INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The ability of magnetotactic bacteria to migrate along magnetic field lines is based on specific intracellular structures, magnetosomes that, in most magnetotactic bacteria, are nanometer sized, membrane-bound magnetic particles consisting of the iron mineral magnetite (Fe₃O₄) (3, 42). The unique characteristics of bacterial magnetosomes have attracted broad interdisciplinary research interest. Their superior crystalline and magnetic properties make them potentially useful as a highly ordered biomaterial in a number of applications, e.g., in the immobilization of bioactive compounds, in magnetic drug targeting, or as a contrast agent for magnetic resonance imaging (24, 29, 45). Recently, the characteristics of bacterial magnetosomes have been used as biosignatures to identify presumptive Martian magnetofossils (15, 51).

The narrow size distributions and uniform, species-specific crystal morphologies of bacterial magnetosomes imply a high degree of biological control over the mineralization process. The biomineralization of magnetosome particles is achieved by a complex mechanism that involves the uptake and accumulation of iron and the deposition of the mineral particle with a specific size and morphology within a specific compartment provided by the magnetosome membrane (MM). In bacteria of the genus Magnetospirillum (40), the MM consists of a bilayer containing phospholipids and proteins (16, 41; D. Schüler, K. Grünberg, and B. M. Tebo, Abstr. 100th Gen. Meet. Am. Soc. Microbiol. 2000, abstr H-111, p. 373, 2000). A number of proteins were identified as specifically associated with the MM in Magnetospirillum magnetotacticum and Magnetospirillum sp. strain AMB-1 (16, 25, 32). The exact role of these magnetosome-specific proteins has not been elucidated, but it has been suggested that they have specific functions in iron accumulation, nucleation of minerals, and redox and pH control (4, 16, 42). Although several genes putatively related to magnetosome formation have been identified (25, 28, 32), the genetic basis of magnetite biomineralization has remained mostly unknown. Recently, the almost complete genome sequences of two magnetotactic alpha-proteobacteria, M. magnetotacticum strain MS-1 and magnetic coccus strain MC-1, have become available (http://www.jgi.doe.gov/tempweb/JGI_microbial/html/index.html), which now allows the study of magnetosome formation at the genomic level. M. magnetotacticum is a microaerophilic spirillum producing cubo-octahedral magnetite particles that are 42 nm in size (8, 40). The size of its genome is about 4.3 Mb (6). Magnetic coccus strain MC-1, which has a genome size of about 3.7 Mb (12), was reported to form pseudohexagonal prismatic magnetite crystals about 70 nm in diameter (13, 26).

The magnetotactic bacterium M. gryphiswaldense, which was isolated from a freshwater sediment (40, 46), produces up to 60 cubo-octahedral magnetosome particles that strongly resemble those found in M. magnetotacticum and other Magnetospirillum species (3, 10, 47). M. gryphiswaldense can be cultivated more readily than most other magnetotactic bacteria, which has facilitated its physiological and biochemical analysis (41, 43, 44, 48).

In this study, we have cloned and analyzed several genes encoding magnetosome proteins from M. gryphiswaldense. Except for MamA, none of these proteins have been previously reported to be related to magnetosome formation in any magnetotactic bacterium. We report here the identification and preliminary analysis of a major gene cluster that encodes a number of these magnetosome proteins and is conserved in M. gryphiswaldense, M. magnetotacticum, and magnetic coccus strain MC-1.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Strains and growth conditions. M. gryphiswaldense strain MSR-1 (DSM 6361) was grown microaerobically at 30°C in a growth medium containing 100 µM ferric citrate as described before (44). The batch culture was exposed to air in 100-ml, 1-liter, and 10-liter bottles containing 50 ml, 500 ml, and 5 liters of medium, respectively, and agitated at 100 rpm on a New Brunswick incubation shaker. An inoculum of 10% of the culture volume was used. Microaerobic conditions arose in the medium at higher cell densities by oxygen consumption of cells (43). Escherichia coli DH5 (GIBCO BRL) was used as the host strain for cloning experiments with pBluescriptSKII (Stratagene). For cloning of PCR products using pCR-TOPO, E. coli TOP10 (Invitrogen) was used. For E. coli strains, the culture conditions used were those described by Sambrook et al. (38).

Isolation of magnetosomes. Approximately 10 g (wet weight) of M. gryphiswaldense cells suspended in 100 ml of 20 mM HEPES-4 mM EDTA, pH 7.4, was disrupted by three passes through a French pressure cell (20,000 lb/in²). All of the buffers used for magnetosome isolation contained 0.1 mM phenylmethylsulfonyl fluoride as a protease inhibitor. Unbroken cells and cell debris were removed from the sample by centrifugation (10 min, 680 × g). The cell extract was passed through a MACS magnetic separation column (Miltenyi Biotec). Columns were placed between two Sa-Co-magnets generating a magnetic field gradient inside the column, which caused the magnetic particles to bind to the column matrix. The absence of any black, magnetosome-like material in the cell extract after passage through the column indicated that the separation of magnetosome particles was complete. To eliminate electrostatically bound contamination, magnetic particles attached to the column were rinsed first with 50 ml of 10 mM HEPES-200 mM NaCl, pH 7.4, and subsequently with 100 ml of 10 mM HEPES, pH 7.4. After removal of the column from the magnets, magnetic particles were eluted from the column by flushing with 10 mM HEPES buffer. Finally, the magnetosome suspension was loaded on top of a sucrose cushion (55% [wt/wt] sucrose in 10 mM HEPES, pH 7.4) and subjected to ultracentrifugation (280,000 × g, 8 h, 4°C) in a swinging-bucket rotor. The magnetic particles sedimented at the bottom of the tube, whereas residual contaminating cellular material was retained by the sucrose cushion.

Isolation of nonmagnetic subcellular fractions. After separation of magnetosomes, an aliquot of the cell extract was subjected to ultracentrifugation (330,000 × g, 1 h, 4°C). The supernatant fluid from this high-speed centrifugation contained the soluble proteins. The membrane fraction contained in the pellet was further separated by isopycnic centrifugation as described by Osborn and Munson (34).

Analytical methods. The iron content of whole cells and isolated magnetosome particles was determined by using a Perkin-Elmer 3110 atomic absorption spectrometer. Air-acetylene flame spectroscopy was used under the following conditions: wavelength, 248.6 nm; bandwidth, 0.2 nm; lamp current, 30 mA. For iron determination, the dried samples were incubated in concentrated nitric acid until digestion of the material was complete (18). The protein concentration of samples was determined by using the bicinchoninic acid protein microassay kit (Pierce) in accordance with the manufacturer's instructions.

Electron microscopy. Purified magnetosomes were adsorbed on carbon-coated copper grids and negatively stained with 2% (wt/vol) uranyl acetate. Samples were viewed and recorded with a Philips CM10 transmission electron microscope at an accelerating voltage of 100 kV.

Sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE) and determination of N-terminal and internal amino acid sequences. Gels were prepared and run in accordance with the Laemmli procedure (20). An amount of magnetosomes equivalent to approximately 20 µg of protein was resuspended in electrophoresis sample buffer containing 2% SDS and 5% 2-mercaptoethanol. After boiling for 5 min, the samples were centrifuged for 3 min to pellet the magnetite particles. The supernatant was loaded on a 10 to 16% gradient polyacrylamide gel, which was stained with Coomassie brilliant blue after running. Digitized gels were analyzed by the ImageMaster 1D software (v.3.0; Amersham-Pharmacia). Amino-terminal protein sequence analysis was performed on an Applied Biosystems 470A amino acid sequencer by F. Lottspeich (Max-Planck-Institut für Biochemie, Martinsried, Germany) as previously described (14). Internal sequences were determined after cleavage with AspN protease (as described in reference 50).

Recombinant DNA techniques. Total DNA of M. gryphiswaldense was isolated as described by Marmur (23). Plasmid isolation, transformation, and DNA manipulations in E. coli were essentially carried out by standard methods (38). Long oligonucleotides for hybridization used in Southern hybridization experiments were DS24 (5'-AAGCCCTCGAACATGCTGGACGAGGTGACCCTGTATACCCACTATGGCCTGTCGGTGGCC-3') and DS33 (5'-ATGAAGTTCGAG AACTGCCGGGACTGCCGGGAAGAGGTGGTCTGGTGGGCGTTC-3'). Plasmid vectors used for cloning were pCR2.1-TOPO (Invitrogen) and pBluescriptSKII (Stratagene).

PCR amplification and DNA sequencing. Degenerate primers for PCR amplification of a 240-bp mamC fragment were DS15F4 (5'-GCCGCBCTSGCBAAGAAYGC-3') and DS15RV3 (5'-CGSAGYTCCTTYTCRATGAARTC-3'). For the amplification of a 960-bp mamD fragment, the primers were 1KGVDF and 4KGCR. 1KGVDF (5'-ATGTGGAGCGTCCTGGCCATG-3') was deduced from the DNA sequence upstream of the homologous region in the genome of M. magnetotacticum. 4KGCR (5'-GCCTCAGGGTGGTGGCGGAT-3') was deduced from the cDNA sequence close to the 3' end of the mamC gene of M. gryphiswaldense. PCR amplification was performed with the Mastercycler Gradient (Eppendorf) by using standard protocols. Automatic sequencing of both strands of the plasmid DNA was carried out by primer walking (primers not shown).

Analysis of DNA sequence data. Assembly of DNA sequences, identification and translation of open reading frames (ORFs), and calculation of the molecular masses of the proteins were done by the MacVector 6.5.3 software package (Oxford Molecular Ltd.). Sequence alignments were carried out by using the ClustalW algorithm (52), which is part of the same software. Protein sequences were compared to the GenBank, EMBL, and SwissProt databases by using the BLASTP program (1). Motif searches were carried out by using the Prosite program (17). Protein location was determined by the PSORT program (27). Preliminary sequence data for M. magnetotacticum MS-1 and magnetic coccus strain MC-1 was obtained from the DOE Joint Genome Institute at http: //www.jgi.doe.gov/tempweb/JGI_microbial/html/index.html (status, 04/20/01). The amino acid sequences of the identified Mam proteins from M. gryphiswaldense were used in TBLASTN similarity searches to identify genes encoding homologous proteins in the preliminary baseline genomic assemblies of these bacteria. The identified regions of sequence homology on the respective contigs were analyzed for ORFs and translated into protein sequences.

Nucleotide sequence accession numbers. The nucleotide sequence of the M. gryphiswaldense mamAB gene cluster has been deposited in the GenBank, EMBL, and DDBJ libraries and assigned accession number AF374354. The nucleotide sequence of the M. gryphiswaldense mamCD region has been deposited under accession number AF374355.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Analysis of magnetosome particles. The magnetosome purification protocol resulted in 9 mg of clean magnetosomes from 1 g of magnetic cells on a dry-weight basis. Approximately 0.04 mg (dry weight) of protein was associated with 1 mg of isolated magnetosomes. The amount of MM-associated protein was equivalent to 0.07% of the total cellular protein content. Magnetosome-bound iron constituted approximately 93% of the total intracellular iron. Transmission electron microscopy indicated that isolated individual magnetite crystals were enclosed by an electron-thin layer representing the MM and were apparently free of contaminating cellular material (Fig. 1). Individual particles remained attached but were separated from each other by the membrane.

View larger version (125K): [in this window] [in a new window]   FIG. 1.   Transmission electron micrograph of purified magnetosomes from M. gryphiswaldense. Note that individual magnetosome particles are enclosed by a membrane and appear to remain attached to each other.

View larger version (88K): [in this window] [in a new window]   FIG. 2.   SDS-PAGE of the MM-associated proteins from M. gryphiswaldense compared to soluble proteins (SP) and the cytoplasmic membrane proteins (CM) and outer membrane proteins (OM). The bands were visualized by staining with Coomassie blue. Thirteen MM-specific proteins were identified in various amounts (arrowheads).

Cloning and sequence analysis of genes encoding MM proteins in M. gryphiswaldense. (i) mamA and mamB Based on the codon usage bias found in previously analyzed genes from Magnetospirillum species, long (50 to 65 bases), nondegenerate oligonucleotides were designed from N-terminal amino acid sequences of several major MM-specific polypeptides. These oligonucleotides were labeled and directly used as probes for hybridization. Two probes (DS24 and DS33), corresponding to the amino acid sequences of MM24.3 and MM33.3, respectively, recognized the same genomic 7.55-kb EcoRI DNA fragment in Southern blotting experiments. It was cloned into plasmid pBluescriptSKII, resulting in pDS902. Sequence analysis by primer walking of the complete 7.55-kb fragment identified eight complete and two truncated consecutive ORFs. The deduced amino acid sequences of two ORFs matched the N-terminal sequences of MM24.3 and MM33.3, respectively. Consequently, these ORFs were designated mamA and mamB (mam for MM).

(ii) mamC and mamD Since long, nondegenerate oligonucleotide probes derived from the MM15.5 N-terminal amino acid sequences failed to identify specific chromosomal DNA fragments in hybridization experiments, a pair of oligonucleotide primers for PCR were deduced from the N-terminal and internal amino acid sequences of this polypeptide. By using these primers, a single 240-bp fragment was amplified from genomic DNA and cloned into pCR2.1-TOPO, generating pMT1. By using the cloned 240-bp fragment as a probe, a 4.3-kb chromosomal EcoRI-fragment was identified by Southern hybridization and cloned into pBluescriptSKII, generating plasmid pKG2. Sequence analysis of the insert identified an ORF that contained the N-terminal and internal peptide sequences of the MM15.5 protein. It was designated mamC. The mamC gene encodes the most abundant polypeptide in the MM of M. gryphiswaldense (MM15.5). The calculated molecular mass of 12.24 kDa was lower than the apparent molecular mass of 15.5 kDa estimated by SDS-PAGE, as is frequently the case with hydrophobic proteins.

Identification and sequence analysis of genes encoding putative MM proteins in the genomes of M. magnetotacticum MS-1 and magnetic coccus strain MC-1. Genes with significant similarity to mamA, mamB, mamC, and mamD of M. gryphiswaldense were identified in the genome sequences of both M. magnetotacticum and strain MC-1. The characteristics of the predicted mam gene products of M. magnetotacticum and strain MC-1, together with gene products of ORFs from adjacent regions, are shown in Tables 2 and 3. Generally, the homologous genes have sizes comparable to those of their respective counterparts in M. gryphiswaldense and encode proteins with characteristics very similar to theirs. Secondary-structure predictions for the equivalent genes using various algorithms gave similar results (data not shown). The alignments of Mam protein sequences are shown in Fig. 3.

View larger version (84K): [in this window] [in a new window]   FIG. 3.   Sequence alignments of identified magnetosome proteins of M. gryphiswaldense (M.g.) and their homologs from M. magnetotacticum (M.m.) and magnetic coccus strain MC-1. If applicable, the most similar homolog from a nonmagnetic organism was included. Identical amino acids are shown on a solid background, and similar amino acid are shaded. Mtherm, Methanobacterium thermoautotrophicum; Bsubt, Bacillus subtilis.

View larger version (19K): [in this window] [in a new window]   FIG. 4.   Molecular organization of the mamAB gene clusters of M. gryphiswaldense, M. magnetotacticum, and magnetic coccus strain MC-1. Arrows indicate the direction of gene transcription. Filled arrows indicate ORFs that belong to families of homologous genes shared by the mamAB clusters of the three magnetotactic bacteria investigated. The dashed lines connect equivalent genes (closest homologs).

Molecular organization of the mamAB gene cluster in M. gryphiswaldense, M. magnetotacticum MS-1, and magnetic coccus strain MC-1. The mamB gene of M. gryphiswaldense MSR-1 was found to be located 1,120 bp downstream of mamA. As mentioned above, both genes are part of a region containing several ORFs of colinear orientation. Likewise, genes homologous to mamA and mamB were both found in the same chromosomal region in M. magnetotacticum (contig 3824) and strain MC-1 (contig 431). Since this finding was suggestive of the clustering of several genes possibly related to magnetite formation, the organization of the mamA and mamB genes, as well as the ORFs adjacent to them, was characterized in more detail. The arrangement of ORFs in the chromosomal mamAB gene clusters of M. gryphiswaldense, M. magnetotacticum, and strain MC-1 is shown in Fig. 4, and the characteristics of the corresponding predicted proteins are given in Table 2.

(i) TPR proteins. The mamA genes of all three strains display similarity to genes encoding TPR (tetratricopeptide repeat) proteins. The mamA gene of strain MC-1 (ORF4) is followed by ORF5, which encodes a deduced protein of 1,025 amino acids. Its C-terminal domain (222 amino acids) was also found to be similar to MAM22 of M. magnetotacticum (32, 33) (identical to MamA [this study]) and other members of the TPR family (21).

(ii) CDF transporters. Besides mamB, two more genes (ORF5 and ORF16) in the mamAB gene cluster of M. magnetotacticum and one more in strain MC-1 (ORF1) display significant similarity to members of the CDF protein family (36). Pairwise sequence alignments revealed that ORF5 of M. magnetotacticum and ORF1 of strain MC-1 are equivalent to each other, whereas the mamB genes of the two bacteria form a group of distinct similarity (data not shown).

(iii) HtrA. The mamE gene (ORF1) is located at the 5' end of the mamAB gene cluster in M. magnetotacticum and is most similar to the mamE gene of strain MC-1. However, in strain MC-1, this gene is located outside the mamAB cluster. Additional genes with similarity to htrA genes were identified in the mamAB regions of M. gryphiswaldense (ORF2), M. magnetotacticum (ORF7), and strain MC-1 (ORF2). In all three organisms, it is immediately followed by an ORF that also bears weak similarity to htrA-like genes.

(iv) lemA In all three magnetotactic strains, an ORF with sequence similarity to lemA-like genes (M. gryphiswaldense, ORF5; M. magnetotacticum, ORF10; strain MC-1, ORF6) is situated between the mamA and mamB genes. lemA-like genes have been identified in the genomes of a number of bacteria and are of unknown function. The LemA protein was first identified as an epitope in the bacterial pathogen Listeria monocytogenes (22).

Organization of the mamC and mamD genes in M. gryphiswaldense, M. magnetotacticum, and strain MC-1. The genes encoding MM proteins MamC and MamD in M. gryphiswaldense and their respective homologs in M. magnetotacticum and strain MC-1 are not closely linked to the mamAB gene cluster. In M. gryphiswaldense and M. magnetotacticum, mamD is immediately followed by mamC (Fig. 5). In the genome of strain MC-1, the identified homologous genes are not linked (mamC, contig 369; mamD, contig 431).

View larger version (13K): [in this window] [in a new window]   FIG. 5.   Molecular organization of the mamC and mamD genes of M. gryphiswaldense. An equivalent arrangement of genes is present in M. magnetotacticum.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The purification protocol reported in this study allowed the efficient isolation of magnetosome particles from M. gryphiswaldense. The isolated magnetosomes of M. gryphiswaldense exhibited characteristics (i.e., size, morphology, presence of the membrane, etc.) similar to those of the magnetosomes from M. magnetotacticum and Magnetospirillum sp. strain AMB-1, as previously described (16, 29). The tendency of isolated magnetosome particles to maintain their chainlike alignment might suggest that individual particles are attached to each other by specific interactions. A total of 13 polypeptide bands could be identified in Coomassie-stained SDS-polyacrylamide gels of the solubilized MM of M. gryphiswaldense, although the possibility cannot be excluded that proteins loosely attached to the MM were lost during preparation or that additional proteins are present below the level of detection by Coomassie staining.

In this study, the genes for four major MM proteins from M. gryphiswaldense were cloned and analyzed. In addition, a gene encoding a putative MM protein in M. magnetotacticum was identified based on sequence data from a homologous MM protein in M. gryphiswaldense. Four of the newly identified genes (mamB, mamC, mamD, and mamE) have not been previously reported to encode MM-specific proteins in other magnetotactic bacteria. None of these genes or neighboring genes from the mamAB cluster in the three magnetotactic bacteria investigated display substantial similarity to the magA and mpsA genes of Magnetospirillum sp. strain AMB-1, which were previously reported to encode MM-associated proteins (25, 28). Genes sharing homology with magA and mpsA of strain AMB-1 were identified in different chromosomal regions of both M. magnetotacticum and strain MC-1 in a preliminary analysis (unpublished data), indicating that these genes are not linked to chromosomal regions comprising the mamAB or mamCD genes. Likewise, the bacterioferritin-encoding gene (bfr) of M. magnetotacticum, which has been speculated to be involved in magnetite biomineralization (5), is also located in a distant genomic region. These findings suggest that the genetic determination of magnetosome formation is complex and involves several different genomic sites in addition to the mamAB and mamCD chromosomal regions identified in this study.

Comparative analysis of the mam gene sequences from M. gryphiswaldense with the almost completed genomic assemblies of M. magnetotacticum and strain MC-1 allowed us to identify homologous genes in the latter organisms. Generally, the Mam proteins of M. gryphiswaldense and M. magnetotacticum have nearly identical sequences (91 to 97% similarity) while the amino acid sequence similarity between the Magnetospirillum species and strain MC-1 is 46 to 67%. Although the biochemical composition of the MM remains to be analyzed in the latter bacterium, the extensive sequence similarity shared by the Mam proteins of all three of these magnetotactic bacteria implies that they are likely to be functionally equivalent. For Mam proteins with homology to known protein families from databases, namely, MamA, MamB, and MamE, the similarity between the equivalent proteins from the magnetotactic bacteria was generally found to be significantly higher than to database homologs from other organisms.

The arrangement of the mamAB genes, as well as the genetic organization of the flanking regions, was found to be conserved in all three magnetotactic strains. In bacteria, functionally related genes are often located close to each other. Therefore, the operon-like arrangement of genes in the conserved mamAB region suggests that the neighboring genes might be related to the formation of magnetosomes. Interestingly, most of the genes identified in the mamAB cluster encode putative membrane proteins, several of them with sizes consistent with the molecular masses of protein bands observed in MM preparations from various Magnetospirillum species (this study; 16, 31, 32). Hence, several of the products of genes from the mamAB cluster might correspond to these unidentified proteins but also could have other functions related to magnetite biomineralization, such as the uptake and transport of iron into the cell and intracellular differentiation during MM formation. In addition to genes that are specific for either the Magnetospirillum species or strain MC-1, the mamAB cluster is characterized by a set of genes found in all three magnetotactic bacteria. These genes can be assigned to six different homology classes. In addition to two unknown classes, four classes of genes correspond to proteins with homology to one of the following families: (i) TPR proteins, (ii) CDF transporters, (iii) HtrA-like serine proteases, and (iv) LemA-like proteins.

TPR motifs, which have been identified across the biological kingdom in a large number of proteins with diverse functions, are known to mediate protein-protein interactions (for a review, see reference 21). Proteins with multiple copies of TPR motifs function as scaffolding proteins and coordinate the assembly of proteins into multisubunit complexes (11, 49). TPR proteins are represented by the mamA genes in all three strains and ORF5 of strain MC-1. MamA of M. gryphiswaldense shares extensive similarity with the previously identified MAM22 protein of M. magnetotacticum (32). Since the nomenclature of this protein does not reflect its actual molecular mass of 24 kDa and its gene was found to be part of a putative operon containing additional mam genes, we propose to reassign the mam22 gene to mamA as in M. gryphiswaldense. By analogy to TPR function in many eukaryotic proteins, Okuda et al. suggested that MAM22 localized in the MM may act as a receptor interacting with proteins from the cytoplasm (32, 33). Alternatively, the function of the MamA proteins in the MM may involve the formation of multiprotein complexes within the MM or between the individual magnetosome particles.

CDF proteins occur ubiquitously in eukaryotes, bacteria, and archaea and are involved in the transport of various heavy metals. CDF proteins are represented by the MamB protein and additional CDF homologs present in the mamAB region of M. magnetotacticum and strain MC-1. Several members of this family are known to confer resistance to Cu, Cd, and Zn (30, 36). Although members of the CDF protein family have not yet been demonstrated to be involved in iron transport, its specific location in the MM suggests that MamB might participate in the transport of iron into the MM vesicle.

Members of the HtrA protein family are widely distributed in nature. In E. coli and other bacteria, they are heat shock-induced serine proteases that are active in the periplasm, where their main function is the degradation of misfolded proteins. Different HtrA proteins have distinct regulatory and housekeeping functions in the cell (9, 35). Besides mamE, several additional, highly divergent genes with sequence similarity to htrA-like genes were identified in the mamAB regions of all three magnetotactic bacteria. The reported N terminus of the 66.2-kDa MM protein from Magnetospirillum sp. strain AMB-1 (25) has no homology to predicted products of the mamAB gene cluster identified in this study but does bear similarity to HtrA-like proteins (unpublished data). Although these findings suggest that HtrA-like proteins are constituents of the MM in several magnetotactic bacteria, their role is not apparent. In addition to the presence of a catalytic domain characteristic of trypsin-like serine proteases, profile searches of the Prosite database with each of the two homologous MamE sequences identified two PDZ domains characteristic of HtrA proteins in the MamE sequences of M. magnetotacticum and strain MC-1, respectively (data not shown) (37, 39). It is generally believed that the role of PDZ domains is to position ion channels, receptors, or other signaling molecules in the correct spatial arrangement (7). Hence, it might be speculated that HtrA-like proteins fulfill similar functions in the MM.

Since magnetosome formation in magnetotactic bacteria is under strict biological control, it has been assumed that a number of different gene functions are involved in this complex process (19). Our data suggest that several of these functions might be contributed by genes with homology to ubiquitous families. In addition to those, there is a set of genes represented by mamC, mamD, and ORF8 and ORF9 of the mamAB cluster of M. gryphiswaldense, whose predicted products lack recognizable homology to any prokaryotic or eukaryotic proteins from databases but are present in all magnetotactic bacteria. Hence, it can be speculated that genes of unknown function are involved in magnetosome formation. Functional studies are required to elucidate the specific role of these candidate genes in bacterial magnetite biomineralization.