AmericanMineralogist, Volume 83,pages 1510-1515,1998 Primary structure of a solublematrix protein of scallop shell: Implications for calcium carbonatebiomineralization I. SlnasurNA ANDK. ENoox GeologicalInstitute, University of Tokyo, Tokyo I l3-0033, Japan Ansrnacr Soluble proteins in the scallop (Patinopectenyessoensis) foliated calcite shell layer were characterizedusing biochemical and molecular biological techniques.SDS PAGE of these molecules revealed three major protein bands, 97 kD, 12 kD, and 49 kD in molecular weight, when stained with Coomassie Brilliant Blue. Periodic Acid Schiff staining and Stains-All staining indicated that these proteins are slightly glycosylated and may have cation-binding potential. N-terminal sequencingof the three proteins revealedthat all three sharethe same amino acid sequenceat least for the first 20 residues.A partial amino acid sequenceof 436 amino acids of one of these proteins (MSP-l) was deduced by charac- terization of the complementary DNA encoding the protein. The deduced sequenceis composed of a high proporton of Ser (3l%o), Gly (25Vo),and Asp (20Vo),typifying an acidic glycoprotein of mineralized tissues. The protein has a basic domain near the N- terminus and two highly conserved Asp-rich domains interspersedin three Ser and Gly- rich regions. In contrast with prevalent expectations,(Asp-Gly)n-, (Asp-Ser)n-, and (Asp- Gly-X-Gly-X-Gly)ntype sequencemotifs do not exist in the Asp-rich domains,demanding revision of previous theories of protein-mineral interactions. INrnooucrroN (l) induction of oriented nucleation (Weiner 1975;Weiner Minerals produced by organisms often have crystal and Addadi l99l); (2) inhibition of crystal growth shapesclearly different from those formed inorganically. (Wheeler et al. 1981; Wheeler 1992); (3) control of ara- Most such biominerals are a compositeof inorganic crys- gonite-calcite polymorphism (Falini et al. 1996; Belcher tals and organic molecules such as lipids, polysaccha- et al. 1996); and (4) enhancementof mechanical prop- rides, and proteins, collectively known as the organic ma- erties of the crystals (Berman et al. 1988; Berman et al. trix. It is generally postulated that the elaborate 1990). Most of the evidence to support these hypotheses fabrication of biominerals arisesfrom specific molecular has been obtained through in vitro experiments. The interactions at inorganic-organic interfaces (Mann et al. weaknessof these studies is that unpurified proteins or 1993), and that the organic matrix representsmany of the protein fractions of dubious homogeneity have been ap- important molecules involved in the interactionscontrol- plied in the biochemical analysesand in the in vitro min- ling crystal growth (e.g., Watabeand Wilbur 1960; Low- eralization experiments, and that the stereochemicalre- enstam 1981; Weiner 1984; Lowenstam and Weiner lationships between the organic and inorganic phases 1989). have been presumed without precise information of the Calcium carbonate is one of the most common bio- fine structuresof the proteins. minerals, and its matrix molecules, especially of mollus- To understandthe underlying mechanismsof the pro- can shells, have been studied to a considerableextent to tein-mineral interactions,it seemsessential first of all to unveil their roles in the mineralizationprocesses. The ma- know the primary structure of the proteins involved. ffix molecules have been classif,ed conventionally into However, only a limited number of amino acid sequences two types based on their solubility in aqueoussolutions: have been determined so far for the calcium carbonate the insoluble matrix is thought to be largely intercrystal- matrix proteins. The available sequencescomprise those line (Krampitz 1982) and provides a framework where from spicules of sea urchin emryo (Sucov et al. 1987; mineralization is to occur, whereas the soluble matrix is Katoh-Fukui etal. l99l; Katoh-Fukui etal.1992; Benson known as intracrystalline or located on the intercrystal- and Wilt 1992), from pearl oyster shell layers (Miyamoto line matrix surfaces,but its functions are still poorly un- et al. 1996; Sudo et al. 1997\, and from the nacre of a derstood (Addadi and Weiner 1997). gastropod shell (Shen et al. 7997), in addition to partial Advocated functions of the mainly proteinaceous,sol- sequencesfrom brachiopod shells (Cusack et al. 1992), uble matrix of the molluscan shell in particular, include: an oyster shell (Wheeler 1992), and gastrolith of a cray- fish (Ishii et al. 1996, Ishii et al. 1998). Here we present * E-mail: endo @geol.s.u-tokyo.ac jp a partial amino acid sequenceof the molluscan shell pro- 0003-004x/98/1112-15 l0$05.00 1510 SARASHINA AND ENDO: MATRIX PROTEIN OF BIO-CALCITE l5l I tein MSP-I, a major soluble matrix protein of the scallop quired to design unique primers to facilitate subsequent foliated calcite shell layer, and discussits bearing on the amplification of the DNA sequenceencoding the entire functions of matrix proteins in calcium carbonate proteln. biomineralization. We extracted the total RNA from the mantle tissue of a single specimenof P. yessoensis,using ISOGEN (Nip- MlrBnrlr,s AND METHoDS pon Gene) and the single-stepmethod for RNA isolation Isolation of MSP-I (Chomzynski 1993). The RNA (5 pg) was applied as a Specimens of the commercial scallop Patinopecten template for reversetranscription to preparecomplemen- yessoensiswere purchasedlocally. A single shell valve tary DNA (cDNA) in a 20-p,L reaction, primed with a was thoroughly cleaned mechanically and incubated for "hybrid" primer, TCGAATTCGGATCC-GAGCTC(T ),,, 48 h at room temperaturein a l0 volVosolution of sodium using the SuperScriptpreamplification system(Life Tech- hypochlorite to desffoy surface contaminants.After thor- nologies). The target cDNA sequence,encoding the N- ough washing with ultrapure water, the marginal portion terminal sequenceof 20 amino acid residuesdetermined of the shell, consisting only of the outer shell layer of for the purified MSP-I, was amplified by a method foliated calcite, was crushedto fine fragments.The matrix known as reverseffanscription-polymerase chain reaction proteins were extractedby dissolution of the shell flakes (RT-PCR).The senseprimer and the antisenseprimer cor- (100 g) in 3 liters of 0.5 M EDTA (ethylenediaminete- responded to the sequence encoding LDTDKD and traacetate),pH 8. The extraction was performed at 4 'C NAAED (for one-letter abbrevation of amino acids, see with continuous stirring for 12 h. The preparation was the caption for Fig. 1), respectively, each being degen- then filtered through cheeseclothto remove viscous in- erate, containing oligonucleotides of all the possible se- soluble materials and desaltedby ultrafiltration using the quencesfor each amino acid sequence.The reaction mix- Minitan tangential flow system (Millipore). In this pro- ture (50 pL) contained2 pL cDNA, 2 p.M of eachprimer, cedure, the amount of EDTA was reduced to less than 1 x Taq DNA polymerasebuffer (Life Technologies),3 l0 6 mol, at which point the sample was concentratedto mM MgC,,,100 p,M dNTB and I unit of Taq DNA poly- about 50 mL, then lyophilized. merase(TOYOBO). A Cetus DNA Thermal Cycler (Per- kin Elmer) was employed with an initial step of 94 "C SDS PAGE analyses for 3 min, then 30 cycles at 94 "C for 30 s, 52 "C for 30 The extracted macromolecules (1 mg per each well) s,72'C for I min, followed by a final extensionstep of '72 were separatedby SDS PAGE (Laemmli 1970), using "C for 5 min. The resulting PCR products of 60 base slab gels (15 x 15 cm) of 2 mm thickness containing pairs (bp) in length (correspondingto 20 amino acids), l27o polyacrylamide. After elecffophoresis, gels were were sequenceddirectly by the chain termination method stained by Coomassie Brilliant Blue R or silver (Silver using the BigDye Terminator Cycle SequencingKit and Stain kit; Bio-Rad) to visualize proteins, by Stains-All to an automated DNA sequencer (Perkin-Elmer Applied visualize cation-binding proteins (Campbell et al. 1983), Biosystems). and by periodic acid and Schiff's reagentto visualize car- bohydrates(Holden et al. l97I). Amplffication and sequencingof cDNA 3'-end Threegene-specific sense primers,Pl,P2, and P3 (Fig. 1), designed based on the sequencedetermined by the N-terminal sequencedetermination above method, and the antisense"adaptor" primer (PA), Following separationby SDS PAGE, the proteins were TCGAATTCGGATCCGAGCTC, were synthesized for electroblotted onto polyvinylidene difluoride membrane the PCR amplification of the region between the point (ProBlott; ABI) in Caps buffer (10 mM, pH l l) contain- correspondingto the N-terminal end of the mature protein ing methanol (10 volTo solution), prior to staining with (MSP-l) and the 3'-end of the transcript(3' RACE: rapid Coomassie Brilliant Blue R. N-terminal amino acid se- amplification of cDNA ends protocol; Frohman 1990). quence analysis of the immobilized protein sampleswas The reactions using the primer pair of PI-PA were per- by Edman degradation using an automated protein se- formed first, under conditions similar to those described quencer (Perkin-Elmer Applied Biosystems). Sequences above,except that the reactionswere catalyzedby Ex Taq were determined at least twice for each protein band re- (TAKARA) in this case. A second and a third round of produced by different SDS PAGE gels. PCR reactions were performed with the P2-PA and P3- PA primer pairs, respectively,using the PCR products of RNA purification and RT-PCR the previous round of reactions as a template to