Patterns of [Ca2+]cyt changes are often remarkably similar across species. For example, when treated with 20 mm caffeine under certain conditions, the green alga Eremosphaera viridis displays [Ca2+]cyt oscillations. These reach [Ca2+]cyt peaks of c. 700 nm, display a period of c. 1 min and are dependent on extracellular Ca2+ (Bauer et al., 1997). They thus bear a remarkable similarity to the response of rat neuronal cells to 5 mm caffeine (Fig. 2), which also show Ca2+ext-dependent [Ca2+]cyt oscillations with periods of c. 1 min, although the peak [Ca2+]cyt is only c. 200 nm (Usachev & Thayer, 1999).
Two extremes bound any explanation for such similarities. The first extreme is homology, in which two patterns in two different organisms are similar because they are generated by the same pathway. This pathway may have been inherited by each organism from a common ancestor, in which case it is called orthologous, or it may have been independently constructed in each organism from the same components, a process which often reflects developmental constraints and which is called parallelism. The second extreme is homoplasy, also known as convergence, in which the efficacy of natural selection in moulding an optimal characteristic leads to the creation of similar patterns from different pathways (Hall, 2003).
1. Probable [Ca2+]cyt signalling homology: nodulation in legumes
Around 30–60 Myr ago (Wikström et al., 2001), one eudicot clade of flowering plants developed the ability to host nitrogen-fixing bacteria in a symbiosis characterized by the formation of distinctive nodules on the root hairs of infected plants (Soltis et al., 1995). Their descendants include the legumes, a group of crop plants, which are commercially important because the nitrogen-fixing nodules allow them to be grown on poor soils. The formation of nodules is triggered by a bacterial invitation to prospective hosts, in which the bacteria release lipo-chito-oligosaccharide chains, called Nod (for nodulation) factors, and the plants respond by diverting the normal process of root hair development to encompass their bacterial allies.
As part of this diversion of normal development, root hairs display a variety of [Ca2+]cyt elevations which are thought to trigger some, or all, of the necessary changes in metabolism and gene expression. It is currently unclear whether all these [Ca2+]cyt elevations are produced by a single Nod-factor-activated pathway, or whether multiple pathways operate in parallel (Wais et al., 2000; Esseling et al., 2004). Nonetheless, challenging root hairs of Medicago sativa (Ehrhardt et al., 1996), Phaseolus vulgaris (Cárdenas et al., 1999), Medicago truncatula (Wais et al., 2000; Shaw & Long, 2003), Pisum sativum (Walker et al., 2000) or Lotus japonicus (Harris et al., 2003) with nanomolar amounts of Nod factors results in a lag of c. 10 min followed by the onset of ‘[Ca2+]cyt spiking’: nuclear localized [Ca2+]cyt elevations which recur, with periods of between 0.5 and 3 min, for a couple of hours (Fig. 3a). In some cases, a distinct second response (Cárdenas et al., 1999; Shaw & Long, 2003) results in [Ca2+]cyt elevations near the root hair tip (Fig. 3b).
Figure 3. Interspecific comparison of Nod factor-induced [Ca2+]cyt spiking. Phaseolus vulgaris root hairs (on the right) were microinjected with fura-2 70 kDa dextran (© Blackwell Science Ltd. Reproduced, with permission, from Cárdenas et al., 1999); Medicago truncatula root hairs (on the left) were microinjected with Calcium Green-1 10 kDa dextran and Texas Red 10 kDa dextran (© American Society of Plant Biologists. Reproduced, with permission, from Shaw & Long, 2003). Each root hair is approx. 10 µm wide. (a) About 10 min after application of 10 nm Nod factor, [Ca2+]cyt spiking is seen in the perinuclear region of root hairs, and oscillations can persist for several hours. (b) Within 10 min of application of 10 nm Nod factor, [Ca2+]cyt elevations are sometimes seen in the tips of root hairs (Ehrhardt et al., 1996; Walker et al., 2000).
Download figure to PowerPoint
This conservation of [Ca2+]cyt patterns seems to be matched by conservation of the toolkit components identified so far, which suggests that these responses are homologous. Work on L. japonicus nodulation mutants has shown that a pair of Nod factor receptor kinases, NFR1 (Radutoiu et al., 2003) and NFR5 (Madsen et al., 2003), are required to trigger nodulation. LYK3 and LYK4, the LysM domain-containing receptor-like kinases of M. truncatula, are NFR1/5 orthologues (Limpens et al., 2003), and, again, without these nodulation will not occur. The identity between NFR1/5 and LYK3/4 is not exact; NFR5 has three LysM domains (Madsen et al., 2003), but NFR1 (Madsen et al., 2003), LYK3 and LYK4 (Limpens et al., 2003) have only two. Given that the LysM domain is thought to be responsible for oligosaccharide binding (Radutoiu et al., 2003), such variation presumably underlies receptor specificity, and may explain why chitin oligomers whose structures are closely related to those of Nod factors elicit [Ca2+]cyt spiking in P. sativum (Walker et al., 2000) and M. truncatula (Oldroyd et al. 2001) but not in P. vulgaris (Cárdenas et al., 1999).
A more contentious analysis of [Ca2+]cyt-spiking mutants (Wais et al., 2000 vs Esseling et al., 2004) has revealed three further loci named DMI1-3, which do not make bacterial infections and which have orthologues in all legumes studied to date. DMI1 is a novel protein which may form a cation channel and probably interacts with other proteins (Anéet al., 2004). DMI2, which is also referred to as SYMRK (symbiosis receptor-like kinase) or NORK (nodulation receptor kinase), is a membrane-bound serine/threonine protein kinase (Endre et al., 2002; Kistner & Parniske, 2002; Stracke et al., 2002); DMI3 is an orthologue of the Pisum sym9 gene (Mitra et al., 2004), and a member of the [Ca2+]cyt sensor-responder family known as the Ca2+ and calmodulin-dependent protein kinases, or CCaMKs (Lévy et al., 2004).
Exactly which [Ca2+]cyt signalling pathway DMI1-3 belong to is still unclear. Until recently, DMI1-3 were thought to be components in the nuclear [Ca2+]cyt-spiking pathway (Wais et al., 2000), but this view may need revision, following the discovery that DMI mutants are able, when handled extremely delicately, to display root hair deformation (Esseling et al., 2004). DMI1-3 may be involved in the tip-high [Ca2+]cyt pathway, rather than the nuclear one, although this remains to be confirmed either way (Shaw & Long, 2003; Esseling et al., 2004).
2. Probable [Ca2+]cyt signalling homoplasy: Ca2+ release from endomembrane stores
While homology is often an intuitively appealing explanation for any similarities which exist between stimulus-evoked [Ca2+]cyt patterns, convergent evolution of complex traits is not unknown in either the animal or plant world, explaining, among other things, similarities between transcriptional regulation in bacteria and yeast (Conant & Wagner, 2003) and intelligence in animals (Emery & Clayton, 2004).
It is likely that a further example of convergence occurs in the mechanisms by which [Ca2+]cyt elevations are propagated. There is little common morphological ground to be found between metazoa and plants, which diverged between 1 and 2 Gyr ago (Feng et al., 1997; Sanderson, 2003), but the early development of both begins with a sperm fertilizing an egg. In metazoa, sperm entry is accompanied by a [Ca2+]cyt fertilization wave, whose speed is conserved at c. 10 µm s−1 in all species studied to date (Jaffe, 2002). A ten-fold slower [Ca2+]cyt fertilization wave has been seen in Zea mays (Digonnet et al., 1997; Antoine et al., 2000), and no wave at all can be seen in the stramenopile Fucus serratus (Roberts et al., 1994) (Fig. 4).
Figure 4. Interspecific comparison of [Ca2+]cyt elevations during fertilization. Fucus serratus eggs (© Company of Biologists Ltd. Adapted, with permission, from Roberts et al., 1994) and metazoan Ciona intestinalis eggs were microinjected with Calcium Green-1 dextran. Zea mays eggs were ester loaded with fluo-3 AM (© Company of Biologists Ltd. Adapted, with permission, from Digonnet et al., 1997). Figs have been adapted to fit on the same timescale, with t = 0 being the time of sperm entry into the egg.
Download figure to PowerPoint
The elucidation of the machinery by which the metazoan fertilization waves, and other metazoan [Ca2+]cyt elevations, are propagated is one of the most elegant achievements of signal transduction research and readers are referred elsewhere for a more detailed description (Berridge et al., 2003). Briefly, metazoan cells have a number of endomembrane Ca2+ stores, and agonists can cause Ca2+ to be released from any of these by the activation of three main pathways. In the first pathway, agonists bind G protein-coupled receptors. The G proteins then stimulate one of a variety of phospholipase C (PLC) isoforms, which cleave phosphatidylinositol 4,5 bisphosphate (PIP2), generating inositol-1,4,5-trisphosphate (IP3), which binds to the IP3 receptor (IP3R) on the endoplasmic reticulum (ER). In the second pathway, agonists cause intracellular generation of cyclic ADP-ribose (cADPR) and nicotinic acid adenine dinucleotide phosphate (NAADP), both of which are synthesized by the same enzyme, ADP ribosyl cyclase. In the third pathway, [Ca2+]cyt elevation causes activation of ryanodine receptors (RyR) on the ER. Activation of IP3R, the NAADP receptor, or RyR will cause increases in their Ca2+ permeablility, and hence release of Ca2+ from the ER and other endomembrane stores (Berridge et al., 2003).
The interplay between Ca2+ influx and efflux determines the spatio-temporal [Ca2+]cyt pattern, so early attempts to mathematically simulate metazoan [Ca2+]cyt elevations tended to focus on describing the behaviour of the IP3R using the so-called De Young/Keizer, or DYK, model (De Young & Keizer, 1992). This assumes that the IP3R consists of three subunits, each having stimulatory IP3- and Ca2+-binding sites, as well as inhibitory Ca2+-binding sites. When the DYK approximation is combined with terms for Ca2+ efflux through a nonlinear pump and Ca2+ diffusion through a homogenous cytosol, model predictions match experiment extremely well. Varying the intracellular IP3 concentration ([IP3]cyt) gives rise to different [Ca2+]cyt patterns: a low [IP3]cyt gives a steady low [Ca2+]cyt; increasing [IP3]cyt gives solitary [Ca2+]cyt waves or oscillations; and high [IP3]cyt leads to cessation of waves and a steady high [Ca2+]cyt (De Young & Keizer, 1992).
Since there is evidence that NAADP (Navazio et al., 2000), cADPR (Allen et al., 1995; Leckie et al., 1998) and IP3 (MacRobbie, 2000) can also stimulate [Ca2+]cyt elevations in plants, it is often assumed that plants and metazoan use the same internal Ca2+-release pathways. Indeed, many of the proteins involved in the metazoan response, such as PLC (Koyanagi et al., 1998; Mueller-Roeber & Pical, 2002) and the IP3-removing inositol phosphatases (Berdy et al., 2001; Xiong et al., 2001; Ercetin & Gillaspy, 2004) have been found in plants, and their expression is both affected by plant hormones (Hunt & Gray, 2001; Ercetin & Gillaspy, 2004) and required for certain Ca2+-dependent pathways such as nodulation (Engstrom et al., 2002) and the stomatal closure response (Sánchez & Chua, 2001; Xiong et al., 2001; Hunt et al., 2003).
Despite these similarities, there are caveats which make us suggest that Ca2+ release from endomembrane stores is an example of homoplasy, and not homology. While cADPR is generated in plants, no orthologue to the metazoan ADP-ribosyl cyclase has been found in plant genomes (Sánchez et al., 2004). Similarly, no study has ever shown the existence of orthologues of IP3R or RyR in plants or algae (Nagata et al., 2004). To complicate matters further, the PLCβ isoform which is activated by a variety of G proteins in metazoa (Hartweck et al., 1997) is not found in sequenced plant genomes, which encode the PLCδ isoform and only one Gα protein (Mueller-Roeber & Pical, 2002). In fact, although it is not seriously doubted that phosphoinositides are important regulators of plant cell physiology, the exact species responsible for plant [Ca2+]cyt elevations have yet to be pinned down. Although a rise in IP3 has been observed to follow hormonal and stress signalling (Hunt & Gray, 2001), and addition of IP3 to isolated vacuoles stimulated Ca2+-permeable channels (Allen et al., 1995; Muir & Sanders, 1997), more recent work suggests that IP6 is also competent to stimulate Ca2+ release from endomembrane stores (Lemtiri-Chlieh et al., 2003), which has lead to the proposal that IP3 is converted to IP6 and is active in that form. For the moment we can only say that if IP3 is involved in plant endomembrane Ca2+ release, it probably isn't generated or perceived as it is in metazoa, which may explain much current confusion among plant physiologists.
It is, however, important to realize that spatio-temporal patterns of [Ca2+]cyt elevations are not inextricably linked to certain molecules. Bearing this in mind, it is instructive to look at a mathematical model which was developed to look at [Ca2+]cyt oscillations in cardiac myocytes, in which the Ca2+ release sites are not IP3 receptors, but the poorly characterized RyR. This is the Fire-Diffuse-Fire (FDF) model (Keizer et al., 1998), in which a cluster of IP3 receptors is replaced by a release unit which is activated when [Ca2+]cyt rises above a certain threshold. All the biology of the receptor model is then approximated by the choice of threshold, and once the threshold is passed, Ca2+ release occurs. No assumptions need to be made about the identity of the Ca2+ release agent or the Ca2+ release site, yet the FDF model is also able to simulate a wide variety of [Ca2+]cyt patterns, including waves (Ponce-Dawson et al., 1999) and oscillations (Keizer et al., 1998).
So, although interspecific stimulus-evoked [Ca2+]cyt patterns are often extremely similar, perhaps best exemplified by the caffeine-induced [Ca2+]cyt oscillations seen in both green algae and mammals, these similarities may not necessarily reflect conserved mechanisms of generation but could easily result from the similarity in wiring between two systems, each composed of very different components. Having persuaded ourselves of this, we now turn to see how the components of [Ca2+]cyt signalling toolkits are organised and may vary.