Purification, Characterization, and in VitroMineralization Studies of a Novel Goose Eggshell Matrix Protein, Ansocalcin*

Biomineralization is an important process in which hard tissues are generated through mineral deposition, often assisted by biomacromolecules. Eggshells, because of their rapid formation via mineralization, are chosen as a model for understanding the fundamentals of biomineralization. This report discusses purification and characterization of various proteins and peptides from goose eggshell matrix. A novel 15-kDa protein (ansocalcin) was extracted from the eggshell matrix, purified, and identified and its role in mineralization evaluated using in vitro crystal growth experiments. The complete amino acid sequence of ansocalcin showed high homology to ovocleidin-17, a chicken eggshell protein, and to C-type lectins from snake venom. The amino acid sequence of ansocalcin was characterized by the presence of acidic and basic amino acid multiplets. In vitro crystallization experiments showed that ansocalcin induced pits on the rhombohedral faces at lower concentrations (<50 μg/ml). At higher concentrations, the nucleation of calcite crystal aggregates was observed. Molecular weight determinations by size exclusion chromatography and sodium dodecyl sulfate -polyacrylamide gel electrophoresis showed reversible concentration-dependent aggregation of ansocalcin in solution. We propose that such aggregated structures may act as a template for the nucleation of calcite crystal aggregates. Similar aggregation of calcite crystals was also observed when crystallizations were performed in the presence of whole goose eggshell extract. These results show that ansocalcin plays a significant role in goose eggshell calcification.

Organisms are capable of developing minerals and biocomposites with complex architecture to fulfill important biological functions, such as skeletal support, protection of soft tissues, and food grinding (1)(2)(3)(4). Often, survival of the organism depends on the structure and strength of these composite materials. Due to the wide range of mechanical and functional properties of these biocomposites, identifying the organic and inorganic components and understanding the structure-function relationships have potential industrial and biotechnological applications (5). The inorganic mineral phase of such materials is formed over an insoluble organic matrix or mold, and the mineral phase is intimately associated with organic macromolecules, such as proteins/glycoproteins, polysaccharides, or proteoglycans (6). These biomacromolecules are highly acidic in nature and have been postulated to control nucleation, growth, crystal size, and shape of the mineral phases (7).
Mann (8) classified the biologically programmed composites into four types (types I, II, III, and IV), based on matrix intervention in nucleation and growth processes. Avian eggshells form type II biocomposites, which is the fastest forming hard acellular composite in nature. For example, in the case of chicken eggshell, about 5 g of the mineral phase is produced within 24 h (9). Their calcified layer consists of ϳ95% mineral and ϳ5% organic phase (10). The mineral phase acts as a mechanical support as well as allows the diffusion of gases, water, and ions and is, therefore, essential for survival of the embryo. The organic phase (proteins and proteoglycans) of the eggshell matrix is believed to be responsible for the nucleation and directed growth of the calcified layer. So far, several matrix proteins from chicken eggshells have been purified and characterized (11)(12)(13)(14)(15)(16)(17). These proteins are subdivided into three groups, namely non-collagenous bone proteins (osteopontin), eggshell-specific proteins (ovocleidins and ovocallyxins), and egg white proteins (ovalbumin, ovotransferrin, and lysozyme). Although their presence within the mineral layer has been demonstrated by immunohistochemistry (11, 12, and 17), the role of these proteins in calcite mineralization is not clearly understood. Hincke et al. (18) showed that egg white lysozyme and ovotransferrin influence the morphology of CaCO 3 crystals. Dermatan sulfate, chondroitin sulfate, and hyaluronic acid were also identified in the chicken eggshell matrix (19 -22). Wu et al. (23) have shown that partially purified dermatan sulfate proteoglycans obtained from the eggshell reduced the size of the calcite crystals. Uterine fluid collected at various stages of eggshell formation (initial, growth, and final stages), reduced the size, and induced curved faces on the calcite crystals (24). Though many reports are available on the ultrastructure, composition, and presence of organic macromolecules in the eggshell matrix (25,26), information on the role of individual proteins and on the molecular mechanism of avian eggshell mineralization are limited (27). This is presumably due to problems associated with the separation and purification of these biomacromolecules (28,29).
A comparative study of various avian eggshells indicates that the basic architecture is identical, but their ultrastructure and composition are different (30). Therefore, one expects the interaction between the organic macromolecules and the mineral phase to be different in different avian species. We believe that understanding the structure and role of the organic matrix would help us to understand the complex process of biomineralization. Here we report the purification and characterization of various matrix proteins and peptides present in the calcified layer of the goose (Anser anser) eggshell. We have identified a 15-kDa protein (ansocalcin), a major constituent of the goose eggshell extract and determined its complete amino acid sequence. For a better understanding of the role of ansocalcin in the eggshell mineralization process, we compared ansocalcin with whole eggshell extract in in vitro crystallization of CaCO 3 . We propose that ansocalcin plays an important role in goose eggshell calcification.

EXPERIMENTAL PROCEDURES
Extraction of Eggshell Matrix Proteins-Commercially available fresh goose eggshells were broken and thoroughly washed with Millipore water. They were decalcified with 1 N HCl for 30 min, filtered, and centrifuged. The supernatant solution was desalted using an Amicon microconcentrator (YCO 5 membrane, 500 M r cutoff) at 4°C. The turbid solution was centrifuged at 4000 rpm for 15 min, and the supernatant liquid was taken for purification using RP-HPLC. 1 Protein Purification-The proteins were fractionated on a Jupiter C 18 reversed phase column (5, 250 ϫ 10 mm) using a Vision Work station (PerkinElmer PerSeptive Biosystems). The column was equilibrated with 0.1% trifluoroacetic acid, and a linear gradient of acetonitrile was used for elution. The microconcentrated sample (ϳ15 mg of protein) was injected onto the column and was eluted at a flow rate of 2 ml/min. The elution of the proteins was monitored both at 215 and 280 nm.
Electrospray Ionization Mass Spectrometry-Precise masses of the proteins and peptides were determined by ESI-MS using a PerkinElmer Sciex API 300 triple quadrupole instrument equipped with an ion spray interface. The ion spray voltage was set at 4.6 kV and the orifice voltage at 30 V. Nitrogen was used as a curtain gas with a flow rate of 0.6 liters/min, while compressed air was utilized as the nebulizer gas. The sample was injected into the mass spectrometer at a flow rate of 50 l/min and scanned from a mass to charge (m/z) ratio of 500 -3000. The multiply charged spectrum was deconvoluted into the mass scale using the Biospec Reconstruct software supplied with the instrument data system.
Reduction and Alkylation of Native Proteins-The native protein (ϳ3 mg) was dissolved in 1000 l of 130 mM Tris-HCl, 1 mM EDTA, and 6 M guanidine-HCl (pH 7.5). 2-Mercaptoethanol (20 l/mg of protein) was added, and the solution was incubated under nitrogen atmosphere for 2 h at 37°C. The alkylating agent, 4-vinylpyridine (200 l/mg of protein), was subsequently added, and the mixture was incubated under nitrogen for another 2 h at room temperature. The S-pyridylethylated protein was separated from the reaction mixture by RP-HPLC on a Jupiter C 18 (250 ϫ 10 mm) column using a linear gradient of acetonitrile.
Enzymatic and Chemical Cleavage-The S-pyridylethylated protein was digested with the enzymes lysyl endopeptidase and trypsin. In a typical digestion method, 500 g of the protein was dissolved in 500 l of 50 mM Tris-HCl, 4 M urea, 5 mM EDTA (pH 7.5). The enzymes were added (enzyme/protein ratio of 1:100), and the digestions were carried out at 37°C for 18 h. The aspartidyl cleavage was performed using 2% formic acid (31). The S-pyridylethylated protein was dissolved in 2% formic acid in a glass vial. The solution was frozen and thawed completely under vacuum. The vial was then vacuum-sealed and heated at 108°C for 2 h. The peptides generated by chemical and enzymatic cleavage were separated by RP-HPLC on a Sephasil C 18 column (100 ϫ 2.1 mm) using a linear gradient of acetonitrile.
N-terminal Sequencing-N-terminal sequencing of the purified proteins or peptides was performed by automated Edman degradation using a PerkinElmer Applied Biosystems 494 pulsed-liquid phase protein sequencer (Procise) with an online 785A PTH-amino acid analyzer.
Size Exclusion Chromatography-All the standards and purified protein were dissolved in 7.5 mM CaCl 2 solution or in 200 mM Tris-HCl (pH 7.5) solution. The SEC was performed on a Sepharose CL-6B column (Sigma, 1.5 ϫ 57 cm) at various concentrations of ansocalcin using CaCl 2 (7.5 mM) solution as the mobile phase. SEC experiments were repeated using 200 mM Tris-HCl (pH 7.5) as the mobile phase in the absence of calcium ions. Ovalbumin, bovine serum albumin, and chicken egg lysozyme were used as standards. The sample was dissolved in eluent and loaded onto the column. Absorbance of the fractions was measured using a Jasco V-560 uv/vis spectrophotometer.

Circular Dichroism (CD) Experiments-
The secondary structure of the protein was analyzed using a Jasco J 700 circular dichroism spectropolarimeter. The instrument was calibrated with 0.05% (ϩ)-10-camphor sulfonic acid solution. The CD spectra of the protein at a concentration range of 10 -500 g/ml in water were collected using a 0.1-mm sample cell. To study the effect of Ca 2ϩ ions, spectra were also recorded in 7.5 mM CaCl 2 solution. The instrument optics were flushed with 30 liters/min nitrogen gas. A total of three scans was recorded, averaged for each spectrum, and baseline subtracted. CD spectral data from 195-240 nm were appended to the reference data set in Convex Constant Analysis program (32). The conformational weight was obtained from 4-component deconvolution, which resulted in best fit with the experimental CD spectrum.
Crystal Growth Experiments-CaCO 3 crystals were grown on glass cover slips placed inside the CaCl 2 solution kept in a Nunc dish, 4 ϫ 6 wells (33). Typically, the lyophilized proteins or protein extract was accurately weighed on a microbalance (Ohaus Analytical Plus, OHAUS Corporation, NJ) and dissolved in a 7.5 mM CaCl 2 solution to give a final concentration of 0.1-1000 g/ml. 1 ml of 7.5 mM CaCl 2 solution was introduced into the wells containing the cover slips, and the whole set up was covered with aluminum foil with a few pinholes on the top. Crystals were grown inside a closed desiccator for 2 days by slow diffusion of gases released by the decomposition of ammonium carbonate placed at the bottom of the desiccator. After 2 days, the glass slides were carefully lifted from the crystallization wells, rinsed gently with Millipore water, air-dried at room temperature, and used for characterization.
Scanning Electron Microscopy-Scanning electron microscopic studies of the CaCO 3 crystals were carried out using a JEOL 2200 scanning electron microscope at 15/20 kV after coating with gold to increase the conductivity. The crystal aggregate size distribution was measured by randomly choosing at least 20 aggregates from each experiment, and the size of each aggregate was measured. The number of crystal aggregates with a particular size at an accuracy of Ϯ 2 m was noted, and the percentage was calculated.
SDS-PAGE-SDS-PAGE was performed on a 4 -20% polyacrylamide gel (Bio-Rad) under non-reducing conditions according to the method of Laemmli (34). The proteins were stained with Coomassie Brilliant Blue R-250.
Fluorescence Spectroscopy-The fluorescence emission spectra were collected on a Shimadzu RF-5301PC spectrofluorometer with the emission and excitation band passes set at 3 nm. The excitation wavelength was set at 295 nm (to selectively excite tryptophan residues), and the spectra were recorded from 300 to 400 nm. Spectra of the proteins were recorded either in 7.5 mM CaCl 2 solution or in Millipore water.

FIG. 1. Purification of ansocalcin from goose eggshell extract.
A, RP-HPLC of goose eggshell extract. The crude extract was loaded onto the Jupiter C 18 column and eluted using a linear gradient of acetonitrile. The various fractions were labeled a, b, c, and d. B, fraction c was pooled and rerun using a shallow gradient. The dashed lines represent the solvent gradient applied during elution.

Purification and Characterization of Eggshell Proteins-
The calcified shell was dissolved in 1 N HCl to extract the entire soluble organic matrix associated with it. After decalcification, the crude extract was microconcentrated and fractionated by RP-HPLC on a Jupiter C 18 column. Fig. 1A shows a typical chromatogram of the various eluting fractions (labeled a, b, c, and d). Using a shallow gradient, we obtained 3 peptides from fraction b (labeled b1, b2, and b3) and one peptide from fraction a (a1, data not shown). Fraction c was separated again on an RP-HPLC column using a shallow gradient into two subfractions (Fig. 1B, c1 and c2). Fraction c1 was identified as a 15-kDa protein (Fig. 2), and the minor component, c2, was identified as a 16-kDa protein as determined from the positive ion ESI-MS (data not shown). We named the 15-kDa protein (fraction c1) as ansocalcin (Anser ovum calcium-binding protein). Fraction d was found to contain a 16-kDa protein (Fig.  2F, labeled d1). This protein is similar to c2 based on its molecular weight and retention time (data not shown). The RP-HPLC profile showed that ansocalcin is the major constituent of the eggshell extract.
The mass spectra of the various proteins/peptides purified from the crude extract are shown in Fig. 2. The mass spectrum of ansocalcin showed an overlapping envelope of positive ion series from 6ϩ to 13ϩ (Fig. 2E). The inset diagram indicates the reconstructed mass of the protein with a value of 15,341.70 Ϯ 1.97 daltons, which confirmed its homogeneity. The mass spectrum of fraction d1, the second major component in the eggshell extract, showed a charge distribution of 6ϩ to 10ϩ. The reconstructed mass of this protein was found to be 16,278.52 Ϯ 1.36 daltons (Fig. 2F). The mass spectra of the peptides purified from fractions a and b are shown in Fig. 2, A-D. Thus the soluble components of the eggshell extract consist of a number of proteins and peptides with a wide range of molecular weights. The mass spectral data indicated that all the proteins/peptides are homogeneous.
Amino Acid Sequence of Ansocalcin-The complete amino acid sequence was determined by sequencing the peptides generated by chemical and enzymatic digestions of S-pyridylethylated ansocalcin. The sequences of the overlapping peptides and other details are shown in Fig. 3. The peptides generated by formic acid digestion gave more or less the complete sequence of ansocalcin. The C terminus of the protein was confirmed by the peptides obtained by both formic acid and trypsin digestion. Alanine was identified as the C-terminal residue because it was the last residue sequenced on peptides A118-FIG. 2. ESI-MS of the proteins purified from goose eggshell extract. Positive ion electrospray ionization mass spectrograph of the various proteins/peptides purified from goose eggshell extract. 20 l of the sample was injected into the spectrometer using a 1:1 mixture of 0.1% trifluoroacetic acid and acetonitrile. The multiply charged spectrum was converted using Biospec Reconstruct software (shown in the inset diagram). The deconvoluted spectra also indicate the homogeneity of the various eggshell matrix proteins/peptides. The mass spectra correspond to fraction a1 (A), fraction b1 (B), fraction b2 (C), fraction b3 (D), fraction c1 (ansocalcin, E), and fraction d1 (F).
A132 and A124-A132 obtained by formic acid and trypsin digestions, respectively. Table I shows the observed and calculated masses of the various peptides generated by chemical and enzymatic digestions. In all cases, the calculated molecular masses of the peptides matched the observed molecular weight. Ansocalcin has a total of 132 amino acid residues with a calculated mass of 15,341.20 daltons. This matches the observed mass of 15,341.70 Ϯ 1.97. Based on the amino acid sequence, ansocalcin is rich in Ala (9%), Glu (8%), Ser (7.6%), and Gly (7.6%) residues. Ansocalcin also has a high content of tryptophan residues (6%) compared with the natural abundance of 1.3% (35). The amino acid sequence of ansocalcin revealed the presence of repeat patterns of acidic and basic amino acids. Structural Similarity with Other Proteins-A search of the NCBI data base revealed that ansocalcin has high identity (31-35%) and homology (48 -52%) to C-type lectin(-like) (CTL) family of proteins from snake venom (Fig. 4). The first two proteins are involved in CaCO 3 biomineralization and have CTL-like domains, whereas the next three proteins are CTL extracted from snake venom. Based on the sequence homology, we propose intrachain disulfide linkages between cysteines 3 and 14, 31 and 128, and 103 and 120. Ansocalcin lacks the sequence QPD, a sequence motif that was interpreted to account for galactose specificity (36). The residues that are involved in Ca 2ϩ ion binding sites 1 and 2 of rat mannose-binding protein (37) are not conserved in ansocalcin. Earlier studies have shown that matrix proteins associated with the calciumrich mineral phase (16, 38 -40) showed homology to C-type lectins. However, except for perlucin (39), these proteins lack the conserved QPD sequence that is required for carbohydrate binding.
The search also indicated ϳ34% identity and 50% homology with the amino acid sequence of ovocleidin 17 (OC-17), the eggshell matrix protein from chicken eggshells (16). Comparison of the sequence of ansocalcin with OC-17 showed four important differences. Firstly, ansocalcin has an extra cysteine (Cys 38 ), which is not present in OC-17. Secondly, the occurrence and number of acidic and basic amino acids repeats is different. In OC-17, we identified one acidic triplet (Asp 118 -Glu 119 -Glu 120 ) and four basic pairs (Arg 34 -Arg 35 , His 80 -Arg 81 , His 102 -Arg 103 , Arg 108 -Arg 109 ). Thirdly, OC-17 is phosphorylated at two serine residues (indicated as S in Fig. 4). On the other hand, ansocalcin is not phosphorylated as the calculated molecular weight matches the one determined by ESI-MS. Finally, the ratio of basic to acidic residues is 1.3 in ansocalcin and 1.9 in OC-17 (including the phosphorylated serines). The acidic amino acid residues constitute about 13.6% for ansocalcin and 8% for OC-17. Thus, ansocalcin and OC-17 do not possess a large number of acidic amino acids compared with other soluble matrix proteins, which possess a large amount (35-50%) of acidic amino acids (41)(42)(43).
Amino Acid Sequence of Other Peptides-We have determined the complete amino acid sequence of fraction b1 (Fig.  5A). The calculated mass of this peptide from the amino acid FIG. 3. Amino acid sequence strategy for ansocalcin. N-terminal amino acid sequence of intact S-pyridylethylated ansocalcin. Peptides derived from the digestion of ansocalcin by formic acid, trypsin, and endopeptidase Lys-C (Lys-C), are indicated. Solid lines represent regions whose sequences were determined by Edman degradation and/or confirmed by mass spectrometry. Broken lines indicate the regions that were not sequenced. Thus, it appears to be a proteolytic product of a protein related to OC-116. The partial amino acid sequence of b3 (Fig. 5B) did not show significant homology to any known proteins. Sequencing of fractions d1 and a1 showed two N-terminal residues, which indicates that these fractions contain two polypeptides, and, thus, their sequence could not be determined.
Effect of the Eggshell Extract on CaCO 3 Crystallization-To understand the role of ansocalcin in eggshell calcification, CaCO 3 crystals were grown in the presence of whole eggshell extract or ansocalcin in different sets of experiments under identical conditions. The experiments were performed by slow diffusion of carbon dioxide (produced via the decomposition of solid ammonium carbonate) into the protein extract dissolved in CaCl 2 solution placed inside a closed desiccator. The concentration of the extract was varied from 0.1-1000 g/ml. Under these conditions, only the calcite phase was nucleated in all crystal growth experiments regardless of the presence or absence of protein(s). However, the growth pattern of the calcite crystals varied with the concentration of the eggshell extract (mixture of proteins). No significant effects were observed below 50 g/ml of the protein extract (Fig. 6B), although pits were observed on the {10.4} rhombohedral faces (Fig. 6B, inset). At a concentration of 100 g/ml, some of the crystals exhibited protrusions parallel to the {10.4} faces (Fig. 6C). As the protein extract concentration was increased to 250 g/ml, calcite crystal aggregates started to appear along with curved edges for the crystals (Fig. 6D). At 500 g/ml, calcite crystal aggregates with various shapes (spherical, ellipsoidal, or dumbbell) were formed (Fig. 6E). At 1000 g/ml, the highest concentration used in these experiments, the individual crystal aggregates formed were smaller, but greater in number and more curved at the corners and edges than the aggregates grown at other concentrations (Fig. 6F).
Effect of Ansocalcin on CaCO 3 Crystallization-The mor-phology of the calcite crystals changed gradually as a function of the concentration of ansocalcin (Fig. 7). At lower concentrations (0.1-1.0 g/ml) spiral pits appeared at the sides of the {10.4} rhombohedral crystals without major change in size and shape (data not shown). As the concentration was increased to 10 g/ml, the calcite crystals that formed had terraced structures on the {10.4} face penetrating toward the center of the crystal (Fig. 7B). At 50 g/ml ansocalcin, the crystal aggregates exhibited similar morphology to that obtained at 500 g/ml whole eggshell extract (mixture of proteins, compare Figs. 6E and 7C). However, in the case of ansocalcin at a concentration of 50 g/ml, the edges and corners remained sharp. The aggregates formed at 100 g/ml ansocalcin contained a higher number of crystallites than at 50 g/ml without significant change in the overall aggregate size (Fig. 7D). As the ansocalcin concentration was increased further, smaller and smaller crystal- lites were formed, and these crystallites randomly aggregated into spherical and ellipsoidal forms (Fig. 7, E and F). Attempts to break the calcite spherules with a sharp blade to identify the core were unsuccessful due to the collapse of the aggregates into small crystallites.
The size distribution of the calcite aggregates measured (accuracy of Ϯ2 m) at the concentration range of 50 -500 g/ml ansocalcin is shown in Fig. 8. The aggregate size distribution was narrow (32-42 m) with a maximum around 38.5 m at ansocalcin concentrations of 50 and 100 g/ml. At 250 g/ml, a reduction in the size of the calcite crystal aggregates was accompanied by a wide aggregate size distribution (10 -32 m). At 500 g/ml ansocalcin, the size distribution was narrower (10 -22 m) with a maximum number of crystals of size 20 m. Thus, an increase in ansocalcin concentration resulted in a decrease in the size of calcite crystal aggregates. It is important to note that the nucleation density (i.e. the number of crystals per unit area) increased significantly at the highest concentration of ansocalcin tested (500 g/ml, Fig. 7F).
Aggregation of Ansocalcin-To determine the noncovalent aggregation of ansocalcin, we examined the elution behavior of ansocalcin by SEC using 7.5 mM CaCl 2 solution as eluent (Fig.  9A). At 500 g/ml ansocalcin, it was observed that ansocalcin separated into two components. The profile was characterized by the presence of predominantly trimeric and a smaller amount of tetrameric species. However, at 100 g/ml, ansocalcin elutes as a dimer. At a concentration of 250 g/ml, chromatogram was complex and indicated the existence of a mixture of monomers and multimeric species, such as dimers, trimers, and tetramers (data not shown). The SEC of ansocalcin at 100 g/ml in 200 mM Tris-HCl (pH 7.5) as eluant in the absence of calcium ions showed a profile similar to that observed in CaCl 2 solution (Fig. 9B). At 500 g/ml, the profile was characterized by the presence of predominantly trimeric and a smaller amount of tetrameric species. However, as observed in the presence of CaCl 2 , a different profile was obtained at a concentration of ansocalcin of 250 g/ml with a strong peak corresponding to the formation of trimers and, to a smaller extent, the presence of higher order multimers (data not shown).  increase in the population of the dimer and, to a lesser extent, the intensity of a trimer band (molecular mass of 42 kDa). The SDS-PAGE profile was unaltered in the presence of Ca 2ϩ ions (Fig. 10). The absence of a change in electrophoretic mobility of ansocalcin in the presence of Ca 2ϩ ions indicates that ansocalcin does not require Ca 2ϩ ions for aggregation. The low amount of trimers may be due to decreased stability of the trimers in the solution used for this experiment or due to presence of large protein domains that could sterically preclude its formation in the SDS micelles. The concentration of ansocalcin required to form dimer is higher in SDS-PAGE than in SEC, indicating that at lower concentration SDS disrupts dimerization.
Far UV-CD studies of ansocalcin in the presence and absence of Ca 2ϩ ions showed significant changes in secondary structure, especially at high concentrations of ansocalcin (Fig. 11).
Analysis of the secondary structure at 50 g/ml showed the presence of 26% ␣-helix, 31% turn conformation, 6.8% aromatic stacking, and 29% random conformation. There were no significant differences in the CD spectra of ansocalcin in the presence and absence of Ca 2ϩ ions, when the experiments were performed up to a concentration of 100 g/ml ansocalcin. This indicates that Ca 2ϩ ions did not affect the conformation of the protein at these concentrations. Above 100 g/ml ansocalcin, some differences in the CD spectra were observed in the presence and absence of Ca 2ϩ ions. The spectra exhibited fine structures at 208 nm as the concentration was increased from FIG. 10. Electrophoretic mobility of ansocalcin. SDS-PAGE (4 -20%) of ansocalcin at various concentrations with and without Ca 2ϩ ions. Lane 1, molecular weight standards (broad range, Bio-Rad); lanes 2, 4, and 6, ansocalcin without Ca 2ϩ ions; lanes 3, 5, and 7, ansocalcin with Ca 2ϩ ions. The amounts of ansocalcin loaded are 15 g (lanes 2 and 3), 90 g  (lanes 4 and 5), and 150 g (lanes 6 and  7). The protein was dissolved in water or incubated in CaCl 2 (7.5 mM) overnight. 20 l of 2ϫ SDS loading buffer was added to an equal volume of the protein sample, the solution was heated at 95°C for 5 min and loaded onto the gel. M, monomers; D, dimers; T, trimers.
FIG. 11. CD spectra of ansocalcin. Far UV-CD spectra of ansocalcin at various concentrations with and without CaCl 2 (7.5 mM). A, 10 g/ml; B, 50 g/ml; C, 100 g/ml; D, 250 g/ml; E, 500 g/ml. The spectra showed significant changes in the secondary structure at higher concentrations of ansocalcin. The black lines represent CD spectra of ansocalcin in water, and the gray lines represent CD spectra recorded in 7.5 mM CaCl 2 . The instrument settings used were: scan range, 190 -260 nm; scan rate, 50 nm/min; sensitivity, 10 millidegrees; response time, 1 s.
100 to 250 g/ml, which may be due to the presence of monomeric and multimeric species (Fig. 11D). Further increases in the concentration of protein to 500 g/ml decreased the amplitude of the peaks accompanied by a strong negative maximum at 230 nm (Fig. 11E). Excessive scattering from the solution complicated the conformational analysis at higher concentrations (250 and 500 g/ml).
Ansocalcin contains eight tryptophan residues. To investigate the changes in the microenvironment around tryptophan residues, we recorded the emission spectrum of ansocalcin at exc of 295 nm (to selectively excite tryptophan). Fig. 12, A and B show the tryptophan fluorescence of ansocalcin in water and CaCl 2 solution. The spectra were characterized by smooth Lorentzian curves indicating more or less homogeneous environments for all of the tryptophan residues. An emission maximum of 346 nm was observed at all concentrations of ansocalcin. The emission spectra were not significantly affected by the presence of Ca 2ϩ ions (Fig. 12B). Fig. 12C indicates the variation in fluorescence emission intensity at 346 nm at various concentration of ansocalcin in water and in CaCl 2 solution. The increase in fluorescence intensity is linear up to 100 g/ml, reaches a maximum at 250 g/ml and decreases with further increases in the concentration of ansocalcin. A reduction in fluorescence intensity was observed at 500 g/ml without any shift in the emission maximum and it was greater in the absence of Ca 2ϩ ions.

DISCUSSION
Avian eggshell is a unique example of an acellular hard bioceramic wherein the mineral phase is sequentially assembled over the shell membrane. Information on the structureproperty relationship of eggshell matrix proteins and their role in the biomineralization process will increase our understanding of the molecular mechanisms of mineralization and enhance our capability to develop new materials. As a first step, we have purified various soluble organic matrix proteins to homogeneity from goose eggshells. The amino acid sequence of ansocalcin, the major matrix protein, was determined by automated Edman degradation of the S-pyridylethylated protein and from peptides obtained by chemical and enzymatic degradation. Ansocalcin has 132 amino acid residues comprising hydrophilic and hydrophobic domains. The sequence showed structural similarity to OC-17 from chicken eggshell matrix and CTL from snake venom (Fig. 4).
At lower concentration, ansocalcin exists in monomeric form (based on ESI-MS, Fig. 2E). As the concentration increases, it oligomerizes to dimers and trimers, as observed in the size exclusion chromatogram of ansocalcin at various concentrations. It is interesting to note that ansocalcin aggregates are stable in 2% SDS solution indicating that the protein-protein interactions are most likely through hydrophobic forces. The aggregation was not influenced by Ca 2ϩ ions at lower concen-trations (Ͻ100 g/ml). The presence of multimers severely hindered the conformational analysis. The appearance of a negative maximum at 230 nm in the CD spectra indicates the probable role of tryptophan-tryptophan interactions in the aggregation of ansocalcin (44). The observation that the wavelength of the emission maxima is 346 nm at all concentrations of ansocalcin indicates that tryptophan residues are exposed to the solvent (45). The presence of multimers in solution is further confirmed from the concentration-dependent fluorescence intensity at 346 nm (Fig. 12C). At lower concentrations, the fluorescence intensity increases linearly with concentration and is maximal at 250 g/ml. However, a significant decrease in intensity was observed in water and in CaCl 2 solution at higher concentrations of ansocalcin. The decrease was greater in water than in CaCl 2 solution (Fig. 12C). Such a reduction in the fluorescence intensity of ansocalcin (in the absence of an external quencher) may be due to the quenching of the tryptophan fluorescence emission by polarizable groups (e.g. -COOH, -COO Ϫ , -NH 2 , -SH, imidazole, and others) present in close vicinity (46). In CaCl 2 solution, calcium ions would bind to some of these polar groups, which reduces the fluorescence quenching and, thus, results in a higher fluorescence intensity.
It is interesting to note the similarities and differences of ansocalcin aggregation under both denaturing (SDS-PAGE) and non-denaturing (SEC) conditions. Gregoire et al. (47) reported that lithostathine protein, a CTL-like protein, in its proteolytic S1 form oligomerizes to dimers and tetramers even under denaturing conditions (47). Thus, we believe that ansocalcin aggregation behavior is similar to that of the S1 form of lithostathine, but ansocalcin forms dimers and trimers instead of tetramers at high concentration. Hattan et al. (48) observed that a glycoprotein purified from the extrapallial fluid of molluscan shell showed Ca 2ϩ ion-dependent oligomerization (48). However, our results from SEC, SDS-PAGE, CD spectra, and fluorescence studies show that the aggregation of ansocalcin does not require Ca 2ϩ ions.
Acidic and basic amino acid multiplets are present in soluble matrix proteins associated with calcium ions in biominerals (Table II and Refs. 49 -52). However, their exact role in the mineralization process has not yet been fully elucidated. Molecular modeling studies indicated that such repetitive arrangements might be responsible for the nucleation and growth regulation of crystals (50). The existence of acidic amino acid multiplets in the amino acid sequence of ansocalcin highlights its influence in CaCO 3 crystallization. In vitro crystal growth experiments showed that individual crystals transformed into polycrystalline aggregates as the concentration of ansocalcin or the eggshell extract was increased. Purified ansocalcin was about 5-10 times more effective than whole eggshell extract toward inducing the aggregation of calcite crystals (Figs. 6 and  7). The formation of spiral pits at lower concentration of anso- calcin (0.1 g/ml) indicates strong interaction between the acidic groups present in the protein and the growing crystal nuclei. Similar effects were observed in the presence of highly acidic polyaspartic acid molecules (53). As the concentration of ansocalcin was increased (Ͼ50 g/ml), calcite crystal aggregates were formed. The concentration-dependent aggregation of ansocalcin may provide templates for calcite crystal aggregates. At high concentrations of ansocalcin (i.e. in its aggregated state), there is a remarkable increase in the number of crystals nucleated (nucleation density) indicating its ability to trigger calcite crystal nucleation. From amino acid sequence, it is clear that ansocalcin possesses amphiphilic geometry comprising hydrophobic and hydrophilic domains. This amphiphilic character facilitates self-assembly, which can give rise to a hydrophilic exterior surface encompassing acidic and basic side chains. Such a feature may then assist specific interactions with the mineral phase accelerating the crystal nucleation. The decrease in fluorescence intensity and large changes in the CD spectra confirm this prediction. The lack of alignment between individual crystallites in the aggregate grown in the presence of ansocalcin (Fig. 7C) shows that the orientation of crystallites is not well controlled during crystal growth. According to Collier et al. (54) the formation of crystal aggregates in a suspension was controlled by two opposing factors: hydrodynamic forces and crystal growth rate. At lower levels of ansocalcin, fluid shear forces separate the crystals from each other through hydrodynamic forces. At higher concentrations, the growth rate dominates the hydrodynamic forces due to protein aggregation, and nucleation of the crystals takes place in a predefined space resulting in the formation of crystal aggregates. Such morphology is prevalent in rapidly mineralizing organisms such as calcareous algae, pennatulid sea fans, scleractinian corals, and other avian eggshells (55).
Overall, ansocalcin is the most abundant protein in goose eggshell matrix and induces formation of calcite crystal aggregates in the in vitro mineralization experiments. A comparison of the effect of ansocalcin and whole eggshell extract in CaCO 3 crystallization indicates that ansocalcin plays a vital role in goose eggshell calcification. The subtle differences observed between the crystals grown in the presence of ansocalcin and whole eggshell extract might be due to the presence of other proteins/peptides found in the eggshell extract. Furthermore, ansocalcin undergoes calcium-independent aggregation in solution, which may be responsible for the nucleation of calcite crystal aggregates. Further studies on the role of other proteins/peptides may help in the understanding of biomineralization in the eggshell as well as in the use of this as a tool toward the development of novel biomaterials.