Advertisement

Dentin Matrix Protein 1 Immobilized on Type I Collagen Fibrils Facilitates Apatite Deposition in Vitro*

  • Gen He
    Affiliations
    Department of Oral Biology, University of Illinois at Chicago, Chicago, Illinois 60612
    Search for articles by this author
  • Anne George
    Correspondence
    To whom correspondence should be addressed: Dept. of Oral Biology (M/C 690), University of Illinois at Chicago, Chicago, IL 60612
    Affiliations
    Department of Oral Biology, University of Illinois at Chicago, Chicago, Illinois 60612
    Search for articles by this author
  • Author Footnotes
    * This study was supported by National Institutes of Health Grants DE 13836 and DE 11657. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Open AccessPublished:December 29, 2003DOI:https://doi.org/10.1074/jbc.M309296200
      During bone and dentin mineralization, the crystal nucleation and growth processes are considered to be matrix regulated. Osteoblasts and odontoblasts synthesize a polymeric collagenous matrix, which forms a template for apatite initiation and elongation. Coordinated and controlled reaction between type I collagen and bone/dentin-specific noncollagenous proteins are necessary for well defined biogenic crystal formation. However, the process by which collagen surfaces become mineralized is not understood. Dentin matrix protein 1 (DMP1) is an acidic noncollagenous protein expressed during the initial stages of mineralized matrix formation in bone and dentin. Here we show that DMP1 bound specifically to type I collagen, with the binding region located at the N-telopeptide region of type I collagen. Peptide mapping identified two acidic clusters in DMP1 responsible for interacting with type I collagen. The collagen binding property of these domains was further confirmed by site-directed mutagenesis. Transmission electron microscopy analyses have localized DMP1 in the gap region of the collagen fibrils. Fibrillogenesis assays further demonstrated that DMP1 accelerated the assembly of the collagen fibrils in vitro and also increased the diameter of the reconstituted collagen fibrils. In vitro mineralization studies in the presence of calcium and phosphate ions demonstrated apatite deposition only at the collagen-bound DMP1 sites. Thus specific binding of DMP1 and possibly other noncollagenous proteins on the collagen fibril might be a key step in collagen matrix organization and mineralization.
      Mineralized tissues such as bone and dentin are hierarchically organized biocomposites possessing unique mechanical properties (
      • Weiner S.
      • Wagner H.D.
      ). Understanding the biomineralization process is of great value for synthesis of bioengineered bone and dentin-like materials with optimum structural properties. Type I collagen accounts for 90% of the total protein in the organic matrix of bone and dentin (
      • Weiner S.
      • Wagner H.D.
      ). It not only provides the structural framework with viscoelastic properties but also defines compartments for ordered mineral deposition. Studies have demonstrated that the apatite crystals are first nucleated in the gap region, and the growing mineral platelets are highly organized in a staggered manner within the collagen fibrils (
      • Traub W.
      • Arad T.
      • Weiner S.
      ,
      • Jager I.
      • Fratzl P.
      ).
      It is well established that type I collagen matrix does not have the capacity to induce matrix-specific mineral formation from metastable calcium phosphate solutions that do not spontaneously precipitate. Several in vitro studies have demonstrated that ordered mineralization of apatite on collagen fibril is impossible without additives (
      • Saito T.
      • Arsenault A.L.
      • Yamauchi M.
      • Kuboki Y.
      • Crenshaw M.A.
      ,
      • Bradt J.
      • Mertig M.
      • Teresiak A.
      • Pompe W.
      ). As a result, much attention has been drawn to the noncollagenous proteins (NCPs)
      The abbreviations used are: NCP, noncollagenous protein; AFM, atomic force microscopy; MALDI, matrix-assisted laser desorption ionization; TOF, time-of-flight; TEM, transmission electron microscopy; DMP, dentin matrix protein; PBS, phosphate-buffered saline; GST, glutathione S-transferase; BSA, bovine serum albumin; TPCK, l-1-tosylamido-2-phenylethyl chloromethyl ketone.
      1The abbreviations used are: NCP, noncollagenous protein; AFM, atomic force microscopy; MALDI, matrix-assisted laser desorption ionization; TOF, time-of-flight; TEM, transmission electron microscopy; DMP, dentin matrix protein; PBS, phosphate-buffered saline; GST, glutathione S-transferase; BSA, bovine serum albumin; TPCK, l-1-tosylamido-2-phenylethyl chloromethyl ketone.
      that are tightly bound to the collagen fibers in mineralized tissues. Bone and/or dentin-specific NCPs are mostly acidic in nature and are rich in glutamic acid, aspartic acid, and phosphoserines (
      • Boskey A.L.
      ,
      • Butler W.T.
      • Ritchie H.
      ). They possess high calcium binding capacity and hydroxyapatite affinity (
      • Chen Y.
      • Bal B.S.
      • Gorski J.P.
      ,
      • Stetler-Stevenson W.G.
      • Veis A.
      ,
      • Wallwork M.L.
      • Kirkham J.
      • Chen H.
      • Chang S.X.
      • Robinson C.
      • Smith D.A.
      • Clarkson B.H.
      ). In vitro mineralization analyses suggest that these NCPs can greatly influence the apatite deposition rate and morphology of crystals, therefore they can be considered as nucleators or inhibitors of mineralization (
      • Hunter G.K.
      • Hauschka P.V.
      • Poole A.R.
      • Rosenberg L.C.
      • Goldberg H.A.
      ). Results from genetic disease analyses and transgenic animal models show that dysfunctional NCPs cause impaired mineralization in vivo (
      • Xiao S.
      • Yu C.
      • Chou X.
      • Yuan W.
      • Wang Y.
      • Bu L.
      • Fu G.
      • Qian M.
      • Yang J.
      • Shi Y.
      • Hu L.
      • Han B.
      • Wang Z.
      • Huang W.
      • Liu J.
      • Chen Z.
      • Zhao G.
      • Kong X.
      ,
      • Sreenath T.
      • Thyagarajan T.
      • Hall B.
      • Longenecker G.
      • D'Souza R.
      • Hong S.
      • Wright J.T.
      • MacDougall M.
      • Sauk J.
      • Kulkarni A.B.
      ). All of these results suggest an active regulatory role for NCPs during the biomineralization process.
      Several NCPs have been reported to bind to the collagen surface (
      • Chen Y.
      • Bal B.S.
      • Gorski J.P.
      ,
      • Dahl T.
      • Sabsay B.
      • Veis A.
      ). Phosphophoryn in particular was shown to bind at the e-band in the middle of the gap region (
      • Traub W.
      • Jodaikin A.
      • Arad T.
      • Veis A.
      • Sabsay B.
      ). Other studies have demonstrated that NCPs bind to charge group constellations favoring binding at the gap zone boundaries in collagen fibrils (
      • Fujisawa R.
      • Nodasaka Y.
      • Kuboki Y.
      ), where the mineral is initiated in vivo (
      • Traub W.
      • Arad T.
      • Weiner S.
      ). However, there is still no in-depth study pertaining to the specific nature of collagen-NCP interaction and how it modulates mineralization.
      Dentin matrix protein 1 (DMP1) is an acidic noncollagenous protein (
      • George A.
      • Sabsay B.
      • Simonian P.A.
      • Veis A.
      ) and is now known to be present in the mineralized matrix of both dentin and bone (
      • Qin C.
      • Brunn J.C.
      • Jones J.
      • George A.
      • Ramachandran A.
      • Gorski J.P.
      • Butler W.T.
      ). In situ hybridization experiments have demonstrated that the initial expression of DMP1 coincides with bone and dentin mineralization (
      • Kamiya N.
      • Takagi M.
      ,
      • George A.
      • Silberstein R.
      • Veis A.
      ), indicating that DMP1 is actively involved in regulating the temporal and spatial aspects of mineral initiation. This assumption was further confirmed by preliminary results showing that recombinant DMP1 can initiate apatite nucleation in vitro (
      • He G.
      • Dahl T.
      • Veis A.
      • George A.
      ).
      To understand DMP1-mediated spatial regulation of matrix-mediated mineralization, we hypothesize that specific collagen fibril-DMP1 interaction is necessary for mineral induction. In this study we have demonstrated that DMP1 binds to the N-telopeptide of fibrillar type I collagen in vitro in a regular, periodic manner. Further, the functional collagen binding domains in DMP1 have also been identified and characterized. These results provide the molecular basis underlying the initial steps involved in mineralized tissue formation, namely the specific binding of NCPs on collagen template.

      MATERIALS AND METHODS

      Solid-phase Binding Assay of DMP1 with Monomeric Collagen— Lathyritic rat skin collagen was prepared as described previously (
      • Payne K.J.
      • Veis A.
      ). The collagen was then dissolved in 0.01 m HCl at a 1 mg/ml concentration. Collagen binding property (
      • Dickeson S.K.
      • Mathis N.L.
      • Rahman M.
      • Bergelson J.M.
      • Santoro S.A.
      ) was determined by coating microtiter plates (Immulon 4, Dynex Technologies) with 5 μg of type I collagen in 100 μl of 0.1 m acetic acid/well overnight at 4 °C. Wells were washed three times with PBS and blocked with 0.5% (w/v) bovine serum albumin in PBS at 37 °C for 2 h before the addition of varying concentrations of the GST-fused recombinant DMP1 (
      • Srinivasan R.
      • Chen B.
      • Gorski J.P.
      • George A.
      ). After incubation at room temperature for 2 h with gentle shaking, the wells were extensively washed with PBST (PBS containing 0.1% (v/v) Tween 20). They were then incubated with mouse anti-GST monoclonal antibody (1:1000, Sigma) for 2 h. The plates were washed three times with PBST, and then alkaline phosphatase-conjugated anti-mouse IgG (Sigma) diluted 10,000-fold with blocking buffer was added to the wells. After incubation at room temperature for 1 h, the wells were washed again with PBST. The color was developed by adding 100 μl of 1.3 m diethanolamine, pH 9.8, containing 1 mg/ml p-nitrophenyl phosphate (Sigma) to the wells. After 15 min of incubation the absorbance at 405 nm (A405 nm) was measured using a Thermomax microplate reader (Molecular Devices Corp.). Experiments were performed in triplicate and repeated with independently prepared DMP1 preparations. Colorimetric readings from wells blocked with BSA without collagen coating were set as the base line and subtracted from reading values of other wells. Bound protein in each well was calculated based on the reference colorimetric readings from wells pre-coated with known amount of the corresponding protein. The KD values were determined from the corresponding Scatchard plot analysis. Data are presented as the mean value ± S.E. from a representative experiment (n = 3).
      Solid-phase Binding Assay of DMP1 with N-Telopeptide—The rat α1(I) N-telopeptide, EMSYGYDEKSAGVAVP, was synthesized and purified as described previously (
      • George A.
      • Malone J.P.
      • Veis A.
      ). The peptide was biotinylated by biotinamidohexanoic acid 3-sulfo-N-hydroxysuccinimide (Sigma) based on the protocol provided by the manufacturer. 10 μg of recombinant DMP1 in 100 μl of PBS was adsorbed onto Immulon 4 flat-bottom plate wells overnight at 4 °C. After washing once with 400 μl of wash buffer (150 mm NaCl, 0.1% Tween 20), the wells were blocked with 400 μl of blocking solution (0.5% BSA in PBS) for 2 h. The wells were then rinsed with 400 μl of washing buffer four times and incubated with 100 μl of biotinylated N-telopeptide for 2 h in a concentration-dependent gradient. After four washes with 400 μl of washing buffer, the wells were incubated with ExtrAvidin®-alkaline phosphatase (Sigma) for 1 h with a dilution at 1:50,000. The wells were then washed with washing buffer, developed with p-nitrophenyl phosphate (Sigma), and read with a colorimetric microtiter plate reader at 405 nm. Wells were set up in triplicate, and the final readings were averaged.
      DMP1 Subcloning—To identify the functional domain in DMP1 that interacts with type I collagen, the coding sequence of DMP1 cDNA was subamplified into two parts (nucleotides 58–1029 and 1009–1365) by PCR with a set of primers flanking the required sequence region. Both of the PCR fragments were then inserted into the pGEX-4T3 vector (Amersham Biosciences) and expressed in Escherichia coli BL21-DE3 cells (Invitrogen). The GST fusion protein, induced and purified by standard procedures, was cleaved by thrombin at 4 °C overnight. The corresponding recombinant peptides were termed N-DMP1 (residues 20–343) and C-DMP1 (residues 337–456).
      Identification of Collagen Binding Domains in DMP1 by Proteolytic Digestion and Mass Spectrometric Analysis—100 μg of C-DMP1 in 100 mm ammonium bicarbonate, pH 8.5, was mixed with 2 μg of Proteomics Grade Trypsin (TPCK-treated, Sigma) and incubated overnight at 37 °C. After incubation, the reaction was terminated by the addition of trypsin inhibitor (T9003, Sigma), and the mixture was loaded on a 1-ml collagen-immobilized Sepharose 4B column equilibrated with 100 mm ammonium bicarbonate, pH 8.5. The column was then washed extensively with 500 mm NaCl to dissociate nonspecific binding. The eluted solution and the original trypsin digests were subjected to MALDI analysis. An aliquot of the peptide mixture (1 μl) was spotted onto the MALDI target plate and overlaid with 1 μl of α-cyano-4-hydroxycinnamic acid solution (20 mg/ml in 100% acetonitrile). A Voyager-DE STR MALDI mass spectrometer (Applied Biosystems) equipped with delayed extraction was employed for peptide mass mapping in positive reflector mode. Peptide mass maps were identified based on the theoretical digests map obtained by the online protein-cutter program (delphi.phys.univ-tours.fr/Prolysis/cutter.html).
      Site-directed Mutagenesis of DMP1 cDNA—Based on MALDI analyses, two amino acid clusters in DMP1 were mutated to examine their collagen binding function. The primers used were: a, GGA TCC GCC AGA TAC CAA AAT ACT GAA; b, GTC CTC CTC ACT GGA CTC ACT GTT CTT TG; c, AGT GAG TCC AGT GAG GAG GAC AGA GCT GAA C; d, GC GGC CGC ATC TTG GCA ATC ATT GTC; e, AGA GTC ACT GTC CCT GTT TTC CTC AGA CGG CT; f, GAC AGT GAC TCT CAG GAC AGT AGC GCG ATC C.
      D1-col-m1 Construct—Using “ab” and “cd” primers separately, DMP1 cDNA was subamplified into two parts (58–1068 and 1048–1440). The PCR products were gel-purified, mixed in a 1:1 ratio, and used as the template for further amplification using “ad” as primers, to generate full-length DMP1 cDNA with mutated sequence. The PCR product was inserted into TA vector (PCR4TOPO, Invitrogen), released with BamHI/NotI, and ligated into pGEX4T-3 vector (Amersham Biosciences) which was digested by BamHI/NotI.
      D1-col-m2 Construct—Using “ae” and “df” primers separately, DMP1 cDNA was subamplified into two parts (58–1296, 1285–1440). The PCR products were gel-purified, mixed in a 1:1 ratio, and used as the template for further amplification using “ad” as primers. The PCR product was inserted into TA vector (PCR4TOPO, Invitrogen), released with BamHI/NotI, and ligated into pGEX4T-3 vector (Amersham Biosciences) which was digested by BamHI/NotI.
      D1-col-m3 Construct—The construction process is the same as for D1-col-m2 except that the PCR template used was D1-col-m1 cDNA.
      All of the constructs were inserted into pGEX4T-3 plasmid and transformed into BL21-DE3 (Invitrogen) E. coli. The recombinant proteins were expressed and purified as mentioned above.
      Fibrillogenesis Experiment—Collagen fibril assembly was monitored by spectrophotometric turbidity measurements at 313 nm as a function of time at 26 °C (
      • George A.
      • Veis A.
      ). The “cold-start” method was used to initiate fibrillogenesis (
      • Payne K.J.
      • King T.A.
      • Holmes D.F.
      ). At 4 °C, 500 μl of a solution containing 0.2 mg/ml type I collagen in 0.01 m HCl was mixed with 500 μl (0, 2, and 10 μm DMP1) of solution in double strength phosphate buffer (9.3 mm Na2HPO4, 5.83 mm KH2PO4, 150 mm NaCl, pH 7.04) separately, resulting in a 0.1 mg/ml collagen solution with 0, 1, and 5 μm DMP1 setups.
      Atomic Force Microscopy (AFM) and Transmission Electron Microscopy (TEM) Analyses—For AFM analyses, reconstituted collagen fibrils were prepared at 26 °C for 15 min as mentioned above. 10 μl of collagen solution was adsorbed onto freshly peeled mica surface for 10 min. After washing with water, the sample was air-dried, and collagen fibrils were imaged by tapping mode using multimode Nanoscope III AFM (Digital Instruments) equipped with a D-scanner. Silicon cantilevers with a 300-MHz resonance frequency were used. The average diameter and bandwidth of the fibrils were analyzed by “off-line section analyses.”
      For TEM analysis, colloidal gold-DMP1 complex was prepared according to the method of Slot and Geuze (
      • Slot J.W.
      • Geuze H.J.
      ) using a 4 μg/ml gold suspension to yield a particle size of 4–6 nm after centrifugation. Carbon-formvar-coated nickel grids were floated on a drop of fibrillar collagen suspension for 2 h at room temperature. The grids were subsequently washed with water and suspended on a drop of DMP1-gold solution surface overnight at 4 °C. The grids were then washed with water and negatively stained with 1% uranyl acetate for 1 min. Colloidal gold-labeled bovine serum albumin prepared according to the same procedure was used as a negative control for collagen binding analysis. The samples were examined with an electron microscope (JEOL 1200 EX) operating at 120 kV.
      In Vitro Apatite Nucleation Assay—The in vitro apatite nucleation assay was conducted as described previously (
      • He G.
      • Dahl T.
      • Veis A.
      • George A.
      ). Briefly, collagen-DMP1-coated grids were inserted into a channel connecting two halves of an electrolytic cell, with one compartment containing calcium buffer (165 mm NaCl, 10 mm HEPES, 2.5 mm CaCl2, pH 7.4) and the other phosphate buffer (165 mm NaCl, 10 mm HEPES, 1 mm KH2PO4, pH 7.4). The ionic concentration and ionic strength mimicked the physiological condition. When an external 1-mA current was applied, the ions traversed over the grid surface continuously. The buffers were changed regularly to maintain a constant ionic environment and were tested periodically. The nucleation experiment was conducted for 5 days. Grids coated with collagen fibrils served as control.

      RESULTS

      Binding of DMP1 to Immobilized Type I Collagen—A solidphase binding assay was used as the method of choice for characterization of interactions between collagen and noncollagenous proteins (
      • Chen Y.
      • Bal B.S.
      • Gorski J.P.
      ,
      • Dickeson S.K.
      • Mathis N.L.
      • Rahman M.
      • Bergelson J.M.
      • Santoro S.A.
      ). The binding curves shown in Fig. 1 demonstrate that the full-length DMP1 bound to collagen in a concentration-dependent and saturated manner (with an apparent KD = 3.8 ± 1.7 μm and maximum binding amount at Bmax = 68 ± 12.6 nmol/5 μg type I collagen). To identify the specific domain responsible for this interaction, DMP1 cDNA was subcloned into two parts and expressed recombinantly. A solid-phase binding assay clearly demonstrated that only the C-terminal end of DMP1 had affinity to the immobilized collagen with an apparent KD = 2 ± 0.5 μm and maximum binding amount at Bmax = 15 ± 2.3 nmol/5 μg of type I collagen (Fig. 1).
      Figure thumbnail gr1
      Fig. 1Binding of full-length recombinant DMP1, N-DMP1, and C-DMP1 with type I collagen. Microtiter plates were coated with 5 μg of type I collagen in 100 μl of 0.1 m acetic acid. After blocking with 0.5% (w/v) BSA in PBS for 2 h, wells were incubated with varying concentrations of the GST-fused protein for 2 h. After incubation with mouse anti-GST monoclonal antibody, wells were detected with alkaline phosphatase-conjugated anti-mouse IgG, and color was developed by p-nitrophenyl phosphate and quantified by measuring the absorbance at 405 nm (A405 nm). Colorimetric readings from wells blocked with BSA without collagen coating were set as the base line and subtracted from the values obtained from other wells. Bound protein in each well was calculated based on the reference colorimetric readings from wells pre-coated with known amount of the corresponding protein. The KD values and Bmax were calculated based on the corresponding Scatchard plot. Data are presented as the mean of triplicate analyses ± S.E.
      DMP1 Specifically Binds to the N-Telopeptide Region of the Collagen Fibril—In an attempt to study the specific domain on the collagen molecule that was responsible for interaction with DMP1, solid-phase binding assays were performed with synthesized N-telopeptide (Fig. 2). Results from this study showed that the N-telopeptide demonstrated strong specific binding affinity with DMP1 with a KD of 0.3 ± 0.12 μm. The binding amount Bmax was found to be 2.2 ± 0.3 nmol/10 μg of coated DMP1, which represents an ∼1:1 molar ratio. This observation suggests that in the collagen fibril assembly the N-telopeptide region could be a major site for specific binding of DMP1.
      Figure thumbnail gr2
      Fig. 2Binding assay of N-telopeptide with DMP1. Microtiter plates were coated with 10 μg of recombinant DMP1 in PBS. The rat α1(I) N-telopeptide EMSYGYDEKSAGVAVP was synthesized. A solid-phase binding assay was conducted with gradients of biotinylated telopeptides as described under “Materials and Methods.” N-Telopeptide showed specific binding to DMP1 with a KD of 0.3 ± 0.12 μm.
      Mapping the Collagen Binding Sites on DMP1—To identify the amino acid residues on DMP1 that are directly responsible for collagen binding, the C-terminal DMP1 was trypsinized and passed through a collagen-immobilized column. MALDI-TOF analysis demonstrated that three peptides bound on the collagen column: 349DSESSEEDR357 (theoretical [M+H]+ = 424SEENR429 (theoretical [M+H]+ = 634.63), 429DSDSQDSSR437 (theoretical [M+H]+ = 996.92) (Fig. 3). Of these, two peptide clusters, 349DSESSEEDR357 and 424SEENRDSDSQDSSR437 at the C-terminal end of DMP1 demonstrated high binding affinity with immobilized collagen. No peptides were found to be retained because of nonspecific affinity to the Sepharose 4B column. To further confirm this data, mutations were introduced in the putative collagen binding clusters by substitution with neutral sequences as shown in Fig. 4A, and three mutated proteins were expressed as recombinant proteins. Two of these recombinant proteins contained 1053.97), individually mutated sites (D1-col-M1, D1-col-M2), and the other recombinant protein contained mutations in both of the collagen binding peptides (D1-col-M3). These recombinant proteins were then employed for solid-phase binding assays, and their collagen-binding efficiency was examined. Compared with native DMP1, D1-col-M1 and D1-col-M2 demonstrated collagen binding properties, but their maximum binding amount (Bmax) was 25 ± 3.2 nmol/5 μg of collagen for D1-col-M1 and 20 ± 2.5 nmol for D1-col-M2. These values were significantly lower than that of native DMP1 (68 ± 12.6 nmol). No collagen binding affinity was seen with D1-col-M3 (Fig. 4B). These results further confirmed that both 349DSESSEEDR357 and 424SEENRDSDSQDSSR437 were responsible for binding collagen monomers.
      Figure thumbnail gr3
      Fig. 3Peptide mapping of C-DMP1. C-DMP1 was trypsinized and passed through a collagen-immobilized column. MALDI-TOF was then used for mapping peptides of total trypsinized mixture (A) and peptides eluted out of the collagen-immobilized column (B). Three peptides were found to bind to collagen (m/z = 634.35, 996.53, and 1053.52).
      Figure thumbnail gr4
      Fig. 4Characterization of the collagen binding sites in DMP1. A, protein constructs were designed with putative collagen binding sites mutated by site-directed mutagenesis as described under “Materials and Methods.” B, solid-state binding assay of the mutated rDMP1s with Type I collagen. The binding profiles demonstrated that both clusters impaired the effect of collagen binding.
      DMP1 Accelerates Type I Collagen Fibrillogenesis and Increases the Fibril Diameter—To determine the role of DMP1 during collagen fibrillogenesis, fibril formation was initiated by the cold-start procedure. In a typical collagen fibrillogenesis experiment, the turbidity plot is characterized by a sigmoidal curve, constituted by a lag phase (defined as the intercept of the line of maximum slope with the horizontal axis) during which the hydrodynamic parameters of the monomers do not appear to change, followed by a rapid fibril growth phase and finally a plateau region (
      • Na G.C.
      • Butz L.J.
      • Carroll R.J.
      ). Under the present experimental conditions, the kinetics of fibril assembly demonstrated that in the absence of DMP1, the lag period appeared longer and was ∼5 min. However, the lag period dropped to 3 and 2 min when 1 and 5 μm DMP1 was added in separate turbidity experiments (Fig. 5). The slopes of the growth phase and the final turbidity increased with the presence of DMP1. These changes were found to be concentration-dependent, with progressively accelerated fibril formation in the presence of increasing amounts of DMP1. AFM analyses have been successful in visualizing biomolecules at nanoscale and possess the advantage of maintaining biological molecules and their interactions at near physiological conditions (
      • Agarwal G.
      • Kovac L.
      • Radziejewski C.
      • Samuelsson S.J.
      ,
      • Paige M.F
      • Rainey J.K.
      • Goh M.C.
      ). When analyzed by AFM, fibrils reconstituted in the presence of DMP1 appeared to be well formed with clear D periodic banding patterns (Fig. 6). However, using the software analysis all of the reconstituted fibrils had a bandwidth of ∼67 nm, irrespective of the presence of DMP1. The increase in the turbidity values suggested that the diameter of the constituted fibrils were larger in the presence of DMP1. In the absence of DMP1 the diameter of the fibrils formed was measured at 150 ± 15 nm. However, when 1 μm DMP1 was added to 0.1 mg/ml collagen solution, the average diameter of the final fibrils was 302 ± 38 nm, with a strong tendency toward aggregating and forming bigger fibrils (Fig. 6). Thus these results demonstrate that the addition of DMP1 markedly accelerates the kinetics of collagen fibril assembly.
      Figure thumbnail gr5
      Fig. 5Turbidity time curves showing the influence of DMP1 on collagen fibrillogenesis. Collagen fibril assembly was initiated by the cold-start method, and the change in turbidity was monitored spectrophotometrically at 313 nm as a function of time at 26 °C. Solutions containing 0.1 mg/ml type I collagen without DMP1 (—) and with 1 μm DMP1 (····) and 5 μm DMP1 (–––) were set up as described under “Materials and Methods.” DMP1 was shown to accelerate collagen fibrillogenesis and increase fibril diameter in a dose-dependent manner.
      Figure thumbnail gr6
      Fig. 6Determination of the diameter of the collagen fibrils by AFM. Reconstituted collagen fibrils prepared without DMP1 (A) and with 1 μm DMP1 (B) were adsorbed onto a freshly peeled mica surface and imaged by tapping mode AFM. Multiple images were obtained, and the average diameter and bandwidth of the fibrils were analyzed by off-line section analyses. The diameter of reconstituted collagen fibrils in the absence of DMP1 was found to be 150 ± 15 nm and in the presence of 1 μm DMP1 was 302 ± 38 nm.
      DMP1 Is Located in the Gap Region of the Reconstituted Collagen Fibril Surface—The localization of DMP1 on collagen fibrils was investigated by colloidal gold labeling as this is the classical method used to characterize collagen-NCP interactions (
      • Fujisawa R.
      • Nodasaka Y.
      • Kuboki Y.
      ). TEM analysis of DMP1-gold complex incubated on grids pre-coated with reconstituted collagen fibrils and negatively stained confirmed that DMP1-gold complex was detected at the edge of the gap region (Fig. 7, A–D). Of the 138 gold particles examined, ∼95% was located in the gap region. These data corroborate well with results showing specific binding of DMP1 to the N-telopeptide, as this region is located at the edge of the gap region. As a negative control, gold-colloid particles with BSA were found to be randomly distributed and did not associate with type I collagen fibrils (Fig. 7E).
      Figure thumbnail gr7
      Fig. 7Binding of DMP1 on the collagen fibril surface as determined by gold colloid-TEM. For TEM analysis, carbon-formvar-coated nickel grids pre-coated with reconstituted collagen fibril suspension were suspended on a drop of 4 μg/ml colloidal gold-DMP1 solution overnight at 4 °C. The grids were then washed with water and negatively stained with 1% uranyl acetate for 1 min. The samples were examined with an electron microscope (JEOL 1200 EX) operating at 120 kV. DMP1-gold particles (indicated by arrowheads) were localized at the edge of the gap region (A–D). As a negative control, BSA-gold particles were randomly distributed on the grid surface and did not associate with collagen fibrils (E). Scale bar = 200 nm.
      In Vitro Mineralization—In vitro mineralization experiments were conducted with our newly established system (
      • He G.
      • Dahl T.
      • Veis A.
      • George A.
      ). The advantage of this in vitro system is that it maximizes specific mineral depositions during the prolonged experimental period. In the current study, nickel grids bound with collagen fibrils, with and without adsorbed DMP1, were inserted into the system and subjected to mineralization for 5 days. The grids were then subjected directly to TEM analyses. Results demonstrate that no mineral deposition was found on the control collagen surface (Fig. 8, A–D), confirming that the main function of collagen in mineralized tissues is to act as a scaffold. On the contrary, numerous mineral particles were found on collagen surface adsorbed with DMP1 (Fig. 8, E–H). This result strongly suggests that DMP1 can temporally and spatially initiate mineralization when immobilized on the surface of the collagen fibril.
      Figure thumbnail gr8
      Fig. 8TEM images of collagen fibril surface with and without DMP1 after in vitro mineralization assay. The in vitro apatite nucleation assay was conducted on collagen-DMP1-coated grids that were inserted into a channel connecting two halves of an electrolytic cell, one compartment containing calcium buffer (165 mm NaCl, 10 mm HEPES, 2.5 mm CaCl2, pH 7.4) and the other phosphate buffer (165 mm NaCl, 10 mm HEPES, 1 mm KH2PO4, pH 7.4) as described previously (
      • He G.
      • Dahl T.
      • Veis A.
      • George A.
      ). An external 1-mA current was applied, and the buffers were changed regularly to maintain a constant ionic environment. The nucleation experiment was conducted for 5 days. Grids coated with collagen fibrils served as control. Electron-dense calcium phosphate mineral particles (indicated by arrowheads) were associated with collagen fibrils in the presence of DMP1 (E–H), whereas no mineral particles were observed in the control (A–D). Scale bar = 200 nm.

      DISCUSSION

      Type I collagen is the most abundant extracellular matrix protein in bone and dentin (
      • Gehron-Robey P.
      ). It is expressed by osteoblastic and odontoblastic cells at all stages during development and throughout life, constituting the fundamental framework that supports cellular proliferation, migration, and mineralization (
      • Sandberg M.
      • Autio-Harmainen H.
      • Vuorio E.
      ). Preosteoblastic cells seeded on type I collagen matrix have been shown to differentiate into mature osteoblasts with induction of osteoblast-specific gene expression such as osteocalcin, bone sialoprotein, and osteopontin (
      • Mizuno M.
      • Kuboki Y.
      ).
      The acidic noncollagenous proteins identified in the mineralized matrix of bone and dentin (
      • Hoshi K.
      • Ejiri S.
      • Ozawa H.
      ) are known to function either as promoters or inhibitors of mineral deposition in in vitro experiments (
      • Hunter G.K.
      • Hauschka P.V.
      • Poole A.R.
      • Rosenberg L.C.
      • Goldberg H.A.
      ). Several of these NCPs have been shown to possess a collagen binding property. These NCPs include bone sialoprotein, osteopontin, phosphophoryn, biglycan, decorin, and tyrosine-rich acidic matrix protein (
      • Chen Y.
      • Bal B.S.
      • Gorski J.P.
      ,
      • Dahl T.
      • Sabsay B.
      • Veis A.
      ,
      • Traub W.
      • Jodaikin A.
      • Arad T.
      • Veis A.
      • Sabsay B.
      ,
      • Fujisawa R.
      • Nodasaka Y.
      • Kuboki Y.
      ,
      • MacBeath J.R.
      • Shackleton D.R.
      • Hulmes D.J.
      ,
      • Schonherr E.
      • Hausser H.
      • Beavan L.
      • Kresse H.
      ,
      • Schonherr E.
      • Witsch-Prehm P.
      • Harrach B.
      • Robenek H.
      • Rauterberg J.
      • Kresse H.
      ,
      • Neame P.J.
      • Kay C.J.
      • McQuillan D.J.
      • Beales M.P.
      • Hassell J.R.
      ,
      • Fujisawa R.
      • Kuboki Y.
      ). However, the role of NCPs in collagen mediated mineralization is not well understood.
      The identification of DMP1 in the extracellular matrix of bone and dentin led us to examine whether specific interactions with collagen could be demonstrated as DMP1 and type I collagen are expressed during the early stages of mineralization. Recently we reported that DMP1 functions as an initiator for apatite nucleation in vitro and generated calcium phosphate deposits with a morphology that mimics the crystals found at the mineralization front of bone and dentin (
      • He G.
      • Dahl T.
      • Veis A.
      • George A.
      ). Therefore, this protein can be considered as a model protein for studying the chemistry and biology of mineralization processes. As collagen matrix forms the basic architecture for mineral deposition, it is of utmost importance to study the interactions between collagen and DMP1 and the role of this complex in initiating mineralization.
      In this manuscript we demonstrate by a variety of techniques that DMP1 interacts specifically with type I collagen with a KD of 3.8 ± 1.7 μm. Two peptide clusters at the C-terminal end of DMP1 were identified as collagen-interactive peptides by peptide mapping. Site-directed mutagenesis of the 349DSESSEEDR357 and 424SEENRDSDSQDSSR437 domains with neutral sequences confirmed that the maximum binding amount was significantly decreased (Fig. 4). Although several acidic noncollagenous proteins have been identified with collagen binding properties, there are no reports thus far on the peptide domains that are responsible for interaction with collagen. The two identified collagen binding domains in DMP1 contain clusters of acidic residues, which might suggest that the binding between DMP1 and collagen are driven by electrostatic interactions. Recently, we also characterized the sequence in DMP1 responsible for mineral induction (
      • He G.
      • Dahl T.
      • Veis A.
      • George A.
      ). These peptides are located between the two collagen binding sites. Therefore, we suggest that binding of DMP1 on the collagen substrate might aid DMP1 to adopt an appropriate conformation conducive for mineral induction.
      Collagen fibril formation requires activation of the collagen molecules. It is known that fibrillogenesis includes two phases, the lag phase and the fibril growth phase. The lag phase is the time required for development of a nucleus capable of participation in fibrillogenesis via linear polymerization (
      • Na G.C.
      • Butz L.J.
      • Carroll R.J.
      ). According to the fibrillogenesis assay, DMP1 was shown to accelerate collagen fibrillogenesis with a shorter lag time, which suggests that DMP1 could facilitate the linear assembly of collagen fibrils. The addition of DMP1 was found to result in fibrils with larger diameters than the control (Fig. 5). Extensive studies have shown that several proteoglycans, glycosaminoglycans, and glycoproteins can modulate the collagen fibrillogenesis process (
      • MacBeath J.R.
      • Shackleton D.R.
      • Hulmes D.J.
      ,
      • Schonherr E.
      • Hausser H.
      • Beavan L.
      • Kresse H.
      ,
      • Schonherr E.
      • Witsch-Prehm P.
      • Harrach B.
      • Robenek H.
      • Rauterberg J.
      • Kresse H.
      ,
      • Neame P.J.
      • Kay C.J.
      • McQuillan D.J.
      • Beales M.P.
      • Hassell J.R.
      ). With reference to NCPs found in the mineralized matrix of bone and dentin, bone sialoprotein was found to accelerate type I collagen fibrillogenesis (
      • Fujisawa R.
      • Kuboki Y.
      ), and phosphophoryn could retard collagen fibrillogenesis (
      • Cocking-Johnson D.
      • Sauk J.J.
      ) but increase the diameter of the reconstituted collagen fibrils (
      • Clarkson B.H.
      • McCurdy S.P.
      • Gaz D.
      • Hand A.R.
      ). Thus noncollagenous proteins might function as non-covalent cross-linkers, bridging collagen fibrils and forming an ordered template, which might be necessary to mediate controlled mineral deposition (
      • Dahl T.
      • Veis A.
      ).
      The N-telopeptide region of the collagen fibril was identified as the DMP1 binding site. The N-telopeptide of the α1(I) chain consists of 16 amino acids, which have been highly conserved in a variety of species. The self-assembly of collagen fibrils requires the short non-helical telopeptides to allow correct molecular registration and cross-link formation, thus ensuring proper packing in the D-staggered array (
      • Orgel J.P.
      • Miller A.
      • Irving T.C.
      • Fischetti R.F.
      • Hammersley A.P.
      • Wess T.J.
      ,
      • Veis A.
      • George A.
      ). Neutron scattering studies of calcifying collagen in turkey leg tendons have demonstrated an initial nucleation site located toward the N terminus of the gap region, which led to the axial propagation of the mineral deposit along the gap region (
      • White S.W.
      • Hulmes D.J.
      • Miller A.
      • Timmins P.A.
      ,
      • Berthet-Colominas C.
      • Miller A.
      • White S.W.
      ). X-ray scattering and electronic microscope tomography studies have also shown that apatite initiates and elongates from the gap region in the collagen fibrils (
      • Traub W.
      • Arad T.
      • Weiner S.
      ,
      • Traub W.
      • Arad T.
      • Weiner S.
      ,
      • Camacho N.P.
      • Rinnerthaler S.
      • Paschalis E.P.
      • Mendelsohn R.
      • Boskey A.L.
      • Fratzl P.
      ). TEM analyses confirmed the localization of DMP1 on reconstituted collagen fibrils. Specifically, DMP1 bound to the edge of the gap region of the collagen fibrils, where the N-telopeptide is located. Therefore, it is possible that DMP1 binds to the N-telopeptide of the collagen fibril and thereby facilitate collagen polymerization and mineral nucleation during calcified tissue formation.
      Binding of DMP1 on the collagen fibril raises an important question regarding its functional role in the mineralization process. When DMP1-collagen complex was incubated in a pseudo-physiological buffer and tested for its mineralization property in vitro, it was observed that only DMP1-collagen complex initiated biomineral deposition on the collagen surface, whereas there was no calcium phosphate deposition on the control collagen surface. Similar results have been achieved when the collagen surface was immobilized with bone sialoprotein-decorin chimeric protein (
      • Hunter G.K.
      • Poitras M.S.
      • Underhill T.M.
      • Grynpas M.D.
      • Goldberg H.A.
      ) and phosphophoryn (
      • Saito T.
      • Arsenault A.L.
      • Yamauchi M.
      • Kuboki Y.
      • Crenshaw M.A.
      ). In all of these cases, mineralization was initiated specifically on the collagen surface adsorbed with noncollagenous proteins. These results further emphasize the specific role of noncollagenous proteins during the collagen-mediated mineralization process.
      In conclusion, the present study suggests that the cooperative interaction between DMP1 and collagen matrix could be an essential step in the biomineralization process in matrix-mediated mineralization. Understanding the principles involved in mineral-matrix interactions could be useful in the biomimetic design of composite materials with mechanical properties emphasizing both strength and toughness.

      References

        • Weiner S.
        • Wagner H.D.
        Annu. Rev. Mater. Sci. 1998; 28: 271-298
        • Traub W.
        • Arad T.
        • Weiner S.
        Matrix. 1992; 12: 251-255
        • Jager I.
        • Fratzl P.
        Biophys. J. 2000; 79: 1737-1746
        • Saito T.
        • Arsenault A.L.
        • Yamauchi M.
        • Kuboki Y.
        • Crenshaw M.A.
        Bone. 1997; 21: 305-311
        • Bradt J.
        • Mertig M.
        • Teresiak A.
        • Pompe W.
        Chem. Mater. 1999; 11: 2694-2701
        • Boskey A.L.
        Connect. Tissue Res. 1996; 35: 357-363
        • Butler W.T.
        • Ritchie H.
        Int. J. Dev. Biol. 1995; 39: 169-179
        • Chen Y.
        • Bal B.S.
        • Gorski J.P.
        J. Biol. Chem. 1992; 267: 24871-24878
        • Stetler-Stevenson W.G.
        • Veis A.
        Calcif. Tissue Int. 1987; 40: 97-102
        • Wallwork M.L.
        • Kirkham J.
        • Chen H.
        • Chang S.X.
        • Robinson C.
        • Smith D.A.
        • Clarkson B.H.
        Calcif. Tissue Int. 2002; 71: 249-255
        • Hunter G.K.
        • Hauschka P.V.
        • Poole A.R.
        • Rosenberg L.C.
        • Goldberg H.A.
        Biochem. J. 1996; 317: 59-64
        • Xiao S.
        • Yu C.
        • Chou X.
        • Yuan W.
        • Wang Y.
        • Bu L.
        • Fu G.
        • Qian M.
        • Yang J.
        • Shi Y.
        • Hu L.
        • Han B.
        • Wang Z.
        • Huang W.
        • Liu J.
        • Chen Z.
        • Zhao G.
        • Kong X.
        Nat. Genet. 2001; 27: 201-204
        • Sreenath T.
        • Thyagarajan T.
        • Hall B.
        • Longenecker G.
        • D'Souza R.
        • Hong S.
        • Wright J.T.
        • MacDougall M.
        • Sauk J.
        • Kulkarni A.B.
        J. Biol. Chem. 2003; 278: 24874-24880
        • Dahl T.
        • Sabsay B.
        • Veis A.
        J. Struct. Biol. 1998; 123: 162-168
        • Traub W.
        • Jodaikin A.
        • Arad T.
        • Veis A.
        • Sabsay B.
        Matrix. 1992; 12: 197-201
        • Fujisawa R.
        • Nodasaka Y.
        • Kuboki Y.
        Calcif. Tissue Int. 1995; 56: 140-144
        • George A.
        • Sabsay B.
        • Simonian P.A.
        • Veis A.
        J. Biol. Chem. 1993; 268: 12624-12630
        • Qin C.
        • Brunn J.C.
        • Jones J.
        • George A.
        • Ramachandran A.
        • Gorski J.P.
        • Butler W.T.
        Eur. J. Oral Sci. 2001; 109: 133-141
        • Kamiya N.
        • Takagi M.
        Histochem. J. 2001; 33: 545-552
        • George A.
        • Silberstein R.
        • Veis A.
        Connect. Tissue Res. 1995; 33: 67-72
        • He G.
        • Dahl T.
        • Veis A.
        • George A.
        Connect. Tissue Res. 2003; 44 (suppl.): 240-245
        • Payne K.J.
        • Veis A.
        Biopolymers. 1988; 27: 1749-1760
        • Dickeson S.K.
        • Mathis N.L.
        • Rahman M.
        • Bergelson J.M.
        • Santoro S.A.
        J. Biol. Chem. 1999; 274: 32182-32191
        • Srinivasan R.
        • Chen B.
        • Gorski J.P.
        • George A.
        Connect. Tissue Res. 1999; 40: 251-258
        • George A.
        • Malone J.P.
        • Veis A.
        Proc. Indian Acad. Sci. (Chem. Sci.). 1999; 111: 121-131
        • George A.
        • Veis A.
        Biochemistry. 1991; 30: 2372-2377
        • Payne K.J.
        • King T.A.
        • Holmes D.F.
        Biopolymers. 1986; 25: 1185-1207
        • Slot J.W.
        • Geuze H.J.
        Eur. J. Cell Biol. 1985; 38: 87-93
        • Na G.C.
        • Butz L.J.
        • Carroll R.J.
        J. Biol. Chem. 1986; 261: 12290-12299
        • Agarwal G.
        • Kovac L.
        • Radziejewski C.
        • Samuelsson S.J.
        Biochemistry. 2002; 41: 11091-11098
        • Paige M.F
        • Rainey J.K.
        • Goh M.C.
        Biophys. J. 1998; 74: 3211-3216
        • Gehron-Robey P.
        Principles of Bone Biology. Academic Press, San Diego, CA1996: 155-166
        • Sandberg M.
        • Autio-Harmainen H.
        • Vuorio E.
        Dev. Biol. 1988; 130: 324-334
        • Mizuno M.
        • Kuboki Y.
        J. Biochem. 2001; 129: 133-138
        • Hoshi K.
        • Ejiri S.
        • Ozawa H.
        J. Electron Microsc. 2001; 50: 33-40
        • MacBeath J.R.
        • Shackleton D.R.
        • Hulmes D.J.
        J. Biol. Chem. 1993; 268: 19826-19832
        • Schonherr E.
        • Hausser H.
        • Beavan L.
        • Kresse H.
        J. Biol. Chem. 1995; 270: 8877-8883
        • Schonherr E.
        • Witsch-Prehm P.
        • Harrach B.
        • Robenek H.
        • Rauterberg J.
        • Kresse H.
        J. Biol. Chem. 1995; 270: 2776-2783
        • Neame P.J.
        • Kay C.J.
        • McQuillan D.J.
        • Beales M.P.
        • Hassell J.R.
        Cell Mol. Life Sci. 2000; 57: 859-863
        • Fujisawa R.
        • Kuboki Y.
        Calcif. Tissue Int. 1992; 51: 438-442
        • He G.
        • Dahl T.
        • Veis A.
        • George A.
        Nat. Mater. 2003; 2: 552-558
        • Cocking-Johnson D.
        • Sauk J.J.
        Biochim. Biophys. Acta. 1983; 742: 49-53
        • Clarkson B.H.
        • McCurdy S.P.
        • Gaz D.
        • Hand A.R.
        Arch. Oral Biol. 1993; 38: 737-743
        • Dahl T.
        • Veis A.
        Connect. Tissue Res. 2003; 44: 206-213
        • Orgel J.P.
        • Miller A.
        • Irving T.C.
        • Fischetti R.F.
        • Hammersley A.P.
        • Wess T.J.
        Structure. 2001; 9: 1061-1069
        • Veis A.
        • George A.
        Mecham R.P. Extracellular Matrix Assembly and Structure. Academic Press, San Diego1994: 15-45
        • White S.W.
        • Hulmes D.J.
        • Miller A.
        • Timmins P.A.
        Nature. 1977; 266: 421-425
        • Berthet-Colominas C.
        • Miller A.
        • White S.W.
        J. Mol. Biol. 1979; 134: 431-445
        • Traub W.
        • Arad T.
        • Weiner S.
        Proc. Natl. Acad. Sci. U. S. A. 1989; 86: 9822-9826
        • Camacho N.P.
        • Rinnerthaler S.
        • Paschalis E.P.
        • Mendelsohn R.
        • Boskey A.L.
        • Fratzl P.
        Bone. 1999; 25: 287-293
        • Hunter G.K.
        • Poitras M.S.
        • Underhill T.M.
        • Grynpas M.D.
        • Goldberg H.A.
        J. Biomed. Mater. Res. 2001; 55: 496-502