Annexins

Authors


Author for correspondence:
Julia Davies
Tel: +44 1223 333939
Email: jmd32@cam.ac.uk

Abstract

Contents

 Summary40
I.Introduction40
II.Expression and localization during development41
III.Response to the environment42
IV.Control of subcellular localization42
V.Structural determinants of function43
VI.Activities and possible functions47
VII.Conclusions and prospects49
 Acknowledgements49
 References49

Summary

Annexins are multifunctional lipid-binding proteins. Plant annexins are expressed throughout the life cycle and are under environmental control. Their association or insertion into membranes may be governed by a range of local conditions (Ca2+, pH, voltage or lipid identity) and nonclassical sorting motifs. Protein functions include exocytosis, actin binding, peroxidase activity, callose synthase regulation and ion transport. As such, annexins appear capable of linking Ca2+, redox and lipid signalling to coordinate development with responses to the biotic and abiotic environment. Significant advances in plant annexin research have been made in the past 2 yr. Here, we review the basis of annexin multifunctionality and suggest how these proteins may operate in the life and death of a plant cell.

I. Introduction

Annexins are soluble proteins, encoded by a large multigene family spread through the Eukaryote and Prokaryote domains (Morgan & Fernández, 1997; Braun et al., 1998; Hofmann, 2004; Bouzenzana et al., 2006). At neutral pH, annexins exhibit reversible calcium (Ca2+)-dependent binding to negatively charged phospholipids such as phosphatidylserine (PS), phosphatidylglycerol and phosphatidylinositol. Entering ‘annexin’ as a literature search term can elicit over 17 000 entries but many will occur because an annexin has been used as an animal cell death or stress marker. In such studies, vertebrate annexin A5 binds to PS as the latter emerges at the extracellular plasma membrane (PM) face. Discarding those references leaves a much smaller selection, dominated by animal studies. There, annexins have been tracked as they move around the cell in response to elevation of cytosolic free Ca2+ ([Ca2+]cyt) (Babiychuk et al., 2009). They might cluster together at a membrane, bind membrane receptors, demarcate membrane domains, regulate traffic, regulate the cytoskeleton or a transport protein or form a transport pathway themselves (reviewed by Kourie & Wood, 2000; Gerke & Moss, 2002; Gerke et al., 2005; Grewal & Enrich, 2009; Monastyrskaya & Babiychuk, 2009). As such, they are implicated in regulating membrane dynamics, mediating [Ca2+]cyt sensing and signalling, linking [Ca2+]cyt dynamics to cytoskeletal responses and mediating immune or stress responses. They may be pro-apoptotic or anti-apoptotic and feature in proliferation and movement of cancerous cells (reviewed by Gerke & Moss, 2002; Gerke et al., 2005; Mussunoor & Murray, 2008; Grewal & Enrich, 2009; Monastyrskaya & Babiychuk, 2009). By contrast, comparatively few reports on plant annexins have been made since their first isolation from tomato (Boustead et al., 1989) and even fewer exist for oomycete and fungal annexins.

Plant annexins (32–36 kDa) are abundant proteins (0.1% of total cell protein) and are widespread in the Plant Kingdom (Morgan & Fernández, 1997; Hofmann, 2004; Mortimer et al., 2008). Annexins have been identified at mRNA and/or protein level in model and crop plants including Arabidopsis (Clark et al., 2001; Lee et al., 2004), barley (Clarke et al., 2008), celery (Seals et al., 1994), maize (Blackbourn et al., 1991), Medicago (Kovács et al., 1998), pea (Clark et al., 1998), rice (Hashimoto et al., 2009), potato (Riewe et al., 2008), tobacco (Seals et al., 1994; Tang et al., 2003) and wheat (Breton et al., 2000). Studies on fungi and oomycetes lag behind but, to date, annexins have been firmly identified in Saprolegnia (Bouzenzana et al., 2006), Phytophthora ramorum and Phytophthora sojae (Meijer et al., 2006; Savidor et al., 2008), Aspergillus niger and Aspergillus fumigatus (Khalaj et al., 2004a,b), and Neurospora crassa (Braun et al., 1998). Sequences for Fusarium, Magnaporthe grisea and Pleurotus have also been identified (Braun et al., 1998; Khalaj et al., 2004b). For oomycetes, fungi and plants the trend is for the genome to harbour at least two annexins. Plant annexins are phylogenetically distinct from animals and form four clades (Moss & Morgan, 2004; Jami et al., 2009) while oomycete and fungal annexins form distinctive, well-separated clades (Bouzenzana et al., 2006). These annexins may have evolved to have quite disparate functions from their animal counterparts. Plant annexin function remains poorly understood. We can be guided by results from animals exemplified by vertebrate annexin 2 which, depending on its position within or outside the cell, its modification, binding partners or even receptor interaction executes discrete operations (Faure et al., 2002; Deora et al., 2004; Grewal & Enrich, 2009; Swisher et al., 2010). Here we review primarily plant annexins; by relating structure to in vitro functions and considering expression/localization we link to possible roles in the plant.

II. Expression and localization during development

There is little information on expression and distribution of fungal or oomycete annexins. Aspergillus niger ANXC3.1 appears to be constitutively expressed (Khalaj et al., 2004a). Abundance of P. ramorum and P. sojae annexin increases in the vegetative mycelium compared with the germinating cyst (Savidor et al., 2008). Plant annexins have been detected in all organs (reviewed by Mortimer et al., 2008). Transcript levels of annexins in Arabidopsis and Brassica differ between tissues as a function of life-cycle stage (Clark et al., 2001, 2005b; Bianchi et al., 2002; Cantero et al., 2006; Jami et al., 2009). Plant annexin expression and localization appear linked to growth and development. Differential expression or abundance has been noted, for example, in the cell cycle (Proust et al., 1999), embryogenesis (Gallardo et al., 2003), pollen and seed germination (Buitink et al., 2006; Dai et al., 2006; Yang et al., 2007), primary root growth and lateral root formation (Clark et al., 2001, 2005a; Bassani et al., 2004), tuber enlargement (Sheffield et al., 2006), development of vasculature (Clark et al., 2001, 2005b), cotton fibre elongation (e.g. Yang et al., 2008), cork formation (Soler et al., 2007), fruit ripening (e.g. Bianco et al., 2009) and petal senescence (Bai et al., 2010).

Transcriptional regulators are now being identified such as MYB98, which appears to regulate Arabidopsis thaliana Annexin8 (AtANN8) expression in synergid cells (Punwani et al., 2007). AtANN1 abundance increases in the ntm1-D mutant (NAC with transmembrane motif 1; Lee et al., 2008). In this T-DNA insertion mutant, a transcription factor that is originally membrane-tethered is constitutively active and present at the nucleus, manifesting as an inhibition of cell division. Cytokinin possibly stabilizes NTM1 during cell division and it will be interesting to see whether it relates to the accumulation of the tobacco annexins (Ntp32.1 and Ntp32.2) that accumulate at the cytokinin-regulated G1/S and G2/M cell cycle stages (Proust et al., 1999). Results from transcriptomic studies are now revealing the influence of plant growth regulators on expression, such as the upregulation of AtANN2 and AtANN4 by zeatin (Kiba et al., 2005).

Several annexin proteins have been identified in elongating cotton fibres (e.g. Yang et al., 2008). Cotton annexin (Gossypium hirsutum) GhANN1 is upregulated during rapid fibre elongation and the protein associates with Golgi-derived vesicles and plasma membrane (Andrawis et al., 1993; Shin & Brown, 1999). Annexins have been detected in the root elongation zone of Arabidopsis (Clark et al., 2005a,b) and maize (Carroll et al., 1998; Bassani et al., 2004) and are associated with expansion of tomato pericarp (Faurobert et al., 2007) and tobacco cells (Proust et al., 1999; Seals & Randall, 1997). They have also been detected (immunologically) at the apex of cells exhibiting polar growth such as fern rhizoids, pollen tubes and root hairs, leading to proposed roles in growth (Blackbourn et al., 1992; Blackbourn & Battey, 1993; Clark et al., 2001, 2005a; Dai et al., 2006). Immunolocalization of AtANN1 and AtANN2 to the periphery of highly secretory cell types (such as root hairs) is striking and correlates with areas of high levels of 3H-galactose incorporation into new wall polysaccharides, consistent with a role in growth (Clark et al., 2005a,b).

III. Response to the environment

Plant annexin expression and abundance change in response to environmental stimuli. Abundance may be lower during daytime (Hoshino et al., 2004). Light affects expression of certain Arabidopsis and tobacco annexins (Tang et al., 2003; Cantero et al., 2006) with AtANN1 and AtANN4 displaying opposite responses downstream of the photomorphogenesis transcription factor HY5 (Lee et al., 2007). Gravity affects expression and abundance of AtANN1 (Clark et al., 2005a; Kamada et al., 2005; Barjaktarovićet al., 2007). Abiotic stress stimuli are also potent regulators with differential effects on the eight Arabidopsis annexins evident at expression level, revealed by quantitative PCR (Cantero et al., 2006). Promoter analysis suggests the presence of at least one stress-responsive cis-acting element in promoter regions of Brassica juncea annexins (Jami et al., 2009). Phosphate starvation upregulates AtANN1 expression in leaves (Müller et al., 2007). Metal stress can increase expression or abundance: zinc affects Thlaspi caerulescens homologues of AtANN1 and AtANN2 (Tuomainen et al., 2010); copper affects AtANN3 and AtANN4 (Weber et al., 2006); cadmium affects AtANN1 and pea root annexin abundance (Repetto et al., 2003; Konopka-Postupolska et al., 2009).

Cold stress has been shown to induce accumulation of wheat and rice annexins (Breton et al., 2000; Hashimoto et al., 2009), affect expression in poplar leaves (Renault et al., 2006) and upregulate AtANN1,3,4,5,7,8 expression (Clark et al., 2005b; Cantero et al., 2006). By contrast, AtANN2 and AtANN6 expression is downregulated (Cantero et al., 2006). Drought changes annexin expression and/or abundance in Arabidopsis (Bianchi et al., 2002; Cantero et al., 2006; Konopka-Postupolska et al., 2009) Medicago (Buitink et al., 2006), Loblolly pine (Watkinson et al., 2003), rice (Gorantla et al., 2005) and B. juncea (Jami et al., 2009). Consistent with this, ABA also increases expression and/or abundance of AtANN1, which contains the ABA responsive cis-acting element (ABRE) in its promoter (Gidrol et al., 1996; Lee et al., 2004; Kiba et al., 2005; Konopka-Postupolska et al., 2009). Abscisic acid also has a positive effect on AtANN4 (Kiba et al., 2005; Xin et al., 2005), Medicago sativa MsANN2 (Kovács et al., 1998), BjANN1-3,6,7 (Jami et al., 2009) and the abundance of tobacco annexin NtANN12 (Vandeputte et al., 2007). AtANN1 accumulation in droughted roots is impaired in the auxin-insensitive mutant axr1-3 (single recessive mutation in chromosome 1, partial loss of function; Bianchi et al., 2002) suggesting that this annexin is downstream of the cross-talk between auxin- and ABA-signalling. Salinity stress positively affects AtANN4–8 (Cantero et al., 2006) BjANN 3 and BjANN 7 (Jami et al., 2009), MsANN2 (Kovács et al., 1998) and NtANN12 (Vandeputte et al., 2007). For AtANN1, response to salinity depends on how and when the stress is administered. An acute stress can upregulate expression but in salt-adapted roots, expression is suppressed (Lee et al., 2004; Cantero et al., 2006; Konopka-Postupolska et al., 2009; Katori et al., 2010). Oxidative stress, which may be downstream of abiotic and biotic challenge, also upregulates AtANN1, BjANN1–3 and MsANN2 (Gidrol et al., 1996; Kovács et al., 1998; Jami et al., 2009; Konopka-Postupolska et al., 2009).

Biotic challenges have distinct effects on annexins. Medicago truncatula MtANN1 is only transcriptionally active upon an application of Nod factor to root tissue and its expression is limited to roots (de Carvalho-Niebel et al., 1998, 2002). Mycorrhizal infection increases annexin abundance in pea roots and the effect of cadmium treatment is synergistic (Repetto et al., 2003). Pathogenic challenges such as Pto expression upregulate tomato annexin (Lycopersicon esculentum) LeANN34 (Xiao et al., 2001). Pseudomonas fluorescens or cucumber mosaic virus (CMV)-Y infection downregulates AtANN4 (Marathe et al., 2004; Wang et al., 2005) but Pseudomonas syringae infection upregulates (Truman et al., 2007). Wounding induces expression of AtANN4 downstream of jasmonic acid production (Yan et al., 2007) and jasmonic acid also induces AtANN2 (Kiba et al., 2005). Wounding upregulates AtANN1 and BjANN3 expression, which for the latter may be via H2O2 production (Jami et al., 2009; Konopka-Postupolska et al., 2009). Salicylic acid positively regulates AtANN1 (Gidrol et al., 1996; Konopka-Postupolska et al., 2009) but has a negative effect on AtANN2 and AtANN4 expression (Kiba et al., 2005). The most striking effect on transcription to date is the 27.9-fold increase in AtANN1 transcription at the infection site of the obligate biotroph Golovinomyces orontii (Chandran et al., 2010), revealed by transcript analysis of laser-dissected tissue. These results clearly point to annexin participation in symbiosis and defence but precise roles have yet to be elucidated.

IV. Control of subcellular localization

The cellular position of an annexin is likely to be a key determinant of its function and role. In addition to being cytosolic, plant annexins have been detected (through proteomic or immunological analyses) at the PM (Thonat et al., 1997; Santoni et al., 1998; Breton et al., 2000; Alexandersson et al., 2004; Marmagne et al., 2007; Carletti et al., 2008), tonoplast (Seals & Randall, 1997; Lin et al., 2001), Golgi and Golgi-derived vesicles (e.g. Shin & Brown, 1999), chloroplast envelope (Seigneurin-Berny et al., 2000), stroma (e.g. Rudella et al., 2006), thylakoid (Friso et al., 2004), nucleus (Clark et al., 1998; Kovács et al., 1998), glyoxysome (Fukao et al., 2003), mitochondria (Ito et al., 2006) and peribacteroid membrane (Wienkoop & Saalbach, 2003). AtANN1 has been recovered from PM, vacuole, glyoxysome, mitochondria, stroma and thylakoid (albeit from different tissues), indicating that a given annexin could exist in several cellular positions (reviewed by Laohavisit & Davies, 2009).

Saprolegnia annexin has been recovered from PM and several 35 kDa isoforms have been recovered from PM detergent-resistant microdomains (DRM; Briolay et al., 2009). Sterol-enriched DRM operate in signal transduction and membrane trafficking. Various animal annexins have been found associated with DRM, where they may be involved in membrane organization or local Ca2+ influx (Zhao & Hardy, 2004). For plants, only MtANN2 has been recovered from a DRM (Lefebvre et al., 2007). In common with animal annexins (Babiychuk et al., 2009), there is now evidence that plant annexins can undergo stimulus-dependent relocation from the cytosol to a membrane. Immunological analysis of cell fractions has revealed that salt stress prompts Ca2+-dependent reversible relocation of AtANN1 from cytosol to root cell membranes (Lee et al., 2004). Mechanical stress causes relocation of Bryonia dioica annexin from the cytosol to the parenchyma PM (Thonat et al., 1997) and this may be Ca2+-dependent. In mimosa, annexins are largely cytosolic at night but occupy the periphery of motor cells of the pulvinus during daytime, possibly to contribute to the latter’s nyctinastic movement (Hoshino et al., 2004). Gravitational stimulus changes the positioning of annexin in cells below the pea plumule apical meristem so that they are more evenly distributed at the periphery (Clark et al., 2000). MsANN2 localizes in the nucleolus upon stress conditions (Kovács et al., 1998). Cold causes accumulation (as integral proteins) of wheat annexin p39 in shoot PM and p22.5 in a microsomal fraction but it is not clear how this is achieved (Breton et al., 2000).

Stimulus-dependent transit time to a membrane can be determined by an animal annexin’s Ca2+ affinity (100 nM to 20 μM; Gerke et al., 2005). Location of membrane association and function are influenced by phospholipid head-group specificity, local [Ca2+], pH, voltage and membrane curvature (Hofmann et al., 1997; Maffey et al., 2001; Fischer et al., 2007; Babiychuk et al., 2009). For plant annexins, [Ca2+] required for half-maximal binding varies with lipid and pH from submillimolar (Capsicum annuum CaANN24, GhANN1; Dabitz et al., 2005) to nanomolar (tobacco VCaB42, Seals et al., 1994; AtANN1, Gorecka et al., 2005). Mildly acidic pH (6.0) decreases the Ca2+ requirement (Blackbourn et al., 1991; Laohavisit et al., 2010). An absence of classical targeting sequences has lent support to the idea that highly local conditions regulate the location of annexin–membrane interaction. A clear example comes from spinach annexin which binds to a chloroplast-specific sulpholipid enabling localization to the organelle’s outer surface (Seigneurin-Berny et al., 2000). Analysis of plant annexin sequences also reveals a C-terminal diacidic motif (Fig. 1) that has recently been recognized as an important ER export signal to the PM (Mikosch & Homann, 2009). This motif (D/E-X-D/E– where X denotes any amino acid) is found in several plant PM K+ channels and aquaporins (Mikosch & Homann, 2009). AtANN1, ZmANN33 and ZmANN35 contain the motif and have been detected at the PM (Santoni et al., 1998; Alexandersson et al., 2004; Benschop et al., 2007; Carletti et al., 2008). However, these three annexins plus AtANN2 have been predicted by secretomep software (http://www.cbs.dtu.dk/services/SecretomeP/) to be non-classical secreted proteins that can potentially become extracellular (Laohavisit et al., 2009). Both AtANN1 and AtANN2 have been identified in the cell wall (Kwon et al., 2005; Bayer et al., 2006) while 15N metabolically labelled AtANN1 has been found in washing fluid from leaves, suggesting an apoplastic location (Bindschedler et al., 2008). An annexin from Phytophthora ramorum has also been detected in the cell wall (Meijer et al., 2006).

Figure 1.

 Alignment of the amino acid sequences of plant annexins with vertebrate annexin A5 (ANXA5) and partial sequence from horseradish peroxidase. The putative annexin repeats (Repeat I to Repeat IV) are shown beneath the sequence. The sequences highlighted are as follows, with supporting references in the main text: teal, Ca2+-binding sequences with the phosphatidylserine (PS)-binding motif demarcated with arrows; black box, conserved tryptophan required for Ca2+-independent membrane binding; purple, putative S3 cluster thought to be involved in redox reactions; pink box, diacidic motif as an important endoplasmic reticulum (ER) export signal to the plasma membrane; green, salt bridges involved in channel function of animal annexins; yellow, IRI motif for binding actin; orange box, putative GTP binding motif; blue box, the K-R-H/G/D motif which interacts with the C2 domain containing protein; red, comparing haem-binding domain of peroxidase from Armoracia rusticana with the N-terminus of annexins with identical residues highlighted and the conserved His residue marked (*). Amino acid sequence alignment was performed using clustalw and edited in jalview. Accession numbers are as: ANXA5 (gi:4502107); AtANN1 (gb:NP174810); AtANN2 (gb:201307); CaANN24 (gb:CAA10210); ZmANN33 (gb:CAA66900); ZmANN35 (gb:CAA66901); GhANN1 (gb:AAC33305); MtANN1 (gi:22859608); LeANN34 (gb:AAC97494); LeANN35 (AAC97493); A. rusticana horseradish peroxidase (gb:CAA00083).

In common with AtANN1, vertebrate A2 can occur at the PM or extracellular matrix but lacks a classical secretion signal. Binding to its p11 partner (S100A10) enables non-vesicular transit to the inner PM face where N-terminal tyrosine phosphorylation (possibly by a membrane-associated kinase) prompts translocation to the outer membrane face (Deora et al., 2004). Release may also be via inclusion in exocytotic vesicles (Faure et al., 2002). Vertebrate A1 can be exported to the extracellular matrix by ABC transporters (Wein et al., 2004), but this possibility remains untested for plants, fungi and oomycetes. A range of factors (notably Ca2+, pH, lipid, reactive oxygen species (ROS), voltage, motifs and post-translational modification) therefore seem likely to control plant, fungal and oomycete annexin positioning and hence position-dependent function.

V. Structural determinants of function

1. Ca2+ and lipid binding

X-ray crystal structures have been obtained for cotton and Capsicum annexins (Hofmann et al., 2000a, 2003; Hu et al., 2008), indicating formation of a slightly curved disc with type II and type III Ca2+-binding sites exposed at the convex side, similar to their animal counterparts (Hofmann et al., 2000a, 2003; Hu et al., 2008; Fig. 2). The short N-terminal domain appears anchored into the C-terminal core by hydrogen bonding (Fig. 2; Hofmann et al., 2003) and is unaffected by coordination of Ca2+ ions (Hu et al., 2008). Up to four ‘annexin repeat(s)’ are involved in Ca2+ binding (Fig. 1). The repeat (c. 70 amino acids), contains a consensus endonexin sequence referred to as K-G-X-G-T-{38}-D/E. Each repeat contains five short α-helices joined together by short loops. These loops are necessary for Ca2+–annexin interaction, particularly the AB loop (i.e. a short loop connected to helix A and helix B) and the DE loop (Hofmann et al., 2003; Hu et al., 2008). The α-helices make up the majority of annexin secondary structure.

Figure 2.

 The crystal structure of Ca2+-bound cotton annexin GhANN1. Structures are shown from the side (top left) or from above (top right). The molecule forms a slightly curved disc with four distinct repeats. The N-terminal (yellow) and the end of the C-terminal domain (dark blue) interact together. The four repeats are shown in colour (red, repeat I; silver, repeat II; magenta, repeat III; cyan, repeat IV). Ca2+-dependent lipid binding is mediated by three Ca2+ ions (green spheres) binding in the first and fourth repeats while the fourth Ca2+ ion is structurally possible in loop I-DE. Some of the important features of plant annexins are highlighted and magnified in the coloured box inserts : green box, N-terminal interaction with C-terminus (Hu et al., 2008); red box, the S3 cluster (Hofmann et al., 2003); yellow box, salt bridge interactions (Laohavisit et al., 2009); blue and magenta boxes, critical residues for plant annexin membrane association (Dabitz et al., 2005; Hu et al., 2008). The residues are marked (using GhANN1 numbering) and black dashed lines represent possible hydrogen bonding. The side-chain residues are marked (using GhANN1 numbering) and coloured with red, blue and yellow atoms on the amino acid side-chain that represent oxygen, nitrogen and sulphur, respectively. The crystal structures are taken from the PBD database (3BRX; Hu et al., 2008) and structures were edited and viewed using PyMol.

The vertebrate annexin A5 PS binding-site ([R/K]XXXK-BC-helices-[R/K]XXXXDXXS[D/E]) is well conserved in some of the fungal annexins (Khalaj et al., 2004b) and variable in plant annexins (Fig. 1). The four repeats of plant annexins are not equivalent in lipid binding (Lim et al., 1998). Repeats I/IV and II/III may act as discrete lipid-binding modules (Hofmann, 2004; Hu et al., 2008). For GhANN1, the I/IV repeat pair is responsible phospholipid binding (Hu et al., 2008) and harbours three Ca2+ co-ordination sites. A fourth site is thought to occur in the I-DE loop (Hu et al., 2008). The I-AB loop may have conformational flexibility important for rapid phospholipid interaction. Trp 35 (W35) (W27 in GhANN1 numbering in Fig. 2) appears critical for Ca2+-dependent lipid binding as well as for Ca2+-independent binding (Hofmann et al., 2003; Dabitz et al., 2005; Hu et al., 2008). The Ca2+-independent binding may also require two other tryptophan residues that are exposed on the convex side of the proteins as well as basic residues, such as K190, R262 and R263 (Dabitz et al., 2005; Fig. 2). Plant annexins show Ca2+-independent membrane binding at neutral and acidic pH; up to 20% of total GhANN1 and CaANN24 are bound without Ca2+ at neutral pH (Blackbourn et al., 1991; Breton et al., 2000; Hofmann et al., 2000a, 2002; Dabitz et al., 2005; Gorecka et al., 2007; Laohavisit et al., 2009). This suggests that there can be resident annexin populations at membranes ready to sequester Ca2+ and perhaps undergo a change of function as a consequence.

Acidic pH induces Ca2+-independent conformational changes and adoption of transmembrane forms by some animal annexins (Golczak et al., 2001; Ladokhin et al., 2002; Isas et al., 2003; Patel et al., 2005). The membrane-inserted form of annexin B12 is a monomer, unlike the trimeric membrane-attached form (Isas et al., 2003). The B12 monomer can also exhibit a partially inserted, membrane peripheral form at pH 5.5 that can be interconverted between the cytosolic form at pH 6.5 and the transmembrane form at pH 4 (Hegde et al., 2006). Such complex interactions with membranes may relate to ion channel function (Section VI.4). Plant annexins have not been studied so systematically but are known to exist in the fully or partially membrane-inserted form (wheat p39 and p22.5, Breton et al., 2000; AtANN1, Santoni et al., 1998; Gorecka et al., 2007).

2. N-terminus and post-translational modification

In common with animal annexins, the N-terminal part of the protein may be also be involved in membrane binding (Kovács et al., 1998; Hofmann et al., 2000a; Hofmann, 2004). Length and structure of animal annexin N-termini are highly variable and thought to function in post-translational modification and protein–protein interactions, particularly with actin and the EF-hand S100 family (Hawkins et al., 2000; Gerke & Moss, 2002). For plant annexins, actin binding could be due to a conserved IRI motif (Calvert et al., 1996; Clark et al., 2001; Fig. 1) but its presence has not always correlated with actin binding in vitro (addressed by Mortimer et al., 2008). The N-terminus of fungal annexin C4 (present in A. fumigatus and probably also Neurospora crassa and Magnaporthe grisea) is significantly longer than most animal annexins (up to 553 amino acids) but the significance of this is unknown (Khalaj et al., 2004b). In comparison, the N-terminal region of plant annexins is short (c. 10 amino acids) but divergent and may be important for functional diversity.

Brassica napus annexin is phosphorylated at the N-terminal of the second annexin repeat (AVMLWT*LDPPER where T is phosphorylated) (Agrawal & Thelen, 2006); this sequence is fully conserved in Arabidopsis AtANN2 (Fig. 1). Sequence analysis suggests that a plant annexin can contain multiple putative sites for phosphorylation (Jami et al., 2009) and phosphorylation status has been linked to regulation of in vitro peroxidase activity (Gorecka et al., 2005; see Section VI.3). GhANN1 can bind to cotton fibre PM and be phosphorylated by a PM-associated kinase (Andrawis et al., 1993). This suggests regulation of annexin activity by post-translational modification at specific cellular positions to direct function. Rice annexins have been shown to interact with several different kinases, including MAPKK and Ste20-related protein kinase, pointing to involvement in a membrane-associated, Ca2+-dependent MAPK signalling cascade (Rohila et al., 2006).

Plant annexins contain two conserved cysteine residues (Fig. 1) that are targets for post-translational modification. The Cys residues of AtANN1 can be S-nitrosylated or S-glutathionylated with the latter occurring downstream of ABA (Lindermayr et al., 2005; Konopka-Postupolska et al., 2009; Clark et al., 2010). Although potentially capable of mediating intramolecular or intermolecular disulphide bonds, the cys residues of AtANN1 and GhANN1 appear in reduced form under standard conditions (Hofmann et al., 2003; Konopka-Postupolska et al., 2009). Mutation of AtANN1 cys residues has no significant impact on tertiary or quaternary structure (Konopka-Postupolska et al., 2009). Their significance may lie in regulating membrane binding as S-glutathionylation decreases AtANN1’s Ca2+ affinity for lipids and may act to restrict recruitment to membranes (Konopka-Postupolska et al., 2009).

3. Sulphur cluster and copper binding

The cys residues may contribute to activity of an S3 cluster (two cys residues with methionine and tyrosine, MCCY; Hofmann et al., 2003; Figs 1, 2) that has been proposed to operate in electron transfer and redox reactions (Hofmann et al., 2003; see Section VI.3). Peroxidase activity of plant annexins was once ascribed to a conserved N-terminal His40 residue (Fig. 1: not present in animal annexins) thought to be involved in haem association but recent analyses show that peroxidase activity is independent of both His40 and haem (Konopka-Postupolska et al., 2009; Laohavisit et al., 2009; Mortimer et al., 2009). Rather, His40 has now been proposed to help maintain secondary structure (Konopka-Postupolska et al., 2009; Clark et al., 2010). Another possibility is that copper coordinated with an annexin could react with peroxide. AtANN1 has been reported to be a copper-binding protein (in common with vertebrate annexins A2, A4 and A5; She et al., 2003) with binding possibly mediated by His- or Met-bounded motifs (Kung et al., 2006). AtANN1 does not bind zinc, suggesting specificity (Kung et al., 2006). The functional implications of this have not been explored either in terms of redox activity or copper sequestration.

4. Nucleotide binding and C2 interaction

Nucleotide phosphodiesterase activity of plant annexins has been reviewed by Mortimer et al., (2008). AtANN1 binds ATP but the structural basis is unclear (Ito et al., 2006). The putative GTP binding motif is ‘GXXXXGKT and DXXG’, the Walker A motif and the GTPase superfamily GTP-binding motif, respectively (Clark et al., 2001). The GTP-binding motif can overlap the Ca2+-binding motif of the fourth repeat (Fig. 1; Shin & Brown, 1999) suggesting that membrane binding could modulate GTPase activity (Calvert et al., 1996; Lim et al., 1998). The binding of F-actin does not affect tomato annexin nucleotide diphosphatase activity (Calvert et al., 1996), possibly allowing actin to ‘place’ GTPase activity in the cell. In zucchini (Cucurbita pepo), two PM-associated annexins also bind to F-actin in vitro (Hu et al., 2000), which could mean that annexins are part of the apparatus connecting the PM to the cytoskeleton. Animal annexins can interact with C2 domain-containing proteins to modulate their activity (Morgan et al., 2006). C2 permits Ca2+-dependent membrane binding and enhances activity of coupled enzymatic domains. The basis for annexin–C2 interaction is thought to be the K/R/H-G-D motif in the Ca2+-binding repeats and N-terminus (Morgan et al., 2006). AtANN1, AtANN7 and CaANN24 contain the consensus K-R-H/G/D motif (Fig. 1). Others contain a slightly modified version (ZmANN33; S/G/D motif) and some lack it (AtANN5, ZmANN35). Plant proteins containing C2 domains include phospholipases and protein kinases. Capsicum annexin p35 inhibits porcine pancreatic phospholipase (PLA) A2in vitro (Hoshino et al., 1995), suggesting that PLAs may be annexin targets in the cell. An association with annexins could have implications for fine spatial control of lipid-based signalling or, by analogy with animal studies, membrane dynamics.

VI. Activities and possible functions

There is a dearth of information on fungal annexins. Aspergillus niger lacking ANXC3.1 shows no obvious phenotype (gene deletion; Khalaj et al., 2004a) and there are no reports on protein activities. Most of what we know about plant and oomycete annexin activities comes from in vitro studies, some of which have been discussed in earlier sections. Protein partners are being identified but structural basis and consequences may not yet be understood (e.g. barley annexin interaction with a 14-3-3 protein; Schoonheim et al., 2007). Here we examine the best-studied activities, relate structure to function as far as possible and synthesize with expression and localization studies in assessments of cellular roles.

1. Membrane dynamics

The simplest assignment of annexin function is, as their name suggests, an ability to bring membranes together. This suggests roles in membrane repair, constitutive or stress-induced autophagy, stromule formation and exo-/endo-cytosis. Secretion from the A. niger anxc3.1 mutant appears normal (Khalaj et al., 2004a), suggesting that other or no annexins participate in exocytosis. However, plant annexins cause aggregation of liposomes and secretory vesicles in vitro (Blackbourn & Battey, 1993; Hoshino et al., 1995). ZmANN33/35 stimulate exocytosis in root cap protoplasts (estimated by membrane capacitance) and this can be prevented by nonhydrolysable GTP analogues (Carroll et al., 1998). This is consistent with Ca2+ and GTP competing for binding (see Section V.4) and suggests a mechanism for fine spatial control of exocytotic activity. Cotton annexin associates with PM and dictyosome-derived coated vesicles involved in fibre elongation (Shin & Brown, 1999) implicating it in exocytosis. Tobacco Ntp32.1 and Ntp32.2 correlate positively with cell division and lie beneath the PM perhaps to contribute to exocytosis (Proust et al., 1999). The presence of annexin during cork formation could reflect a role in cell proliferation and expansion, perhaps through exocytosis, but could also indicate involvement in these cells’ irreversible commitment to cell death (Soler et al., 2007).

A loss of function Atann1 mutant has a short root phenotype and AtANN1 is implicated in gravitropic-induced root bending by stimulating oligosaccharide secretion (T-DNA knockout; Clark et al., 2005a,b). Given that AtANN1 is downstream of AXR1 in roots (Bianchi et al., 2002; see Section III), it is feasible that it contributes to drought-induced changes in root architecture. The increased abundance of AtANN1 in the ntm1-D mutant (Lee et al., 2008; see Section II) could relate to the increased cell length of this mutant rather than its inhibition of cell division. In addition, Atann1 displays stress-induced germination delay (T-DNA knockout; Lee et al., 2004) that could implicate this annexin in exocytosis or cell cycle control. The mutation can be complemented by 35S-driven expression of a nematode annexin (Patel et al., 2010) which may in future help explain mode of action. Whether AtANN1 is involved in PM dynamics at the site of Golovinomyces penetration (where its transcript is dramatically upregulated: Chandran et al., 2010; see Section III) now needs testing.

An Atann2 mutant has reduced hypocotyl elongation and AtANN2 is implicated in hypocotyl oligosaccharide secretion during gravitropism, suggesting a role in exocytosis (T-DNA knockout; Clark et al., 2005a,b). Gravistimulated redistribution of pea annexin could relate to regulation of exocytosis for growth (Clark et al., 2000, Clark et al., 2005a,b). Increased longitudinal growth of potato tubers (as a consequence of manipulating extracellular ATP as a growth regulator) is associated with upregulation of annexin transcription (Riewe et al., 2008), which may reflect a role for the annexin in cell elongation. For endomembranes, tobacco annexin VcaB42 at the tonoplast (associated with a ROP GTPase that could act as a molecular ‘on/off’ switch) is likely to play a role in vacuolar biogenesis for cell expansion (Seals & Randall, 1997; Lin et al., 2001). With so many incidences of annexin expression and abundance at growth points, further research on their function in growth is merited.

2. Regulation of complex carbohydrates

A recent study revealed that barley annexin p33 expression is upregulated in the endosperm of the starch synthase IIa mutant during seed filling (Clarke et al., 2008). The carbohydrate profile of the grain is perturbed and although harder, the grain is shrunken. Whether this points to a role for the annexin in regulating carbohydrates or as a negative regulator of growth remains to be determined but deserves further investigation as the mutant’s starches appear beneficial to animal health (Clarke et al., 2008). The impact of annexins on callose is better documented. Cotton (GhANN1) and Saprolegnia annexins have been shown to inhibit or stimulate callose ((1→3)-β-d-glucan) synthase activity (determined through radiolabelled UDP-glucose incorporation), respectively. GhANN1 association with cotton fibre PM callose synthase was Ca2+-dependent, while the oomycete annexins existed in a DRM with the callose synthase (Andrawis et al., 1993; Shin & Brown, 1999; Bouzenzana et al., 2006; Briolay et al., 2009). Oxylipin increases AtANN7 expression and callose formation (the latter is determined with aniline blue staining or using anticallose antibodies) but there is no direct link to callose synthase regulation (Vellosillo et al., 2007). Similarly, the timing of increased annexin abundance in flax bast fibre (phloem) growth is consistent with a role in secondary wall synthesis but this has not been tested directly (Hotte & Deyholos, 2008). Regulation of callose synthase implicates annexins in cell plate formation, sieve tube occlusion, growth of pollen tubes, cotton and bast fibre growth, response to pathogens, wounding and cell death. Hence it may be significant that annexins are evident in the cell cycle, are found in phloem, localize at polar growth points and are upregulated during biotic stress responses. Whether conferring resistance to oomycete infection by ectopic expression of plant annexin (Jami et al., 2008) involves increased callose production has not yet been tested.

3. Redox reactions and regulation of reactive oxygen species

Peroxide induces dimerization of AtANN1 and oligomerization of ZmANN33/35 (Gorecka et al., 2005; Mortimer et al., 2009), and peroxide is also an in vitro substrate. Recombinant AtANN1 has weak peroxidase activity regulated by phosphorylation (Gidrol et al., 1996; Gorecka et al., 2005; Konopka-Postupolska et al., 2009). In vitro peroxidase activity has also been detected from the maize annexin doublet ZmANN33/35, BjANN1, GhANN1 and CaANN24 (Laohavisit et al., 2009; Mortimer et al., 2009). Maximal reaction velocity of (soluble) ZmANN33/35 peroxidase is two orders of magnitude lower than that of horseradish peroxidase (HRP) but the Km (15 μM) is half and in line with the ascorbate peroxidases that help regulate cellular redox levels (Mortimer et al., 2009). Calcium stimulated the peroxidase activity of soluble and lipid-associated ZmANN33/35 (Mortimer et al., 2009). Populations of membrane-bound and soluble annexins may therefore allow spatial fine-tuning of ROS within cells and organelles such as the chloroplast. As actin is sensitive to ROS (Franklin-Tong & Gourlay, 2008), it is feasible that actin-bound annexin peroxidase could affect the cytoskeleton’s role in ROS signalling.

Consistent with peroxidase function, heterologous AtANN1 expression offers protection against oxidative stress to the H2O2-sensitive Escherichia coliΔoxyR mutant or mammalian cells (Gidrol et al., 1996; Kush & Sabapathy, 2001; Konopka-Postupolska et al., 2009; Clark et al., 2010). In guard cells, ABA acts to close stomata with ROS generation as an important positive step. Guard cells of a T-DNA insert Atann1 knockout mutant accumulate more intracellular ROS when challenged with ABA or exogenous H2O2 than wild type, which in turn accumulate more than AtANN1 overexpressor lines (Konopka-Postupolska et al., 2009). This is broadly consistent with cytosolic AtANN1 peroxidase function. However, while it might be predicted that ABA-induced stomatal closure would be impaired in the overexpressors as they accumulate less ROS and hence they would be more drought sensitive, they are more drought tolerant (Konopka-Postupolska et al., 2009). Atann1 is more drought-sensitive than wild type. This points to a more complex relationship between ROS-mediated ABA-induced stomatal closure and altered AtANN1 expression. Constitutive expression of BjANN1 (a possible peroxidase) in tobacco confers drought, salinity, cadmium (Cd2+) and oxidative stress tolerance on the T1 seedlings and better ability to resist Phytophthora-induced necrosis (Jami et al., 2008). Lipid peroxidation was reduced and chlorophyll content was maintained suggesting enhanced ability to sense/signal or degrade ROS at membranes and in the chloroplast (Jami et al., 2008). The mustard annexin found in the chloroplast RNA polymerase complex has been proposed to protect transcription from oxidative damage (Pfannschmidt et al., 2000).

4. Ion transport and Ca2+ signalling

Several animal annexins form Ca2+-permeable channels in artificial membranes, regulated by ATP, GTP, peroxide, pH and voltage (reviewed by Kourie & Wood, 2000; Hegde et al., 2006; Kirilenko et al., 2006). Mechanisms of channel formation include monomer attachments disturbing the bilayer allowing ion flux through the protein’s central pore (Kourie & Wood, 2000), GTP-dependent nonpenetrative trimer formation (Golczak et al., 2001; Kirilenko et al., 2006) and low pH-dependent insertion (Golczak et al., 2001; Gorecka et al., 2007). Channel gating and selectivity involves two conserved salt bridges (D92-R117 and E112-R271) located within the central pore (Liemann et al., 1996; Fig. 1). These are conserved in plant annexins (Figs 1, 2). Early work showed that recombinant CaANN24 mediates a Ca2+-permeable pathway in vesicles (Hofmann et al., 2000a). Recombinant AtANN1 inserts into planar lipid bilayers at acidic pH to form a K+-permeable conductance (Gorecka et al., 2007). As a pH-sensitive K+-channel, AtANN1 has been proposed to act in stress scenarios involving cytosolic acidosis or as a pH sensor (Gorecka et al., 2007). Its activity could help regulate membrane voltage or contribute to plasma membrane K+ efflux as a consequence of pathogen attack. Whether nitric oxide (NO) regulates activity through S-nitrosylation (Lindermayr et al., 2005) is unknown. However, S-glutathionylation appears to decrease AtANN1’s Ca2+ affinity (Konopka-Postupolska et al., 2009); this could decrease local numbers of AtANN1 capable of membrane insertion, limiting the extent of channel activity.

Maize annexins ZmANN33/3 can increase [Ca2+]cyt (possibly as a signalling second messenger) and form a Ca2+- and K+-permeable conductance in planar lipid bilayers with a low selectivity for Ca2+ over K+, resembling nonselective Ca2+-permeable cation channels (NSCC) (Laohavisit et al., 2009, 2010). The NSCC are thought to have diverse roles including root Na+ influx (Demidchik & Maathuis, 2007). Incorporating malondialdehyde (MDA; a breakdown product of peroxidized polyunsaturated fatty acids) changes the voltage dependence of maize annexins so that the conductance resembles the ROS- and membrane hyperpolarization-activated NSCC that are implicated in ROS signalling and Ca2+-driven polar growth (Laohavisit et al., 2010; Fig. 3). The occurrence of PM ROS-activated Ca2+-permeable NSCC agrees well with AtANN1 distribution making this annexin a prime candidate for the molecular origin of these channels. In addition, the occurrence of Medicago annexins in nuclear membranes, and the possibility that they too form channels (Fig. 1), has lead to the proposal that they could be involved in symbiotic Ca2+ signalling (de Carvalho-Niebel et al., 2002; Talukdar et al., 2009). To establish these links, the gap between in vitro channel studies and electrophysiological analysis of native (mutant) plasma and endomembranes must be bridged.

Figure 3.

 A stylized current–voltage relationship illustrating the effect of membrane peroxidation on maize annexin channel activity. Based on data from Laohavisit et al. (2009, 2010), the annexin conductance may show a near-ohmic relationship in a control plasma membrane mimetic (dotted line) but rectifies inwardly (solid line) in response to membrane peroxidation. With Ca2+ as the charge carrier, this would alter the amount of Ca2+ entering the cell (negative current) over the physiological membrane voltage range thus affecting cytosolic free Ca+, [Ca2+]cyt, signal generated. Vm, membrane voltage; I, current.

Annexins that have undergone Ca2+-dependent membrane binding could also contribute to [Ca2+]cyt signalling on their release. The mechanisms of release have not been tested in plants, fungi or oomycetes but could involve phospholipase activity such as PLA2 which targets PC and promotes cell death (Reina-Pinto et al., 2009). Eversion of PS to the extracellular PM face during cell death (Ning et al., 2002; Reina-Pinto et al., 2009) could promote binding and perhaps channel formation of the extracellular plant annexins. Indeed, the ZmANN33/35 Ca2+ conductance resembles that thought to operate at the PM in maize root cell death and these annexins could be extracellular (Laohavisit et al., 2009; see Section IV). In animal cells, caspase can trigger the release of annexin A2 to the extracellular membrane face (Keller et al., 2008), and with caspase-like enzymes featuring in plant cell death (Franklin-Tong & Gourlay, 2008) it will be interesting to see whether this mechanism pertains in plant cells. More simply, cytosolic acidosis during cell death could promote channel formation at the PM from the intracellular face. A homolog of CaANN24 has recently been found to increase in abundance during pollination-induced petal senescence of Petunia (Bai et al., 2010), suggesting a role in cell death. Again, position and local conditions will be critical to determining annexin function.

VII. Conclusions and prospects

The emergent pattern is of annexin operation in development, stress and defence, with acute and long-term responses evident. Whether plant annexins serve to integrate these processes now needs to be examined. Plant annexins may promote other proteins to bind tightly to a membrane, forming a ‘multidomain cooperation’. Exploring annexin interaction with lipids will be pivotal to understanding their interdependence, both as a function of membrane remodelling during stress and as a way of interlinking Ca2+, ROS, actin and lipid signalling. The lipid bilayer has a significant role to play in modulating Ca2+ transport by annexins and this could fine-tune a resultant [Ca2+]cyt increase to encode a specific signal. The interaction of annexins with lipid-modifying enzymes such as the PLAs or PLDs will help define the regulatory systems governing annexin/lipid association and elucidate downstream responses. That animal extracellular annexins may act through receptors (Swisher et al., 2010) introduces another direction to explore in plants. With secretion of annexins now shown for plant pathogens such as Phytophthora (Meijer et al., 2006), the interplay of plant and pathogen annexins at the infection point could be relevant to immunity and resistance. Resolving real-time changes in distribution within an individual cell is now required as a first step in elucidating events. Determining the extent to which stimulus-dependent relocation to a membrane is determined by targeting and vesicle delivery over local recruitment of a soluble form will be a significant challenge.

Acknowledgements

We thank Dr Haruko Tamura for her assistance in crystal structure analysis. This work was supported by the University of Cambridge, UK.

Ancillary