Formation of Framework Nacre Polypeptide Supramolecular

ARTICLE pubs.acs.org/Biomac

Formation of Framework Nacre Polypeptide Supramolecular Assemblies That Nucleate Polymorphs Fairland F. Amos, Christopher B. Ponce, and John Spencer Evans* Laboratory for Chemical Physics, New York University, 345 E. 24th Street, New York, New York 10010, United States

bS Supporting Information ABSTRACT: The formation of aragonite in the mollusk shell nacre layer is linked to the assembly of framework protein complexes that interact with β-chitin polysaccharide. What is not yet understood is how framework nacre proteins control crystal growth. Recently, a 30 AA intrinsically disordered nacre protein sequence (n16N) derived from the n16 framework nacre protein was found to form aragonite, vaterite, or ACC deposits when adsorbed onto β-chitin. Our present study now establishes that n16N assembles to form amorphous nonmineralized supramolecular complexes that nucleate calcium carbonate polymorphs in vitro. These complexes contain unfolded or disordered (54% random coil, 46% β structures) n16N polypeptide chains that self-assemble in response to alkaline pH shift. The pH-dependent assembly process involves two stages, and it is likely that side chain salt-bridging interactions are a major driving force in n16N self-association. Intriguingly, Ca(II) ions are not required for n16N assembly but do shift the assembly process to higher pH values, and it is likely that Ca(II) plays some role in stabilizing the monomeric form of n16N. Using preassembled fibril
spheroid n16N assemblies on Si wafers or polystyrene supports, we were able to preferentially nucleate vaterite at higher incidence compared to control scenarios, and it is clear that the n16N assemblies are in contact with the nucleating crystals. We conclude that the framework nacre protein sequence n16N assembles to form supramolecular complexes whose surfaces act as nucleation sites for crystal growth. This may represent a general mineralization mechanism employed by framework nacre proteins in general.

I

n some mollusks, such as the Japanese pearl oyster, Pinctada fucata, the mineralized shell is composed of two opposing layers of calcium carbonate known as the nacre (aragonite) and prismatic (calcite) layers.1,2 The material properties of each shell layer are different (nacre is fracture-resistant; prismatic is puncture-resistant), and these properties arise in part from the presence of biomacromolecule assemblies in each shell layer.3
5 With regard to the nacre layer of P. fucata, the three major biomacromolecular components that are in close proximity to the aragonite mineral phase are β-chitin polysaccharide,1,2,4,6 a silk-like fibroin protein that exists as a hydrogel,1,2,4 and an assemblage of intracrystalline and framework proteins.1,2,4
11 In vitro experiments have demonstrated that this multicomponent biomacromolecular system nucleates aragonite crystals.12 This result suggests that protein-based assemblies are responsible for polymorph selection. If true, then one key to understanding the polymorph selection process is to learn how nacre components assemble into a functional nucleation framework. In achieving this goal, the scientific community will possess the necessary information that will enable the development of biomimetic approaches for manipulating crystal growth, designing supramolecular assemblies, and constructing biocomposites in the laboratory. Given the complexity of the nacre organic matrix, reductionist approaches have been utilized to examine the assembly and r 2011 American Chemical Society

nucleation properties of individual intracrystalline and framework nacre proteins. For example, recent studies conducted with intracrystalline aragonite protein 7 (AP7)13 and P. fucata mantle gene product 1 (PFMG1) terminal sequences14 have demonstrated that these sequences nucleate aragonite within selfassembled supramolecular complexes. In contrast to this is the n16 framework protein (P. fucata), which associates with β-chitin polysaccharide and other nacre proteins to form a biomacromolecular complex that nucleates aragonite on the surface of the complex.3,5 Hence, compared to the intracrystalline proteins, the framework proteins utilize a slightly different strategy for aragonite formation, and understanding the functional differences between both sets of proteins will help us understand the nacre formation process. To understand the mechanism of protein
β-chitin aragonite nucleation, it becomes necessary to pursue investigations of framework proteins such as n16. However, since the n16 protein is not yet available in sufficient quantities for experimental studies, alternatives such as the use of peptide mimics of n16 sequence regions10 were developed to understand n16 framework-based aragonite nucleation. One of these mimics, a 30 AA Received: February 18, 2011 Revised: April 7, 2011 Published: April 07, 2011 1883

dx.doi.org/10.1021/bm200231c | Biomacromolecules 2011, 12, 1883–1890

Biomacromolecules intrinsically disordered polypeptide (IDP) derived from the N-terminus of n16 (n16N),6,10,11,15,16 infrequently forms porous polypeptide layers that nucleate aragonite in a lamellar inorganic
organic structure similar to that found in nacre itself.15 This n16N sequence binds to β-chitin and promotes aragonite and vaterite on β-chitin surfaces, and vaterite and amorphous calcium carbonate (ACC) on β-chitin surfaces in the presence of silkfibroin gel.6,16 For these reasons, the n16N sequence represents both an important domain of the n16 framework protein and an excellent model for understanding the assembly and nucleation properties of framework nacre protein sequences. This present study identifies the self-association and nucleation capabilities of the intrinsically disordered n16N framework sequence. Using in vitro conditions similar to those described in previous n16N mineralization studies,6,10,11,15,16,20,21 we observe that n16N self-assembles to form amorphous fibril
spheroidal supramolecular complexes in solution. These complexes can further associate into higher-ordered structures that resemble sheets or films and this coincides with reported observations of n16N “layers” forming within in vitro aragonite-n16N lamellar assemblies15 or on β-chitin surfaces.6,16 These n16N assemblies are nonmineralized, contain unfolded n16N molecules, and form in response to an alkaline pH shift that accompanies the mineralization process. This assembly process is modulated by but not dependent upon Ca(II) ions. Using mineralization experiments, we demonstrate that vaterite and calcite nucleate on the surfaces of preassembled fibril
spheroidal n16N complexes, and this surface nucleation feature is consistent with nacre framework protein function.

’ EXPERIMENTAL SECTION Polypeptide Synthesis, Purification, and Sample Preparation. The 30-mer polypeptide n16N (AYHKKCGRYSYCWIPYDIERDRYDNGDKKC), representing the 1
30 AA N-terminal domain of n16, was synthesized at the 100 μmol synthesis level using the protocol described in our earlier work.6,10,11,15,16 This peptide featured free amino termini and CR amide “capping” to mimic its attachment to the protein and negate the charge contribution from the R-carboxylate C-terminus.10,11 For subsequent studies, stock solutions of n16N were created by dissolving the lyophilized peptide into H2/high purity N2-flushed unbuffered deionized distilled water (UDDW), with subsequent heating of the sealed stock solution at 90 °C for 30 min, followed by slow bench cooling of these solutions at room temperature.11 The inclusion of reducing agents such as dithiothreitol (DTT) was omitted from these stock solutions, since previous studies indicated that reducing agents inhibit the n16N aggregation process.11 Stock solution were stored in airtight sealed vials at
20 °C until needed. n16N Single-Stage Mineralization Studies. Using the same assay conditions employed in n16N aragonite formation studies,6,15,16 we monitored the formation of n16N complexes in solution. These mineralization assays are single-stage and use solid (NH4)2CO3 that vaporizes over time and yields carbon dioxide gas that subsequently dissolves in the assay solution. Excess dissolution of CO2 increases the pH of the solution while producing CO32
.6,10,16 The initial pH of the assay solutions containing n16N peptide and 12.5 mM CaCl2 was found to be 3.3 and reached a value of 8.0
8.3 at the conclusion of the assay.20 These assay solutions contained either no peptide (negative control) or final assay concentrations of 10, 50, and 100 μM n16N as employed in previous mineralization studies.10,11,16 The collection procedure for assay precipitates involved the use of Si wafer fragments (1 cm2 or less in size, “P” type [1 0 0], 20 Ω-cm, 250
350 μm, Silicon Quest Intl., Santa Clara, CA) that were placed

ARTICLE

shiny side up at the bottom of assay wells prior to the start of the assay.14,21 Si wafer fragments were also utilized to capture floating polypeptide films that were observed at the air/water interface. The rescued assay materials were washed with UDDW and then dried at 37 °C overnight. Negative control samples were washed with 50% ethanol/50% UDDW and then dried at 37 °C for the same period of time. Dried samples were coated with a thin layer of carbon or gold and then imaged using Hitachi S-3500N or Zeiss EVO 50 XVP scanning electron microscopes at an accelerating voltage of 20 kV. Energy dispersive spectra (EDS) were collected on carbon-coated samples using the Zeiss SEM equipped with a PGT-Bruker X-ray microanalysis system at 20 kV. For atomic force microscopy, samples floating at the air
water interface were scooped using a Si wafer, washed with UDDW and then neat methanol, and dried overnight at 37 °C. These samples were then analyzed with Si3N4 tips using a Veeco Multimode Nanoscope III atomic force microscope (AFM) in contact mode at 25 °C, and the raw data were visualized using WSxM v2.2 software.

Dynamic Light Scattering Studies of n16N Oligomerization. The oligomerization of n16N was determined by measuring the hydrodynamic radius under different pH conditions (pH 4.0
9.0) and in the presence of Ca(II) (12.5 mM CaCl2) using a Wyatt DynaPro MS/X Dynamic Light Scattering (DLS) instrument. Sodium acetate, Tris-HCl, and sodium carbonate/sodium bicarbonate buffers (all 10 mM) were used to create pH ranges of 4
6, 7
8.5, and 9.0
11.3, respectively. Stock solutions of CaCl2 (99.9%, Sigma/Aldrich) were utilized to create final Ca(II) concentrations of 12.5 mM for each 100 μM peptide sample at each investigated pH point, with the exception of the carbonate/bicarbonate buffer, which would induce unwanted calcium carbonate precipitation. Prior to DLS measurements all buffer/peptide samples were prefiltered using 0.22 μm polyvinylidene fluoride syringe filters (Fisher Scientific) and then placed into quartz curvettes. All samples were equilibrated at 16 °C for 10 min in the quartz curvette prior to acquisition. This temperature corresponds to the same mineralization assay temperature utilized in the present study and in previous n16N mineralization studies.10,11,16 For each sample ten acquisitions were taken at 16 °C. Analysis of the data and determination of hydrodynamic radius (RH)22,23 was performed using the regularization analysis in the Dynamics v6.0 software provided with the instrument. By measuring the fluctuations in the laser light intensity scattered by the sample, the instrument is able to detect the speed (diffusion coefficient) at which the particles are moving through the medium. This value is converted to hydrodynamic radius (RH) using the Stokes
Einstein relation:22,23 D¼

kT 6πηRH

where D is the is the diffusion coefficient, k is the Boltzmann constant, T is the absolute temperature, η is the viscosity, and RH is the sphereequivalent hydrodynamic radius.23 Structural Studies of n16N Assemblies. Circular dichroism (CD) spectrometry was utilized to determine the secondary structures present in 100 μM n16N samples at pH 4, 6, and 7.4, using the same buffers and temperature as per the DLS study. For each sample, CD spectra were taken from the average of 8 scans, with a scan rate of 0.5 nm/s from 185
260 nm, on an AVIV Stopped Flow 202SF CD Spectropolarimeter. Spectra were obtained with appropriate background buffer subtraction performed, and averaged spectra were smoothened using the Savitzky
Golay algorithm. Ellipticity is reported as mean residue ellipticity, θM (deg cm2 dmol
1). Secondary structure estimation of n16N was performed using a constrained least-squares fit of the spectra on a five-component model (R helix, β sheet, β Type I, Type II, and random coil).24 The resulting fit was performed using IGOR Pro 6.0 and is reported as the fractional weight plus or minus the 1884

dx.doi.org/10.1021/bm200231c |Biomacromolecules 2011, 12, 1883–1890

Biomacromolecules standard deviation of the five reference components. The β sheet, β turn Type I, Type II are combined under the label “β structures.” n16N Two-Stage Mineralization Experiments. To determine if preformed n16N complexes can control calcium carbonate nucleation, we implemented a variation on the standard single-stage mineralization assays described in the preceding section. We will refer to this experiment as the two-stage mineralization scheme (Figure S1, Supporting Information). During Stage 1, standard mineralization assay conditions are used to generate n16N fibril
spheroidal deposits onto two kinds of supports: Si wafers or polystyrene assay well fragments (50 μM n16N, 16 h, 16 °C).10,11,16,21 These deposits are washed with UDDW and then dried at 37 °C overnight. In Stage 2, we transfer the washed supports to new mineralization solutions and initiate a second round of mineralization (with or without 50 μM n16N, 18 h, 16 °C; total assay time = 34 h). At the conclusion of Stage 2 the supports are removed, washed, and dried as previously described for SEM characterization in the preceding section. Like the single-stage mineralization experiment, the initial pH of the two-stage assay solution was found to be 3.3 and reached a value of 8.0
8.3 at the conclusion of the assay. For the two-stage mineralization experiment we developed two control scenarios. Control 1 are supports that were never exposed to n16N during both mineralization stages and thus do not contain n16N deposits. Control 2 are supports that were exposed to n16N during Stage 1 and thus possess n16N deposits, but these deposits are not exposed to additional n16N polypeptide during Stage 2. X-ray Diffraction. Washed and dried Control 1 and n16N experimental supports from double mineralization experiments were analyzed using a Bruker D8 DISCOVER GADDS Microdiffractometer equipped with a VANTEC-2000 area detector in a j rotation method. The X-ray generated from a sealed copper tube is monochromated by a graphite crystal and collimated by a 0.5 mm MONOCAP (λ Cu KR = 1.54178 Å). The sample detector distance is 150 mm. Two runs with θ1 = θ2 = 15° and 30° are collected for each specimen and the exposure time is 600 s per run. Data were merged and integrated by the XRD2EVAL program in the Bruker PILOT Software. The raw file was converted to UXD format by the DIFFRACplusFileExchange, which was later analyzed by the WINPLOTR program. Calcite, aragonite, and vaterite-specific X-ray diffraction data were obtained from the Powder Diffraction File (PDF), a database of X-ray powder diffraction patterns maintained by the International Center for Diffraction Data (ICDD), http://www.icdd.com/.

’ RESULTS AND DISCUSSION Formation of n16N Peptide Supramolecular Assemblies. Under nonreducing conditions n16N induces aragonite and vaterite6 or ACC17 on β-chitin films (10 μM n16N), and in limited instances lamellar polypeptide
aragonite composites are observed without the need for β-chitin films (100 μM n16N).16 Using this range of n16N concentrations we repeated these singlestage mineralization assays and analyzed protein deposits that settled to the bottom of the assay wells. As shown in Figure 1, we find that n16N forms supramolecular assemblies that have an amorphous-appearing architecture. These deposits consist of spherical particles and fibrils, and the presence of these species increases as a function of n16N concentration (Figure 2). The appearance of the fibril
spheroidal assemblies is reminiscent of complexes formed by the sponge-associated R-silicatein biomineralization protein,25 and the spherical regions qualitatively resemble the spherical particles formed by the tooth enamel proteins.26 In some cases these n16N particles assemble to form higherordered structures that resemble sheet- or film-like assemblies that still retain fibril
spheroid characteristics (Figure 1B). At higher peptide concentrations these sheet-like assemblies are

ARTICLE

Figure 1. Scanning electron micrograph of n16N supramolecular complexes captured by Si wafers (10 μM n16N assay concentration, 16 °C, 18 h). This is the same peptide concentration utilized in n16N
β-chitin nucleation studies.6,16 In panel A, note the scattered fibril
spheroidal assemblies. In panel B, higher magnification reveals that the sheet or film consists of fibril
spheroidal components. Inset image represents rhombohedral calcite crystal recovered from assays containing no peptide.

buoyant and are found floating at the air/water interface (Figure 3). These floating polypeptide sheet assemblies contain pores or holes (Figure 3), and although these porous regions may represent artifacts of sample drying, we note that the detection of porous regions was also reported within the intracrystalline n16N peptide layers of lamellar aragonite deposits.16 Using AFM, TEM, and electron diffraction, we confirm that the sheet-like assemblies are amorphous in structure and do not contain crystalline domains (Figure 4), as evidenced by the absence of sharp diffraction rings or spots in the electron diffraction pattern and the absence of organized or patterned fibrils or spheroids within these films. We found that these films or sheets were hard to manipulate due to their adhesion to surfaces and folding and creasing artifacts (Figure 3), and thus further characterization of these complexes was not pursued at this time. EDS measurements confirm that the sheet-like assemblies consist of n16N peptide, as evidenced by the detection of peptide-associated nitrogen (N) and sulfur (S, Cys thiol) peaks (Figure 3). However, due to the low density and small size of the fibril
spheroidal complexes, the detected S content is very low and the N content, which usually gives rise to a weaker signal, is not discernible. Thus, the fibril
spheroidal complexes are most likely composed of n16N polypeptide, but this will require future clarification. Interestingly, both types of deposits possesses 1885

dx.doi.org/10.1021/bm200231c |Biomacromolecules 2011, 12, 1883–1890

Biomacromolecules

ARTICLE

Figure 2. Concentration-dependent formation of n16N fibril
spheroidal complexes captured on Si wafers. n16N concentrations are provided on the images. Scale bars = 50 μm.

Figure 3. Scanning electron micrographs of 50 μM n16N supramolecular complexes captured (A) at the air/water interface or (B) as fibril
spheroidal deposits on Si wafers. Backscatter imaging was utilized on the sheet/film samples to clearly show the folding and creasing of the film. Secondary electron mode imaging mode was used for the deposits. Corresponding EDS analyses are shown below each image. Note that the low density of fibril
spheroidal material gives rise to weaker EDS signals, particularly for S. The Si peak in the EDS spectra represents the Si wafer background signal. The chlorine (Cl) peak arises from CaCl2 in the starting solution that is retained by the film.

Figure 4. Imaging of dried 50 μM n16N film obtained at the air/water interface from mineralization assays. (A) TEM image. Inset image represents electron diffraction pattern taken from this film. Note the absence of diffraction spots. (B) Three-dimensional contact mode AFM image.

no significant levels of Ca, indicating that both assemblies are nonmineralized. However, we cannot rule out the possibility that these complexes may contain lower levels of Ca(II) ions that are not detected by EDS. The absence of mineral deposition within these films and deposits is in contrast to other nacre polypeptide studies where mineral deposits were identified within supramolecular assemblies.13,14

Oligomerization of n16N in Assay Solutions. Using dynamic light scattering, we discovered that the hydrodynamic radius, RH,22,23 of apo-n16N at pH 3.3 is 1.0 nm (polydispersity 40.3%). On the basis of the regularization analysis, this RH value corresponds to a 4-kDa peptide, which is close to the MW value of the n16N monomer (3.7 kDa).10,11 Subsequently, we determined that the RH for n16N at the same pH in the presence of 12.5 mM CaCl2 was slightly higher (1.4 nm, polydispersity 39.7%). The RH and high polydispersity values (>15%) indicate that n16N is heterodisperse at low pH in the presence and absence of Ca ions and most likely exists as a mixture of monomeric and oligomeric species at the start of the mineralization assays. Knowing that the assay pH shifts to 8.0 or higher after the introduction of carbonate vapor, we omitted both CaCl2 and (NH4)2CO3 from the assays and simply adjusted the initial pH of a n16N solution (pH 3.3) to 8.1 and incubated the polypeptide at 16 °C for 16 h (Figure S2, Supporting Information). We found that the formation of n16N fibril
spheroidal and sheet-like complexes are reproduced using a simple pH shift. Hence, there appears to be no real requirement for Ca(II) or CO32
/HCO3
ions to generate n16N supramolecular complexes, only an alkaline pH shift. We note that other polypeptide supramolecular assemblies also form in response to pH.27
33 From these results, we conclude that n16N forms amorphous nonmineralizing fibril
spheroidal polypeptide assemblies in response to pH conditions that accompany carbonate vapor diffusion assays. Mapping the pH-Dependent n16N Oligomerization Process. The fact that alkaline pH can induce n16N self-assembly (Figure S2) prompted us to map out the pH range for n16N self-assembly in solution at 16 °C (Figure 5A). We find that apo-n16N oligomerization is pH-dependent and appears to be a two-step process. The first step occurs from pH 4.0 to 4.5, where the average RH increases from 1.8 to 97 nm, indicating that significant oligomerization is taking place.22,23 From pH 4.5 to 8.5 the polydispersity index is >15%, indicating that the size ranges of these complexes are heterogeneous.22,23 The second step in the oligomerization process occurs at pH 8.5
9.0. This is schematically drawn on the figure as an arrow pointing toward higher RH values. Here, we found that the RH scattering intensity exceeds the maximum detection limit of the instrument. This dramatic increase reflects either a significant increase in the number or the average size of the polypeptide particles in solution, but in either case, the high scattering index at pH g 9 represents a further increase in n16N oligomerization and correlates with the appearance of micrometer-sized supramolecular assemblies in mineralization assays (Figures 1 and 2). We note that n16N contains Asp, Glu, His, Lys, Arg, and Cys residues, and given that other charged polypeptides assemble in response to pH variation,27
33 it is likely that side chain electrostatics plays an important role in n16N oligomerization. 1886

dx.doi.org/10.1021/bm200231c |Biomacromolecules 2011, 12, 1883–1890

Biomacromolecules

Figure 5. (A) Dynamic light scattering analysis of n16N oligomerization (100 μM) as a function of pH in the presence and absence of 12.5 mM CaCl2, 16 °C. Arrow indicates that the RH value exceeds the detection limit of the instrument [pH > 8.5 for apo state, pH > 8.0 for Ca(II) state]. The primary amino acid sequence of n16N is shown above this figure, with red and blue indicating the position of anionic and cationic residues, respectively. (B) pH-dependent CD spectra of apon16N (100 μM) in DLS buffers (see Experimental Section) at 16 °C.

A somewhat different pH-dependent oligomerization process occurs in the presence of Ca(II) (Figure 5A). At pH 4
6, the average RH = 1.7 nm with a polydispersity of 47%, which corresponds to a mixture of monomeric and small oligomeric species of heterogeneous size ranges. Thus, significant oligomerization does not occur at pH < 6 in the presence of Ca(II). This suggests that Ca(II) is somehow stabilizing monomeric and small oligomeric complexes at pH < 6 and preventing further assembly. Interestingly, the first oligomerization step now occurs between pH 6 and 7, where we observe a significant transition point (average RH increases from 1.7 to 110.4 nm) (Figure 5A). The second oligomerization step occurs between pH 8 and 8.5, which represents a
0.5 pH shift from the apo state. These findings were replicated in several DLS measurements between pH 8.0 and 8.5. Because of the potential for formation of calcium carbonate in carbonate/bicarbonate buffer systems, beyond pH 8.5 we were unable to test the effect of Ca(II) ions on n16N oligomerization (i.e., the limit of Tris HCl buffer range). In general, we note that the polydispersity is greater than 15% for RH measurements in the presence of Ca(II), indicating that the forming n16N oligomers are heterogeneous in size. Thus, although Ca(II) is not required for n16N assembly, this cation stabilizes monomeric and small oligomeric n16N complexes at higher pH compared to the apo state. Secondary Structure of n16N within Oligomeric Assemblies. Over different concentration ranges and under reducing

ARTICLE

and nonreducing conditions, n16N exists as an intrinsically disordered polypeptide17,18 consisting of random coil structure in equilibrium with β structures such as extended β strand.10,11 Interestingly, this random or unfolded structure persists in the presence of excess Ca(II) ions, so presumably n16N exists as an unfolded species at the start of the mineralization assays.11 However, it is known that some intrinsically disordered proteins will fold during interprotein interactions with other proteins or during self-association.17,18 If true, then the question is do individual n16N molecules remain disordered within polypeptide assemblies, or does folding occur in response to oligomerization? To address this, we estimated the secondary structure preferences for apo-n16N at monomeric-oligomeric conditions (pH 4) and under conditions known to promote supramolecular complex formation (pH = 6.0, 7.4, 16 °C) (Figure 5B). Here, as we transition from a mixture of monomeric and oligomeric species (pH 4.0) to a heterogeneous, polydisperse species (pH 6, 7.4), we note that there is minimal change in the π
πi* or n
π* transition bands and molar ellipticity intensities. This consistency in both values indicates that the disordered conformation of n16N polypeptide persists in the monomeric and oligomeric states. Estimation of the secondary structure of n16N using the reference data set24 reveals a nearly equal content of intrinsic disorder (54% random coil) and β structures (46% β sheet þ β turn Types I, II), from pH 4 to 7.6, similar to that reported for the low pH, nonaggregated form of n16N.10,11 These results suggest that the n16N molecule retains its unfolded state within polypeptide assemblies and does not undergo detectable folding or secondary structure formation during oligomerization. This finding explains why there is an amorphous structure for these assemblies and the absence of crystalline domains (Figure 4). We note that other intrinsically disordered protein sequences exhibit similar structural tendencies in protein
protein complexes.17,18 Nucleation Activity of Preformed n16N Supramolecular Assemblies. It is believed that nacre framework proteins, such as n16, assist in the nucleation of aragonite after these proteins become incorporated on or within β-chitin
silk-fibroin macromolecular assemblies.3,5,6,16 Thus, we were interested in testing the inherent nucleation activity of n16N assemblies themselves. Initially, we attempted to use n16N sheets or films as a nucleation surface, but due to the creasing, folding, and difficulty in manipulation, we were unsuccessful in obtaining reproducible results with these films. However, we found that fibril
spheroidal n16N deposits adhere to both Si wafers and polystyrene supports after washing and were easier to manipulate. Thus, we reasoned that we could use these washed, n16N-coated supports for an additional round of mineralization (i.e., two-stage or double mineralization experiments, Figure S1, Supporting Information) and determine if preassembled fibril
spheroidal n16N complexes (Figures 1
3) could nucleate calcium carbonate crystals relative to control scenarios. As shown in Figure 6, the results obtained for these double experiments are quite interesting. Silicon wafers obtained from negative control assays (no n16N present during Stage 1 or 2, designated as Control 1) primarily feature rhombohedral calcite and some vaterite (Figure 6A). The presence of calcite and vaterite and the high ratio of calcite to vaterite are confirmed by powder X-ray diffraction spectra of these Si wafer controls (Figure 7; Table S1, Supporting Information). When n16N complexes are generated on Si wafers in Stage 1, washed, and 1887

dx.doi.org/10.1021/bm200231c |Biomacromolecules 2011, 12, 1883–1890

Biomacromolecules

Figure 6. SEM images of calcium carbonate mineral deposits formed on Si wafers in two-stage mineralization assays (Figure S1, Supporting Information). (A) Negative control (i.e., Control 1, n16N absent during both stages of mineralization). Note that both randomly oriented calcite and vaterite (v) appear in these assays. (B) n16N present during the first stage of mineralization but not during the second stage (i.e., Control 2). Note both the higher incidence of vaterite in panel B compared to panel A and the appearance of fibril
spheroidal n16N deposits on the surface of the support. (C, D) n16N present during both stages of the experiment. “c” denotes calcite crystals. Note the presence of fibril
spheroidal complexes in the background of both images. In panel D, arrows denote the close spatial relationship between n16N deposits and mineral crystals and evidence of fibril
spheroidal deposit formation on the exposed surfaces of the crystals.

Figure 7. Representative powder micro X-ray diffraction spectra of Si wafers taken from Control 1 (red) and 50 μM n16N (blue) two-stage mineralization assays. Calcite (c)- and vaterite (v)-specific peaks are labeled and were assigned from known data sets (Table S1, Supporting Information). The X-ray diffraction pattern for Control 2 is nearly identical to that obtained from the 50 μM n16N sample (data not shown). Note the intensities of the vaterite peaks relative to the calcite peaks in the n16N sample and in Control 1.

then reintroduced to fresh CaCl2 solution during Stage 2 of the double mineralization experiment (Control 2), we note the presence of scattered n16N fibril
spheroidal assemblies, rhombohedral calcite, and a slightly higher incidence of vaterite deposits (Figure 6B). These results were duplicated on polystyrene supports as well (Supporting Information, Figure S3).

ARTICLE

Hence, the preassembled n16N deposits are nucleating both calcite and vaterite in these assays. An even more striking result is obtained when these same preassembled n16N complexes are exposed to fresh n16N solution during Stage 2 of the double mineralization experiment. Here, compared to controls, we note an overall increase in the incidence of vaterite nucleation (Figure 6C,D); however, we note that calcite still forms in these assays. Qualitatively similar results were also obtained for polystyrene supports (Supporting Information, Figure S3). Closer examination confirms that the vaterite crystals are in contact with n16N fibril
spheroidal assemblies (Figure 6D), and X-ray diffraction spectra of these Si wafers confirms a higher vaterite to calcite ratio compared to the Control 1 scenario (Figure 7). Even more interesting, we note that new n16N deposits have assembled and formed on the surfaces of the vaterite crystals (Figure 6D). This implies that additional n16N assembly occurred during this second stage of mineralization when n16N was present in the assay solution.

’ CONCLUSIONS The contemporary view of n16 is that of a framework protein that cooperatively participates with other nacre proteins (Pif)5 and β-chitin polysaccharide to form a biomolecular multicomponent system that facilitates aragonite nucleation in the molluscan nacre layer.3,5 Recent studies with the intrinsically disordered n16N sequence10,11,31 confirm that the 30 AA N-terminal region of n16 possesses β-chitin binding capability6 and can facilitate the nucleation of vaterite and aragonite6,15 and, in the presence of a silk-fibroin gel, vaterite and ACC on β-chitin surfaces.16 Our present study now demonstrates that n16N also exhibits key properties associated with framework proteins: this sequence forms supramolecular assemblies of heterogeneous particle sizes that facilitate polymorph nucleation. These polypeptide assemblies are nonmineralized (Figure 3), amorphous in structure (Figure 4), and consist of fibril
spheroidal components (Figures 1
3) that are reminiscent of particles formed by other biomineralization proteins.25,26 These n16N particles can undergo further assembly into sheet or film-like complexes (Figures 1
3; Figure S2, Supporting Information). At this time the pathway for fibril
spheroidal-to-sheet/film assembly is not known. The formation of n16N films supports previous observations that porous peptide layers of n16N were integral components of aragonite deposits,15 and it is likely that either the fibril
spheroidal or sheet-like assemblies were responsible for ACC, vaterite, and aragonite nucleation on n16N-coated β-chitin surfaces.6,16 Although we were unable to successfully test n16N films for nucleation activity, we verified that n16N fibril
spheroidal assemblies promote calcite and vaterite nucleation on supports (Figures 6 and 7; Figure S3, Supporting Information), with evidence of contact between the mineral phase and the polypeptide assemblies (Figures 6C,D). Thus, n16N assemblyinduced nucleation exhibits the molecular characteristics expected for a nacre framework protein, i.e., crystal formation occurs on or near the surface of the protein assembly and not within it. This is in contrast to the mechanism observed for intracrystalline aragonite-associated protein sequences such as AP713 and PFMG114 where nucleation was observed within the polypeptide assemblies. Surface nucleation may represent a mineralization mechanism employed by framework nacre proteins in general. 1888

dx.doi.org/10.1021/bm200231c |Biomacromolecules 2011, 12, 1883–1890

Biomacromolecules We note that support-captured n16N assemblies nucleate vaterite and calcite but not aragonite (Figure 4). This would be consistent with the hypothesis that n16N functions like a framework protein and thus must associate with other biomacromolecules, such as β-chitin, in order for aragonite to form.6 The ability of n16N assemblies to form aragonite without the participation of other biomacromolecules is extremely rare, as noted in lamellar aragonite
n16N studies.15 Thus, when adsorbed onto polystyrene or Si wafer supports, we believe that n16N lacks essential associations with biomacrmolecules that are required for aragonite nucleation. The fact that a higher incidence of vaterite formation occurs instead of aragonite (Figure 6) is intriguing, given that vaterite is considered to be an intermediate in the ACC-to-calcite or -aragonite formation pathways.34
36 If true, then the presence of vaterite in our present assays and in previous n16N studies6,16 suggests that vaterite may be the first crystalline polymorph to form but is unable to transform to aragonite. Further studies will be required to establish why vaterite preferentially forms in n16N systems that lack a β-chitin substrate. We believe that polypeptide intrinsic disorder17
19 (54% random coil, 46% β structures, Figure 5B)11 influences the formation of n16N supramolecular assemblies on two levels. First, the presence of disordered polypeptide chains within these assemblies explains why n16N sheets or films appear to be amorphous and lack crystalline organization (Figures 1, 3, and 4). Second, it is known that disordered protein sequences will form protein
protein complexes in order to achieve internal polypeptide stability yet retain some degree of polypeptide chain disorder in order to participate in additional interactions.17
19 The fact that the n16N polypeptide persists in an unfolded state as a monomer11 and oligomer (Figure 5B) is consistent with this notion and suggests that this polypeptide complex could participate in additional matrix interactions, such as with β-chitin and/or the mineral phase over time.6,16
19 This ability to interact with multiple targets over time may be an important feature of nacre multicomponent systems and will require further investigation. Another major feature of the n16N assembly process is the contribution from side chain electrostatics, as evidenced by pHdependent n16N assembly formation (Figure 5A). This data indicates that side chain salt bridging/ion pairing is at work and the pH range where oligomerization occurs yields clues as to the side chain participants in this process. We note that the first stage of oligomerization (pH 4
4.5) corresponds with the pKa’s for most protein Asp and Glu residue carboxylate groups.37
39 In n16N, these residues are found at positions 17, 19, 21, 24, 27 (Figure 5A), and we believe that these carboxylate groups form intermolecular salt bridges with corresponding cationic groups (e.g., H3, K4, K5, R8, R20, R22, K28, K29) within other n16N molecules. Once salt bridging is achieved, it is possible that other nonbonding interactions (e.g., van der Waals/hydrophobic or hydrogen bonding)27
33 further stabilize the n16N oligomers. In the second stage of oligomerization (pH g 8.5) a number of potential side chain participants may be involved. Examples include Cys sulfhydryl groups37 and Lys amino groups,38,39 both of which reside within the n16N sequence and have pKa’s in the alkaline pH range (Figure 5A). However, since these side chain groups possess overlapping pKa’s, it is difficult to discern which charged groups participate in the second stage of n16N assembly. Therefore, additional studies using sequence variants of the n16N polypeptide will be required to clearly identify side chain participants for the second assembly step.

ARTICLE

Finally, our studies demonstrate that mineralized calcium deposits do not form within n16N assemblies (Figure 3) and that Ca(II) ions are not necessary for initiating the assembly process (Figure S2, Supporting Information; Figure 5A). However, Ca(II) does have a significant impact on the pH range where assembly occurs (Figure 5A), and one cannot rule out the possibility that low levels of Ca(II) ions exist within or on the surface of these n16N deposits (Figure 3). So the question is, how does Ca(II) contribute to the assembly process? We believe there are two possibilities. First, we know from previous biophysical studies that Ca(II) binds weakly to Asp and Glu residues in the n16N sequence and does not induce global folding of this polypeptide but does introduce local conformational changes within specific regions of the sequence.11 Thus, Ca(II) ions may induce conformational changes in n16N that do not lead to global folding but affect the equilibrium between monomeric and oligomeric species and shift the initiation of the assembly process (Figure 5A). Second, there may be Ca(II) ion charge screening effects at Asp, Glu carboxylate sites that compete with intermolecular interactions with Lys, Arg, and His cationic groups and thus interfere with the assembly process. Other possibilities may be likely, and thus further experimentation will be required to elucidate Ca(II)-induced modulation in more detail.

’ ASSOCIATED CONTENT

bS

Supporting Information. Experimental scheme for twostage mineralization assays (Figure S1), SEM images of n16N supramolecular assemblies formed by alkaline pH in the absence of Ca(II) (Figure S2), SEM images of n16N two-stage mineralization experiments captured on polystyrene supports (Figure S3), and powder X-ray diffraction data for calcite, vaterite, and aragonite (Table S1). This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*Tel: 212-998-9605. Fax: 212-995-4087. E-mail: [email protected].

’ ACKNOWLEDGMENT Research supported by the U.S. Department of Energy, Office of Basic Energy Sciences, Division of Materials Sciences and Engineering under Award DE-FG02-03ER46099. This work represents contribution number 59 from the Laboratory for Chemical Physics. ’ ABBREVIATIONS n16N = 1
30 AA N-terminal domain of the Pinctada fucata Japanese pearl oyster nacre protein, n16; IDP = intrinsically disordered protein; AFM = atomic force microscopy; UDDW = unbuffered deionized distilled water; DLS = dynamic light scattering; DTT = dithiothreitol; EDS = energy dispersive spectra; ACC = amorphous calcium carbonate; AP7 = aragonite protein 7, Haliotis rufescens; PFMG1 = Pinctada fucata mantle gene product 1; CD = circular dichroism spectrometry; PDF = powder diffraction file 1889

dx.doi.org/10.1021/bm200231c |Biomacromolecules 2011, 12, 1883–1890

Biomacromolecules

’ REFERENCES (1) Mann, S.; Webb, J.; Williams, R. J. P., Eds. In Biomineralization: Chemical and Biochemical Perspectives; VCH: Weinheim, Germany, 1989. (2) Lowenstam, H. A.; Weiner, S. On Biomineralization; Oxford Press, New York, NY, 1989. (3) Samata, T.; Hayashi, N.; Kono, M.; Hasegawa, K.; Horita, C.; Akera, S. FEBS Lett. 1999, 462, 225–232. (4) Nudelman, F.; Gotliv, B. A.; Addadi, L.; Weiner, S. J. Struct. Biol. 2006, 153, 176–187. (5) Suzuki, M.; Saruwatari, K.; Kogure, T.; Yamamoto, Y.; Nishimura, T.; Kato, T.; Nagasawa, H. Science 2009, 325, 1388–1390. (6) Keene, E. C.; Evans, J. S.; Estroff, L. A. Cryst. Growth Des. 2010, 10, 1383–1389. (7) Michenfelder, M.; Fu, G.; Lawrence, C.; Weaver, J. C.; Wustman, B. A.; Taranto, L.; Evans, J. S. Biopolymers 2003, 70, 522–533; Errata2004, 73, 522. (8) Fu, G.; Qiu, S. R.; Orme, C. A.; Morse, D. E.; DeYoreo, J. J. Adv. Mater. 2007, 17, 2678–2683. (9) Shen, X.; Belcher, A. M.; Hansma, P. K.; Stucky, G. D.; Morse, D. E. J. Biol. Chem. 1997, 272, 32472–32481. (10) Kim, I. W.; DiMasi, E.; Evans, J. S. Cryst. Growth Des. 2004, 4, 1113–1118. (11) Collino, S.; Evans, J. S. Biomacromolecules 2008, 9, 1909–1918. (12) Falini, G.; Albeck, S.; Weiner, S.; Addadi, L. Science 1996, 271, 67–69. (13) Amos, F. F.; Evans, J. S. Biochemistry 2009, 48, 1332–1339. (14) Amos, F. F.; Destine, E.; Ponce, C. B.; Evans, J. S. Cryst. Growth Des. 2010, 10, 4211–4216. (15) Metzler, R. A.; Evans, J. S.; Kilian, C. E.; Zhou, D.; Churchill, T. H.; Appathurai, P. N.; Coppersmith, S. N.; Gilbert, P. U. P. A. J. Am. Chem. Soc. 2010, 132, 6329–6334. (16) Keene, E. C.; Evans, J. S.; Estroff, L. A. Cryst. Growth Des. 2010, 10, 5169–5175. (17) Uversky, V. N. Protein Sci. 2002, 11, 739–756. (18) Uversky, V. N.; Gillespie, J. R.; Fink, A. L. Proteins 2000, 41, 415–427. (19) Meng, J.; Romero, P.; Yang, J. Y.; Chen, J. W.; Vacic, V.; Obradovic, Z.; Uversky, V. N. BMC Genomics 2008, 9, 1–26. (20) Kim, I. W.; Darragh, M. R.; Orme, C.; Evans, J. S. Cryst. Growth Des. 2006, 6, 5–10. (21) Ndao, M.; Keene, E.; Amos, F. A.; Rewari, G.; Ponce, C. B.; Estroff, L.; Evans, J. S. Biomacromolecules 2010, 11, 2539–2544. (22) Borgstahl, G. E. O. How to use dynamic light scattering to improve the likelihood of growing macromolecular crystals. In Macromolecular Crystallography Protocols; Doublie, S., Ed.; Humana Press, Inc.: Totowa, NJ, 2010; Vol. 1, Preparation and Crystallization of Macromolecules, pp 109
129. (23) Onuma, K.; Kubota, T.; Tanaka, S.; Kanzaki, N.; Ito, A.; Tsutsui, K. J. Phys. Chem. B. 2002, 106, 4138–4324. (24) Reed, J.; Reed, T. A. Anal. Biochem. 1997, 254, 36–40. (25) Murr, M. M.; Morse, D. E. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 11657–11662. (26) He, X.; Li, W.; Habelitz, S. J. Struct. Biol. 2008, 164, 314–321. (27) Colfer, S.; Kelly, J. W.; Powers, E. T. Langmuir 2003, 19, 1312–1318. (28) Shera, J. N.; Sun, X. S. Biomacromolecules 2009, 10, 2446–2450. (29) Zimenkov, Y.; Dublin, S. N.; Ni, R.; Tu, R. S.; Breedveld, V.; Apkarian, R. P.; Conticello, V. P. J. Am. Chem. Soc. 2006, 126, 6770– 6771. (30) Nakano, M.; Shen, J. R.; Kamino, K. Biomacromolecules 2007, 8, 1830–1835. (31) Dalmau, M.; Lim, S.; Wang, S. W. Biomacromolecules 2009, 10, 3199–3206. (32) Kayser, V.; Turton, D. A.; Aggeli, A.; Beevers, A.; Reid, G. D.; Beddard, G. S. J. Am. Chem. Soc. 2004, 126, 336–343. (33) Rajagopal, K.; Lamm, M. S.; Haines-Butterick, L. A.; Pochan, D. J.; Schneider, J. P. Biomacromolecules 2009, 10, 2619–2625.

ARTICLE

(34) Radha, A. V.; Forbes, T. Z.; Killian, C. E.; Gilbert, P. U. P. A.; Navrotsky, A. Proc. Natl. Acad. Sci. U.S.A. 2010, 107, 16438–16443. (35) Pouget, E. M.; Bomans, P. H. H.; Goos, J. A. C. M.; Frederik, P. M.; de With, G.; Sommerdijk, N. A. J. M. Science 2009, 323, 1455– 1458. (36) Gebauer, D.; Volkel, A.; Colfen, H. Science 2008, 322, 1819– 1822. (37) Song, Y.; Mao, J.; Gunner, M. R. J. Comput. Chem. 2009, 30, 2231–2247. (38) Merz, K. M. J. Am. Chem. Soc. 1991, 113, 3572–3575. (39) Li, H.; Robertson, A. D.; Jensen, J. H. Proteins: Struct., Funct., Bioinf. 2005, 61, 704–721.

1890

dx.doi.org/10.1021/bm200231c |Biomacromolecules 2011, 12, 1883–1890