Self-healing, an intrinsic property of biomineralization processes


  • Werner E. G. Müller,

    Corresponding author
    1. ERC Advanced Investigator Grant Research Group at Institute for Physiological Chemistry, University Medical Center of the Johannes Gutenberg University Mainz, Mainz, Germany
    • Institute for Physiological Chemistry, University Medical Center of the Johannes Gutenberg University Mainz, Duesbergweg 6, Mainz D-55128, Germany
    Search for more papers by this author
    • Tel.: +49-6131-39-25910. Fax: +49-6131-39-25243.

  • Xiaohong Wang,

    1. ERC Advanced Investigator Grant Research Group at Institute for Physiological Chemistry, University Medical Center of the Johannes Gutenberg University Mainz, Mainz, Germany
    2. National Research Center for Geoanalysis, Chinese Academy of Geological Sciences, Beijing, People′s Republic of China
    Search for more papers by this author
  • Klaus Peter Jochum,

    1. Max-Planck-Institut für Chemie, Mainz, Germany
    Search for more papers by this author
  • Heinz C. Schröder

    1. ERC Advanced Investigator Grant Research Group at Institute for Physiological Chemistry, University Medical Center of the Johannes Gutenberg University Mainz, Mainz, Germany
    Search for more papers by this author


The sponge siliceous spicules are formed enzymatically via silicatein, in contrast to other siliceous biominerals. Originally, silicatein had been described as a major structural protein of the spicules that has the property to allow a specific deposition of silica onto their surface. More recently, it had been unequivocally demonstrated that silicatein displays a genuine enzyme activity, initiating and maintaining silica biopolycondensation at low precursor concentrations (<2 mM). Even more, as silicatein becomes embedded into the biosilica polymer, formed by the enzyme, it retains its functionality to enable a controlled biosilica deposition. The protection of silicatein through the biosilica mantel is so strong that it conserves the functionality of the enzyme for thousands of years. The implication of this finding, the preservation of the enzyme function over such long time periods, is that the intrinsic property of silicatein to display its enzymatic activity remains in the biosilica deposits. This self-healing property of sponge biosilica can be utilized to engineer novel hybrid materials, with silicatein as a functional template, which are more resistant toward physical stress and fracture. Those hybrid materials can even be used for the fabrication of silica dielectrics coupled to optical nanowires. © 2013 IUBMB Life, 65(5):382–396, 2013.


Biomineralization is one essential element in living organisms to protect themselves against predators and to form multicellular aggregates, for example during bacterial biofilm formation, or in multicellular organisms, for example in animals and plants. The crucial processes that control biomineral formation proceed on the interphase between hard tissue and soft tissue. Focusing on metazoan animals, the key innovative process introduced during the evolution from the unicellular to the multicellular organisms was the development of a hard skeleton. The earliest metazoans, the sponges (phylum Porifera) and the corals (phylum Cnidaria), as well as the derived Deuterostomia, including the vertebrates, developed an inorganic hard skeleton, whereas the Protostomia, including the insects, form an organic hard skeleton. In all those skeletal elements, the hard biological materials, irrespectively if these are the chitin-based skeletons of insects or the bone-based systems, are formed onto an organic template. In most cases, the initiation of the hard skeleton formation starts intracellularly, and the growing nanoscaled primordial structures are completed extracellularly. This insight came with the improvement of the imaging as well as of the spectroscopic techniques and solved the opposing hypotheses. In the “matrix theory,” it had been assumed that the hard (crystallite) skeleton formation starts on the surface of a fibrillar macromolecule, whereas in the “compartment theory” it had been postulated that the formation of the initial skeletal element proceeds in vacuoles and does not occur in close association with structural elements (1). Most experimental data show that the hard skeleton formation starts at the interphase between an organic template and the soluble precursors, for example inorganic ions or monosaccharides (2, 3), either intracellularly and subsequently extracellularly, or only extracellularly. The structural and function guiding of the deposition of the soluble precursors is controlled by the functional groups on the organic templates.

Lowenstam (4) distinguished biologically induced mineralization and biologically controlled mineralization. During the process of biologically induced mineralization, deposition of minerals occurs on the surface of cells or multicellular plants and animals; known examples are the mineral formation on the surface of bacteria (5) which can result in the deposition of large mineralic natural deposit, for example deep-sea crusts or nodules (reviewed in ref.6). Biologically controlled mineralization refers to organic matrix-mediated mineralization, a process which often occurs in Metazoa during their skeleton formation (7). A closer examination shows that the complex structures of the biominerals formed by eukaryotes are hierarchically built. Investigations of the structural hierarchy must finally end in the understanding of the underlying genetic hierarchy; usually, such studies are restricted only to the analyses of the chemical and mechanical properties (e.g., strength/toughness) of these structures. It turned out to be difficult to analyze and to characterize the organic templates involved in a biomineralization process down to the level of gene expression. As examples, even intensive studies on bone formation in vertebrates (reviewed in: ref.8) or shell development in molluscs (9) cannot go beyond the level of biochemical characterization and cannot give an answer concerning the master gene(s) that drive the respective mineralization process. Until now, it was possible only to trace and to follow up a structural hierarchy down to the genetic level for the sponge siliceous spicules (10). Thanks to the high resolution of the microscopic and spectroscopic instruments available today, it is possible to resolve the nature of the organic templates within a biomineral. In most biological models studied, the organic templates remain embedded into the mineralic phase even after the completion of the mineral formation. This finding implies that the functionality of the templates remains unaffected because the organic macromolecules forming these templates are protected against denaturation and degradation after becoming encased within the mineral structures. As a consequence, the organic mineral-forming polymers become functionally active again, after uncovering from their mineralic envelope. As a consequence, the organic template has a self-healing function after the destruction of the inorganic shell.

Self-Healing Processes in the Inorganic Minerals

The most used structural materials, if concrete or steel is addressed, are vulnerable to cracking and are not repaired, if not special supplements are added to those materials. Only with the help of very advanced technologies, the cracks can be repaired by insertion of tight-bonding metals/alloys into the damaged material. It is well known that the restoration of breaks in concrete bricks by the reformation of the cracks under using new bricks (Fig. 1A) is by nonself-healing. In the last years, it has been proposed that additives containing a repairing agent can be incorporated into the concrete, allowing the self-healing of cracks and, in turn, restoration of the strength of the concrete structure (11). Subsequently, self-healing agents have been incorporated into microcapsules that trigger self-healing (12). These microcapsules consisting of polyurethane, containing a Na-silicate core, are ruptured after a mechanical stress, allowing the Na-silicate to be released. The silicate reacts with Ca-hydroxide of the concrete material and produces a calcium–silica–hydrate gel (x•[CaO•SiO2] •H2O) that forms the binding material to the concrete.

Figure 1.

Three different concepts for repairing damaged solid concrete-like material. Nonself-healing (A): a crack formed in a concrete or steel is repaired by filling-in into the open defect new repair parts (B). Extrinsic self-healing process (A). The self-healing material is added to defective concrete extrinsically, for example via bacteria (B). By this process, bacteria, covered by a Fe- or Mn-shell formed by those microorganisms, are penetrating the crack and fill-in the replacement material. Intrinsic self-healing process (A). This self-healing process is based on a functionalized organic template, for example silicatein (sil), which had been embedded into the (inorganic) hard material during its formation (B). After cracking, the internal organic template is exposed and displayed its functional activity, by forming a solid deposit that closes the fissure. The precursors for the repair, like orthosilicate (si) for silicatein, are delivered from the environment to the enzyme. It is important to note that the functionalized template that is repairing the damage is identical to the template that has initially formed the inorganic deposits. [Color figure can be viewed in the online issue, which is available at]

This is different from biomaterials which are formed and maintained in organisms, where the integrity of the materials can be restored to some extent. Concrete (calcium silicates [CaO•SiO2])-related materials, such as silica- or Ca-based minerals, can be repaired in sponges of molluscs (ref.13). As self-healing initiating and maintaining functional material, the organic template, resides inside of the hard (inorganic) material, this organic platform starts an intrinsic self-healing process (Fig. 1C). This review will discuss this process more extensively later, using the sponge silicatein as an example for an enzyme that is embedded into an inorganic (biosilica) matrix and regains its enzymatic function to form this matrix after cracking this inorganic deposit. Similarly, the iron particles in bacteria retain their integrity via a self-healing property (14). The major reason why the biominerals hold their property of self-healing/self-repair is the fact that these materials represent composites, formed of two components, and usually include the template onto which the mineralic phase has been initially formed.

By using nature as a model, biomimetic concepts have been developed that supplement abiological materials with organic components to provide them with self-healing functions (15), or to repair developed cracks by delivering/filling-in protecting organic materials via bacteria (16). The self-healing material is added to defective concrete extrinsically and via that route initiates an extrinsic self-healing process (Fig. 1B). The self-healing strategy via synthesis of nanocontainers, suitable for the encapsulation of self-healing agents, was introduced first for microcapsules (17); in this approach, the healing agent had been embedded into structural composite matrices. On the organismic level, the autonomic healing concept was conferred on bacteria, for example Bacillus subtilis, whereby the properties of the bacteria were utilized to precipitate Ca-carbonate or urea on their surface (16, 18).

The Sponge Spicules and their Intrinsic Self-Healing Properties

In contrast to the self-healing processes to be applied for minerals and metals, a process during which foreign components are added secondarily to the material (extrinsic self-healing process), the self-healing processes occurring in biological systems are an intrinsic property of their biominerals (intrinsic self-healing process). An established model to demonstrate such an intrinsic self-healing process is the siliceous sponge spicule.

The Siliceous Sponge Spicules

The two taxa of the sponges (phylum Porifera), the Demospongia and the Hexactinellida, that comprise a siliceous skeleton, which is composed of individual skeletal elements, the spicules, form their inorganic hard matrix of amorphous silica (SiO2•nH2O), a finding that has been reported first by Gray (19). All siliceous spicules comprise an axial canal in their center, into which the axial filament is embedded. The sizes of the intricately formed spicules vary in length, from around 1 μm (e.g., in the demosponge Geodia cydonium) to 3,000,000 μm (e.g., in the hexactinellid Monorhaphis chuni) and in diameter, from 0.5 to 10,000 μm. It had been determined that the hexactinellid M. chuni has a life span of 11,000 years (20) during which these sponges form their giant basal spicules, with a size of up to 10 mm × 3 m. These spicules represent the largest biosilica structures on Earth (6, 21). In most sponges, the growth rates of the spicules are very fast about 1–10 μm/h for spicules of the demosponge Ephydatia fluviatilis having a size of 200–350 μm in length and a thickness of 15 μm (22).

It is self-evident that those structures, the sponge spicules with their tiny or their giant dimensions, are prone to fractures. However, surprisingly, broken or semi-broken spicules are only rarely seen in nature. Such a rare instance is shown in Fig. 2, with the example of Euplectella aspergillum. E. aspergillum is a deep-sea hexactinellid sponge that lives on rocky hard bottoms (Fig. 2A). In about 2–4% of the animals, which are collected from the ground, malformations are seen (Figs. 2B and 2C). In the specimen shown in Fig. 2, the thin-walled tube, covered by a colander-like sieve plate, shows a re-healed lateral wall rupture (Fig. 2B). The tube is formed of individual spicules that secondarily fuse together to a continuous square mash. The breakage is restored by the formation of new spicules that form a new continuum together with the existing spicular network (Fig. 2C).

Figure 2.

Healing of the sponge tube after breakage; for example, E. aspergillum. (A) Collecting an E. aspergillum specimen from a depth of approximately 400 m, close to Zanzibar with a submarine. (B) Skeleton of a specimen of E. aspergillum disclosing a healed rupture (><) at the lateral side. The sponge skeletal tube is closed with a sieve plate (siv). (C) The damaged region is restored by new formation of spicules that secondarily fuse together to the original intact skeletal framework. [Color figure can be viewed in the online issue, which is available at]

The Dynamics and Spatial Growth Phases of Sponge Spicule Formation

The following four types of cells are involved in the spicule formation: sclerocytes, archaeocytes, chromocytes and lophocytes (23). The cells where the spicules start to be formed are the sclerocytes. These cells are the producers of silicatein, the enzyme that mediates the biocatalytic formation of biosilica, the inorganic matrix of the spicules (A more extensive description of silicatein, its structure and function, is given below). The archaeocytes also produce galectin, an essential component which forms the organic cylinder with silicatein during the appositional layering of silica (24). The chromocytes store carotenoids/retinoids derived from bacteria and are involved in the synthesis of silintaphin-2, a Ca2+-binding protein. Finally, the lophocytes contribute to spicule formation through the formation of collagen, a fibrillar protein that is required for the shaping of the spicules.

The formation of the spicules of siliceous sponges can be divided into two phases, an intracellular initiation phase leading to the formation of immature spicules and the extracellular “completion” phase of the mature spicules (25). The enzymatic/silicatein-mediated spicule formation starts intracellularly and is completed extracellularly. Using the sponge Suberites domuncula, we showed that the axial growth of the spicules proceeds in three phases: (i) formation of an axial canal, (ii) evagination of a cell process into the axial canal and (iii) assembly of the axial filament composed of silicatein. During these phases, the core part of the spicules, which surrounds the axial filament within the axial canal, is synthesized. Intracellularly, the axial filament is formed from nanofibrils through lateral assembly (26). Around these 10-nm thick nanofibrils and, after assembly, around the axial filaments the enzymatic synthesis of biosilica proceeds, a process during which thin-walled ≈7-μm long immature spicules are formed.

These growing spicules are extruded via a process of evagination from the intracellular compartment to the extracellular space (25, 27). Here, two silicatein-mediated biosilica deposition processes occur (Fig. 3); the first process is axially directed (Fig. 3A), whereas the second growth process occurs through appositional layering of new biosilica lamellae (Fig. 3B). The axial growth is guided by an elongation of the axial filament which pushes away the growing spicule from the cell. During this dynamic process, the diameter of the axial canal narrows down from about 2 to 0.7 μm. The axial filament, protruding with the cell protrusion, mediates the enzymatic biosilica synthesis from silicate monomers that are released from the cell, where these monomers are stored in silicasomes (28). Besides this biosilica deposition process that proceeds within the axial canal and is mediated by the axial filament, a second biosilica deposition process occurs on the surface of the growing spicule. This second process which has been termed concentric appositional layering of silica is similarly mediated enzymatically using silicatein (28). However, the silicatein molecules involved in this process are not produced by the cell that forms the initial immature spicule, but by neighboring cells that surround the axially growing spicules in the extracellular space (27). During this process of appositional biosilica layering, the spicule reaches their final thickness. These two processes of spicule growth are schematically shown in Fig. 3. First, during the axial growth of the immature spicule, the major dynamic force is originating from the dynamic elongation of the axial filament, which pushes away the spicule from the sclerocyte (Fig. 3A). In the second phase, the appositional layering of silica, the spicule reaches its final morphology and thickness; the driving force comes from the amount of silicatein that is released from the surrounding silicatein-producing cells (sclerocytes) (Fig. 3B).

Figure 3.

The two growth phases of a spicule, after the release of the immature spicule into the extracellular space via evagination of a cell protrusion. (A) Initially, the spicule is pushed away from the cell (sclerocyte) that initially has formed the spicule. This process is driven by an elongation of the cell protrusion and in turn an elongation of the axial filament (af) within the axial canal. In this phase, a narrowing of the axial canal takes place mediated by an enzymatic polycondensation via silicatein molecules that compose the axial filament; the spicule remains slim. (B) The appositional layering of the growing spicule occurs in radial direction and is driven by silicatein that is released from silicasomes (sis) of cells that surround the growing spicule. [Color figure can be viewed in the online issue, which is available at]

Sponge Silica—A Hybrid Material

The sponge spicules, as rigid biological materials, can be best characterized as structural hybrid composites, provided with unusual combinations of mechanical properties (29). If the spicules are exposed to hydrofluoric acid (HF) for a limited period, the organic scaffold becomes progressively visible (Figs. 4A and 4B). After a short exposure period of 1 h, the decomposition of the silica lamellae starts from the surface of individual lamellae, setting free wrinkled structures (30). After a longer exposure period of about 2 h, a progressive dissolution of the silica lamellae occurs, during which the proteinaceous scaffold within the siliceous matrix of each lamella is uncovered (Figs. 4A and 4B). These scaffolds display organized structural pattern/sheets. The curtain-like sheets comprise holes with a diameter of approximately 100 nm. The rims of the holes are reinforced with thicker spheres. If the dissolution process is performed under strict avoidance of shear forces, complete proteinaceous scaffolds of lamellae can be obtained from the spicules. This result, supported by spectroscopic data, is taken as a strong indication that a proteinaceous material exists within and between the individual lamellae that is building an organized web (30, 31).

Figure 4.

Micro- and nanostructure of the silica matrix of M. chuni. (A) scanning electron microscope (SEM) image of a giant basal spicule, after treatment at 600 °C for 20 min. The individual lamellae (la) separate from each other. In the center, the axial canal (ac) is located. (B) Dissolution of lamellae via limited HF exposure. The thickness of one lamella is shown (double pointed line; la). Proteinaceous bundles (>< pr) are exposed within one lamella. (C) High-resolution magnification (SEM) of the surface of a silica lamella, displaying individual silica nanoparticles (><). (D) Proteinaceous framework around the silica nanoparticles (SEM). The silica particles had been dissolved by HF, leaving the protein scaffold behind (-pr). (E–G) NanoSIMS analysis of lamellae in a cross-section through a M. chuni spicule. The dimension of one lamella is marked (double pointed line; la). (E) Image taken simultaneously with the NanoSIMS by secondary electrons; the lamella is marked (la) and has a thickness of 5 μm. (F) The 16O/28Si-ratio (mapping) shows the homogeneity of the “biological glass” within the lamellae. (G) Mapping of sulfur/silicon; the 32S/28Si ratios are computed. The ratios of concentrations are given as pseudo-color images. Different colors correspond to different ratios of intensities increasing from black to red. [Color figure can be viewed in the online issue, which is available at]

A high-resolution magnification of the surface of M. chuni lamella displays that it is composed of 5–7 nm large nanoparticles (Fig. 4C). If those samples are exposed to HF, resulting in dissolution of the silica matrix, the organic scaffold, surrounding the nanoparticles becomes visible (Fig. 4D). The dimension of the holes, formed by the organic material, varies between 5 and 7 nm, supporting the view that organic proteins are surrounding the silica nanoparticles.

Our high-resolution spectroscopic analysis of cross-sections through a M. chuni spicule, using the NanoSIMS technology, supports the view that the silica material composing the spicules is not a solid siliceous scaffold, but an organized structure of silica nanoparticles that are surrounded by organic material (32) (Figs. 4E–4G). NanoSIMS element distribution analysis for carbon, oxygen, sulfur and silicon revealed that the borders of the lamellae are especially highlighted in scans obtained from the 16O/28Si mapping (Fig. 4F); here, the relative oxygen level is highest. The 32S/28Si ratios were determined along the same lines (Fig. 4G). It is noticeable that a considerable basal level for the element S is found, rising toward the upper right-hand side of the scan. Sulfur is not homogeneously distributed but highly enriched at the borders of the lamellae, but also within each lamella, indicating that in the mature spicules a protein also exists within the silica material.

Hardness of the Spicules

The sponge hybrid silica material of the spicules has been characterized as an exceptional structural composite because of the unusual combination of mechanical properties, such as strength, stiffness and toughness (reviewed in ref.29).

A load–displacement curve from a nanoindentation experiment at 16.1 mN on the lamellar region of a giant basal spicule form M. chuni is shown in Fig. 5. In the first dwell period, the penetration depth increases to a maximum depth of 538 × 3 nm, which attributes to some creep effect of the material. The calculated and averaged Martens hardness from those measurements is 2.73 × 0.05 GPa and the reduced modulus has an average value of 38.572 × 0.475 GPa. Previous studies using spicules from the same species found the same hardness (31). The authors reported a hardness between 2.5 und 3.0 GPa.

Figure 5.

A typical load-penetration depth curve for a M. chuni sample with an indentation load 16 μN. The following load-penetration stages are marked: (a) loading stage, (b) dwell period at maximum load, (c) unloading part and (d) dwell period correcting for the thermal drift.

The measurement for the flexural properties of an individual giant basal spicule from M. chuni of a length between 30 and 35 mm was performed in the three-point bend assay system (32). An image of a nonbroken spicule, shown in a cross-section, is shown in Fig. 6A. Spicule samples were bent with a crosshead rate of 0.2 mm/min. The representative load–displacement curve recorded with untreated spicules showed the characteristic sequential breaking pattern (Fig. 6C; red curve). Owing to the lamellar organization of the spicules, the surface lamellae break first (Fig. 6B), a process which requires 230 N (Fig. 6C; point B). It is typical that the deflection B–C does not drop to zero. Following the viscoelastic view, the more central lamellae do not burst and cause resistance. In the following consecutive time interval, the subjacent lamellae break and the displacement curve reaches a value of 145 N (at D). Usually, a third flexural stress peak is reached at 128 N which has to be attributed to the inner layers of the lamellar region of the spicules (at F). Finally, the lamellae of the axial cylinder break, which display no distinct linear/peak fracture curve; the resistance of the lamellae is much lower in that region and simulates a flexible displacement (at H). Very much in contrast is the crosshead displacement record obtained with heat-treated spicules (600 °C) (Fig. 6C, black line). The initial resistance of the spicules is much higher in comparison with the nontreated controls. A force of 365 N is needed to break the spicule (at B′), when the heat-treated spicule undergoes an almost complete breakage. A deflection B′–C′ occurs, which is only slightly retarded at 30 N (C′–D′), and followed by a final resistance at 5-0 N (E′–F′). These data demonstrate that the heat-treated spicules have lost the characteristic properties of the natural spicules to comprise stiffness, flexibility and toughness.

Figure 6.

Bending properties of spicules from M. chuni. (A) A cross-section through a spicule that had not been cracked. (B) Force–displacement relationship of a giant basal spicule, as measured in three-point bending assay. The SEM analysis of a partially cracked spicule revealed three different crack zones, including several lamellae (la) each; they are marked (1–3) in addition to the central axial cylinder (cy). After the third crack, the sample was subjected to SEM analysis. (C) Force–displacement relationship of giant basal spicules, measured in three-point bending assay. Records from two representative experiments with nontreated (red curve; lower graph) and heat-treated (600 °C, 20 min; black curve) spicules are shown. In the nontreated sample, the first cracking event is recorded and is reflected by the linear increase A–B (termed “elastic response”). The first crack [1] (B) occurs and is followed by the deflection to C. Two additional breakage events [2,3] (2: C–D–E; and 3: E–F–G) are recorded before a random cracking of the different lamellae within the axial cylinder occurs. The final breakage (G–H) can be ascribed to the flexible fracture of the axial filament. In contrast, the heat-treated spicule shows, in the bend test, only one dominant cracking event (A–B') which is followed by minor fractures (C'–D'; E'–F').

The load–extension curve of the spicule shows a biphasic breakage/cracking pattern. The outer lamellar zone cracks in several distinct steps, showing high resistance in concert with comparably low elasticity, whereas the axial cylinder breaks with high elasticity and lower stiffness. In contrast, the heat-treated spicules which contain denatured protein behave differently and show a higher stiffness and a lower elasticity.

Silicatein, the Basis for the Self-Healing Property of the Sponge Spicules

Silicatein, the Major Protein Existing Within the Silica Material of the Spicules

The major organic constituent within the siliceous spicules from both demosponges and hexactinellid is silicatein. The first identification of this protein had been achieved in the demosponge Tethya aurantium by Shimizu etal. (33) and Cha etal. (34). After identification of the protein sequence, the corresponding gene has been obtained. It was found that this protein has sequence similarity to the hydrolytic enzyme, cathepsin L. Previously, the latter protein has been isolated from the sponge G. cydonium. Silicatein has been described to be a silica-binding protein able to function as a template for silica to be deposited. Later on, this silicatein has been isolated from other demosponges, for example S. domuncula (35), as well as hexactinellidan sponges, for example Crateromorpha meyeri (36) and M. chuni (37). This protein exists not only in the axial filament, in the center of the spicules where it is located within the axial canal, but also in the silica lamellae surrounding this central part (32, 38).

The silicateins comprise about 325 amino acids (aa) with a molecular weight of about 35 kDa (Fig. 7A). During maturation, the primary translation product (proenzyme) is processed by cleaving off a signal peptide (aa1–aa17; S. domuncula [demosponge] silicatein-α) and an adjacent propeptide (aa18–aa112), resulting in the mature enzyme that has a size of 24–25 kDa. Similar to cathepsins, the catalytic center of silicatein contains His and Asn. However, the Cys of the cathepsins' catalytic triad is exchanged by Ser in silicatein. In addition to about 10 putative protein kinase phosphorylation sites, silicateins display a cluster of serine residues that is found to be close to the central aa residue of the catalytic triad, but is otherwise missing in cathepsins. Subsequent phylogenetic analyses revealed that silicateins form a separate branch from cathepsins (39). Demosponge silicateins are subdivided into two/three isoforms, termed silicatein-α, -β and -γ (33). Aligning polypeptides, deduced from the different cDNAs, a grouping of the demosponge silicatein-α and silicatein-β in two different branches of the phylogenetic tree could be computed. The hexactinellidan silicateins fall within silicatein-α and -β (Fig. 7B).

Figure 7.

The grouping of the demosponge and hexactinellidan silicateins using a molecular tree construction. (A) Alignment of the two S. domuncula silicateins (silicatein-α [SILCAa_SUBDO, accession number AJ272013] and silicatein-β [SILCAb_SUBDO, AJ547635.1]) with papain from the papaya plant (PAPAIN_CARPA, AAB02650.1). Residues conserved (similar or related to their physicochemical properties) in all sequences are shown in white on black, and those in at least four sequences in white on gray. The characteristic sites in the sequences, that is the catalytic triad (CT⧫) amino acids, Ser (S) in the silicateins (and Cys (C) in cathepsin), as well as His (H), and Asn (N), as well as the segments of the signal peptide (▴signal▴), the propeptide segment (∼∼∼) as well as the mature enzyme (+++) are marked. Finally, the serine clusters (S) are highlighted. (B) A phylogenetic tree was constructed with the deduced polypeptides from T. aurantium (SILCAa_TETHYA, AAD23951), from L. baicalensis silicatein-a2 (SILCAa2_LUBAI, CAI91571) and the demosponge/lithistid silicatein from Discodermia japonica (SILCA_DISCO, CBY80151) as well as the two hexactinellid sequences from M. chuni (SILCA_MONCUN, CAZ04880) and C. meyeri (SILCA_CRAME, CAP49202.2). The cathepsin L from the hexactinellid Aphrocallistes vastus (CATL_APHRVAS, CAI91577) was used as outgroup. The rooted tree was computed by neighbor-joining and distance matrix determinations; scale bar indicates an evolutionary distance of 0.1 amino-acid substitutions per position in the sequence. [Color figure can be viewed in the online issue, which is available at]

Silicatein, a Genuine Enzyme

After having identified the major protein within the spicules as silicatein, our group succeeded to show that silicatein is a true enzyme with all characteristics known for other enzymes (40, 41). In general, enzymes are defined as (i) three-dimensionally structured proteins that have the same structure at the start and at the end of the reaction; (ii) to be substrate-specific, implying that each enzyme is optimized for a particular reaction transition state, (iii) to lower the free energy of thermodynamics to overcome the transition states, (iv) to be characterized by a defined Michaelis–Menten constant (Km), reflecting the binding affinity to the substrate, (v) to show a maximum velocity (vmax) and (vi) to be inhibitable (ref.42).

The silicateins can be classified into the large class of hydrolases (Enzyme Commission [EC] number 3.x.x.x), either as hydrolases acting on ether bonds (supposing that the substrate of silicatein is orthosilicic acid; EC number: 3.3.x.x) or as hydrolases acting on ester bonds (orthosilicate substrate; EC: 3.1.x.x). The substrate of silicatein, being a polycondensing enzyme, is considered to be the natural monomeric silica, orthosilicate (41), of the synthetic tetraethoxysilane (TEOS) (33, 34). The enzyme reaction can be inhibited by an excess molar concentration of hexafluorosilicic acid (H2SiF6) (40). In addition, the enzymatic reaction is inhibited by the Ser proteinase inhibitor E-64 at a concentration of 10 μM (40).

The key enzyme kinetic parameters of the enzyme-catalyzed polycondensation reaction have been determined (40, 41). For the enzyme assays, either native silicatein or recombinant protein has been used at adequate final concentration of around 10 μg/mL in the reaction mixture. In our studies, we used TEOS that had been prehydrolyzed to form orthosilicate. The temperature optimum was found to be in the range of 20–25 °C; the temperature coefficient (Q10) decreases by 2.5-fold above 25 °C and increases by 2.9-fold below 25 °C. The Michaelis constant (Km) was determined with 22.7 μM. This low concentration for the substrate concentration at which the initial reaction rate is half maximal is very indicative for the silicatein-mediated reaction. The present-day maximal concentration of silica in the marine ocean is within this concentration range (43) and hence allows the biosilica polycondensation, driven by silicatein, already around this value. The turnover value (molecules of converted substrate per enzyme molecule per second) for silicatein in the silica esterase assay is 5.2; in comparison, the rate for the human cathepsin L enzyme is with 20, slightly higher (44). The specificity of the silica-metabolizing reaction was determined by showing that a replacement of silicatein in the assays by bovine serum albumin resulted in no measurable silica synthesis (40). The pH value of the enzyme-driven biopolycondensation reactions was kept at 7.4/7.5 (40, 41) and not varied, considering the fact that the autocatalytic polycondensation reaction is strongly dependent on small changes of the pH (45). The data, listed here, conclusively show that the silicatein-mediated biosilica formation is enzyme driven.

The Reaction Mechanism of Silicatein

Silicatein catalyzes the polymerization/polycondensation process by using the monomeric units, orthosilicate, under formation of biosilica. The first reaction mechanism has been proposed by Cha et al. (34). These authors suggested a two-step mechanism for silicatein-mediated silica formation from silicon alkoxide (TEOS) substrates. Next, a second model (46) has been deduced from mutational studies with a “chimeric” mutant of human cathepsin L. Recently, we proposed a new model (4749) which allows to explain both the initial condensation/disilicic acid formation and the subsequent oligo-/polymerization process. Our model is based on the assumption that the initial step of the catalytic reaction involves a nucleophilic attack (SN2 type) of the (electronegative) oxygen atom of the hydroxy group of the active site Ser to the (electropositive) silicon atom of a silicic acid molecule, a process that occurs in the catalytic pocket of the silicatein molecule. This reaction is presumably facilitated by hydrogen bridge formation between the Ser OH and the His of the catalytic center, which increases the nucleophilicity of the Ser hydroxy group (Fig. 8A, step 1). Next, a proton transfer from the His nitrogen (Ser–His hydrogen bond) to one of the silicon OH ligands of the pentavalent intermediate (transition stage) occurs, resulting in the release of a water molecule and the formation of a covalent binding of the enzyme Ser to the silicic acid. Subsequently, the Ser-bound silicic acid molecule undergoes a nucleophilic attack on the silicon atom of a second orthosilicic acid molecule, entering the substrate pocket of the enzyme. The resulting loss of a water molecule generates a disilicic acid molecule, which remains bound to silicatein through the ester-like linkage. Now, rotation of the ester bond allows the interaction of a second OH ligand of the enzyme-bound silicic acid unit with the nitrogen imidazole of the His in the catalytic center, giving rise to further growth from the disilicic acid by addition of a third orthosilicic acid molecule with release of a water molecule. During this step, a final cyclization intermediate is formed (Fig. 8A, step 2). The products released from the enzyme by hydrolytic cleavage (cyclic trisilicic acid and higher oligomers) are much more reactive than the silicic acid monomers and dimers and promote the further condensation reaction.

Figure 8.

Proposed reaction mechanism of the silicatein-catalyzed polycondensation of orthosilicate. (A) Crucial steps in the silicatein-mediated biosilica reaction. Step 1: nucleophilic attack of the (negatively charged) Ser oxygen atom at the (positively charged) silicon atom of the orthosilicic acid substrate, and transfer of one proton (originating from the Ser/His hydrogen bridge) from the imidazole nitrogen of the His to an OH ligand of the silicic acid molecule. There is a release of one water molecule from a pentavalent intermediate, followed by a nucleophilic attack of the oxygen atom of one of the OH ligands of the covalently bound silicic acid molecule at the silicon atom of a second orthosilicic acid molecule. After rotation of the Si[BOND]O[BOND]C bond between the resulting disilicic acid and the Ser of the enzyme, nucleophilic attack of an oxygen atom of a second OH ligand of the first silicic acid molecule, to a third orthosilicic acid moiety, takes place. Step 2: cyclization of the resulting (enzyme-bound) trisilicic acid is initiated by nucleophilic attack of the (negatively charged) oxygen of an OH ligand of the third condensated orthosilicic acid (data not shown). Finally, a reactive trisiloxane ring is released after hydrolysis of the Si[BOND]O[BOND]C bond. (B) The silicatein polycondensation reaction is reversible. Focusing on the silicatein-mediated self-healing process in spicules, the anabolic (condensation) reaction results in a filling-in of biosilica into a cracked damage. Prior to that, silicatein is assumed to (partially) dissolve the surface of the remaining fracture hole (catabolic reaction) and by that allows an accelerated bonding of the existing biosilica at the fracture and the newly synthesized biosilica filler. [Color figure can be viewed in the online issue, which is available at]

By using synthetic substrates, for example bis(p-aminophenoxy)-dimethylsilane which contains two silicic ester-like and two silane bonds (40), we got strong evidence that silicatein mediates not only the anabolic formation of oligo/polymers linked with siloxane bonds, but also their catabolic hydrolysis. Based on this finding, silicatein to catalyze two reactions, first acting as silica polymerase and second acting as a silica esterase, we proposed that biosilica in sponges allows the formation of an inorganic biosilica polymer that has the advantageous property to act as a biogenic shell that provides a flexible coating for organic matrices/templates and even for organisms that can be bioencapsulated. In continuation, we established that bacteria, transfected with silicatein, form a biosilica shell around them and, by that, provide them with a shield against mechanical stress (47). Although surrounded by the biosilica coat, the proliferation capacity of the bacteria is not impaired. This finding is a strong indication that the biosilica coat formed by the bacteria can be actively broken up prior to cell division and subsequently can be resealed again after the termination of the division process.


Hardening of biosilica, synthesized by silicatein, is an essential step during spicule formation. During the process of silica biopolycondensation, 1 mol of reaction water is produced with each siloxane bond formed. To maintain the polymerization reaction, the reaction water has to be extruded from the primary biosilica product (50, 51). The biochemical mechanism by which the accumulation of the reaction water, at the site of biosilica formation, is prevented has been clarified using the sponge 3D cell-culture system (primmorphs) (52). Exposure of primmorphs to Mn-sulfate resulted in a change of the morphology of these cell aggregates; in addition, the cells started to synthesize abnormal silica deposits. This newly formed biosilica material was rough and porous and abnormally linked/glued the spicules together (53, 54). On the basis of these observations and additional spectroscopic analyses in primmorphs, it is reasonable to propose that in Mn-sulfate-treated primmorphs, the aquaporin water channels become poisoned. Additionally, it was shown that in Mn-sulfate-treated primmorphs post-transcriptional modification processes are impaired; more specifically, an increased phosphorylation of silicatein was found. The observed increase in phosphorylation of silicatein presumably results in aggregation/assembly of silicatein molecules to less compact axial filaments.

It should be noted that, during spicule formation, the diameter of the biosilica mantel decreases. Although the initial diameter of the developing spicule is ≈10 μm, it subsequently diminishes to approximately 2–3 μm (54). Such a shrinking process is also observed during nonenzymatic sol–gel polycondensation processes at higher temperatures, and can be quantified with theoretical calculations, as follows. The diameter of a silicate monomer is 4.48 Å, whereas the dimension of one silica unit embedded in an almost completely polycondensated silica mesh is only 3.56 Å (55). As an example, and assuming that 30 mol of silicate monomers was arranged in a space-filling manner, the same number of silica molecules organized in a completely polycondensated mesh would occupy a 30% smaller space. The mechanism underlying the removal of water from biosilica via the process of syneresis might also help in understanding the formation of the various species-specific morphologies of the spicules formed in different sponge taxa. In both classes of siliceous sponge, spicule types exist that are richly ornamented and armed with hooks or thorns, or are bent in their longitudinal axes (56), a process that is very likely caused by a localized removal of water from the newly formed biosilica matrix of the growing spicules.

Phase Separation

Microscopic phase separation has been recognized to be essential for the template-directed silica precipitation in diatoms (57), and surely is likewise important for the understanding of the process of hardening of biosilica in sponges as well. Recently, we reported for the first time that the synthetic polyether polyethylene glycol (PEG) causes a rapid transition of the lucid biosilica to the opaque/turbid state (49). Subsequently, a sponge-specific protein, a 22–24 kDa polypeptide, termed spicule-binding protein, was identified which similarly caused an increased sol–gel transition of silicatein-synthesized biosilica. The gene corresponding to this protein was identified in S. domuncula and found to encode for a protein containing the nidogen domain. The recombinant protein, the nidogen-related protein, was prepared and found to induce gelation of biosilica owing to phase separation. Based on these data, it is proposed that the PEG-induced phase separation process follows a mechanism during which PEG together with silicatein neutralizes the negative surface charges of the formed biosilica nanoparticles. Further on, it is adopted that the phase-separation process, caused by the nidogen-like protein (spicule-binding protein), can be ascribed best to a polymerization-induced phase-separation process (49).

With the transition of the biosilica material to gel-like flocs, biosilica undergoes syneresis and a distinct phase separation. The material becomes compact and gains a higher viscosity. This process of induced gel formation is crucially important for the application of biosilica in biomimetic approaches.

Silicatein-Mediated Self-Healing Process in Sponge Spicules

As mentioned above, silicatein, the causative protein mediating enzymatic polycondensation reaction, remains embedded in the axial canal of the mature spicules where it forms the axial filament (33), as well as in the mature silica mantel (30, 38, 58).

Self-healing Property of Silicatein Embedded in Spicules

The central canal of the spicules contains a proteinaceous axial filament (Fig. 9A) that is laterally assembled by about 10 nanofibrils (26). By completion of the formation of the mature spicules, the axial filament remains embedded in this central structure. After treatment of the spicules with HF, the proteinaceous filaments can be recovered (26) and stained with Coomassie brilliant blue. With progressing time, the silica is dissolved, leaving behind only the organic axial filament. It is surprising that the proteinaceous axial filament remains preserved in the spicules, even in spicules of an age of about 2 Ma, as demonstrated for the Baikalian sponge Lubomirskia baicalensis (59). The silicatein in these axial filaments remains functionally active as silica-binding protein (34), or as enzyme (41) after dissolution of the biosilica shell. This functional activity of the axial filament can be demonstrated by incubating them with prehydrolyzed TEOS (orthosilicate). If the incubation is performed in the presence of 6 μg/mL of axial filament protein in a 50 mM Tris/HCl buffer (pH 7.4; 150 mM NaCl), containing 200 μM orthosilicate bulky depositions of biosilica onto the axial filament structures can be visualized (Figs. 9C–9E). Those structures cover almost completely the axial filaments after an incubation period of 180 min (Fig. 9E). Axial filaments do not exist only as rods, like those isolated from the tylostyles from S. domuncula (Figs. 9A–9E) but also appear as radial tufts, as in asters form G. cydonium (Figs. 9F and 9G) (60). Also, these actinoid-like arranged filaments retain the property to deposit large amounts of biosilica. It should be stressed that the concentration used here for demonstration of the functional activity is 200 μM orthosilicate, and in turn, is far below of the autopolycondensation concentration for silica which is around 2 mM (61). Consequently, the biosilica formation can be attributed to the enzymatic activity of silicatein (41).

Figure 9.

Deposition of biosilica aggregates (bs) onto silicatein, assembled to axial filament (af) bundles, isolated from the demosponges S. domuncula (A–E) and G. cydonium (F–G) (SEM). (A) Broken spicule with a protruding/projecting axial filament. (B) A large number of axial filaments, isolated form spicules through dissolution with HF. (C–E) Time-dependent accumulation of biosilica deposits onto axial filaments that had been incubated with 200 μM orthosilicate (prehydrolyzed TEOS). Samples of axial filaments were removed from the assays after 5 min (C), after 60 min (D) and after 180 min (E). The newly formed biosilica deposits are marked (bs). (F and G) Axial filaments, isolated from sphaerasters of G. cydonium. The axial filaments were subjected to an incubation assay containing 200 μM orthosilicate and subsequently incubated for 60 min; some biosilica bricks (bs) are marked.

The experimental data are schematically summarized in Fig. 10. In intact spicules, the axial filament remains embedded and—by that—is protected from potentially adverse physical and chemical conditions through the silica mantel (Fig. 10A). After indenting and partial cracking the spicule, the silicatein molecules become exposed to the environment (Fig. 10B) and the silica mantel crumbles (Fig. 10C). In case if orthosilicate is present in the aqueous environment at sufficiently high concentrations, the silicatein can associate with those monomers and can form biosilica. The concentrations of silica in the present-day oceans, with about 5–20 μM (43), might be sufficient to realize an extensive biosilica deposition as the Km value for silicatein was determined to be 22.7 μM (40). However, sponges have, like diatoms (62), a cell membrane-associated silica pump that allows the intracellular accumulation silica within special organelles, the silicasomes (63). In turn, even spicules directly exposed to the aqueous environment have the capacity to self-heal (partially) broken spicules. If existing as skeletal elements in the body, the high concentrations of silica in the spicule-synthesizing cells, the sclerocytes, are sufficient to provide silicatein with the appropriate level of silica for the enzymatic synthesis of biosilica.

Figure 10.

Schematic outline of the silicatein-driven self-healing process in sponge spicules. (A) In an intact spicule, the silicatein, assembled to the axial filament (af), is embedded into the biosilica (bs) material. (B) Setting an indent into the spicule and rupturing of the biosilica mantel. (C) Oozing of silicatein molecules from the axial filament to the fracture zone. (D) Resealing of the fracture by enzymatic biosilica deposition within the hole. [Color figure can be viewed in the online issue, which is available at]

Repair, self-healing, processes taken place within spicules can be frequently recognized within hexactinellid spicules in their lamellar regions, for example within the giant basal spicules of M. chuni (Fig. 11A). Fractures that have been introduced into the spicules, followed by a stay in a dry environment for a short time, remain nonsealed (Fig. 11B). However, fractures within the lamellae that occur in spicules during the growth phase fuse almost completely together (Fig. 11C).

Figure 11.

Cracking of lamellae within giant basal spicules form M. chuni. (A and B) Cross-section through a spicule, displaying fracture tracks (><), caused during a three-point bending assay. (C) Self-healing of cracked lamellae within a spicule (><) growing in a living animal. [Color figure can be viewed in the online issue, which is available at]


The property of self-healing is an intrinsic property of any biomineralization process. The prerequisite that this property sustains is that the organic template with its selective functionalities, which initially functions as matrix onto which the inorganic minerals, form a structured mineral layer, remains conserved and intact. A prominent example for such a sustainable conservation of the functionalities of the organic template is silicatein. This enzyme remains embedded into the inorganic matrix for years, surely for thousands of years, for example in M. chuni (20), or even millions, for example in the Baikalian sponge L. baicalensis (59). The awareness of this basically expected, but not yet fully appreciated, phenomenon should have distinct implications for future biomimetic and bioprocessing developments to fabricate, or more precisely, to bioengineer novel materials with self-healing properties. In the next future, realization can be expected in the field of synthesis of hybrid nanowires. Recently, it could be demonstrated that biosilica layers exhibit electrically insulating properties onto nanowires, allowing a chance to construct in situ selective patterning of dielectrics, under physiological conditions (64). This potential can be combined with the optical properties of biosilica spicules (reviewed in ref.65). The tools are now elaborated to fabricate in a controlled way optical components under physiological processing conditions (66) that are surrounded by a doped dielectric. Finally, the layer-by-layer technology, using a functionalized silicatein template for the controlled fabrication of a mineralic phase 1, and again a silicatein template for a further mineralic phase 2, may provide useful elements for the further advancement in the production of electrical insulating and conducting elements for microelectronics, again under mild conditions (67).


W.E.G.M. is a holder of an ERC Advanced Investigator Grant (No. 268476 “BIOSILICA”). This study was supported by grants from the German Bundesministerium für Bildung und Forschung (project “Center of Excellence BIOTECmarin”), the Deutsche Forschungsgemeinschaft (SCHR 277/10-2), the European Commission (Seventh Framework Programme, Marie-Curie Initial Training Network “BIOMINTEC”, Grant No. 215507; and Industry-Academia Partnerships and Pathways “CoreShell”; Grant No. 286059), the International Human Frontier Science Program, the Public Welfare Project of Ministry of Land and Resources of the People′s Republic of China (Grant No. 201011005-06) and the International S & T Cooperation Program of China (Grant No. 2008DFA00980).