Reference: Biol^ Biill^ 194: 304-316. (June. 1998) Oyster Shell Protein and Atomic Force Microscopy of Oyster Shell Folia C. S. SIKES' *. A. P. WHEELER'. A. WIERZBICKI', R. M. DILLAMAN', AND L. De LUCA' ' The Mineralization Center. Department ofBiological Sciences, and 'Department ofChemistry, University ofSouth Alabama. Mobile. Alabama 36688: -Department ofBiological Sciences. Clemson Universit}: Clemson. South Carolina 29634-1903: and ^ Department ofBiological Sciences. University ofNorth Carolina at Wilmington. Wilmington. North Carolina 28403-3297 Abstract. The organic layers within biominerals often also has increasingly attracted the attention of materials are viewed as sheets that may function in part to Hmit and scientists (Wheelerand Koskan, 1993; Mann, 1996: Stupp define theunderlyingcrystal structure, as well astopromote and Braun, 1997; Weiner and Addadi, 1997). For exam- formation of the next mineral layer. Some insights into ple, there is interest in nanoscale approaches to making the nature of the sheets were revealed by atomic force composite materials that have increased durability, ele- microscopy (AFM) ofsurfaces offreshlycleavedfragments vatedresistance to fracture, andotherdesirableproperties. ofoyster shell folia. Visible in the micrographs were arrays In shell, such properties are thought to be derived from of globular structures that resembled the globules seen in the interplay ofthe mineral phaseswiththe organic layers: isolated oyster shell protein bound to calcite. mica, and the latter are viewed mainly as sheets that may nucleate glass. The results of chemical treatment showed that the the original crystal phase, then regulate and limit its foliar globules slowly dissolved in 5.25% NaOCl or 1 A^ growth so that a specific morphology results. NaOH, reacted with an antibody prepared against an iso- The organic and inorganic phases ofthe carbonate bio- lated oyster shell protein, and were hydrolyzed by several minerals that occur in shells are well suited to atomic- proteolytic enzymes. These morphological and chemical level imaging by atomic force microscopy (AFM), as observations suggested that protein was a significant com- shown by a number of studies that probed the hard, flat ponent of the foliar globules. Although they might also mineral surfaces (Friedbacher et al. 1991: Hilner et ai. have a significant mineral content, the foliar globules were 1992: Drake et al.. 1992; Donachy et ai. 1992; Sikes et AFM not effective as nucleators of CaCO, crystal formation at al.. 1993). There also have been observations of low levels of supersaturation in artificial seawater. Overall, protein molecules isolated from calcite oyster shell, as the results suggested that molecules ofoyster shell protein well as ofpeptide analogs ofthe protein, bound to specific may agglomerate and combine with mineral to form a sur- surfaces of calcite (Wierzbicki et ai. 1994; Sikes and face ofcomplex topography that coats the calcite laths but Wierzbicki. 1995a, b). The isolated oyster shell protein exhibits no obvious correspondence to any specific crystal- had an ellipsoid, globular appearance when attached to lographic plane. calcite grown //; vitro and viewed by AFM. The polygonalcrystalline arraysofaragonitefrom mol- AFM Introduction luscan nacre also have been studied by (Giles et Information about the organization ofshell and regula- al. 1994, 1995; Manne et ai. 1994). The emphasis in tion ofits growth is not only central to biomineralogy but these smdies was on bleached biominerals, which thus were presumably free ofthe associated proteins. Although logically interpreted as composed of mineral only, many R*eTcoeivwehdo3m0cAoprrirlesp19o9n7d;enaccecesphtoeudld6 bMearacdhdre1s9s9e8d.. E-mail; ssikes® globules were observed that at least resembled the glob- usamail.usouthal.edu ules of isolated matrix proteins that we had reported, as 304 C. S. SIKES ET AL. 305 mentioned above. Along these lines, Gutmannsbauer and (Sikes and Wierzbicki, 1995a. b; 1996). Moreover, the Hanni (1994) viewed nacreous tablets by a variety of foliar globules were susceptible to enzymatic hydrolysis techniques and interpreted both the X-ray reflections and (Sikes et al.. 1997). the AFM images as showing an ordered layer of organic The purpose of the present investigation was to estab- globules that coats each tablet. Shell organic layers have lish the identity of the foliar globules more clearly as alsobeen described in detail in many electron microscopy proteinaceous, mineral, orperhaps acombination ofboth. studies over the years. Approaches included chemical, enzymatic, and immuno- For example, Watabe and coworkers studied transmis- logic treatments together with AFM and SEM observa- sion electron micrographs ofFormvarreplicas taken from tions. In addition, the possible function ofthe foliarglob- the whole, inner surface of oyster shell (1958), Formvar ules and their relationship to the adjacent mineral phases replicas of fracture surfaces of oyster shell (1961), and was investigated by AFM and SEM studies ofnucleation diamond-knife sections ofpieces offoliar layers ofoyster of calcite on foliar chips. The results suggested that the shell with and without a decalcification treatment ( 1965), proteinaceous layerofoystershell foliais itselfacompos- usually with silicon monoxide or carbon coatings. The ite ofundefined mineral and agglomerations ofindividual imagesrevealeddistinctrelationshipsbetweentheorganic protein molecules, which together form a layer of com- andinorganic layers, in somecases includingthe presence plex topography, rather than a linearized sheet, that over- of organic layers at the fracture surfaces. Although the lies the calcite laths. As nucleators ofcrystal growth, the spatial resolution ofthe surface topography was necessar- foliar globules were less effective than the areas of ex- ily affected by the limitations of electron microscopy, posed calcite that may occur on the foliar surfaces. No especially when replicated and coated specimens were obviousrelationshipofthe foliarglobules tospecific crys- examined, the clarity of the images was nonetheless ex- tallographic planes was found, although several possibili- cellent, making possible the detection of globular and ties were considered. "reticular" structures on the foliar surfaces. Assignments of spatial dimensions of organic layers were made only Materials and Methods of cross-sections and not of surface views; the assign- ments were based on the relative electron density of the Oyster shellfolia layers. By this approach, the width of the proteinaceous The outer surface of freshly shucked shells of the layer between adjacent folia was estimated to be 12 to American oyster, Crassostrea virginica. was ground us- 20 nm. ing a hand-held Dremel to remove residual periostracum Similarly, Taylor et al. (1969) and Carriker and co- and outer prismatica. The shell was then fractured with a workers (1979, 1980) used transmission and scanning hammer. White, pearlescent foliated chips were separated electron microscopy (SEM) ofcoated replicas, diamond- from pigmented and chalky chips and stored dry in a vial. wheel sections, and fracture surfaces in monographic The chips were several millimeters in linear dimensions; studies of oyster shell ultrastructure. Many high-resolu- they typically weighed between 20 and 30 mg, and did tion images of all layers of the shell were produced. The not exceed 100 mg. large inner layer of calcite sheets, or folia, was one area ofemphasis. Spatial relationships again were evident be- Chemical treatments tween the organic and inorganic constituents, with the observations and interpretations consistent with earlier Foliar chips were incubated in 10 ml of 1 A^ NaOH or studies. In a number ofthe images, globules are apparent 5.25% NaOCl (Clorox) in a glass vial for 3 weeks. The on the surfaces of the folia, but no particular attention solution was changed daily. At intervals, a chip was re- was drawn to them. moved with forceps, placed on a glass-fiber filter, and Previously, we reported AFM images offolia from the washed under gentle vacuum with 10 ml of 0.1 mM inner layer of oyster shell. Foliar chips, which are pro- NaOH. Next, the chip was soaked for lOmin, 3 times, duced in abundance when a shell is cracked open, evi- in 10 ml ofan aqueous solution saturated with respect to dently cleave mainly along the interfaces where the pro- calcite. This solution was the supernatant of a slurry of teins occur rather than through the mineral itself, and reagent-grade calcite crystal, 25 g/liter of water, stirred therefore might be coated with protein (Watabe, 1965; for 3 weeks. The chip then was placed on a piece of Taylor et ai. 1969; Carriker and Palmer, 1979; Carriker absorbent paper, air-dried for 1 h, and glued to a 12-mm et al., 1980; Kuhn-Spearing et al.. 1996). We had ob- glass disc by gently placing it on 10/j1 of 3;1 dichloro- served thatthe foliar surface contained globularellipsoids methane and commercial polyurethane (Minwax), just resembling the globules of isolated oyster shell protein after the dichloromethane had mostly evaporated, leaving imaged both in fluids and dry on calcite, glass, and mica a flat, nonwicking adhesive surface. This produced a firm. 306 AFM OF OYSTER SHELL FOLIA insoluble attachment to the glass disc, which had been buffered saline. pH 7.4). and then transferred to a 1:200 previously attached with superglue (cyanoacrylate) to an solution of the soluble matrix primary antibody. After electron-microscopy stub. Epoxy was also acceptable for 1 h, the chips were washed twice with TBS and trans- adhering the chips to glass, but superglue was not because ferred for 1 h to a 1:1200 solution of horseradish-perox- it dCiosnstorlovledchdiuprsinwgeraequiencouubsatiemdagiinncga.lcite-saturated water i(dSaisgem-ac)o.njAufgtaetredagofaitnalanTtiB-Schicwkaesnh,setchoendDarAyBanstyisbtodeym and washed as above prior to viewing. They also were (Sigma) was used to detect antibody binding to the foliar imaged directly, with no treatment, to ensure that the chips. Additional blocked (BSA-incubated) and un- soaking and washing did not affect the control morphol- blocked (no BSA incubation) chips, incubated with sec- DAB ogy. Some chips and control crystals ofcalcite also were ondary antibody only and developed using the sys- rinsed in distilled water rather than calcite-saturated wa- tem, confirmed that there was very little nonspecific bind- ter. This had no effect on the overall appearance of the ing of the secondary antibody to the foliar shell. Tfohleirae,fobruet,dfiodreAtcFhMtheimsaugrifnagceatofthteheatcoonmtircollecvreyls,tatlhse. ofM0a.t1cAh/edNfaoOliHarcohrip0s.0w8erMe pEreDtrTeAatesdolfourti2o4n h(wpiHthab1omult calcite-saturated solution was preferableforrinsing. How- 9.0) and then assayed as above. The EDTA solution was ever, if the saturated solution was not quickly absorbed prepared so as to dissolve no more than 10% ofthe 80 mg into the paper, the samples would occasionally exhibit of foliar shell. These treatments yielded ELISA results ectopic crystals that formed from drying droplets. that were nearly as dark as those for untreated chips, indicating that most sites for primary antibody binding Enzyiiuitic treatments were still available. The slightly diminished reactivity of the treated chips suggested that some binding sites had Foliar chips were incubated for 48 h in 0.5 ml ofphos- been alteredorremovedviadissolutionofproteinorshell. phate buffer (0.05 M. pH 7.5) that contained 1.95 units Foliar chips to be imaged by AFM were incubated in ofcarboxypeptidase B (Sigma), 2 units ofendoproteinase primary antibody only. glu-C (Boehringer-Mannheim). or 1.5 units of subtilisin (Boehringer-Mannheim). Carboxypeptidase is a general protease of peptide bonds at the C-terminus. Endopro- Atomicforce microscopy teinase glu-C and subtilisin both cleave internal peptide Constant-force, contact-mode AFM (Nanoscope III, bonds, particularly ofacidic residues, which are common Digital InstiTjments) was used to image foliar chips, both in the oyster shell protein. dry and in calcite-saturated artificial seawater (ASW: priTohretochgilpusinwgeornetoritnhseedglwaistshddiissctiallneddAwaFtMersatnudb.aCirondtrrioeld aagbaoivne,,tbhuetspurpeepranarteadntinof0.a5 MslurNrayClo.f 0ca.l0c1it1eMcrKysCtla.lspaHs chips were incubated in buffer alone. 8.3). Tapping-mode AFM was not needed for foliar im- ages because the imaged surfaces and adsorbates were Immiinohistochemical treatments firm and stationary to contact-mode AFM. However, tap- An ELISA assay was performed directly on foliar ping-mode images were obtained for the various treat- chips, using an antibody prepared against a48-kD protein ments and, in all cases, revealed morphologies similar to band obtained by SDS-PAGE of whole, soluble protein those obtained by contact mode. The images reported extracts of folia (Myers et ai. 1996). The electroeluted herein were all obtained by contact-mode AFM. Standard protein was mixed 1:2 with Freunds incomplete adjuvant tips of SisNj were used for contact mode and tips of and injected intramuscularly (three injections) into single- etched silica for tapping mode (both from Digital Instru- comb white leghorn hens to stimulate antibody produc- ments). The tip forces in contact mode were minimized at tion. Approximately 2 weeks after the final injection. IgG ~10 '' N; however, the samples were stable under higher was chloroform-extracted from egg yolks and purified forces as well. The AFM scanner was adapted for use of by ethanol precipitation (after Mohammed et ai. 1986). SEM stubs. Standard ELISA confirmed the reactivity ofthe extracted Procedures to guard against artifacts included variation antibody with the 48-kD antigen and demonstrated cross- of the scan angle to make sure that the image rotated reactivity of the antibody with other isolated oyster shell accordingly. In addition, for contact-mode imaging, tips matrix proteins (Johnstone and Wheeler, unpubl. data). were first used to obtain an atomic image of mica to be The antibody did not react with /3-lactoglobulin or BSA, sure that the tip was performing optimally before imaging which served as negative controls. a sample. This procedure was repeated after an imaging Foliarchips were incubated for 1 h in 1% BSA to block session, as well as during a session if questionable fea- nonspecific binding, washed in twochanges ofTBS (Tris- tures began to appear, to confirm that the tip was still C. S. SIKES ET AL. 307 capable of producing images at the angstrom level. Tips rapid, making real-time monitoring by AFM more diffi- that would not generate an atomic pattern of mica were cult. In addition to rapid crystal formation over large re- discarded. gions of the foliar chip, the turbid suspension of calcite Morphology at the micrometer level as seen by AFM crystals attheseconcentrations ofDIC interfered with the was corroborated by direct comparison ofAFM and SEM laser signal of the AFM. images of similar surfaces. This was more satisfactory than use ofseveral commercialand shareware deconvolu- Scanning electron microscopy AtiFonMpriomgagreasm.s Tthoesreemporvoegraanmystoifpt-ernelaytieedldfeedatiurmeasgefsroomf Foliar chips were dehydrated by vacuum at 1 mTorr for 15 min, then coated with a layer of approximately oystershell foliaorfields ofrandomlyorientedcrystalsof 20 nmofgoldpalladium inaPolaron sputtercoater. Spec- calcite that conflicted with images of the same materials imens were viewed with an ISI SX40 scanning electron viewed by SEM. microscope operated at 30 kV. AFM crystal growth assay Results Metastable and spontaneously nucleating solutions of Atomic force micrographs of dry, control, untreated calcite-supersaturated ASW were gently pumped through foliar chips of oyster shell are shown in Figures 1, 3, 5, a fluid cell (Digital Instruments, volume ~150^1) by and 6. The folia are composed of individual crystalline use of a peristaltic pump (Cole-Parmer) at a flow rate of laths lyingside-by-sidetoformasheet. Laths aretypically ~150 /.^1/min. The calcium concentration was lOmM, about 2 /vm in width, with lath heights in the imaged with total dissolved inorganic carbon (DlC) varied from areas varying from 150 to 350 nm. Figures 1 and 3 are 2 to 10 mM. The solutions were first prepared in a three- comparable in magnification to Figures 2 and 4, which necked, 50-ml round-bottom flask by addition of 30 ml are scanning electron micrographs of control, untreated of ASW, with smooth magnetic stirring. To this was foliar chips. Both Figures 1 and 2 are low magnifications M added 0.3 ml of 1.0 CaCl,•2H;0, followed by appro- that show the arrays of folia revealed by the fracturing M priate volumes of 0.5 NaHCO,. The pH was adjusted process. Figures 3 and 4 show individual folia and reveal to 8.30 with microliter amounts of 1 A' NaOH and moni- randomly broken ends, straight foliar margins, andagran- tored by pH electrode and meter (Fisher 911) equipped ular surface texture. Figures 5 and 6 are highermagnifica- with a strip chart (Cole-Parmer). The apparatus was not tion AFM images that resolve the surface texture, identi- thermostated, but room temperature was recorded at 23° fied in the lower magnification images from both SEM ± 2°C. At the initiation of the experiment, the fluid from and AFM, into a continuous layer of discrete globules. the flask was pumped through the flow-cell of the AFM. These foliar globules typically were about 10 to 15 nm The foliar chip was already in place and being imaged in in height, but ranged up to about 40 nm. calcite-saturated ASW. The susceptibility of the foliar globules to dissolution In separate experiments, the metastability of the fluids in 1 A' NaOH is shown in Figure 7. The foliar globules was demonstrated by growth of 15 mg of CaCOj seeds, were seentobe relatively resistanttothe NaOH treatment addedtothe flaskas 1.5 ml ofa 10 mg/ml seed suspension at 5 days. However, by 10 days (not shown), the globules in saturated ASW, The primary stock ofcalcite seeds was were reduced to remnants with indistinct edges and preparedby stirring 100 gofreagent-gradecalcite (Baker) heights of 1 to 3 nm instead ofthe typical globularheight Iinn 1th1eofabAsSenWcefoorfasteleedastcr3yswtaelesk,sat(W2heteol5ermeMt alD..IC1,99t1h)e. ofTahbeouteff1e0cttoof15NanOmClontcroenattrmoelntfoolniarthseurffoalcieasr.globules solutions exhibited no spontaneouscrystal formationover as viewed by AFM is shown in Figures 8 and 9. After 1 periodsofat least 2 h ofmost imaging sessions. However, day, the globules remained relatively intact. By 5 days, these solutions were shown to be supersaturated and were the globules were blurred andreduced in height to arange defined as metastable because crystal growth did occur of 3 to 8 nm. In both the NaOH and NaOCl treatments, in the presence ofseed crystals. The growth was observed the surfaces became smoother and relatively featureless as a downward drift in pH resulting from incorporation with time. of COj"^ ions into the seed-crystal lattice. Treatment of foliar chips for 48 h with the proteolytic At a concentration of 7 mM DIC at pH 8.3, the calcite enzymes (carboxypeptidase B, endoproteinase glu-C, and crystals did spontaneously nucleate after an induction pe- subtilisin) also led to the partial removal of the foliar riod of 15 to 20 min. At concentrations of DIC higher globules. In each case, as illustrated for carboxypeptidase than 7 mM, the induction of crystal growth, again moni- B (Figs. 10 and 1 1 ), the globules became indistinct and tored by downward pH drift, became increasingly more reduced in height relative to control globules. 308 AFM OF OYSTER SHELL FOLIA Figures 1-6. Micrographs ofunireated (control) chips ofoyster shell folia. C. S. SIKES ET AL. 309 The surfaces of foliar chips that had been treated with than only mineral material. That is, the globules exhibited the antibody to the oyster shell protein were coated with (1) slow dissolution in NaOH and NaOCl, (2) reactivity globular material as seen by AFM (Figs. 12 and 13). to an antibody specific to oyster shell protein, (3) partial These globules of immunoglobulins were 4 to 5 times as hydrolysis by proteolyticenzymes, and (4) morphological large as the untreated foliar globules (compare Figs. 6 similarity to ellipsoids and globules of isolated protein and 13). from oyster shell as observed in prior studies. AFM and SEM images ofafoliarchip on which calcite The correspondence in size and appearance between crystals were grown can be compared (Figs. 14-16). In the foliar globules and the isolated molecules of the pro- both types of micrographs, the newly formed crystals tein was evident in AFM images that were prepared in a appeared to be very smooth and to emerge from the foliar variety of ways. For example, the isolated protein was surface at an oblique angle. Becausetheseectopic crystals bound to calcite and viewed by AFM both in fluids and were not coated with protein, the atomic pattern of the on rinsed, then dried crystals. The protein exhibited ellip- AFM lattice surface could be resolved by (Fig. 17). soid and globular morphologies with lengths and widths The spacings and angles between the positions ofhun- in the range of30 to 100 nm, similarto those ofthe foliar dreds of atoms of the surface of crystal growth were globules (Donachy et al.. 1992; Wierzbicki et al.. 1994; measureddirectlyon several imagesbymanualplacement Sikes et al.. 1994). The isolated protein also exhibited of the cursor onscreen and use of the measurement tools comparable moiphologies when viewed either bound on ofthe software. In addition, Fourier analysis was applied glass and mica in fluids or when dried onto these sub- totheaverageperiodicitiesoftheentireimages. TheAFM strates (Sikes and Wierzbicki, 1995a, b; 1996; Sikes et software readily supplies measurements by both of these aI.. 1997). approaches. The protein used in the prior studies was obtained as The measured atomic spatial relationships were then a distinct, reversed-phase peak from the EDTA-soluble, comparedto the theoretical spatial relationshipsgenerated proteinaceous matrix of the shells (Wheeler et al.. 1988; bycomputermodelsofvariouspossible surfacesofcalcite Wheeler and Sikes, 1989; Rusenko et al.. 1991). The (Cerius", molecular modeling software. Molecular Simu- peak is polydisperse, with an estimated gel-permeation lations, Inc.). A similarapproachhasbeen helpful in iden- molecular mass of approximately 50 kD. The protein is tifying other calcite surfaces such as the (104) cleavage anionic with about 30% of the residues being aspartate surface of control calcite rhombohedrons and the (001) and nearly 30% being serine, much ofwhich is phosphor- basal plane of calcite nucleated on glass (Sikes et al.. ylated. 1994), as well as the (1 -1 0) surface ofcalcite crystals If the foliar globules are proteinaceous in part, they that were stabilized by the presence of polyaspartate would of necessity seem to be agglomerations. That is, (Sikes and Wierzbicki, 1996). the AFM volume (4/3 k abc) of the foliar globular ellip- The first identifiable surface of the ectopic foliar crys- soids of about 100 nm length {a = 50 nm), 50 nm width tals appeared to be the (1 -1 0) plane of calcite. The (b = 25 nm), and 10 nm height (c = 5 nm) is 2.62 X AFM atomic image (Fig. 17) wascompared to acomputer 10"'^cm\ The theoretical molecular volume (4/3 tt r") model of the (1 -10) surface (Fig. 18). The measured ofa globularprotein ofthe size ofa typical soluble oyster spacings and angles between the various atomic positions shell protein (M^, 50 kD, diameter —5.4 nm; Cantor and matched within 5% of the theoretical values in all cases, Schimmel, 1980) is about 8.24 X 10"-Vm'. Comparison as explained in the legend of Figure 18. ofthese values,assumingforthemomentthattheglobules Discussion areentirely protein, yields an estimate ofabout 318 mole- The results supportedthe identity ofthe foliar globules, cules ofoyster shell protein per globule. Lower estimates at least in part, as molecules ofoyster shell protein, rather would result if larger proteins of the shell matrix were Figure 1. Atomic force micrograph ofadry. untreatedchip. Range oflath heights: 150-350nm. Scale bar = 15 /jm Figure 2. Scanning electron micrograph ofan untreated chip. Scale bar = 25jxm. Figure3. Atomic force micrographofadry, untreated chip. Range oflath heights: 150-250nm. Scale bar = 2/jm. Figure 4. Scanning electron micrograph ofan untreated chip. Scale bar = 2^tm. Figure5. Atomic force micrographofadry, untreated chip.Total rangeofelevation within the imaged area = 70nm. Scale bar = 1 /jm. Figure 6. Atomic force micrograph of a dry, untreated chip. Typical globular height = 10 to 15nm: maximum globular height in the imaged area = 40nm. Scale bar = 250nm. 310 AFM OF OYSTER SHELL FOLIA Figures 7-12. Atomic force micrographs ofdry chips ofoyster shell folia treated with vanous sub- stances. C. S. SIKES ET AL 311 considered. For example, the proteins from oyster shell of hypochlorite-resistant, "calcified"" organic matrices exhibit a continuum ofmolecularweights that ranges into associated with the surfaces of nacreous tablets of mol- the millions for soluble fractions. Also present are insolu- luscan shell, and Towe (1990) commented on the resis- ble fractions with an amino acid composition similar to tance to household bleach of some matrix-like organic that of the soluble proteins (Wheeler el al.. 1988). molecules, particularly ifthey were intimately associated Phosphoproteins similar to the oyster shell protein are withthe mineral.Thus itseemsthat, in addition toprotein, known to form micellar agglomerations in solution. For the foliar globules may contain a phase of mineral salts example. Marsh (1989a. b) reviewed the associative be- and perhaps water, as discussed below. havior of casein and other phosphoproteins and demon- One assignment of the relative amounts ofthe mineral strated that the phosphophoryn from tooth dentin forms and organic phases of shell can be made by quantifying agglomerations that are held together via ionic interac- the weight of each component. Another assignment can tions with cations s—uch as calcium and magnesium. Ph—os- be made by comparing the volumes of each layer, taken phophoryn, which like the oyster shell protein is from the linear dimensions ofeach phase as seen in both highly enriched in phosphoserine and aspartic acid, SEM and AFM images, correcting for density differences formed particles of about 25 nm in diameter that con- of the mineral and the proteinaceous material. The ob- tained perhaps 75 monomers per particle. served linear dimensions as determined by SEM and AFM and TEM observations by Fincham et al. (1994, AFM, although subject todifferent kindsofartifacts, were 1995) of 15-20 nm "nanospheres" on the calcium phos- in agreement, lending credence to the estimates. phate surfaces ofdeveloping enamel have also been inter- The protein content of oyster shell has been variously preted as proteinaceous, composed of amelogenin pro- measured as ranging from perhaps a few tenths of 1% to teins. The amelogenins are smaller (~20kD) and more no more than 3% by weight in whole shells (Korringa, hydrophobic than the typical oyster shell protein. The 1951: Weiner and Hood, 1975; Price et al., 1976; Ru- amelogenic nanospheres are thought tobe agglomerations senko, 1988; Rusenko et al.. 1991 ), with the protein con- of greater than 100 monomers. tent of the foliar layers alone placed at <1%. A lower Globular ellipsoids in the range of 100 nm have also protein content for foliar layers is consistent with the been observed on the surfaces of CaCOj otoconia from electron microscopic observations, all of which revealed the inner ear of a newt (Hallworth et al.. 1995). The that the foliar layer lacks the thicker, "interlamellar"" otoconiawere isolated intact and are known to have asso- organic layer of the prismatic regions of shell. ciated proteins. Observed with AMF, the otoconial sur- Given this range ofvalues reported for protein content, faceglobules were quite like the foliarglobules in appear- an analysis ofthe apparent, relative volumes of "nonmin- ance and were attributed to crystal formation as mediated eral" and mineral layers in a foliar lath as seen in both by the otoconial proteins. electron micrographsand atomic forcemicrographs yields Ourresults indicate thatthefoliarglobulesareprobably an estimate of nonmineral content that is too high to be not composed solely ofprotein. The comparative images attributable only to organic matter. For example. Watabe revealed that the foliar globules were generally taller at and Wilbur (1961) and Watabe (1965) in electron micro- 10 to 40 nm than the ellipsoids and globules of isolated scopic studies observed the laths to be composed ofcrys- protein on calcite, mica, and glass, which had heights tal blocks that were surrounded by "intercrystalline"" or- generally in the range ofa few nanometers. Furthermore, ganic material. Thedimensions ofeach block ranged from the foliar globules resisted dissolution in NaOCl and 10 to40 nm in width, 15 to 200 nm in height, and ~4 /jm NaOH. Mutvei (1977, 1978) also reported the presence in length. The dimensions of the organic matrix ranged Figure 7. Treatment: 1 NNaOH for5 days. Total range ofelevation within the imaged area = 80nm: range ofglobular heights = 4 to 30nm. Scale bar = 500nm. Figure 8. Treatment: 5.25% NaOCl for 24 hours. Total range ofelevation within the imaged area = 18nm: heights ofglobular remnants = 3 to 6nm. Scale bar = 250nm. Figure 9. Treatment: 5.25% NaOCl for 5 days. Total range of elevation within the imaged area = 16nm: heights ofglobular remnants = 2 to4nm. Scale bar = 250nm. Figure 10. Treatment: Carboxypeptidase B for 48 hours. Total range of elevation within the imaged area = 30nm. heights ofglobular remnants = 3 to 5nm. Scale bar = 1 ^m. Figure 11. Treatment: Carboxypeptidase B for 48 hours. Total range of elevation within the imaged area = 15 nm. Scale bar = 250nm. Figure 12. Treatment: An antibody to the isolated oyster shell protein. Foliar laths are obscured by copiously bound antibody molecules. Scale bar = 5 /im. 312 AFM OF OYSTER SHELL FOLIA "^i Figures 13-18 C. S. SIKES ET At. 313 from 12 nm to 20 nm, as also later recorded by Taylor et This convolution phenomenon is not a simple function al. (1969). These values for the dimensions both of the and does not always occur: as demonstrated by the com- crystal blocks and of the organic layers are similar to the parison of the micrographs produced by SEM and AFM, dimensions of the crystal laths and the foliar globules AFM showed true foliar morphology. Similarly, AFM AFM observed by and recorded herein. micrographs of peptides bound to calcite (Wierzbicki et Based on average values of these dimensions and a al.. 19^94: Sikes et al.. 1993. 1994. 1997)—and of various simple calculation ofthe volume ofthe crystal block and other molecules bound to flat substrates for example, the volume ofthe encapsulating proteinaceous layer, cor- DNA —to mica (Hansma and Hoh, 1994: Hansma et ai, rected for the density ofcalcite (2.71: Weast et ai. 1988) 1995) revealed molecular morphologies that were con- — and typical proteins ( 1.35; White et al.. 1973). the pro- sistent with theory. tein content would be about 17% by weight, which of The height measurements at the top of an elevation, course is too high. A plausible explanation for the high particularly of Hrm surfaces such as the foliar globules, AFM estimate is that the foliar globules seen with and are considered to be generally reliable, depending on the the organic layerseen in electron micrographs might con- method of standardization. We calibrate our scanners at sist of both protein and mineral, with the latter filling in the micrometer level by use of commercial standards spaces around and between the molecules of protein. (Digital Instruments) and at the angstrom level by use of Another possible complication is that actual volumes well-characterized crystals such as mica for the .v and v of the foliar globules and the AFM volumes might be axes, and calcite [the step height of the cleavage surface different, as sometimes occurs in AFM studies. The dis- (104)] for the z axis. crepancy is most likely to reflect an overestimate of the Overall, the results and analysis suggest that the foliar length and width of molecules. This kind of tip artifact globules are composed in significant part of protein. An results when the inverted pyramidal AFM tip slides up undefined mineral phase that fills in the spaces around and down the sides of objects that are actually sharp. and within the foliar globules also may be a significant giving a smoothed and widened appearance (Hansma et component, along with water, gases, and other mineral al.. 1995: Giles efa/.. 1994. 1995) and perhapsexaggerat- salts as minor components. In fact, Galtsoff (1964) ing the length or width by about 20%. viewed initiation ofoyster shell growth as crystallization Figure 13. Atomicforcemicrographofadrychipofoystershell foliathatwastreatedwithan antibody to the isolated oyster shell protein. Total range ofelevation within the imaged area = 150nm (compare to Fig. 6. a control chip al the same magniticalion). Scale bar = 250nm. Figure 14. Atomic force micrograph of calcite crystals that were nucleated on an untreated chip of oyster shell folia at 10mM Ca"*, 7 mM inorganic carbon, initial pH 8.3. in artificial seawater (imaged in this fluid). Height of the central ectopic crystal = 340nm. with a plane angle <10° between the crystal surface and the underlying foliar surfaces. Scale bar = 2.5/jm. Figure 15. Scanningelectronmicrographofcalcitecrystalsthatwerenucleatedontheoystershell foliar surface of Figure 14. Scale bar = 3^m. Figure 16. Scanningelectronmicrograph ofcalcitecrystalsthat werenucleatedontheoystershell foliar surface ofFigure 14. Scale bar = 2/jm. Figure 17. Atomic force micrograph ofthe lattice atoms ofthe surface ofthe central crystal ofFigure 14, imaged in fluid as above. Total range ofelevation within the imaged area = 3 A. Scale bar = 5 nm. Figure 18. Computer diagram of the (1 -10) surface of calcite showing calcium atoms (grey) and carbonate groups (with white oxygens) in alternating rows. The spacing between lattice positions in the rows (diagonal upward, left to right) ofboth calcium atoms and carbonate groups is 4.99 A. The spacing between lattice positions perpendicularto the rows (that is, between acalcium atom and the nearest oxygen ofan adjacent carbonate group) is 4.265 A. Both ofthese distances matched the spacings observed in the AFM ofFigure 17. The computer model showed that an oxygen ofeach carbonate group is elevated ~1 A relative to the plane ofcalcium atoms, and therefore the most distinct and separate atoms ofFigure 17 are thought to be oxygen atoms. The specific carbonate oxygens that are most protuberant in each row differ in alternating rows, creating a slight zig-zag pattern (the oxygens ofalternating rowsdo not line upexactly along the diagonal from upper left to lower right). This is visible in both the image of Figure 17 and the model. Similarly, the lower atoms were measured in Figure 17 to be in agreement with theory at ~1 A below the plane of the higher atoms, and thus are thought to be calcium atoms that blended together somewhat in the image, probably owing to the interaction between the AFM tip and the lattice during imaging in the troughs. The agreement between these and other features of the experimental image (Fig. 17) and the theoretical model (Fig. 18) lead to the assignment of the surface ofthe crystal as most likely the (1 —1 0) surface ofcalcite.