In vitro Remineralization of Severely Compromised Bonded Dentin

Abstract

Biomimetic remineralization is potentially useful for the remineralization of incompletely resin-infiltrated collagen matrices created by etch-and-rinse adhesives. In this study, we tested the hypothesis that structurally altered dentin collagen cannot be remineralized to the same hierarchical order and dimension seen in structurally intact dentin collagen. The remineralization medium consisted of a set Portland cement/simulated body fluid system containing polycarboxylic acid and polyvinylphosphonic acid as biomimetic analogs. Remineralization of air-dried, collapsed hybrid layers was apparent after one month, with hybrid layers remineralized to 80-90% of their thickness after 2-4 months. A hypermineralized layer was seen on the hybrid layer surface, and tubular orifices were occluded with apatite deposits that resembled those present in non-carious cervical dentin. Structurally altered collagen is unlikely to be remineralized to the same hierarchical order and dimension as seen in intact dentin. The aggressively air-dried acid-etched dentin remineralization model also sheds light on the mechanism of sclerotic dentin formation.

Introduction

Natural mineralized dentin exhibits excellent strength and toughness because the collagen fibrils are reinforced by intrafibrillar and interfibrillar minerals (Kinney et al., 2003). These carbonated apatite nanoplatelets have to be sufficiently small and arranged in a specific order to be incorporated into the gap zones of collagen fibrils (Xu et al., 2007). Recent biomineralization studies have shown that fluidic amorphous nanoprecursors and mesocrystalline intermediates play critical roles in hard tissue formation (Cai and Tang, 2008; Cölfen, 2008).

A Portland-cement-based (Tay et al., 2007) biomimetic remineralization protocol has recently been developed (Tay and Pashley, 2008). This protocol involves the binding of two biomimetic analogs to dentin collagen so that the doped collagen can guide the scale and distribution of apatite remineralization. In this protocol, a low-molecular-weight polyacrylic acid (PAA) was used to mimic the stabilization function of dentin matrix protein 1 (DMP1) on amorphous calcium phosphate precursors (He et al., 2005). The PAA reduces these precursors released by interaction of set Portland cement with the simulated body fluid to a nanoscale (Olszta et al., 2003). This sequestration step prevents the nanoprecursors from aggregation and precipitation (Cai and Tang, 2008). The other biomimetic analog, polyvinylphosphosphonic acid (PVPA), is a polyanion that mimics the negative charges of phosphoproteins such as DMP1, phosphophoryn, or bone sialoprotein (Gajjeraman et al., 2007; Baht et al., 2008) and their roles in organizing aspects of mineralization.

Resin-dentin bonds created by contemporary etch-and-rinse adhesives are not durable (Pashley et al., 2007), since degeneration of denuded collagen fibrils within hybrid layers can occur in vivo (Carrilho et al., 2007). Biomimetic remineralization provides a potential means for extending the longevity of these bonds by remineralizing incompletely infiltrated hybrid layers (Tay and Pashley, 2009). Demineralized collagen fibrils may collapse via inter-peptide hydrogen bonding when they are inadvertently air-dried during bonding, which restricts the infiltration of adhesive resin monomers (Pashley et al., 2007). It is unknown whether a similar process prevents the penetration of the PAA-stabilized amorphous calcium phosphate nanoprecursors into the resin-dentin interfaces produced after extensive air-drying.

There has been a recent paradigm shift on the clinical treatment of deep caries lesions. It is now acceptable to leave behind caries-affected or even caries-infected dentin during restorative procedures (Thompson et al., 2008). Carious dentin may contain structurally altered collagen of various thicknesses, depending on the clinician’s perception of the end-point of caries excavation. Since the order between gap zones and collagen molecules is irreversibly altered in denatured collagen (i.e., gelatin) (Zeugolis et al., 2008), the hierarchy of apatite remineralization within the latter may be different from that observed in intact mineralized collagen. In the present study, the limits of biomimetic remineralization of resin-dentin interfaces were examined by means of a demineralized dentin model consisting of a thin surface layer of structurally altered collagen and a subsurface network of collapsed, but structurally intact, collagen fibrils. Extensive air-drying of acid-etched dentin is far removed from clinical practice. However, an understanding of the mechanisms of biomimetic remineralization in collapsed and structurally altered collagen matrices provides critical information that facilitates future attempts to remineralize thick layers of carious dentin. Thus, the hypothesis tested was that structurally altered dentin collagen cannot be remineralized to the same hierarchical order and dimension as seen in structurally intact dentin collagen.

Materials & Methods

Remineralization Medium

Type I white Portland cement (Lehigh Cement Company, Allentown, PA, USA) was sieved and mixed with de-ionized water in a water-to-powder ratio of 0.35:1, placed in flexible silicone molds, and allowed to set and age at 100% relative humidity for 1 wk before use (Tay et al., 2007). We prepared a simulated body fluid (SBF) by dissolving 136.8 mM NaCl, 4.2 mM NaHCO₃, 3.0 mM KCl, 1.0 mM K₂HPO₄·3H₂O, 1.5 mM MgCl₂·6H₂O, 2.5 mM CaCl₂, and 0.5 mM Na₂SO₄ in de-ionized water (Kokubo et al., 1990) and adding 3.08 mM sodium azide to prevent bacterial growth. The SBF was buffered to pH 7.4 with 0.1 M Tris Base and 0.1 M HCl. For biomimetic remineralization, 500 µg/mL of polyacrylic acid (Mw 1800; Sigma-Aldrich, St. Louis, MO, USA) and 200 µg/mL of polyvinylphosphonic acid (Mw 24,000; Sigma-Aldrich) were added to the SBF as biomimetic analogs, with the pH of the latter adjusted to 7.4.

Dentin Bonding

Twenty recently extracted human third molars were collected after patients’ informed consents were obtained under a protocol reviewed and approved by the Human Assurance Committee of the Medical College of Georgia. A flat dentin surface was prepared perpendicular to the longitudinal axis of each tooth by means of a slow-speed Isomet saw (Buehler Ltd., Lake Bluff, IL, USA) under water-cooling. The exposed tooth surface was further wet-polished with a 320-grit silicon carbide paper (Oliveira et al., 2003) to create an enamel-free bonding surface in mid-coronal dentin. Two etch-and-rinse adhesives were used: One-Step (Bisco, Schaumburg, IL, USA) and Single Bond Plus (3M ESPE, St. Paul, MN, USA), with 10 teeth randomly assigned to each adhesive. Each dentin surface was etched with 32% phosphoric acid (Uni-Etch, Bisco) for 15 sec, rinsed with water, and air-dried with 0.207 MPa compressed air for 60 sec applied at 5 mm away from the etched dentin. Each adhesive was then applied to the air-dried dentin with 2 adhesive coats and light-cured for 20 sec by means of a light-curing unit with an output intensity of 600 mW/cm2. This was followed by the incremental placement of two 2-mm-thick layers of a resin composite. Each layer was light-cured separately for 40 sec. Each tooth was then sectioned occluso-gingivally to produce four 1-mm-thick slabs, each containing the resin-dentin interface (i.e., 4 x 10 = 40 specimen slabs per adhesive). These 40 specimens were randomly assigned to experimental (20 specimens/adhesive) and control groups (20 specimens/adhesive).

Biomimetic Remineralization

Each experimental specimen slab was placed over a set Portland cement block (ca. 1 g) inside a glass scintillation vial. The latter was filled with 15 mL of SBF containing the 2 biomimetic analogs and incubated at 37°C. The remineralization medium was changed every month, with its pH (after inclusion of the Portland cement blocks) monitored weekly so that it was above 9.25. This ensured that apatite was formed instead of octacalcium phosphate (Meyer, 1983; Eanes, 2001). The set-up for the control specimens was the same, except that the liquid medium was replaced with SBF only. For each adhesive, 4 experimental specimens were retrieved after 1-4 mos for ultrastructural examination of the extent of remineralization (i.e., 4 x 4 specimens per mo = 16 non-demineralized experimental specimens). For each adhesive, 2 control specimen slabs were examined at the baseline period (i.e., before immersion) for the extent of incomplete resin infiltration via the ammoniacal silver nitrate tracer protocol reported previously (Tay et al., 2002). In addition, 4 different control specimen slabs were examined after 1-4 mos of immersion in the Portland cement/SBF for evaluation of the extent of remineralization (i.e., 2 baseline control specimens + 4 x 4 control specimens/mo = 18 non-demineralized control specimens).

Transmission Electron Microscopy (TEM)

Following reduction of the diamine silver ions ([Ag(NH₃)₂]⁺), the silver-impregnated control specimen slabs were fixed in Karnovsky’s fixative, post-fixed in 1% OsO₄, dehydrated in ascending ethanol (50-100%), immersed in propylene oxide, and embedded in epoxy resin. For examination of the extent of remineralization, experimental specimen slabs were processed for epoxy resin embedding in the manner described previously, but without silver impregnation. Non-demineralized, 90-nm-thick sections were examined without staining in a JEM-1230 TEM (JEOL, Tokyo, Japan) operated at 110 kV.

To examine the conditions of the collagen matrix in the heavily remineralized experimental specimens, the last 4 experimental slabs in each adhesive that had undergone 4 mos of biomimetic remineralization were completely demineralized in a formic acid/sodium formate buffer (i.e., 16 non-demineralized experimental specimens + 4 demineralized experimental specimens = 20 experimental specimens/adhesive). The last 2 control slabs in each adhesive that had been immersed in SBF were completely demineralized in a similar manner (i.e., 18 non-demineralized control specimens + 2 demineralized control specimens = 20 control specimens/adhesive). After TEM processing, 90- nm-thick sections were stained with phosphotungstic acid/uranyl acetate before examination.

Results

Control Specimens

Intense air-drying resulted in extensive areas of incomplete resin infiltration in hybrid layers created by both adhesives (Figs.1A, 1B), with extensive silver uptake within the denuded collagen matrices. Control specimens examined after 4 mos of immersion in the remineralization medium exhibited no remineralization within the hybrid layers (Figs. 1C, 1D).

Figure 1.

Transmission electron micrographs (TEMs) of non-demineralized, silver-impregnated resin-dentin interfaces created by the application of (A) One-Step and (B) Single Bond Plus to intensively air-dried acid-etched dentin. Extensive regions of incomplete resin infiltration, as indicated by the severe silver uptake (L), could be seen within the unstained hybrid layers (between open arrows). In (A), the surface and basal portions of the hybrid layer (open arrowheads) in One-Step were well-infiltrated by adhesive resin and did not exhibit silver deposits. In (B), the slightly electron-dense globules (arrow) in the adhesive represent polyalkenoic acid copolymer phase separation droplets that are characteristic of this adhesive. Unstained, non-demineralized TEM micrographs of (C) a One-Step-bonded control specimen and (D) a Single Bond Plus-bonded control specimen showing no remineralization within the hybrid layers after 4 mos of immersion in the Portland cement/SBF medium that did not contain biomimetic analogs. Abbreviations: C, resin composite; A, adhesive; T, dentinal tubule; H, hybrid layer; D, mineralized dentin base; P, polyalkenoic acid copolymer.

Experimental Specimens

Resin-dentin interfaces exhibited evidence of remineralization as early as 1 mo (Figs. 2A, 2B), which became very distinct after 2 mos (Figs. 2C, 2D), with both intrafibrillar and interfibrillar remineralization within the denuded collagen matrix. Initially, intrafibrillar minerals appeared as nanocrystals (Fig. 2E; Tay and Pashley, 2008) that eventually coalesced to produce larger mineral platelets within the collagen fibrils (Fig. 2F). These platelets were identified by selected-area electron diffraction to be apatite (Fig. 2F-inset).

Figure 2.

TEMs taken from unstained, non-demineralized sections of representative experimental resin-dentin slabs that had undergone biomimetic remineralization (BR) for 1-2 mos. (A) One-Step after 1 mo of BR; (B) Single Bond Plus after 1 mo of BR; (C) One-Step after 2 mos of BR; (D) Single Bond Plus after 2 mos of BR. Note the progressive increase in electron density of the remineralized regions within the hybrid layer (asterisks) with time. Better-resin-infiltrated regions along the top and basal portions of the hybrid layers (pointers) were not remineralized. From 2 mos onward, the dentin surfaces of the One-Step specimens appeared hypermineralized (open arrowhead; Fig. 2C), while some tubular orifices in both the One-Step and Single Bond Plus specimens were occluded with mineral casts (arrows in Figs. 2C and 2D). These features are further elaborated in Fig. 3. C, composite; A, adhesive; T, dentinal tubule; D, mineralized dentin base; P (for Single Bond Plus), polyalkenoic acid copolymer. (E) Representative example of intrafibrillar (open arrowheads) and interfibrillar minerals (open arrow) formed after 1 mo of BR. (F) Representative example of denser intrafibrillar (open arrowheads) and interfibrillar minerals (open arrow) formed after 1 mo of BR. At this stage, the intrafibrillar minerals appeared as overlapping nano-platelets. Selected-area electron diffraction of these crystallites with indexing confirmed the presence of the apatite phase.

By 3-4 months, most of the hybrid layers had remineralized to 80-90% of their entire thickness, except for a thin basal portion that appeared well-infiltrated with resin and did not exhibit any signs of remineralization. Additional features of hypermineralization (Kwong et al., 2000) could be seen. One-Step-bonded specimens in which a hypermineralized layer was present along the surface of the remineralized hybrid layer are shown in Figs. 3A and 3B. Cohesive failure of the hypermineralized layers occurred during ultramicrotomy for all One-Step-bonded specimens. These hypermineralized layers lacked the hierarchical crystallite arrangement found in a mineralized, intact collagen matrix. An atypical ordered arrangement could be identified (Fig. 3C), with the larger mineral platelets (ca. 25-30 nm) aligning linearly and parallel to one another (Tay et al., 2000). Along the crystallization front (Fig. 3C- inset), each platelet was formed by the fusion of multiple nanocrystals, providing evidence for involvement of mesocrystalline phases in the crystallization process. Hypermineralization features in the Single Bond Plus specimens appeared more subtle (Fig. 3D). The dentinal tubule orifices in these specimens exhibited a phenomenon similar to that in the One-Step specimens, in that they were often occluded by electron-dense mineral casts (Figs. 3D, 3E). The larger platelets that formed these mineral casts were identified via selected-area electron diffraction to be apatites (Fig. 4F).

Figure 3.

Representative unstained, non-demineralized TEMs of resin-dentin interfaces prepared from intensely air-dried acid-etched dentin that had remineralized after more than 2 mos of biomimetic remineralization (BR). (A) A One-Step-bonded specimen that had undergone 3 mos of BR. The adhesive (A) had separated from the remineralized hybrid layer (asterisk) by a space (S) due to cohesive failure of a hypermineralized surface layer (pointers). Dentinal tubule orifices were heavily filled with minerals (open arrowhead), while the underlying tubule (T) remained patent. D, mineralized dentin base; open arrow, original demineralization front. (B) Another One-Step-bonded specimen that had undergone 4 mos of BR. The hypermineralized layer (between open arrows) was continuous with the underlying remineralized hybrid layer (asterisk), and extended into the adhesive (A). The latter was partially dislodged and resulted in a space (S). (C) A high-magnification view of the boxed region in Fig. 3B, showing the density of the mineral platelets (ca. 10 nm in length) within the hypermineralized layer. A, adhesive. A very high magnification of the boxed region (inset) revealed the possible involvement of mesocrystalline phases along the crystallization front before fusion of these phases into larger crystal platelets. (D) A Single Bond Plus-bonded specimen after 4 mos of BR. A hypermineralized layer (pointer) could be seen on the surface of the partially remineralized hybrid layer (asterisk). In this section, dentinal tubules were oriented parallel to the bonded surface. Tubular orifices and subsurface tubules (open arrowheads) were heavily occluded with minerals, while those beyond the demineralization front remained filled with resin (T). A, adhesive; P, polyalkenoic acid copolymer; D, mineralized dentin base. (E) A higher-magnification view of a dentinal tubule (T) in a One-Step-bonded specimen after 4 mos of BR. The tubular orifice was occluded heavily with minerals (S). The surface hypermineralized layer (between open arrowheads) was 500-750 nm thick and was continuous with the less-highly-remineralized hybrid layer (asterisks). (F) A high-magnification view of the mineral platelets within the dentinal tubule in Fig. 3E. Selected-area electron diffraction (inset) of the platelets with indexing confirmed the presence of the apatite phase.

Figure 4.

TEMs of phosphotungstic acid/uranyl acetate-stained, demineralized sections taken from experimental specimens that had undergone biomimetic remineralization (BR) for 4 mos revealed the presence of a denatured surface collagen layer created by intense air-drying of acid-etched dentin. All the original and remineralized mineral components were completely dissolved to reveal the status of the collagen network. (A) Low-magnification view of a Single Bond Plus-bonded specimen showing a 500-nm-thick denatured collagen layer (between open arrows) along the dentin surface. Part of this layer (open arrowheads) had been displaced into the tubular orifice and the dentinal tubule (T). Asterisks, faint background-section folding artifacts; A, adhesive; P, polyalkenoic acid copolymer. (B) A high-magnification view of a One-Step-bonded surface showing a 750-nm-thick denatured layer along the hybrid layer surface, where collagen banding could not be recognized. Here, the collagen fibrils were denatured into gelatin microfibrils (asterisk). Banded collagen could be seen in the underlying ”remineralized” hybrid layer (H). The adhesive layer (A) separated from the dentin surface, resulting in a gap (G). (C) Cross-section of a dentinal tubule (T) located at 500 nm away from the dentin surface in a One-Step-bonded specimen. A zone of denatured collagen (between open arrowheads), caused by intense air-drying of the acid-etched dentin, could be identified along the periphery of the tubule. This zone of denatured collagen appeared as unraveled microfibrillar strands. D, demineralized dentin. (D) Cross-section of a dentinal tubule (T) located 10 µm away from the dentin surface (i.e., beneath the hybrid layer) in a One-Step-bonded specimen, showing the presence of intact banded collagen along the periphery of the tubule. D, demineralized dentin.

Structurally Altered Surface

Stained demineralized sections of four-month remineralized specimens revealed a 500- to 750-nm-thick structurally altered surface layer created by intense air-drying of acid-etched dentin (Figs. 4A, 4B). Fibrillar architecture and banding characteristics were absent within this layer, and a bed of unraveled microfibrillar strands (gelatin) was observed (Figs. 4B). A similar altered layer was seen around dentinal tubules that were located directly beneath the etched dentin surface (Fig. 4C). Conversely, collagen fibrils were intact in dentinal tubules that were located more than 10 µm beneath the etched dentin (Fig. 4D).

Discussion

When collapsed air-dried specimens were immersed in the aqueous remineralization medium, re-expansion of the collagen matrix created a sponge- like effect (Inaba et al., 1995) that probably absorbed the biomimetic analogs from the SBF into the denuded matrix. The intrafibrillar remineralization noted in this investigation provides indirect evidence to support that rehydration of a collapsed, but intact, collagen matrix is a reversible process (Pashley et al., 2007). These principles do not apply to the remineralization of the surface layer of structurally altered, demineralized collagen. This layer is only 500-750 nm thick, despite the presence of a 5- to 8-µm-thick layer of acid-etched dentin. Dehydration of the collagen fibrils could have introduced additional reversible interpeptide hydrogen bonding that increases the denaturing resistance of the bulk of the demineralized collagen matrix (Miles et al., 2005; Armstrong et al., 2006).

The remineralized structurally altered surface layer bore a morphologic resemblance to the surface of naturally occurring non-carious cervical dentin lesions (Kwong et al., 2000; Tay and Pashley, 2004). For both entities, dentinal tubules were occluded with apatite aggregates. Tubular occlusion by apatite crystallites is a feature also shared by transparent dentin beneath caries lesions (Daculsi et al., 1987) and age-induced transparent dentin (Kinney et al., 2005). The etiology of tubular occlusion has been ascribed to a “dissolution and reprecipitation” mechanism (Porter et al., 2005; Zavgorodniy et al., 2008). The results of the present study further suggest that structurally altered collagen plugs must be present in the tubular orifices before tubular occlusion can occur via a “dissolution and reprecipitation” mechanism.

Apatite crystallites within natural hypermineralized intertubular dentin exhibit an atypical head-to-tail alignment (Tay et al., 2000; Tay and Pashley, 2004). Similar features were seen in the remineralized surface layer of intensively air-dried bonded dentin. The association of the remineralized surface layer with a layer of structurally altered collagen provides an alternative mechanism to account for the etiology of hypermineralized dentin. Binding of phosphoproteins to collagen results in conformation changes in the collagen triple helix (Dahl et al., 1998), possibly increasing the flexibility of the hole zones (Landis et al., 2006) to receive calcium phosphate nanoprecursors. Interaction of biomimetic phosphoprotein analogs such as PVPA with intact dentin collagen may achieve similar results. Apatite deposition was considerably more difficult to achieve in type I collagen gels than gelatin gels due to the steric effects of intact collagen on apatite nucleation and growth (Blumenthal et al., 1991). Presumably, alignment of the collagen molecules in a quarter-staggered configuration imposes steric constraints on the apposition of apatites in the hole zone regions. When collagen is denatured, most of these triple helical arrangements are lost as the latter progressively dissociates into the 3 randomly coiled peptide α-chains (gelatin) (Johns and Courts, 1977). When these steric constraints are removed, phosphoprotein analogs such as PVPA may bind to alternative sites along the dissociated gelatin peptide chains. Since there may be as many as 3 times the number of α-chains compared with the original number of triple helices, this may result in hypermineralization when apatite nanocrystals are deposited around the PVPA-bound gelatin.

The results of the present study indicated that although structurally altered collagen matrices are amenable to remineralization in the presence of biomimetic analogs, they are unlikely to be remineralized to the same hierarchical order and dimension as seen in structurally intact dentin collagen. It is important to recognize, however, that these morphologic results were based on two-dimensional TEM images. We are aware that native tissues that have similar morphologic features may have different properties and compositions (Katz et al., 2007). Thus, the nanomechanical properties of remineralized hybrid layers in structurally intact and structurally altered collagen matrices should be further investigated by nanoscopic Dynamic Mechanical Analysis.

Biomimetic remineralization represents a revolutionary approach to improving the durability of resin-dentin bonds by a particle-mediated, non-classic crystallization strategy. We are still at a proof-of-concept stage, since planar remineralization was achieved via placement of a sectioned specimen slab on top of a Portland cement block. Understanding how compromised resin-bonded dentin responds to biomimetic remineralization will help establish the foundation for more sophisticated translational strategies to be designed for delivery of the different components of biomimetic remineralization in a three-dimensional manner to resin-dentin interfaces.