APPLIED AND ENVIRONMENTAL MICROBIOLOGY,0099-2240/97/$04.0010

Sept. 1997, p. 3367–3373 Vol. 63, No. 9

Copyright © 1997, American Society for Microbiology

Distribution of Bacterioplankton in Meromictic LakeSælenvannet, as Determined by Denaturing Gradient

Gel Electrophoresis of PCR-Amplified GeneFragments Coding for 16S rRNA

LISE ØVREÅS,1* LARRY FORNEY,2 FRIDA LISE DAAE,1 AND VIGDIS TORSVIK1

Department of Microbiology, University of Bergen, N-5020 Bergen, Norway,1 and Center for Microbial Ecology,Michigan State University, East Lansing, Michigan2

Received 19 March 1997/Accepted 30 June 1997

The community structure of bacterioplankton in meromictic Lake Sælenvannet was examined by PCRamplification of the V3 region of 16S rRNA from microbial communities recovered from various depths in thewater column. Two different primer sets were used, one for amplification of DNA from the domain Bacteria andanother specific for DNA from the domain Archaea. Amplified DNA fragments were resolved by denaturinggradient gel electrophoresis (DGGE), and the resulting profiles were reproducible and specific for the com-munities from different depths. Bacterial diversity estimated from the number and intensity of specificfragments in DGGE profiles decreased with depth. The reverse was true for the Archaea, with the diversityincreasing with depth. Hybridization of DGGE profiles with oligonucleotide probes specific for phylogeneticgroups of microorganisms showed the presence of both sulfate-reducing bacteria and methanogens throughoutthe water column, but they appeared to be most abundant below the chemocline. Several dominant fragmentsin the DGGE profiles were excised and sequenced. Among the dominant populations were representativesrelated to Chlorobium phaeovibrioides, chloroplasts from eukaryotic algae, and unidentified Archaea.

Meromictic lakes, such as Lake Sælenvannet, are character-ized by two permanent strata separated by a steep salinitygradient called the chemocline (39). The mixing of the upperand lower strata are limited because of the physical surround-ings and differences in water density. The interfaces in ameromictic lake can appear as zones with elevated bacterialactivity and where bacterial community composition maychange. Such zones are therefore suitable for studies of bac-terial diversity and community structure. A fundamental un-derstanding of aquatic ecosystems requires knowledge of thediversity, distribution, and function of bacterioplankton be-cause of their importance in biogeochemical processes. Pro-karyotic organisms are known to be functionally and metabol-ically diverse, exceeding that of all the eukaryotes (10).Assessing the bacterial diversity within an ecosystem is difficultbecause for most bacteria, taxonomic classification requirestheir isolation and cultivation. This approach is impractical,particularly with natural ecosystems, due to the high number ofbacteria that are unculturable (11, 36).

In recent years, investigators have employed methods basedon molecular phylogeny as a means to assess the diversity anddistribution of prokaryotes in marine ecosystems (9, 11). Thesemethods rely largely upon sequence information of genes thatare universally conserved, yet sufficiently different to reflect thephylogeny of the prokaryotes. The nucleotide sequences ofrRNA have been used to develop group-specific oligonucleo-tide probes (3, 6, 18, 30). Such probes can be used to assess thediversity and distribution of bacterial and archaeal populationsby Southern hybridization (30) as well as fluorescence in situhybridization (2, 6, 24, 25). Alternatively, amplified regions ofthe rrn operon can be cloned, and the species abundance pro-

files of communities can be determined by sequence analysis ofrepresentative clones (9, 11, 36). These studies have routinelyshown that novel and previously uncharacterized microbialpopulations dominate aquatic ecosystems (5, 7, 9, 11).

Denaturing gradient gel electrophoresis (DGGE) has beenused to resolve PCR-amplified regions of genes coding for 16SrRNA (16S rDNA) based solely on differences in nucleotidesequence (22). This has proven to be a simple approach toobtain profiles of microbial communities that can be used toidentify temporal or spatial differences in community structureor to monitor shifts in structure that occur in response toenvironmental perturbations (19, 20, 37). Moreover, since eachDNA fragment in the profile is likely to be derived from one(or few) phylogenetically distinct populations, one can readilyobtain an estimate of species number and abundance based onthe number and intensity of amplified fragments in the profile.It has also been possible to infer the phylogeny of communitymembers by DNA sequence analysis of amplified fragmentsafter they have been excised (8, 21, 28, 32).

Few studies of community diversity have been performed formeromictic lakes. Studies of a stratified fjord have focused onthe distribution of sulfate-reducing bacterial populations (24,32). By in situ hybridization, the maximum number of sulfate-reducing bacteria was found in and below the chemocline (24).However, in the same study, sulfate-reducing bacteria werealso cultivated from the oxic part of the water column by themost-probable-number method. DGGE analysis of the samestratified fjord revealed that the distribution of sulfate-reduc-ing bacteria was continuously changing in composition but thatthe total mass remained the same throughout the water col-umn (32). Previous studies of Lake Sælenvann showed highrates of chemoautotrophic productivity at the interface and thehighest rates of sulfate reduction just below the interface (14,15). These old studies, however, lack a solid taxonomic classi-fication based on molecular data.

* Corresponding author. Mailing address: Department of Microbi-ology, University of Bergen, Jahnebakken 5, N-5020 Bergen, Norway.Fax: 47 55 589671. E-mail: Lise.Ovreas@im.uib.no.

3367

on June 16, 2015 by guesthttp://aem

.asm.org/

Dow

nloaded from

In this study, we applied the DGGE method to determinethe relative genetic complexity of microbial communities atdifferent depths in the meromictic Lake Sælenvannet. Thetechnique was also used to monitor community changes inspace and time. Two primer sets, one specific for DNA fromthe domain Bacteria and another specific for DNA from thedomain Archaea, allowed depth profiles for these groups oforganisms to be compared. By hybridizing blots of the DGGEprofiles with group-specific probes against sulfate-reducingbacteria and methanogens, the distribution of these bacteria inthe water column was determined. The taxonomic affiliation ofsome putative dominating members of the community wasdetermined by sequencing dominating DGGE bands.

MATERIALS AND METHODS

Water samples. Water samples were collected from various depths in LakeSælenvannet three times between March 1995 and April 1996 with a submersiblepump as described by Tuomi et al. (35). One-liter samples were collected fromthe upper oxic water layer, the chemocline, and the anoxic bottom layer at depthsof 1.5, 2.5, 3, 5, and 15 m. Temperature and salinity were measured with asalinoterm and a YSI model 33 S-C-T meter (Yellow Springs Instrument Co.,Yellow Springs, Ohio). Oxygen was measured with an oxygen electrode and aYSI model 57 oxygen meter (Yellow Springs Instruments). Sulfide was deter-mined colorimetrically as described by Grasshoff et al. (12). Samples for micro-biological analyses were transported to the laboratory within 1 h. The bacterio-plankton community was harvested by centrifugation of 1.5-ml samples at10,000 3 g for 30 min. The cell pellets were washed once in phosphate-bufferedsaline (137 mM NaCl, 2.7 mM KCl, 4.3 mM Na2HPO4 z 7H2O, 1.4 mMMKH2PO4 [pH 7.3]), resuspended in 50 ml of sterile distilled water, and frozen.The cell suspensions were thawed and used for whole-cell PCR amplification. Toverify that the centrifugation of the water samples efficiently recovered the cells,a microscopic examination was performed. The supernatants were stained withDAPI (49,6-diamidino-2-phenylindole) (23) and examined by fluorescent micros-copy. Less than 3% of the total bacterial population was found to be left in thesupernatant that was discarded.

PCR primers and probe sequences. The V3 region of 16S rRNA genes fromthe bacterioplankton community was amplified by PCR with two different primersets (Table 1). One primer set, PRBA338f and PRUN518r, amplified a product236 bp in length. The PRBA338f primer complements a region conserved amongmembers of the domain Bacteria. This primer does not perfectly match all knownbacterial sequences, and the stringency has been reduced to permit thesematches. In addition, the reverse primer, PRUN518r, is based on a universallyconserved region. PCR mixtures used for amplification of bacterial sequencescontained 10 ml of washed cell suspension, 0.5 mM each primer, 100 mM eachdeoxynucleoside triphosphate, 10 ml of 103 Red Hot PCR buffer, 0.5 U of RedHot DNA polymerase (Advanced Biotechnologies Ltd., Surrey, United King-dom), 0.1% bovine serum albumin, and sterile water to a final volume of 100 ml.PCR amplification was done with the following program: 92°C for 2 min; 30cycles of denaturation at 92°C for 1 min, annealing at 55°C for 30 s, and extensionat 72°C for 1 min; and a single final extension at 72°C for 6 min. Archaeal 16SrRNA sequences were obtained by nested PCR amplification with the conditionsdescribed above, but 5% (wt/vol) of acetamide was added to minimize nonspe-cific annealing of the primers to the bacterial template present in the samples(27), and the annealing temperature was lowered to 53.5°C. For nested PCR,PRA46f and PREA1100r were first used to produce a 1,072-bp fragment, which

was then used as a template in combination with the primers PARCH340f andPARCH519r. This amplification was performed with the same settings as thoseused in the first round of amplification except that acetamide was excluded. TheGC clamps described by Muyzer et al. (20) were included on the 59 end of theforward primers for both the Bacteria and Archaea (PRBA338f andPARCH340f). The presence of PCR products was confirmed by analyzing 5 ml ofproduct on 1.5% agarose gels and staining with ethidium bromide.

DGGE. DGGE was performed with a Hoefer Scientific SE600 vertical dual-cooler system. PCR samples were loaded onto 8% (wt/vol) polyacrylamide gelsin 0.53 TAE (20 mM Tris, 10 mM acetate, 0.5 mM Na2 EDTA [pH 7.4]). The8% (wt/vol) polyacrylamide gels (bisacrylamide gel stock solution, 37.5:1; Bio-Rad Laboratories, Inc.) were made with denaturing gradients ranging from 15 to55% (where 100% denaturant contains 7 M urea and 40% formamide). Theelectrophoresis was run at 60°C, first for 10 min at 20 V and subsequently for 3 hat 200 V. After electrophoresis, the gels were soaked for 45 min in SYBR GreenI nucleic acid gel stain (1:10,000 dilution; Molecular Probes, Eugene, Oreg.),rinsed for 10 min in water, and photographed on a UV transillumination tablewith a Polaroid (Cambridge, Mass.) MP4 Land camera.

Blotting and hybridization analyses of DGGE gels. DGGE-separated PCRproducts were blotted and hybridized to rRNA-specific probes. After electro-phoresis, the gel was allowed to equilibrate in 13 TBE buffer (89 mM Tris-borate[pH 8.0], 89 mM boric acid, 2 mM Na2EDTA) for 15 min. The nucleic acids weretransferred electrically to a nylon membrane (Hybond N; Amersham, LittleChalfont, Buckinghamshire, United Kingdom) with 400 mA of current for 15 minby use of a Sammy Dry electroblotter apparatus (Schleicher & Schuell). Themembrane was then baked for 2 h at 80°C to cross-link the DNA fragments to themembrane.

The membranes were hybridized with oligonucleotide probes that were spe-cific for either sulfate-reducing bacteria or for the methanogenic order Meth-anococcales or Methanosarcinales and Methanomicrobiales (Table 1) (37). Probeswere labeled with digoxigenin with the DIG oligonucleotide 39-end labeling kitby the protocol recommended by the manufacturer (Boehringer, Mannheim,Germany). Hybridization was performed as described by Stahl and Amann (30)in 53 SSC (13 SSC is 0.15 M NaCl plus 0.015 M sodium citrate)–0.1% (wt/vol)sodium dodecyl sulfate (SDS). The probe specific for the d subdivision of thesulfate-reducing bacteria was hybridized at a temperature of 57°C. Under thesehybridization conditions, the probe did not cross-react with DNAs from mem-bers of the a, b, or g subdivision of the class Proteobacteria, high-G1C-contentgram-positive bacteria, or the cytophaga-flexibacter or bacteroides groups. Theprobe against Methanosarcinales and Methanomicrobiales, M(SA/MI)355, washybridized at 49°C, and the probe against Methanococcales, M(CO/BA)377, washybridized at 58.5°C. By using these temperatures, the two probes did notcross-hybridize. After hybridization, the membranes were washed twice for 5 minin 23 SSC–0.1% SDS at room temperature and twice for 15 min in 0.13SSC–0.1% SDS at the hybridization temperatures. Detection of hybrids wasperformed as described by the manufacturer. Hybridized digoxigenin-labeledprobes were detected with a digoxigenin-specific antibody coupled to alkalinephosphatase, which acts on the substrate CDP-star (Boehringer). After themembranes were rinsed, they were wrapped in plastic wrap and exposed toHyperfilm-ECL (Amersham) for 5 to 30 min. To remove hybridized probe, themembranes were boiled in 0.5% SDS, and the solution was allowed to cool toroom temperature.

Sequencing of DGGE fragments. DNA fragments to be nucleotide sequencedwere punched from the gel with sterile pipette tips and placed in sterilized vials,and 20 ml of sterilized water was added. The DNA was allowed to passivelydiffuse into the water at 4°C overnight. Ten microliters of the eluate was used astemplate DNA in a PCR with the primers and conditions described above.Following amplification, the PCR products were analyzed by DGGE to confirmtheir electrophoretic mobility relative to the fragment from which they were

TABLE 1. Primer and probe sequences and positions

Primer or probea Positions(bases)b Target Sequence Reference

PRBA338f 338–357 Bacteria, V3 region 59ACTCCTACGGGAGGCAGCAG 16PRUN518r 518–534 Universal, V3 region 59ATTACCGCGGCTGCTGG 20PRA46f 46–60 Archaea 59(C/T)TAAGCCATGC(G/A)AGT 34PREA1100r 1100–1117 Archaea 59(T/C)GGGTCTCGCTCGTT(G/A)CC 34PARCH340f 340–357 Archaea, V3 region 59CCCTACGGGG(C/T)GCA(G/C)CAG This paperPARCH519r 519–533 Archaea, V3 region 59TTACCGCGGC(G/T)GCTG This paperSRB-385 385–402 SRB, d-Proteobacteria 59CGGCGTCGCTGCGTCAGG 3M(CO/BA)377 377–395 Methanococcales 59CCCCCGTCGCACTT(T/G)CGTG 33M(SA/MI)355 355–372 Methanomicrobiales 59GTAAAGTTTTCGCGCCTG 33GC clampc 59CGCCCGCCGCGCGCGGCGGGCGGGGCGGGGGCACGGGGGG 20

a f, forward primer; r, reverse primer.b The numbering of positions is based on E. coli 16S rRNA.c The GC clamp was attached to the 59 end of the PRBA338f and PARCH340f primers.

3368 ØVREÅS ET AL. APPL. ENVIRON. MICROBIOL.

on June 16, 2015 by guesthttp://aem

.asm.org/

Dow

nloaded from

excised. The PCR products were directly sequenced with an AmpliCycle se-quencing kit (Perkin-Elmer, Roche Molecular Systems, Branchburg, N.J.) asdescribed in the manufacturer’s instructions. The products of the sequencingreactions were separated on a 6% (wt/vol) acrylamide-bisacrylamide (19:7.1)–7M urea wedge-shaped gradient gel (0.2 to 0.6 mm) with an LKB 2010 sequencingsystem. The gels were run at 55°C for 3 h at a constant voltage of 1,750 V,sequentially washed for 20 min in 10% acetic acid, 20 min in 10% acetic acid with1% glycerol, and briefly in distilled water, and then dried overnight on a glassplate. The gels were then exposed overnight to Hyperfilm-bmax (Amersham).The sequences were aligned to 16S rRNA sequences obtained from the NationalCenter for Biotechnology Information database by the BLAST search (1).

Nucleotide sequence accession numbers. The sequences obtained in this studyare available in GenBank under accession numbers AF007938 (fragment 2),AF007939 (fragment 5), AF007940 (fragment 7), and AF007941 (fragment 13).

RESULTS

Characteristics of Lake Sælenvannet. Lake Sælenvannet is apermanently stratified estuary in western Norway (14) whereinthe oxic surface water is separated from the anoxic bottomwater by a permanent chemocline. A narrow channel (roughly0.5-m deep and 2-m wide) that connects the lake to the searestricts vertical water transport and deep mixing, thus limitingthe exchange of oxygenated water. Consequently, the surfacelayer (0 to 2 m) is oxic and brackish with a salinity of 2 to 12%,whereas the bottom layer (2.5 to 26 m) is anoxic with a salinityof 20 to 22% (Fig. 1). Bacterial numbers based on microscopiccounts ranged from 106 to 107 ml21, with the highest numberfound at the interface between the oxic and anoxic water layers(35).

Optimization of whole-cell PCR amplification. To optimizethe PCR amplification of microbial populations, the effect ofthe number of cells used as a template was determined. Bac-terial suspensions containing between 104 and 107 cells per mlall gave a good yield of amplified product, regardless of thesampling depth (Fig. 2A). The DGGE profiles of replicatesamples taken from the same depth were essentially identical(Fig. 2B). These results indicate that the number of cellspresent in samples used as a template for PCR did not signif-icantly affect the analysis of microbial community structure byDGGE.

Effect of depth on microbial community structure. 16SrDNA fragments obtained by PCR amplification of whole-cellmicrobial communities at different depths in Lake Sælenvan-net were analyzed by DGGE. Two different primer sets wereused in this study, one for DNA from Bacteria and the other forDNA from Archaea. The bacterial populations from differentdepths showed distinctly different profiles based on the num-ber and distance of migration of the PCR products (Fig. 2B).The bands in the DGGE profile correspond to 16S rRNAfragments that differ in nucleotide sequence. They thus reflectdistinct numerically dominant microbial populations in thecommunity. Moreover, the intensity of the PCR product isbelieved to be proportional to the abundance of the templateand therefore the abundance of each population (20). Hence,the appearance and disappearance of fragments in the DGGEprofiles approximate shifts in the microbial community struc-ture. Comparison of DGGE profiles shows that the structuresof bacterial communities at various depths differ from oneanother (Fig. 2B). In general, the number of populations de-creased with increasing depth. The oxic zone had the greatestnumber of bacterial populations, and they were present inapproximately equal numbers. Some populations were presentonly in the oxic zone (Fig. 2B, fragments 1, 3, and 4). At adepth of 3 m, no unique numerically dominant populationswere observed. In contrast, the anoxic zone had three domi-nant populations (Fig. 2B, fragments 5 to 7) in addition to

FIG. 1. Depth profile of Lake Sælenvann. The vertical distribution of oxygen(E), sulfide (h), temperature (F), and salinity (■) is shown.

FIG. 2. Agarose gel (A) and DGGE (B) analyses of 16S rDNA fragments ofdifferent bacterioplankton communities after PCR amplification with the Bacte-ria primers. The bacterioplankton communities from five different depths wereinvestigated, and from each sample, three different dilutions were analyzed (1,1021, and 1022). nk, negative control; pGEM, size marker. The arrowhead inpanel A indicates the 236-bp amplified fragment. The positions and numberingof bands discussed in the text are indicated with arrows in panel B.

VOL. 63, 1997 DISTRIBUTION OF BACTERIOPLANKTON IN A MEROMICTIC LAKE 3369

on June 16, 2015 by guesthttp://aem

.asm.org/

Dow

nloaded from

other minor populations. These dominant populations werealso present in the oxic zone but at lower numbers. Somepopulations appeared to be common throughout the watercolumn (Fig. 2B, fragments 6 and 7).

Analyses of 16S rDNA fragments amplified with the primersspecific for DNA from Archaea showed that this part of themicrobial community structure as well varied between differentzones of the water column (Fig. 3). One dominant populationwas common to all the depths sampled (fragment 9). Thediversity of members of Archaea was higher in samples takenfrom the anoxic zone than from the oxic zone. This was oppo-site from the findings obtained with primers specific for DNAfrom Bacteria, which showed the greatest diversity in the oxiczone. All the archaeal populations present in the upper oxiczone were present in the other zones as well (Fig. 3, fragments8, 9, 11, and 12). At a depth of 2.5 m, there was one minorpopulation that appeared to be unique (fragment 10). In theanoxic zone, the number of distinct populations increased withdepth. Three populations appeared to be unique to the anoxiczone (Fig. 3, fragments 14 to 16). One of these (fragment 16)was also present at 3 m and, given the width of the fragment,may consist of several different archaeal populations.

Temporal variations in microbial community structure. Thecommunity structure in the different regions of the water var-ied at the different sampling times (Fig. 4). Three samples werecollected from March 1995 until March 1996, and major dif-ferences were observed at each depth for every sampling time.Differences from one month to another were also detected(Fig. 4). The temporal variation showed that the chemoclinewas at a higher position in the water column in 1996 (2.5 m)than in 1995 (3.0 m). This was illustrated by the salinity profile(data not shown) and the position of the dominating popula-tion represented by fragment 17 (Fig. 4). Based on the num-bers of DGGE fragments, higher numbers of bacterial popu-lations were present at a depth of 2.5 m than at other depths.This illustrates that some of the temporal differences observedare due to changes in the depth of the chemocline.

Hybridization analysis of microbial populations. There is apermanent chemocline in Lake Sælenvannet caused by no

mixing between oxic and anoxic zones. Because of this, onewould expect anaerobic processes such as sulfate reduction andmethanogenesis to be dominant below the chemocline and thestructure of the microbial communities to reflect this. Thepresence of sulfate-reducing bacteria in water samples takenfrom various depths was determined by Southern hybridizationof DGGE gels with a digoxigenin-labeled 16S rRNA oligonu-cleotide probe. The probe used was specific for gram-negativesulfate-reducing bacteria of the d subdivision of the class Pro-teobacteria. The probe hybridized with two fragments presentin samples from all the different depths (Fig. 2B, fragments 5and 6). Weak hybridization signals were obtained with samplesfrom the oxic water phase, whereas samples from the chemo-cline and to a depth of 15 m gave intense hybridization signals.These results indicate that there were sulfate-reducing popu-lations within the entire water column but that they were moredominant in the anaerobic zone.

Similar experiments were done with oligonucleotide probesspecific for various groups of methanogens. Two fragmentshybridized with the M(SA/MI)335 probe (Fig. 3, fragments 11and 13), indicating that methanogenic bacteria in the watercolumn belonged to the Methanomicrobiales and Methanosar-cinales. The M(CO/BA)377 probe did not hybridize to anyfragments, suggesting that common marine methanogens ofthe Methanococcales group were absent or in low abundance.However, if we reduced the hybridization temperature by 1°C,we found that some of the target for the M(SA/MI)355 andM(CO/BA)377 probes cross-hybridized and that several of theArchaea DNA fragments hybridized with both probes. Thismay indicate that there are bacteria present with high sequencesimilarity to both groups of methanogens. The probes stronglyhybridized to fragments amplified from microbial communitiesin the 3-, 5-, and 15-m samples. The same hybridization patternwas observed in samples from oxic zones but at a much lowerintensity. These data indicate that the majority of methanogensare present below the chemocline.

Sequencing and identification of DGGE fragments. To con-firm results from hybridization experiments, four bacterialfragments and two archaeal fragments were excised from theDGGE gels and amplified by PCR and the nucleotide se-

FIG. 3. DGGE analyses of 16S rDNA fragments of Archaea obtained afterusing nested PCR amplification on samples of the microbial community from fivedifferent depths. The positions and numberings of bands discussed in the text areindicated with arrows.

FIG. 4. DGGE analysis showing variations in time and space of bacterio-plankton community structure from Lake Sælenvannet. Bacterial primers wereused for PCR amplification. Five different depths and three different samplingtimes were processed. D1, 22 March 1995; D2, 29 January 1996; D3, 5 March1996. The arrow labeled 17 indicates the fragment with high sequence similarityto C. phaeovibrioides DNA.

3370 ØVREÅS ET AL. APPL. ENVIRON. MICROBIOL.

on June 16, 2015 by guesthttp://aem

.asm.org/

Dow

nloaded from

quences were determined. None of the sequences obtainedprecisely matched those found in the database, suggesting thatpreviously uncharacterized populations can be present in LakeSælenvannet. Fragment 7 (Fig. 2B) (and fragment 17 [Fig. 4])showed 98% sequence similarity to a Chlorobium phaeovibri-oides DNA fragment. Fragment 2 (Fig. 2B) showed 90% se-quence similarity to 16S rRNA chloroplast genes from theeukaryotic alga Euglena gracilis. This fragment was present atboth 1.5 and 2.5 m but not at greater depths. The absorptionspectra of the water masses at this depth also suggest that thereare green algae and cyanobacteria present, and these are im-portant components of the picoplankton in the lake. Fragment5 (Fig. 2B) showed 96% similarity to an unidentified marinebacterium (Hstpl 4). This is an uncultured microorganism as-sociated with the seagrass Halophila stipulaceae (38). Fragment13 showed 91% similarity to a DNA fragment of an unidenti-fied Archaea clone (SBAR5) isolated from coastal marine en-vironments (5). Fragments 8 and 11 (Fig. 3) yielded a PCRproduct, but further sequencing failed due to impurities of theamplified product.

Visually dominant microbial populations. Nucleotide se-quence analysis indicated that C. phaeovibrioides, a green sul-fur bacterium, was dominant at and below the chemocline. Thepresence of green sulfur bacteria in the water column was alsosuggested by absorption spectral analysis of the water. Theabsorption maxima at 720 nm is consistent with the absorptionmaxima of bacteriochlorophylls c, d, and e found in greensulfur bacteria and Chloroflexus of between 715 and 755 nm(29). Moreover, electron microscopic analysis of water samplesfrom a 2.5-m depth showed that organisms with morphologysimilar to that of C. phaeovibrioides were dominant members ofthe community. Analyses of water samples at other depthsshowed significantly lower number of Chlorobium-like cellsand total number of organisms (Fig. 5). This clearly shows thatthis Chlorobium population has an affinity for the chemocline.

DISCUSSION

In this study, the microbial diversity and community struc-ture within the water column in a meromictic lake were esti-mated. The microbial ecology of such lakes is of specific inter-est because the chemocline is believed to spatially separatedistinct microbial communities. Spatial and temporal studies ofthe heterogeneity of bacterial populations in marine environ-ments have been limited mainly due to the labor intensivenessof cloning and sequencing, oligonucleotide probing, and culti-vation approaches. DGGE separation of PCR products al-lowed a rapid assessment of the relative changes in the com-munity structure of the amplifiable bacterioplankton within thelake. By use of this approach, we could study the distribution ofmicrobial populations along a depth profile, and because thetechnique is rapid and reproducible, we were also able to studythe temporal variations in the community. For determinationof more specific traits in the community structure, hybridiza-tion and sequencing analysis of individual bands were per-formed. There has been a concern with chimera formationduring the PCR amplification process (17, 27, 31). To avoidthis problem, we chose to amplify the shorter and highly vari-able V3 region of 16S rRNA for DGGE analysis. The primerswere designed to specifically amplify DNA from either Bacteriaor Archaea. To check the reproducibility of PCR products andthe subsequent DGGE patterns, many replicate amplificationsand DGGE gels were run. All were highly reproducible. If abias in the amplification process exists, it is constant and thebanding profiles can be reliably compared. The shifts that we

are detecting by the different gel patterns are indeed due tochanges in the environmental samples.

The lake system is permanently stratified with an oxic-anoxicinterface in the photic zone at about 2.5 to 3 m. The moststriking results were the shift in DGGE pattern between theoxic and anoxic parts of the lake. In the oxic part, oxygenicphotosynthetic activities occur, as reflected by the presence ofphotosynthetic organisms. The DGGE profile of Bacteria inthe oxic zone, compared to that in the anoxic zone, showed ahigher number of individual bands (15 to 20) with an equiva-lent intensity. In the anoxic zone, the number of bands de-creased, but the intensity of some of the bands increased. Thus,the highest diversity was found in the oxic zone, with no dom-inant populations. Hofle and Brettar (13) found the highestnumber of dominant taxa in the biologically most active re-gions of the water column. Their data indicated that three toseven taxa comprised as much as 85 to 90% of the total com-munity. A rather limited number of taxa in marine systems hasalso been reported by Rehnstam et al. (26).

Our results showed a decrease in diversity but an increase inintensity of some of the populations of Bacteria; we also foundthat few dominant bacterial populations were present in andbelow the chemocline. The chemocline in Lake Sælenvannet isa unique habitat where H2S, light penetration, and low oxygenare present. This habitat is ideal for chemoautotrophs as wellas photoautotrophs. The photoautotrophic bacteria that do notevolve oxygen during photosynthesis are able to use a wholerange of electron donors such as H2S and organic acids. Hy-bridization with group-specific probes showed that sulfate-re-ducing bacteria were present in the entire water column, butthe majority of them were present in the anoxic zone. Thenumber of sulfate-reducing organisms in the anoxic part of the

FIG. 5. Electron micrographs of air-dried cells (communities) from differentdepths in Lake Sælenvannet. (A) 2.5 m; (B) 5 m. Arrowheads indicate Chloro-bium-like cells.

VOL. 63, 1997 DISTRIBUTION OF BACTERIOPLANKTON IN A MEROMICTIC LAKE 3371

on June 16, 2015 by guesthttp://aem

.asm.org/

Dow

nloaded from

water has earlier been estimated to be 10 to 50 ml21 (14).Ramsing et al. (24) found by in situ hybridization that themaximum number of sulfate-reducing bacteria occurred in andbelow the chemocline of a stratified fjord. Teske et al. (32)studied the same fjord and found that the population compo-sition of sulfate-reducing bacteria was continuously changingthroughout the water column, even though the total mass re-mained fairly constant. They also found that DGGE analysis of16S rDNA-derived PCR products revealed a highly differenti-ated pattern of active versus dormant subpopulations.

In the anoxic part of Lake Sælenvannet, reductive processestake place. The DGGE method may distinguish those micro-organisms responsible for the process. The numbers of Bacteriapopulations decreased in the anoxic zone. On the other hand,the numbers of Archaea populations were lowest in the oxiczone and increased with depth. These results are in agreementwith observations made by others. Evidence from the Antarc-tic, Atlantic, and Pacific Oceans suggest that uncultivated Ar-chaea may dominate in marine environments (5, 7, 9) and thusmay be the most abundant group of organisms on earth.

The DGGE profile of the Archaea showed several bands thathybridized with probes for methanogens. Methanogenesis ismore common in freshwater and terrestrial environments thanin the sea (4), which contains high levels of sulfate, becausesulfate-reducing bacteria effectively compete with the meth-anogens for available acetate and H2. In previous studies,methane was detected only in the anoxic zone of Lake Sælen-vannet (15), and the results indicated that methanogens werepresent in the anoxic water. The technique used in this study istherefore an easy way to detect methanogens in natural envi-ronments since these organisms can be slow-growing and dif-ficult to isolate (15, 40).

When the community structure of bacterioplankton overtime was observed, we found shifts within the same sampledepths due to temporal variations. We were able to detectmajor changes from samplings in 1995 to samplings in 1996,and we could also see seasonal changes in 1996. The chemo-cline was at a higher level in 1996 (2.5 m) than in 1995 (3.0 m).This shift is reflected by the dominating population repre-sented by band 15 (Fig. 4). The nucleotide sequence of thisband had a high similarity to the DNA sequence of C. pha-eovibrioides (98%). This bacterium has a characteristic mor-phology; therefore, direct microscopic observations confirmedthat this population was indeed numerically dominant at thechemocline. The red water at the oxic-anoxic interface alsoindicates that the photosynthetic bacteria are the dominatingpopulation in this community. On top of this red layer, a densepopulation of Euglena sp. was observed (35). Sequencing dataof fragment 2 (Fig. 2B) showed 90% similarity to Euglenagracilis DNA sequence data, and this band was present only inthe oxic zone. The temporal shifts in the anaerobic zone wereless significant.

DNA fractions obtained by this approach were further ana-lyzed with DNA probes to detect populations of sulfate-reduc-ing bacteria and methanogens. The maximum number of sul-fate-reducing bacteria and methanogens occurred in and belowthe chemocline. Some of the populations were present in all ofthe different water layers. Communities from different zonescould have been mixed due to sedimentation of organisms inlarge aggregates of cells, buoyancy of cells with gas vesicles, ortransportation with gas bubbles. Although our results suggestdiversity of metabolic types, these findings address the ecolog-ical question of whether bacterial populations observed in thecommunity based on phylogenetic analysis are those metabol-ically active in these communities.

ACKNOWLEDGMENTS

We thank Cindy Nakatsu and Roald Kåre Nielsen for criticallyreading the manuscript. We thank Mikael Heldal for preparing thesamples for electron microscopy. We are grateful to Mikael Heldal,Gunnar Bratbak, and Tron Frede Thingstad for helpful discussion.

This work was supported by grants from the Research Council ofNorway to Lise Øvreås.

REFERENCES

1. Altschul, S. F., W. Gish, W. Miller, E. W. Myers, and D. J. Lipman. 1990.Basic local alignment search tool. J. Mol. Biol. 215:403–410.

2. Amann, R. I., B. J. Binder, R. J. Olson, S. W. Chisholm, R. Devereux, andD. A. Stahl. 1990. Combination of 16S rRNA-targeted oligonucleotideprobes with flow cytometry for analyzing mixed microbial populations. Appl.Environ. Microbiol. 56:1919–1925.

3. Amann, R. I., J. Stromley, R. Devereux, R. Key, and D. A. Stahl. 1992.Molecular and microscopic identification of sulfate-reducing bacteria onmultispecies biofilms. Appl. Environ. Microbiol. 58:614–623.

4. Brock, T. D., and M. T. Madigan. 1988. Biology of microorganisms, 5th ed.Prentice Hall International Inc., London, United Kingdom.

5. DeLong, E. F. 1992. Archaea in costal marine environments. Proc. Natl.Acad. Sci. 89:5685–5689.

6. DeLong, E. F., G. S. Wickham, and N. R. Pace. 1989. Phylogenetic stains:ribosomal RNA-based probes for the identification of single cells. Science243:1360–1362.

7. DeLong, E. F., K. Y. Wu, B. B. Prezelin, and R. V. M. Jovine. 1994. Highabundance of Archaea in Antarctic marine picoplankton. Nature 371:695–697.

8. Ferris, M., G. Muyzer, and D. Ward. 1996. Denaturing gradient gel electro-phoresis profiles of 16S rRNA-defined populations inhabiting a hot springmicrobial mat community. Appl. Environ. Microbiol. 62:340–346.

9. Fuhrman, J. A., K. McCallum, and A. Davis. 1993. Phylogenetic diversity ofsubsurface marine microbial communities from the Atlantic and PacificOceans. Appl. Environ. Microbiol. 59:1294–1302.

10. Fuhrman, J. A., K. McCallum, and A. A. Davis. 1992. Novel major archae-bacterial group from marine plankton. Nature 345:60–63.

11. Giovannoni, S. J., T. B. Britschgi, C. L. Moyer, and K. G. Field. 1990.Genetic diversity in Sargasso Sea bacterioplankton. Nature 345:60–62.

12. Grasshoff, K., M. Ehrhardt, and K. Kremling. 1983. Methods of seawateranalysis, 2nd ed. Verlag Chemie, Weinheim, Germany.

13. Hofle, M. G., and I. Brettar. 1995. Taxonomic diversity and the metabolicactivity of microbial communities in the watercolumn. Limnol. Oceanogr.40:868–874.

14. Indrebø, G., B. Pengerud, and I. Dundas. 1979. Microbial activities in apermanently stratified estuary. I. Primary production and sulphate reduction.Mar. Biol. 51:295–304.

15. Indrebø, G., B. Pengerud, and I. Dundas. 1979. Microbial activities in apermanently stratified estuary. II. Microbial activities at the oxic-anoxicinterface. Mar. Biol. 51:305–309.

16. Lane, D. J. 1991. 16S/23S rRNA sequencing, p. 115–175. In E. Stackebrandtand M. Goodfellow (ed.), Nucleic acid techniques in bacterial systematics.John Wiley & Sons Ltd., West Sussex, United Kingdom.

17. Liesack, W., H. Weyland, and E. Stackebrandt. 1991. Potential risks of geneamplification by PCR as determined by 16S rRNA analysis of a mixed cultureof strict barophilic bacteria. Microb. Ecol. 21:191–198.

18. Ludwig, W., S. Dorin, N. Springer, G. Kirchhof, and K.-H. Schleifer. 1994.PCR-based preparation of 23S rRNA-targeted group-specific polynucleotideprobes. Appl. Environ. Microbiol. 60:3236–3244.

19. Murray, A. E., J. T. Hollibaugh, and C. Orrego. 1996. Phylogenetic compo-sitions of bacterioplankton from two California estuaries compared by de-naturing gradient gel electrophoresis of 16S rDNA fragments. Appl. Envi-ron. Microbiol. 62:2676–2680.

20. Muyzer, G., E. C. Dewaal, and A. G. Uitterlinden. 1993. Profiling of complexmicrobial populations by denaturing gradient gel electrophoresis analysis ofpolymerase chain reaction-amplified genes coding for 16s rRNA. Appl. En-viron. Microbiol. 59:695–700.

21. Muyzer, G., and N. B. Ramsing. 1995. Molecular methods to study theorganization of microbial communities. Water Sci. Technol. 32:1–9.

22. Myers, R. M., T. Maniatis, and L. S. Lerman. 1987. Detection and localiza-tion of single base changes by denaturant gradient gel electrophoresis. Meth-ods Enzymol. 155:501–527.

23. Porter, K. C., and Y. S. Feig. 1980. The use of DAPI for identifying andcounting aquatic microflora. Limnol. Oceanogr. 25:943–948.

24. Ramsing, N. B., H. Fossing, T. G. Ferdelmann, F. Andersen, and B. Tham-drup. 1996. Distribution of bacterial populations in a stratified fjord (Mar-iager Fjord, Denmark) quantified by in situ hybridization and related tochemical gradients in the water column. Appl. Environ. Microbiol. 62:1391–1404.

25. Raskin, L., R. I. Amann, L. K. Poulsen, B. E. Rittmann, and D. A. Stahl.1995. Use of ribosomal RNA based molecular probes for characterization of

3372 ØVREÅS ET AL. APPL. ENVIRON. MICROBIOL.

on June 16, 2015 by guesthttp://aem

.asm.org/

Dow

nloaded from

complex microbial communities in anaerobic biofilms. Water Sci. Technol.31:261–272.

26. Rehnstam, A.-S., S. Backman, D. C. Smith, F. Azam, and Å. Hagstrom. 1993.Blooms of sequence specific culturable bacteria in the sea. FEMS Microbiol.Ecol. 102:161–166.

27. Reysenbach, A.-L., L. J. Giver, G. S. Wickham, and N. R. Pace. 1992.Differential amplification of rRNA genes by polymerase chain reaction.Appl. Environ. Microbiol. 58:3417–3418.

28. Rolleke, S., G. Muyzer, C. Wawer, G. Wanner, and W. Lubitz. 1996. Iden-tification of bacteria in a biodegraded wall painting by denaturing gradientgel electrophoresis of PCR-amplified gene fragments coding for 16S rRNA.Appl. Environ. Microbiol. 62:2059–2065.

29. Schlegel, H. G. 1986. General microbiology. Cambridge University Press,Cambridge, United Kingdom.

30. Stahl, D. A., and R. Amann. 1991. Development and application of nucleicacid probes, p. 205–248. In E. Stackebrandt and M. Goodfellow (ed.), Nu-cleic acid techniques in bacterial systematics. John Wiley & Sons Ltd., WestSussex, United Kingdom.

31. Suzuki, T. M., and S. J. Giovannoni. 1996. Bias caused by template annealingin the amplification of mixtures of 16S rRNA genes by PCR. Appl. Environ.Microbiol. 62:625–630.

32. Teske, A., C. Wawer, G. Muyzer, and N. B. Ramsing. 1996. Distribution ofsulfate-reducing bacteria in a stratified fjord (Mariager Fjord, Denmark) asevaluated by most-probable-number counts and denaturing gradient gelelectrophoresis of PCR-amplified ribosomal DNA fragments. Appl. Environ.

Microbiol. 62:1405–1415.33. Torsvik, T., V. Torsvik, J. Keswani, and W. B. Whitman. 1993. Oligonucle-

otide probes to 16S rRNA of methanogenic bacteria, abstr. N-19, p. 300. InAbstracts of the 93rd General Meeting of the American Society for Micro-biology 1993. American Society for Microbiology, Washington, D.C.

34. Torsvik, T., V. Torsvik, J. Keswani, and W. B. Whitman. Unpublished re-sults.

35. Tuomi, P., T. Torsvik, M. Heldal, and G. Bratbak. 1997. Bacterial populationdynamics in a meromictic lake. Appl. Environ. Microbiol. 63:2181–2188.

36. Ward, D. M., R. Weller, and M. M. Bateson. 1990. 16S rRNA sequencesreveal numerous uncultured microorganisms in a natural community. Nature345:63–65.

37. Wawer, C., H. Ruggeberg, G. Meyer, and G. Muyzer. 1995. A simple andrapid electrophoresis method to detect sequence variation in PCR-amplifiedDNA fragments. Nucleic Acids Res. 23:4928–4929.

38. Weidner, S., W. Arnold, and A. Puhler. 1996. Diversity of uncultured micro-organisms associated with the seagrass Halophila stipulacea estimated byrestriction fragment length polymorphism analysis of PCR-amplified 16SrRNA genes. Appl. Environ. Microbiol. 62:766–771.

39. Wetzel, R. G. 1983. Limnology. Saunders College Publishing, Philadelphia,Pa.

40. Whitman, W. B., T. L. Bowen, and D. R. Boone. 1992. The methanogenicbacteria, p. 719–767. In A. Balows, H. G. Truper, M. Dworkin, W. Harder,and K. H. Schleifer (ed.), The procaryotes. Springer-Verlag, Berlin, Ger-many.

VOL. 63, 1997 DISTRIBUTION OF BACTERIOPLANKTON IN A MEROMICTIC LAKE 3373

on June 16, 2015 by guesthttp://aem

.asm.org/

Dow

nloaded from