Flexible Minerals: Self-Assembled Calcite Spicules with Extreme Bending Strength

See allHide authors and affiliations

Science  15 Mar 2013:
Vol. 339, Issue 6125, pp. 1298-1302
DOI: 10.1126/science.1216260


Glass or metal fibers can show incredible flexibility. Natalio et al. (p. 1298; see the Perspective by Sethmann) used the protein silicatein-α, which is responsible for the biomineralization of silicates in sponges, to guide the formation of spicules made of calcite. These synthetic spicules could be bent to a high degree because of their inherent elasticity, whilst retaining the ability to guide light.


Silicatein-α is responsible for the biomineralization of silicates in sponges. We used silicatein-α to guide the self-assembly of calcite "spicules" similar to the spicules of the calcareous sponge Sycon sp. The self-assembled spicules, 10 to 300 micrometers (μm) in length and 5 to 10 μm in diameter, are composed of aligned calcite nanocrystals. The spicules are initially amorphous but transform into calcite within months, exhibiting unusual growth along [100]. They scatter x-rays like twinned calcite crystals. Whereas natural spicules evidence brittle failure, the synthetic spicules show an elastic response, which greatly enhances bending strength. This remarkable feature is linked to a high protein content. With nano-thermogravimetric analysis, we measured the organic content of a single spicule to be 10 to 16%. In addition, the spicules exhibit waveguiding properties even when they are bent.

The organisms of various phyla have developed complex and intriguing strategies to deposit minerals in their structural frameworks. Most of them are directed by organic matrices, which control size, shape, organization, and even the mineral phase (13). A simple biological design strategy is to use single crystals. Amorphous precursors (47) are used to mold biominerals into their final forms (68) and to prevent unintentional calcification by rapid precipitation. (5) Organic macromolecules have been shown to play a pivotal role. They stabilize the transient phases (9, 10), influence the shape, and overcome the intrinsic brittleness of the crystalline phase (11). Macromolecules create defects in the lattice, which strengthen the crystal against fracture by absorbing stress and stopping the propagation of cracks (12). Occluded macromolecules are—at least in part—responsible for the shape of biominerals by inhibiting their growth in certain directions. In addition, the observed anisotropic distribution of defects indicates that the macromolecules are arranged in certain crystal directions (3, 12). An early structural study of calcitic sponge spicules revealed a morphological and textural symmetry different from the hexagonal symmetry of calcite, indicating an additional level of biological control (13).

The intricate structures observed in biominerals have inspired the development of synthetic strategies to mimic nature. The precipitation of minerals in the presence of soluble organic additives has proven to be highly successful in the regulation of crystal morphology and nanostructures (14). Although nature produces sophisticated structures with remarkable ease and fidelity, the synthesis of their artificial counterparts is still a challenge. The fabrication of materials that resemble spicules—calcareous or siliceous—is even more difficult because it involves dissimilar organic and inorganic nanophases. The principle of controlled nucleation occurs in both siliceous (noncrystalline) and calcareous (crystalline) biomineralization independently of their chemical nature.

One important protein responsible for the formation of siliceous spicules in sponges is silicatein-α. This ~23-kD protein, which has been shown to form oligomers and fibrous structures through self-aggregation (15), catalyzes and structurally directs the formation of silica spicules. (16) Here, we demonstrate that silicatein-α can also be used to form synthetic spicules of calcium carbonate. The mechanical properties of the synthetic spicules proved to be superior to those of biogenic material. For comparison, we used the spicules (monoaxa) of the calcareous sponge Sycon sp. (Porifera, Calcarea) (12, 13), which are morphologically very similar to our synthetic calcite spicules.

A solution of freshly prepared CaCl2 (5 mM) was mixed on a cleaned mica surface with refolded recombinant silicatein-α (210 μg/mL, pH 7.4). This was then exposed to CO2 partial pressure above solid ammonium carbonate [(NH4)2CO3] for 4 hours at room temperature (RT) in a sealed desiccator (17). Needle-like crystals ("synthetic spicules") with smooth surfaces, diameters of 5 to 10 μm, and lengths of 10 to 300 μm were formed (Fig. 1, A and B). A scanning electron microscopy (SEM) image (Fig. 1B, inset) of a cross section of a synthetic spicule prepared by use of a focused ion beam (FIB) revealed a uniform, homogeneous, and circular structure. In the absence of silicatein-α, only rhombohedral calcite crystals were formed (fig. S1A).

Fig. 1

(A) Light microscopy overview of synthetic spicules. (B) SEM image of a representative synthetic spicule. (Inset) SEM image of a cross section through a synthetic spicule prepared with a FIB. (C) FT-IR mapping for amide I band at 1650 cm−1 and ν3 of carbonate ions in calcite at 1456 cm−1 on a synthetic spicule. High intensity is indicated in red. (D) Confocal microscopy image of immunostained synthetic spicules after cross-reaction with polyclonal antibodies (PoAb-aSILIC) raised against natural silicatein-α. (E) SEM of a synthetic spicule after NaOCl treatment. (F) Gel electrophoresis (SDS–polyacrylamide gel electrophoresis, 12%) analysis after NaOCl treatment (1% v/v, 10 min). A band with a molecular weight of ~23 kD after Coomassie staining shows the presence of silicatein (lane a). After the second and third washing steps (lanes b and c), no protein was observed. M, molecular markers for calibration.

The presence of silicatein-α is evident from the Fourier transform infrared (FT-IR) spectrum (fig. S4B, blue line), in which the most substantial bands can be assigned to amide I, II, and III at 1650, 1580, and 1370 cm−1, respectively (18). Because of the complexity of the FT-IR spectra, a specific three-dimensional conformation cannot be assigned to silicatein-α (such as α helix, β sheets, or random coils). The distribution of the amide I band (1650 cm−1) and the ν3 band of carbonate ions (1456 cm−1) of a spicule over the mapped area are shown in Fig. 1C, top and bottom, respectively. The highest intensities (red) are observed both on the synthetic spicule, indicating the presence of silicatein-α and its participation in the control of the mineralization process.

To localize silicatein-α more precisely, we removed surface-bound silicatein-α by treating the synthetic spicules with NaOCl (1% v/v, 10 min). After washing with distilled water, the supernatant was kept, and the synthetic spicules were analyzed by means of SEM and immunochemistry. Fluorescence microscopy showed a positive cross-reaction between the synthetic spicules and polyclonal antibodies raised against axial filaments (PoAb-aSILIC) (Fig. 1D) (19). NaOCl treatment caused a minor decrease in the size of the synthetic spicule by rounding edges (Fig. 1E). No fluorescence was observed in the controls, in which the antibody was replaced by preimmune serum. Subsequently, the synthetic spicules were dissolved in acetic acid solution (5%, 10 min). The solution was analyzed by means of gel electrophoresis after neutralization. The presence of a band at ~23 kDa corresponding to silicatein-α is shown in Fig. 1F, lane a, confirming that the protein was in fact occluded in the synthetic spicules. Electrophoretic analysis of the supernatant from the NaOCl treatment did not show any band, which demonstrates the absence of surface-bound silicatein-α (Fig. 1F, lane b and c).

To estimate the organic content of synthetic spicules, we used a nano-thermogravimetric analysis (nano-TGA) procedure on single, freshly prepared spicules (fig. S5). A single spicule was placed onto the end of an atomic force microscope (AFM) cantilever. Its mass was determined by measuring the resulting decrease in the resonance frequency of the cantilever. The density of the spicules, calculated from their mass and their dimension as determined with optical microscopy, was found to be 2.5 and 0.6 g/cm3 for the natural and synthetic spicules, respectively. The density of a natural spicule is close to that of pure calcium carbonate (2.7 g/cm3). Synthetic spiculae are highly porous on the microscopic scale. The spicules were then calcinated at 500°C for 10 min to remove all organic content, which was confirmed by means of attenuated total reflectance (ATR) μFT-IR analysis (fig. S6A). The mass loss amounted to 1.5% for natural spicules and 10 to 16% for synthetic spicules. Macroscopically, the synthetic spiculae shrunk in size. Their density increased to 1.7 g/cm3 (fig. S6B).

The CaCO3 mineralization process in the presence of silicatein-α was monitored by taking transmission electron microscopy (TEM) "snapshots" at given time intervals. After 5 min, CaCO3 nanocrystals with diameters of ~5 nm were formed [Fig. 2A and fig. S7, high-resolution (HR) TEM image]. After 30 min, fractal-like structures were formed (Fig. 2B) proceeding through a continuous oriented assembly process (Fig. 2C) until elongated, compacted, amorphous, and granular microsized structures had grown (Fig. 1A).

Fig. 2

TEM "snapshots" of synthetic spicule formation. (A) Overview of nanocrystals after 5 min, (B) 30 min, and (C) 1 hour. (D) Reconstructed x-ray diffraction pattern of the (hk0)-plane of a mature synthetic spicule (10 months). (Inset) Reconstructed diffraction pattern of the (0kl)-plane (bottom left), rocking curve of the (104) reflection displaying a split peak with peak widths (FWHM) of ≈0.6 to 0.7°, respectively (top right), and zoom of the intensity distribution of some selected reflections (bottom right). Data from which the rocking curve was derived are reconstructed from the full volume of the Ewald sphere. Therefore, they show the reflections of all domains. As a consequence, we see only a few broad maxima rather than multiple sharp maxima. (E) HRTEM image of a cross section of a mature spicule showing small calcite crystals (~5 nm) embedded into an amorphous protein matrix. (F) Cross-polarized light microscopy images of the aged synthetic spicule (10 months) at different angles (0° to 90°).

The amorphous nature of the immature spicules and the subsequent aging of the synthetic spicules were demonstrated by means of a series of x-ray diffraction experiments at different aging stages. Synthetic spicules that had matured for more than 6 months diffracted, whereas spicules younger than 5 months were amorphous, as shown with optical polarization microscopy (fig. S8A). Despite their smooth surface, mature synthetic spicules scattered like a calcite twin crystal. According to reconstructed layers from reciprocal space ("precession plots," hk0, hk1, 0kl, and 1kl), virtually all reflections are split into separated maxima as typically observed for non-merohedral twins with slightly skewed lattices (Fig. 2D). The individual maxima of the split reflections are relatively broad [full width at half maximum (FWHM) of 0.6 to 0.8°] pointing to small coherent scattering domains of 5 to 7 nm. Such small crystalline domains were also observed with TEM (Fig. 2E and fig. S7). This indicates that the synthetic spicules consist of crystallographically aligned nanodomains (20). The growth direction of the synthetic spicule is [100], which corresponds to no naturally observed morphology of calcite, thus suggesting the presence of mesocrystallinity (20). Because no Mg2+ was incorporated into the crystal (supplementary materials), one may attribute the reflection splitting in mature synthetic spicules to lattice strain from the occluded protein.

To further analyze the structure, we prepared ion-milled cross sections of natural spicules (fig. S9) as well as of immature and mature synthetic spicules (Fig. 2E). In Sycon sp. monoaxons, we observed filamentous structures and voids (fig. S9A). Electron diffraction (ED) patterns of different areas showed a mixture of amorphous and crystalline domains (fig. S9D). This, together with HRTEM images (fig. S9E), proves the mesocrystalline nature of the calcitic structures. Thin cuts of immature synthetic spicules showed no clear boundaries between organic and inorganic material, thus confirming the virtually amorphous character (fig. S10A). HRTEM analysis of mature synthetic spicules revealed the presence of small crystalline domains surrounded by an amorphous layer that we attribute to protein (Fig. 2E and fig. S10B), reflecting the mesocrystalline ultrastructure of mature synthetic spicules.

Although the outward appearance of a synthetic spicule kept under ambient conditions for 10 months remained virtually unchanged (fig. S11, SEM and AFM), polarized light microscopy (Fig. 2F and fig. S8A) and ATR μFT-IR spectra (fig. S8B) revealed a transition from a noncrystalline to a crystalline phase. The crystalline phase grows at the expense of transient spherical amorphous calcium carbonate (ACC) particles according to Ostwald's step rule via a dehydration step combined with a dissolution-recrystallization process (fig. S12), as described before for sea urchin larvae (21).

The mechanical properties of single spicules have been studied before. These studies were, however, limited to spicules large enough to allow macroscale testing. The Young's modulus of siliceous sponge spicules was found to be close to 40 GPa in most cases (2224), but values of 14 (25) and 68 GPa (26) have also been reported. Little is known about the mechanical properties of calcareous spicules from marine species; a value of 36 GPa has been reported for spicules from echinoderm larvae (27).

We probed the mechanical properties of natural and synthetic spicules with an AFM. Individual spicules were attached to the edge of a silicon wafer by using a small drop of two-component epoxy glue. The spicule acted as a flexible beam of length L (Fig. 3A). A tipless AFM cantilever was used to apply a force at a distance x from the wafer edge (Fig. 3A, inset). We recorded force-versus-beam deflection curves for a series of positions along the spicules. According to beam theory, the applied force F and the induced beam deflection δ at the loading position x of a one-sided clamped beam with a circular cross section are related byF=3πER4δ/(4x3). Here, E is the Young's modulus of the beam, and R is its radius. We observed a linear relation between applied force and induced beam deflection (Fig. 3A). The effective stiffness ks = F/δ increases with decreasing distance x. From linear fits of many such force-versus-deflection curves, we obtained plots of the effective stiffness as a function of the normalized position x/L (Fig. 3, B and C). The Young's modulus of the spicules was calculated by fitting withks=3πER44(xx0)3(1)Fit parameters were E and x0, where x0 is an offset to correct for uncertainties in the determination of the distance from the clamping point. Using this method, we obtained E = (14.2 ± 0.8) GPa for a synthetic spicule 1 month after synthesis (Fig. 3B). The Young's modulus of a natural spicule was (8.8 ± 0.9) GPa (Fig. 3C). The Young's modulus of our synthetic spicule was lower than the one found in natural echinoderm spicules (36 GPa) (27) but higher than for spicules from Sycon sp. monoaxons. The Young's modulus of the synthetic spicules increased with time, which is a strong indication of aging. We measured E = (3.0 ± 0.5) GPa for a fresh spicule (less than a week old) and E = (19 ± 4) GPa for a 7-month-old sample (fig. S13).

Fig. 3

(A) The flexural response of a fixed synthetic spicule of length L to a well-defined load applied at different positions x along its main axis (inset) was probed by using a tipless AFM cantilever. From the linear relationship between applied force and deformation, the stiffness ks of the clamped spicule was obtained. (B and C) Measured values of ks versus x/L and the fit (red dashed) for (B) a synthetic spicule sample and for (C) a biogenic Sycon sp. monoaxon. (D) The fracture properties of spicules were probed with a micromanipulator and recorded in situ with SEM for natural (i to vi) and synthetic spicules (vii to xii). The synthetic spicule did not fracture even under extreme loading and deformation conditions (xi) that lead to plastic deformation.

To probe fracture properties and ultimate strength, spicules were bent in an SEM by using a micromanipulator. Natural spicules fracture in a brittle way when deflected by δ/L ≥ 11% (Fig. 3D, top inset, and movie S1). Assuming a linear stress-strain curve up to fracture and using the measured E modulus for the natural spicules, we calculated a maximum stress of σ = 89 to 125 MPa at the clamping point before fracture. SEM analysis of the natural spicule shows, instead of a crystal facture, a clear glassy conchoidal fracture (fig. S14A). Similar behavior has been observed for glass sponge fibers and echinoderm spicules and attributed to a combination of mesocrystallinity and organic content (11).

Synthetic spicules (1 and 7 months) showed no signs of fracture or plastic deformation, at least up to δ/L = 18 and 27% (Fig. 3D, bottom inset, 1 month, and fig. S15, 7 months). This corresponds to a maximum applied stress of σ = 170 and 442 MPa, respectively. Even when bending a synthetic spicule to a U shape, we were not able to fracture it. Instead of brittle failure, we observed elastic bending with a small plastic component; after 180° bending and release, the spicule showed a residual plastic deformation of δ/L ~ 10% (Fig. 3D, bottom inset, and movie S2) and cannot be broken even under extreme deformations (fig. S14, C and D). We attribute the remarkable bending strength of synthetic spicules as compared with that of natural spicules to the almost 10-times-higher content of organic material (10 to 16% in contrast to 1.5% for natural spicules). Furthermore, AFM imaging revealed a partial fusion of the initial granular structural units. This "sintering" effect can be attributed to a structural reorganization (fig. S16, A to I).

Using a micromanipulator (such as the one described above), we performed bending experiments on a calcinated (at 500°C) synthetic spicule that was partially glued with epoxy onto a silicon surface. (fig. S16J, i to vi for sequence, and movie S3). The calcinated synthetic spicule could still be bent to some extent before fracturing near its clamping point (fig. S16J, i to vi for sequence, and M, and movie S3). The fact that the synthetic spicule is composed of nanoscale calcite building blocks that are (to some extent) displaceable to each other because of interspersed protein contributes to the remarkable mechanical properties of the synthetic spicules before calcination (fig. S17).

A high flexibility in calcite biomaterials has been reported before—for example, in sea urchin spines (11) or the ossicles of the sea star Echinaster spinulosus (28). The flexibility of the sea urchin spine is due in part to the entrapment of proteins in the crystal. Similarly, echinoderms possess a rigid endoskeleton composed of calcite and small amounts of occluded matrix proteins. In the synthetic spicules, the mineral phase is composed of ACC, which then transforms to an anhydrous ACC and eventually to calcite. (29)

The uniform circular cross section of mature synthetic spicules with a diameter of a few micrometers is an optimal range for efficient waveguiding of visible light. To demonstrate the ability to guide light, a continuous laser (λ = 632 nm, incident spot size 44 μm at an angle of θ = 16°) was focused onto the fixed end of an epoxy-clamped mature synthetic spicule (Fig. 4A and fig. S18). The free end of the spicule was bent with an AFM cantilever (kc = 42 N/m, tip length of 17 μm) mounted on a three-axis micromanipulator (Fig. 4B). The bending process was recorded with an optical microscope placed above the sample. In a typical experiment, we pushed the synthetic spicule with the front side of the AFM tip (Fig. 4C, ii), allowed it to return to its initial position (Fig. 4C, iii), and then pulled it with the back side of the AFM tip (Fig. 4C, iv). Even after deformations of δ/L = 24 to 33%, the synthetic spicule was still light guiding, without considerable loss of intensity at its free end. We attribute the light-guiding properties of synthetic mature spicules to a gradient in the refractive index due to the protein content, as apparent from our data.

Fig. 4

(A and B) Schematic of a spicule (A) fixed at one end and (B) bent by an AFM cantilever to test its waveguiding properties. (C) Microscopy images of the waveguiding capacity of the synthetic spicule during deformation by (i and ii) pushing and (iii and iv) pulling it away from its original position (i).

We have described the formation of needle-like calcite crystals that resemble naturally occurring Sycon sp. monoaxons by using the self-assembly properties of silicatein-α. Synthetic spicules are composed initially of amorphous building blocks that undergo an aging process. At the mature stage, these spicules consist of aligned calcite nanocrystals with silicatein-α occluded in the domain boundaries. The Young's modulus of 14 GPa for synthetic spicules is at least as high as that for natural spicules. Synthetic spicules sustain a fracture stress at least three times higher than that of natural spicules, without any sign of brittle fracture. The presence of 10 to 16% organic components accounts for these remarkable mechanical properties.

Supplementary Materials

Materials and Methods

Figs. S1 to S18


Movies S1 to S3

References and Notes

  1. Acknowledgments: This work was partially supported by the Deutsche Forschungsgemeinschaft within the SPP 1420. T.P.C. was supported by a Deutscher Akademischer Austauschdienst scholarship. We acknowledge the use of the facilities of the EM Center in Mainz (EZMZ) supported by the Center for Complex Matter and the SFB 625. We are grateful to R. Jung-Pothmann for performing single-crystal x-ray measurements and J. Ally for insightful discussions. The authors declare that they have no competing financial interests.

Stay Connected to Science

Navigate This Article