Silica precipitation induced by silaffins ( 14 ). ( A ) Low molecular mass fraction of HF ex- tract from C. fusiformis cell walls. The extract was subjected to Tricine–SDS-PAGE ( 13 ) and stained with Coomassie blue. ( B ) Correlation between silaffin concentration ( 27 ) and the amount of silica precipitated from a silicic acid solution. The dotted line represents the result obtained for the silaffin mixture; the solid line shows the result for pure silaffin-1A ( 15 ). ( C and D ) SEM micrographs of silica precipitated by silaffin-1A (C) and the mixture of silaffins (D). The diameter of silica particles is 500 to 700 nm (C) and Ͻ 50 nm (D). The protein concentration was 5 mg/ml. Bar: 1 ␮ m. 

Context in source publication

Context 1

... chemical synthesis of silica-based materials like resins, molecular sieves, and catalysts requires extremes of temperature, pressure, and pH. In contrast, biosilicification proceeds at ambient temperatures and pressures, producing an amazing diversity of nanostructured frame- works ( 1 – 4 ). The silica cell wall of diatoms consists of two overlapping valves, and their structure is precisely controlled by the cell. During cell division a new valve is formed in minutes by controlled precipitation of silica within a specialized membrane vesicle called silica deposition vesicle (SDV) ( 5 ). Electron microscopic evidence indicates that the silica is initially deposited in the form of nanoscale spheres, suggesting the presence of components within the diatom cell that control silica sphere formation ( 6 , 7 ). However, the molecular structure of these components has remained elusive. It is well established that amorphous silica in diatom cell walls is intimately associated with organic substances that have been hypothesized to act as regulating molecules in biosilicification ( 8 – 10 ). We examined the protein composition of cell walls from the diatom Cylindrotheca fusiformis (Fig. 1). Extraction of purified cell walls with EDTA led to the characterization of a Ca 2 ϩ binding protein family called frustulins ( 11 ). Even after harsh extraction procedures with boiling SDS solutions the resulting silica prep- aration still contains tightly bound organic material that is recovered only after solubilizing silica with anhydrous hydrogen fluoride (HF). This HF-extractable material consists of two main protein fractions: a high molecular mass protein family, termed HEPs, that is localized at a specific substructure of the cell wall ( 12 ), and a low molecular mass fraction with apparent masses ranging from 4 to 17 kD. Here we show that the latter (poly)peptides have affinity to silica; they are thus named silaffins. Silaffins appear to be the most abundant protein component within the HF extract of the cell wall, contributing about 50 ␮ g per milligram of dry weight of EDTA-SDS–extracted cell wall material. Using Tricine-SDS polyacrylamide gel electrophoresis (PAGE) ( 13 ), silaffins fraction- ate into three components (Fig. 2A). NH -ter- 2 minal amino acid sequencing indicates a high degree of homology between the 4-kD component, termed silaffin-1A [NH -terminus in sin- 2 gle-letter amino acid symbols: SSXX Ј SG- SYSG(S/Y)], and the 8-kD component, termed silaffin-1B (NH -terminus: SSXX Ј SGSYYSY- GT). The fact that serine and tyrosine are simul- taneously present in amino acid position 11 of silaffin-1A indicates that this component represents a mixture of almost identical polypeptides. Edman degradation of silaffin-1A and -1B, respectively, did not produce any signal at position 3 (X), whereas an unknown amino acid derivative was obtained at position 4 (X Ј ). NH 2 terminal sequencing of silaffin-2 mainly produced unidentified phenylthiohydantoine (PTH) derivatives, indicating the presence of a different amino acid sequence with an even higher degree of posttranslational modification. Each of these silaffin species is able to precipitate silica within seconds when added to a freshly prepared solution of metastable silicic acid. The silicic acid solution without added silaffins did not form any precipitate and remained homog- enous for at least a few hours. The amount of precipitated silica is proportional to the amount of silaffin applied (Fig. 2B). At any protein concentration, silaffins completely coprecipitate with the silica as long as silicic acid is present in excess. In the precipitate, the molar ratio silica/ silaffin-1A is about 12 ( 14 , 15 ). As shown by scanning electron microscopy (SEM) (Fig. 2C), the silaffin-1A–induced precipitate is composed of a network of nearly spherical silica particles with diameters of 500 to 700 nm. Within the networks neighboring silica spheres are closely attached to each other or partly fused. When the unfractionated mixture of silaffins (Fig. 2A) was used for precipitation, aggregates of much smaller silica particles (diameters Ͻ 50 nm) were obtained (Fig. 2 D). Using sequence information from the NH - 2 terminus of silaffin-1B, we amplified by poly- merase chain reaction (PCR) a corresponding cDNA fragment from a C. fusiformis cDNA library. This fragment was then used to screen a C. fusiformis genomic library, which led to the identification of a 795– base pair (bp) open reading frame, termed sil1 , encoding a polypeptide (sil1p) of 265 amino acids (Fig. 3). Amino acid residues 1 to 19 represent a typical signal sequence followed by a peptide sequence (residues 20 to 107) highly enriched in acidic amino acid residues. The remaining COOH-termi- nal part of sil1p has a striking repetitive structure. It is composed of seven highly homolo- gous repeating units (R1 to R7) containing 33 (R1), 22 (R2), and 19 (R3 to R7) amino acids, respectively. Basic amino acid residues predominate throughout this part of sil1p, particu- larly Lys-Lys and Arg-Arg clusters, which are spaced in a highly regular manner. In between these clusters, the hydroxy-amino acids serine and tyrosine predominate. The NH -terminal 2 sequence of silaffin-1B exactly matches amino acids 108 to 120 of repeat unit R1, with both X and X Ј representing lysine residues. Also, the NH -terminal sequences of silaffin-1A are con- 2 tained within sil1p, being identical to the first 11 amino acids of each of the repeats R2 to R7, with X and X Ј again representing lysine residues. Thus, comparison of sil1p sequence with the NH -terminal sequences of silaffin-1A and 2 -1B led to the following conclusions. First, silaffin-1A and -1B originate from endoproteo- lytic processing of sil1p. This has been further confirmed by experiments described below. Second, the unidentified amino acid residues X and X Ј in the NH 2 -terminal sequences of silaffin-1A and -1B represent posttranslationally modified lysine residues. To elucidate these modifications, we again subjected silaffin-1A to Edman degradation. The unknown lysine derivative liberated in cycle 4 (denoted X Ј ) was then analyzed by elec- trospray ionization mass spectrometry (ESI- MS). A single peak of mass ( m ϩ H) ϩ ϭ 292 was detected. This mass exactly matches the value calculated for the PTH derivative of ε - N , N -dimethyl-lysine. Collision-induced fragmentation of this compound by tandem mass spectrometry (MS/MS) confirmed this conclu- sion, as the same fragment ion pattern was obtained when compared with authentic - N , N - dimethyl-lysine ( 16 ). In contrast, no lysine derivative was detected in cycle 3 (denoted X) during Edman degradation of silaffin-1A. Therefore, a different approach was selected to identify this lysine derivative. Chymotryptic peptides derived from silaffin-1A were fractionated by reversed-phase high-pressure liquid chromatography (HPLC), which allowed the isolation of six peptides eluting at about 12% acetonitrile ( 16 ). All six peptides had the amino acid sequence SSXX Ј SGSY but different molecular masses. These masses differed by multiples of 71, indicating the presence of a repeated unit oligomer attached to these peptides. Complete acid hydrolysis of silaffin-1A (6 N HCl at 110°C for 16 hours) did not hydrolyze the oligomeric structure. ESI-MS of the products of acid hydrolysis detected a series of singly charged ion masses that differed again by multiples of 71 (Fig. 4A). The values exactly matched the ( m ϩ H) ϩ ion masses obtained by 5 to 11 repeats of a 71-dalton unit linked covalently to a lysine residue. The resistance of this oligomer to acid hydrolysis indicated a lysine- N -alkyl linkage. 1 H-Nuclear magnetic resonance (NMR) spectroscopy in D O pro- 2 duced a singlet resonance at 2.4 parts per mil- lion as well as a multiplet resonance at 2.65 ppm, indicating the presence of both N-CH 3 and N-CH -alkyl elements within the oligomer. 2 A repeated N -methyl-propylamine unit attached to the ε -amino group of the lysine at position 3 of the chymotryptic peptides would explain both the NMR and the mass spectrometric data. Final proof for this structure was obtained by MS/MS analysis of the ( m ϩ H) ϩ ϭ 573 ion (six repeats of the unit element) of the acid hydrolysate (Fig. 4A). Fragmentation yields two series of subfragments with masses differ- ing again by 71 daltons (Fig. 4B), confirming the proposed polyamine structure (Fig. 4C). Taking into account the masses of these lysine modifications allows the correct prediction of the experimentally determined masses of the chymotryptic peptides (see above; Fig. 4D). Reversed-phase HPLC combined with ESI- MS and NH -terminal sequencing demon- 2 strates that silaffin-1A is a mixture of peptide isoforms with molecular masses ranging from 2500 to 3500 daltons. These peptide isoforms represent the individual, covalently modified repeat units R2 to R7 of si/1p (Fig. 3). We have previously found that certain peptides terminat- ing with the sequence motifs RHL or RQL become cleaved off from diatom cell wall proteins in vivo, suggesting that these motifs serve as recognition ...