Cultivationindependent characterization of Candidatus …sourcedb.biomag.igg.cas.cn/fbwz/201502/P02015020967804501189… · sandy sediment sampled at 1–3 m water depth close to

Cultivation-independent characterization of‘Candidatus Magnetobacterium bavaricum’ viaultrastructural, geochemical, ecological andmetagenomic methodsemi_2220 2466..2478

C. Jogler,1 M. Niebler,1 W. Lin,2 M. Kube,3

G. Wanner,1 S. Kolinko,1 P. Stief,4 A. J. Beck,4

D. de Beer,4 N. Petersen,5 Y. Pan,2 R. Amann,4

R. Reinhardt3 and D. Schüler1*1LMU Department Biology I, Microbiology,Großharderner-Str. 2, 82152, Planegg-Martinsried,Germany.2Biogeomagnetism Group, Paleomagnetism andGeochronology Laboratory, State Key Laboratory ofLithospheric Evolution, Institute of Geology andGeophysics, Chinese Academy of Sciences, Beijing100029, China.3Max Planck Institute for Molecular Genetics, Ihnestr.73, 14195 Berlin, Germany.4Max Planck Institute for Marine Microbiology,Celsiusstr. 1, 28359 Bremen, Germany.5LMU Department of Earth and Environmental Sciences,Geophysics, Theresienstr. 41, 80333 Munich Germany.

Summary

‘Candidatus Magnetobacterium bavaricum’ is unusualamong magnetotactic bacteria (MTB) in terms of cellsize (8–10 mm long, 1.5–2 mm in diameter), cell archi-tecture, magnetotactic behaviour and its distinct phy-logenetic position in the deep-branching Nitrospiraphylum. In the present study, improved magneticenrichment techniques permitted high-resolutionscanning electron microscopy and energy dispersiveX-ray analysis, which revealed the intracellular orga-nization of the magnetosome chains. Sulfur globuleaccumulation in the cytoplasm point towards a sulfur-oxidizing metabolism of ‘Candidatus M. bavaricum’.Detailed analysis of ‘Candidatus M. bavaricum’ micro-habitats revealed more complex distribution patternsthan previously reported, with cells predominantlyfound in low oxygen concentration. No correlation toother geochemical parameters could be observed. Inaddition, the analysis of a metagenomic fosmid library

revealed a 34 kb genomic fragment, which contains 33genes, among them the complete rRNA gene operon of‘Candidatus M. bavaricum’ as well as a gene encodinga putative type IV RubisCO large subunit.

Introduction

Magnetotactic bacteria (MTB) synthesize membrane-enclosed organelles, so-called magnetosomes, consist-ing of magnetite crystals (Fe3O4), or less commonlygreigite (Fe3S4) (Bazylinski and Frankel, 2004). Thesecomplex subcellular organelles are normally aligned inchains along cytoskeletal structures (Komeili et al., 2006;Scheffel et al., 2006; Schüler, 2008). Thus, the passivemagnetic alignment of MTB along the earth magneticfield lines facilitates navigation in a stratified environmentwithin freshwater and marine sediments (Bazylinski andFrankel, 2004; Faivre and Schüler, 2008). The formationand positioning of magnetosomes is under strict geneticcontrol. Most of the genes controlling magnetosome for-mation and magnetotaxis in Magnetospirillum gryph-iswaldense and other freshwater magnetospirilla areclustered and part of a large genomic magnetosomeisland (MAI) (Schübbe et al., 2003; Ullrich et al., 2005;Fukuda et al., 2006; Richter et al., 2007). It was recentlyshown that the MAI is also conserved in marine MTB,such as the magnetotactic vibrio MV-1 and the magneticcoccus MC-1 (Jogler et al., 2009a; Nakazawaet al., 2009; Schübbe et al., 2009). It has been sug-gested that the MAI had been transferred horizontallybetween different MTB (Schübbe et al., 2003; Jogleret al., 2009b).

Despite continued efforts by many laboratories, themajority are still not available in pure culture. This isprobably due to the fact that many MTB are typical gra-dient organisms and require complex growth conditions.However, unlike other uncultivated bacteria, MTB can bemagnetically enriched from environmental sampleswithout cultivation (Flies et al., 2005a). This has beenutilized in a number of earlier studies uncovering themorphological and phylogenetic diversity of MTB found inenvironmental populations (Spring et al., 1992; DeLonget al., 1993; Spring et al., 1994; 1998; Thornhill et al.,

Received 12 October, 2009; accepted 20 February, 2010. *For cor-respondence. E-mail [email protected]; Tel. (+49) 89 218074502; Fax (+49) 89 2180 74515.

Environmental Microbiology (2010) 12(9), 2466–2478 doi:10.1111/j.1462-2920.2010.02220.x

© 2010 Society for Applied Microbiology and Blackwell Publishing Ltd

mailto:[email protected]

1994; 1995; Spring and Scheifer, 1995; Flies et al.,2005b). Recently, a metagenomic strategy for the analysisof MTB was described, which revealed a similar, yet dis-tinct, organization of magnetosome genes in two unculti-vated MTB and provided further evidence for horizontaltransfer of magnetosome genes (Jogler et al., 2009b).

One of the most interesting, uncultivated MTB is therelatively large magnetotactic rod ‘Candidatus Magneto-bacterium bavaricum,’ discovered in sediment samplesfrom Lake Chiemsee and Lake Ammersee, in southernGermany (Vali et al., 1987a; Petersen et al., 1989). ‘Can-didatus M. bavaricum’ is unique with respect to its mor-phology, ultrastructure, magnetic characteristics andphylogenetic position. Compared with the about 10–50magnetosome crystals found in most other MTB, cells of‘Candidatus M. bavaricum’ contain between 600 and 1000crystals made of magnetite (Fe3O4). Unlike the isotropiccrystals of cultivated magnetospirilla, which are organizedin a single linear chain, ‘Candidatus M. bavaricum’ formselongated bullet-shaped, kinked crystals that are highlyanisotropic with respect to their shape and magnetizationand are arranged in several parallel chains that in turnform rope-like strands or bundles. Fossilized magneto-some crystals of ‘Candidatus M. bavaricum’-like bacteriaare suspected to account for a large proportion (up to10%) of the total magnetization in certain aquatic sedi-ments (Petersen et al., 1989; Pan et al., 2005). As discov-ered by Spring and colleagues (1993), ‘Candidatus M.bavaricum’ is affiliated with the independent deep-branching Nitrospira phylum, which currently encom-passes few other isolates including the iron oxidizerLeptospirillum ferrooxidans and the chemolithoau-totrophic nitrite oxidizer Nitrospira moscoviensis (Daimset al., 2006). However, 16S rRNA gene-based analysesshowed that several phylogenetic sublineages of thephylum Nitrospira are widely distributed in many kinds ofaquatic and terrestrial habitats (Daims et al., 2001). Anumber of further reports, based merely on the retrieval ofenvironmental 16S rRNA sequences, provide growing evi-dence that ‘Candidatus M. bavaricum’-like bacteria (up to99% identity) are abundant cosmopolitans and wide-spread in a number of different habitats, such as fresh-water lakes (Schwarz et al., 2007), marine deep-seahydrothermal systems (Suzuki et al., 2004), petroleum-hydrocarbon spill sites (Kao et al., 2008) and benzenedegrading reactor columns (Kleinsteuber et al., 2008).However, in none of these examples it was revealedwhether the detected phylotypes were forming magneto-somes. In addition, recent findings indicated that thegeographical distribution of magnetosome-forming ‘Can-didatus M. bavaricum’-like bacteria is not restricted tocalcareous sediments of Lake Chiemsee, but closelyrelated magnetic organisms were also reported from sedi-ments in Bremen, Germany (Flies et al., 2005a), Beijing,

China (Lin and Pan, 2009; Lin et al., 2009), the SeineRiver near Paris, France (Isambert et al., 2007) and Brazil(Lins et al., 2000). However, since the seminal descriptionof ‘Candidatus M. bavaricum’ (Spring et al., 1993), noprogress was made concerning its further ecological,physiological and genetic characterization. Since cultiva-tion attempts failed so far, we applied culture-independentcharacterization techniques, such as microsensor mea-surements, metagenomic analysis and MTB mass collec-tion to further characterize ‘Candidatus M. bavaricum’. Inaddition, refined structural analyses of magnetosomeorganization and sulfur globule content verified the previ-ously hypothesized arrangement of magnetosome chainsadjacent to the cell envelope and provides further evi-dence for a sulfur oxidizing metabolism of ‘Candidatus M.bavaricum’ (Spring et al., 1993; Hanzlik et al., 1996).

Results

Spatial distribution of ‘Candidatus M. bavaricum’-likebacteria and their morphological characterization

In a screening of various habitats, ‘Candidatus M.bavaricum’-like bacteria were only detected in sedimentssamples from Lake Chiemsee (Germany) and samplesfrom ponds in Unterlippach and Coburg (Germany).However, only a single sample from the Coburg pondcontained ‘Candidatus M. bavaricum’ cells after a twoyears incubation period. Thus detailed analysis was notpossible due to the low sample volume. We failed todetect ‘Candidatus M. bavaricum’ in all other examinedhabitats (Table S1). However, nearly all samples con-tained numerous MTB other than ‘Candidatus M. bavari-cum’ with diverse morphotypes, which were not analysedfurther within the present study. In Lake Chiemsee, theoccurrence of ‘Candidatus M. bavaricum’ was confined tosediments at 10–20 m water depth, where the sedimentwas grey and sandy to loamy. No ‘Candidatus M.bavaricum’-like bacteria were observed in the brownishsandy sediment sampled at 1–3 m water depth close tothe beach, or in other nearshore areas, which did notcontain any MTB. In fresh samples from Lake Chiemsee,up to 1.8 ¥ 103 cells ml-1 sediment ‘Candidatus M. bavari-cum’ cells were detected. In the small shallow pond inUnterlippach, which is a habitat different from Lake Chi-emsee, we found high numbers (6 ¥ 103 cells ml-1) of‘Candidatus M. bavaricum’-like cells only in the northernarea of the pond in shallow water sediments (depth about10 cm). This is most likely due to drainage close to thenorthern part of the pond and frequent sediment deposi-tion in this area. Preliminary 16S rRNA gene sequenceanalysis using the PCR primers CJ160 and CJ161revealed no significant difference between ‘Candidatus M.bavaricum’ cells from both sampling sites (above 99%

Candidatus Magnetobacterium bavaricum 2467

© 2010 Society for Applied Microbiology and Blackwell Publishing Ltd, Environmental Microbiology, 12, 2466–2478

cut-off) and sequences retrieved from both sample siteswere also identical to the previously reported ‘CandidatusM. bavaricum’ sequence from Lake Chiemsee (99% cut-off) (Spring et al., 1993). However, ‘Candidatus M. bavari-cum’ populations from Unterlippach were not stable, andafter microcosm incubation for several weeks, the numberof cells decreased until none was observed after severalmonths. Thus, all further analyses focused on ‘CandidatusM. bavaricum’ from Lake Chiemsee.

Our refined enrichment technique enabled us to collect‘Candidatus M. bavaricum’ cells in sufficient numbers (upto 108 cells) to facilitate extensive electron microscopicanalysis (Jogler et al., 2009b). Until now no SEM micro-graphs of ‘Candidatus M. bavaricum’ existed and TEMpictures were used for the morphological characterizationof magnetosome arrangement (Vali et al., 1987a; Springet al., 1993; Hanzlik et al., 2002; Pan et al., 2005). Tiltingof the specimen by� 45° during TEM analysis led to themodel of 3–5 rope-shaped bundles of magnetosomesseparated by the maximum possible distance from eachother and positioned adjacent to the cell envelope(Hanzlik et al., 1996). To further investigate the magneto-some arrangement of ‘Candidatus M. bavaricum’, we per-formed SEM analysis while applying an acceleratingvoltage of 30 kV in combination with the detection of back-scattered electrons and tilting of the specimen, allowing3D visualization of the shape and arrangement ofelectron-dense magnetosome crystals. Figure 1 showsan exemplary micrograph with three magnetosomechains per bundle and five bundles arranged as linearchains putatively attached to the cytoplasmic membrane.This is in agreement with the postulated model of mag-netosome positioning from Hanzlik and colleagues(1996).

Furthermore, TEM analysis revealed that cells fromboth sampling sites (Chiemsee and Unterlippach) werevirtually identical with respect to cell size (3–10 mm), sizeof projectile shaped magnetosomes (110–150 nm), theirnumbers (up to 1000 per cell) and arrangement (data notshown).

Similar to the description of Spring and colleagues1993, we observed a morphological heterogeneitybetween cells from same samples due to variations inintracellular sulfur content (Fig. 2A–F). We distinguishedbetween cells completely filled with sulfur globules(between 20–25 globules; morphotype I), without anysulfur inclusions (morphotype II) and an intermediateforms with variable numbers of less than 20 inclusions.Sulfur globules were hardly visible in SEM pictures(Fig. 2C and D) but became obvious as dark areas bybackscattered electron detection (Fig. 2E and F). EDXanalysis revealed a correlation between the observed twomorphotypes and the presence or absence of sulfur(Fig. 2, panels E and F, and G and H, respectively). In‘Candidatus M. bavaricum’ samples stored at room tem-perature in sterilized Lake Chiemsee water, the proportionof cells containing sulfur globules decreased dramaticallyover time, while cells without sulfur inclusions becamemore abundant (data not shown).

Magnetotaxis of ‘Candidatus M. bavaricum’

The observation of ‘Candidatus M. bavaricum’ in thehanging drop assay revealed unusual swimming charac-teristics. While cells were initially predominantly north-seeking, intermittent backwards swimming was frequentlyobserved. The artificial magnetic field was kept constantbut ‘Candidatus M. bavaricum’ cells swam back and forthin a somehow coordinated manner, while the periods ofbackwards swimming always being much shorter than offorwards swimming. This ‘La-Ola’ behaviour is differentfrom the ‘ping-pong’ motility observed in multicellularmagnetotactic prokaryotes (Wenter et al., 2009).

Characterization of the ‘CandidatusM. bavaricum’ microhabitat

The initial study of Spring and colleagues (1993) corre-lated ‘Candidatus M. bavaricum’ occurrence to low oxygenconcentrations. However, besides detecting no sulfide, nofurther geochemical parameters were analysed (Springet al., 1993). We measured vertical cell distributions inconjunction with selected geochemical parameters, suchas dissolved O2, NO3

-, Fe, Mn and NH4+ and pH within

selected microcosms. The occurrence of ‘Candidatus M.bavaricum’ in various microcosms was very diverse withrespect to abundance and distribution, despite the fact thatall microcosms were initially prepared from the same

1µm

Fig. 1. SEM micrograph of ‘Candidatus M. bavaricum’. Signalmixing of secondary electrons (green) which show topography, andbackscattered electrons (red) visualize the magnetosomes bymaterial contrast. The white box indicates the area of magnificationshowing three parallel magnetosome chains.

2468 C. Jogler et al.


C

BA

D

FE

keV 5.695

coun

ts

C

O Na

MgAl

Si

S

K

Ca

0

400

keV 5.695

coun

ts

C

O Na

Al

Si

S K

Ca

0

400C

Mg

G H

1µm

1µm

1µm

1µm

6µm6µm

Fig. 2. Analysis of two different morphotypes of ‘Candidatus M. bavaricum’ with bright field microscopy (A), differential interference contrastmicroscopy DIC (B), SEM by secondary electron detection (C and D), and by backscattered electron detection (E and F) and energydispersive X-ray spectroscopy (EDX) (G and H).A and B. Open triangles indicate representative cells of morphotype I, which is full of sulfur globules, while white triangles point towardsintermediate forms containing several globules. Arrows indicate morphotype II without any visible sulfur globules.C and E. SEM of morphotype I; white squares indicate the position of the corresponding EDX spectra (G).D and F. SEM of morphotype II; white squares indicate the position of the corresponding EDX spectra (H).



homogenized sediment slurry (data not shown). Somemicrocosms showed high (5 ¥ 105 cells ml-1), while othersshowed low to moderate cell numbers (3 ¥ 103 cells ml-1),or even lacked ‘Candidatus M. bavaricum’ cells entirely. Todetermine whether different geochemical characteristicswere responsible for the observed variations in cell abun-dance and vertical localization, six microcosms selectedduring a pilot study were analysed in detail. Two of themicrocosms lacked any ‘Candidatus M. bavaricum’ cells(A and B in Fig. 3), while one exhibited moderate cellnumbers (C in Fig. 3), and three showed high cell numbers(D, E, F in Fig. 3). For example, a maximal cell number of4.8 ¥ 105 was measured at 6 mm depth in microcosm F,while the moderate amount of 1.8 ¥ 103 cells was detectedat 5 mm depth in microcosm D (Fig. 3). A common patternof correlation between ‘Candidatus M. bavaricum’ and theoxic-anoxic transition zone (OATZ) was observed, butdepending on the microcosm or even the lateral samplingposition within a single microcosm, one or multiple peakswere observed. In microcosm D (Fig. 3), the vertical dis-tribution of ‘Candidatus M. bavaricum’ cells had amaximum at 5 mm below the sediment surface, where theoxygen concentration was about 25 mM. In contrast,microcosms C, E and F (Fig. 3), showed two maximalocated at depth of 6 and 11, 5 and 8, and 6 and 11 mm,respectively. The oxygen concentrations at both cellmaxima in microcosms C, E and F differed between30/ < 5 mM, 30/5 mM and 20/5 mM respectively. Thus, noclear link between the vertical ‘Candidatus M. bavaricum’distribution and the exact oxygen concentration wasobserved. In addition to oxygen, no strict correlation withother geochemical parameters, such as pH, nitrate,ammonium, manganese, and iron was obvious (Fig. 3 anddata not shown). A slight decrease in pH with sedimentdepth was observed in all microcosms, independent of thepresence or absence of ‘Candidatus M. bavaricum’ cells.The absolute values varied between about 8 (A, E), 8.5 (B,C) and between 8.5 and 9 (D, F) at the sediment surface.At 24 mm pH values of about 7.0 (E), 7.3 (A, C), 7.7 (B, F)and 8.0 (D) were detected. Nitrate was hardly detectable inmicrocosms A, C, D, E and F, while microcosm B showedhigh fluctuations and the presence of nitrate even at20 mm depth, where the oxygen concentration wasalready bellow detection limit. However, in microcosm Bbiological perturbation of the sediment by tubificid wormswas observed. The concentration of ammonium increasedbelow the layers with maximal ‘Candidatus M. bavaricum’cell counts, whereas it was low below the OATZ in micro-cosms A and B, which lacked any cells, but were inhabitedby tubificid worms. However, microcosm B showed twoammonium peaks, while microcosm A displayed only onepeak above the regions where ‘Candidatus M. bavaricum’maxima would be expected based on the depth of oxygenpenetration.

The concentrations of dissolved and solid-phase ironand manganese showed inconsistent patterns (data notshown), and sulfide was not detected in any of the sixanalysed microcosms (data not shown).

Metagenomic analysis

The high cell yields obtained by our improved magneticcollection methods allowed the construction of metage-nomic fosmid libraries, which were found to contain 16SrRNA genes of ‘Candidatus M. bavaricum’. To reduce theinterference of Escherichia coli rRNA genes with thescreening of fosmids for 16S rRNA genes of ‘CandidatusM. bavaricum’, specific primers were designed by com-paring the 16S rRNA gene of ‘Candidatus M. bavaricum’(NCBI: X71838) with the 16S rRNA genes a to h of E. coliK12 (NCBI: AP009048). The resulting primer CJ160(TAAAAGGAGTAATTCACCTG) and CJ161 (TATTTT-TAGGGATTTGC TCCA) had matches of 40% and 48% toE. coli 16S rRNA gene sequence (position 210–229 and1285–1306 respectively). The primer CJ160 showed only80% sequence identity to any other bacterial sequence inthe NCBI database, while CJ161 shows 100% identity toseveral 16S rRNA genes of yet uncultivated members ofthe phylum Nitrospira. Thus, CJ160 is ‘Candidatus M.bavaricum’ specific, while CJ161 is Nitrospira specific dueto the lack of base 1296 of the 16S rRNA gene sequenceof E. coli. This strategy allowed high-throughput screeningof liquid fosmid containing 96-well E. coli cultures withPCR. Thus, the screening of 5000 clones from the LakeChiemsee fosmid library with the primers CJ160 andCJ161 yielded one positive clone. Fosmid AJB2 was con-firmed positive for the presence of a ‘Candidatus M.bavaricum’ 16S rRNA gene by PCR amplification andsubsequent sequencing. The sequence of the entire 34 kbgenomic fragment was determined (Fig. 4A), and 33genes were predicted (Table S2). The fosmid carries a fullrRNA gene operon with the standard 16S, 23S-5Sarrangements of rRNA genes. Two putative tRNA encod-ing sequences (tRNA-Ile and tRNA-Ala) are locatedbetween the 16S and the 23S rRNA genes.

From the residual genes, five were putatively phage ortransposon associated, while 12 genes encode hypotheti-cal proteins. As shown in Table S2, the best BLAST results,based on the comparison of the encoded proteins with theNCBI database, are diverse. Two of the highest e-valueswere obtained from protein sequences related to eukary-otes, while 12 genes encode proteins with high similarityto proteobacterial sequences. In addition, three proteinsgave best hits to cyanobacterial proteins and one to aprotein from green sulfur bacteria. Another gene encodesa protein with similarity to Planctomycetes, while sixothers are closely related to Gram-positive bacteria, andthree more to the deep-branching Aquificae phylum.



A B C

D E F

Fig. 3. Vertical distribution of oxygen-, nitrate- and ammonium- concentrations in combination with pH values measured with microsensors inmicrocosms A–F relative to ‘Candidatus M. bavaricum’ cell numbers. The cell numbers in microcosm C are in the range of 103 cells ml-1, whilemicrocosms D, E and F are presented in the dimension of 105 cells ml-1.



Remarkably, none of the encoded proteins generated abest BLAST hit to proteins from the Nitrospira phylum. Inaddition, throughout the 34-kb-long ‘Candidatus M.bavaricum’ fosmid, no magnetosome-related genes wereobserved.

One gene, encoding a protein with similarity toa RubisCO (Ribulose-1,5-bisphosphat-carboxylase/-oxygenase) large subunit, was analysed in more detail.Phylogenetic analysis of orf ajb2F003 indicates an affilia-tion with type IV deep-branching RubisCO proteins(Fig. 4B and C). The phylogenetic tree shown in Fig. 4Bconsists of representative examples for each type ofRubisCO as defined by Hanson and Tabita (2001) andshows the same topology as trees consisting of moresequences (data not shown) (Tabita et al., 2008). Thisobservation is in contrast to all Magnetospirillum species,

which exclusively encode type II RubisCOs as shown inFig. 4B (Bazylinski et al., 2004; Matsunaga et al., 2005).RubisCO types I and II are both capable of CO2 and O2

fixation, although they differ in the specificity factor foreach substrate (Jordan and Ogren, 1981). Type IV formsare also called ‘RubisCo-like proteins’ (RLP), but they lackRubisCO activity (Hanson and Tabita, 2001). Since little isknown about the function of RLPs, the type IV RubisCOfrom ‘Candidatus M. bavaricum’ was compared with twotype IV forms involved in thiosulfate oxidation in Chloro-bium tepidum and C. limicola (Hanson and Tabita, 2001).The alignment of RLP protein sequences from Chloro-bium species and ‘Candidatus M. bavaricum’ revealed asimilar, yet distinct amino acid composition concerningresiduals involved in catalysis and RuBP binding (C and Rin Fig. 4C), suggesting that the ‘Candidatus M. bavaricum’

C R CSyn. PCC 6301 (51) GAAIAAESSTGTWTTVW (118) VGNVFGFK-AIRSLR. rubrum (42) AAHFAAESSTGTNVEVC (109) MGNNQGMG-DVEYAM. jannaschii (36) ANEIAGESSIGTWTKVQ (105) AGNIFGMK-IAKGLA. fulgidus (40) AGAVAAESSTGTWTSLH (106) AGNVF-GMKRVKGLB. subtilis (30) AEQIATGLTVGSWTDLP (99) FGKLS----LDGKI C. tepidum (38) LAHFCSEQSTAQWKRVG (117) CGEGTYFTPGVPVV C. limicola (38) LAHFCSEQSTAQWRRVG (117) CGEGAFFTPGVPVV M. bavaricum (23) IEALLREIALEQTVEVP (91) FGNIS----FKGNV

C CSyn. PCC 6301 (131) IKPKLGLSAKNYGRAVYECLRGGLDFTKDDENINSQPFQRWRDRFLFVR. rubrum (122) IKPKLGLRPKPFAEACHAFWL-GGDFIKNDEPQGNQPFAPLRDTIALVM. jannaschii (118) VKPKVGLKTEEHAKVAYEAWVGGVDLVKDDENLTSQEFNKFEDRIYKTA. fulgidus (119) PKPKVGYSAEEVEKLAYELLSGGMDYIKDDENLTSPAYCRFEERAERIB. subtilis (109) FKGVIGRDLSDIKEQLRQQALGGVDLIKDDEIFFETGLAPFETRIAEGC. tepidum (131) VKPNIGLSPGEFAEIAYQSWLGGLDIAKDDEMLADVTWSSIEERAAHLC. limicola (131) VKPNIGLKPSCFAEIAYQSWLGGLDIAKDDEMLADVDWSTLEERSRELM. bavaricum (101) LKP-MGLSSADLANLAAEFAFGGADIIKDDHGLVDQTFCQRLTLCQQA RubisCO motif=GXDFXKXDE

CR R C Syn. PCC 6301 (288) HIHRAMHAVIDR-QRNHGIHFRVLAKCLRLSGGDHLHSGTV-VGKLEGR. rubrum (286) HYHRAGHGAVTSPQSKRGYTAFVHCKMARLQGASGIHTGTMGFGKMEGM. jannaschii (274) HAHRAMHAAMTR-SRDFGISMLALAKIYRLLGVDQLHIGTV-VGKMEGA. fulgidus (275) HGHRAMHAAFTR-NAKHGISMFVLAKLYRIIGIDQLHIGTAGAGKLEGB. subtilis (265) MAHPAVSGAFTSSPFYGFSHALLLGKLNRYCGADFSLFPSPYGSVALP C. tepidum (287) IGHFPFIASFSR-MEKYGIHSKVMTKLQRLAGLDAVIMPGFGDRMMTP C. limicola (287) IGHFPFIAAFSR-LEKYGVHSRVMTKLQRLAGLDSIIMPGFGSRMMTP M. bavaricum (256) MAHPAMTGAYFHDPNYGIRPSVLLGSVFRLMGADLSIFPNAGGRFNFT

R R RRSyn. PCC 6301 (369) PGVLPVASGGIHVWHMPALVEIFG-DDSVLQFGGGTLGHPWGNAPGATR. rubrum (361) KACTPIISGGMNALRMPGFFENLGNANVILTAGGGAFGHIDGPVAGARM. jannaschii (352) KPVFPVSSGGVHPRLVPKIVEILG-RDLIIQAGGGVHGHPDGTRAGAKA. fulgidus (357) KPAMPVSSGGLHPGNLEPVIDALG-KEIVIQVGGGVLGHPMGAKAGAKB. subtilis (330) NQTFAVPSAGIHPGMVPLLMRDFG-IDHIINAGGGVHGHPNGAQGGGRC. tepidum (350) KPCLPVPGGSDSALTLQTVYEKVGNVDFGFVPGRGVFGHPMGPKAGAKC. limicola (350) RTSLPVPGGSDSALTLETVYRKVGSFDFGFVPGRGIFGHPMGPKAGAAM. bavaricum (320) KPAFPAPAGGMSLSNIPDMAAMYG-VDTVYTIGGSLLKHPDDIRDSTR

phage related / transposase RubisCO

5S

23S 16S

rRNA genes hypothetical proteindivers tRNA

MV-1

MSR-1

Rhodospirillum rubrum

MS-1AMB-1

Pseudomonasputida F1

Synechococcus sp. PCC 7002

Nitrospira multiformis ATCC 25196

Methanosarcina acetivorans C2A

Magnetobacterium bavaricum

Heliobacillus mobilis

Bacillus subtilis subsp. subtilis s

Chlorobium tepidum TLS

Chlorobium chlorochromatii CaD3

Methanosarcina acetivorans C2A

RLPsRLPs

type I

type III type II

type IV non photo

type IV ykrW/MfnW

type IV phototype IV-deep ykr

A

B C

Fig. 4. A. Schematic illustration of the predicted genes located on the sequenced fosmid AJB2, such as the ribosomal rRNA gene operon anda putative RubisCO large subunit- encoding gene.B. Phylogenetic analysis of the RubisCO protein encoded on AJB2. Species synthesizing RubisCO proteins from type I to IV in comparison tothe ‘Candidatus M. bavaricum’ RubisCO.C. Partial alignment of amino acid sequences from Synechococcus sp. strain PCC 6301 (Syn. PCC 6301) and other selected bacteriaincluding cand ‘M. bavaricum’. Modified after Hanson and Tabita (2001) and Tabita and colleagues (2008).



RLP might be involved in thiosulfate oxidation rather thanCO2 fixation.

Discussion

Previous studies reported that ‘Candidatus M. bavaricum’dwells in Lake Chiemsee (Vali et al., 1987b; Spring et al.,1993). This motivated us to perform a more extendedanalysis of its regional occurrence and vertical distributionin sediments. Surprisingly, we identified ‘Candidatus M.bavaricum’ cells only in Lake Chiemsee and in sedimentsof very different eutrophic ponds in Unterlippach andCoburg, but failed to detect these bacteria in Lake Starn-berg and Lake Attersee. In addition, ‘Candidatus M.bavaricum’ cells were only detected at distinct spots inUnterlippach and Lake Chiemsee, indicating a patchy dis-tribution even within a contiguous habitat. We failed toidentify correlations other than between oxygen concen-tration and cell numbers, which was previously describedby Spring and co-workers (Spring et al., 1993). However,Spring and colleagues (1993) discovered that the verticaldistribution of ‘Candidatus M. bavaricum’ exhibited a clearmaximum below the oxic- anoxic transition zone (OATZ),while nearly a third of the cells were found in anoxicregions. Our data argues for a more complex and variabledistribution, which cannot easily be explained by a definedlayer where ‘Candidatus M. bavaricum’ dwells. Withrefined cell counting methods and a threefold higher ver-tical resolution, two peaks were frequently observed, withone predominantly in the OATZ and the other close to theanoxic zone (Fig. 3), which might be explained via elec-tron shuttling, which means swimming of ‘Candidatus M.bavaricum’ between microoxic and anoxic sedimentregions. However, both peaks contained cells from allthree different morphotypes and thus did not representdistinct populations. In addition, correlation betweenoxygen and ‘Candidatus M. bavaricum’ cell numbers doesnot explain why some microcosms (A and B in Fig. 3) lackcells, while cell numbers increased in other microcosmswith similar oxygen concentrations and profiles (E and Fin Fig. 3). Such variations in ‘Candidatus M. bavaricum’occurrence and the resulting inconsistent dynamics wereobserved in microcosms set up from identical sedimentsamples, thus, may be an unknown chemical compoundrelated to the presence or absence of other organismsmight be responsible. Alternatively, phages or eukaryoticgrazing might lead to dramatic variations in ‘CandidatusM. bavaricum’ cell numbers between and within individualmicrocosms.

Consistent with previous studies, we failed to detectdissolved sulfide in microcosms from Lake Chiemsee(Spring et al., 1993). Thus, it is likely that sulfate-reducingbacteria, which are known to dwell in the microoxic zone(Widdel and Bak, 1992), produce hydrogen sulfide, but

the concentrations are kept low (< 1 uM) by biotic orabiotic removal processes (Sørensen and Barker Jør-gensen, 1987). This hydrogen sulfide might be oxidizedby ‘Candidatus M. bavaricum’ in a reaction leading to thedeposition of sulfur as described for other chem-olithotrophic sulfur-oxidizing gradient organisms such as,e.g. Beggiatoa (Nelson and Castenholz, 1981). Weobserved three different morphotypes of ‘Candidatus M.bavaricum’, which had no, low or high accumulation ofinclusions. Since sulfur globules disappeared during pro-longed storage of ‘Candidatus M. bavaricum’ cells underoxic conditions, we provided further indications for thehypothesis of Spring and colleagues (1993) that the intra-cellular deposition of sulfur serves as a reservoir forfurther oxidation. Spring and colleagues (1993) postu-lated that hydrogen sulfide is used as a reductant coupledto intracellular redox-cycling of iron. Porewater-dissolvedFe concentrations were negligible in surface layers(<5 mm) and greatly elevated but scattered at depth, sug-gesting limited removal (data not shown). However, withsulfide concentrations below detection limit (< 1 mM),coupled Fe-S cycling is unlikely to visibly affect Fe con-centrations on the order of 20 mM. However, the storedsulfur globules might be utilized via subsequent oxidationwith oxygen or nitrate, which is energetically favourableand the distribution of ‘Candidatus M. bavaricum’ onlycorrelated with oxygen in our studies (Fig. 3). In addition,the disappearance of sulfur globules during incubation inan oxic environment at room temperature provides furtherindications for an oxidation of sulfur by oxygen.

In addition to refined EDX and SEM analysis of sulfurglobules, our microscopic analysis allowed the first SEM3D reconstruction of magnetosome arrangements in‘Candidatus M. bavaricum’. Hanzlik and co-workers pos-tulated a position of the magnetosome chains in ‘Candi-datus M. bavaricum’ adjacent to the cell envelope andseparated by the maximum possible distance from eachother (Hanzlik et al., 1996). Three-dimensional recon-struction of magnetosome arrangements is difficult byTEM since the drying and shrinking of the sample mightcause artefacts. However, our SEM analysis further con-firms Hanzlik’s observations.

Our present study provides the first insights into thegenome of ‘Candidatus M. bavaricum’. We failed to detectany magnetosome genes within this 34 kb genomicregion, although this is not a big surprise, as there is noreason why magnetosome genes should colocalize withthe rRNA operon. However, while protein sequencesencoded by all 33 predicted genes were compared withthe NCBI database, none of them had the highest simi-larity to a protein from the Nitrospira phylum. This findingis not so surprising, since the abundance of genomesequences from this deep-branching phylum is ratherlimited. In addition, a mosaic-like genome structures has



been described for other MTB such as the magneticcoccus MC-1 as well (Esser et al., 2007; Schübbe et al.,2009). Alternatively this finding questions the phyloge-netic position of ‘Candidatus M. bavaricum’. Among themetabolic genes, the putative large type IV RubisCOsubunit is most interesting (Fig. 4B). While these so called‘RubisCO-like protein’ (RLP) lack RubisCo activity(Hanson and Tabita, 2001), little is known about theirenzymatic activity. However, it has been proposed thatphotosynthetic RubisCOs evolved from RLPs, which wereinvolved in sulfur metabolism in non-photosyntheticorganisms (Ashida et al., 2005). Physiological results indi-cate that a RLP from the green sulfur bacterium C.tepidum is involved in some aspect of thiosulfate oxidation(Hanson and Tabita, 2001). The alignment in Fig. 4C dem-onstrates that the RLP of ‘Candidatus M. bavaricum’shares similarities with respect to conserved residualexchanges with the RubisCo from Bacillus subtilis and C.tepidum. Thus, it might be involved in sulfur metabolismas well.

However, a CO2 fixation capability of ‘Candidatus M.bavaricum’ cannot be excluded, since all of the MTBinvestigated thus far contain a type II RubisCO protein(Fig. 4B). Thus, ‘Candidatus M. bavaricum’ might containa second RubisCO gene elsewhere in the genome, like,e.g. Rhodospirillum rubrum.

The sequence obtained in the present study can beused to train algorithms for assigning metagenomicsequences to ‘Candidatus M. bavaricum,’ and thus mightfacilitate a genome sequencing project in the future. Thisis of particular interest, since no other MTB is as deepbranching as ‘Candidatus M. bavaricum’. Its genomesequence will provide further insights into the evolution ofmagnetotaxis and will improve our understanding ofprokaryotic biomineralization.

Conclusions

Our study is the second comprehensive analysis of ‘Can-didatus M. bavaricum’ and significantly improves theknowledge of its cell biology, spatial localization andmetabolism. Thus, we revealed a more complex verticaldistribution of ‘Candidatus M. bavaricum’ than previouslysuggested (Spring et al., 1993). The observed cell distri-bution is not limited to a narrow layer, but displayed mul-tiple discrete maxima. However, both maxima consist ofcells with and without sulfur globules, which weredescribed before (Spring et al., 1993; Hanzlik et al.,1996). We demonstrated their consumption during pro-longed storage, which provides further evidence for asulfide oxidation energy metabolism of ‘Candidatus M.bavaricum’. We provided first insights into the genome ofthis extraordinary bacterium and identified 33 genes.None of the genes showed a best BLAST hit in to the

Nitrospira phylum. One of these genes encodes a RLP,possibly involved in sulfur metabolism. This sequenceinformation can serve as a starting point to reveal thewhole-genome sequence of ‘Candidatus M. bavaricum’.

Experimental procedures

Sediment sampling, set-up of microcosms and magneticcollection of ‘Candidatus M. bavaricum’ and ‘CandidatusM. bavaricum’-like bacteria

We analysed sediment samples from nine different locationsfor the presence of ‘Candidatus M. bavaricum’-like morpho-types. The set-up of microcosms from the ponds in Unterlip-pach, Nymphenburg and Staßfurt, as well as the samplingsite in Cuxhaven (North Sea) was previously described(Jogler et al., 2009b). Table S1 summarizes all samplinglocations and their characteristics. Sediment samples weretaken as previously described (Jogler et al., 2009b). Briefly,samples from ponds were taken with a shovel, whereas LakeChiemsee sediment was collected from a boat at 10–20 mdepth by a bottom sampler. Cuxhaven samples were taken atlow tide near the shore. Microcosms were set up in thelaboratory with aliquots of sediment slurry in three parallel500 ml plastic flasks. Each flask contained about 200 ml sedi-ment and 200 ml of supernatant water. Additionally, in thecase of Lake Chiemsee and Unterlippach, seven and threeaquaria, respectively, were filled with 18–25 l of sedimentslurry. Microcosms were immediately examined for ‘Candida-tus M. bavaricum’-like bacteria by hanging drop assay(Frankel et al., 1997). Two additional screening rounds wereperformed after incubation at room temperature in the darkfor 1 month and 1 year. In case of Lake Chiemsee and thepond in Unterlippach, additional in situ analyses were per-formed. Samples were collected either via scuba diving alongthe waterside of Lake Chiemsee or via direct sampling with amodified 10 ml syringe in Unterlippach.

Subsequent analysis required a high amount of ‘Candida-tus M. bavaricum’ cells. Thus, seven aquaria were surveyed,and the aquarium containing the highest number of ‘Candi-datus M. bavaricum’ was harvested using a two-step mag-netic enrichment technique as described before (Jogler et al.,2009b). In contrast to previous reports, magnets wereattached slightly above the sediment surface directly at theaquarium wall.

Light microscopy

‘Candidatus M. bavaricum’ cell counts in environmentalsamples and those in microcosms were made using a LeitzSM-LUX microscope and the hanging drop assay asdescribed before (Frankel et al., 1997). Differentiation of‘Candidatus M. bavaricum’ morphotypes was done with cellsimmobilized on agarose pads (sterile filtered water from LakeChiemsee supplemented with 1% agarose). The immobilizedcells were imaged with an Olympus BX61 microscopeequipped with a 100_UPLSAPO100XO objective allowingdifferential interference contrast microscopy. Documentationwas done with an Olympus F-View II camera, while imageswere captured and analysed using Olympus Cell and AdobePhotoshop software.



Scanning electron microscopy

Immediately after collection, cells were fixed with 2.5% glutar-dialdehyde in fixative buffer (75 mM sodium cacodylate, 2 mMMgCl2, pH 7.0), for 1 h at room temperature. Afterwards,samples were rinsed several times in fixative buffer and post-fixed at room temperature for 1 h with 1% osmium tetroxide infixative buffer. After two washing steps in distilled water, thecells were stained en bloc for 30 min with 1% uranyl acetate in20% acetone. Dehydration was performed with a gradedacetone series. Samples were then infiltrated and embeddedin Spurr’s low-viscosity resin. Ultrathin sections were cut witha diamond knife and mounted onto uncoated copper grids. Thesections were post-stained with aqueous lead citrate(100 mM, pH 13.0). Transmission electron micrographs weretaken with an EM 912 electron microscope (Zeiss,Oberkochen, Germany) equipped with an integrated OMEGAenergy filter operated at 80 kV in the zero loss mode.

For scanning electron microscopy (SEM), drops of thesample were placed onto a glass slide, covered with a cov-erslip and rapidly frozen with liquid nitrogen. The coverslipwas removed with a razor blade and the glass slide wasimmediately fixed with 2.5% glutaraldehyde in 75 mMcacodylate buffer (pH 7.0), postfixed with 1% osmium tetrox-ide in fixative buffer, dehydrated in a graded series of acetonesolutions and critical-point dried after transfer to liquid CO2.Specimens were mounted on stubs, coated with 3 nm plati-num using a magnetron sputter coater, and examined with aHitachi S-4100 field emission scanning electron microscopeoperated at 5 kV.

For energy-dispersive X-ray (EDX) specimens were exam-ined with SEM operated at 30 keV, equipped with a backscattered electron detector of the YAG type (Autrata) and aNoran Vantage system for EDX analysis (Thermo Fisher Sci-entific, Waltham, USA).

Microsensor measurements and determination of‘Candidatus M. bavaricum’ occurrence

Six microcosms filled with sediment from Lake Chiemseewere incubated in the dark and at 15°C. An air jet wasdirected onto the water surface to stir and oxygenate thewater column of each microcosm. Microsensors for O2 (Revs-bech, 1989), NO3

-, NH4+ and pH (summarized in De Beer

et al., 1997) were constructed in the laboratories of the MPIBremen. The sensors were calibrated and used for profiling inan automated measuring set-up as previously described(Stief et al., 2002). Using a custom-written software(m-Profiler, provided by Lubos Polerecky), vertical micropro-files were recorded at increments of 1 mm from 5 mm aboveuntil 25 mm below the sediment surface. One profile witheach of the four sensor types was measured in the sixmicrocosms.

Following cell counts and microsensor measurements,subcores were taken with an acid-washed, modified 60 mlpolypropylene syringe. Subcores were sliced at 1 mmintervals in a nitrogen-flushed glovebag, and sedimentstransferred to 2 ml polypropylene microcentrifuge tubes.Sediments were centrifuged, and the supernatant was filtered(0.22 mm) and acidified to pH < 2 with ultrapure nitric acid.The remaining sediment pellet was dried at 60°C, ground to

a fine powder with an agate mortar and pestle, and 50 mgaliquots were leached with 0.5 M HCl at room temperature.Leaches were centrifuged, and the leachate removed bypipette for analysis. Fe and Mn were analysed by ICP-OES.Instrumental accuracy was assessed by analysis of certifiedreference material NIST-1643e, and values were alwayswithin 10% of those certified. Blanks were less than 1% ofaverage sample concentrations.

The distribution of ‘Candidatus M. bavaricum’ was deter-mined in sediment cores, which were taken with a thin strawto prevent sediment compression within the core. A plungermade of modelling clay and a slat was used to push out thesediment core millimetre by millimetre onto a sterile glasssurface. A scalpel was used to slice the core into 1 mmincrements, and slices were resuspended and diluted insterile filtered Lake Chiemsee. Cell numbers were deter-mined in the hanging drop assay, and a dilution with about100 cells was analysed.

Fosmid library construction

The construction of the fosmid library was previouslydescribed (Jogler et al., 2009b). In brief, MTB were collectedmagnetically from Lake Chiemsee sediment samples. AfterDNA extraction, a fosmid library was constructed using theCopyControl Fosmid Library Production Kit (Epicenter) fol-lowing the specifications of the manufacturer. Ten cloneswere randomly selected for insert length determination byrestriction analysis. The LCh library contained an averageinsert size of 32 kb. Fosmid DNA was prepared by alkalinelysis (Sambrook et al., 2001) and contaminating genomicDNA was removed with the plasmid safe DNase (Epicenter)according to the manufacturer’s instructions.

Design of ‘Candidatus M. bavaricum’ screening primers

The alignment of 16S rRNA gene sequences was performedwith the AlignX algorithm of the Vector NTI software (10.3.1).Primer sequences were selected after manual inspection ofthe alignment and subsequent comparison of putative primersequences to the NCBI nucleotide database using the BLAST

algorithm. PCR reaction was performed with the FermentasTaq polymerase according to the manufacturer’s instructions.

PCR product sequencing

PCR products were purified via gel electrophoresis and sub-sequent band excision, while DNA preparation was done withthe NucleoSpin gel extraction kit (Macherey-Nagel, Düren,Germany) according to the manufacturer’s instructions. Thesequence of PCR products was determined using the CJ160forward and CJ161 reverse primer with an ABI systemaccording to the instructions of the manufacturer.

Fosmid sequencing

Recombinant fosmid DNA was isolated using the QiagenLarge-Construct kit (Qiagen, Hilden, Germany). For shotgunsequencing the isolated DNA was fragmented by sonication.



Resulting fragments were end-repaired with T4 and Klenowpolymerase (New England Biolabs, Beverly, MA, USA), sizeselected, ligated into pUC19 vectors and transformed into E.coli DH10B (Invitrogen, Carlsbad, CA, USA). Two plasmidlibraries with 1.3- and 2.5-kb inserts were obtained. Plas-mids were isolated by alkaline lysis (BigRobby system, MPIfor Molecular Genetics, Berlin, Germany) and used as tem-plates for sequencing with Big Dye Terminator chemistry andABI 3730XL capillary sequencers (Applied Biosystems,Foster City, CA, USA). Sequence quality assessment andassembly were performed using PHRED (Ewing and Green,1998; Ewing et al., 1998), PHRAP (http://www.phrap.org/phredphrapconsed.html) and Consed (Gordon, 2003).

Gene prediction

Structural rRNAs and tRNAs were determined using Rfam(Griffiths-Jones et al., 2005) and tRNAscan-SE (Lowe andEddy, 1997). Protein-coding sequences were predicted byGlimmer3 (Delcher et al., 1999), manually curated in Artemis(http://www.sanger.ac.uk/Software/Artemis/) and annotatedin HTGA (Rabus et al., 2002).

The annotated fosmid sequence has been deposited inGenBank, EMBL and DDBJ under accession numbersFP929063.

Phylogenetic analysis of RubisCO proteins

The phylogenetic analysis and alignment of RLPs andRubisCO proteins was done as previously described (Hansonand Tabita, 2001; Tabita et al., 2008). In brief, multiplesequence alignments were generated with Jalview (Clampet al., 2004). The same RubisCO selection was used for theanalysis, as described before (Hanson and Tabita, 2001;Tabita et al., 2008), but with the newly identified RLP from‘Candidatus M. bavaricum’ in addition. Phylogenetic treeswere generated with the MEGA software version 4.0 (Tamuraet al., 2007), using the minimum evolution method with ap-distance model of amino acid substitution and a gammaparameter of 1.55 for rate distributions between lineages asdescribed before (Tamura et al., 2007; Tabita et al., 2008).

Acknowledgements

We thank Silvia Dobler and Cornelia Niemann (LMUMünchen) as well as Tina Moser and Julia Cekanov (MPIBerlin) for excellent technical assistance. We are grateful forthe help of Ramon Egli, Kerstin Reimer, Isabel Bauer, AnnaPollithy and the Freiwillige Feuerwehr Prien with sedimentsampling at Lake Chiemsee and we thank Emanuel Katzmannfor TEM analysis. This work was funded by the GermanResearch Foundation (DFG) and the Max Planck Society(MPG).

References

Ashida, H., Danchin, A., and Yokota, A. (2005) Was photo-synthetic RuBisCO recruited by acquisitive evolution fromRuBisCO-like proteins involved in sulfur metabolism? ResMicrobiol 156: 611–618.

Bazylinski, D.A., and Frankel, R.B. (2004) Magnetosome for-mation in prokaryotes. Nat Rev Microbiol 2: 217–230.

Bazylinski, D.A., Dean, A.J., Williams, T.J., Long, L.K.,Middleton, S.L., and Dubbels, B.L. (2004) Chemolithoau-totrophy in the marine, magnetotactic bacterial strainsMV-1 and MV-2. Arch Microbiol 182: 373–387.

Clamp, M., Cuff, J., Searle, S.M., and Barton, G.J. (2004) TheJalview Java alignment editor. Bioinformatics 20: 426–427.

Daims, H., Nielsen, J.L., Nielsen, P.H., Schleifer, K.H., andWagner, M. (2001) In situ characterization of Nitrospira-likenitrite-oxidizing bacteria active in wastewater treatmentplants. Appl Environ Microbiol 67: 5273–5284.

Daims, H., Maixner, F., Lucker, S., Stoecker, K., Hace, K.,and Wagner, M. (2006) Ecophysiology and niche differen-tiation of Nitrospira-like bacteria, the key nitrite oxidizers inwastewater treatment plants. Water Sci Technol 54: 21–27.

De Beer, D., Schramm, A., Santegoeds, C.M., and Kuhl, M.(1997) A nitrite microsensor for profiling environmental bio-films. Appl Environ Microbiol 63: 973–977.

Delcher, A.L., Harmon, D., Kasif, S., White, O., and Salzberg,S.L. (1999) Improved microbial gene identification withGLIMMER. Nucleic Acids Res 27: 4636–4641.

DeLong, E.F., Frankel, R.B., and Bazylinski, D.A. (1993) Mul-tiple evolutionary origins of magnetotaxis in bacteria.Science 259: 803–806.

Esser, C., Martin, W., and Dagan, T. (2007) The origin ofmitochondria in light of a fluid prokaryotic chromosomemodel. Biol Lett 3: 180–184.

Ewing, B., and Green, P. (1998) Base-calling of automatedsequencer traces using phred. II. Error probabilities.Genome Res 8: 186–194.

Ewing, B., Hillier, L., Wendl, M.C., and Green, P. (1998)Base-calling of automated sequencer traces using Phred.I. Accuracy assessment. Genome Res 8: 175–185.

Faivre, D., and Schüler, D. (2008) Magnetotactic bacteria andmagnetosomes. Chem Rev 108: 4875–4898.

Flies, C., Peplies, J., and Schüler, D. (2005a) A combinedapproach for the characterization of uncultivated magneto-tactic bacteria from various aquatic environments. ApplEnviron Microbiol 71: 2723–2731.

Flies, C.B., Jonkers, H.M., de Beer, D., Bosselmann, K.,Böttcher, M.E., and Schüler, D. (2005b) Diversity and ver-tical distribution of magnetotactic bacteria along chemicalgradients in freshwater microcosms. FEMS Microbiol Ecol52: 185–195.

Frankel, R.B., Bazylinski, D.A., Johnson, M.S., and Taylor,B.L. (1997) Magneto-aerotaxis in marine coccoid bacteria.Biophys J 73: 994–1000.

Fukuda, Y., Okamura, Y., Takeyama, H., and Matsunaga, T.(2006) Dynamic analysis of a genomic island in Magneto-spirillum sp. strain AMB-1 reveals how magnetosome syn-thesis developed. FEBS Lett 580: 801–812.

Gordon, D. (2003) Viewing and editing assembled sequencesusing Consed. Curr Protoc Bioinform Chapter 11: Units11–12.

Griffiths-Jones, S., Moxon, S., Marshall, M., Khanna, A.,Eddy, S.R., and Bateman, A. (2005) Rfam: annotating non-coding RNAs in complete genomes. Nucleic Acids Res 33:D121–D124.

Hanson, T.E., and Tabita, F.R. (2001) A ribulose-1,5-bisphosphate carboxylase/oxygenase (RubisCO)-like



http://www.phrap.org/phredphrapconsed.html

http://www.phrap.org/phredphrapconsed.html

http://www.sanger.ac.uk/Software/Artemis/

protein from Chlorobium tepidum that is involved with sulfurmetabolism and the response to oxidative stress. Proc NatlAcad Sci USA 98: 4397–4402.

Hanzlik, M., Winklhofer, M., and Petersen, N. (1996) Spatialarrangement of chains of magnetosomes in magnetotacticbacteria. Earth Planet Sci Lett 145: 125–134.

Hanzlik, M., Winklhofer, M., and Petersen, N. (2002) Pulsed-field-remanence measurements on indivdual magnetotac-tic bacteria. J Magn Magn Mater 248: 258–267.

Isambert, A., Menguy, N., Larquet, E., Guyot, F., and Valet,J.-P. (2007) Transmission electron microscopy study ofmagnetites in a freshwater population of magnetotacticbacteria. Am Mineral 92: 621–630.

Jogler, C., Kube, M., Schübbe, S., Ullrich, S., Teeling, H.,Bazylinski, D.A., et al. (2009a) Comparative analysis ofmagnetosome gene clusters in magnetotactic bacteria pro-vides further evidence for horizontal gene transfer. EnvironMicrobiol 11: 1267–1277.

Jogler, C., Lin, W., Meyerdierks, A., Kube, M., Katzmann, E.,Flies, C., et al. (2009b) Toward cloning of the magnetotac-tic metagenome: identification of magnetosome islandgene clusters in uncultivated magnetotactic bacteria fromdifferent aquatic sediments. Appl Environ Microbiol 75:3972–3979.

Jordan, D.B., and Ogren, W.L. (1981) Species variation in thespecificity of ribulose biphosphate carboxylase/oxygenase.Nature 291: 513–515.

Kao, C.M., Chen, C.Y., Chen, S.C., Chien, H.Y., and Chen,Y.L. (2008) Application of in situ biosparging to remediate apetroleum-hydrocarbon spill site: field and microbial evalu-ation. Chemosphere 70: 1492–1499.

Kleinsteuber, S., Schleinitz, K.M., Breitfeld, J., Harms, H.,Richnow, H.H., and Vogt, C. (2008) Molecular character-ization of bacterial communities mineralizing benzeneunder sulfate-reducing conditions. FEMS Microbiol Ecol66: 143–157.

Komeili, A., Li, Z., Newman, D.K., and Jensen, G.J. (2006)Magnetosomes are cell membrane invaginations orga-nized by the actin-like protein MamK. Science 311: 242–245.

Lin, W., and Pan, Y. (2009) Uncultivated magnetotactic coccifrom Yuandadu Park in Beijing, China. Appl Environ Micro-biol 75: 4046–4052.

Lin, W., Li, J., Schüler, D., Jogler, C., and Pan, Y. (2009)Diversity analysis of magnetotactic bacteria in Lake Miyun,northern China, by restriction fragment length polymor-phism. Syst Appl Microbiol 32: 342–350.

Lins, U., Freitags, F., Keim, C.N., and Farina, M. (2000)Electron spectroscopic imaging of magnetotactic bacteria:magnetosome morphology and diversity. Microsc Microa-nal 6: 463–470.

Lowe, T.M., and Eddy, S.R. (1997) tRNAscan-SE: a programfor improved detection of transfer RNA genes in genomicsequence. Nucleic Acids Res 25: 955–964.

Matsunaga, T., Okamura, Y., Fukuda, Y., Wahyudi, A.T.,Murase, Y., and Takeyama, H. (2005) Complete genomesequence of the facultative anaerobic magnetotactic bac-terium Magnetospirillum sp. strain AMB-1. DNA Res 12:157–166.

Nakazawa, H., Arakaki, A., Narita-Yamada, S., Yashiro, I.,Jinno, K., Aoki, N., et al. (2009) Whole genome sequence

of Desulfovibrio magneticus strain RS-1 revealed commongene clusters in magnetotactic bacteria. Genome Res 19:1801–1808.

Nelson, D.C., and Castenholz, R.W. (1981) Use of reducedsulfur compounds by Beggiatoa sp. J Bacteriol 147: 140–154.

Pan, Y., Petersen, N., Davila, A.F., Zhang, L., Winklhofer, M.,Liu, Q., et al. (2005) The detection of bacterial magnetite inrecent sediments of Lake Chiemsee (southern Germany).Earth Planet Sci Lett 232: 109–123.

Petersen, N., Weiss, D.G., and Vali, H. (1989) Magnetic bac-teria in lake sediments. In Geomagnetism and Paleomag-netism. Lowes, F.J. (ed.). Boston, MA, USA: KluwerAcademic Publishers, pp. 231–241.

Rabus, R., Kube, M., Beck, A., Widdel, F., and Reinhardt, R.(2002) Genes involved in the anaerobic degradation ofethylbenzene in a denitrifying bacterium, strain EbN1. ArchMicrobiol 178: 506–516.

Revsbech, N.P. (1989) An oxygen microsensor with a guardcathode. Limnol Oceanogr 34: 474–478.

Richter, M., Kube, M., Bazylinski, D.A., Lombardot, T., Glöck-ner, F.O., Reinhardt, R., and Schüler, D. (2007) Compara-tive genome analysis of four magnetotactic bacteriareveals a complex set of group-specific genes implicated inmagnetosome biomineralization and function. J Bacteriol189: 4899–4910.

Sambrook, J., and Russel, D.W. (2001) Molecular Cloning: ALaboratory Manual, 3rd edn. Cold Spring Harbor, NY, USA:Cold Spring Harbor Laboratory Press.

Scheffel, A., Gruska, M., Faivre, D., Linaroudis, A.,Graumann, P.L., Plitzko, J.M., and Schüler, D. (2006) Anacidic protein aligns magnetosomes along a filamentousstructure in magnetotactic bacteria. Nature 440: 110–114.

Schübbe, S., Kube, M., Scheffel, A., Wawer, C., Heyen, U.,Meyerdierks, A., et al. (2003) Characterization of a spon-taneous nonmagnetic mutant of Magnetospirillum gryph-iswaldense reveals a large deletion comprising a putativemagnetosome island. J Bacteriol 185: 5779–5790.

Schübbe, S., Williams, T.J., Xie, G., Kiss, H.E., Brettin, T.S.,Martinez, D., et al. (2009) Complete genome sequence ofthe chemolithoautotrophic marine magnetotactic coccusstrain MC-1. Appl Environ Microbiol 14: 4835–4852.

Schüler, D. (2008) Genetics and cell biology of magnetosomeformation in magnetotactic bacteria. FEMS Microbiol Rev32: 654–657.

Schwarz, J.I., Lueders, T., Eckert, W., and Conrad, R. (2007)Identification of acetate-utilizing Bacteria and Archaea inmethanogenic profundal sediments of Lake Kinneret(Israel) by stable isotope probing of rRNA. Environ Micro-biol 9: 223–237.

Sørensen, J., and Barker Jørgensen, B. (1987) Early diagen-esis in sediments from Danish coastal waters: microbialactivity and Mn-Fe-S geochemistry. Geochim CosmochimActa 51: 1583–1590.

Spring, S., and Scheifer, K.L. (1995) Diversity of magneto-tactic bacteria. Syst Appl Microbiol 18: 147–153.

Spring, S., Amann, R., Ludwig, W., Schleifer, K.H., andPetersen, N. (1992) Phylogenetic diversity and identifica-tion of nonculturable magnetotactic bacteria. Syst ApplMicrobiol 15: 116–122.



Spring, S., Amann, R., Ludwig, W., Schleifer, K.-H., vanGemerden, H., and Petersen, N. (1993) Dominating role ofan unusual magnetotactic bacterium in the microaerobiczone of a freshwater sediment. Appl Environ Microbiol 50:2397–2403.

Spring, S., Amann, R., Ludwig, W., Schleifer, K.-H., Schüler,D., Poralla, K., and Petersen, N. (1994) Phylogeneticanalysis of uncultured magnetotactic bacteria from thealpha-subclass of Proteobacteria. Syst Appl Microbiol 17:501–508.

Spring, S., Lins, U., Amann, R., Schleifer, K.H., Ferreira,L.C.S., Esquivel, D.M.S., and Farina, M. (1998) Phyloge-netic affiliation and ultrastructure of uncultured magneticbacteria with unusually large magnetosomes. Arch Micro-biol 169: 136–147.

Stief, P., De Beer, D., and Neumann, D. (2002) Small-scaledistribution of interstitial nitrite in freshwater sedimentmicrocosms: the role of nitrate and oxygen availability, andsediment permeability. Microb Ecol 43: 367–378.

Suzuki, Y., Inagaki, F., Takai, K., Nealson, K.H., and Horiko-shi, K. (2004) Microbial diversity in inactive chimney struc-tures from deep-sea hydrothermal systems. Microb Ecol47: 186–196.

Tabita, F.R., Hanson, T.E., Satagopan, S., Witte, B.H., andKreel, N.E. (2008) Phylogenetic and evolutionary relation-ships of RubisCO and the RubisCO-like proteins and thefunctional lessons provided by diverse molecular forms.Philos Trans R Soc Lond B Biol Sci 363: 2629–2640.

Tamura, K., Dudley, J., Nei, M., and Kumar, S. (2007)MEGA4: molecular evolutionary genetics analysis (MEGA)software version 4.0. Mol Biol Evol 24: 1596–1599.

Thornhill, R.H., Burgess, J.G., Sakaguchi, T., and Matsu-naga, T. (1994) A morphological classification of bacteriacontaining bullet-shaped magnetic particles. FEMS Micro-biol Lett 115: 169–176.

Thornhill, R.H., Burgess, J.G., and Matsunaga, T. (1995)PCR for direct detection of indigenous uncultured magnetic

cocci in sediment and phylogenetic analysis of amplified16s ribosomal DNA. Appl Environ Microbiol 61: 495–500.

Ullrich, S., Kube, M., Schübbe, S., Reinhardt, R., andSchüler, D. (2005) A hypervariable 130 kb genomic regionof Magnetospirillum gryphiswaldense comprises a magne-tosome island, which undergoes frequent rearrangementsduring stationary growth. J Bacteriol 187: 7176–7184.

Vali, H., Forster, O., Amarantidis, G., and Petersen, N.(1987a) Magnetotactic bacteria and their magnetofossils insediments. Earth Planet Sci Lett 86: 2–4.

Vali, H., Forster, O., Amarantidid, G., and Petersen, H.(1987b) Magnetotactic bacteria and their magnetofossils insediments. Earth Planet Sci Lett 86: 389–400.

Wenter, R., Wanner, G., Schüler, D., and Overmann, J.(2009) Ultrastructure, tactic behaviour and potential forsulfate reduction of a novel multicellular magnetotacticprokaryote from North Sea sediments. Environ Microbiol11: 1493–1505.

Widdel, F., and Bak, F. (1992) Gram-negative mesophilicsulfate-reducing bacteria. In The Prokaryotes. Balows, A.,Trüper, H.G., Dworkin, M., Harder, W., and Schleifer, K.-H.(eds). New York, NY, USA: Springer, pp. 3352–3378.

Supporting information

Additional Supporting Information may be found in the onlineversion of this article:

Table S1. Sampling sites analysed for the presence of ‘Can-didatus M. bavaricum’-like bacteria.Table S2. Annotation of the fosmid AJB2 from ‘CandidatusM. bavaricum’.

Please note: Wiley-Blackwell are not responsible for thecontent or functionality of any supporting materials suppliedby the authors. Any queries (other than missing material)should be directed to the corresponding author for the article.



Documents

Cultivationindependent characterization of Candidatus …sourcedb.biomag.igg.cas.cn/fbwz/201502/P02015020967804501189… · sandy sediment sampled at 1–3 m water depth close to