Evolutionary Perspective on Annexin Calcium-Binding Domains

Home , Annexin, Annexin A11, Annexin A13

View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector

Biochimica et Biophysica Acta 1742 (2004) 133–140 http://www.elsevier.com/locate/bba Review Evolutionary perspective on annexin calcium-binding domains

R.O. Morgan*, S. Martin-Almedina, J.M. Iglesias, M.I. Gonzalez-Florez, M.P. Fernandez

Department of Biochemistry and Molecular Biology, Edificio Santiago Gascon, Faculty of Medicine, University of Oviedo, 33006 Oviedo, Spain

Received 7 September 2004; accepted 13 September 2004 Available online 22 September 2004

Abstract

Molecular systematic analysis of the annexin gene superfamily characterized the evolutionary origin, frequency and range of structural variation in calcium interaction domains that are considered intrinsic for membrane targeting and ion channel function. Approximately 36% of annexin repeat domains in an estimated 100 distinct subfamilies contained amino acid changes consistent with the functional loss of type two calcium-binding sites. At least 11% of annexin domains contained a novel K/H/RGD motif conserved in particular subfamilies and manifest in all phyla, apparently via convergent evolution. The first yeast annexin from Yarrowia lipolytica was classified in the ANXC1 subfamily with fungal and mycetozoan representatives. This clade had intact calcium-binding sites but disruption of the normally well- conserved, mid-repeat 4 region implicated in calcium channel regulation. Conversely, a tandem pair of novel annexins from the amphioxus Branchiostoma floridae resembled annexin A13 in gene structure and conserved the charged amino acids associated with the internal hydrophilic pore, but were devoid of external type 2 calcium-binding sites and incorporated K/RGD motifs instead, like annexin A9. The selective erosion of calcium-binding sites in annexin domains and the occurrence of alternate ligands in the same exposed, interhelical loops are pervasive features of the superfamily. This suggests greater complexity than previously appreciated in the mechanisms controlling annexin membrane interaction and calcium channel operation. D 2004 Elsevier B.V. All rights reserved.

Keywords: Annexin; Calcium-binding domain; Hidden Markov model; Molecular evolution; Phylogenetic analysis; Protein modeling

1. Introduction or predictions about their mechanism and function, even when empirical data are lacking. The basis of this approach Protein domains are conserved structural elements, is that functional constraints ordain the conservation of generally associated with a perceived function or mecha- residues that are vital for structure and function, be they nism of interaction. Even where such a reductionist concept universal or class-specific. The methodology involves may be coherent with proteomic analysis, it should be determining the evolutionary relationships among homolo- tempered by the realization that individual domains and gous members, defining molecular profiles for individual integral proteins may exhibit a broad range of diversity and clades or subfamilies, and mapping site-specific changes specificity in both structure and function. This is achieved into three-dimensional structures to infer which sites are through sometimes subtle structural variations, context- responsible for the preservation or divergence of function. sensitive differences in domain architecture, target-condi- The diversity of calcium-binding proteins is further tioned conformational changes and host status affecting complicated by the independent evolution of some protein gene expression, subcellular localization, posttranslational domains (e.g., C2 and EF-hand) within multiple, distinct modification, and pathophysiology. Molecular systematic gene families [1,2]. The identification of calcium-inde- examination of homologous proteins can greatly aid in the pendent members in virtually all families of calcium- characterization of protein domains and facilitate inferences binding proteins argues for roles that extend beyond mere targeting domains, to proffer alternative mechanisms that * Corresponding author. Tel.: +34 985 104214; fax: +34 985 103157. accomplish the same function, or to introduce accessory E-mail address: [email protected] (R.O. Morgan). features that modify function. The annexin superfamily

[3,4] is particularly amenable to systematic evolutionary visual inspection and integration of contig data based on study because its universally conserved, four-repeat archi- existing models [16–18]. Site-specific rate variations to tecture may represent the basic functional unit (possibly an detect functional divergence made use of DIVERGE [19] atypical calcium channel?) and the internally homologous and CONSURF [20] and protein modeling of evolutionary repeats normally harbor unique, discontiguous (type 2) information was performed by SWISS-MODEL and DEEP- calcium-coordinating residues. However, the selective VIEW [21]. ablation of these sites in annexins A9 and A10 [5–8] and the importance of other calcium-binding residues in B- helices [9] forecast a range of variation in calcium depend- 3. Results and discussion ence and mechanism of action. A systematic molecular analysis of the annexin superfamily was therefore under- 3.1. Phylogenetic survey of calcium-binding domains taken to ascertain the origin, extent and consequences of such variations to better comprehend the multifaceted Conserved domains pertaining to distinct gene families functional profile of annexins. Associated fundamental can serve as a basis for surveying the phylogenetic questions include, for example, whether major clades such distribution of calcium-binding protein families with unique as plant, fungal and animal annexins have evolved differ- domain profiles and architectures using the PFAM database ently, what key features differentiate individual subfamilies, [15]. The known species distribution of calcium-binding and whether there is evidence of convergent evolution by proteins with C2, copine, calreticulin or annexin domains which novel features have appeared independently. spans all eukaryotic kingdoms (Table 1), while S100 and calsequestrin families are confined to vertebrates and the EF-hand and its Excalibur analog [22] extend to thermo- 2. Materials and methods philic archaea and eubacteria. The true origin of most calcium-binding domains still remains uncertain due to the The bioinformatic analysis of the annexin gene super- absence of extant homologs in primitive eukaryotes and family began with the identification of extant homologs by knowledge of domain architectures and ancestral life forms BLAST (esp. PSI-BLAST) and FASTA sequence compar- near the root of the Tree of Life. A replete and statistically isons of authenticated members with protein, cDNA tran- validated phylogenetic tree for the entire annexin super- script and genomic trace databases, and their compilation family comprising almost 100 subfamilies and 1500 cognate into multiple sequence alignments by CLUSTALW [10,11]. orthologs is nearing realization. This will serve as a scaffold Phylogenetic analysis initiated with neighbor-joining anal- upon which to relate structural and functional changes ysis of bootstrapped alignments using MEGA [12] followed among diverse homologs in individual subfamilies by by maximum likelihood analysis of clades with TREE- defining the cladistic order of paralogous gene duplications PUZZLE [13]. Molecular profiles of individual subfamilies and orthologous speciation events and measuring the extent were created as hidden Markov models (HMMs) by of evolution as the branch length for each taxon. HMMER [14] and visualized with sequence logos, as The basic evolutionary unit is the gene subfamily and applied to domain families by the PFAM database [15]. species orthologs typically cluster in the tree topology Variant domains were detected by unusual branch patterns because they represent structural and functional equivalents and lengths in phylogenetic trees and visually compared that coevolved in different species. The subtree branching with the multiple sequence alignments and HMMs. Gene for species members of the annexin C1 subfamily (Fig. 1) structures and chromosomal linkage maps were deduced by portrays their close relationship in evolution, structure and

Table 1 Phylogenetic distribution of calcium-binding domains C2 EF-hand Excalibur Annexin Copine Calreticulin S100 Calsequestrin Eukarya 1086 2540 – 229 68 126 107 20 Metazoa 762 1727 – 160 43 69 107 20 Planta 241 499 – 45 17 40 – – Fungi 51 116 – 4 – 6 – – Mycetozoa 13 38 – 1 3 2 – – Yeasts 27 38 – 1 – 4 – – Alveolata 12 92 – 1 2 – – – Diplomonad – 19 – 12 4 – – – Archaea – 6 – – – – – – Eubacteria – 68 17 – – – – – Viruses – 5 1 – – – – – Architectures 115 171 6 8 3 1 4 1 The numbers of domain entries and representative architectures (bottom line) were extracted from the current PFAM database. R.O. Morgan et al. / Biochimica et Biophysica Acta 1742 (2004) 133–140 135

Fig. 1. Phylogenetic tree of the annexin ANXC1 subfamily. The tree was constructed by the neighbor-joining algorithm of MEGA from 1000 bootstrap alignments of the 314-amino-acid core tetrad region for the genus–species indicated. Sequences were compiled from public databases at NCBI provided by the Ge´nolevures Consortium (France), the Whitehead-Broad Institute (USA) and the DOE-Joint Genome Institute (USA). Bootstrap percentages at the nodes signify statistical support for the branching topology. The yeast/slime mold outgroup oriented the tree in accordance with a supertree for all annexin subfamilies (not shown) placing the more divergent 12 ascomycete fungi above. function. It classifies the first yeast annexin from Yarrowia their structural divergence in a site-specific manner and to lipolytica sequenced by the Ge´nolevures Consortium [23]. identify synapomorphies (conserved inherited character- The branch lengths and NJ-bootstrap values indicate that all istics) and convergent evolution (independently emerged are typical ANXC1 members with slightly greater proximity features). between yeast and mycetozoa (i.e., slime mold) compared to A profile HMM for the ANXC1 subfamily can be basidiomycete and ascomycete fungi [24], despite only 35– visualized as a sequence logo (Fig. 2), which reflects the 40% amino acid (aa) identity among members. The aa composition and frequency at each site with the observation that these species genomes host one, two or consensus sequence on top. The total stack height is three annexins, respectively, highlights the vagarious nature especially important as it conveys the relative information of genome inheritance. This is not restricted to unicellular content at each site, based on commonly observed organisms as vertebrate genome analysis points to the replacement values, to indicate the overall degree of selective absence (loss?) of annexins ANXA3 and A9 from structural conservation and hence functional significance birds, ANXA8 from amphibians and annexins A7, A8 and at each site. The yeast annexin can be seen to conform to A10 from fish. These reductions contrast with the selective the sequence model, and the model itself conforms to most amplification of the annexin A1 subfamily in fish, amphib- of the key structural aspects common to annexins such as ians and some birds (unpublished findings). the highly conserved type 2 calcium-binding sites, the recently reevaluated sites in adjacent B-helix [9], and the 3.2. Subfamily profiles from hidden Markov models global four-repeat architecture. However, one cluster of normally well-conserved aa involved in the formation of Protein family databases such as PFAM [15] process internal salt bridges related to putative calcium channel multiple sequence alignments by algorithms that utilize activity [25,26] appears particularly anomalous. The hidden Markov models to create statistical molecular typical vertebrate annexin mid-repeat 4 sequence spanning profiles of homologous families as a basis for the structural helices 4B and 4C bGaGTDDktliRimvsRsEiDQ [27] and functional annotation of emerging genomes. Such converts to bGmGTKDerliYRvvRlHWNrQ in ANXC1 family signatures are efficient, powerful tools for the members (i.e., key conserved residues capitalized and classification and alignment of unidentified sequences and underlined), with displacement of the vital Arg residues by the models can also be used (like PSI-BLAST) to ferret bulky, aromatic aa such as Tyr and Trp and followed by a more distant homologs. A logical but unpracticed exten- 2-aa insertion (Fig. 2). Thus, ANXC1 exhibits general sion of this approach is to create unique molecular profiles conservation of a-helices 2AB and 4AB comprising the that characterize orthologs pertaining to individual sub- hydrophilic pore of the putative calcium channel and the families, thereby refining the structural prototypes to Dictyostelium homolog (only) retains at least the first correspond more precisely to the divergent phenotypes conserved Arg correctly positioned [28], but the relative and functions inculcated within the family. We have capacity of these subfamily members to function as generated such profile HMMs for each of the 100 annexin calcium channels remains to be verified. The yeast annexin subfamilies delineated by phylogenetic analysis, to define also suffers a 2-aa deletion in the loop between repeats 2 136 R.O. Morgan et al. / Biochimica et Biophysica Acta 1742 (2004) 133–140

Fig. 2. Sequence logo of a hidden Markov model for the annexin ANXC1 subfamily. The complete, deduced aa sequence from Y. lipolytica (Ge´nolevures Consortium, GenBank accession no. CR382130) is shown above the four aligned repeat blocks for 16 representative species, listed in Fig. 1 and compiled by HMMER. The model logo summarizes the aa distribution and frequency by letter height and the functional information content at each site is given by total stack height, based on expected aa replacements. The predicted a-helical regions (labels 1A through 4E), the calcium-binding sites of type 2 (solid arrows) and B-helix (open arrows), the disrupted mid-repeat 4 region (stars) and the 15 most conserved, charged sites (numbered symbols) were based on analysis ofa master alignment of 1500 annexins. and 3, but still retains the highly conserved Arg here, conserved aa changes affecting calcium-binding sites in the presumably critical for inter-repeat structural stability. four internal repeats of each subfamily. Visual examination of the individual HMMs and multiple sequence alignments 3.3. Variable domain structure in the annexin family for these subfamilies confirmed the loss of critical aa at these sites in all repeats of ANXA9 members, repeats 1 and HMM models of the 12 subfamilies comprising bAQ 4 of ANXA2, repeat 1 of ANXA1, repeats 1, 3 and 4 of family annexins were similarly compiled from alignments of ANXA10 and repeat 3 of the 3V tetrad of ANXA6. The all known vertebrate orthologs and consensus sequences novel K/H/RGD motif previously detected [18] was con- from each were used as operational taxonomic units to firmed to reside in the same interhelical loops as the compute the neighbor-joining bootstrap tree. This fully calcium-coordinating residues, as an adjacent ligand in resolved tree is congruent to those generated under different repeats 3 and 4 of ANXA9, A2 and A1, in an overlapping models of evolution and with more extensive taxon position for ANXA4 and A5, and in some orthologs of representation, except for some remaining uncertainty in ANXA3 repeat 1 and ANXA13 repeat 2. Its presence did the branch position of ANXA3, which occasionally bifur- not appear to depend on the presence or absence of type 2 cated with ANXA4. The ANXA13, A7 and A11 subfamilies calcium-binding sites and it was additionally located in the form the basal group closest to the trunk of the superfamily inter-repeat 2–3 loop of all ANXA1 orthologs and the amino tree, consistent with their ancestral gene structures [16–18] terminus of ANXA13 encoded by an alternatively spliced and the former two exhibit strong phylogenetic association cassette exon present only in late-diverging mammals [17]. with invertebrate annexins (not shown). The large ortholog A broader survey of all 100 annexin subfamilies for the sample for each subfamily facilitated the identification of loss of type 2 sites and incorporation of KGD motifs R.O. Morgan et al. / Biochimica et Biophysica Acta 1742 (2004) 133–140 137 revealed that 36% of annexin domains had mutations lies. We further investigated the origin of this arrangement in consistent with a loss of calcium-binding capacity, and the extraordinary example of amphioxus annexins because more than 11% of domains contained a KGD motif that was this marine invertebrate species was expected to possess at also represented by either HGD or RGD equally in about least some cognate orthologs of A family annexins. As a 10% of these latter cases. Since the occurrence of this motif primitive chordate, phylogenetically positioned just prior to was shared by evolutionarily related ANXA1, A2 and A9, it the purported genome duplications that conditioned verte- is important to emphasize that all other cases indicated a brate emergence around 500–550 million years ago, it was stochastic appearance of the KGD motif in evolutionarily expected to show some resemblance to vertebrate annexins. independent lineages (but always at the same critical sites) The genomic contig containing these homologs (Genbank and ostensibly independent of the presence or absence of accession no. AC150387) was analyzed bioinformatically to type 2 calcium-binding sites. Interestingly, earlier diverging ascertain close physical linkage between the two tetrads phyla representing invertebrate B family, fungal C family, (Fig. 4) with some uncertainty about the presence of a stop plant D family and protist E family exhibited a somewhat codon or intron splice at the end of the 5V tetrad. Their gene greater proportion of annexin domains lacking type 2 structures were both congruent with that of vertebrate calcium-binding sites and incorporating the KGD motif. ANXA13, except for one extra and one absent intron in The selected examples described beneath ANXA domains the gene encoding the 5V tetrad. This suggested a unique (Fig. 3) illustrate some of the more extreme variations tandem duplication event unlike any observed for vertebrate observed in selected, novel subfamilies from amphioxus, annexins and the level of pairwise aa identity between the fungus, thale cress and protists. two tetrads (i.e., 30.6%) indicated that this was an ancient event, perhaps accompanied by accelerated divergence due 3.4. Conserved, alternate motifs in annexin domains to the loss of calcium-binding sites. An examination of the deduced protein structures The preceding domain analysis highlighted the extent to established unequivocally that the calcium-coordinating which bcanonicalQ type 2 calcium-binding sites are altered residues in all four annexin domains in both proteins were and presumably dysfunctional throughout the annexin super- replaced with functionally incompatible aa and, surprisingly, family. The prevalence of an alternate KGD motif in the KGD and RGD motifs were positoned in some of these very same loop regions and in other critically exposed regions sites and in the inter-repeat 2–3 loop (Fig. 4). Thus, both between repeats 2 and 3 or the amino and carboxy termini genes had structures reminiscent of the most primitive was also a conserved characteristic within diverse subfami- vertebrate subfamily ANXA13, but had fully calcium-

Fig. 3. Phylogenetic tree of the ANXA family and domain architecture of representative annexins. The cognate orthologs of all 12 human annexins (numbers in brackets) were aligned and compiled by HMMER to yield consensus sequences used as taxa. Maximum likelihood branch lengths were computed by TREE- PUZZLE using the topology obtained from neighbor-joining analysis of 1000 bootstrap alignments with MEGA (percentage branching statistics at nodes). Symbols to the right describe the domain architecture conserved in ANXA subfamilies and for the selected species representing B, C, D and E subfamilies. The domain subtypes were characterized by visual examination of a master sequence alignment of 1500 annexins by the absence (solid symbol) or presence of a type 2 calcium-binding motif (CBD) and/or a K/H/RGD motif (i.e., KGD or inter-repeat bKQ). The source GenBank accession nos. are AC150387 for Branchiostoma, CO031583 for Coccidioides,NM_103274 for Arabidopsis and AF514360 for Giardia. 138 R.O. Morgan et al. / Biochimica et Biophysica Acta 1742 (2004) 133–140

Fig. 4. Gene structure and calcium-binding site aa replacements for a pair of amphioxus annexins. (A) Exon locations were deduced by comparison of an amphioxus contig sequence (GenBank accession no. AC150387) with known annexin proteins and the resulting transcript exon sizes were compared to the three prototype structures of human annexins A11, A7 and A13. (B) The canonical type 2 calcium-binding site present in most annexin domains is aligned with four repeats of the two amphioxus annexins to illustrate loss of these sites with incorporation of alternate K/RGD motifs defined by bioinformatic analysis. independent protein domains containing the alternate KGD heritage, it remained important to visualize the location of motif reminiscent of ANXA9. Thus, neither was a clear modified sites to fully appreciate their potential impact on ortholog of any vertebrate annexin, and the complete annexin function. The protein model in Fig. 5 portrays the annexin repertoire of the amphioxus genome will be three-dimensional structure of the amphioxus 3V tetrad, required for rigorous phylogenetic analysis to determine provisionally designated Bfl1.2. It was created by SWISS- their progenitor and any common ancestor with vertebrate annexins. It is noteworthy that our analysis of annexins in the urochordate Ciona intestinalis and the purple sea urchin Strongylocentrotus purpurata also reveals gene exon structures compatible with vertebrate annexins but without any obvious cognate orthologs in phylogenetic analyses of their transcripts or proteins (unpublished findings). It is apparent that calcium site depletion and KGD emergence are widespread properties of convergent evolution in the annexin superfamily.

3.5. Molecular modeling of functional ligands in annexin domains Fig. 5. Protein model of amphioxus annexin bBfl1.2Q. The deduced protein sequence (GenBank accession nos. are AC150387) was threaded by We have previously used site conservation and rate- SWISS-MODEL [21] through the crystal structure template of calcium- change analyses to determine annexin sites likely to be free pig annexin A1 (RCSB Protein DataBank accession no. 1HM6) responsible for conserved function or subfamily divergence, together with an alignment of 100 annexin subfamily consensus sequences. respectively, and have mapped this evolutionary information The backbone model of tetrad repeats (numbered) was rendered by DEEPVIEW to highlight the disrupted calcium-binding sites on the upper into threaded crystal structure models to visualize these convex surface (black sticks with adjacent sequences), the two KGD patterns [27]. For the divergent amphioxus annexins clusters (grey spheres), and the 15 most conserved, charged aa sites in the characterized above as having an obscure subfamily entire annexin superfamily (black spheres). R.O. Morgan et al. / Biochimica et Biophysica Acta 1742 (2004) 133–140 139

MODEL, based on the crystallography coordinates of itself may also be relevant. The search for specific, annexin A1 [29] and an alignment of 100 annexin subfamily functionally relevant membrane receptor molecules (i.e., consensus sequences. It clarifies the location of the four lost proteins especially) may encounter clues from studies of calcium-binding sites in the AB and DE interhelical loop KGD biological properties in the context of some annexin regions exposed on the upper convex surface of the protein domains. The present study emphasizes that different and shows that one KGD motif lies in an adjacent position annexin subfamilies exhibit conserved differences in their while the other lies near the lower concave surface domain properties and architecture, and that this must be connecting repeats 2 and 3. The former case invites kept in mind when trying to decipher the mechanism and speculation that the KGD ligand may substitute for the lost role for each subfamily. calcium-binding site for either a similar or distinct function, while the latter may create a novel functional group involved in either the maintenance of inter-repeat protein Acknowledgments structure or even cytosolic receptor interaction. Finally, it is noteworthy that of the 15 charged residues normally highly This work was supported by grant BMC2002-00827 conserved in all annexins (see Fig. 2), the four situated in from the Ministry of Science and Technology of Spain. the type 2 calcium-binding loops are mutated in this amphioxus annexin whereas the remainder are essentially intact and positioned in the central core of the tetrad References structure where they could still maintain the hydrophilic pore of a putative calcium channel. [1] J.L. Jime´nez, G.R. Smith, B. Contreras-Moreira, J.G. Sgouros, F.A. Meunier, P.A. Bates, G. Schiavo, Functional recycling of C2 domains throughout evolution: a comparative study of synaptotagmin, protein 4. Conclusion kinase C and phospholipase C by sequence, structural and modelling approaches, J. Mol. Biol. 333 (2003) 621–639. [2] H. Kawasaki, S. Nakayama, R.H. Kretsinger, Classification and This first phylogenetic survey of domain structure in the evolution of EF-hand proteins, BioMetals 11 (1998) 277–295. eukaryotic annexin superfamily has revealed universal [3] T.E. Hawkins, C.J. Merrifield, S.E. Moss, Calcium signaling and conservation of the four-repeat architecture for these unique annexins, Cell Biochem. Biophys. 33 (2000) 275–296. 68–69-aa domains. However, extensive disruption of the [4] V. Gerke, S.E. Moss, Annexins: from structure to function, Physiol. Rev. 82 (2002) 331–371. type 2 calcium-binding sites in approximately one third of [5] V. Goebeler, D. Ruhe, V. Gerke, U. Rescher, Atypical properties annexin domains and the incorporation of novel KGD displayed by annexin A9, a novel member of the annexin family of motifs pose the question of functional consequences and Ca(2+) and lipid binding proteins, FEBS Lett. 546 (2003) 359–364. physiological roles. The pervasiveness of both changes [6] R.O. Morgan, M.P. Fernandez, Expression profile and structural throughout the superfamily, the high conservation of each divergence of novel human annexin 31, FEBS Lett. 434 (1998) 300–304. within particular subfamilies, and their strategic locations [7] R.O. Morgan, D.W. Bell, J.R. Testa, M.P. Fernandez, Human annexin accentuate their potential significance. Their apparently 31 genetic mapping and origin, Gene 227 (1999) 33–38. independent (i.e., convergent) evolution may mean they [8] R.O. Morgan, N.A. Jenkins, D.J. Gilbert, N.G. Copeland, B.R. have distinct roles, that they represent alternate mechanisms Balsara, J.R. Testa, M.P Fernandez, Novel human and mouse annexin to achieve the same end (e.g., membrane binding) or that A10 are linked to the genome duplications during early chordate evolution, Genomics 60 (1999) 40–49. they may act in concert to achieve greater specificity for [9] M. Jin, C. Smith, H.Y. Hsieh, D.F. Gibson, J.F. Tait, Essential role of membrane receptor interactions. In this regard, it is B-helix calcium binding sites in annexin V-membrane binding, J. Biol. interesting to note that there are significant distinctions in Chem. 279 (2004) 40351–40357. the Ca/KGD architecture of annexin domains between major [10] S.F. Altschul, T.L. Madden, A.A. Sch7ffer, J. Zhang, Z. Zhang, W. clades, exempified by the characteristic loss of type 2 Miller, D.J. Lipman, Gapped BLAST and PSI-BLAST: a new generation of protein database search programs, Nucleic Acids Res. calcium-binding sites in repeats 2 and 3 of virtually all plant 25 (1997) 3389–3402. annexin subfamilies. [11] J.D. Thompson, D.G. Higgins, T.J. Gibson, CLUSTAL W: improving A more detailed understanding is needed as to how these the sensitivity of progressive multiple sequence alignment through structural variations, singly or in combination, influence sequence weighting, positions-specific gap penalties and weight annexin-membrane binding or insertion and electrophysio- matrix choice, Nucleic Acids Res. 22 (1994) 4673–4680. [12] S. Kumar, K. Tamura, M. Nei, MEGA3: Integrated software for logical activity. In the same way that a systematic molecular evolutionary genetics analysis and sequence alignment, investigation of mutated annexin calcium-binding sites can Brief. Bioinform. 5 (2004) 150–163. unveil novel mechanistic contributions [9], a knowledge of [13] K. Strimmer, A. von Haeseler, Quartet puzzling: a quartet maximum the role of exposed K/H/RGD motifs in annexin membrane likelihood method for reconstructing tree topologies, Mol. Biol. Evol. interactions and receptor targeting may shed light on 13 (1996) 964–969. [14] S.R. Eddy, Profile hidden Markov models, Bioinformatics 14 (1998) annexin function. If the assumption of a mere membrane- 755–763. targeting role is too simplistic, then probing the biophysical [15] A. Bateman, L. Coin, R. Durbin, R.D. Finn, V. Hollich, S. Griffiths- consequences of annexin binding on membrane function Jones, A. Khanna, M. Marshall, S. Moxon, E.L.L. Sonnhammer, D.J. 140 R.O. Morgan et al. / Biochimica et Biophysica Acta 1742 (2004) 133–140

Studholme, C. Yeats, S.R. Eddy, The Pfam protein families database, [22] D.J. Rigden, M.J. Jedrzejas, M.Y. Galperin, An extracellular calcium- Nucleic Acids Res. 32 (2004) D138–D141. binding domain in bacteria with a distant relationship to EF-hands, [16] P. Bances, M.R. Fernandez, M.I. Rodriguez-Garcia, R.O. Morgan, FEMS Microbiol. Lett. 221 (2003) 103–110. M.P. Fernandez, Annexin A11 (ANXA11) gene structure as the [23] B. Dujon, D. Sherman, G. Fischer, et al., Genome evolution in yeasts, progenitor of paralogous annexins and source of orthologous cDNA Nature 430 (2004) 35–44. isoforms, Genomics 69 (2000) 95–103. [24] V. Khalaj, L. Smith, J. Brookman, D. Tuckwell, Identification of a [17] J.M. Iglesias, R.O. Morgan, N.A. Jenkins, N.G. Copeland, D.J. novel class of annexin genes, FEBS Lett. 562 (2004) 79–86. Gilbert, M.P. Fernandez, Comparative genetics and evolution of [25] P. Demange, D. Voges, J. Benz, S. Liemann, P. Gottig, R. Berendes, annexin A13 as the founder gene of vertebrate annexins, Mol. Biol. A. Burger, R. Huber, Annexin V: the key to understanding ion Evol. 19 (2002) 608–618. selectivity and voltage regulation? Trends Biochem. Sci. 19 (1994) [18] M.P. Fernandez, R.O. Morgan, Structure, function and evolution of 272–276. the annexin gene superfamily, in: J. Bandorowicz-Pikula (Ed.), [26] S. Liemann, R. Huber, Three-dimensional structure of annexins, Cell. Annexins: Biological Importance and Annexin-Related Pathologies, Mol. Life Sci. 53 (1997) 516–521. Landes Bioscience/Kluwer Academic/Plenum, 2003, pp. 21–37. [27] S.E. Moss, R.O. Morgan, The annexins, Genome Biol. 5 (2004) 219. [19] X. Gu, Statistical methods for testing functional divergence after gene [28] S. Liemann, I. Bringemeier, J. Benz, P. Gottig, A. Hofmann, R. Huber, duplication, Mol. Biol. Evol. 16 (1999) 1664–1674. A.A. Noegel, U. Jacob, Crystal structure of the C-terminal tetrad [20] F. Glaser, T. Pupko, I. Paz, R.E. Bell, D. Bechor-Shental, E. Martz, N. repeat from synexin (annexin VII) of Dictyostelium discoideum, Ben-Tal, ConSurf: identification of functional regions in proteins by J. Mol. Biol. 270 (1997) 79–88. surface-mapping of phylogenetic information, Bioinformatics 19 [29] A. Rosengarth, V. Gerke, H. Luecke, X-ray structure of full-length (2003) 163–164. annexin 1 and implications for membrane aggregation, J. Mol. Biol. [21] T. Schwede, J. Kopp, N. Guex, M.C. Peitsch, SWISS-MODEL: an 306 (2001) 489–498. automated protein homology-modeling server, Nucleic Acids Res. 31 (2003) 3381–3385.