ORIGINAL ARTICLE

Niche differentiation amongsulfur-oxidizing bacterial populationsin cave waters

Jennifer L Macalady1, Sharmishtha Dattagupta1, Irene Schaperdoth1, Daniel S Jones1,Greg K Druschel2 and Danielle Eastman2

1Department of Geosciences, Pennsylvania State University, Pennsylvania, PA, USA and 2Department ofGeology, University of Vermont, Burlington, VT, USA

The sulfidic Frasassi cave system affords a unique opportunity to investigate niche relationshipsamong sulfur-oxidizing bacteria, including epsilonproteobacterial clades with no cultivatedrepresentatives. Oxygen and sulfide concentrations in the cave waters range over more than twoorders of magnitude as a result of seasonally and spatially variable dilution of the sulfidicgroundwater. A full-cycle rRNA approach was used to quantify dominant populations in biofilmscollected in both diluted and undiluted zones. Sulfide concentration profiles within biofilms wereobtained in situ using microelectrode voltammetry. Populations in rock-attached streamersdepended on the sulfide/oxygen supply ratio of bulk water (r¼ 0.97; Po0.0001). Filamentousepsilonproteobacteria dominated at high sulfide to oxygen ratios (4150), whereas Thiothrixdominated at low ratios (o75). In contrast, Beggiatoa was the dominant group in biofilms at thesediment–water interface regardless of sulfide and oxygen concentrations or supply ratio. Ourresults highlight the versatility and ecological success of Beggiatoa in diffusion-controlled niches,and demonstrate that high sulfide/oxygen ratios in turbulent water are important for the growth offilamentous epsilonproteobacteria.The ISME Journal advance online publication, 20 March 2008; doi:10.1038/ismej.2008.25Subject Category: microbial population and community ecologyKeywords: Beggiatoa; epsilonproteobacteria; Frasassi cave; microelectrode voltammetry; Thiothrix;Thiovirga

Introduction

Sulfidic caves form in limestone (CaCO3) rockswhere sulfide-rich groundwater interacts with oxy-gen at the water table. The Frasassi cave systemhosts a rich, sulfur-based lithoautotrophic microbialecosystem (Sarbu et al., 2000; Vlasceanu et al., 2000;Macalady et al., 2006, 2007; Jones et al., 2008).Previous studies of the geochemistry of the cavewaters have revealed that they are mixtures ofslightly salty, sulfidic groundwater diluted 10–60%by oxygen-rich, downward-percolating meteoricwater (Galdenzi et al., 2007). Initial observations ofthe abundant biofilms in cave streams and poolssuggested that they respond dynamically to seasonaland episodic hydrologic changes. In particular, wenoted that changes in specific conductivity (trackingfreshwater dilution) and water flow characteristics

correspond with morphological differences in thebiofilms. Our initial observations motivated asystematic, multi-year study of the populationstructure of biofilms collected from cave waterswith a wide range of hydrological and geochemicalcharacteristics. The goal of this study was to identifythe ecological niches (defined as the range ofenvironmental conditions that permit growth) ofmajor biofilm-forming populations.

Unraveling the effects of changing hydrologicconditions on microbial growth within the cavesystem is not trivial because dilution of the sulfidicgroundwater has multiple effects relevant to micro-bial metabolism. The input of meteoric waterincreases water depth and flow rates, dilutesdissolved species in the sulfidic aquifer and addsdissolved oxygen. Water flow conditions that in-crease turbulence also increase sulfide degassingfrom the water and oxygen transport into the waterfrom the oxygenated cave air. We investigatedmicrobial populations using a full-cycle rRNAapproach. Environmental conditions were measuredusing microelectrode voltammetry and bulk watergeochemical analyses. We found that both

Received 18 October 2007; revised 14 February 2008; accepted 20February 2008

Correspondence: JL Macalady, Department of Geosciences,Pennsylvania State University, University Park, PA 16802, USA.E-mail: jmacalad@geosc.psu.edu

The ISME Journal (2008), 1–12& 2008 International Society for Microbial Ecology All rights reserved 1751-7362/08 $30.00

www.nature.com/ismej

sulfide/oxygen ratios and physical water flowcharacteristics are important for determining thedistributions of sulfur-oxidizing groups.

Materials and methods

Field site, sample collection and geochemistryThe Grotta Grande del Vento-Grotta del Fiume(Frasassi) cave system is actively forming in Jurassiclimestone in the Appennine Mountains of theMarches Region, Central Italy. The waters of thecave system are near neutral (pH 6.9–7.4) and havespecific conductivities ranging from 1200 to3500 mS cm�1, or roughly 4–5% of average marinesalinity. The major ions are Naþ , Ca2þ , Cl�, HCO3

�

and SO42� (Galdenzi et al., 2007). Concentrations of

electron donors and acceptors other than sulfurspecies and oxygen are as follows: bicarbonate(5.3–7.1 mM), ammonium (30–175mM), methane(1–20 mM), dissolved iron (o0.1 mM), dissolved man-ganese (o0.04 mM), nitrate (not detected, o0.7 mM)and nitrite (not detected, o2.0 mM). Organic carbonconcentrations range between 0.16 and 4.5 mg l�1.

Biofilms from cave springs and streams werecollected from 5 to 40 cm water depth at samplelocations shown in Figure 1 in May (wet season) andAugust (dry season) in 2005, 2006 and 2007.Biofilms were harvested using sterile plastic transferpipettes into sterile tubes, stored on ice andprocessed within 4–6 h of collection. Subsamples

for fluorescence in situ hybridization (FISH) werefixed in 4% (w/v) paraformaldehyde and stored at�20 1C. Samples for clone library construction werepreserved in four parts RNAlater (Ambion/AppliedBiosystems, Foster City, CA, USA) to one partsample (v/v). Water samples were filtered (0.2 mm)into acid-washed polypropylene bottles and storedat 4 1C until analyzed. Conductivity, pH and tem-perature of the waters were measured in the fieldusing sensors attached to a 50i multimeter (WTW,Weilheim, Germany). Dissolved sulfide and oxygenconcentrations were measured in the field using aportable spectrophotometer (Hach Co., Loveland,CO, USA) using the methylene blue and indigocarmine methods, respectively. Duplicate sulfideanalyses were within 1% of each other. Replicateoxygen analyses were within 20% of each other.Nitrate, nitrite, ammonium and sulfate were mea-sured at the Osservatorio Geologico di ColdigiocoGeomicrobiology Lab using a portable spectrophot-ometer within 12 h of collection according to themanufacturer’s instructions (Hach Co.). Light micro-scopy was performed on live samples within 8 h ofcollection on a Zeiss Model 47-30-12-9901 opticalmicroscope (� 1250) at the Osservatorio Geologicodi Coldigioco Geomicrobiology Lab.

Microelectrode voltammetryVoltammetric signals are produced when dissolvedor colloidal species interact with the surface of agold amalgam working electrode. Electron flowresulting from redox half-reactions at a 100 mmdiameter tip is registered as a current that isproportional to concentration (Skoog et al., 1998;Taillefert and Rozan, 2002). The gradients associatedwith microbial metabolism in biofilms are thusreadily measured using this technique. Aqueousand colloidal species that are electroactive at goldamalgam electrode surfaces include: H2S, HS�, S8,polysulfides, S2O3

2�, S4O62�, HSO3

�, Fe2þ , Fe3þ ,FeS(aq), Mn2þ , O2 and H2O2 (Luther et al., 1991,2001; Xu et al., 1998; Taillefert et al., 2000; Druschelet al., 2003, 2004; Glazer et al., 2004, 2006).Voltammetric analyses of dissolved sulfur, iron andother species in the field were accomplished using aDLK-60 potentiostat powered with a 12 V batteryand controlled with a GETAC ruggedized computer.To protect the electrochemical system from dripwaters and high humidity, the potentiostat wascontained inside a storm case containing dryritehumidity sponges and modified with rubber strip-ping to allow the communication ribbon cable andelectrode cables to go outside the case while keepingthe inside sealed. Electrodes were constructed asdescribed in Brendel and Luther (1995).

Voltammetric analyses in the cave systeminvolved placing working (gold amalgam) electro-des into a narishege three-axis micromanipulatorwith a two-arm magnetic base on a steel plate. Ateach sampling location, the reference and counter

Figure 1 Map of the Frasassi cave system showing samplelocations (open circles). The small inset shows the location of theFrasassi cave system in Italy. Major named caves in the Frasassisystem are shown in different shades of gray. Topographic linesand elevations in meters refer to the surface topography. Base mapcourtesy of the Gruppo Speleologico CAI di Fabriano.

Niches of sulfur-oxidizing cave bacteriaJL Macalady et al

2

The ISME Journal

electrodes were placed in the flowing water near thebiofilm. The working electrode was lowered to theair–water interface, then to the biofilm–water inter-face and subsequently lowered in increments toprofile the biofilm. We used both cyclic voltammetrybetween �0.1 and �1.8 V (vs Ag/AgCl) at scan ratesfrom 200 to 2000 mV s�1 with a 2 s conditioning step,and square wave voltammetry between �0.1 and�1.8 V (vs Ag/AgCl) at scan rates from 200 to1000 mV s�1, with a pulse height of 25 mV. Analyseswere carried out in sets of at least 10 sequentialscans at each sampling point in space, with the firstthree scans discarded.

Clone library constructionEnvironmental DNA was obtained using phenol–chloroform extraction as described in Bond et al.(2000) using 1� buffer A (100 mM Tris, 100 mM

NaCl, 25 mM EDTA, 1 mM sodium citrate, pH 8.0)instead of phosphate-buffered saline for the firstwashing step. Small subunit ribosomal RNA geneswere amplified by PCR from the bulk environmentalDNA. Libraries were constructed from four samplesusing the bacteria-specific primer set 27f and 1492r.Each 50ml reaction mixture contained environmen-tal DNA template (1–150 ng), 1.25 U ExTaq DNApolymerase (TaKaRa Bio Inc., Shiga, Japan), 0.2 mM

each dNTPs, 1� PCR buffer, 0.2 mM 1492r universalreverse primer (50-GGTTACCTTGTTACGACTT-30)and 0.2 mM 27f primer (50-AGAGTTTGATCCTGGCTCAG-30). A universal library was constructed fromsample FS06-12 using universal forward primer 533f(50-GTGCCAGCCGCCGCGGTAA-30) and 1492r. Ther-mal cycling was as follows: initial denaturation5 min at 94 1C, 25 cycles of 94 1C for 1 min, 50 1C for25 s and 72 1C for 2 min followed by a finalelongation at 72 1C for 20 min. PCR products werecloned into the pCR4-TOPO plasmid and used totransform chemically competent OneShot MACH1T1 Escherichia coli cells as specified by themanufacturer (TOPO TA cloning kit; Invitrogen,Carlsbad, CA, USA). Colonies containing insertswere isolated by streak plating onto Luria–Bertoniagar containing 50 mg ml�1 kanamycin. Plasmidinserts were screened using colony PCR withM13 primers (50-CAGGAAACAGCTATGAC-30 and50-GTAAAACGACGGCCAG-30). Colony PCR pro-ducts of the correct size were purified using theQIAquick PCR purification kit (Qiagen Inc., Valen-cia, CA, USA) following the manufacturer’s instruc-tions. Full-length sequences for 70–80 clones fromeach bacterial library were obtained, in addition to60 sequences from the universal library constructedfrom sample FS06-12.

Sequencing and phylogenetic analysisClones were sequenced at the Penn State UniversityBiotechnology Center using T3 and T7 plasmid-specific primers. Sequences were assembled with

Phred base calling using CodonCode Aligner v.1.2.4(CodonCode Corp., Dedham, MA, USA) and manu-ally checked for ambiguities. The nearly full-lengthgene sequences were compared against sequences inpublic databases using BLAST (Altschul et al., 1990)and submitted to the online analyses CHIMERA_CHECK v.2.7 (Cole et al., 2003) and Bellerophon 3(Huber et al., 2004). Putative chimeras were ex-cluded from subsequent analyses. Non-chimericsequences were aligned using the NAST aligner(DeSantis et al., 2006), added to an existing align-ment containing 4150 000 nearly full-length bacter-ial sequences in ARB (Ludwig et al., 2004), andmanually refined. Alignments were minimizedusing the Lane mask (1286 nucleotide positions)(Lane, 1991). Phylogenetic trees were computedusing neighbor joining (general time-reversiblemodel) with 1000 bootstrap replicates. Neigh-bor-joining trees were compared with maximumlikelihood trees (general time-reversible model, site-specific rates and estimated base frequencies). Bothanalyses were computed using PAUP* 4.0b10(Swofford, 2000).

Probe design and FISHProbes were designed and evaluated as described inHugenholtz et al. (2001), including checks againstall publicly available sequences using megaBLASTsearches of the non-redundant databases at NationalCenter for Biotechnology Information (NCBI). FISHexperiments were carried out as described inAmann (1995) using the probes listed in Table 1.Oligonucleotide probes were synthesized andlabeled at the 50 ends with fluorescent dyes (Cy3,Cy5 and FLC) at Sigma-Genosys (St Louis, MO,USA). Cells were counterstained after hybridizationwith 40,60-diamidino-2-phenylindole (DAPI),mounted with Vectashield (Vectashield LaboratoriesInc., Burlingame, CA, USA) and viewed on a NikonE800 epifluorescence microscope. Images werecollected and analyzed using NIS Elements AR2.30, Hotfix (Build 312) image analysis software.The object count tool was used to measure areascovered by cells hybridizing with specific probes.Ten images were collected for each sample, takingcare to represent the sample variability, and a totalDAPI-stained area of approximately 3� 104mm2

(equivalent to the area of 5� 104 E. coli cells) wasanalyzed for quantitation.

Statistical analysesThe program MINITAB (Minitab Inc., State College,PA, USA) was used for all statistical analyses. Atwo-sided Student’s t-test was used to comparesulfide/oxygen ratios associated with Thiothrixand epsilonproteobacteria, and correlations betweenparameters were analyzed using the Pearsonmethod.

Niches of sulfur-oxidizing cave bacteriaJL Macalady et al

3

The ISME Journal

Nucleotide sequence accession numbersThe 16S rRNA gene sequences determined in thisstudy were submitted to the GenBank databaseunder accession numbers EF467442–EF467519 andEU101023–EU101289.

Results

Field observations and geochemistryTwo common biofilm morphologies were collectedfrom a variety of sample sites in the cave system(Figure 1) over a 3-year period: (1) rock-attachedstreamers and (2) biofilms developed on the surfaceof fine sediment. Streamers 1–5 mm thick and5–20 cm long were attached to rocks in quicklyflowing or turbulent water (Supplementary Figure 1,upper panel). Sediment surface biofilms o1 mmthick were present in eddies or stream reaches withslower flow, at the interface between the watercolumn and fine gray sediment (SupplementaryFigure 1, lower panel). We commonly observed thetwo biofilm types coexisting in a patchwork corre-sponding to spatial variations in water flow. All ofthe biofilms contained abundant elemental sulfurparticles, as evidenced by microscopic observationsof the particles under polarized light and by rapiddissolution of the particles in ethanol (Nielsen et al.,2000).

Dissolved ion concentrations, including concen-trations of electron donors and acceptors such assulfides, ammonium and sulfate, were stronglycorrelated with specific conductivity (0.89oro0.98; Po0.0001). This result is consistent withprevious work showing that the cave water geo-

chemistry is controlled primarily by physical mix-ing processes rather than biological activity(Galdenzi et al., 2007). In contrast, dissolved oxygenconcentrations were not strongly correlated withspecific conductivity (r¼�0.39; P¼ 0.04). Totaldissolved sulfide and oxygen concentrations foreach biofilm sample collected in the study areplotted in Figure 2.

Voltammetric microelectrodes were used toinvestigate the spatial variability of sulfide andother redox-active species within biofilms and thesurrounding bulk water. Iron species were notdetected (Fe2þ and Fe3þ o5 mM, FeS oB0.5 mM asFeS monomer), consistent with total dissolved Feconcentrations o0.1 mM measured using inductivelycoupled plasma mass spectrometry. Oxygen con-centrations were at or below 15 mM (detection limit)for all waters analyzed, as expected based on valuesobtained from spectrophotometric tests at the samesites. Microelectrode profiles of sulfide concentra-tions within biofilms (Figure 3) are discussed below.

16S rDNA clone librariesClone libraries were constructed to investigateevolutionary relationships among the most abun-dant biofilm populations and to facilitate theevaluation of 16S rRNA probes. We recentlydescribed the phylogeny of clones from two Frasassistream biofilms dominated by Beggiatoa species(Macalady et al., 2006). Four additional biofilmswere selected for 16S rDNA cloning to capture awide range of geochemical conditions and biofilmmorphologies (Figure 2, cloned samples indicatedby large circles). Libraries were constructed using

Table 1 Oligonucleotide probes used in this study

Probe Target group Sequence (50-30) %formamide

(%)

Target site Reference

EUB338a Most bacteria GCTGCCTCCCGTAGGAGT 0–50 16S (338–355) Amann (1995)EUB338-IIa Planctomycetales GCAGCCACCCGTAGGTGT 0–50 16S (338–355) Daims et al. (1999)EUB338-IIIa Verrucomicrobiales GCTGCCACCCGTAGGTGT 0–50 16S (338–355) Daims et al. (1999)ARCH915 Archaea GTGCTCCCCCGCCAATTCCT 20 16S (915–934) Stahl and Amann

(1991)GAM42a Gammaproteobacteria, including

Frasassi Thiothrix clonesGCCTTCCCACATCGTTT 35 23S (1027–1043) Manz et al. (1992)

cGAM42a Competitor GCCTTCCCACTTCGTTT 35 23S (1027–1043) Manz et al. (1992)DELTA495a Most deltaproteobacteria, some

Gemmatimonas groupAGTTAGCCGGTGCTTCCT 45 16S (495–512) Macalady et al.

(2006)cDELTA495a Competitor AGTTAGCCGGTGCTTCTT 45 16S (495–512) Macalady et al.

(2006)SRB385 Some deltaproteobacteria, some

Actinobacteria and Gemmatimonasgroup

CGGCGTCGCTGCGTCAGG 35 16S (385–402) Amann (1995)

EP404 Epsilonproteobacteria AAAKGYGTCATCCTCCA 30 16S (404–420) Macalady et al.(2006)

EP404mis Negative control for EP404 AAAKGYGTCTTCCTCCA 30 16S (404–420) Macalady et al.(2006)

BEG811 Frasassi Beggiatoa clade CCTAAACGATGGGAACTA 35 16S (811–828) Macalady et al.(2006)

aCombined in equimolar amounts to make EUBMIX.

Niches of sulfur-oxidizing cave bacteriaJL Macalady et al

4

The ISME Journal

bacteria-specific primers because FISH analysesindicated that the biofilms contained few archaea(described below). Sample FS06-12 was also clonedusing universal primers in an attempt to retrievearchaeal sequences, but we only obtained bacterialsequences. Bacterial and universal primer setsyielded similar clones, with sequences most closelyrelated to ‘Thiobacillus baregensis’ accounting for91% in the universal library and 86% in thebacterial library.

The taxonomy of clones from each library issummarized in Figure 4 and Supplementary Table1. Between 25 and 97% of the clones in each librarywere associated with known or putative sulfur-oxidizing clades within gamma-, beta- and epsilon-proteobacteria (Figure 4). Gammaproteobacterialclones (Figure 5) include representatives of sulfur-oxidizing groups Beggiatoa (86–92% identity),Thiothrix (92–99% identity) and an unnamed cladecontaining ‘Thiobacillus baregensis’ (94–99% iden-tity) and the recently described sulfur-oxidizinglithoautotroph Thiovirga sulfuroxydans (86–93%identity) (Ito et al., 2005). Beggiatoa clones wereretrieved from four sample locations (Figure 4) and

form a coherent clade most closely related to non-vacuolate, freshwater Beggiatoa strains (Ahmadet al., 2006). Most betaproteobacterial clones wererelated to species of the sulfur-oxidizing generaThiobacillus (497% identity) or Thiomonas (90–99% identity).

Epsilonproteobacterial clones (Figure 6) werephylogenetically related to Arcobacter species or tomembers of the Sulfurovumales, Sulfuricurvalesand 1068 groups, which have few or no cultivatedrepresentatives. The majority of the clones wereassociated with the Sulfurovumales clade (Figure 4)and were distantly related to cultivated strains,including the named species Sulfurovum lithotro-phicum (88–94% identity). Sulfuricurvales groupclones were rare and shared 96–97% identity withSulfuricurvum kujiense. Frasassi clones in bothSulfurovumales and Sulfuricurvales were mostclosely related to clones from other sulfidic cavesand springs (98–99% identity), including filamentsfrom Lower Kane Cave (LKC) groups I and II (Engelet al., 2003). Arcobacter clones were diverse andonly distantly related to the closest cultivatedstrains (91–94% identity). The 1068 group has no

GS06-3

GS06-205

RS06-3

GS06-23CS05-6

RS06-102PC06-112

GS07-5

PC05-14

GS07-32

PC07-20

PC07-11PC07-27

PC07-24

LKC

PC06-110

PC05-16

PC05-11

FS06-12

FS06-10

FS06-4

CS06-2CS05-2

CS05-4CS06-101

RS06-101

0

100

200

300

400

500

600

0 5 10 15 20 25 30

Lower Kane Cave waters

sediment surface

streamers

Thiothrix

Beggiatoa

filamentous

epsilonproteobacteria

O2 (µM)

Σ H

2S (

µM)

RS05-6

RS05-22RS05-21

Figure 2 Dissolved oxygen and total sulfide concentrations for waters hosting Frasassi biofilm samples. Concentration field for LowerKane Cave (LKC) (Engel et al., 2003) is shown in gray for comparison. Symbols are colored if more than 50% of the biofilm cell area iscomposed of a single population or group based on FISH. Colored squares with error bars show the mean±1 s.d. for each major biofilmtype. Samples analyzed by 16S rDNA cloning are circled. The open diamond symbol represents a filamentous epsilonproteobacterialbiofilm from LKC reported in Engel et al. 2004. The range of conditions inside the RS05-21 (Thiothrix) biofilm based on voltammetry isindicated by the blue dashed box.

Niches of sulfur-oxidizing cave bacteriaJL Macalady et al

5

The ISME Journal

cultivated representatives and contains clones fromdeep subsurface igneous rocks, sulfidic caves andsprings, groundwater and wetland plant rhizo-spheres. Frasassi clones associated with the 1068group included phylotypes that shared less than92% identity with each other, and add significantlyto the known diversity within this clade. There wassupport in both neighbor-joining and maximumlikelihood phylogenies for the placement of the1068 group at the base of the epsilonproteobacteria(Figure 6).

Biofilm morphology and population structureTwenty-eight biofilms, including those selected for16S rDNA cloning, were homogenized and exam-ined using epifluorescence microscopy after FISH.Probes and hybridization conditions are listed inTable 1. Probe BEG811 was designed to bindspecifically to Beggiatoa populations in environ-mental samples from Frasassi (Macalady et al.,2006), and matches new Beggiatoa clones retrievedin this study (Figure 5). Probe EP404, targeting

epsilonproteobacteria, has no mismatches withFrasassi clones from this or previous studies(n4120), with the exception of seven clones withinthe Arcobacter and 1068 groups (Figure 6). TheEP404 probe does not match any publicly availablesequences outside the epsilonproteobacteria.

FISH experiments revealed three major biofilmtypes, as shown in Figure 2. The dominant group ineach biofilm sample accounted for more than 50%of the total DAPI cell area (Figure 2, coloredsymbols) with one exception (PC05-11). Sedimentsurface biofilms (n¼ 15) were dominated by 5–8 mmdiameter Beggiatoa filaments with abundant largesulfur inclusions and gliding motility. Streamers(n¼ 13) were dominated either by 1.5 mm diametergammaproteobacterial filaments with holdfasts andsulfur inclusions (Thiothrix), or by filamentousepsilonproteobacteria with holdfasts and no sulfurinclusions (1–2.5 mm diameter). Non-filamentouscells targeted by EP404 made up less than 5% ofthe EP404-positive cell area in each sample. Asreported previously (Macalady et al., 2006), the 23SrDNA probe GAM42a produces no signal from

RS05-22

GS06-205

marine sediment

-1200

-1000

-800

-600

-400

-200

0

200

0 50 100 150 200 250

-2700

-2200

-1700

-1200

-700

-200

300

0 500 1000 1500

Dep

th (

µm)

Σ H2S (µM)

PC06-110

Beggiatoa

filamentousEpsilonproteobacteria

to sediment−water interface

biofilm

sediment−water interface

sediment

water

water

water

Figure 3 Vertical sulfide concentration profiles measured using voltammetric microelectrodes. (a) Beggiatoa biofilm developed in thesediment and at the sediment–water interface. Zero depth on the y axis corresponds to the upper surface of the biofilm. Dissolved oxygenconcentrations were o15mM (detection limit) for all points. Marine sediment curve is from a Beggiatoa mat described in Jorgensen andRevsbech (1983). (b) Filamentous epsilonproteobacterial mat (‘streamers’) developed in turbulent water. The biofilm (gray box) isattached at the upstream end to a rock several centimeters above the sediment–water interface. Zero depth on the y axis corresponds tothe upper surface of the biofilm. Stream water flows both above and below the biofilm.

Niches of sulfur-oxidizing cave bacteriaJL Macalady et al

6

The ISME Journal

Frasassi Beggiatoa filaments at 35% (v/v) formamideconcentration. GAM42a-positive filaments withholdfasts and sulfur inclusions did not bind withprobes EP404 or Delta495a. Given these observa-tions, GAM42a-positive filaments could be assumedto be members of the Thiothrix clade. Archaeal cellsin the biofilms were rare or not detected using theprobe ARCH915. Consistent with this result, bacter-ial cell area measured using the EUBMIX probe wasconsistently within 15% of the area measured usingthe nucleic acid stain DAPI, which stains botharchaeal and bacterial DNA. Representative FISHphotomicrographs of the three major biofilm typesare shown in Supplementary Figure 2.

Discussion

Niches of sulfur-oxidizing populationsNiche differentiation is defined as the tendency forcoexisting populations to have different niches orenvironmental requirements. This concept is rarelyexplored in field studies of microorganisms but isone of the processes commonly invoked to explainthe enormous complexity of natural microbialcommunities. This study reveals that at least threeniche dimensions (sulfide, oxygen and water flowcharacteristics) are critical for niche differentiationamong major groups of sulfur-oxidizing bacteria pre-sent in the cave waters. Filamentous epsilonproteo-

bacteria dominate in waters with high sulfide andlow oxygen, while Thiothrix dominate in waterswith low sulfide and high oxygen. A similar patternwas suggested by 16S rDNA clone frequencies in astudy of LKC (Engel et al., 2004), but has not beendemonstrated until now. Figure 2 also shows thateither sulfide or oxygen concentrations alone arepoor predictors of biofilm compositions. All three ofthe dominant sulfur-oxidizing groups tolerate verylow oxygen concentrations (o5 mM). Furthermore,Beggiatoa-dominated biofilms colonize the entirerange of sulfide and oxygen concentrations mea-sured in the cave waters, but only in locations wherethe shear stress caused by the flowing water is lowenough to permit the accumulation of fine sediment.

The role of sulfide and oxygen concentrations indetermining the composition of the biofilms is mostclearly demonstrated in Figure 7, showing biofilmcommunity composition plotted against sulfide/oxygen ratios. We observed a strong linear correla-tion between filamentous epsilonproteobacterialarea % and the sulfide/oxygen ratio of water hostingstreamers (r¼ 0.97; Po0.0001). Correlations be-tween epsilonproteobacterial area % and eithersulfide or oxygen concentrations alone were weaker,with r values of 0.82 (P¼ 0.002) and �0.80(P¼ 0.003), respectively. In sharp contrast to strea-mer populations, Beggiatoa filaments colonized theentire range of sulfide, oxygen and sulfide/oxygenratios observed in the cave waters (Figures 2 and 7).

PC06-110 (streamers) PC05-11 (streamers)FS06-12 (streamers)

Sulfurovumales

Arcobacter

Sulfuricurvales

1068 group

“T.baregensis”

Beggiatoa

Thiothrix

Betaproteobacteria

S reducers

other

GS02-zEL (sedimentsurface mat)

GS02-WM (sedimentsurface mat)RS06-101 (streamers)

Figure 4 Distribution of 16S rDNA clones in Frasassi stream biofilms. Shaded wedges represent known or putative sulfur-oxidizingclades. Deltaproteobacteria associated with sulfate-reducing clades are shown in black. White wedges include all other clones (seeSupplementary Table 1). GS02-zEL and GS02-WM clone libraries are described in Macalady et al. (2006). Clones from both bacterial anduniversal libraries are combined for sample FS06-12.

Niches of sulfur-oxidizing cave bacteriaJL Macalady et al

7

The ISME Journal

Consistent with this result, Beggiatoa biofilms wereobserved immediately adjacent to both Thiothrixand filamentous epsilonproteobacterial streamersin situ, always in less turbulent or more slowlyflowing water.

Although changes in turbulence and water depthhave the potential to alter gas exchange and there-fore water chemistry, our data suggest that physicaleffects of water flow are more significant under therange of conditions present in the cave system.Because Beggiatoa filaments lack holdfasts, they can

be washed out by flows that are too strong to allowthe accumulation of fine sediment. Likewise, non-motile Thiothrix or epsilonproteobacteria filamentsmay become buried below the zone where oxidantsare available in stream reaches that are accumulat-ing sediment.

Our results support the idea that morphologicaland behavioral adaptations to physical constraintsare responsible for the separate niches colonized bylarge, filamentous bacteria (Schulz and Jorgensen,2001; Preisler et al., 2007). Similar to Beggiatoa

Figure 5 Neighbor-joining phylogenetic tree showing gamma(beta)proteobacteria. Frasassi clones are shown in bold followed by thenumber of clones represented in each phylotype. Neighbor-joining bootstrap values 450% are shown. Filled circles indicate nodespresent in the maximum likelihood phylogeny. Sequences identical to the probe BEG811 are indicated by the dashed line.

Niches of sulfur-oxidizing cave bacteriaJL Macalady et al

8

The ISME Journal

described from other environments, Beggiatoa atFrasassi inhabit diffusion-controlled sediments, andcan respond to changing geochemical conditions bygliding vertically in the sediment column. Sulfideconcentration profiles through Beggiatoa mats re-flect diffusion-controlled transport, although Fra-sassi sediments differ from typical marine or

lacustrine sediments in that sulfide diffuses bothfrom water above and sediment below the biofilms(Figure 3). Non-motile filaments with holdfasts(Thiothrix, filamentous epsilonproteobacteria) colo-nized niches with strong currents and a narrowersupply ratio of turbulently mixed sulfide andoxygen (Figure 7). Interestingly, vacuolated marine

Figure 6 Neighbor-joining phylogenetic tree showing epsilonproteobacteria. Frasassi clones are shown in bold followed by the numberof clones represented in each phylotype. Neighbor-joining bootstrap values 450% are shown. Filled circles indicate nodes present in themaximum likelihood phylogeny. Clades which hybridize with probe EP404 are indicated by the dashed line.

Niches of sulfur-oxidizing cave bacteriaJL Macalady et al

9

The ISME Journal

Beggiatoa with holdfasts have recently been identi-fied at cold seeps (Kalanetra et al., 2004). Frasassiclones are only distantly related (B88% identity) toBeggiatoa species with holdfasts (Ahmad et al.,2006), and we did not observe Beggiatoa withholdfasts in any of the cave samples.

Among streamer populations, we observed astrong niche differentiation between Thiothrix andfilamentous epsilonproteobacteria based on sulfide/oxygen ratios of bulk water. Microbial activitywithin biofilms modifies sulfide/oxygen ratios on asubmillimeter scale due to oxygen consumption atthe biofilm surface and sulfide production fromsulfate reduction or sulfur disproportionation dee-per in the biofilm. Sulfide concentrations based onmicroelectrode voltammetry varied up to severalfold with depth in individual biofilms, typicallyreaching the highest values in the center. Arepresentative sulfide concentration profile throughepsilonproteobacterial biofilm sample PC06-110 isshown in Figure 3b. Sulfide concentrations withinThiothrix streamers could not be profiled due to

their small size and rapid motions in the streamflow. The sulfide concentration just outside biofilmRS05-21 was approximately 200 mM, compared to350 mM in the interior. On the basis of these values,the approximate range of conditions inside theThiothrix biofilm are shown as a blue dashed boxin Figure 2, and clearly overlap with those permit-ting the growth of filamentous epsilonproteobacter-ia. Nonetheless, Thiothrix-dominated biofilmscontained at most 3.6 area % epsilonproteobacterialfilaments. The lack of epsilonproteobacteria in theinterior of Thiothrix biofilms with appropriatesulfide/oxygen ratios can be explained by factorssuch as competition for other limiting resources, orantagonistic interactions such as antibiotic produc-tion.

Both Thiothrix and filamentous epsilonproteobac-teria tolerate extremely low oxygen (o3 mM) and lowsulfide (o50 mM) concentrations. However, Thio-thrix-dominated biofilms do not occur at sulfideconcentrations above 210 mM, suggesting that sulfidetoxicity may play a role in excluding them fromhigh-sulfide environments. The absence of epsilon-proteobacterial filaments in waters with oxygenconcentrations above 3 mM is also striking, suggest-ing that oxygen toxicity may be limiting to epsilon-proteobacteria. Functional genomic studies providesome evidence that epsilonproteobacteria are un-iquely sensitive to oxygen among proteobacteriainhabiting sulfidic and microoxic environments dueto electron transport proteins with the potential toproduce millimolar levels of superoxide anionsduring oxidative stress (St Maurice et al., 2007).Filamentous epsilonproteobacteria are also appar-ently unable to store elemental sulfur intracellularly,an attribute that may limit their ability to consumetoxic levels of oxygen in the absence of high sulfideconcentrations. The inability to store intracellular So

may also be a disadvantage in permanently ortransiently low-sulfide environments such as thosewhere Thiothrix thrive.

Filamentous epsilonproteobacteria have pre-viously been described from LKC, Wyoming (Engelet al., 2003). Frasassi sequences differ significantlyfrom LKC clones, and do not hybridize withpreviously published oligonucleotide probesLKC59 and LKC1006 targeting environmentalgroups (Engel et al., 2003). LKC waters have anorder of magnitude lower sulfide concentrationsthan those hosting filamentous epsilonproteobacter-ia at Frasassi (Figure 2). Nonetheless, Figure 7shows that the LKC epsilonproteobacteria colonizewaters within the niche defined by high sulfide/oxygen supply ratios. As in LKC, no sulfur inclu-sions were observed in epsilonproteobacterial fila-ments, suggesting that this is a consistentphysiological attribute.

Frasassi biofilms host a wide variety of otherknown or putatitive sulfur-oxidizing taxa as shownin Figure 4. Close relatives of ‘Thiobacillus bare-gensis’ are present in all clone libraries analyzed to

r = 0.97

0

10

20

30

40

50

60

70

80

90

0 100 200 300 400

LKC

0 100 200 300 400sulfide/O2 ratio

area

% fi

lam

ento

us e

psilo

npro

teob

acte

ria

PC06-110

FS06-12

PC05-11

RS06-101

Figure 7 Sulfide/oxygen ratios for Frasassi biofilms analyzedusing fluorescence in situ hybridization (FISH). The upper panelshows a linear correlation (Po0.0001) between sulfide/oxygenratio and the microbial composition of streamers. The opendiamond (not included in correlation) represents a biofilm fromLower Kane Cave (Engel et al., 2004) and assumes 0.2 mM O2

(detection limit). The dashed arrow shows how the sulfide/oxygen ratio for the sample would change assuming an O2

concentration of 0.1 mM. The lower panel shows the distributionof biofilm types compared based on sulfide/oxygen ratios.Colored boxes and associated bars show the mean±1 s.d. foreach major biofilm type. Mean sulfide/oxygen ratios associatedwith Thiothrix and filamentous epsilonproteobacteria habitats aresignificantly different (P¼ 0.0004).

Niches of sulfur-oxidizing cave bacteriaJL Macalady et al

10

The ISME Journal

date, sometimes comprising the majority of clones(for example, sample FS06-12). Circumstantial evi-dence suggests that members of this novel clade aresulfur oxidizers (Elshahed et al., 2003; Ito et al.,2005). This study is the first to our knowledge toretrieve a significant percentage of Arcobacterclones from a freshwater environment. A marinestrain (Candidatus Arcobacter sulfidicus) that growsattached to solid substrates via filamentous sulfurstrands in high-flow, microoxic, sulfidic environ-ments has recently been described (Wirsen et al.,2002). Sulfide concentrations for growth of themarine Arcobacter (400–1200 mM) are broadly con-sistent with the Frasassi environment. Further workwill be required to evaluate the ecological roles ofArcobacter and other sulfur cycling populations inFrasassi cave waters.

Acknowledgements

This paper was improved by the comments of threeanonymous reviewers. We thank A Montanari for logis-tical support and the use of facilities and laboratory spaceat the Osservatorio Geologico di Coldigioco (Italy) and SMariani, S Galdenzi and S Cerioni for expert advice andfield assitance. We also thank P D’Eugenio, M Mainiero, SRecanatini, R Hegemann, H Albrecht, K Freeman and RGrymes for assistance in the field. We thank B Thomas andJ Moore for water analyses, and L Albertson and T Stofferfor laboratory assistance. E Fleming provided valuablecomments on the paper. DE contributed to this research asan undergraduate student and was supported in 2006 by aBarrett Foundation scholarship. This work was supportedby grants to JLM from the Biogeosciences Program of theNational Science Foundation (EAR 0311854 and EAR0527046) and NASA NAI (NNA04CC06A). GKD acknowl-edges support from the American Chemical SocietyPetroleum Research Fund (43356-GB2) and NSF-EPS-CoR-VT (EPS 0236976).

References

Ahmad A, Kalanetra KM, Nelson DC. (2006). CultivatedBeggiatoa spp. define the phylogenetic root of mor-phologically diverse, noncultured, vacuolate sulfurbacteria. Can J Microbiol 52: 591–598.

Amann RI. (1995). In situ identification of microorganismsby whole cell hybridization with rRNA-targetednucleic acid probes. Mol Microb Ecol Manual 3: 1–15.

Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ.(1990). Basic local alignment search tools. J Mol Bio215: 403–410.

Bond PL, Smriga SP, Banfield JF. (2000). Phylogeny ofmicroorganisms populating a thick, subaerial, predo-minantly lithotrophic biofilm at an extreme acid minedrainage site. Appl Environ Microbiol 66: 3842–3849.

Brendel PJ, Luther GW. (1995). Development of a goldamalgam voltammetric microelectrode for the deter-mination of dissolved Fe, Mn, O2, and S(-Ii) inporewaters of marine and fresh-water sediments. EnvSci Technol 29: 751–761.

Campbell BJ, Engel AS, Porter ML, Takai K. (2006). Theversatile e-proteobacteria: key players in sulphidichabitats. Nature Rev Microbiol 4: 458–468.

Cole JR, Chai B, Marsh TL, Farris RJ, Wang Q, Kulam SAet al. (2003). The Ribosomal Database Project (RDP-II):previewing a new autoaligner that allows regularupdates and the new prokaryotic taxonomy. NucleicAcids Res 31: 442–443.

Daims H, Bruhl A, Amann R, Schleifer K-H, Wagner M.(1999). The domain-specific probe EUB338 is insuffi-cient for the detection of all Bacteria: developmentand evaluation of a more comprehensive probe set.Syst Appl Microbiol 22: 434–444.

DeSantis TZ, Hugenholtz P, Larsen N, Rojas M, Brodie EL,Keller K et al. (2006). Greengenes, a chimera-checked16S rRNA gene database and workbench compatiblewith ARB. Appl Environ Microbiol 72: 5069–5072.

Druschel GK, Hamers RJ, Luther GW, Banfield JF. (2003).Kinetics and mechanism of trithionate and tetrathio-nate oxidation at low pH by hydroxyl radicals.Aquatic Geochem 9: 145–164.

Druschel GK, Sutka R, Emerson D, Luther GW, Kraiya C,Glazer B. (2004). Voltammetric investigation of Fe–Mn–S species in a microbially active wetland. In:Wanty RB, Seal RR (eds). Eleventh InternationalSymposium on Water–Rock Interaction WRI-11. Balk-ema: Saratoga Springs, NY, USA, pp 1191–1194.

Elshahed MS, Senko JM, Najar FZ, Kenton SM, Roe BA,Dewers TA et al. (2003). Bacterial diversity and sulfurcycling in a mesophilic sulfide-rich spring. ApplEnviron Microbiol 69: 5609–5621.

Engel AS, Lee N, Porter ML, Stern LA, Bennett PC, WagnerM. (2003). Filamentous ‘Epsilonproteobacteria’ dom-inate microbial mats from sulfidic cave springs. ApplEnviron Microbiol 69: 5503–5511.

Engel AS, Porter ML, Stern LA, Quinlan S, Bennett PC.(2004). Bacterial diversity and ecosystem function offilamentous microbial mats from aphotic (cave) sulfi-dic springs dominated by chemolithoautotrophic‘Epsilonproteobacteria’. FEMS Microbiol Ecol 51: 31.

Galdenzi S, Cocchioni M, Morichetti L, Amici V, Scuri S.(2007). Sulfidic groundwater chemistry in the Frasassicaves, Italy. J Cave Karst (in press).

Glazer BT, Luther GW, Konovalov SK, Friederich GE, NuzzioDB, Trouwborst RE et al. (2006). Documenting thesuboxic zone of the Black Sea via high-resolution real-time redox profiling. Deep Sea Res II 53: 1740–1755.

Glazer BT, Marsh AG, Stierhoff K, Luther GW. (2004). Thedynamic response of optical oxygen sensors andvoltammetric electrodes to temporal changes in dis-solved oxygen concentrations. Analyt Chim Acta 518:93–100.

Huber T, Faulkner G, Hugenholtz P. (2004). Bellerophon; aprogram to detect chimeric sequences in multiplesequence alignments. Bioinformatics 20: 2317–2319.

Hugenholtz P, Tyson GW, Blackall LL. (2001). Design andevaluation of 16S rRNA-targeted oligonucleotideprobes for fluorescent in situ hybridization. In: deMuro MA, Rapley R (eds). Gene Probes: Principles andProtocols. Humana Press: London. pp 29–42.

Ito T, Sugita K, Yumoto I, Nodasaka Y, Okabe S. (2005).Thiovirga sulfuroxydans gen. nov., sp. nov., a chemo-lithoautotrophic sulfur-oxidizing bacterium isolatedfrom a microaerobic waste-water biofilm. Int J SystEvol Microbiol 55: 1059–1064.

Jones DS, Lyon EH, Macalady JL. (2008). Geomicrobiology ofsulfidic cave biovermiculations. J Cave Karst (in press).

Niches of sulfur-oxidizing cave bacteriaJL Macalady et al

11

The ISME Journal

Jorgensen BB, Revsbech NP. (1983). Colorless sulfurbacteria, Beggiatoa spp. and Thiovulum spp., in O2

and H2S microgradients. Appl Environ Microbiol 45:1261–1270.

Kalanetra KM, Huston SL, Nelson DC. (2004). Novel,attached, sulfur-oxidizing bacteria at shallow hydro-thermal vents possess vacuoles not involved inrespiratory nitrate accumulation. Appl Environ Micro-biol 70: 7487–7496.

Lane DJ. (1991). 16S/23S rRNA sequencing. In: Stack-ebrandt E, Goodfellow M (eds). Nucleic Acid Techni-ques in Bacterial Systematics. Wiley: New York. pp115–175.

Ludwig W, Strunk O, Westram R, Richter L, Meier H,Yadhukumar et al. (2004). ARB: a software environ-ment for sequence data. Nucleic Acids Res 32:1363–1371.

Luther GW, Ferdelman TG, Kostka JE, Tsamakis EJ, ChurchTM. (1991). Temporal and spatial variability ofreduced sulfur species (FeS2, S2O3

2�) and porewaterparameters in salt-marsh sediments. Biogeochem 14:57–88.

Luther GW, Glazer BT, Hohmann L, Popp JI, Taillefert M,Rozan TF et al. (2001). Sulfur speciation monitored insitu with solid state gold amalgam voltammetricmicroelectrodes: polysulfides as a special case insediments, microbial mats and hydrothermal ventwaters. J Environ Monitoring 3: 61–66.

Macalady JL, Jones DS, Lyon EH. (2007). Extremely acidic,pendulous microbial biofilms from the Frasassi cavesystem, Italy. Environ Microbiol 9: 1402–1414.

Macalady JL, Lyon EH, Koffman B, Albertson LK, Meyer K,Galdenzi S et al. (2006). Dominant microbial popula-tions in limestone-corroding stream biofilms, Frasassicave system, Italy. Appl Environ Microbiol 72:5596–5609.

Manz W, Amann R, Ludwig W, Wagner M, Schleifer K-H.(1992). Phylogenetic oligodeoxynucleotide probes forthe major subclasses of Proteobacteria: problems andsolutions. Syst Appl Microbiol 15: 593–600.

Nielsen PH, de Muro MA, Nielsen JL. (2000). Studies onthe in situ physiology of Thiothrix spp. present inactivated sludge. Environ Microbiol 2: 389–398.

Preisler A, de Beer D, Lichtschlag A, Lavik G, Boetius A,Jorgensen BB. (2007). Biological and chemical sulfideoxidation in a Beggiatoa inhabited marine sediment.ISME J 1: 341.

Sarbu S, Galdenzi S, Menichetti M, Gentile G. (2000).Geology and biology of the Frasassi caves in centralItaly: an ecological multi-disciplinary study of ahypogenic underground karst system. In: Wilkens H(ed). Ecosystems of the World. Elsevier: New York, NY.pp 359–378.

Schulz HN, Jorgensen BB. (2001). Big bacteria. Annu RevMicrobiol 55: 105–137.

Skoog D, Holler F, Neiman T. (1998). Principles ofInstrumental Analysis, 5. Brooks/Cole–Thomas Learn-ing. Pacific Grove: CA.

Stahl DA, Amann R. (1991). Development and applicationof nucleic acid probes. In: Stackebrandt E, GoodfellowM (eds). Nucleic acid techniques in bacterial systema-tics. John Wiley & Sons Ltd: Chichester, England,pp 205–248.

St Maurice M, Cremades N, Croxen MA, Sisson G, SanchoJ, Hoffman PS. (2007). Flavodoxin:quinone reductase(FqrB): a redox partner of pyruvate:ferredoxin oxidor-eductase that reversibly couples pyruvate oxidation toNADPH production in Helicobacter pylori and Cam-pylobacter jejuni. J Bacteriol 189: 4764–4773.

Swofford DL. (2000). PAUP*: Phylogenetic Analysis UsingParsimony and Other Methods (Software). SinauerAssociates: Sunderland, MA.

Taillefert M, Bono AB, Luther GW. (2000). Reactivity offreshly formed Fe(III) in synthetic solutions and(pore)waters: voltammetric evidence of an agingprocess. Env Sci Technol 34: 2169–2177.

Taillefert M, Rozan TF. (2002). Electrochemical methodsfor the environmental analysis of trace elementsbiogeochemistry. In: Taillefert M, Rozan TF (eds).Environmental Electrochemistry: Analyses of TraceElement Biogeochemistry. American Chemical So-ciety: Washington DC, pp 3–14.

Vlasceanu L, Sarbu SM, Engel AS, Kinkle BK. (2000).Acidic cave wall biofilms located in the FrasassiGorge, Italy. Geomicrobiol J 17: 125–139.

Wirsen CO, Sievert SM, Cavanaugh CM, Molyneaux SJ,Ahmad A, Taylor LT et al. (2002). Characterization ofan autotrophic sulfide-oxidizing marine Arcobactersp. that produces filamentous sulfur. Appl EnvironMicrobiol 68: 316–325.

Xu K, Dexter SC, Luther GW. (1998). Voltammetricmicroelectrodes for biocorrosion studies. Corrosion54: 814–823.

Supplementary Information accompanies the paper on The ISME Journal website (http://www.nature.com/ismej)

Niches of sulfur-oxidizing cave bacteriaJL Macalady et al

12

The ISME Journal