|Home | About | Journals | Submit | Contact Us | Français|
Microstructure of the teeth of the sea urchin Lytechinus variegatus was investigated using optical microscopy, SEM (scanning electron microscopy) and SIMS (secondary ion mass spectroscopy). The study focused on the internal structure of the first-stage mineral structures of high Mg calcite (primary, secondary and carinar process plates; prisms) and on morphology of the columns of second-stage mineral (very high Mg calcite) that cement the first-stage material together. Optical micrographs under polarized light revealed contrast in the centers (midlines) of carinar process plates and in prisms in polished sections; staining of primary and carinar process plates revealed significant dye uptake at the plate centers. Demineralization with and without fixation revealed that the midlines of primary and carinar process plates (but not secondary plates) and the centers of prisms differed from the rest of the plate or prism, and SIMS showed proteins concentrated in these plate centers. SEM was used to study the morphology of columns, the fracture surfaces of mature teeth and the 3D morphology of prisms. These observations of internal structures in plates and prisms offer new insight into the mineralization process and suggest an important role for protein inclusions within the first-stage mineral. Some of the 3D structures not reported previously, such as twisted prisms and stacks of carinar process plates with nested wrinkles, may represent structural strengthening strategies.
Most sea urchin and other echinoderm ossicles (skeletal elements) are single crystal calcite, and functionality superior to inorganic calcite is obtained by modulating crystal growth to produce very complex geometry, e.g., the stereom of test plates and jaw structures aside from teeth (Smith 1980) and the wedge structures of certain spines. Sea urchin teeth, the subject of this paper, differ from other ossicles and are ceramic matrix composites incorporating an array of parallel and aligned, geometrically-complex single crystal elements. Initially, the many high Mg calcite elements of the tooth grow spatially separate from the other elements; these are the first-stage mineral of the tooth. Later in development, the parallel first-stage structural elements are connected by separately formed, second-stage mineral of very high Mg calcite. These very high Mg columns (also called disks) cement the tooth into a dense compositionally modulated single crystal.*
This paper describes aspects of the geometrically complex composite present in the teeth of the sea urchin Lytechnius variegatus. One focus is on the internal structure of the first-stage skeletal elements (primary plates, secondary plates, carinar process plates, needles, prisms) and on what this reveals about the growth process and mechanical functionality of the calcite in the tooth. A second focus is observation of the second-stage mineral (very high Mg calcite columns) and the interface between first- and second-stage mineral. Some of the observations on L. variegatus have been anticipated on teeth of other urchins, but it is nonetheless useful to extend documentation of microstructural similarities across the clade.
The structural elements making up each tooth form inside an extracellular space surrounded by multi-nucleated cell syncytia which provide the proteins and ions required for biomineralization. For the primary, secondary and carinar process plates in camarodont sea urchins, these spaces are large parallel sheets originating in the semi-mineralized aboral end of the tooth known as the plumula. The mature tooth is a stack of curved plates surrounding a channel of needles and prisms, all connected by very high Mg columns that provide resistance against inter-element shear. Curved plate arrays are removed from the incisal edge through abrasion during feeding, leaving behind a continuously sharpened surface. The continuous growth, the single crystal nature of the ensemble of tooth elements, the differing non-equilibrium concentrations of Mg in the calcite, and the level of control over the orientation and shape of the elements makes the sea urchin tooth a novel model in which to study the processes of biomineralization.
The geometry of the structural elements of L. variegatus teeth is illustrated in Fig. 1. The sea urchin jaw consists of five pyramids with one tooth each (Fig. 1a, right). In Fig. 1a, left, an isolated tooth (in the same orientation as the pyramid) is shown with the keel “K” (leg of the “T”-shaped cross-section) projecting toward the viewer and the flange “F” (bar across the “T”) behind the keel. The tooth schematic has adoral end up and shows very roughly the orientations of primary plates “PP,” secondary plates “SP,” carinar process plates “CPP,” needles “N” and prisms “Pr” of high Mg calcite. Figure 1b shows a transmission optical micrograph of a polished section of mature L. variegatus tooth with the different skeletal elements labeled.
The tooth’s are aligned to improve functionality. The tooth is a “T”-shaped girder; like an I-beam, this cross-section provides good bending resistance (Märkel and Gorny 1973) with the flange experiencing compressive and the keel taking tensile loading. The fine structure of the stone part is the working edge of the scraping tool (Giesbrecht 1880). The primary plates are aligned more-or-less edge-on to the compression stresses as are the secondary plates (Märkel and Gorny 1973). These two sets of plates may also confine the stone part laterally, limiting displacements of the very fine elements perpendicular to the tooth axis (preventing buckling of these micrometer-sized structures?). During tooth bending, the prisms in the keel are loaded in tension, a geometry that is employed in engineered monofilament composites (Märkel and Gorny 1973). On either side of the prisms of the keel, the carinar process plates may limit bending along a secondary axis (Stock et al. 2003).
Primary plates are the first mineralized elements that form in the plumula. Next, the secondary plates and stone part form, completing most of the volume of the flange. Subsequently, the secondary plates extend forming the carinar process plates while needles emerge from the end of primary plates and broaden into prisms as they extend into the keel between the carinar process plates. At about midshaft where the keel is forming, very high Mg columns, x ~ 0.32 (Stock et al. 2002), grow in the cellular space between first-stage skeletal elements and begin to cement the disparate crystals into a rigid structure. Interestingly, the high Mg and the later-forming very high Mg phases in L. variegatus are in crystallographic alignment, demonstrating that the tooth is a crystal with sharply modulated composition in the three spatial directions (Stock et al. 2002).
Proteins play a pivotal role in mineral formation; and the well-defined onset of second-stage mineral, the large difference in composition between the calcite of the two stages and the precise crystallographic alignment of the two phases suggests very interesting protein modulations occur. Occluded proteins account for up to 0.2–0.25 wt. % of calcite skeletal elements in sea urchins (Weiner 1985; Veis et al. 1986). In L. variegatus tooth plates, internal cavities were observed that most likely contain hydrated proteins (Robach et al. 2005). It is not known exactly how the proteins become buried in a growing calcite single crystal, nor is the spatial distribution of protein fully understood. Earlier studies identified an optically different region at the middle of primary plates (Giesbrecht 1880), a portion of the plate that etched more rapidly forming a midline cleft in the plates (Giesbrecht 1880; Märkel 1990; Wang et al. 1997). Some proposed that this layer might be amorphous calcium carbonate (ACC) trapped in a shell of crystalline calcite (Wang et al. 1997). Anticipating findings presented in Results, it is also possible that the layer is protein-enriched calcite.
Secondary ion mass spectroscopy (SIMS) was applied to mature teeth of L. variegatus to map Ca, Mg and amino acid fragments such as Asp and Ser (Robach et al. 2006). Aspartic acid was strongly colocalized with the very high Mg content regions of the tooth (i.e., the columns). Demineralized specimens showed reduced Mg and Asp signals whereas Ser content had the same distribution in demineralized and mineralized sections, emphasizing the association of very high Mg calcite and readily soluble Asp-rich protein(s).
X-ray diffraction revealed coherence length (i.e., crystallite size) of biogenic calcite of ~200 nm (Berman et al. 1988; Berman et al. 1990; Berman et al. 1993; Aizenberg et al. 1997) but did not consider microstrain (i.e., broadening from dislocation density gradients, etc.) clearly present if one makes a Williamson-Hall plot from the data (see (Cullity and Stock 2001) for what this analysis entails). More recent diffraction studies on L. variegatus teeth (Stock et al. 2002; Stock et al. 2003; Stock et al. 2004) showed coherence lengths similar to those of Berman et al. as well as microstrain of ~0.35%.
Studies of sea urchin teeth (Salter 1861) and larval spicules (Okazaki and Inoue 1976) revealed that the calcite grew in concentric layers, indicating incremental growth of the element (Seto et al. 2004). This implies that the elements of the sea urchin tooth, however they are nucleated, grow incrementally through progressive deposition of calcium carbonate and protein. What, if anything, lies between these layers is not known, and but the initial form of the deposited material may be amorphous calcium carbonate, ACC, (Politi et al. 2004).
The goal of this paper is to describe and interpret certain aspects of the structure of L. variegatus teeth. Optical transmission microscopy, scanning electron microscopy (SEM) and SIMS are employed.
Live L. variegatus sea urchins were pre-fixed in 1% glutaraldehyde in seawater (Instant Ocean®, www.instantocean.com/Instantocean.home) for one hour or until expiration. The teeth were removed and placed into Karnovsky's perfusion fixative overnight at 4°C. After washing three times with 0.1 M sodium cacodylate buffer, the teeth were placed into 18 MΩ deionized water and dehydrated in an ethanol series (30%, 50%, 70%, 90%, 100%) in 15 min. steps. The ethanol was removed and replaced with propylene oxide for 30 minutes and repeated three times with fresh propylene oxide. Teeth were then placed in a 1:1 mixture of propylene oxide and Embed 812 (Epon replacement, Electron Microscopy Sciences) resin mixture and left overnight. The mixture was replaced with a 1:2 propylene oxide:Embed 812 mixture and left uncovered for 4 hours. The teeth were placed into sample capsules and covered in Embed 812 resin and cured overnight at 60°C. A few samples were embedded in Epon following this same procedure. Cured blocks were sectioned with a diamond wafering saw and polished with alumina to a 0.3 µm finish. Samples were sonicated to remove any embedded polishing compound. Some specimens were polished on both sides, allowing for transmission optical microscopy; and others were etched with 0.2% aqueous EDTA (room temperature, 20–60 min depending on the specimen) and were mounted on steel disks using conductive tape for SEM imaging.
Another group of samples (tooth sections cast in plastic) was demineralized in the presence of fixative to reveal organic consitutents (Albeck et al. 1996). 10 ml of cation ion exchange resin beads (Dowex 50 ×8, mesh 50–100) were added to a 50 ml tube with an aqueous solution of 4% formaldehyde and 0.5% cetylpyridinium chloride. Samples (cast in plastic, sectioned and mechanically polished) were placed into dialysis tubing filled with the fixative solution, and the ends were clamped closed. The preparations were then placed into the 50 ml tube, and the ion exchange resin was agitated on a magnetic stirrer for varying times to produce progressively demineralized specimens.
Scanning electron microscopy (SEM) was performed with a Hitachi S-4500 system operated by Northwestern’s EPIC Center. Specimens were imaged with secondary electrons (at 10 or 15 kV) after light coating with sputtered carbon (~ 1 s duration). Electron back scattered patterns (EBSP) could be detected from some carefully prepared surfaces, but because the patterns were too indistinct for analysis or even for useful reproduction, nothing further will be written here.
Secondary ion mass spectrometry was performed using a PHI TRIFT-III TOF-SIMS instrument operated at 25 keV with charge compensation (via an electron beam) at the Keck-II facility at Northwestern University. The Ga ion beam probe size was approximately 0.4 µm, and mass calibration was performed at peaks of CH3 (m=15.0235), C2H3 (m= 27.0235), and C3H5 (m=41.0391). Image sizes were 256×256 pixels, and the images were subject to automatic linear rescaling to maintain contrast. Characteristic positive and negative ions from amino acids (from published poly[amino acid] spectra, i.e., (Mantus et al. 1993; Samuel et al. 2001)) and inorganic material were mapped to the various components of the L. variegatus tooth following the procedure described previously (Robach et al. 2006).
Microscopy of the internal structure of first-stage mineral elements begins the section. Via the fracture surfaces of plate-column assemblies in Fig. 7, the coverage transitions into optical and SEM imaging of the second-stage mineral microstructure, i.e., the columns. The next data related are SIMS images of polished samples containing plates and columns, data that focuses on the organic constituents associated with the ensemble. The section closes with SEM images of carinar process plates (Fig. 15), appended primarily as a cautionary note relating to interpretation of 2D sections through 3D objects.†
Optical and SEM micrographs of a polished and etched tooth (transverse cross section) appear in Fig. 2. At the left is a low magnification optical micrograph of a mature L. variegatus tooth; the symbols have the same meanings as in Fig. 1. Compared to Fig. 1, the ratio of keel length to flange breadth is larger in Fig. 2, indicating that the Fig. 2 section is somewhat more adoral (closer to the incisal end) than that in Fig. 1 and is more mature and contains a greater density of columns linking adjacent first-stage mineralized elements. At the right of Fig. 2, structures are shown from the three boxed areas.
Area 1 in Fig. 2, an SEM image, shows the effect of differential etching at the boundary between the primary and secondary plates. In this transverse section, the secondary plates are significantly inclined, but the primary plates are nearly perpendicular and etching produced a groove running through the center of each primary plate. Three such grooves are labeled by arrowheads. Grooves are not produced in the secondary plates. In the upper right corner of area 1, a few thin strands of unidentified material bridge the gaps between adjacent primary plates.
Area 2 of Fig. 2 shows a cross-polarization (transmission) optical micrograph of carinar process plates. Several columns “C” link adjacent plates. The faint gray contrast at the center of plates (arrowheads) is characteristic of the plates and is much more prominent in the microscope than in micrographs.
In area 3 of Fig. 2, the prisms of the keel show several types of features under cross-polarization optical microscopy. The lowest two rows show small cores with dark contrast at the center of the prisms. These cores do not have sharp edges in the micrograph, but their outer diameter is on the order of 5 µm. Observations elsewhere on this section and on other specimens suggest that the precise core diameter may be related more to observational conditions or specific section thickness than to an intrinsic characteristic dimension within the prism. The authors identify these cores as the “axial cylinders” described by Giesbrecht (1880) and consider the structure producing this contrast in the Discussion. Giesbrecht’s drawings from thin sections literally show small cylinders of contrast running along the center of prisms.
In the upper portion of area 3 in Fig. 2, bands of strong contrast extend from the center to the outer face of various prisms. These features contrast differs from polishing scratches, and in area 3 these bands are parallel or nearly parallel to each other. In a few prisms (one immediately left of and one immediately to the right of the “P”), both core and band contrast are seen together. Note that the prisms containing both features separate the region containing only the bands from that containing core contrast.
Figure 3 shows SEM images of a lightly etched keel before columns have linked adjacent prisms. The arrowhead in a. shows the region enlarged in b. The enlargement shows an onion-skin structure within the prisms. Etching is greatest near the center of the prisms, about where the dark cores appeared optically, and least on the outer portions of each prism. Grooves “G” recapitulate the structure seen in dark contrast bands of optical micrographs (area 3 of Fig. 2).
Figure 4 shows transmission optical micrographs of the keel with wider field of view and/or higher magnification than shown earlier. The transverse sections from two different teeth were oriented slightly differently, and adjacent plates were not yet cemented together. The pink staining in Fig. 4b follows midline positions more-or-less equivalent to those in Fig. 4a. In Fig. 4b, a small area of calcite pulled out from the section (labeled “po”). A series of small cracks “cr” cut many of the carinar process plates in the same vicinity. Several cracks cross entire plates, but most, at this magnification, appear to stop at the midline features.
Figure 5 shows SEM images of differential polishing of a carinar process plate by 48 hr exposure to water. This transverse section is somewhat inclined relative to sections of carinar process plates shown elsewhere in this paper. In Fig. 5a, steps in the fracture surface run across the thickness of the plate. The plate etched most its center and least toward its adoral and aboral surfaces; note that only one surface is shown in its entirety (right side). Onion-skin layering is present, similar to that pictured in Fig. 3b. In Fig. 5b, a higher magnification view is shown of the midline portion of a plate adjacent to that in a. A high density of equiaxed‡ nodules and significant empty space is present.
Figure 6 shows secondary electron images of carinar process plates in a longitudinal section of a tooth (i.e., a section parallel the tooth axis and perpendicular to the transverse section shown in Fig. 1b, ,2,2, ,3,3, etc.). Concurrent fixation and partial demineralization with ion exchange resin retains organic material that was removed in the procedure used to expose the midline structure in Fig. 5. The deep groove at the midline shows this was the most rapidly removed portion of the tooth, and onion-skin layers appear to either side of the groove. The retained cross-linked material is concentrated at the midline, and the prominent arcs of charging material, despite carbon coating and as covered in the Discussion, is interpreted as condensation of occluded macromolecules released during mineral dissolution. Further, arcs of charging material appear in all of the plates exposed in the section from which Fig. 6 was recorded, and all of these arcs have the same orientation as those in Fig. 6.
The SEM micrographs of Fig. 7 show unetched (artificial) fracture surfaces of primary plates “PP”; Fig. 7a–c are from mature sections of the tooth (high concentration of columns “C” joining adjacent primary plates “PP”) while 7d shows a plate from a position before columns were formed. As expected from many earlier studies, cleavage characteristic of inorganic calcite fracture is not observed. The crack sometimes followed the column-primary plate interface, exposing the surface of columns or primary plates to view (“ic” and “ip”, respectively in Fig. 7a,b). These surfaces appear quite flat and emphasize the tendency for column-plate interfacial fracture when the externally imposed stress field allows. At several places the columns cantilever over positions where primary plates have pulled away (e.g., “hc” in Fig. 7a,b). Elsewhere (e.g., upper left of Fig. 7a and Fig. 7c), the fracture plane cuts directly across adjacent plates and columns, but examination of many other fractured surfaces suggests this is a relatively minor fracture mode. In Fig. 7a,b, the thickness of the columns and of the plates vary somewhat from plate to plate.
Several features visible in Fig. 7 characterize primary plate fracture surfaces. A small amount of material adheres to primary plates midlines. Every couple of micrometers, a larger “hanging chad” decorates the plates’ fracture surfaces; these are particularly prominent to the left of “PP” in Fig. 7a. Just to the right of “ic” in Fig. 7a, a primary plate has fractured in a series of regular undulations spanning the plate’s surface. Small steps are also sometimes on the fracture surface (Fig. 7d) and appear to angle from the midline (which exhibits strong charging). The surface roughness might be termed “river lines” and is a commonly encountered surface characteristic in brittle fracture. The fracture surface steps echo those seen in the etched surface of Fig. 5a.
When fracture exposes intact columns, the depth-of-focus of SEM allows one to peer between adjacent undisturbed columns and observe the shape of the growth surface of the columns. At the stage where the columns nearly fill the volume between primary plates (Fig. 7), the very high Mg columns have slightly concave growth surfaces.
Figure 8 focuses on columns at mid-stage of formation (Fig. 8a,b) and after nearly complete consolidation (Fig. 8c,d). Figure 8a,b shows a negative cast of the columns between plates, the polymer filling the space occupied in the living tooth by the cellular syncytia and being subsequently exposed by complete demineralization of the tooth. Although there is considerable variation in column diameter, the column centers appear to be fairly regularly spaced. As the secondary electron image of Fig. 8d shows, positions where adjacent columns have recently linked are indistinguishable from positions in the middle of the exposed columns.
Figure 9 shows SEM micrographs of a mature portion of the keel after it has been fractured to expose the columns. Figure 9a shows a lower magnification view of the prisms (left side) and carinar process plates (right side) of the keel with the positions of the higher magnification images in 9b and 9c indicated by the asterisk and black disk, respectively. The lengths of both the prisms and the carinar process plates are evident in Fig. 9a. In Fig. 9b columns densely cover the plate in most places but have adhered to the opposite fracture surface elsewhere. A number of adjacent columns have grown together (lower right portion of Fig. 9b), and some fracture debris remains where columns have been pulled away from the plate. The field of view in Fig. 9c spans the prisms and carinar process plates and shows pronounced curvature of the medial portions of these plates (see also Fig. 4), a subject that is taken up again in the Discussion. By and large, the prism cross-sections are equiaxed, but the prism adjacent to the carinar process plates in Fig. 9c grew distorted to fill the space at the medial edges of the carinar process plates (see also the following paragraph). At some locations the columns cemented two prisms together; these columns appear flat in Fig. 9c. Elsewhere, columns fold across adjacent prisms on the exposed surface, i.e., these columns connected three bordering prisms. Some columns on the prisms have grown together.
A negative casting of the plates and prisms in a keel appears in Fig. 10 and shows the interlocking waviness of adjacent carinar process plates (i.e., amn Epon cast tooth, transversely sectioned and demineralized). The sharp wrinkles “*” are particularly noticeable and confirm the observation of substantial medial curvature of the carinar process plates described in Fig. 9c.
Figure 11 shows SEM images of prisms fractured from a keel. In Fig. 11a, a group of more than a dozen prisms is pictured, prisms that remain attached or only very slightly separated. The longitudinal axes of these prisms run from lower left to upper right in Fig. 11a, and an enlargement of the area to the right of the black disk is inset at the upper left. A set of parallel, uniformly spaced surface indentations run across the prisms. These make an angle of 20–30° with the prism axes and appear every 20 µm or so along the prisms. An isolated prism is shown at higher magnification in Fig. 11b.
Many prisms do not have the pattern of surface indentations revealed in Fig. 11a,b. For example, the ends and shafts of the exposed prisms in Fig. 9 lack indentations. Some prisms isolated from the keel have well defined faces (e.g., Fig. 11c). In Fig. 11c, the disks on the faces indicate that this prism is from a relatively mature portion of the keel. The prism twists along its length, and, if a series of transverse thin sections were prepared from a keel containing such a prism, the prism’s cross-section would change gradually across the series of sections. Among the prisms fractured from teeth, twisted prisms are neither rare nor common, but it is not possible with the data on hand to estimate what fraction of prisms twist, what fraction are indented or what fraction have neither feature. Serial sections (thin sections, sequential polished surfaces or microCT) remain to be produced that are adequate to show the spiraling of prisms in situ.
The variable spatial distribution of protein was studied with SIMS in polished sections of tooth. Figure 12a shows a negative ion image tracking CN− ions over a polished longitudinal section of carinar process plates. The CN− ions reflect all proteins, and the lighter the pixel in Fig. 12a,b, the higher the ion signal. The dark areas are holes or depressions in the surface between interplate columns. Three bright parallel lines (arrowed) at the center and surfaces of the carinar process plates can be seen in Fig. 12a. Figure 12b is a plot of the white profile line in Fig. 12a, confirming the three peaks of local protein enrichment.
Positive ion SIMS was then used to track the characteristic fragmentation product of glycine, another indicator of total protein, to confirm the negative ion data. Figure 13 (left and center panels) of primary plates show SIMS images from all ions and from the glycine characteristic peak (m=30.03), respectively. A broad midline peak is consistently present in both the total ion and glycine images. The midline peaks seen in the total ion image are enhanced slightly due to light etching of the surface during polishing. Despite this, the glycine image exhibits a broad peak along the center of the plates, with localized peaks on the surfaces of the interplate columns (Fig. 13, center). Figure 13 (right) is a transmission optical micrograph of a similar area on the same sample used for SIMS; this section was stained with toluidine blue. Faint staining can be seen along the midlines of the plates, further confirming a local enrichment of occluded organic macromolecules within the single crystal elements.
Figure 14 presents SIMS maps of total ions, Asp (asparagine/aspartic acid) and Ser (serine) peaks and shows additional details of the primary plate-column assembly. Total ion yield is enhanced at plate mid-line (left panel of Fig. 14) and decreased at boundary between very high Mg (vh, columns) and high Mg (h, plates). Note the nonuniformities in total ion map due to a glitch in SIMS system; these do not affect interpretation of the maps. The sharp pair of peaks running along the center of the plates in the total ion image (Fig. 14, left, and labeled with asterisks) appear in maps for many masses, none of which could be unambiguously attributed to a particular amino acid fragment. In the right panel, the Asp signal is low at column/plate interface (dashes) with weak broad maximum at the plate center (asterisks). The middle panel shows the Ser signal is high at column/plate interfaces (positions where the total ion and Asp signals are low), i.e., it is anti-correlated with the latter two signals in plate structure.
Figure 15 shows carinar process plates in a polished section transverse to the tooth axis. One plate in this field of view appears to bifurcate, and two regions of midline structure can be seen (labeled by the white disks in Fig. 15b). The plate does not, in fact, split, but this section intersects the plate at either side of a wrinkle like in Fig. 4, ,9c9c or or10.10. Considerable caution must be exercised, therefore, in interpreting single sections through structures such as sea urchin tooth.
Many of the observations on L. variegatus epand the range of known common microstructural characters of sea urchins with “T”-shaped teeth. Similarities between different order and families cannot be assumed because major differences do exist: carinar process plate geometries diverge in different families (Stock et al. 2004).§
Salter (1861) described soldering particles (very high Mg calcite disks or columns), and Giesbrecht (1880) expanded on Salter’s description of columns and observed internal structure in prisms, both studies on Echinus esculentus (family Echinidae). Märkel and coworkers (1973, 1976) discussed tooth microarchitecture in Paracentrotus lividus (family Parechnidae) and in Sphaerechinus granularis, an urchin from the same family as L. variegatus (Toxopneustidae). Wang et al. (1997) also discussed 3D microstructure in P. lividus and noted that holes etched in the centers of primary plates and of prisms.
The internal structure of needle-prisms and what these observations may reveal about their growth is discussed in the first subsection. Carinar process plate structure is discussed next followed by primary plate structure. The fourth subsection focuses on the columns and their surroundings, and the fifth on inferences about growth in the syncytical spaces that can be made from the observed microstructures. The last subsection discusses evolution and the notion of observed microstructures being optimized for functionality.
The central regions of dark contrast at the prism centers (area 3 of Fig. 2) suggest that the prisms are centered on a Volterra dislocation (Volterra 1907), that is, a dislocation with screw character, a hollow core and a Burgers vector a multiple of b, the magnitude of the smallest closure gap in a circuit of lattice positions around the core of the dislocation ** (Weertman and Weertman 1992). The basis for this identification follows.
Threading dislocations (with screw character) exiting the surface of a crystal allow it to grow at lower supersaturations and much more rapid rates than would be possible if each successive layer had to nucleate independently on the newly formed surface (Frank 1949; Burton et al. 1951). The screw dislocation displacement produces a step on the surface that acts as a preferred nucleation site and, as more atoms or molecules are added, the step spirals around the dislocation core. In flat-faced crystals, growth pyramids form. Long (up to 1 cm), very large aspect ratio crystals called whiskers (diameters 1–200 µm) have their axes following the line of a screw dislocation; the same spiraling elongates the whisker rapidly along the dislocation line direction while growth along the whisker’s other axes is much slower. Whiskers grow in molecular as well as elemental crystal systems, for example, in silicon nitride (Kawai and Yamakawa 1998), calcium carbonate (Yu et al. 2004; Shen et al. 2007; Chen and Xiang 2009) and apatite (Suchanek et al. 1995; Teshima et al. 2006).
Threading dislocations can have Burgers vectors larger than b, but, given that the elastic energy of a dislocation is ~GB2, where G is the shear modulus and actual B is the Burgers vector (b, 2b, 3b or larger), formation of super dislocations is energetically very unfavorable (Weertman and Weertman 1992). The elastic energy of a crystal-growth-related super dislocation is reduced if it forms a hollow core screw dislocation, i.e., a Volterra dislocation (Volterra 1907). Hollow core dislocations occur in SiC (Vetter and Dudley 2001; Presser et al. 2008) and in steroid (Stoica et al. 2007) crystals, the former having diameters larger than 10 µm and the latter 100 nm. Hollow core dislocation sources for spiral crystal growth were seen in inorganic calcite (Hillner et al. 1992). The optical contrast in Fig. 2, area 3, and the etched prism surface in Fig. 3 suggest, therefore, that hollow core dislocations with dislocation-related surface steps have been observed and that the prisms of sea urchin teeth are based on this crystal defect. The cores of the prisms might then be decorated or even filled with macromolecules or mixed with poorly crystalline calcium carbonate or with remnant ACC.
Current understanding of biocalcite formation was recently reviewed (Meldrum and Cölfen 2008), including ACC. Amorphous calcium carbonate growth was documented in sea urchin embryos (Beniash et al. 1997) and has been advocated as the precursor for single crystal calcite in sea urchin stereom and teeth. In freshly extracted lamellar needle complexes from sea urchin teeth, optical birefringence developed gradually over 24 hr in water; this indicates ACC was crystallizing (Ma et al. 2007). It is possible, therefore, that ACC fills the prism syncytial spaces initially and subsequently transforms to calcite, but this mechanism is difficult to test.
Crystal growth rate can provide important clues about non-equilibrium processes producing materials such as sea urchin calcite. The effect of various aqueous environments on growth hillocks’ step morphologies and growth rates has been examined, e.g. (Orme et al. 2001). In young adults of Strongylocentrotus purpuratus, teeth renew in approximately 75 days (Holland 1965); growth rates are similar in P. lividus (Märkel 1969). Assuming that this growth rate pertains to L. variegatus, a 20 mm long tooth would elongate at ambient temperatures at a rate of just over 10 µm/hr.
The extension rate of prisms is independent of the tooth replacement rate: the prisms grow at nearly right angles to the tooth’s longitudinal axis. Based on the series of transversely oriented microCT slices reported earlier for a similarly-sized L. variegatus tooth (Stock et al. 2002), prisms grow to a length of ~ 1 mm while the tooth extends ~ 7 mm, i.e., over a period of about 25 days. This equates to a mean growth rate of ~ 2 µm/hr, about one-fifth of that cited above for tooth replacement. Unlike the formation of the very high Mg calcite columns between impermeable primary skeletal elements, sources of cations and anions interior to the sea urchin are not far away.
The prisms widened from very small diameter needles growing from the lamellar-needle complex. Figure 16 shows needles attached to the lamellae in a L. variegatus tooth; the mean initial needle diameter is <dp> ~ 1 µm, consistent with that of P. lividus at an equivalent state of development (Wang et al. 1997). Mature prism diameters are 10–15 µm in L. variegatus (e.g. Fig. 3), not too different from the diameters of ~ 20 µm reported for P. lividus (Wang et al. 1997). By way of comparison, inorganic chlorapatite whiskers have been grown with lengths 0.2 mm < l < 4 mm and with aspect ratio between 20 and 80 (Teshima et al. 2006).
Needle-prisms with initial diameters of ~ 1 µm that widen to ~15 µm diameter over 1 mm length, suggest that the structure extends 60–70 times more than it widens. One interpretation is that the axial prism (growth) rate is that much more rapid than expansion along the other directions; this would be a whisker growth mechanism as is seen in isolated inorganic materials. Another possibility is that lateral spatial confinement (adjacent distinct and well-defined syncytial spaces restrict enlargement in all directions except along the prisms’ axes, i.e., as in inorganic growth of single crystal calcite in replicas of stereom (Park and Meldrum 2002)).Either or both mechanisms could be operating, but further work is required before one can provide a definitive answer.
The greater etching of the centers vs. peripheries of prisms (Fig. 3) has been observed before (Wang et al. 1997). The surface steps seen in Fig. 3 and the bands of contrast seen in Fig. 2, are, to the best of the authors’ knowledge, new observations (Surface steps on growing crystals, especially those linked to dislocations, are a mechanism of rapid crystal growth). As mentioned in the Introduction, etching reveals layering in echinoderm calcite, interpreted in terms of successive additions of material (Salter 1861; Okazaki and Inoue 1976; Seto et al. 2004), and the structures observed in L. variegatus certainly do not differ in this respect.
The prisms’ external shapes may strengthen the tooth. Twists in the prisms, i.e., spiraling of flat faces as seen in Fig. 11c, may hinder slippage of adjacent prisms past each other during loading of the tooth. The indentations in prisms where they grew adjacent to the carinar process plates (Fig. 9 and 11a, b) may also impede prisms’ sliding past the carinar process plates during bending and reduce any structural weakness of this structural transition. In engineering, this type of structural strengthening is termed mechanical interlocking and appears in applications ranging in scale from tens or hundreds of meters (bridges) to hundreds of nanometers through tens of micrometers (reinforcements in composite materials).
Crossed polarization reveals different optical properties at midline vs outer thirds of the carinar process plates (Fig. 4a). The uptake of dye at the midline (Fig. 4b) may reflect a region containing significant organic material. The contrast of the midlines in Fig. 4b, however, is not that different from that of the small cracks, probably an artefact of the sectioning and polishing, running across the plates. Closer examination of some stained midlines suggests these are actually microcracks following tracking plate center (second and fourth arrows from the left). The stained carinar process plates in Fig. 13, however, do not appear to be cracked. These data plus the results of simultaneous fixation and demineralization demonstrate not only that the centers of the primary and carinar process plates differ from the surrounding material but also that macromolecules segregate there.
Figure 4 clearly shows the surfaces of the carinar process plates are quite wavy and the contours of adjacent plates follow each other quite closely. The nesting, non-planar surfaces, in particular, the sharp wrinkle in each carinar process plate (labeled “*” in Fig. 10), appear to prevent sliding of the plates past each other and to strengthen the structure during multiaxial loading (e.g., that might be encountered during tooth rasping on uneven surfaces). This is a second example (twisted prisms being the first) where mechanical interlocking plays a role in strengthening the tooth.
In Fig. 6, the charging material at the center of carinar process plates lies in parallel, curved arcs at the plate midline. The arcs in all of the plates in this specimen have the same alignment (concave right in Fig. 6) and reflect the topography of the midline groove (a slight valley) and oblique SEM viewing angle. The charging material is clumped macromolecules from previously removed calcite. Given that only 0.2–0.25 wt.% of organic macromolecules are occluded in the tooth (Weiner 1985; Veis et al. 1986), the data in Fig. 6 and in other figures suggest strong concentration of the macromolecules in the midline volume.
The water-etched structure within the carinar process plate in Fig. 5 appears granular with the more soluble material removed. These granules have diameters on the order of 100 nm. This may relate to the mesoscopically structured crystal model for biogenic calcite: small nanocrystals of crystalline calcite surrounded by thin layers of ACC with the array of the calcite nanocrystals having a single crystallographic orientation (Meldrum and Cölfen 2008). Such mesocrystal structures have been documented in sea urchin ossicles (Oaki and Imai 2006). Dissolution of ACC is much more rapid than calcite, so what is seen in Fig. 5 is consistent with a mesoscopic calcite –ACC structure. If plate is mesocrystalline and the central portion of the plate is ACC rich, the macromolecules localized there (Fig. 6) may stabilize the ACC. Further, the observed crystallite size in L. variegatus teeth (Stock et al. 2002) and in other echinoderm ossicles (Berman et al. 1990) differ little from the granule dimensions in Fig. 5.
Figure 2 showed the centers of primary plates etched more than the surrounding material of the plate. Figure 13 shows the mid-plate volume stains with an organic dye, much the same as the centers of the carinar process plates; and the total ion map of Fig. 13 showed two lines of enhanced yield bounding the mid-plate volume within each primary plate. Similar pairs of parallel lines were visible in the total ion map of Fig. 14, and increased Asp signal was observed from the volumes between these pairs of lines. There may be very marginally more Mg in the Asp-rich primary plate centers than in the outer parts of the plates (data not shown), but this difference may be nothing more than an artefact of enhanced ion yield at the plate centers and is, at any rate, much smaller than the difference in Mg between columns and plates reported earlier (Robach et al. 2006). These observations are consistent with growth of paired proto-plates that fuse into a single primary plate (Märkel 1969; Wang et al. 1997). Further, this suggests that differences between reported L. variegatus primary plate spacings (Stock et al. 2003) and those of other species (Märkel 1969) reflect the former being on teeth before proto-plate fusion and the latter on post-fusion plates.
In L. variegatus like in other sea urchins, fracture surfaces in primary plates have conchoidal not cleavage character, i.e., the undulating surfaces (Fig. 7a). Similar sized crack steps in fractured P. lividus teeth (Wang et al. 1997) appeared more crystallographic than the glassy fracture path of the undulations. Fracture surface curvature may be from interaction of the advancing crack front with occluded proteins, i.e., protein containing nanocavities observed recently (Robach et al. 2005). The accompanying small flakes of material adhere to the surface and tend not to cross the midline (Fig. 7a–c). Small fracture steps forming chevrons centered at the primary plate midlines (Fig. 7d) also suggest somewhat different fracture properties for the midline and its surroundings. The undulations, flaked material and fracture steps confirm that energy absorption during fracture is increased relative to inorganic calcite, and fracture paths in primary plates may be quite anisotropic for different imposed 3D stress fields.
The fracture plane sometimes cuts directly across the high Mg plates and the adjacent very high Mg columns (Fig. 7c), suggesting that their fracture properties do not differ appreciably. Elsewhere, geometry may force propagating cracks along the interface between first- and second-stage structures (Fig. 7a,b). The L. variegatus data, therefore, are not at all clear with respect to the relative fracture resistances of the two compositions of calcite or of the interface between the two. That being said, fracture parallel to the primary plates’ and secondary plates’ planes is firmly established as the sharpening mechanism in sea urchin teeth.
The slightly concave growth surfaces of the columns (Fig. 7) suggest that columns grow slightly more rapidly near their border with the bounding plates than at the center of the channels between the plates. The SIMS data (Fig. 12) show remnants of organic material (protein) at the column-plate interfaces, and diffraction data demonstrate that the columns and first-stage skeletal elements (plates of all types, needle-prisms) are crystallographically aligned (Stock et al. 2002).
Together, these data suggest that the columns are not in continuous contact with the bordering plates but have some decoration of organic material at the interface and some atomic-scale contact between the two phases, contact that nucleates growth of the later-forming, very high Mg phase (second-stage mineral) in the same crystallographic orientation as the first-stage mineral. If the orientation of the very high Mg calcite is dictated by a narrow crystalline bridge (from the first-stage mineral and) spanning the membrane of the syncytial space, then the situation is analogous to the stacked brick and mortar structures of calcium carbonate in mollusks, e.g. (Lin et al. 2008). Alternatively, the very high Mg calcite of the columns might grow completely independently of the previously present high Mg calcite with an orientation mediated by the syncytical space membranes.
In Fig. 8a,b, the casting of columns shows that they covered the visible area fairly uniformly but they varied somewhat in size. The columns in Fig. 8c,d show even more variability in diameter and have more anisotropic dimensions than in Fig. 8a,b. At the stage of growth shown in Fig. 8c,d, adjacent columns have merged smoothly at numerous points, at least at the magnifications accessible to SEM, consistent with their forming an extended single crystal.
The internal structures and various forms of mineral elements follow from the spaces and environments created by the odontogenic cell syncytia whose structure changes in various parts of the tooth. How the same cell syncytia first grow the first-stage mineral of the tooth (primary, secondary and carinar process plates) of high Mg calcite and later grow the second-stage mineral (columns) of very high Mg calcite is an important question.
A parallel study focused examined the proteins that might be involved in these steps and found distinct mineral-phase-related proteins that could serve these functions (Alvares et al. 2009). Two of these, already reported by others (Illies et al. 2002; Cheers and Ettensohn 2005), were characterized, cloned and immunohistochemically localized within different tooth sections. These data, to be reported elsewhere, clearly showed that the mineralization-related proteins may have direct action in the mineralization process and regulatory activity on cell function.
Quantification of the rate of column growth and what it would reveal about the rate of calcium, magnesium and carbonate transport within the cell syncytia and between the impervious first-stage skeletal elements deserves further study. So far as the authors are aware, this avenue of research has not yet been investigated.
In needle-prisms, x-ray diffraction suggests that Mg content rises as they elongate (Wang et al. 1997; Stock et al. 2002; Ma et al. 2009). Unless the syncytia surrounding the needle-prisms are replaced, they must gradually switch the composition of their crystalline product by introducing more Mg as the prisms extend. The columns, the second-stage mineral, and their single composition of the highest Mg content found in the tooth, therefore, represent the end composition of this continuum. If this scenario is correct, then the questions remain: why does the calcite composition begin to change, and why do the columns begin to form at a certain developmental stage?
Note that the prisms and carinar process plates elongate in parallel as the keel develops. This suggests the following hypothesis: Mg content varies along the length of the carinar process plates. The hypothesis could be directly tested via microbeam x-ray diffraction on an isolate carinar process plate.
The orientation of the calcium carbonate crystal “bricks” in abalone nacre is controlled by the formation of bridges in perforations in the organic layers defining the crystal growth space, e.g., (Lin et al. 2008). The present authors speculate that such bridges may cross cellular syncytia in the plumula and transfer crystallographic orientation to adjacent syncytial spaces, i.e., to the array of primary plates. This is essentially the same idea as was discussed with respect to column orientations relative to adjacent first-stage mineral elements (preceding subsection). If such bridges form, they occur very early in organization of the plumula, are very small and may be transient. This mechanism would apply to crystal growth either from solution or from solid-state transformation of ACC to calcite.
Indentations were observed in one side of prisms in Fig. 9 and and11.11. The indents occur every ~20 µm, consistent with measured carinar process plate spacing in L. variegatus (Stock et al. 2003). Because the array of prisms do not extend beyond the end of the stack of carinar process plates in the developing keel (e.g., Fig. 9–11 of (Stock et al. 2003)), the indents appear to result from syncytia distorting slightly to fill open space at the medial edges of the stack of carinar process plates. The twisted cross-section of prisms away from the carinar process plates (e.g., Fig 11c) likewise may reflect distortions of the syncytial spaces in which the prisms grow. If distortions in the unsupported syncytial envelopes have a consequence of strengthening the structure, then more rigid syncytial organization might not evolve.
Attachment of needle-prism through lamellar needle complex has been long identified as reducing the tension transmitted to the anchoring structure, the plates of the flange (Märkel and Gorny 1973). Although this type of connection is desirable, the attachment point is near the structure’s neutral plane for the primary bending mode. With this in mind, the lamellar zone coupling needles to the primary plates may be better viewed as a transitional crystal growth region where the syncytial cell organization for plates morphs to an organization delimiting small diameter, equiaxed growing spaces. What controls the switch is an interesting, unresolved question. Stress related to action of the mature, chewing portion of the tooth is unlikely to be the trigger because the lamellar-needle complex forms aboral to the pyramid’s tooth slide, and one presumes this portion of the tooth is protected from the external environment.
Slight concavity of the growth surfaces of columns (Fig. 7) suggests growth near the interface with the first-stage mineral (high Mg calcite) is more rapid than at the center of the growth plates. The difference in growth rates cannot be very large or the morphology of distinct columns would not be observed. Instead, the columns only coalesce at very late stage when most of the volume of the syncytial space has been filled with very high Mg calcite (Fig. 8c,d).
The SIMS data (Fig. 12–14) showing protein decorating the interface between high and very high Mg calcite suggest remnants of syncytia persist after mineralization is complete. This material presumably defined the growth space for high Mg calcite and may or may not have isolated the compartment for growth of very high Mg columns from the previously formed mineral.
The fact that a structure exists may not indicate that a structure provides an evolutionary advantage. Consider as an example whether teeth possessing carinar process plates provide an evolutionary advantage. It was suggested that the carinar process plates provide resistance against keel fracture during multiaxial stressing (Stock et al. 2003). More robust teeth would be advantageous, but this interpretation remains tentative because, at the same time that teeth with carinar processes were evolving, Aristotle’s lantern (the jaw structure of sea urchins) developed a stiffer structure through cross-bracing, i.e., bridging of the epiphysis across the foramen magnum (Smith 1984).
Interpretation of improved functionality of biomineralized structures must, therefore, be placed in (or tested in light of) the evolutionary or developmental context lest undue weight be given to incidental (evolutionary) changes or to patterns retained from earlier growth stages. This consideration applies to the aforementioned carinar process plates in camarodonts, the formation of high and very high Mg calcite in all sea urchin teeth, the mirrored crystallographic orientations across the tooth’s plane of symmetry, and other aspects of unique tooth structure.
The introduction and refinement of mineralized tissues has given a wide array of organisms powerful evolutionary advantages over competitors without such tissue. Prominent are biominerals based on the cation Ca. In the case of vertebrates, very fine nanocrystallites of a calcium phosphate phase, carbonated apatite, are the mineral building block used in bones and tooth. As discussed above, invertebrate echinoderms employ a calcium carbonate phase, calcite, in the form of micrometer- to millimeter-sized single crystals.
From a materials science and engineering perspective, vertebrate and echinoderm mineralization involve production of contrasting material types, the evolution of which confer particular strengths and weaknesses. Vertebrate bone is a discontinuously reinforced composite, specifically, a matrix predominantly of water-stabilized type I collagen “filled” with nanoplatelets of apatite. Sea urchin teeth are, in the authors’ opinion, continuously reinforced composites.
In manufacturing, discontinuously reinforced composites offer much greater latitude in changing component geometry and dimensions compared to other, more complex composite designs (see below). Specialized functionality in collagen-apatite structures is obtained via different levels of structural hierarchy as discussed elsewhere (Weiner and Wagner 1998; Currey 2002), and this allows considerable plasticity in modifying organs of mineralized tissue (bone) to respond to changing environments, not only over evolutionary time scales but also to some extent during the life of an individual††.
The mineralization of sea urchin teeth differs from that of vertebrates not only in the calcium salt employed but also in the complex morphology of the mineral phase. In engineering, composites with continuous reinforcements are used when performance is required that is superior to that of the same matrix and the same reinforcement in discontinuous form. Quite complex geometries and further optimized performance can be obtained with continuous reinforcements by first forming complex, 3D preforms (via stacking cloths or weaving filaments) and subsequently joining the reinforcements in second processing step such as liquid or vapor phase infiltration of the matrix phase, e.g., (Lee et al. 1998). The sea urchin tooth, therefore, can be regarded (however loosely) as a continuously reinforced composite with the first-stage mineral of the sea urchin tooth as the preform (reinforcement) and the second-stage mineral (columns) as the consolidating phase (matrix).
Calcite is a wretched structural material, and it is remarkable that the sea urchin obtains such functionality from it. There are, however, certain consequences inherent in the sea urchin’s mineralization strategy, consequences suggested by the manufacturing analogy. Continuously reinforced manufactured composites require much more complex processing systems than discontinuously reinforced composites. Such processing systems are enormously expensive and are not easily altered, i.e., they are not what one would choose for flexible manufacturing. In the context of biology, one wonders obtaining whether achieving reasonable functionality from calcite via the complex composite design route has required sea urchin to develop a concomitant, inflexible mineralization process. If so, this is consistent with the sea urchin’s success in populating specific niches but not evolving to the range of sizes of vertebrates or to occupy the range of habitats in which vertebrates are found.
Despite the difference in composite “design” between vertebrates and echinoderms, the proteins related to mineralization of calcium-based skeletal tissue have remarkable similarities. Therefore, studying differences in the mineral end products of one group will continue to illuminate processes in the other.
This research was funded by NICDR grant DE001374 (to AV). Use of Northwestern University SIMS and SEM facilities (listed in Materials and Methods) are also acknowledged.
*Three points should be noted here. First, the terminology of “first-stage” and “second-stage mineral” is adopted for the high Mg phase (composition Ca1−xMgxCO3 where x ~0.13 for Lytechnius variegatus) and for the very high Mg phase (x ~0.32, see Stock et al. (2002)), respectively, instead of the older terminology “primary” and “secondary calcite”. This avoids possible confusion with primary plates and secondary plates, both of which are first-stage mineral. Second, as discussed by Märkel (1969), the tooth could be described as a bi-crystal that developed from two stacks of plates, each stack having a slightly different crystallographic orientation. Third, although some authors insist that the disks (columns) are polycrystalline, x-ray diffraction demonstrates that the columns plus plates or prisms are crystallographically aligned in addition to being in contact, see the Background.
†There is a Fig. 16 that appears in the Discussion. It is placed in that section because the figure is important only insofar as it supports a point of discussion and otherwise would not have been included.
‡Equiaxed, a common materials science/engineering term, literally means “with equal axes”, i.e. structures that do not have any dimension greatly different from the other directions.
§Stock et al. (2004) showed: L. variegatus (order Temnopleuroida) carinar process plates are only slightly inclined from flange to the base of the keel; Stronglylocentrotus purpuratus and Heterocentrotuts trigonarius (order Echinoida, former family Strongylocentrotus and latter family Echinometrida) plates inclined more than L. variegatus and may incline differently from each other; P. lividus (order Echinoida, family Echinida) plate inclinations reverse as one approaches the base of the keel. Whether these differences reflect altered functionality would require numerical analysis. All taxonomy is taken from Smith (2005)
**For a given crystal system, b has a well-defined magnitude somewhat smaller than the lattice parameter in most cases. A dislocation with Burgers vector magnitude larger than b is termed a super dislocation.
††Osteoporosis and disuse-mediated bone loss (prolonged weightless periods in space) are examples of the latter.