Diversity and distribution of idioblasts producing calcium oxalate crystals in Dieffenbachia seguine (Araceae)†
The author thanks Dr. J. Guinan for help with statistical analyses and for helpful comments on the manuscript. He also thanks the undergraduate students who worked in the laboratory over the past several years and contributed toward this work: A. Woodson, M. deMilt, L. Bowling, and A. Andrews. The work was supported by the Thomas F. and Kate Miller Jeffress Memorial Trust.
Abstract
Although cells that synthesize crystals are known throughout the plant kingdom, their functional significance is still unknown. Mechanical support, mineral balance, waste sequestration, and protection against herbivores have all been proposed as crystal functions. To seek clues to their role(s), I systematically examined all organs except fruit of Dieffenbachia seguine (Araceae) for crystals. Crystals were found in nearly every organ. Raphides (long, slim, pointed crystals) were most common, but druses (crystal aggregates) and prisms were also found. Raphides varied in size by a factor of 10 and also in organization from tightly bundled to loosely organized. Biforines, a type of cell capable of expelling raphides, or biforine-like cells, were found in nearly all organs, but especially in leaves, spathes, and anthers. Different organs had different crystal complements, and characteristic crystals were found at specific locations, such as among pollen, along the undersides of leaf veins, and at root branch points. All crystals appeared to be composed of calcium oxalate, based on acid solubility. Possible roles of the crystals are discussed in light of these findings.
Many plant tissues contain cells, known as idioblasts, which differ markedly from neighboring cells (reviewed by 7). Among the most common idioblasts are those that synthesize crystals of calcium oxalate (reviewed by 9; 15; 10). Many crystal forms are known (9), including druses, which are spherical crystal aggregates; raphides, which are long pointed needles found in bundles within cells; prisms; and very fine crystalline sand. In the Araceae, many species have an unusual type of raphide-containing idioblast, first described by 32, called a biforine. These spindle-shaped idioblasts with nipple-like projections at each end are capable of forcibly expelling the crystals (23; 28), which may explain the ability of plants containing these cells to damage oral tissues when eaten (12).
Crystal idioblasts are widely distributed in the plant kingdom (9); 22 listed 215 families in which they had been reported. Crystals are particularly diverse in the Araceae (27), and their distribution in this family has been reviewed (17, 18). Despite the ubiquity and variety of crystals in plants, the purpose of these structures is poorly understood. It is often assumed that crystals might deter herbivory, and biforines, in particular, seem well suited to this role. However, whether crystals are truly effective as deterrents has not been clearly demonstrated. Furthermore, it is not known whether all crystal types can deter herbivores or whether crystals have other important roles.
To help elucidate the possible roles of crystals in plants, it would be helpful to understand the variety of different crystal types made by a single plant and where these crystals are localized. I have been studying the crystal idioblasts produced by Dieffenbachia seguine, an aroid from the American tropics. Reports of crystals in the vegetative organs of Dieffenbachia species have been reviewed by 17, who also made additional observations on D. lentii, D. longispatha, and D. oerstedtii. The occurrence of crystals within the reproductive organs, however, has not been well documented.
Dieffenbachia produces inflorescences typical of the Araceae. Flowers are minute, borne on a long column, the spadix, which is surrounded by a leaf-like spathe (Fig. 1A, B). The spadix is fused along its back, from the base to about halfway up, to the surrounding spathe. Approximately the upper half of the spadix is covered with closely crowded staminate flowers. Four or five staminate flowers are fused together to form a flat-topped structure termed a synandrium (Fig. 1C). Most of the lower spadix is covered with pistillate flowers consisting of a single gynoecium, surrounded by sterile staminodia (Fig. 1E). Between staminate and pistillate flowers is a region of scattered sterile flowers, consisting solely of staminodia (Fig. 1D). Dieffenbachia is reported to be pollinated by scarab beetles (reviewed by 13). The plants are protogynous, and the spathe begins to loosen around the spadix when the female flowers are receptive. The beetles enter the partially opened spathe, eat staminodia, mate, and pollinate receptive stigmata. Later, the spathe loosens further, and the beetles leave, picking up pollen on their way out.
Reproductive structures of Dieffenbachia seguine. (A) Inflorescence on plant. The columnar spadix is surrounded by the spathe, open like a cowl at top, closed toward base. (B) With spathe partially excised, flowers are visible. Top of inflorescence is at left. (C) Top of spadix is covered with fused male flowers, synandria, and pollen is visible. (D) Sterile flowers, staminodia, at center constriction of spadix. (E) Female flowers at base consist of mushroom-shaped gynoecia surrounded by smaller bulbous staminodia. A and B, bar = 1 cm; C–E were photographed through a dissecting microscope; bar = 1 mm.
I here describe the variety of crystal idioblasts found within both vegetative and reproductive organs of D. seguine and discuss the implications of this diversity in understanding the possible role(s) of these cells.
MATERIALS AND METHODS
Plant material
Plants of Dieffenbachia seguine (Jacq.) Schott cv. Compacta (Araceae) [synonyms: D. maculata (Lodd.) G. Don; D. picta (Lodd.) Schott], obtained from Florida Plant Growers (Apopka, Florida, USA), were grown in approximately 16-cm pots either under controlled conditions in a growth chamber or under ambient conditions in a lathe house within a greenhouse. In the growth chamber a light/dark cycle of 14 h/10 h under fluorescent lights (approximately 1700–2000 lux at plant level) was maintained; temperature was 28°C in the light period and 22°C in the dark period. Plants were fertilized approximately every six months with Osmocote 13–13–13 (Scotts-Sierra, Marysville, Ohio, USA). Plants were divided and repotted as needed twice a year. Plants were transferred freely between greenhouse and growth chamber as needed for experiments. Under greenhouse conditions, many plants, although not all, flower spontaneously in May.
Tissue preparation and microscopy
Organs were hand-sectioned with a single-edged razor blade. Both cross sections and transverse sections were prepared of stems and roots. Petioles and leaf midveins were cross-sectioned. Leaves and petiole wings were cut into pieces about 1 cm2 or smaller. Adventive roots, induced by growing the plants in plastic bags, were either transverse sectioned or processed whole.
To examine reproductive tissues, I dissected spathes from the spadices. Samples of spathe, about 1 cm square were cut out. Spadices were cross-sectioned through staminate flowers, through pistillate flowers, and through sterile flowers.
All samples were cleared using a modification of the NaOH tissue clearing method of 29 as follows. They were placed in 5% NaOH and shaken on a rotary platform shaker at 30 rpm for 2 to 4 days. Sufficient liquid was used to allow free movement of the tissue samples, generally about 30 mL. Samples were then bleached in full strength household chlorine bleach (3–6% sodium hypochlorite), usually for 5–10 min, but no more than 30 min, and then washed three times in water for at least 10 min each time. Samples were maintained on the shaker at 30 rpm during bleaching and washing.
Cleared samples were then dehydrated by passing them through 25%, 50%, 75%, and 100% reagent grade ethanol, and rinsed again in 100% ethanol. Each solution was applied for at least 10 min with shaking at 30 rpm. They were then placed in 1:1 reagent grade ethanol:xylene (xylol) under a fume hood for 10 min with occasional swirling, and then in two changes of xylene for 10 min each, with swirling, under the fume hood. Finally, the samples were transferred to a pool of Permount (Fisher Scientific, Fairlawn, New Jersey, USA) on a microscope slide and covered with a glass cover slip.
Slides were examined with bright field and polarization microscopy. The latter was performed by placing two or three aligned polarization filters (Ward's Natural Science, Rochester, New York, USA) over the microscope light source, and viewing the slides through polarizing sunglasses. The filters on the light source were rotated to achieve extinction of background illumination when viewed through the sunglasses. Digital micrographs were taken with a Nikon DXM1200C digital camera mounted on a Nikon Eclipse 80i microscope (Nikon, Tokyo, Japan). Polarization with the Eclipse 80i was achieved by mounting three aligned polarization filters within an unused filter cubicle of the epifluorescence turret and placing three aligned filters over the light source in the correct orientation to achieve background extinction. Some low magnification digital micrographs were taken with an Olympus QColor5 digital camera mounted on an Olympus BH-2 compound microscope (Olympus, Tokyo, Japan). Polarization was achieved in this case by placing three filters over the light source and one filter, rotated to achieve extinction, on top of the slide. Although a single filter could be used in these applications, two or three combined gave darker backgrounds.
Chemical identification of crystals
The chemical identity of the crystals was tested by the acid treatment described by 36 for renal tissue. This method has successfully been used with plant material (21). After the tissue clearing with NaOH and bleach as described previously, samples were transferred to 2% hydrochloric acid (dissolves crystals of all calcium compounds) or 5% acetic acid (dissolves calcium carbonate and phosphate, but not oxalate), and incubated overnight on a rotary shaker at approximately 30 rpm. They were then rinsed three times with pure water for at least 15 min each time, dehydrated with ethanol, and preserved as permanent slides as described previously.
Statistical analyses
Mean druse diameter and mean length and width of biforine and box-like bundles were compared among different tissues. Mean crystal densities in various parts of leaves were also compared. All comparisons were made by one-way ANOVA followed by multiple posthoc comparisons between tissues or between leaf sections using the Tukey–Kramer honestly significant difference (HSD) test. Statistical tests were performed using the JMP 7 statistical software package (SAS Institute, Cary, North Carolina, USA).
RESULTS
Types of crystal-containing idioblasts found
Idioblasts containing druses
Druses, that is, spherical crystal aggregates (Fig. 2A), were abundant in nearly all organs examined. Diameters of representative druses were measured in stem epithelium, stem pith, leaves, petioles, petiole wings, roots, spathe tips, upper spathe, upper and mid spadix columns, synandria, gynoecia, staminodia from pistillate flowers, and staminodia from sterile flowers. After one-way ANOVAs, post hoc comparisons showed no significant difference between druses in stem pith and epithelium, between druses in petioles and petiole wings, between druses in either part of the spathe, between druses in either part of the spadix, or between druses in either kind of staminodium (data not shown). It is reasonable that druses in each of these tissue pairs could represent common biological populations, and given the lack of any significant difference between them, the tissue pairs were lumped for further analysis.
Variety of crystals in Dieffenbachia seguine. All crystals are from leaves. (A) Typical druse and a very small druse (upper left). (B, C) Free raphides isolated from leaves. Thick raphide in (B) is typical of those in giant bundles. Those in (C) were being shot from an isolated biforine. (D) A large bundle of thick raphides called a giant bundle. (E, F) Small bundles of raphides; crystals in (E) are tightly bundled while those in (F) are rather disorganized. (G) A biforine; note that some crystals have been expelled from either side. (H) A box-like bundle typical of those that line midveins. Crystals in (B) and (C) were isolated from leaves and are floating in buffer; others were photographed in situ in cleared leaves. All micrographs were taken with bright field illumination. Bar = 20 μm.
The analysis was then repeated with the druses categorized as being from stem, leaves, petioles, roots, spathes, spadix columns, synandria, gynoecia, or staminodia, with between 12 and 44 druses measured in each tissue. There was significant variation between the means of druse diameter by tissue (F8,195 = 47.2, P < 0.0001). Post hoc comparisons of the means indicated that the stem druses were significantly larger than druses from any of the other tissues, while druses in the roots were significantly smaller (Fig. 3). There was no evidence for significant variation in the mean size of druses in other parts of the plant. The mean druse diameter for all locations except stem and root was 31.8 ± 5.3 μm (N = 138).
Variation in diameter of druse crystals in different tissues of Dieffenbachia seguine. Bars represent mean diameter of between 12 and 44 druses from each tissue, with error bars representing standard error of mean. Bars annotated with same letter are not significantly different.
Leaves also contained occasional druse-like aggregates obviously much smaller than typical druses in leaves or other tissues, in the range of 13 μm diameter. These microdruses were not common, although they were often clustered. They were sometimes seen in the same microscopic field as typical druses (Fig. 2A).
Idioblasts containing raphides
Raphides released by disrupting leaf tissue in a mortar and pestle, or in a blender, could be divided into two categories by their cross-sectional diameter. The thicker raphides (Fig. 2B) measured about 3 μm, the thinner (Fig. 2C), about 2 μm. At high magnification, a central, longitudinal groove was visible in the 3-μm raphides. In leaves, thicker raphides were found in regular cylindrical bundles, 131 ± 23 μm long (N = 8), a characteristic form I call giant bundles (Fig. 2D). Thinner raphides were seen in all reproductive and vegetative organs examined and were generally arranged parallel in regular cylindrical bundles (Fig. 2E); however, some cells appeared to contain overlapping bundles of raphides, and others contained apparently disorganized bundles (Fig. 2F). Raphide length varied widely, from about 16 μm to as much as 160 μm.
Biforines
Biforines, spindle-shaped idioblasts with terminal nipples, capable of ejecting their crystals, could be easily recognized in leaves and other tissues because they generally expelled a portion of their crystals during tissue clearing (Fig. 2G). Biforine raphides were the 2-μm variety (Fig. 2C) and were arranged parallel in cylindrical bundles, similar to the arrangement of the thicker raphides in giant bundles. I use the term biforine-like to indicate idioblasts that resembled biforines, but had not expelled any crystals and did not have easily discerned end nipples. It is possible that these actually were biforines, but were not induced to expel their crystals, especially in thicker tissue sections. Biforine-like idioblasts were common in synandria, where many of them were associated in long strand-like clusters.
The lengths and widths of crystal bundles in typical biforines and biforine-like cells were measured in leaves, stems, spathes, and synandria (with those associated in strands recorded separately); between 5 and 19 bundles were measured in each tissue. Mean lengths differed significantly (F4,50 = 42.0, P < 0.0001), and post hoc comparisons of the means indicated that biforines in spathes were longer, while biforine-like cells occurring in strands in synandria were shorter than those in other locations (Fig. 4). Mean widths also differed significantly (F4,50 = 15.6, P < 0.0001). Biforine-like cells in strands in synandria were also narrower than those elsewhere, although not significantly narrower than those in leaves (Fig. 4). Leaf biforines also appeared to be narrower than those in stems or spathes. Because some of the crystals had been expelled, these variations in width may indicate the degree of expulsion of the crystals during the clearing process.
Variation in size of biforine crystal bundles in different tissues of Dieffenbachia seguine. Bars represent mean length and mean width of between five and 19 crystal bundles from each tissue, with error bars representing standard error of mean. Bars annotated with same letter are not significantly different. “Syn. strands” indicates crystal bundles occurring in strands in synandria, which were counted separately from those occurring individually.
Idioblasts containing box bundles
Crystals that appeared to be raphides were also seen in box-like bundles (Fig. 2H). These bundles were unusual in that they often were pale brown, even following leaf clearing. These crystals were associated with major veins in leaves and spathes and were also found in petioles, stems, spadix columns, and at the bases of secondary roots.
Box bundle length and width were measured for typical bundles in leaves (along midveins), petioles, roots, spadix, spathe, and stem, with between 6 and 16 bundles measured in each tissue. Populations varied significantly in mean length (F5,54 = 4.67, P = 0.0013) and mean width (F5,54 = 10.34, P < 0.0001). Root bundles tended to be shorter and narrower than others; while midvein and petiole bundles tended to be wider. However, there was a continuum in both width and length (Fig. 5).
Variation in size of box-like crystal bundles in different tissues of Dieffenbachia seguine. Bars represent mean length and mean width of between six and 16 crystal bundles from each tissue, with error bars representing standard error of mean. Bars annotated with same letter are not significantly different.
Distribution of crystals within particular tissues
Leaves
Leaf blades contained a variety of crystal idioblasts (Fig. 6A). Manipulating the focus at high power showed that most crystals, of all types, were distributed primarily within the upper mesophyll, with only occasional crystals coming into focus lower in the leaf.
Crystals in vegetative tissues of Dieffenbachia seguine. (A) Leaf lamina contains scattered biforines and giant bundles. Small crystals are mostly druses, but include some small raphide bundles. (B) Leaf margin shows a greater density of druses, as well as articulated (overlapping) bundles; (C) higher magnification of articulated bundle. (D) Cross section of midvein (lower surface at left) shows clustered box-like bundles. (E) Cross section of petiole near lower surface shows similar clusters of box-like bundles (lower right); vascular tissue and druses can be seen near surface at left. (F) Transverse section of petiole shows internal box-like bundles, aligned parallel to long axis of petiole. (G) Internal petiole box-like bundles at greater magnification. (H) Cross section of petiole near upper surface shows druses, somewhat irregular raphide bundles, and some free raphides. (I) Transverse section of petiole at higher magnification shows druses and a long, overlapping bundle. (J) Petiole wings contain many druses. (K) Cross-section of upper stem, near epidermis shows druses and box-like bundles scattered under surface, which runs along bottom of micrograph. The cells on both sides of the crystal-containing layer are free of crystals. (L) Transverse section of upper stem epidermis through crystal-containing layer. Stem axis runs horizontally, and crystals are predominately oriented parallel to or radial to stem axis. (M) Pair of box-like bundles near the stem epidermis. (N) Internal pith of stem cross section with scattered druses and raphide bundles. (O) Transverse section of secondary root shows aligned druses. (P) Transverse section of a major root shows occasional raphide bundles, sometimes apparently overlapping; (Q) overlapping bundles at higher magnification. (R) Transverse section of primary root, with secondary root emerging at right. Clumps of box-like bundles are present at base of secondary root. Files of druses can also be seen in secondary root. (S) Higher magnification view of bundles at base of secondary root. (T) Adventive root, covered with root hairs, contains no apparent crystals. The root epidermis is near the bottom, under a thick forest of root hairs. Micrographs A, C, and M were taken with polarization microscopy; the rest are bright field. C, G, M, O, Q, S: bar = 20 μm; all others: bar = 100 μm. Abbreviations: b, biforine; g, giant raphide bundle; fr, free raphide; ob, overlapping (articulated) raphide bundle; rb, raphide bundle; vb, vascular bundle.
Druses were the most common crystal type in leaves, and occasionally very small druses were also seen. Raphide-containing idioblasts included biforines, giant bundles, and those containing bundles of small raphides. The small bundles varied from tightly to loosely bundled (Fig. 2E, F), and varied in length from about 15 to 50 μm. Small bundles were the most common raphide-containing idioblasts in leaves, followed by biforines; giant bundles were rare.
Different parts of the leaf blade seemed to contain the same kind of crystal idioblasts, in similar densities, except that leaf margins were marked by excess druses and scarce biforines and other raphide bundles. A leaf margin is shown in Fig. 6B; the increased density of druses along the immediate edge can be seen, and giant bundles and biforines are seen only many micrometers in from the margin. In cleared leaves, the increased druse density was visible to the naked eye as a white marginal band. Moreover, there were articulated or overlapping bundles of idioblasts, about 70 μm to 100 μm long, along the leaf margin, which were not seen internally (Fig. 6B, C).
To test whether the density of different idioblasts varied between different parts of the leaf blades, leaves were split at the midvein, one half was cut into four pieces from petiole to tip, and each piece, except the tip, was cut lengthwise into an inner (toward the midvein) and an outer section. In addition, the area between the leaf margin and the outermost vein, a strip about 250–400 μm wide, was examined separately. For each piece, all biforines and giant bundles were counted in 15–25 microscope fields at 100×, and all smaller idioblasts were counted in 15–25 fields at 400×. Four different leaves from four different plants were used, and, to reduce human error, four different individuals replicated the counting. One-way ANOVA indicated highly significant variation between the different sections in the mean densities of all idioblast types, except for the rare giant bundles. (giant bundles: F10,32 = 1.07, P = 0.41; druses: F10,32 = 7.65, P < 0.0001; biforines: F10,33 = 8.53, P < 0.0001; small bundles: F10,33 = 3.02, P < 0.0001.) Overlapping (articulated) bundles were not analyzed because they occurred only at the margins. Post hoc comparisons suggested that differences in crystal densities were primarily between the internal sections and the edges, in agreement with visual observation. In particular, there was no significant difference, for any idioblast type, between any two internal sections, between any two outer sections, or between any two marginal sections. Therefore, for each leaf, these sections were averaged, and a new analysis was performed to compare inner, outer, and marginal densities. In all cases, except again for the rare giant bundles, there was significant variation. (giant bundles: F2,9 = 0.43, P = 0.66; druses: F2,9 = 12.86, P = 0.002; biforines: F2,9 = 14.49, P = 0.0015; small bundles: F2,9 = 45.18, P < 0.0001.) Post hoc comparisons showed a significant difference between the margin and either of the two internal sections, but no differences between the two internal sections, again except for the giant bundles. Mean values for idioblast density at the leaf edges and internally are given in Table 1 .
| Mean density of crystal idioblasts (mm−2) | |||||
|---|---|---|---|---|---|
| Location | Druses | Biforines | Giant raphide bundles | Small raphide bundles | Articulated bundles |
| Internal lamina | 92 ± 6 | 3.68 ± 0.25 | 0.46 ± 0.04 | 16.5 ± 0.7 | 0 |
| Leaf margin | 248 ± 21 | 0.76 ± 0.23 | 0.22 ± 0.21 | 3.44 ± 0.99 | 5.11 ± 0.70 |
The midvein of the leaf also differed from the lamina in idioblast distribution. While typical leaf idioblasts occurred in the upper surface, the lower surface (Fig. 6D) was lined with clusters of box bundles of raphides (Fig. 2H) about 70–100 μm in the long dimension.
The leaf petiole was lined along its back with a loose strand of box bundles about 300 μm below the surface, paralleling the long axis of the petiole (Fig. 6E). These were similar to the box bundles lining the midvein. Box-like bundles were also found internally, in rows running parallel to the long axis (Fig. 6F, G). Petiole internal tissue also contained druses and small raphide bundles, about 35 μm long (Fig. 6H). Some free raphides were also seen, possibly released when the petiole was cut. There were also some very long bundles, 30–60 μm wide and generally 70–150 μm, but up to 600 μm long, which had the appearance of regular cylindrical bundles that had been pinched and drawn out so that the raphides overlapped instead of lining up (Fig. 6I). The petiole cortex was rich in druses (Fig. 6E). Biforines or biforine-like idioblasts were rarely seen in petioles. Petioles are winged distally, and the wings contained a dense concentration of druses and small raphide bundles typical of petioles; druses were more common (Fig. 6J).
Stems
Beneath the stem epidermis was a layer of parenchyma well endowed with druses, box-like raphide bundles and occasional small bundles of raphides, about 50 μm across (Fig. 6K–M). The cortical cells outside this crystal-containing layer, and the longitudinally lengthened collenchyma cells inward from this layer contained few or no crystal idioblasts (Fig. 6K). The internal pith of stems contained many druses and small raphide bundles (Fig. 6N). Box bundles similar to those in the outer parenchyma were occasionally found, usually oriented parallel to the stem. Biforines were rare.
Roots
Roots contained many druses lined up in files (Fig. 6O), small bundles of raphides about 20 μm wide by 60 μm long (Fig. 6P), overlapping raphide bundles (Fig. 6Q), and very rare biforines. Giant raphide bundles were not seen. Within the primary roots, the bases of lateral roots were marked by clusters of box bundles (Fig. 6R, S). Secondary roots had strings of druses near their bases (Fig. 6O, R), and overlapping bundles of raphide further from the bases. Adventive roots, which grow from the stem under humid conditions, contained no clearly visible crystals (Fig. 6T). Roots in general and adventive roots in particular had highly birefringent cell walls, making polarization microscopy difficult to use. Consequently, crystals may have been missed in adventive roots if they were also highly transparent with bright field microscopy.
Spathes
Leaf-like spathes surrounding inflorescences contained all the idioblasts typical of a leaf: biforines, druses, and both large and small raphide bundles. However, unlike in leaves, these cells were not evenly distributed. Above the point where the spathe was fused with the spadix, druses and biforines were densely distributed (Fig. 7A). Up to 400 druses and up to 30 biforines per square millimeter could be counted. Both kinds of cells were most common toward the tip, becoming rarer toward the junction with the spadix. Cells containing large raphide bundles (giants) were also found, and those with small raphides were common near the junction with the spadix. Below the junction (Fig. 7B), druses were rarer than above (about 20 mm−2), and biforines were completely absent. Small raphide bundles were also rarer, but large ones were more common. Along the spathe edge near the base were many large bundles (Fig. 7C). Where the spathe was fused to the spadix were plentiful strands of box-like raphide bundles, similar to those in petioles and stems (Fig. 7D), aligned parallel to the spadix axis.
Crystals in tissues of Dieffenbachia seguine inflorescence. (A) Spathe sample from near tip shows many druses, along with a number of biforines and one giant raphide bundle. (B) Spathe sample from near base shows many small to large bundles of raphides, but no druses and no biforines. (C) Along spathe edge near base is a scattering of large raphide bundles. (D) Where spathe is fused to spadix, a dense line of box-like bundles runs parallel to axis of spadix. (E) Cross section of spadix at level of staminate flowers shows many scattered druses and small raphide bundles in central pith. (F) Tissue of connective between anthers is capped with a dense crowd of crystals, including druses and biforine-like cells. (G) Anthers are associated with strands of biforine-like cells that run along long axis of organ adjacent to pollen sacs; another example is visible in (F) at upper right. (H) Strand of biforine-like cells at higher magnification. (I and J) Scattered among pollen are prismatic crystals. They are difficult to see under bright field microscopy, as in (I), but are strongly birefringent under crossed polarizers (J). (K) Pair of anthers which have released their pollen; strands of biforine-like cells are still apparent. (L, M) Gynoecia are dense with crystals in their outer walls, but crystals are absent from ovules in center of gynoecia. (L) Single gynoecium at low power. (M) Thin section of gynoecium wall. Crystals include raphide bundles and druses. (N) Crystals are common just beneath stigma, but stigmatic hairs (upper left) contain no crystals. (O and P) Staminodia are dense with crystals. Staminodium in (O) is from a female flower, while that in (P) is one of those constituting a sterile flower; their crystals do not appear to differ. Micrographs A, F, and J were taken with polarization microscopy; the rest are bright field. H–J: bar = 20 μm; all others: bar =100 μm. Abbreviations: d, druse; p, pollen; str, strand of biforine-like cells; others as in Fig. 6 .
Spadices
The central column of the spadix was relatively sparse in crystals. Small raphide bundles were found around the cortex of the spadix. Internally, there were druses and some small bundles of raphides, about 30–60 μm across (Fig. 7E).
Synandria
The outer surfaces of the connectives of the fused staminate flowers were dense with crystals, especially druses (Fig. 7F). Small and large raphide bundles, including some biforine-like, as well as true biforines clearly identified by their expelled crystals, were found lower down, aligned perpendicular to the surface. The lower portion of the connective was nearly devoid of crystals. Anthers were associated with narrow, crystal-dense strands that ran longitudinally alongside the pollen sacs (Fig. 7G). These strands contained primarily biforine-like idioblasts mixed with some bundles of large raphides and bundles of small raphides (Fig. 7H). As noted previously, these biforine-like bundles were significantly smaller than true biforines or biforine-like cells elsewhere in the plant. The pollen sacs contained only prismatic crystals (Fig. 7I, J). The number of prisms in the pollen sacs varied from section to section from many per anther to none. After the pollen was released, the strands of biforine-like bundles were still apparent in the anthers (Fig. 7K), and many prismatic crystals could still be observed in some empty chambers.
Gynoecia
The walls of gynoecia contained a dense concentration of crystals (Fig. 7L–N), including druses and large and small raphide bundles; however, biforines were rare. The ovules and the stigmata lacked crystals.
Staminodia
Staminodia within the pistillate flowers (Fig. 7O) as well as staminodia comprising the sterile flowers (Fig. 7P) contained very dense concentrations of the same kinds of crystals found in gynoecia, including very rare biforine-like cells.
Chemical identity
Crystals from stem, petiole, leaf blade, spathe, synandria, staminodia, and gynoecia were tested with acetic acid and hydrochloric acid (Fig. 8). All tissues treated overnight with acetic acid retained the usual complement of crystals (Fig. 8A, D, E, G), whereas tissues treated overnight with hydrochloric acid generally lost their crystals (Fig. 8B, C, F, H, I). Biforines, box bundles, and druses vanished on hydrochloric acid treatment; however, small raphide bundles sometimes remained after treatment (Fig. 8H). These appeared degraded and disorganized compared to bundles in acetic-acid-treated samples. Crystals often persisted in thicker sections or in areas of leaf distant from the cut edges. Crystals of all types persisted in some leaf tip samples, but only toward the tip and not near the cut edges. The cell walls of biforines were birefringent and were still visible under polarization, empty of crystals, after hydrochloric acid treatment (Fig. 8B, C, F, H, I).
Stability of crystals in selected tissues to acid treatment. Samples in (A, D, E, and G) were treated overnight with acetic acid as described in Materials and Methods, while other samples were treated overnight with hydrochloric acid. (A–C) Leaf samples. (A) Normal complement of crystals is present on acetic acid treatment; compare to Fig. 6A. (B) Only empty biforine cell walls, which are birefringent, are seen after treatment with hydrochloric acid. (C) Empty biforine at higher magnification. (D–F) Samples of synandria. (D) Strands of biforine-like cells and other crystals remain after acetic acid treatment; compare to Fig. 7G. (E) Prisms remain among pollen after acetic acid treatment; compare to Fig. 7J. (F) After hydrochloric acid treatment, only empty biforine-like cell walls are seen near anther. (G) Pistillate flower, gynoecium at lower left, stigma at upper right, and staminodium at far upper right. After acetic acid treatment, crystals are still visible in gynoecium and staminodium; compare to Fig. 7L, O. (H) After hydrochloric acid treatment, only some small raphide bundles remain visible in a gynoecium; stigma is at right, ovary at left. (I) After hydrochloric acid treatment, only empty biforine-like cell walls are seen in staminodium. All micrographs were taken with polarization microscopy. Abbreviations: ecw, empty biforine-like cell walls; srb, small raphide bundles; others as in Fig. 6. C, E: bar = 20 μm; all others: bar =100 μm.
DISCUSSION
The findings reported here agree with previous studies, but extend these studies to other vegetative and reproductive tissues.
Diversity of crystals
The diversity of crystals in Dieffenbachia seguine is remarkable. Three basic crystal forms were found: druses and raphides, found essentially throughout the plant, and prismatic crystals, found mixed with pollen. Raphides show diversity in their arrangement within the idioblast cell, from neatly bundled (Fig. 2E) to somewhat disorganized (Fig. 2F). Raphides occur in at least two thicknesses and a diversity of lengths. Druses also come in different sizes, larger in stems and smaller in roots, and with occasional, very small druses in leaves. Table 2 summarizes the variety of crystals found in different tissues of D. seguine.
| Organ | Druses | Giant bundles | Overlappingbundles | Biforines or biforine-like | Box bundles | Small raphide bundles | Prismatic crystals |
|---|---|---|---|---|---|---|---|
| Leaf lamina | ++ | + | + a | + | + b | + | − |
| Petiole | + | − | − | (+) | + | + | − |
| Petiole wing | ++ | − | − | − | − | + | − |
| Stem | + | − | − | (+) | + | + | − |
| Root | + | − | + | (+) | + | − | − |
| Adventive root | − | − | − | − | − | − | − |
| Spathe | + | + | − | − | + | + | − |
| Spadix column | + | − | − | − | + | − | − |
| Synandrium | + | − | − | + | − | + | + c |
| Gynoecium | ++ | − | − | (+) | − | + | − |
| Staminodium | ++ | − | − | (+) | − | + | − |
- a Notes: ++ = predominant; + = present; (+) = rare; − = not seen
- a Along edge.
- b Along midvein.
- c Among pollen.
Some crystal types could represent developmental variation. Small or less organized raphide bundles, for example, might be early stages of biforine or giant bundle development. Similarly, microdruses might represent developing druses.
The variety of precise crystal morphologies in D. seguine, and in plants generally, which are not seen in commercially crystallized calcium oxalate (35) suggests that plants impose the morphology upon the developing crystal (3). Idioblasts produce crystals by cocrystallizing oxalate and calcium ions within membrane-bound crystal chambers (35). Particular proteins and/or carbohydrates within the crystal chamber might further control crystal development (reviewed by 35). The diversity of crystal morphologies here described for D. seguine argues that this species must have different crystal developmental programs in different organs and even in different cells within the same organ.
Along with great diversity, crystals in D. seguine show tissue specificity; each tissue has a particular complement of crystals. For example, while druses appear to be nearly ubiquitous throughout the plant, adventive roots, pollen sacs, and ovaries completely lack them. Furthermore, different portions of the same organ may have different crystal types. For example, leaf margins have a greater density of druses than the lamina, as well as overlapping raphide bundles not present in the lamina, but far fewer biforines and small raphide bundles (Table 1). Spathes show even more dramatic variation; druses and biforines are abundant in the upper spathe, but druses are rare and biforines absent in the lower part of the same organ.
In contrast to their diversity in morphology, crystals show no diversity in chemical composition. Three minerals, calcium carbonate, calcium oxalate, and silica (2), commonly form crystals in plants, although calcium phosphate and other calcium salts have been reported in some species (reviewed in 3). The calcium salts of carbonate and phosphate are soluble in both hydrochloric and acetic acids; calcium oxalate is also soluble in hydrochloric acid but not in acetic acid (36). Every crystal type in every tissue of D. seguine could be destroyed by overnight soaking in hydrochloric acid but not in acetic acid, consistent with all of them being composed of calcium oxalate. In some experiments, partially degraded crystals persisted in hydrochloric acid, most commonly distant from cut edges or in thicker sections where the acid would take more time to diffuse, suggesting that overnight incubation was not always sufficient to destroy all crystals. It is possible that persistent crystals were specifically partially protected by some material surrounding them but absent from around other nearby crystals.
We do not know the hydration state of the calcium oxalate in the crystals, whether monohydrate or dihydrate (11). 26 identified both hydrates in tissues of the same plant, Dracaena sanderiana, so it is possible that both hydrates occur in D. seguine as well.
We cannot say whether the crystals found are characteristic of all varieties of D. seguine, including wild forms; however, observations on the cultivar Camille showed the same patterns of crystals in both vegetative and reproductive tissues (data not shown). We also cannot say whether the pattern of crystals in this species varies with different environmental conditions. However, we have not found any differences between plants in the greenhouse and in the growth chambers (not shown). Also, we have examined leaves from plants infested with sucking insects, treated with hormones, or subjected to simulated herbivory and found no differences in the variety and localization of crystals produced (not shown).
Roles of the crystals
A number of roles have been suggested for crystal-containing idioblasts (discussed in 3; 10). The crystals could provide long-term storage of calcium because the crystals apparently can be mobilized and degraded as needed (9; 8; 35). The crystals might also serve as a calcium sink, immobilizing excess calcium because plants regularly absorb more calcium than needed (35).
If the crystals are merely sequestration or storage depots, careful control of crystallization might be essential to avoid disrupting metabolism in adjacent cells, hence the imposition of a specific developmental program on the crystals. However, the diversity of developmental programs seen in D. seguine, even within a single organ like leaf or spathe, remains unexplained.
Because calcium is a critical cross-linking ion in the cell wall, one intriguing suggestion is that crystals play a role in sequestering calcium during cell wall destruction during development of particular tissues such as the air space-filled aerenchyma of cattail leaves (16). Similarly, in the anthers of pepper (Capsicum annuum), crystal formation is associated with the breakdown of cell walls that contributes to pollen release (14). In D. seguine, the strings of biforines seen lining anthers, and/or the prismatic crystals within the pollen sacs could be associated with the breakdown of walls leading to the development and release of pollen.
A structural role for crystals as tissue stiffeners has been suggested (31). The box bundle (Fig. 2H) would seem a good candidate for this role in D. seguine; it is found specifically in locations where extra support would be useful (Table 2)—along the midvein of the leaf, along the back of the petiole, within the stem, along the back of the spadix, and at the branch points of secondary roots (Fig. 6R).
It has been suggested that the crystals could function as deterrents to herbivores (9). In this role, the crystals certainly capture the imagination. Chewing druses would be like chewing sand, and the needle-like raphides would be ideal for penetrating herbivore mouthparts. The ability of biforines to forcibly expel their raphides could function in vivo to drive crystals into the tissues of anything chewing on them. It has been suggested that proteolytic enzymes (33), other toxic proteins (6), glucosides (30), or other toxins (20) may be incorporated into the organic matrix surrounding the crystal. Forcible expulsion from biforines would thus turn raphides into microscopic poisoned darts.
Dieffenbachia, like other crystal-bearing aroids, is reported to be poisonous to humans, causing oral irritation and inflammation, (1; 12). The irritant factor in Dieffenbachia can be destroyed by cooking or drying (28). Indeed, taro (Colocasia esculenta) and konjac (Amorphophallus konjac), among other aroids, are grown for human consumption in the tropics, although they must be dried or cooked before they can be eaten. Drying or cooking prevents active ejection of the crystals from biforines (28) and could conceivably also destroy any poisons associated with the crystals, especially proteins.
34 found that both vertebrate and invertebrate herbivores avoid the portions of desert lily Pancratium sickenbergeri that contain the most crystals. Furthermore, populations of this plant undergoing higher levels of herbivory also had higher levels of crystals (34). Similarly, seedlings of the mallow Sida rhombifolia (Malvaceae) subject to simulated herbivory had a greater concentration of crystals than control seedlings (24). It should be pointed out that neither of these plants have biforines, which are known only in aroids; however, P. sickenbergeri does have raphides and S. rhombifolia has druses.
Druses could function as deterrents simply by abrasion of mouthparts. 25 isolated a mutant form of Medicago truncatula lacking the druses produced by the wild type. 19 showed that the beet armyworm (Spodoptera exigua) prefers leaves of the crystal-free mutant over those of the wild type, and, further, that armyworms eating the wild type leaves gained less mass and had far greater wear on their mouthparts. Even vertebrates might prefer more crystal-free leaves; dental wear from dietary crystals has been documented in human archeological finds (5).
The present findings are largely consistent with crystals acting as deterrents to herbivores. Druses are found in every tissue and could abrade mouthparts of anything chewing them. Crystal-expelling biforines are also widespread, but especially common in leaves and spathes, likely the most commonly eaten portions of the plant. Within the flower, unshed pollen and both fertilized and unfertilized ovules would need particular protection. Consistent with this hypothesis, both synandria and gynoecia are packed with crystals, including druses, and in the case of synandria, biforine-like cells.
Staminodia, on the other hand, should be devoid of any herbivore protection because these are provided as a reward to pollinator beetles and are consumed by the beetles while trapped within the spathe. However, staminodia are packed with crystals, particularly druses and raphide bundles (Fig. 7O, P), strongly suggesting that crystals, in this organ at least, do not deter herbivores. However, crystals in staminodia might lack associated toxins, or there might be some neutralizing agent in staminodia that renders the crystals and/or associated toxins innocuous. It would be interesting to compare the secondary metabolites associated with staminodial crystals with those of crystals from other parts of the plant. It is also possible that pollinator beetles have evolved immunity to staminodial crystals, so that the crystals function as preservatives, deterring herbivores other than the intended recipients.
An unusual crystal in Dieffenbachia is the prismatic crystal found mixed with pollen, but seen nowhere else in the plant. Presumably, this type of crystal plays a role in pollen development, perhaps by sequestering calcium. Although prismatic crystals can be observed in the pollen sacs after pollen extrusion, some may have been extruded with the pollen and may play some role in pollen dispersal and/or in the ability of pollen to germinate and develop upon reaching a receptive stigma; however, it is not clear what such role(s) might be. 4 reported both raphides and prismatic crystals mixed with pollen on the stigmata of a number of aroids, although they did not examine any Dieffenbachia species.
Conclusions
Dieffenbachia sequine contains a diversity of calcium oxalate crystals differing in morphology, size, and tissue localization. It seems likely that they might play a number of different roles in the plant. In particular, box bundles might play structural roles, prismatic crystals might be important for pollen function, and druses and biforines might deter herbivores. Crystals within the staminodia eaten by pollinator beetles, however, suggest either that druses and raphide bundles do not deter herbivores or that there is a more complex relationship between crystals, herbivores, and pollinators that remains to be elucidated.
