell lineages. The promoter region of Drosophila and mosquito cortactin shares putative transcription factors found both in mammalian cortactin and HS1. Thus at least in mammalians, the nature of the promoters seemed to determine the broad distribution of cortactin expression in various tissues except most hematopoietic cells and the limited expression of HS1 to hematopoetic cells.

We recently reported the identification of two alternative splice variants of human cortactin; SV1-cortactin lacking the 6^th repeat and SV2 lacking the 5^th and 6^th repeat resulting in a different F-actin binding properties and decreased cell migration 14. As shown in Table 1, cortactin splice variants exist in other mammalians as well as in chicken and frog. So far, splice variants in other species have not been identified, suggesting that alternative splicing of cortactin seems to be restricted to higher metazoans. All intron sequences of cortactin bordering the splice site junctions follow the general GT/AG rule 53 except for intron 11 (GC/AG) [see Additional file 1]. As has been shown for other genes, a GT-to-GC transition might be responsible for the generation of an alternatively mRNA transcript 54. However, in frog (Xenopus laevis ), the SV1-cortactin variant exists despite the splice donor of intron 11 begins with a GT 20. Thus, concerning the genome of these different species, alternative splicing of the actin-binding domain of cortactin seems to be facilitated during evolution by modulating the splicing machinery by a GT-to-GC transition to create cortactin related variants that influences cellular properties 14. The relative expression of cortactin splice variants by tissue origin 14 suggested that splice variants might have tissue-specific functions such as fine-tuning the organization of the F-actin cytoskeleton and consequently regulating cell adhesion and migration.

Alternative splicing also occurs in human HS1. Recently a splice variant lacking the 3^rd repeat (exon 7) has been found in an SLE patient 44, resulting in enhanced BCR-mediated cell death. This alternative splicing event was due to a germ line mutation. In contrast, the splice donor of HS1 intron 6 begins with a GC [see Additional file 2]. With respect to the similarities between cortactin and HS1, it might be of interest to investigate the occurrence of splicing of HS1 exon 6 and possible biological consequences. The 3^rd repeat and its NLS links HS1 to a role in apoptosis, while such a role has not been described for cortactin lacking a NLS. Since the cytoskeleton architecture in hematopoietic lineage cells is very different from that in adherent cells, it is likely that HS1 plays an important role in the construction of tissue-type specific actin networks. Other types of actin cytoskeleton factors, such as the Arp2/3 complex activators of the WASP family have been reported to have distinct tissue specific expression profiles as well. Thus, the apparent role of HS1 in apoptosis is likely due to its actin remodeling related function. Additionally, our genomic comparisons revealed that the 3^rd repeat of HS1 corresponds with the 5^th repeat of cortactin, and therefore it might be of interest to investigate whether cortactin SV2 variant (lacking the 5^th and 6^th repeat) might be involved in apoptosis.

The 4^th repeat of cortactin has been suggested to be required for F-actin-binding 17. Genomic comparisons revealed that HS1 lacks this 4^th repeat. Nonetheless, HS1 does bind to F-actin and activate the Arp2/3 complex, although at a lower efficiency than cortactin 43. This suggests that not only a single repeat but the number of repeats is crucial for the F-actin-binding affinity 1418. In addition, HS1 contains a PIP₂ binding site in each of its 3 repeats, whereas cortactin has only one in the 4^th repeat. PIP₂ reduces F-actin cross-linking by cortactin, probably due to competition for the same binding site. Due to its higher affinity for PIP₂ 36, HS1 restores this cortactin/F-actin cross-linking process by trapping PIP₂. This might be of importance in platelets and megakaryocytes where both, cortactin and HS1 are expressed. Taken together, the composition of the repeat domain is also involved in diverting the functions of both genes.

An elegant way to study the function of a protein is to perform loss-of-function experiments. So far, cortactin knock-out models have not yet been generated successfully, because deletion of one allele of cortactin leads to premature differentiation of embryonic stem cells (personal communication in 55). However, complete loss-of-function mutants of the Drosophila cortactin gene were viable and fertile, except impaired border cell migration during oogenesis 56. Down-regulation of cortactin by RNA interference, revealed an essential role for cortactin in dendritic spine morphogenesis 57 and in E-cadherin mediated contact formation in epithelial cells 58. Mice lacking HS1, showed normal development of the lymphoid system 38, however, the antigen-receptor induced clonal expansion and deletion of B and T lymphocytes were impaired. Thus, loss of function studies underscores the divergent functions of HS1 and cortactin in different cell systems.

To examine the genesis of the cortactin family, we studied the relationship between the cortactin and HS1 homologues by generating a phylogenetic tree based on a multi-sequence alignment with the ClustalW 1,83 program [see Additional file 3]. We compared the N-terminal regions including repeat 3 (corresponding to nucleotide 1 to 190 of human cortactin), because this is the best-conserved region among all homologues (Figure 3). One cluster contains all known HS1 proteins and appeared to be closest related to a cluster composed by insects (Mosquito (Ag), Drosophila (Dm)), urochordate (sea urchin, (Sp)) and sponge (Sd) cortactin. In this last cluster all the species with only one gene (with the highest similarity with cortactin) are present. This suggests that with the appearance of the vertebrates, an ancestral gene became duplicated to create two genes, which later evolved into cortactin and HS1. This hypothesis is supported by the fact that many genes duplicated at this stage in the evolution, the overall amino-acid sequence in both genes is very similar and the introns are located at the same amino acid position. Furthermore, gene duplication often correlates with a tissue specific expression pattern of the duplicated genes, which is true for mammalian cortactin and HS1.

Figure 4 displays a hypothetical model for the origin of the cortactin and HS1 genes during evolution. The oldest ancestor is the sponge that, like sea squirt (urochordate), carries one cortactin protein with 5 c1/2 repeats. Insects have also one cortactin gene and evolved to 4 1/2 repeats. During evolution, after the creation of the sponge and the worms, the coelomata divided into insects and urochordates (that evolved later into vertebrates). This suggests that during the evolution, the number of repeats decreased in the insects. Unfortunately, no genomic sequences of ancestors of the coelomata that could reveal the roots of cortactin evolution are available yet to perform more detailed genomic analysis.

The genome of pufferfish Takifugu rubripes contains two cortactin-related genomic sequences both including 5 1/2 repeats. Most likely, an ancestor vertebrate cortactin gene underwent gene duplication. From this moment on during evolution, two cortactin/HS1-releated genes are present in all higher species. One gene evolved to mammalian HS1 with a specific function in apoptosis in hematopoietic cells. For its function, exon 8 and 9 (encoding repeat 3 and 4) were not useful and lost during evolution. However, the HS1 protein in pufferfish Takifugu rubripes and frog Xenopus laevis contains 5 1/2 repeats, while chicken and pufferfish Tetraodon nigroviridis HS1 carries 4 1/2 repeats. It might be of interest to investigate the function of these HS1 proteins and their functional differences to mammalian HS1. The other gene evolved to a ubiquitously expressed mammalian cortactin protein with a vital function in the organization of the cytoskeleton and cell migration. The 6^th repeat of cortactin most likely originated from a duplication event of the 5^th repeat, since the 6^th repeat is most similar to the 5^th repeat in all species with 6 1/2 repeats. We recently demonstrated that 6 1/2 repeats are necessary for optimal F-actin cross-linking activity and cell migration, while the splice variant lacking both the 5th and 6th repeats (SV2) was less efficient 14. Thus, the number of repeats in the F-actin binding domain of cortactin fine-tunes its function in cytoskeletal remodeling. For that reason, in higher metazoans, alternative splicing of the F-actin binding domain is most likely facilitated by a GT-GC transition in the splice donor. Alternatively, we can not exclude that gene duplication might have taken place after duplicated of the 5^th repeat (dotted arrows), since both zebrafish cortactin and HS1 contain 6 1/2 repeats.

To determine the genomic structure of the human cortactin gene, an algorithm was applied based on the consensus sequence of exon-intron junctions ('gt ... ag' rule of intronic sequence) as well as on the codon usage within ORF. Nucleotide sequence comparisons with human cortactin sequences (NCBI, GenBank accession no. M98343) using BLASTn 59 revealed homology with two genomic clones (GenBank accession no. AP000487 and AP000405). With these clones, we determined all exon/intron boundaries and size of all introns and exons (Table 2A) of the human cortactin gene by (1) performing BLAST comparisons with the cDNA against the genomic DNA and (2) using the GeneFinder program 60 based on the consensus sequence of exon-intron junctions ('gt ... ag' rule of intronic sequence) as well as on the codon usage within ORF 61.

To confirm the predicted genomic structure, we determined the intron/exon boundaries using a cloning procedure as described 62. Genomic DNA of two cosmid clones COS-7.12 and COS-3.72 covering the cortactin gene as determined by the full-length cDNA 5, was amplified with randomly selected primers from the cDNA sequence (GeneBank accession no. M98343). All PCR products that were larger than the cDNA control sample were considered to be caused by intron sequences and compared to genomic sequence (accession number AP000487 and AP000405) using BLASTn 59. The size of intron 1, 5, 8, 12 and 13 was too large to obtain a reliable sequence.

Because no overlapping genomic sequences immediately 5' of the first exon were present in the database, we performed sequence analysis of a 2.7-kb HincII-HincII fragment representing the first exon and its 5'-flanking sequences from cosmid COS-7.12 cloned into pUC18 (p5'EMS_3135). In addition, we sequenced a 5-kb PCR product using a 5'-primer in the vector (within the TET gene) and 3'-primer (p3135p601: 5'-ccgggtcggccctggattcc-3') within exon 1, subcloned in pUC18 (p5'EMS_4911). Nucleotide sequences of both products were compared with the genomic clones representing the cortactin gene present in the NCBI database (Accession number AP000487 (GI 8118774 and GI 6277297) and AP000405 (GI 8118742)) and used to define the 7.4 kb 5'-flanking region. The PROSCAN program 63 from BIMAS was used to define the 316 bp promoter region preceding exon 1. Putative transcription factor binding sites where determined by the TFSEARCH program 64 and graphically represented in figure 2. Sequences from human cortactin were submitted to NCBI GenBank 65 as accession No. M98343 (cDNA) and AJ288897 (promoter).

The (deduced) protein and genomic sequences of all cortactin and HS1 genes were retrieved from various WEB-sites and their available sequence data are summarized in Table 1. In addition, partial cortactin sequences (ESTs and/or genomic) of various organisms were identified based on amino acid sequence homology with existing cortactin proteins. The genomic organization of the sea squirt and Takifugu rubripes could not be completely elucidated, because cDNA/genomic sequences were only partially available. All data were compiled using BLAST searches of the following databases: National Center for Biotechnology Information (NCBI) (Bethesda, MD, USA) 65; The Wellcome Trust Sanger Institute (Cambridge, UK) 66; EnsEMBL of The Wellcome Trust Sanger Institute (Cambridge, UK) 67; DOE Joint Genome Institute (Walnut Creek, CA, USA) 68; TIGR: The Institute for Genomic Research (Rockville, MD, USA) 69; DNA Data Bank of Japan (Mishima, Shizuoka, Japan) 70; Nematode.net Genome Sequencing Center (St. Louis, MO, USA) 71; Wormbase (NY, USA) 72; European Bioinformatics Institute (EBI) (Cambridge, UK) 73; Genoscope National Sequencing Center (Evry, France) 74; The U.S. Poultry Gene Mapping Project (MI, USA) 75 and UCSC Genome Bioinformatics (Santa Cruz, CA, USA) 76.

To determine the exon/intron boundaries of all cortactin and HS1 genes, available genomic sequences were subjected to sequence alignments of each species-specific cDNA sequence using the BLAST program of NCBI. Using the same algorithms, as described for human cortactin, the exon/intron-boundaries could be predicted. The complete genomic sequences of the 5' flanking region of cortactin of human, chimpanzee, mouse, rat, fruit fly, and mosquito were determined using the various accession numbers of genomic DNA in Table 1. Putative transcription factor binding sites of 800 bp of the 5' flanking regions where determined by the TFSEARCH program (Figure 2). The predicted exon in intron 6 of HS1 was predicted by the bio-informatics program HMMER 77) The human cortactin 6 1/2 repeats of the actin-binding domain were aligned, resulting in a consensus sequence: (kfGvqkdrvDksAvGfdyqekvekhesqkDysk). With HMMER this consensus sequence was 'tBLASTn' to intron 6 of human HS1. With an acceptable probability (E-value 0.095), the program predicted an exon in this intron 6 (at location 3271–3381).

Sequence alignments were carried out using the BLAST program of NCBI. The multiple sequence alignments of various cortactin proteins were constructed using Basic GeneBee ClustalW 1.83 78. The genome, cDNA or protein was completed for all cortactin homologues and the number of repeats differs across species and between HS1 and cortactin. Only the N-terminal of cortactin and HS1 proteins including repeat 3 (corresponding to amino acid 1–190 of human cortactin) was used to generate a phylogenetic tree, because this is the most conserved part. Predicted nuclear localization signals sequences were obtained using Predict NLS program 79.