Biomimetic remineralization of dentin

Abstract

Objectives

Remineralization of demineralized dentin is important for improving dentin bonding stability and controlling primary and secondary caries. Nevertheless, conventional dentin remineralization strategy is not suitable for remineralizing completely demineralized dentin within hybrid layers created by etch-and-rinse and moderately aggressive self-etch adhesive systems, or the superficial part of a caries-affected dentin lesion left behind after minimally invasive caries removal. Biomimetic remineralization represents a different approach to this problem by attempting to backfill the demineralized dentin collagen with liquid-like amorphous calcium phosphate nanoprecursor particles that are stabilized by biomimetic analogs of noncollagenous proteins.

Methods

This paper reviewed the changing concepts in calcium phosphate mineralization of fibrillar collagen, including the recently discovered, non-classical particle-based crystallization concept, formation of polymer-induced liquid-precursors (PILP), experimental collagen models for mineralization, and the need for using phosphate-containing biomimetic analogs for biomimetic mineralization of collagen. Published work on the remineralization of resin–dentin bonds and artificial caries-like lesions by various research groups was then reviewed. Finally, the problems and progress associated with the translation of a scientifically sound concept into a clinically applicable approach are discussed.

Results and significance

The particle-based biomimetic remineralization strategy based on the PILP process demonstrates great potential in remineralizing faulty hybrid layers or caries-like dentin. Based on this concept, research in the development of more clinically feasible dentin remineralization strategy, such as incorporating poly(anionic) acid-stabilized amorphous calcium phosphate nanoprecursor-containing mesoporous silica nanofillers in dentin adhesives, may provide a promising strategy for increasing of the durability of resin–dentin bonding and remineralizing caries-affected dentin.

Introduction

Teeth are the most heavily mineralized tissues in the human body. Demineralization and remineralization processes coexist in teeth during the entire life of an individual. In pathological conditions, demineralization outweighs remineralization . Fermentation of dietary carbohydrates by acidogenic bacteria results in the production of acids such as lactic acid, acetatic acid and propionic acid that demineralize enamel and dentin. As the carious lesion progresses into dentin, activation of endogenous, bound matrix metalloproteinases and cysteine cathepsins will lead to the degradation of collagen fibrils and decrease in the mechanical properties of dentin . Prevention and treatment of dental caries is a major challenge because as many as nine out of ten adults in Western countries suffer from dental caries . In the United States alone, more than 100 million dollars is spent annually on dental service. Despite significant advances in preventive and restorative dentistry, replacement of tooth fillings constitutes a substantial share of this annual dental expenditure, due to limited durability of contemporary resin-based restorative materials, particularly when these materials are applied to damaged dentin in the absence of a superficial enamel layer .

Apart from caries, resin–dentin bonding is another major reason for dentin demineralization . The formation of resin–dentin bonds is accomplished predominantly by micromechanical retention via resin penetration and entanglement of exposed collagen fibrils in the partially or completely demineralized dentin. This is achieved by etching dentin with acids or acidic resin monomers derived from self-etching primers/adhesives to expose the collagen fibrils . To date, it is impossible for resin monomers to completely displace water within the extrafibrillar and particularly the intrafibrillar compartments of a demineralized collagen matrix, and infiltrate the collagen network completely . Even if this may be achieved, the limited intermolecular space (1.26–1.33 nm) between collagen molecules renders it challenging to accommodate even small, extended resin monomer molecules such as triethyleneglycol dimethacrylate (≈2 nm long) . This invariably results in the presence of mineral-depleted, resin-sparse, water-rich collagen fibrils along the bonded interface . Under the combined challenges of enzymes, temperature and functional stresses, regions of incomplete resin infiltration within the dentin hybrid layer is susceptible to degradation, resulting in damage of interfacial integrity, reduction in bond strength and ultimately, the failure of resin–dentin bonds. Thus, remineralization of demineralized dentin has important consequences for control of dentinal caries as well as improvement of dentin bonding stability .

Different strategies have been employed for remineralizing demineralized dentin. For instance, fluoride, amorphous calcium phosphate (ACP)-releasing resins or resin-based adhesives containing bioactive glass have been used to improve the resistance of bonded restorations to secondary caries However, most of these studies focused on remineralizing partially demineralized carious dentin, which was based on the epitaxial deposition of calcium and phosphate ions over existing apatite seed crystallites . With these traditional ion-based strategies, remineralization does not occur in locations where seed crystallites are absent . Thus, the classical ion-based crystallization concept may not be applicable for remineralizing completely demineralized dentin within hybrid layers created by etch-and-rinse adhesive systems or the superficial part of a caries-affected dentin lesion left behind after minimally invasive caries removal, due to the unavailability of seed crystallites in those regions for accomplishing homogeneous nucleation of apatite crystallites .

Biomimetic remineralization represents a different approach to this problem by attempting to backfill the demineralized dentin collagen with liquid-like ACP nanoprecursor particles that are stabilized by biomimetic analogs of noncollagenous proteins . This is achieved by adopting the recently discovered, non-classical particle-based crystallization concept utilized by Nature in various biomineralization schemes, ranging from the mineralization of sea-shells (calcium carbonate), siliceous shells of diatoms and sponges (amorphous silica) to the deposition of calcium phosphate salts in fish scales and bone . Intrafibrillar mineralization of fibrillar collagen not only significantly increases its mechanical properties , but also protects the collagen molecules from external challenges, such as temperature, endogenous enzymes, bacterial acids and other chemical factors. Using this biomimetic remineralization strategy, both hybrid layers created by etch-and-rinse adhesives and moderately aggressive self-etch adhesives , as well as 250–300 μm thick completely demineralized dentin lesions can be remineralized . This bottom-up remineralization strategy does not rely on seed crystallites, and may be considered as a potentially useful mechanism in extending the longevity of resin–dentin bonds via restoring the dynamic mechanical properties of the denuded collagen within the hybrid layer to approximate those of mineralized dentin . This paper reviews the changing concepts in calcium phosphate remineralization and the progress in clinical translation of the biomimetic dentin remineralization strategy.

Changing concepts of calcium phosphate biomineralization

Biomineralization is the process by which living organisms secrete inorganic minerals in the form of biominerals (e.g. magnetite, silica, oxalates, various crystalline forms of calcium carbonate and carbonated apatite) within cell cytoplasm, shells, teeth and bony skeletons . This process exhibits a high level of spatial and hierarchical control as mineralization usually takes place in a confined reaction environment under ambient temperature and pressure conditions. Calcified human tissues consist of the collagen matrix and the hierarchically arranged carbonated apatite inorganic phase; deposition of the latter is regulated by non-collagenous proteins . It is generally believed that non-collagenous proteins, along with specific MMPs and other important enzymes secreted by odontoblasts, play critical roles to orchestrate dentin mineralization. They possess carboxylic acid and phosphate functional groups that act as preferential sites for Ca/P nucleation and subsequent apatite crystallization . As the therapeutic use of native or recombinant non-collagenous proteins for in situ biomineralization is not yet economically viable, research scientists have resorted to the use of polyelectrolyte and poly(acid) macromolecules to mimic the functional domains of these naturally occurring proteins, in biomimetic mineralization . In the past few years, this field of research has attracted a lot of attention, resulting in changing concepts of calcium phosphate biomineralization.

Particle-based vs ion-based crystallization

Traditional collagen mineralization studies were based on the classical pathway of ion-mediated crystallization, or classical nucleation theory . The classical model of crystal formation begins with crystal nucleation, followed by crystal growth. This process starts from primary building blocks like atoms, ions or molecules, forming clusters, which may grow or disintegrate again, depending on the counter-play of surface and crystal lattice energies. Eventually, some clusters reach the size of a so-called critical crystal nucleus. These primary nuclei grow further via ion-by-ion attachment and unit cell replication. While the classical crystallization model was relatively successful in controlling the dimensions of calcium carbonate or calcium phosphate, they achieved limited success in reproducing the structural hierarchy of intrafibrillar apatite deposition within the collagen matrix.

In contrast to the ion-mediated classical crystallization pathway, the non-classical crystallization pathway is particle-mediated and involves a mesoscopic transformation process . The term mesoscopic refers to materials of an intermediate length scale, the lower limit may be the size of individual atoms. Whereas macroscopic objects usually obey the laws of classical mechanics, a large number of particles can interact in a quantum-mechanically correlated fashion for mesoscopic objects. Evidence for these pathways is rapidly increasing in the literature. To date, mesocrystals of a wide range of materials including CaCO 3 , BaSO 4 , metal oxides, metal tungstates and chromates, NH 4 TiOF 3 , (NH 4 ) 3 PW 12 O 40 , LiFePO 4 , metal chalcogenides (e.g. semiconductor cadmium sulphide), noble metals, and organic materials have been successfully synthesized, as described in several excellent reviews . The contemporary concept of calcium phosphate biomineralization has been advanced by parallel studies on the biomineralization of calcium carbonate (calcite and aragonite) from amorphous calcium carbonate . In the context of calcium phosphate biomineralization, calcium and phosphate ions are sequestered by biomimetic analogs of non-collagenous proteins involved in hard tissue mineralization into nanoparticles that exist in nanoscopical units known as prenucleation clusters . These prenucleation clusters (≈1 nm in diameter) eventually aggregate into larger (10–50 nm in diameter) liquid-like ACP nanoparticles. The latter, on penetration into the intrafibrillar water compartments of a collagen fibril, utilize the latter as a mineralization template, and undergo self-assembly and crystallographic alignment to form a metastable crystalline phase via mesoscale assembly. These mesocrystals probably fused to form iso-oriented crystal intermediates, and finally to single apatite crystallites within the 40 nm wide gap zone between the collagen molecules . This ordered arrangement of apatite crystallites results in the manifestation of a banded appearance in unstained, mineralized fibrillar collagen .

Polymer-induced liquid precursor

Formation of amorphous nanoprecursors is the fundamental step in many forms of biomineralization. Existence of transient ACP nanoprecursors has been identified from enamel and bone . In biomimetic mineralization of type I collagen, Dr. Gower and colleagues pioneered a process based on formation of a polymer-induced liquid-precursor (PILP) system . The plasticity of liquid amorphous mineral precursors allows them to take the shape of their containers, resulting in a variety of biominerals with different hierarchical structures . Using the PILP concept, Gower and colleagues were successful in mineralizing a variety of organic matrices with both calcium carbonate and calcium phosphate, including intrafibrillar mineralization of collagen matrices by carbonated apatite .

Based on the aforementioned new concepts of biomineralization introduced over the past decade, biomimetic mineralization of type I collagen may be envisaged to proceed in several stages : (1) the collagen fibril serving as an active template for hierarchical intrafibrillar mineralization; (2) calcium and phosphate ions in the calcifying medium self-assemble into stable particulate units known as pre-nucleation clusters. In the presence of a polyanionic analog of matrix proteins such as poly(aspartic acid), these pre-nucleation clusters further condense into larger particles of fluidic amorphous ACP precursors (in the range of 10–30 nm) that are capable of diffusing into the intrafibrillar compartments of type I collagen; (3) the negatively charged polyanion-stabilized ACP precursors interact with positively charged sites along the collagen molecules, inducing solidification and nucleation of the ACP inside collagen; (4) these nucleated amorphous mineral precursor phases further growth and maturation into apatite nanocrystals along the intrafibrillar space of collagen via non-classical crystallization pathways; (5) Growth of the intrafibrillar apatite results in heavy intrafibrillar mineralization by apatite crystallites, with concomitant extrafibrillar mineralization between adjacent collagen fibrils.

Models of collagen mineralization

In our early studies, totally or partially demineralized dentin collagen matrices were employed for remineralization of hybrid layers and caries-like dentin lesions . However, it is not known if the remineralization demonstrated in those studies was caused by the presence of remnant phosphoproteins that remain bound to collagen matrices after demineralization . As self-assembled purified collagen fibrils do not contain bound matrix proteins, the authors employed a two-dimensional model of biomimetic mineralization of collagen based on reconstituted collagen fibrils deposited on grids used for transmission electron microscopy . Briefly, bovine skin-derived type I collagen dissolved in acetic acid was reconstituted on formvar-coated grids by ammonia diffusion and cross-linked with 0.3 M carbodiimide. Then, the grid with a single layer of collagen was floated over a drop of a prospective biomimetic analog-containing biomineralization medium. The grid was retrieved at different time points (usually within 72 h) and rinsed with water. They were examined unstained for mineral deposition or stained for correlation of mineral deposition with collagen cross-striation patterns. With this model, we demonstrated a highly hierarchical assembly of intrafibrillar apatite crystallites in reconstituted collagen mineralized ex situ with biomimetic analogs of extracellular matrix proteins ( Fig. 1 ). This 2-D single-layer collagen mineralization model is useful as a rapid screening tool for potential biomimetic analogs for collagen mineralization . Moreover, the technique may be modified for creating intrafibrillarly mineralized collagen coatings on the surfaces of orthopedic and dental implants as a more bio-compatible alternative to the use of hydroxyapatite coatings .

Fig. 1
A two-dimensional model for examining mineralization of reconstituted collagen fibrils, using a dual biomimetic analog mineralization protocol. Transmission electron microscopy (TEM) images of unsectioned reconstituted collagen fibrils. (A) After 24 h of mineralization. The fibril was briefly stained with uranyl acetate to highlight the gap and overlap zones. Amorphous calcium phosphate (ACP) prenucleation clusters (open arrowhead) could be seen along the periphery of the fibril. Pointer: initial needle-shaped intrafibrillar apatite. (B) Unstained collagen fibril after 48 h of mineralization. Prenucleation clusters have coalesced into ACP droplets (arrow) that continued to infiltrate the fibril. Denser intrafibrillar apatite deposition could be seen but the banding periodicity caused by hierarchical arrangement of the apatite crystallites was still present. (C) Schematic depicting replacement of intrafibrillar water with fluidic polyacrylic acid-stabilized ACP nanoprecursors. (D) Unstained fibril showing heavy intrafibrillar mineralization with apatite platelets after 72 h. Banding characteristics was obscured by the heavy mineralization.

Although the single-layer collagen model is a convenient and rapid method for studying biomimetic mineralization of individual collagen fibrils, the simple 2-D structure of the collagen model limits it further application. To further mimic natural collagen matrices, 3-D models of reconstituted collagen ( Fig. 2 ) were used by different research groups to evaluate the ability of the biomimetic mineralization scheme to mineralize collagen bundle assemblies within a three-dimensional scaffold . Briefly, commercially available type I collagen sponges derived from bovine tendon were cultured in analog-containing biomineralization medium for 3–14 days. The mineralization assembly was kept in a 37 °C oven to emulate physiological conditions. At predetermined reaction times, mineralized samples were removed from the solution, copiously washed with deionized water and further prepared for different analysis, including transmission electron microscopy, thermogravimetric analysis, attenuated total reflection-Fourier transform infrared spectroscopy, nuclear magnetic resonance spectroscopy and X-ray diffraction, which cannot be easily achieved with the 2-D collagen model . The formed collagen–hydroxyapatite composites indicated that polyaspartic acid-stabilized ACP fluid droplets can diffuse into collagen bundles and mineralize a relatively thick collagen scaffold .

Fig. 2
A three-dimensional reconstituted collagen model of biomimetic mineralization. TEM images of an unstained section taken from a highly porous collagen scaffold that had been mineralized with a calcium and phosphate-containing medium containing dual biomimetic analogs for 14 days. (A) Low magnification of the mineralized collagen scaffold, showing interconnecting mineralized collagen bundles. (B) High magnification taken from a part of a mineralized collagen bundle. Collagen fibrils on the left are less heavily mineralized, with hierarchical deposition of apatite platelets in the gap zones, producing a periodic banding pattern. Collagen fibrils on the right are heavily mineralized with apatite platelets. Banding characteristic can no longer be identified.

Although the mechanical properties of 3-D collagen sponges increase significantly after functional mineralization, they are still far less than those of the natural mineralized tissues because of their highly porous nature. Thus, attempts have been made by various groups to produce biomaterials that structurally mimic bone and dentin . However, the hierarchical complexity of natural composites is usually far beyond contemporary bioengineering processing capability. One approach is to utilize the hierarchical structure of natural materials as a template for the development of high-performance engineering materials using in vitro biomimetic methods . Thus, researchers revisited the use of natural collagen model, including rat tail tendon (a soft tissue collagen 3D model; Fig. 3 ), demineralized manatee bone and demineralized dentin collagen . It was found that the amount of mineral present in the rat tail tendon samples represented the maximum amount of mineral available in the PILP solution . Conversely, for demineralized manatee bone specimens, mineral distribution and penetration across the bulk of the bone matrix appears to be a challenge for the PILP process. The mineral penetration depth was limited to 100 μm . The mineral content for the remineralized manatee bone specimens (about 45 wt%) was below the theoretical value for cortical bone, even though enough calcium and phosphate ions were available in the PILP solution to restore the mineral value of bone. It was proposed that auto-transformation and solidification of the surface-infiltrated ACP nanoprecursors into crystalline apatite with time could potentially block further infiltration of the additional nanoprecursors into the bulk of the demineralized bone matrix, thereby limiting their ultimate depth of penetration.

Fig. 3
A three-dimensional natural soft tissue collagen model of biomimetic mineralization. TEM images of unstained sections taken from mineralized rat tail tendon with parallel collagen fibrils at different time periods. (A) Incomplete mineralization at 7 days. Asterisk: Unmineralized collagen fibrils; Arrow: extrafibrillar mineralization; Open arrowhead: Intrafibrillar mineralization recapitulating the D-spacings of fibrillar collagen. (B) Heavy mineralization of the parallel collagen fibrils after 14 days. (C) High magnification of B, showing a collagen fibril with extrafibrillar (arrow) and intrafibrillar mineralization by discrete apatite platelets (open arrowhead). (D) Selected area electron diffraction of the mineralized fibril in (C) produced arc-shaped diffraction patterns in the 0 0 2 and 0 0 4 plane of apatite. The C -axis of the apatite platelets is arranged almost parallel to the longitudinal axis of the collagen fibril.

The need for using phosphate-containing biomimetic analogs in biomimetic mineralization of collagen

The functional role played by collagen in calcium phosphate biomineralization has been conjectural based on the results of the past several decades of research. Earlier studies indicated that fibrillar collagen does not initiate biomineralization on its own, but serves as a passive depot for the housing of apatite crystallites . This led to a plethora of studies that examined the biological control of intrafibrillar collagen mineralization by noncollagenous proteins . Noncollagenous proteins are believed to play a crucial role in the mineralization of bone and dentin. This assertion is supported by studies demonstrating that mutations in genes that code for these proteins result in abnormal bone and dentin mineralization . The extracellular matrix of bone and dentin contains small quantities of noncollagenous proteins such as osteopontin, bone sialoprotein, dentin matrix protein 1 (DMP1), and dentin sialophosphoprotein (DSPP) . The latter is further cleaved into dentin sialoprotein and dentin phosphoprotein (DPP; aka phosphophoryn) . These proteins are highly anionic due to the prevalence of carboxylate groups on the polyaspartic acid residues that comprise the protein backbone. Post-translation phosphorylation of the serine residues produces phosphoserines, which further augments their anionic character . For example, dentin phosphoprotein, being the most abundant NCP in dentin, contains a large number of aspartic acid (Asp) and phosphoserines (Pse) in the repeating sequences of (Asp-Pse) n and (Asp-Pse-Pse) n . Dentin matrix protein 1, another member of the Small Integrin-Binding Ligand Interacting Glycoproteins (SIBLING) family, also contains high levels of Asp and Pse . A unifying feature of the SIBLING proteins is that they all contain an acidic serine aspartate-rich MEPE (ASARM)-associated motif. The ASARM motif in SIBLING genes and the released ASARM peptide play roles in mineralization of bone and teeth . The highly anionic nature of noncollagenous proteins involved in biomineralization enables them to sequester and bind calcium ions and presenting them to collagen fibrils at the mineralization front during the formation of bone and dentin.

In the authors’ laboratory, biomimetic mineralization of collagen was performed using a dual biomimetic analog strategy . Polyacrylic acid or polyaspartic acid was used as an analog for sequestering calcium ions released by set calcium silicate cements or a supersaturated calcium phosphate mineralization solution. These analogs function by acting as surfactants to prevent fluidic ACP nanoparticles from aggregating into larger particles, and to inhibit auto-transformation of the ACP nanoparticles into apatite (i.e. apatite nucleation inhibitor) prior to their entry into the intrafibrillar water compartments of the collagen fibril . In addition, a polyphosphate-containing biomimetic analog such as polyvinylphosphonic acid, sodium trimetaphosphate or sodium ascorbyl phosphate was employed as a templating biomimetic analog of matrix phosphoproteins. These phosphoprotein analogs were allowed to bind to the collagen fibrils prior to immersion of the fibrillar matrices in the poly(anionic) acid-containing mineralization medium . Using the aforementioned collagen mineralization models, attachment of ACP nanoprecursors to the D-spacings of non-mineralized collagen fibrils was observed within 24 h. After 48 h, hierarchical apatite deposition was observed from unstained collagen fibrils that resulted in banded mineral arrangement. After 72 h, highly mineralized collagen fibrils could be seen in which the periodicity of the mineral arrangement was obscured as the fibrils were completely mineralized .

In contrast to the earlier work, more recent studies indicate that type I collagen plays an active role in biomineralization by acting as templates for physicochemical (electrostatic) attraction of ACP nanoparticles and direct formation of intrafibrillar apatite within the gap zones, without the intervention or mediation of other noncollagenous proteins present in vertebrate calcifying tissues . Silver and Landis reported specific sites in the e2 band of the gap zones (corresponding to the a, e and d bands) of type I collagen that have the potential to sequester and bind calcium ions . Nudelman et al. calculated that there are net positive charges close to the C-terminal end of the collagen molecules that favor infiltration of the collagen fibrils with negatively charged ACP nanoprecursors. The arrangement of charged amino acids within the gap and overlap zones produce nucleation sites that control the conversion of the ACP nanoprecursors into oriented apatite crystals . Indeed, several research groups have demonstrated that it is possible to produce intrafibrillar mineralization of type I collagen using poly(aspartic acid)-stabilized ACPs alone as an apatite nucleation inhibitor, without the adjunctive use of phosphorylated DDP, phosphorylated DMP1 or their polyphosphate analogs . Although this simplified biomimetic collagen mineralization strategy is more economical and reduces time in preparing mineralized collagen scaffolds from a tissue engineering perspective, it begs the provision of a rationale for the existence of highly phosphorylated noncollagenous proteins in bone and dentin. The use of the simplified biomimetic mineralization strategy also challenges earlier biological studies that reported critical roles played by the phosphorylated forms of these noncollagenous proteins in the formation of mineralized collagenous tissues .

A recent study compared the effects of phosphorylated vs non-phosphorylated forms of DPP and DMP1 on collagen mineralization using a 2D model . Although differences existed between DPP and DMP1 in the locations in which collagen fibrils were mineralized (intrafibrillar vs extrafibrillar), both phosphorylated proteins facilitated highly organized intrafibrillar mineralization of collagen fibrils. Conversely, the use of non-phosphorylated forms of these proteins resulted in randomly oriented intrafibrillar crystallites, with no particular organization of their crystallographic axes to the longitudinal axis of the collagen fibrils. Nevertheless, this study failed to explain why non-phosphorylated biomimetic apatite nucleation inhibitors such as polyaspartic acid or fetuin-A are capable of producing highly organized intrafibrillarly mineralized collagen. It must be emphasized that while the concepts of particle-based ACP prenucleation clusters and the PILP phenomenon of fluidic ACP infiltration are new in collagen biomineralization research, the theoretical basis for interpretation of ultrastructural results of intrafibrillar apatite deposition is not , and is based upon the classic model of collagen molecular packing proposed by Petruska and Hodge . This straight and rod-like model of steric arrangement of collagen molecules has since been replaced by a synchrotron X-ray diffraction-derived model in which collagen molecules are arranged in a right-handed helically twisted, discontinuous manner along the length of the microfibrils . In the latter model, interdigitation of adjacent microfibrils placed geometric constraints on the availability of lateral intermolecular spacings between collagen molecules, with no room to accommodate apatite platelets outside the gap zones.

A recent study examined the effect of using single (polyacrylic acid) vs dual biomimetic analogs for mineralization of collagen fibrils (polyacrylic acid and sodium trimetaphosphate) . In that study, polyacrylic acid was employed as an analog to inhibit apatite nucleation, and sodium trimetaphosphate was used as a templating analog for guiding intrafibrillar apatite deposition. The use of polyacrylic acid without a templating analog resulted only in intrafibrillar mineralization with continuous apatite strands. Conversely, the use of both analogs resulted in intrafibrillar mineralization with discrete apatite crystallites. While both methods resulted in intrafibrillar mineralization of collagen fibrils, the authors opined that in the absence of polyphosphate as a templating analog, infiltration of poly(anionic) acid-stabilized ACP nanoprecursors via a PILP process into the interconnecting water-filled volume within a collagen fibril appeared to have resulted in molding of ACP nanoprecursors into a continuum. This, in turn, resulted in crystallization of the carbonated apatite into a monolithic crystalline structure. Such a crystallization mechanism produces mineralized collagen entities that resemble the monolithic single-crystal structure in sea urchin spines or siliceous bio-skeletons. It is possible that release of poly(anionic) acid into the intrafibrillar milieu results in osmotic swelling and relaxation of the helical collagen microfibrillar arrangement, that facilitates continuous apatite deposition or growth from the gap zones into overlap zones into continuous strands. Conversely, electrostatic binding of polyphosphate analogs to discrete sites along the collagen molecules may have caused the bound analogs to act as inhibitor to discourage continuous growth of the apatite crystallites along the overlap zones, thereby resulting in constraining apatite platelets to the gap zones of the collagen fibril. This hypothesis is speculative and awaits further validation.

Nevertheless, the aforementioned hypothesis is consistent with the timely report comparing the ultrastructure of mineralized collagen in bone with the use of ion-milled sections vs sections prepared by conventional ultramicrotomy . In ion-milled sections, the authors reported that approximately 70% of the minerals in bone are extrafibrillar, confirming previous models proposed by Lees et al. and Hellmich and Ulm . With respect to intrafibrillar apatite, crystallite platelets were only identified in the gap zones, without extension into the overlap zones, in contrast to what was previously proposed by Landis et al. . Similar results were reported in another study using steric modeling to estimate the packing density of apatite within the gap zones . Those modeling results were further confirmed using electron energy loss spectroscopy associated with scanning transmission electron microscopy. Taken together, these novel findings highlight that mechanisms in the control of discrete intrafibrillar apatite platelet deposition in the gap zones by phosphorylated NCPs are not completely understood. It is possible that the current success of intrafibrillar collagen mineralization with a single poly(anionic) acid analog may not result in truly biomimetically mineralized collagen that resembles those present in bone and dentin. These challenging issues require further in-depth investigations to decipher the riddles.

Changing concepts of calcium phosphate biomineralization

Biomineralization is the process by which living organisms secrete inorganic minerals in the form of biominerals (e.g. magnetite, silica, oxalates, various crystalline forms of calcium carbonate and carbonated apatite) within cell cytoplasm, shells, teeth and bony skeletons . This process exhibits a high level of spatial and hierarchical control as mineralization usually takes place in a confined reaction environment under ambient temperature and pressure conditions. Calcified human tissues consist of the collagen matrix and the hierarchically arranged carbonated apatite inorganic phase; deposition of the latter is regulated by non-collagenous proteins . It is generally believed that non-collagenous proteins, along with specific MMPs and other important enzymes secreted by odontoblasts, play critical roles to orchestrate dentin mineralization. They possess carboxylic acid and phosphate functional groups that act as preferential sites for Ca/P nucleation and subsequent apatite crystallization . As the therapeutic use of native or recombinant non-collagenous proteins for in situ biomineralization is not yet economically viable, research scientists have resorted to the use of polyelectrolyte and poly(acid) macromolecules to mimic the functional domains of these naturally occurring proteins, in biomimetic mineralization . In the past few years, this field of research has attracted a lot of attention, resulting in changing concepts of calcium phosphate biomineralization.

Particle-based vs ion-based crystallization

Traditional collagen mineralization studies were based on the classical pathway of ion-mediated crystallization, or classical nucleation theory . The classical model of crystal formation begins with crystal nucleation, followed by crystal growth. This process starts from primary building blocks like atoms, ions or molecules, forming clusters, which may grow or disintegrate again, depending on the counter-play of surface and crystal lattice energies. Eventually, some clusters reach the size of a so-called critical crystal nucleus. These primary nuclei grow further via ion-by-ion attachment and unit cell replication. While the classical crystallization model was relatively successful in controlling the dimensions of calcium carbonate or calcium phosphate, they achieved limited success in reproducing the structural hierarchy of intrafibrillar apatite deposition within the collagen matrix.

In contrast to the ion-mediated classical crystallization pathway, the non-classical crystallization pathway is particle-mediated and involves a mesoscopic transformation process . The term mesoscopic refers to materials of an intermediate length scale, the lower limit may be the size of individual atoms. Whereas macroscopic objects usually obey the laws of classical mechanics, a large number of particles can interact in a quantum-mechanically correlated fashion for mesoscopic objects. Evidence for these pathways is rapidly increasing in the literature. To date, mesocrystals of a wide range of materials including CaCO 3 , BaSO 4 , metal oxides, metal tungstates and chromates, NH 4 TiOF 3 , (NH 4 ) 3 PW 12 O 40 , LiFePO 4 , metal chalcogenides (e.g. semiconductor cadmium sulphide), noble metals, and organic materials have been successfully synthesized, as described in several excellent reviews . The contemporary concept of calcium phosphate biomineralization has been advanced by parallel studies on the biomineralization of calcium carbonate (calcite and aragonite) from amorphous calcium carbonate . In the context of calcium phosphate biomineralization, calcium and phosphate ions are sequestered by biomimetic analogs of non-collagenous proteins involved in hard tissue mineralization into nanoparticles that exist in nanoscopical units known as prenucleation clusters . These prenucleation clusters (≈1 nm in diameter) eventually aggregate into larger (10–50 nm in diameter) liquid-like ACP nanoparticles. The latter, on penetration into the intrafibrillar water compartments of a collagen fibril, utilize the latter as a mineralization template, and undergo self-assembly and crystallographic alignment to form a metastable crystalline phase via mesoscale assembly. These mesocrystals probably fused to form iso-oriented crystal intermediates, and finally to single apatite crystallites within the 40 nm wide gap zone between the collagen molecules . This ordered arrangement of apatite crystallites results in the manifestation of a banded appearance in unstained, mineralized fibrillar collagen .

Polymer-induced liquid precursor

Formation of amorphous nanoprecursors is the fundamental step in many forms of biomineralization. Existence of transient ACP nanoprecursors has been identified from enamel and bone . In biomimetic mineralization of type I collagen, Dr. Gower and colleagues pioneered a process based on formation of a polymer-induced liquid-precursor (PILP) system . The plasticity of liquid amorphous mineral precursors allows them to take the shape of their containers, resulting in a variety of biominerals with different hierarchical structures . Using the PILP concept, Gower and colleagues were successful in mineralizing a variety of organic matrices with both calcium carbonate and calcium phosphate, including intrafibrillar mineralization of collagen matrices by carbonated apatite .

Based on the aforementioned new concepts of biomineralization introduced over the past decade, biomimetic mineralization of type I collagen may be envisaged to proceed in several stages : (1) the collagen fibril serving as an active template for hierarchical intrafibrillar mineralization; (2) calcium and phosphate ions in the calcifying medium self-assemble into stable particulate units known as pre-nucleation clusters. In the presence of a polyanionic analog of matrix proteins such as poly(aspartic acid), these pre-nucleation clusters further condense into larger particles of fluidic amorphous ACP precursors (in the range of 10–30 nm) that are capable of diffusing into the intrafibrillar compartments of type I collagen; (3) the negatively charged polyanion-stabilized ACP precursors interact with positively charged sites along the collagen molecules, inducing solidification and nucleation of the ACP inside collagen; (4) these nucleated amorphous mineral precursor phases further growth and maturation into apatite nanocrystals along the intrafibrillar space of collagen via non-classical crystallization pathways; (5) Growth of the intrafibrillar apatite results in heavy intrafibrillar mineralization by apatite crystallites, with concomitant extrafibrillar mineralization between adjacent collagen fibrils.

Models of collagen mineralization

In our early studies, totally or partially demineralized dentin collagen matrices were employed for remineralization of hybrid layers and caries-like dentin lesions . However, it is not known if the remineralization demonstrated in those studies was caused by the presence of remnant phosphoproteins that remain bound to collagen matrices after demineralization . As self-assembled purified collagen fibrils do not contain bound matrix proteins, the authors employed a two-dimensional model of biomimetic mineralization of collagen based on reconstituted collagen fibrils deposited on grids used for transmission electron microscopy . Briefly, bovine skin-derived type I collagen dissolved in acetic acid was reconstituted on formvar-coated grids by ammonia diffusion and cross-linked with 0.3 M carbodiimide. Then, the grid with a single layer of collagen was floated over a drop of a prospective biomimetic analog-containing biomineralization medium. The grid was retrieved at different time points (usually within 72 h) and rinsed with water. They were examined unstained for mineral deposition or stained for correlation of mineral deposition with collagen cross-striation patterns. With this model, we demonstrated a highly hierarchical assembly of intrafibrillar apatite crystallites in reconstituted collagen mineralized ex situ with biomimetic analogs of extracellular matrix proteins ( Fig. 1 ). This 2-D single-layer collagen mineralization model is useful as a rapid screening tool for potential biomimetic analogs for collagen mineralization . Moreover, the technique may be modified for creating intrafibrillarly mineralized collagen coatings on the surfaces of orthopedic and dental implants as a more bio-compatible alternative to the use of hydroxyapatite coatings .

Fig. 1
A two-dimensional model for examining mineralization of reconstituted collagen fibrils, using a dual biomimetic analog mineralization protocol. Transmission electron microscopy (TEM) images of unsectioned reconstituted collagen fibrils. (A) After 24 h of mineralization. The fibril was briefly stained with uranyl acetate to highlight the gap and overlap zones. Amorphous calcium phosphate (ACP) prenucleation clusters (open arrowhead) could be seen along the periphery of the fibril. Pointer: initial needle-shaped intrafibrillar apatite. (B) Unstained collagen fibril after 48 h of mineralization. Prenucleation clusters have coalesced into ACP droplets (arrow) that continued to infiltrate the fibril. Denser intrafibrillar apatite deposition could be seen but the banding periodicity caused by hierarchical arrangement of the apatite crystallites was still present. (C) Schematic depicting replacement of intrafibrillar water with fluidic polyacrylic acid-stabilized ACP nanoprecursors. (D) Unstained fibril showing heavy intrafibrillar mineralization with apatite platelets after 72 h. Banding characteristics was obscured by the heavy mineralization.

Although the single-layer collagen model is a convenient and rapid method for studying biomimetic mineralization of individual collagen fibrils, the simple 2-D structure of the collagen model limits it further application. To further mimic natural collagen matrices, 3-D models of reconstituted collagen ( Fig. 2 ) were used by different research groups to evaluate the ability of the biomimetic mineralization scheme to mineralize collagen bundle assemblies within a three-dimensional scaffold . Briefly, commercially available type I collagen sponges derived from bovine tendon were cultured in analog-containing biomineralization medium for 3–14 days. The mineralization assembly was kept in a 37 °C oven to emulate physiological conditions. At predetermined reaction times, mineralized samples were removed from the solution, copiously washed with deionized water and further prepared for different analysis, including transmission electron microscopy, thermogravimetric analysis, attenuated total reflection-Fourier transform infrared spectroscopy, nuclear magnetic resonance spectroscopy and X-ray diffraction, which cannot be easily achieved with the 2-D collagen model . The formed collagen–hydroxyapatite composites indicated that polyaspartic acid-stabilized ACP fluid droplets can diffuse into collagen bundles and mineralize a relatively thick collagen scaffold .

Fig. 2
A three-dimensional reconstituted collagen model of biomimetic mineralization. TEM images of an unstained section taken from a highly porous collagen scaffold that had been mineralized with a calcium and phosphate-containing medium containing dual biomimetic analogs for 14 days. (A) Low magnification of the mineralized collagen scaffold, showing interconnecting mineralized collagen bundles. (B) High magnification taken from a part of a mineralized collagen bundle. Collagen fibrils on the left are less heavily mineralized, with hierarchical deposition of apatite platelets in the gap zones, producing a periodic banding pattern. Collagen fibrils on the right are heavily mineralized with apatite platelets. Banding characteristic can no longer be identified.

Although the mechanical properties of 3-D collagen sponges increase significantly after functional mineralization, they are still far less than those of the natural mineralized tissues because of their highly porous nature. Thus, attempts have been made by various groups to produce biomaterials that structurally mimic bone and dentin . However, the hierarchical complexity of natural composites is usually far beyond contemporary bioengineering processing capability. One approach is to utilize the hierarchical structure of natural materials as a template for the development of high-performance engineering materials using in vitro biomimetic methods . Thus, researchers revisited the use of natural collagen model, including rat tail tendon (a soft tissue collagen 3D model; Fig. 3 ), demineralized manatee bone and demineralized dentin collagen . It was found that the amount of mineral present in the rat tail tendon samples represented the maximum amount of mineral available in the PILP solution . Conversely, for demineralized manatee bone specimens, mineral distribution and penetration across the bulk of the bone matrix appears to be a challenge for the PILP process. The mineral penetration depth was limited to 100 μm . The mineral content for the remineralized manatee bone specimens (about 45 wt%) was below the theoretical value for cortical bone, even though enough calcium and phosphate ions were available in the PILP solution to restore the mineral value of bone. It was proposed that auto-transformation and solidification of the surface-infiltrated ACP nanoprecursors into crystalline apatite with time could potentially block further infiltration of the additional nanoprecursors into the bulk of the demineralized bone matrix, thereby limiting their ultimate depth of penetration.

Fig. 3
A three-dimensional natural soft tissue collagen model of biomimetic mineralization. TEM images of unstained sections taken from mineralized rat tail tendon with parallel collagen fibrils at different time periods. (A) Incomplete mineralization at 7 days. Asterisk: Unmineralized collagen fibrils; Arrow: extrafibrillar mineralization; Open arrowhead: Intrafibrillar mineralization recapitulating the D-spacings of fibrillar collagen. (B) Heavy mineralization of the parallel collagen fibrils after 14 days. (C) High magnification of B, showing a collagen fibril with extrafibrillar (arrow) and intrafibrillar mineralization by discrete apatite platelets (open arrowhead). (D) Selected area electron diffraction of the mineralized fibril in (C) produced arc-shaped diffraction patterns in the 0 0 2 and 0 0 4 plane of apatite. The C -axis of the apatite platelets is arranged almost parallel to the longitudinal axis of the collagen fibril.

The need for using phosphate-containing biomimetic analogs in biomimetic mineralization of collagen

The functional role played by collagen in calcium phosphate biomineralization has been conjectural based on the results of the past several decades of research. Earlier studies indicated that fibrillar collagen does not initiate biomineralization on its own, but serves as a passive depot for the housing of apatite crystallites . This led to a plethora of studies that examined the biological control of intrafibrillar collagen mineralization by noncollagenous proteins . Noncollagenous proteins are believed to play a crucial role in the mineralization of bone and dentin. This assertion is supported by studies demonstrating that mutations in genes that code for these proteins result in abnormal bone and dentin mineralization . The extracellular matrix of bone and dentin contains small quantities of noncollagenous proteins such as osteopontin, bone sialoprotein, dentin matrix protein 1 (DMP1), and dentin sialophosphoprotein (DSPP) . The latter is further cleaved into dentin sialoprotein and dentin phosphoprotein (DPP; aka phosphophoryn) . These proteins are highly anionic due to the prevalence of carboxylate groups on the polyaspartic acid residues that comprise the protein backbone. Post-translation phosphorylation of the serine residues produces phosphoserines, which further augments their anionic character . For example, dentin phosphoprotein, being the most abundant NCP in dentin, contains a large number of aspartic acid (Asp) and phosphoserines (Pse) in the repeating sequences of (Asp-Pse) n and (Asp-Pse-Pse) n . Dentin matrix protein 1, another member of the Small Integrin-Binding Ligand Interacting Glycoproteins (SIBLING) family, also contains high levels of Asp and Pse . A unifying feature of the SIBLING proteins is that they all contain an acidic serine aspartate-rich MEPE (ASARM)-associated motif. The ASARM motif in SIBLING genes and the released ASARM peptide play roles in mineralization of bone and teeth . The highly anionic nature of noncollagenous proteins involved in biomineralization enables them to sequester and bind calcium ions and presenting them to collagen fibrils at the mineralization front during the formation of bone and dentin.

In the authors’ laboratory, biomimetic mineralization of collagen was performed using a dual biomimetic analog strategy . Polyacrylic acid or polyaspartic acid was used as an analog for sequestering calcium ions released by set calcium silicate cements or a supersaturated calcium phosphate mineralization solution. These analogs function by acting as surfactants to prevent fluidic ACP nanoparticles from aggregating into larger particles, and to inhibit auto-transformation of the ACP nanoparticles into apatite (i.e. apatite nucleation inhibitor) prior to their entry into the intrafibrillar water compartments of the collagen fibril . In addition, a polyphosphate-containing biomimetic analog such as polyvinylphosphonic acid, sodium trimetaphosphate or sodium ascorbyl phosphate was employed as a templating biomimetic analog of matrix phosphoproteins. These phosphoprotein analogs were allowed to bind to the collagen fibrils prior to immersion of the fibrillar matrices in the poly(anionic) acid-containing mineralization medium . Using the aforementioned collagen mineralization models, attachment of ACP nanoprecursors to the D-spacings of non-mineralized collagen fibrils was observed within 24 h. After 48 h, hierarchical apatite deposition was observed from unstained collagen fibrils that resulted in banded mineral arrangement. After 72 h, highly mineralized collagen fibrils could be seen in which the periodicity of the mineral arrangement was obscured as the fibrils were completely mineralized .

In contrast to the earlier work, more recent studies indicate that type I collagen plays an active role in biomineralization by acting as templates for physicochemical (electrostatic) attraction of ACP nanoparticles and direct formation of intrafibrillar apatite within the gap zones, without the intervention or mediation of other noncollagenous proteins present in vertebrate calcifying tissues . Silver and Landis reported specific sites in the e2 band of the gap zones (corresponding to the a, e and d bands) of type I collagen that have the potential to sequester and bind calcium ions . Nudelman et al. calculated that there are net positive charges close to the C-terminal end of the collagen molecules that favor infiltration of the collagen fibrils with negatively charged ACP nanoprecursors. The arrangement of charged amino acids within the gap and overlap zones produce nucleation sites that control the conversion of the ACP nanoprecursors into oriented apatite crystals . Indeed, several research groups have demonstrated that it is possible to produce intrafibrillar mineralization of type I collagen using poly(aspartic acid)-stabilized ACPs alone as an apatite nucleation inhibitor, without the adjunctive use of phosphorylated DDP, phosphorylated DMP1 or their polyphosphate analogs . Although this simplified biomimetic collagen mineralization strategy is more economical and reduces time in preparing mineralized collagen scaffolds from a tissue engineering perspective, it begs the provision of a rationale for the existence of highly phosphorylated noncollagenous proteins in bone and dentin. The use of the simplified biomimetic mineralization strategy also challenges earlier biological studies that reported critical roles played by the phosphorylated forms of these noncollagenous proteins in the formation of mineralized collagenous tissues .

A recent study compared the effects of phosphorylated vs non-phosphorylated forms of DPP and DMP1 on collagen mineralization using a 2D model . Although differences existed between DPP and DMP1 in the locations in which collagen fibrils were mineralized (intrafibrillar vs extrafibrillar), both phosphorylated proteins facilitated highly organized intrafibrillar mineralization of collagen fibrils. Conversely, the use of non-phosphorylated forms of these proteins resulted in randomly oriented intrafibrillar crystallites, with no particular organization of their crystallographic axes to the longitudinal axis of the collagen fibrils. Nevertheless, this study failed to explain why non-phosphorylated biomimetic apatite nucleation inhibitors such as polyaspartic acid or fetuin-A are capable of producing highly organized intrafibrillarly mineralized collagen. It must be emphasized that while the concepts of particle-based ACP prenucleation clusters and the PILP phenomenon of fluidic ACP infiltration are new in collagen biomineralization research, the theoretical basis for interpretation of ultrastructural results of intrafibrillar apatite deposition is not , and is based upon the classic model of collagen molecular packing proposed by Petruska and Hodge . This straight and rod-like model of steric arrangement of collagen molecules has since been replaced by a synchrotron X-ray diffraction-derived model in which collagen molecules are arranged in a right-handed helically twisted, discontinuous manner along the length of the microfibrils . In the latter model, interdigitation of adjacent microfibrils placed geometric constraints on the availability of lateral intermolecular spacings between collagen molecules, with no room to accommodate apatite platelets outside the gap zones.

A recent study examined the effect of using single (polyacrylic acid) vs dual biomimetic analogs for mineralization of collagen fibrils (polyacrylic acid and sodium trimetaphosphate) . In that study, polyacrylic acid was employed as an analog to inhibit apatite nucleation, and sodium trimetaphosphate was used as a templating analog for guiding intrafibrillar apatite deposition. The use of polyacrylic acid without a templating analog resulted only in intrafibrillar mineralization with continuous apatite strands. Conversely, the use of both analogs resulted in intrafibrillar mineralization with discrete apatite crystallites. While both methods resulted in intrafibrillar mineralization of collagen fibrils, the authors opined that in the absence of polyphosphate as a templating analog, infiltration of poly(anionic) acid-stabilized ACP nanoprecursors via a PILP process into the interconnecting water-filled volume within a collagen fibril appeared to have resulted in molding of ACP nanoprecursors into a continuum. This, in turn, resulted in crystallization of the carbonated apatite into a monolithic crystalline structure. Such a crystallization mechanism produces mineralized collagen entities that resemble the monolithic single-crystal structure in sea urchin spines or siliceous bio-skeletons. It is possible that release of poly(anionic) acid into the intrafibrillar milieu results in osmotic swelling and relaxation of the helical collagen microfibrillar arrangement, that facilitates continuous apatite deposition or growth from the gap zones into overlap zones into continuous strands. Conversely, electrostatic binding of polyphosphate analogs to discrete sites along the collagen molecules may have caused the bound analogs to act as inhibitor to discourage continuous growth of the apatite crystallites along the overlap zones, thereby resulting in constraining apatite platelets to the gap zones of the collagen fibril. This hypothesis is speculative and awaits further validation.

Nevertheless, the aforementioned hypothesis is consistent with the timely report comparing the ultrastructure of mineralized collagen in bone with the use of ion-milled sections vs sections prepared by conventional ultramicrotomy . In ion-milled sections, the authors reported that approximately 70% of the minerals in bone are extrafibrillar, confirming previous models proposed by Lees et al. and Hellmich and Ulm . With respect to intrafibrillar apatite, crystallite platelets were only identified in the gap zones, without extension into the overlap zones, in contrast to what was previously proposed by Landis et al. . Similar results were reported in another study using steric modeling to estimate the packing density of apatite within the gap zones . Those modeling results were further confirmed using electron energy loss spectroscopy associated with scanning transmission electron microscopy. Taken together, these novel findings highlight that mechanisms in the control of discrete intrafibrillar apatite platelet deposition in the gap zones by phosphorylated NCPs are not completely understood. It is possible that the current success of intrafibrillar collagen mineralization with a single poly(anionic) acid analog may not result in truly biomimetically mineralized collagen that resembles those present in bone and dentin. These challenging issues require further in-depth investigations to decipher the riddles.

Only gold members can continue reading. Log In or Register to continue

Stay updated, free dental videos. Join our Telegram channel

Nov 25, 2017 | Posted by in Dental Materials | Comments Off on Biomimetic remineralization of dentin

VIDEdental - Online dental courses

Get VIDEdental app for watching clinical videos