Author for correspondence: Vincent R. Franceschi Tel: +1 509 335 3052 Fax: +1 509 335 3184 Email: email@example.com
• The function of calcium oxalate (CaOx) raphide crystal formation, and structural features related to regulation of crystal formation, were studied in Lemna minor fronds using physiological and microscopy techniques.
• Specialized crystal-forming cells (crystal idioblasts) increased in number and size; CaOx, but not soluble oxalate, increased in response to increasing calcium in the growth medium. Size and number of idioblasts had a distinct upper limit.
• The CaOx crystals are formed in membranous ‘chambers’ and connected in rows by parallel membrane sheets, both forming de novo in the vacuole. The chambers, but not parallel membranes, had calcium associated with them. A calcium-binding matrix protein was associated with idioblast vacuoles and crystal formation.
• Lemna crystal idioblasts function as calcium-inducible, specialized high-capacity but saturable sinks for bulk regulation of calcium, and crystal deposition is a highly controlled process requiring intravacuolar membrane systems and calcium-binding organic matrix materials.
To be able to fully understand this common biomineralization phenomenon, it is necessary to identify the processes involved in induction and accumulation of substances used for crystal formation, the way in which these substances are delivered to the site of calcium oxalate precipitation, and mechanisms regulating precipitation in the vacuole. Calcium must enter from the apoplast, while oxalate could be transferred from adjacent cells or synthesized within the idioblasts (Li & Franceschi, 1990). Various studies suggest ascorbic acid is the immediate precursor of oxalic acid used for crystal formation (Franceschi & Horner, 1979; Franceschi, 1987; Horner et al., 2000; Keates et al., 2000) and that crystal idioblasts contain the pathway for ascorbate synthesis (Kostman et al., 2001; Kostman & Koscher, 2003), thus making them autonomous for oxalic acid production. Crystal formation will require rapid transport of Ca and oxalate across a number of compartments, and the specialized ultrastructural features of crystal idioblasts are undoubtedly involved in the synthesis and/or movement of these compounds to the vacuole. Crystal idioblasts are commonly reported to have greater amounts of endoplasmic reticulum (ER) and golgi, and recent studies on the raphide idioblasts of Pistia stratiotes have indicated the idioblasts are enriched in calreticulin, which occurs in subdomains of the abundant ER (Quitadamo et al., 2000; Kostman et al., 2003; Nakata et al., 2003). The calreticulin is proposed to be involved in keeping cytoplasmic Ca activities low, while allowing for rapid accumulation of Ca used for CaOx formation (Nakata et al., 2003). The abundant golgi in these idioblasts have also been found to be involved in transporting a calcium-binding crystal idioblast specific protein, matrix protein, to the vacuole (Li et al., 2003). This protein becomes incorporated into the crystals, so it is possible that part of the flux of Ca to the vacuole is in association with protein. There are few studies of how the crystalization process is regulated, although it is clear that raphide crystal growth is coordinated with cell growth (Kostman et al., 2001), and that an organic matrix and macromolecules are associated with the crystals and may influence nucleation or growth (Arnott, 1982; Webb & Arnott, 1983; Webb et al., 1995; Bouropoulos et al., 2001; Volk et al., 2002; Li et al., 2003). However, it remains to be demonstrated how the various crystal idioblast-specific structures participate in transport and precipitation processes.
There is also evidence that the process of CaOx precipitation can be reversed, releasing Ca during periods when calcium is limiting to growth of the plant. For example, CaOx crystals disappear during rapid growth of shoots, such as at the beginning of growth of deciduous perennial vines when the transpiration stream is lacking, suggesting the Ca is mobilized to support initial growth (Assailly, 1954; Calmes, 1969; Calmes & Carles, 1970). Direct demonstrations of loss of Ca from idioblasts under induced Ca deficiency have been made in Lemna minor roots (Franceschi, 1989) and P. stratiotes shoots (Volk et al., 2002). These studies indicate that, at least in some species, crystal idioblasts function as mobilizable Ca sinks (Franceschi, 1989), although the type of crystal produced may affect how readily Ca can be released (Volk et al., 2002). Oxalate oxidase was found to be induced and accumulated in the vacuole of idioblasts under Ca starvation, and is probably involved in degradation of the oxalic acid that is released (Volk et al., 2002).
Whole L. minor (duckweed) plants can be easily manipulated, thus avoiding wound induced effects, and we have used this species to study various aspects of the physiology and biochemistry of CaOx formation (Franceschi & Schuren, 1986; Franceschi, 1987, 1989; Li & Franceschi, 1990). It has been proposed that crystals in duckweed are involved in detoxification of oxalic acid (Ledbetter & Porter, 1970). The purpose of this study was to determine the role of CaOx in the Ca nutrition of whole Lemna plants and the capacity of the Lemna plant for CaOx precipitation and crystal idioblast formation; and to further elucidate the possible functions of structures commonly observed in crystal-forming cells, with emphasis on the mechanisms related to CaOx deposition. We demonstrate that crystal idioblast formation is induced by Ca and has an upper limit in capacity; and, using cytochemical techniques, we show that calcium is associated with idioblast-specific elements. The results provide new information on the functional and ultrastructural basis of CaOx formation in plants.
Materials and Methods
Axenic Lemna minor L., a free-floating, freshwater aquatic angiosperm, was grown in aseptic culture on a liquid medium, called E medium, as previously described (Franceschi & Schuren, 1986). The plants rapidly reproduce clonally under the conditions used.
Calcium availability and crystal idioblast formation
Plants from maintenance cultures were subcultured at a density of three plants per culture onto liquid medium containing 0.5, 1, 2.5, 5, 7.5 and 10 mm Ca as CaCl2. The cultures were grown for 4 wk, at which time plant number had increased to about 100 per culture dish. Growth was strictly by clonal propagation. There was some slight depression in size of the fronds at the highest Ca concentrations. About 40 plants from each culture were cleared, as described by Franceschi & Schuren (1986). Using a projection microscope, the area of each frond of each plant was determined and the number of raphide crystal idioblasts per frond counted. The length and width of the crystal bundles in the idioblasts was measured using an ocular micrometer on a Leitz Orthoplan microscope fitted with polarizing filters. Representative images of mature, cleared fronds were taken with crossed polarizing filters, which makes the crystal idioblasts appear as bright objects against a dark background, using a black-and-white CCD camera interfaced to a computer. The polarity of the image was reversed so that the crystals appeared as dark objects, which allowed for a better image of the frond outline.
HPLC analysis of calcium oxalate
Lemna plants were transferred from E medium to sterile distilled water for 5 d to deplete the plants of Ca, which stops CaOx crystal idioblast formation and plant growth (Franceschi, 1989). Plants were then transferred to culture flasks containing E medium without sucrose and minus calcium, or E medium minus sucrose with 15 mm total Ca as CaCl2. Plants began growing again, even without Ca, as indicated by new frond formation. Three replicate flasks were cultured for each condition. Plants were grown for 2 and 6 d, after which all plants in each flask were removed, washed with distilled water, then lyophilized. Lyophilized Lemna plants were finely ground in a chilled mortar and suspended in 4 mm H2SO4 (HPLC mobile phase) containing 5 mm DTT and insoluble polyvinylpolypyrrolidone at 10 mg mL−1. Extracts were spun at 15 000 g for 30 min at 4°C to produce a clear supernatant containing soluble oxalate, and a pellet containing CaOx crystals. Soluble oxalate was precipitated as CaOx from the supernatant by titration with Ca acetate. The pellet that resulted was rinsed in dH2O and recovered by centrifugation. The CaOx in the pellet was converted to free oxalic acid by stirring it with a suspension of cation exchange resin (Dowex-50Wx8, 200 ± 400 mesh, H+) for 30 min at 60°C. The solubilized oxalic acid was adjusted to original volume, the resin removed by centrifugation, and the supernatant analyzed for oxalic acid. The pellet of extracted tissue containing the CaOx crystals was suspended in dH2O and pelleted to remove residual soluble oxalate. Oxalic acid from CaOx crystals was recovered as free oxalic acid by the procedure just described. Analysis for oxalic acid was by HPLC (Bio-Rad Aminex HPX-87H ion exclusion column, 300 × 7.8 mm, 0.6 mL min−1, 35°C, 900 psi; variable wavelength UV detector, Thermal Separation Products SpectraChrom 100, San Jose, CA, USA; and electrochemical detector, BioAnalytical Systems Model LC-4C, West Lafayette, IN, USA). HPLC data were integrated and analyzed with Gilson 715 controller software v. 1.21 (Gilson USA, Middleton, WI, USA). Oxalic acid was measured at 210 nm (retention time = 7.0 min). External standards at known concentrations were used to prepare calibration plots for assaying amounts of oxalic acid. Triplicate injections were done for each sample time and each sample flask.
Transmission electron microscopy (TEM)
Small developing fronds were cut from individual plants and plunged directly into fixative solution containing 2% (v/v) paraformaldehyde and 2.5% (v/v) glutaraldehyde buffered with 50 mm PIPES (pH 7.2) and left for 12 h at 4°C. The specimens were postfixed in 1% (w/v) osmium tetroxide in 50 mm sodium cacodylate buffer (pH 7.2) for 2 h at room temperature. For better visualization of membrane distribution, some specimens were fixed in potassium permanganate. Plants were fixed overnight in 1.2% KMnO4 in 93 mm veronal acetate buffer pH 7.4. After fixation, the tissue was rinsed three times in veronal acetate buffer. The specimens from both fixation schemes were dehydrated stepwise with acetone and infiltrated with Spurr epoxy resin (Spurr, 1969). Thin sections were cut on a diamond knife, picked up onto nickel grids and poststained sequentially with 2% (w/v) aqueous uranyl acetate and 1% (w/v) aqueous lead citrate before examination and photography on a Hitachi 300 or JEOL 1200 EX TEM.
Freshly cut young fronds were fixed for 6 h at 21°C in darkness in modified antimonated Karnovsky's fixative (Karnovsky, 1965). This fixative contained 2% (v/v) glutaraldehyde, 2.5% (v/v) formaldehyde in potassium phosphate-buffered (0.1 m, pH 7.6) potassium antimonate (2% w/v). Tannic acid was dissolved in the fixative (0.1% w/v) just before use. Fronds were washed twice in phosphate-buffered antimonate solution for 15 min, and postfixed in 1% (w/v) osmium tetroxide, which was prepared by diluting a 2% osmium tetroxide stock 1 : 1 with phosphate-buffered antimonate just before use. Postfixation was for 3 h at 21°C in the dark. After osmication, samples were briefly washed in phosphate-buffered antimonate, then in 0.01 m phosphate buffer without antimonate for 30 min. This last step was carried out to remove the relatively soluble sodium and potassium antimonate salts from the tissue (Klein et al., 1972). Samples were dehydrated with an acetone series and embedded in Spurr epoxy resin. Thin sections were viewed unstained or after poststaining as described under ‘Transmission electron microscopy’. After examination and photography, some sections were incubated with 10 mm EGTA in 50 mm PIPES buffer (pH 8) for 4 h. This treatment selectively removes Ca, including Ca salts of antimonate. The sections were then re-examined on the TEM. Dark-staining deposits in some of the sections were examined by energy-dispersive X-ray analysis to determine if the deposits contained Ca and antimony. A Tracor Northern 2000 detection system (Noran, Middleton, WI, USA) and Esstek Inc. software (Esstek Ohio Inc., North Olmstead, OH, USA) were used. For comparison, spectra were generated from areas of the same section free of precipitate.
For light microscopy, whole Lemna plants were frozen in liquid nitrogen, then freeze-substituted with acetone at c. 80°C, brought to room temperature, and infiltrated with xylene followed by paraffin. Sections were immunolabeled as described by Li et al. (2003) using preimmune serum or an antiserum raised against a calcium-binding matrix protein isolated from P. stratiotes, which has previously been shown specifically to label a protein in the vacuole and crystals of developing raphide crystal idioblasts. A secondary antibody conjugated to alkaline phosphatase was used to detect the primary antibody, using color development. For TEM, young Lemna fronds were fixed for 6 h at 4°C in 2% (v/v) paraformaldehyde and 1.25% (v/v) glutaraldehyde in 50 mm PIPES buffer (pH 7.2). The samples were dehydrated with an ethanol series and embedded in LR White acrylic resin. Thin sections on uncoated nickel grids were incubated for 1 h in TBST/BSA (10 mm Tris–HCl, 150 mm NaCl, 0.1% v/v Tween 20, 1% w/v bovine serum albumin) to block nonspecific protein-binding sites on the sections. They were then incubated for 4 h with either preimmune serum or antimatrix protein antisera. The dilution used for the sera was 1 : 100 with TBST/BSA. After extensive washing with TBST/BSA, the sections were incubated for 1 h with Protein A-gold (15 nm) diluted 1 : 100 with TBST/BSA. The sections were washed with TBST/BSA, TBST, and distilled water before poststaining with a 1 : 4 mix of 1% (w/v) potassium permanganate and 2% (w/v) aqueous uranyl acetate.
Western immunoblots were done on total protein extracted from whole plants as described by Li et al. (2003). The blots were probed with antibodies to Pistia matrix protein and visualized by a chemiluminescence procedure.
Calcium induces oxalate synthesis and crystal idioblast formation
Calcium oxalate raphide crystal idioblasts of L. minor are located throughout the frond tissue (Fig. 1a–d) and contain a bundle of needle-shaped raphide crystals in the vacuole (Fig. 1e–g), as previously shown (Arnott & Pautard, 1970; Franceschi & Schuren, 1986; Franceschi, 1987, 1989). The mature idioblasts are roughly cylindrical, and are about twice as long as adjacent cells (Fig. 1e,g) and their size is quite variable (Fig. 1f). The mature idioblasts are generally elongated in the direction of the long axis of the frond (Fig. 1f). The idioblasts lack chloroplasts (Fig. 1g) and contain very small plastids with few internal membranes (Fig. 1h). The plastids remain small compared to chloroplasts of surrounding mesophyll cells, and contain only small amounts of RuBisco (Li & Franceschi, 1990).
Crystal idioblast formation was found to be dependent on the Ca concentration of the growth medium. Casual observation of the clearings of entire mature fronds strongly suggests that the size and number of crystal idioblasts increase as Ca in the growth medium is increased (Fig. 1a–d). This is confirmed by analysis of the number of idioblasts per unit area of frond, and the length and width of the crystal bundles relative to Ca concentration in the growth medium (Fig. 2a–c). Each of these parameters increases up to a certain point as the Ca concentration increases. The number of idioblasts per plant is quite variable, but shows a general upward trend from 0.5 to 5 mm Ca. The apparent drop at 7.5 and 10 mm Ca (although not significant) could be caused by toxic effects of high Ca (or Cl) on prolonged growth of the plants. Length increases significantly from 0.5 to 5 mm Ca, then levels off. Width increases significantly from 0.5 to 1.5 mm and 1.5–5 mm Ca, then also levels off. The large standard errors seen in this analysis of size are a result of the fact that there is a naturally large amount of variation in size of crystal idioblasts and crystal bundles throughout the plant, as clearly indicated in Fig. 1f. This analysis indicates that both the number of crystal idioblasts and the capacity of individual idioblasts increase in response to Ca, although there is a limit to both the number and size of idioblasts that can be produced.
Biochemical analysis of oxalate content of the plants in response to Ca nutrition is consistent with the anatomical analysis. Soluble oxalate levels remain about the same after 2 and 6 d exposure to 15 mm Ca, while CaOx increases by c. 50% compared to plants on 0 Ca medium (Fig. 3). This increase occurs within 2 d of transfer of plants from distilled water to 15 mm Ca medium, and indicates that oxalate is synthesized in response to increased Ca in the plant.
Crystal form is associated with unique structures in developing idioblasts
Even the youngest crystal idioblasts are easily distinguished from surrounding cells. The idioblasts begin to enlarge earlier than other cells, have a larger central vacuole and a thin layer of more densely staining cytoplasm, and develop numerous cytoplasmic features not seen in other cells (Fig. 4a,b) while chloroplast fail to develop in contrast to the surrounding mesophyll (Figs 1g,h, 4a). The ER (Fig. 4b,c), golgi bodies and vesicles of 50–100 nm diameter (Fig. 1h) are more abundant in the idioblasts. Calcium oxalate crystals form within the central vacuole of developing idioblasts (Fig. 4b–g), which also contains a flocculent material (Fig. 4b) and many small vesicles of c. 100 nm diameter in the early stages of idioblast development (Fig. 4g). The vesicles and flocculent material are less common in mature crystal idioblasts.
Calcium oxalate is formed exclusively in the idioblast vacuole and occurs as bundles of needle-like raphide crystals with a rectangular cross-section during early growth stages (Figs 4b, 5), but later develop grooves along two opposite sides and become H-shaped in section (not shown). In sections made for TEM the crystals, which do not infiltrate with resin, are generally lost during staining so that only holes remain where the crystals reside (Fig. 4). There are two distinct membrane systems found in the idioblast vacuole that are involved in crystal formation. One membrane system directly surrounds the developing crystal and enlarges as the crystal grows (Fig. 4d–f). We refer to this as a crystal chamber (Arnott & Pautard, 1970). The other membrane system consists of parallel sheets of membrane between which the crystal chambers and crystals are produced (Fig. 4b,d,g). These have a bilayer structure typical of TEM images of other cellular membranes. The membrane systems organize formation of files of crystals, each in its own chamber, connected in linear arrays when viewed in cross-section (Fig. 4b,d,g). The membranes at the end of the file are not joined to form an enclosure, but remain free and end blindly in the vacuolar sap (Fig. 4b,g). Numerous files of crystals form within the vacuole of a developing idioblast, and there does not appear to be any specific orientation of one file with respect to another at early stages of differentiation. However, in mature idioblasts the crystals are more closely packed together and more organized. New crystals appeared to be formed at the ends of the files and at the boundary of the increasingly large bundle of crystals.
The parallel membranes within the vacuole are formed de novo in association with a flocculent material that is dispersed throughout the vacuole (Figs 4b,e, 5a,b). This material was not detectable in vacuoles of nonidioblast cells. The crystal chambers form at a subsequent stage at discrete intervals along the parallel membranes (Fig. 4d–g). Thus both the parallel membranes and the crystal chambers form de novo within the vacuole. They are not extensions of the tonoplast, as no direct connection between tonoplast and crystal membrane systems was seen with any of the three fixation protocols used.
Fully differentiated crystal idioblasts are living cells, as indicated by their intact cytoplasmic contents. The large bundle of crystals dominates the cell volume, while the thin peripheral layer of cytoplasm accounts for < 5% of the idioblast volume.
Calcium localization within developing idioblasts
The antimonate procedure for Ca localization gave very specific staining of components within the crystal idioblast vacuoles, but not vacuoles of adjacent cells (Fig. 5a,b). The flocculent material and crystal chambers within the developing idioblasts were always very intensely stained, and were the most heavily stained components of all cells (Fig. 5a). Staining of the crystal chamber membrane was interesting as it gave an exceptionally sharp image of the chamber shape during crystal formation (Fig. 5b), especially when fairly thick sections were examined (Fig. 6a). The parallel membranes connecting the crystal chambers in files did not stain with the antimonate procedure (Figs 5b, 6a), nor did any other biological membrane within the cell. Poststaining with lead and uranyl salts did make the parallel membranes and other organelle membranes show up (Fig. 6b,c), but the crystal chambers were always more intensely stained (Fig. 6b). Energy-dispersive X-ray analysis of idioblast vacuoles gave peaks for both antimony and Ca, although the peaks were quite small (Fig. 5a; see Kostman et al., 2003 for similar analysis in P. stratiotes). Neither element could be detected in blank areas of resin from the sections (not shown). Treatment of sections with EGTA, a relatively selective chelator of Ca, removed the staining from the crystal chambers and flocculent material (Fig. 5c), further indicating that the dark staining is at least partly caused by calcium antimonite precipitation.
Sites of Ca accumulation were also made evident within the cytoplasm of idioblast cells by the antimonate procedure. Most noticeably, dense deposits were formed within golgi bodies and associated membranes and vesicles (Fig. 6c). The stained material in the golgi vesicles had the same structural appearance as the flocculent material in the vacuole. Calcium deposits were also found within the ER, where they occurred as large isolated particles within the ER lumen (Fig. 6c). Dense staining for Ca was also found within the Ca pectate-rich regions of the wall junctures between two or more cells throughout the Lemna samples (Fig. 5a).
A crystal matrix protein is found in Lemna idioblasts
An antibody raised against the calcium-binding matrix protein from Pistia raphide crystals gave a positive signal in Western blots of total Lemna protein, with a strong band of the same molecular weight as seen with Pistia extracts and purified Pistia matrix protein (Fig. 7). This protein was previously found to be very difficult to keep in solution and does not run as a focused band (Li et al., 2003). It is probable that a combination of aggregation and degradation of the matrix protein during extraction is responsible for smearing of the reactive protein on the gel. There is also the possibility that other related proteins may be detected. Comparison of labeling with preimmune (Fig. 8a) and immune serum (Fig. 8b) on paraffin sections of Lemna demonstrates that the antibody very strongly labels developing crystal idioblasts, but not other cell types. Using TEM level immunolabeling procedures, it was shown that the developing crystals had label associated with their surfaces (Fig. 8c) as well as with the flocculent material in the vacuole. While the amount of label was less than that associated with the crystals in Pistia idioblasts, it was consistent and specific among all idioblasts examined. No appreciable label could be found in other cells, and preimmune serum did not lead to labeling of any sort.
A number of functions have been ascribed to CaOx deposition in plant cells, with one likely function being bulk Ca regulation in tissues and organs (Zindler-Frank, 1975; Borchert, 1985, 1986; Franceschi, 1989; Franceschi & Loewus, 1995; Volk et al., 2002). Calcium oxalate formation in L. minor has been definitively shown here to be responsive to Ca availability, which indicates a role in Ca regulation. Both the number of crystal idioblasts and their capacity, as indicated by the amount of CaOx deposited in each cell, increases in response to increasing extracellular Ca. By contrast, soluble oxalate levels remain constant when Ca is present, further indicating that CaOx is formed in response to excess Ca and not excess oxalic acid. While the number of crystal idioblasts increases with increasing calcium in the medium, they generally tend to remain evenly distributed throughout the frond. This supports a hypothesis that each idioblast serves as a calcium sink for a given volume of tissue, with the tissue volume per idioblast decreasing as the Ca in the medium, and thus apoplast, increases (Borchert, 1985; Franceschi, 1989).
The data also show that there are limitations to both the number of idioblasts that can be produced within the frond and the maximum capacity of individual idioblasts, as indicated by the size the idioblasts can achieve. Once these limitations are reached, further increases in Ca in the medium can lead to toxic effects such as decreased growth rate, reduced overall plant size and senescence (data not shown). Thus, although it can be considered a high capacity mechanism for Ca sequestration, as in any homeostatic mechanism the capacity of the regulatory process can be exceeded in extreme conditions, resulting in detrimental effects on the plant.
Calcium oxalate-forming idioblasts have been shown to be enriched in golgi bodies and ER relative to adjacent cells, and some unusual structures, such as intravacuolar membranes and flocculent or paracrystalline material, are also commonly associated with the formation of CaOx in plant cells (reviewed by Arnott & Pautard, 1970; Franceschi & Horner, 1980; Horner & Wagner, 1995; Webb, 1999). While these structures are assumed to be important to the precipitation process, their actual functions have not been identified. Using a combination of cytochemical techniques, we have demonstrated that some of these ultrastructural features are involved in Ca transport and/or sequestration in idioblasts. Calcium localization demonstrates that the flocculent material in the vacuoles of developing Lemna idioblasts has Ca associated with it. This material, or at least a component of it, has structural, functional and antigenic properties similar to the calcium-binding matrix protein isolated from P. stratiotes raphide crystals (Li et al., 2003), and we propose that it is a heterologous protein. In Pistia, this protein is transported to the vacuole via golgi-derived vesicles (Li et al., 2003). At least some of the abundant vesicles seen in Lemna cytoplasm are presumably performing a similar function. It is notable that Ca localization indicates the golgi vesicles contain Ca which is probably bound to the matrix protein, as indicated by the somewhat flocculent nature of the labeling pattern. These results suggest that at least part of the Ca used for CaOx formation is transported to the vacuole via fusion of golgi vesicles with the tonoplast, and is released into the vacuole in a complex with the matrix protein. The abundance of golgi and associated vesicles in the idioblasts is thus directly related to part of the mechanism of Ca transfer to the vacuole. However, it is also likely that other mechanisms (tonoplast Ca channels and transporters) for transfer of Ca to the vacuole are required to accommodate high fluxes of Ca into the vacuole during rapid crystal formation. For example, the large amounts of ER commonly seen close to the tonoplast may allow for an efficient leak/pump system for transfer of Ca from ER to vacuole.
Various proteins have been shown to be important in biomineralization in animal systems, often forming a complex matrix (Weiner, 1984, 1986; Addadi & Weiner, 1985; Lowenstam & Weiner, 1989; Weiner & Addadi, 1991). Matrix materials are also associated with plant crystals (Webb & Arnott, 1983; Webb et al., 1995; Volk et al., 2002); proteins that can affect CaOx precipitation have been shown to exist in plant crystals (Bouropoulos et al., 2001); and a calcium-binding crystal matrix protein has been isolated and characterized from raphide crystals (Li et al., 2003). Our results indicate that the crystal chamber directly surrounding the crystals is probably a precipitation membrane consisting partly of matrix protein, rather than a typical lipid bilayer membrane. This is supported by immunolabeling and Ca localization results that indicate that the matrix protein becomes associated with the surfaces of the developing crystals. Whether the protein chamber is involved in crystal nucleation or control of growth remains to be determined, although results from Pistia raphide matrix protein indicate that it could play a role in both processes (Li et al., 2003).
Membranes are common in the vacuole of crystal idioblasts, and parallel membranes as shown here are seen associated with raphides in other duckweed species (Ledbetter & Porter, 1970; Chiu & Falk, 1975; Vintejoux & Shoar-Ghafari, 1985) and the larger, free-floating aquatic P. stratiotes (Li et al., 2003), as well as other raphide-producing species. The parallel membranes that form a boundary on either side of the Lemna crystals do not show any appreciable Ca binding. However they do appear to be involved in initiation of crystal formation, as crystal chambers are produced early between the parallel membranes before crystal precipitation. This was also shown in a micrograph of Lemna in the review of Arnott & Pautard (1970), and in a micrograph of the related species Spirodela oligorrhiza in the monograph of Ledbetter & Porter (1970) on plant cell fine structure. Unfortunately, although an abstract appeared (Arnott & Pautard, 1965), a full research article on the process was never published. The parallel membranes may also be involved in the orientation of the developing raphide crystals in the idioblast vacuole (Ledbetter & Porter, 1970). As demonstrated in Pistia, the raphide crystals show bidirectional coordinated growth along the same axis as the idioblast elongation (Kostman & Franceschi, 2000), so the precipitation and orientation of the crystals within the vacuole must be tightly controlled. It is possible that the parallel membranes may also serve to anchor and orient the crystal bundle in the middle of the vacuole, so that they do not ‘settle’ against the tonoplast caused by the effects of gravity on the relative mass of the bundle. Settling could lead to rupture or puncture of the tonoplast. Attachment of some of the parallel membranes to the tonoplast would be necessary to accomplish anchoring of the bundle. We have never seen parallel membranes directly attached to the tonoplast; however a relatively few attachments could serve this purpose, and serial sections would be needed to observe them.
Once the Ca enters the cytoplasm, its activity must be rapidly regulated to ensure normal cellular metabolism continues. The abundant ER in the Lemna crystal idioblast is proposed to be involved in this process, as indicated by Ca localization results. The ER is known to have calcium pumps that allow this organelle to accumulate calcium, and thus help to regulate free Ca activity in the cytosol. A high-capacity Ca-binding protein such as calsequestrin or calreticulin would reduce the solubility of the Ca once transported to the ER. We have previously identified such a protein in ER of Pistia raphide idioblasts, and have shown that the idioblasts are enriched for calreticulin relative to surrounding cells, and the ER has calreticulin-rich subdomains and bound Ca associated with it (Quitadamo et al., 1999; Kostman et al., 2003; Nakata et al., 2003). This is consistent with a proposed mechanism of the transfer of Ca to matrix protein, which is synthesized on the RER and transferred to the vacuole by golgi vesicles (Li et al., 2003), and transport of some Ca to the vacuole in association with golgi vesicles as seen here.
In summary, Lemna raphide crystal idioblasts are shown to play a role in bulk Ca regulation in the plant rather than oxalate detoxification, and it was found that there is a limitation to the number and size of the crystal idioblasts that can be formed, which limits the ultimate capacity of the system. The data presented also provide some clues on the function of organelles and structures previously identified as characteristic of this specialized cell type, but whose function relative to CaOx formation remained undefined. This work provides further support for the role of CaOx formation in high-capacity calcium regulation in plants, and further characterizes some of the developmental and biochemical aspects of specialized subcellular features of idioblasts with respect to Ca transport and precipitation.
This work was supported in part by NSF grants MCB 9632027 and MCB 9904562 to V.R.F. Electron microscopy was done in the facilities of the Washington State University Electron Microscopy Center. The authors thank Nathan Tarlyn for technical assistance with HPLC analysis of oxalate content.