Diversity and Seasonal Variability of β-Proteobacteria in Biofilms of Polluted Rivers: Analysis by Temperature Gradient Gel Electrophoresis and Cloning

Abstract

The β-subgroup of the Proteobacteria has been shown to be important in aquatic habitats and was investigated in depth here by molecular 16S rRNA techniques in biofilms of the Elbe River and its polluted tributary, the Spittelwasser River. The bacterial 16S rRNA genes were cloned from each site, screened for β-proteobacterial clones and sequenced. River biofilm clones from both rivers grouped into 9 clusters (RBFs). RBFs 1, 2, and 3 fell into the recently described βI cluster of cosmopolitan freshwater bacteria, where they represented new species related to Rhodoferax, Aquaspirillum, and Hydrogenophaga. RBFs 4 to 7 affiliated with Aquabacterium commune, Ideonella dechloratans, and Sphaerotilus natans, respectively. The two remaining RBFs were uncultivated clusters, one of them being distantly related to Gallionella ferruginea. Seasonal changes in the relative intensity of the β-proteobacterial 16S rRNA genes of biofilms harvested monthly for 1 year were determined by specific amplification and separation by temperature gradient gel electrophoresis (TGGE). Bands were identified by comparison of clones to community fingerprints by TGGE. Eight of 13 identified bands were shared by both habitats but showed different relative abundance and seasonal variability in the two rivers, probably caused by differences in temperature and pollutants. The data indicate new not-yet-cultivated clusters of river biofilm organisms, some of them probably distributed globally. They confirm the importance of certain known freshwater genera in river biofilms. The high phylogenetic resolution obtained by clone library analysis combined with the high temporal resolution obtained by TGGE suggest that the observed microdiversity in the river biofilm clone libraries might be caused by phylogenetically closely related microbial populations which are adapted to ecological parameters.

The number of investigations concerning the microbial ecology of lotic ecosystems (creeks, rivers, and streams) is increasing, but compared to lakes or marine habitats, it is still rather small (22). This is partly due to their changeability, which is ruled by the water flow. The fact that stream bacteria originate both from autochthonous sources (biofilms, water column, and benthos) and allochthonous sources (soil, groundwater, and wastewater) also complicates efforts to investigate microbial interactions in rivers (17).

In streams, bacteria live planktonically or integrated into biofilms which are either suspended freely as so-called suspended particulate matter (SPM) or growing on solid surfaces. The importance of biofilms on solid surfaces for the total metabolic activity and carbon cycling in rivers is largest in small rivers and decreases with increasing river size due to the surface-to-volume ratio (24). Leff et al. (23) described biofilms on solids as important stations for planktonic bacteria on their way down the river. Moreover, river biofilms also exchange SPM with the water phase and therefore represent a combination of planktonic and sessile communities over the whole growth period of the biofilm, which in our case was 1 month (6). By analyzing such biofilms, the observation of medium-term instead of momentary states of the river was feasible.

In a preceding study with quantitative FISH (fluorescence in situ hybridization) analysis, it was shown that β-proteobacteria comprised on average 31% of all cells in microbial biofilms of the extremely polluted Spittelwasser River and 18% in the less polluted Elbe River (6). The globally high abundance of this group has also been confirmed in similar studies of other rivers (2, 3, 18, 26). β-Proteobacteria have been identified in various freshwater and terrestrial habitats via cloning of the amplified 16S rRNA genes from the total microbial community DNA (for examples, see references 4 and 10), whereas they are rarely found in marine environments (reviewed in reference 31). Comparative sequence analysis of environmental clones revealed that some of the β-proteobacteria belonged to globally distributed freshwater clusters (14, 45, 46).

Therefore, the aim of this study was to investigate the phylogenetic composition of the β-proteobacteria community in river biofilms in depth. In particular, we compared the extremely polluted Spittelwasser River with the less polluted Elbe River to determine the effect of industrial pollution. Moreover, biofilm samples were analyzed over the period of 1 year to observe seasonal changes in the community structure which might be related to changes in chemical and physical parameters. We chose a multiple approach, combining community profiling by temperature gradient gel electrophoresis (TGGE) (29) with analyses of clone libraries of total biofilm community DNA. First, a β-proteobacterium-specific TGGE method for specific community profiling was established, which allowed us to monitor Elbe and Spittelwasser β-proteobacterial 16S rRNA genes throughout the year. Subsequently, one clone library was made for each river by using bacterial primers, and the library was screened for β-proteobacterial clones via specific PCR. β-Proteobacterial clones were sequenced, and their migration behaviors on TGGE were compared to bands in the β-proteobacterial community fingerprints. In such a way, it was possible to identify corresponding clones for predominant bands in the community fingerprints and thus obtain high-resolution phylogenetic information on those bands.

MATERIALS AND METHODS

Sampling site and sampling procedure.

Sampling sites were the Elbe River, upstream of the city of Magdeburg (11° 40′ E, 54° 4′ N; Germany), and the Spittelwasser River, a highly polluted tributary of the Elbe (12° 17′ E, 52° 4′ N; Germany). Massive industrial pollution occurred from about 1900 until 1990. In the Spittelwasser, sediments are still highly contaminated, and this river also receives polluted groundwater. Water quality in the water phase and the SPM in the Elbe is regularly monitored by several water authorities assembled in the Arbeitsgemeinschaft zur Reinerhaltung der Elbe (www.arge-elbe.de). Biofilms were grown in situ on glass slides and harvested about every month from spring 1997 until spring 1998 (6). Sampling dates, corresponding river parameters, and concentrations of selected pollutants were noted (see Table 3). Biofilm samples were aliquoted and directly frozen at the sampling site as described.

TABLE 3.

Sampling dates and coinciding values for water flow, conductivity, temperature, and selected pollutants^e of the Elbe and Spittelwasser Rivers

Sample no.	Sample date (mo/day/yr)	Flow rate (m3 s−1) of Elbe Rivera	Conductivitya (μS cm−1) in river:		Temperaturea (°C) in river:		p,p′- DDTb (μg kg−1)	HCBb (μg kg−1)	CODd (mg l−1 O2)	Znc (mg kg−1)	Cuc (mg kg−1)	Crc (mg kg−1)
			Elbe	Spittelwasser	Elbe	Spittelwasser
	04/08/97	975	855	1,647	5.8	9.2			31	1,300	87	150
1	05/13/97	591	900	1,650	15.0	18.0	180	210	21	1,300	120	140
2	06/16/97	279	1,480	1,610	18.6	20.1	190	220	27	1,400	110	120
3	07/18/97	481	1,020	1,770	21.0	21	31	140	28	960	89	84
4	08/21/97	301	1,400	1,640	23.6	20.2	12	30	21	1,400	110	140
5	09/25/97	246	1,890	1,560	15.5	15.8	140	150	18	1,500	140	160
6	11/03/97	259	1,800	1,890	6.1	9.1	290	180	15	1,800	170	190
7	12/11/97	305	1,423	1,886	6.1	9.3	180	360	17	1,600	200	250
8	01/16/98	512	1,073	1,842	6.3	9.0	130	300	18	1,900	210	330
9	02/19/98	448	1,288	1,897	6.8	10.8	160	350	17	1,400	130	160
10	03/30/98	637	924	1,446	8.2	10.2	97	140	25	940	87	110
11	05/04/98	333	1,489	1,573	15.0		43	37

^aData from sampling day.

^bData from SPM collected over 1 month.

^cData from settled SPM of single water samples nearest to sampling dates (sampling interval, every 2 weeks).

^dCOD, chemical oxygen demand.

^eValues for pollutants are for the Elbe River only.

DNA extraction from river biofilms.

The DNA was extracted from river biofilms by a modified direct lysis procedure including mechanical disruption and removal of humic substances by cetyltrimethylammonium bromide precipitation (11, 35) and further purification by phenol-chloroform extraction. Two milliliters of each biofilm sample was subdivided into 0.5-ml aliquots and treated as described in reference 11, only differing in the use of 200 mg of lysozyme per tube. The purified DNA extracts of each sample were pooled to give a final volume of 100 μl. Diluted DNA extracts (1:100) were used in PCR.

Primers for TGGE PCR.

For the partial amplification of bacterial 16S rRNA genes, the primers F-968-GC and R-1401 were used (32). For amplification of the 16S rRNA genes of β-proteobacterial positions 8 to 381 (Escherichia coli numbering) was chosen. As a forward primer, we used primer 27f-GC (5′-CGCCCGGGGCGCGCCCCGGGCGGGGCGGGGGCACGGGGGGAGAGTTTGATCCTGGCTCAG-3′) without degenerate position (20, 32). The β-proteobacterium-specific primer bP365r (5′-GCGCCCATTGTCCAAA-3′) served as the reverse primer. Its main feature was a signature nucleotide at the 3′ end. While β-proteobacteria have a complementary U at this position, most others have an A or a C mismatch. A few β-proteobacteria had mismatches at the 5′ end of the primer with no effect on amplification efficiency (data not shown). Only a few β-proteobacteria with a mismatch at the 3′ end (e.g., Thiomonas perometabolis, Iodobacter fluviatile, some wastewater clones, and occasionally members of the genera Delftia, Comamonas, and Acidovorax) may have been excluded by this primer.

PCR for TGGE analysis.

The PCR mix for partial amplification of the 16S rRNA genes of most bacteria (50 μl) contained 0.1 to 1 ng of template DNA, 3 mM MgCl₂, 5% (vol/vol) dimethyl sulfoxide (DMSO), 0.1 mM concentrations of each deoxynucleoside triphosphate, 0.2 μM concentrations of each primer, 1 U of AmpliTaq Stoffelfragment (Perkin Elmer, Branchburg, N.J.), and Stoffel buffer, with final concentrations of 10 mM Tris-HCl (pH 8.3) and 10 mM KCl. Amplification conditions were initial denaturation at 94°C for 1 min, 30 cycles consisting of denaturation at 94°C for 10 s, annealing at 54°C for 20 s, and extension at 68°C for 40 s, and final extension at 68°C for 2 min. PCR amplification was performed with a Mastercycler gradient (Eppendorf, Hamburg, Germany).

High-stringency amplification of β-proteobacterial 16S rRNA gene fragments was achieved by optimizing the PCR conditions by using reference strains with one or more mismatches. An annealing temperature of 61°C and a formamide concentration of 5% gave the best results (data not shown). Amplified 16S rRNA genes from the biofilm DNA and reference strains were used as templates for community fingerprints because the PCR product yield of most DNA extracts from Elbe biofilm samples was too low for TGGE analysis. The comparison of TGGE fingerprints obtained from direct and indirect amplification of all Spittelwasser samples and a single Elbe sample showed negligible differences (data not shown). The corresponding fragment (E. coli positions 8 to 381) from β-proteobacterial clones was amplified from the diluted PCR product obtained with vector-specific primers T7 and SP6 (see below) in a total volume of 10 μl.

PCR of environmental 16S rRNA genes.

16S rRNA genes in environmental samples and reference strains were amplified with bacterial primer 27f and the universal primer 1492r (20). The PCR mixtures (50 μl) contained 0.1 to 1 ng of environmental DNA, 3 mM MgCl₂, 0.125 mM (each) dATP, dCTP, dGTP, and dTTP, 0.1 μM concentrations of each primer, 5% DMSO, 0.5 U of recombinant Taq DNA polymerase (LifeTechnologies, Paisley, United Kingdom), 20 mM Tris-HCl (pH 8.4), and 50 mM KCl. PCR conditions were the same as described for bacterial TGGE PCR, except the extension time was increased to 2 min.

TGGE.

For sequence-specific separation of PCR products, a TGGE system (Qiagen, Hilden, Germany) was used. Gels contained 6% (wt/vol) acrylamide, 0.1% (wt/vol) bisacrylamide, 8 M urea, 20% (vol/vol) formamide, and 2.2% (vol/vol) glycerol in 1× morpholinepropanesulfonic acid (MOPS) electrophoresis buffer (20 mM MOPS, 1 mM EDTA [pH = 7.5]).

Aliquots of 4 μl of amplified fragments of environmental 16S rRNA genes with the same amount of DNA (checked on agarose gels) or 0.7 μl of DNA extracted from β-proteobacterial clones (volume adjusted with sterile Milli Q water) and 1 μl of loading buffer (0.08% bromophenol blue, 0.08% xylencyanol, 10 mM EDTA, 8 M urea, 6× MOPS) were applied on the horizontal gel. The river-specific reference pattern (ΣElbe and ΣSpittelwasser; see below) was applied to every third or fourth lane.

Runs were performed at 350 V in 1× MOPS electrophoresis buffer. For separating the amplified fragments of the bacterial 16S rRNA genes (E. coli positions 968 to 1401), a temperature gradient of 33 to 43°C was applied for 6 h. Amplified β-proteobacterial 16S rRNA gene fragments (E. coli positions 8 to 381) were separated with a temperature gradient of 35 to 44°C applied for 6 h 45 min. After the run, the gels were silver stained (34) and dried over night.

Cloning of 16S rRNA genes.

For improved blue-white screening of transformands, the forward primer t-BAC27f (5′-ttttgactgactGAGTTTGATCCTTGGCTCAG-3′) and reverse primer t-Uni1493r (5′-ttttgactgactgACGGCTACCTTGTTACGAC-3′) were used, which contained a nonmatching tail (small letters) with stop codons (bold letters) and 3 additional thymidine residues to promote ligation (4, 12). The PCR mix for amplifying 16S rRNA genes was used, with 0.5 μl of template and 0.5 U of recombinant Taq DNA polymerase (LifeTechnologies) in a total volume of 20 μl. PCR was performed as follows: predenaturation at 94°C for 1 min followed by 35 cycles of 94°C for 10 s, 50°C for 20 s, and 68°C for 120 s and final extension at 68°C for 6 min. As a template, 0.1 to 1 ng of biofilm DNA from each sampling was used.

The template for ligation was prepared by pooling equal amounts of the PCR products from each sampling date for each river followed by purification via gel extraction (Qiagen). Both PCR mixes were ligated into the linear plasmid vector pGEM-T (Promega, Madison, Wis.), and plasmids were transformed in competent E. coli JM109 cells by following the manufacturer's instructions. The pooled template was also used to create the river-specific references called ΣElbe and ΣSpittelwasser for TGGE analysis. Two hundred eighty-seven transformands were picked for the Elbe River biofilm clone library, and 300 transformands were picked for the Spittelwasser River biofilm clone library. The lysate of the clones (picked clone solved in 50 μl of sterile water and incubated at 95°C for 15 min) served as the template for the vector-specific PCR.

To confirm positive clones, check the insert size, and produce the template for subsequent analyses of the clones, a PCR with the pGEM-T-specific primers T7 and Sp6 targeting bacteriophage-derived promoters (T7-Sp6 PCR) (44) was carried out in a total volume of 10 μl under the same conditions as described for the amplification of 16S rRNA genes, with the exception that no DMSO was added.

Differential signature PCR and further screening of clones.

Each clone library was screened for β-proteobacteria with the multiplex diagnostic differential signature PCR (39). The PCR mix was the same as for cloning except for the primer mix and 1 μl of template (1:5 diluted T7-Sp6 PCR product). The touchdown PCR protocol was performed as described. A clone was tentatively classified as belonging to the β-proteobacteria if a fragment of 700 to 750 bp was amplified.

Since primer βɛ774f binds not exclusively with β-proteobacteria but also with some others, positive clones were subjected to a second screening via the β-proteobacterium-specific TGGE PCR with primers 27f-GC and βP365r. Here, positive clones had a strong band with a fragment size of about 370 bp.

16S rRNA gene sequencing.

From selected clones, a 50-μl T7-Sp6 PCR was performed and PCR products were purified with a PCR purification kit (Qiagen) and used as templates for the sequencing reaction with the BIG dye terminator cycle sequencing kit (Applied Biosystems, Inc., Foster City, Calif.). Samples were sequenced either by a company (GATC Biotech AG, Konstanz, Germany) or after precipitation, drying, and resolving in loading dye on an ABI PRISM 373 or 377 DNA sequencer (Applied Biosystems) according to the manufacturer's instructions. The following primers were used: 27f, 518r, and 1492r (20). Sequencing resulted in 25 nearly full-length sequences and 28 sequences of >480 bp.

Sequence analysis.

For a first identification and to find the closest relatives of the clones, the unaligned sequences were submitted to the sequence match programs Blast (National Center for Biotechnology Information, http://www.ncbi.nlm.nih.gov/blast) and FASTA (33). Chimera formation was checked with the Chimera check program (version 2.7; Ribosomal Database Project, www.rdp.cme.msu.edu/cgis/chimera.cgi). β-Proteobacterial sequences from this study as well as recently published sequences were imported into the database (release June 2002) of the ARB program package (release 1999; www.arb-home.de) and added up to a total of 2,056 sequences for this group. Sequences were automatically aligned and subsequently corrected manually. Partial phylogenetic trees were calculated from 6 (Fig. 1A) and 16 (Fig. 1B) nearly complete small-subunit sequences, based on distance matrix, maximum-parsimony, neighbor-joining, and maximum-likelihood analysis by using the intrinsic majority rule filter for β-proteobacteria. Subsequently, biofilm clone sequences (>480 bp) and related published sequences were added to trees with the parsimony tool without changing the overall tree topology. Finally, a consensus tree was constructed, where uncertain branching orders were replaced by a multifurcation.

FIG. 1.

Rooted phylogenetic consensus trees of 16S rRNA sequences showing affiliation of Elbe (Elb) and Spittelwasser (Spb) biofilm clones within the β-proteobacteria. The scale bar corresponds to a 10% difference in nucleotide sequence. E. coli was used as an outgroup (data not shown). The number of identical sequences (above 99% similarity) not shown in the tree is given in parenthesis (+1, etc.). Corresponding bands with the community pattern and accession numbers are indicated. Clamps mark river biofilm clusters (RBFs 1 to 9) containing at least two different biofilm clones, together with their most closely related environmental clones and cultured organisms (above 96% similarity). All sequences are available from the EMBL database. (A) Partial tree of the Comamonadaceae showing RBFs 1 to 3 within the cosmopolitan freshwater cluster βI (14). (B) Partial tree showing RBFs 4 to 7 in the Leptothrix subcluster of the β-proteobacteria as well as the affiliation of RBFs 8 and 9 and freshwater clusters βII to βIV (14).

Nucleotide sequence accession numbers.

The sequences determined in this study have been deposited in the EMBL database under the accession numbers AJ421912 to AJ421941 and AJ422152 to AJ422177.

RESULTS

Clone libraries and screening for β-proteobacterial clones.

As templates for cloning amplified 16S rRNA gene fragments from the various sampling times were pooled for each river because TGGE patterns from PCR products obtained with bacterial TGGE primers (E. coli positions 968 to 1401) did not show strong seasonal differences (data not shown). From this template, which was also used to generate the river-specific reference for TGGE analysis (Fig. 2), two clone libraries were constructed with bacterial primers. Affiliation of clones to the β-proteobacteria was subsequently determined. About 30% (96) of the Elbe biofilm clones and 25% (68) of the Spittelwasser biofilm clones were assigned to the β-proteobacteria in a prescreening with a multiplex signature PCR (39). Finally, about half of these were identified as β-proteobacteria after applying the β-proteobacterium-specific PCR for TGGE analysis, which was confirmed later on by sequencing. As a control, clones which were excluded by the β-proteobacterium-specific PCR were also sequenced. Except for one sequence with missing bases in the beginning (5′ end), no additional β-proteobacterial sequences were found. Thus, the primer set 27F-GC-bP365r was highly specific for β-proteobacteria. From 84 identified β-proteobacterium clones, 67 clones were partially sequenced (some of the TGGE duplicates and clones without a match in the community fingerprint were omitted from sequencing). After checking for chimera formation, 28 clones of the Elbe and 25 clones of the Spittelwasser (> 480 bp) were further analyzed.

FIG. 2.

Community profiles of β-proteobacteria based on specific amplification and TGGE separation of 16S rRNA gene sequences (positions 8 to 381) from 1-month-old biofilms grown in the Elbe and Spittelwasser Rivers. Numbers indicate bands for which corresponding clones were found, and letters mark bands for which no corresponding clones were found. EΣ and SΣ are abbreviations for river-specific references (ΣElbe and ΣSpittelwasser) generated from pooled PCR products from all sampling dates. The month of sampling is indicated above the lanes (Jun, June; Jul, July; Aug, August; Sep, September; Nov, November; Dec, December; Jan, January; Feb, February; Mar, March). Hyphens between river-specific references mark bands which were present in both rivers.

β-Proteobacteria community composition.

Generally, river biofilm clones were more closely related to cloned environmental sequences than to cultivated organisms. The nearest published relatives were from aquatic habitats like rivers, lakes, aquifers, and activated sludge, but some were also derived from soil. Several clones were found which affiliated closely (> 98% similarity) with cultivated species, i.e., Rhodoferax antarcticus, Hydrogenophaga taeniospiralis, Hydrogenophaga palleronii, Sphaerotilus natans, Dechloromonas sp., and methyl tert-butyl ether-degrading strain PM1 (7).

The phylogenetic affiliation of the river biofilm clones is shown in Fig. 1. Clamps mark river biofilm clusters (RBFs 1 to 9) containing two or more not identical river biofilm clones and related sequences of environmental clones and isolates. Clamps did not always group sequences of the same similarity level in order to show the phylogenetic affiliation and a number of closest relatives in more detail.

Many of the analyzed clones (42%) fell into the recently suggested cosmopolitan freshwater cluster βI (14) (Fig. 1A). Clones affiliating with the other freshwater clusters βII to βIV were not found (Fig. 1B). Within the βI cluster, three subclusters (RBFs 1 to 3) could be identified. Most biofilm clones of RBFs 1 and 3 were highly related (>99% similarity) to described species or environmental clones. On the other hand, RBF 2, consisting exclusively of very similar sequences (>98.5%) of the sampled habitats, stood a little to the side, affiliating more or less equidistantly with R. antarcticus and Aquaspirillum delicatum (about 97%). This cluster, which also contained the largest number of biofilm clones (18 clones), was clearly dominated by sequences from the Elbe, only two sequences were contributed by the Spittelwasser.

Approximately the same number of clones (45%) fell in the Leptothrix subgroup of the Comamonadaceae (13) in equal shares from both habitats and grouped into 4 clusters (RBFs 4 to 7) (Fig. 1B). RBF 4 contained the second largest number of biofilm clones (14 sequences), which were mainly retrieved from the Spittelwasser. Clones in RBF 4 representing bands 7 and 11 had almost the same sequence similarity to Aquabacterium commune and the methyl tert-butyl ether-degrading bacterium PM1. But all treeing methods and the remaining sequences indicated that clones of this cluster affiliated with PM1, isolated from compost biofilters, which seems to represent a separate lineage within the Leptothrix subgroup (7). Searching for the nearest published relatives showed that these were mostly clones retrieved from soil habitats.

The phylogenetic positions of RBFs 5 to 7 were not unambiguous, as indicated by multifurcations. While similarities of biofilm clones in RBFs 4 to 6 to cultivated species ranged between 96 and 98%, biofilm clones of cluster RBF 7, mainly from the Elbe, showed similarities of 98 to 99% to S. natans. These clones showed fuzzy bands on TGGE and could therefore not be detected in the community fingerprints. Nearest cloned environmental sequences and cultivated species in RBF 5 to 7 clusters were retrieved mainly from freshwater and activated sludge.

Most of the remaining river biofilm clones could be grouped into 2 monophyletic clusters (RBF 8 and 9) which consisted exclusively of cloned environmental sequences. While RBF 8 was an uncultivated cluster with clones of different origins and not near to any species, biofilm clones of RBF 9 were solely aquatic clones and affiliated distantly (91 to 94%) with Gallionella ferruginea and Ferribacterium limneticum.

With the exception of RBF 6, all river biofilm clusters contained sequences retrieved from both rivers or representing TGGE bands which were found in both rivers.

Clones which were only found once and thus were not included in the clusters were Spb283 (97.7% similarity to strain A0640), Elb271 (97.0% similarity to Rubrivivax gelatinosus), Spb272 (98.3% similarity to Dechloromonas sp. SIUL), Spb253 (99.6% similarity to Westerscheldt estuary clone Ws12).

TGGE profiles.

Figure 2 shows the TGGE patterns of the β-proteobacterial population of river biofilms monitored for 1 year. Strikingly, the community fingerprints of each river were more similar throughout the year than samples of the two sites taken at the same time. Characteristic for the Elbe biofilms was the broader pattern of the fingerprints compared to those of the Spittelwasser, which showed generally a more compact pattern consisting of strong bands.

Identification of TGGE bands.

Amplicons of the 16S rRNA genes of the β-proteobacterial clones were compared with bands in the river reference pattern by TGGE. Most clones migrated with a band of the reference pattern of the Elbe (60% comigration of clones) or the Spittelwasser (71%). For the remaining clones with either sharp bands (Elbe, 20%; Spittelwasser, 18%) or more or less fuzzy bands (Elbe, 20%; Spittelwasser, 11%), no corresponding band in the river community fingerprints could be found. Altogether, 13 bands in the community patterns were identified (Table 1). Bands could be unambiguously assigned to a single phylotype, since comigrating clones had nearly identical or very similar sequences. One exception was Spittelwasser biofilm clones migrating with band 7, which affiliated with two different taxa falling into different branches of the β-proteobacteria (Table 1; Fig. 1). In some cases, clones with different TGGE bands had high similarity values, e.g., Elbe clones corresponding to bands 2 to 4 and Spittelwasser clones corresponding to bands 12 and 13 had similarities to each other above 99%. Small sequence heterogeneity in the 16S rRNA genes can result in large differences in migration behavior in TGGE. Thus, the data demonstrate that there is no simple correlation between the number of TGGE bands on a gel and the species diversity of a community.

TABLE 1.

Identification of Elbe and Spittelwasser biofilm clones comigrating with bands in the TGGE community fingerprints and their affiliation with river biofilm clusters

TGGE banda	Clone(s)b	Sequence length (bp)	RBF	Nearest published relative (accession no.)	% Similarityc	Nearest described relative (accession no.)	% Similarityc
1	Elb168, −203, (−217), Spb06	1,440	9	Clone SG2-128 (AY135925)	97.3	G. ferruginea (L07897)	94.0
2	Elb263, −24, −213, (−29), (−63), (−64), (−74), (−277)	1,411	2	Clone CRE-FL50 (AF141469)	96.5	A. delicatum (AF078756)	96.9
3	Elb154, −100, −180, (−7, −30)	1,407	2	Clone CRE-FL50 (AF141469)	96.5	A. delicatum (AF078756)	96.7
4	Elb117	536	2	Clone CRE-FL50 (AF141469)	97.1	A. delicatum (AF078756)	95.5
5	Elb270, −216	1,413	1	Strain Wuba 139 (AF336363)	99.1	R. antarcticus (AF084947)	97.7
6	Eb252, −197	1,406	6	Clone Sta4-36 (AJ416246)	98.2	I. dechloratans (X72724)	97.0
7	Elb111, −07, −49, −69, (−143), Spb283	1,420	4	Clone WR839 (AJ292856)	97.3	A. commune (AF035054)	95.9
		1,400	—d	Strain A0640 (AF236010)	97.7	I. dechloratans (X72724)	97.1
8	Spb54	1,437	3	H. taeniospiralis (AF078768)	99.1
9	Spb280, −87, −293	1,435	5	Clone T10 (Z93979)	97.5	A. commune (AF035054)	97.0
10	Elb205, −236	1,408	8	Clone A21b (AF234707)	96.3	Uncultivated cluster
11	Spb132, −153, −298, Elb192, (−275)	1,434	4	Clone WR839 (AJ292856)	97.6	A. commune (AF035054)	96.5
12	Spb223	1,432	3	H. palleronii (AF019073)	99.4
13	Spb256, −266	1,402	3	H. palleronii (AF019073)	99.5

^aBold numbers indicate bands occurring in fingerprints from both rivers.

^bClones in parentheses had the same migration behavior on TGGE but were not sequenced. Elb, Elbe; Spb, Spittelwasser.

^cThe similarity value is for the clone listed first, as determined with the distance matrix ARB tool by using all available nucleotides.

^d—, single clone which is not in a cluster.

Shared and unique bands in β-proteobacterial community fingerprints of Elbe and Spittelwasser biofilms.

Eight bands of the 13 designated bands for each river were shared by both habitats (Fig. 2; Table 1). Corresponding bands between the rivers were bands 1, 2, 5, 8, 9, 10, 11, and 13. Bands which were only present in one river were bands 3, 4, and 6 (Elbe) and bands 7 and 12 (Spittelwasser). However, even if a band was only detected in the community fingerprint of one river, related clones were sometimes found in the clone library of the other river. For example, clones corresponding to Elbe bands 3 and 4 were located in the uncultivated cluster RBF 2, which also contained two closely related sequences from the Spittelwasser River (Fig. 1A). Thus, although the community fingerprint did not show bands for these organisms, they or very closely related organisms must have been present in the community, presumably at a low abundance.

Conversely, band 12 was unique for the Spittelwasser River and was identified as H. palleronii (99.4% similarity). No corresponding clone was found in the Elbe clone library, but band 13, which was shared by both rivers, was also assigned to H. palleronii. Thus, a highly related strain of H. palleronii must have been present in the Elbe. To summarize, the combination of community fingerprints and clone library analysis revealed that the dominant members of the β-proteobacteria appeared to be the same in both rivers or were at least closely related; differences occurred with respect to the relative abundance and microdiversity of each of these phylogenetic clusters throughout the year (see below).

Seasonal variability of β-proteobacterial 16S rRNA genes in Elbe and Spittelwasser biofilms.

In both rivers, strong seasonal variability was observed (Fig. 2). Bands 2 (RBF 2), 5 (RBF 1), 8 (RBF 3), and 10 (RBF 8) were mainly present in the colder season. While bands 2 and 5 were more pronounced in the Elbe, band 8 was stronger in the Spittelwasser, and band 10 was always relatively weak. Seasonal patterns could also be observed for bands for which no clones were found. These occurred especially in the warmer season (e.g., bands b, d, e, and g). Some bands showed no clear seasonal trend but appeared sporadically, e.g., bands 1 (RBF 9), 9 (RBF 5), a, and c. For mixed band 7, seasonal variability was possibly disguised. By contrast, shared bands 11 (RBF 4) and 13 (RBF 3), Elbe band 6 (RBF 6), and Spittelwasser band 12 (RBF3) were strong throughout the year without a clear seasonal trend.

Differences and seasonal changes in river parameters.

While the Elbe River is a stream with a large water volume, high flow rate, and medium level of contamination, the Spittelwasser River is a second-order tributary to the Elbe with a small water volume, low flow rate, and high level of contamination. Table 2 shows characteristic river parameters during the investigated time period. In the water phase, the Elbe had a slightly higher biological oxygen demand, higher concentrations of nitrate and phosphate, and a much higher content of SPM than the Spittelwasser. Ammonia concentrations were 10-fold and AOX (adsorbable organic halogens) values were about 5-fold increased in the Spittelwasser River compared to the Elbe, indicating diverse organic pollutants. Concentrations of metals and xenobiotics in the SPM were much higher in the Spittelwasser River, with extreme concentrations for tributyltin (a highly toxic component of antifouling paints). Table 3 shows selected river parameters and pollutants for the dates of biofilm sampling. Clear seasonal patterns were observed for temperature in both rivers, but the temperature in the Spittelwasser River never dropped below 9°C due to pollution with processed waste water. In wintertime, elevated values for organochlorine pesticides (e.g., p,p′-dichlorodiphenyltrichloroethane [p,p′-DDT] and hexachlorobenzene [HCB]) and heavy metals (Zn, Cu, and Cr) were observed in the SPM in the Elbe; a similar trend can be suspected in the Spittelwasser from occasional measurements, but continuous data are lacking.

TABLE 2.

Average values for a selection of physical and chemical parameters during the investigated time period in the water phase and SPM of the Elbe and Spittelwasser Rivers

Phase	Parameterb (units)	Mean (min, max) value for rivera:
		Elbe	Spittelwasser
Water	pH	8.0 (7.1, 8.7)	7.0 (6.6, 7.5)
	O2 content (mg liter−1)	10.5 (7.5, 13.0)	6.3 (3.5, 10.3)
	TOC (mg liter−1)	8 (5.6, 10)	8.83 (7.16, 10.6)
	BOD7 (mg liter O2−1)	6.5 (2.9, 12.6)	4.1 (2.4, 7.4)
	AOX (μg liter−1)	37 (30, 49)	162 (93, 280)
	NH4−-N (mg liter−1)	0.27 (0.01, 0.95)	2.0 (1.1, 2.9)
	NO3−-N (mg liter−1)	4.6 (3.4, 6.6)	2.6 (1.9, 3.4)
	Total P (mg liter−1)	0.25 (0.14, 0.36)	0.14 (0.1, 0.19)
	SPM (mg liter−1)	25 (10, 68)	6 (4, 9)
SPM	Mercury (mg kg−1)	4.5 (3.2, 7,6)	15.3
	Lead (mg kg−1)	149 (121, 173)	295
	Arsenic (mg kg−1)	33.6 (25.5, 44)	149
	γ-HCH (μg kg−1)	2.5 (0.4, 5.8)	40
	p,p′-DDT (μg kg−1)	140 (12, 290)	1,700
	HCB (μg kg−1)	198 (30, 360)	590
	AOX (mg kg−1)	199 (130, 260)	480
	Tributyltin (μg kg−1)	22.8 (18, 30)	20,000

^aData were collected every 2 and 8 weeks for water-phase results from the Elbe and Spittelwasser Rivers, respectively. SPM data from the Elbe River were collected monthly, and SPM data from the Spittelwasser River are from 1 month (February to March 1998).

^bBOD, biological oxygen demand; TOC, total organ carbon; HCH, hexachlorohexane.

DISCUSSION

Community profiling and clone libraries.

For the 13 dominant bands in the community fingerprints, corresponding β-proteobacterial clones were found in the clone library. The eight bands for which no corresponding clones were found occurred only in some samples and it can therefore be assumed that the template concentration was too low, since the clone library was made from pooled DNA from all sampling times. Thus, the dominant bands in the TGGE profiles represented those rRNA genes which were present at higher concentrations in the community, as found also by Casamayor et al. (9).

Conversely, most clones identified as belonging to the β-proteobacteria in the clone library had a corresponding band in the community profile. This confirms the specificity of the β-proteobacterial PCR amplification for TGGE analysis and allowed us to identify these bands with good phylogenetic resolution. Clones matching with the same band of the fingerprints were, with one exception, identical or very similar among each other. It can therefore be assumed that bands which exhibited the same mobility in fingerprints of one or both sites represented identical or very similar 16S rRNA sequences. The appearance of such a band in all community fingerprints suggested that the respective organism was present throughout the year.

The number of TGGE bands in the community fingerprints was not directly correlated to the number of species in the microbial biofilm but represented the predominant, PCR-amplifiable 16S rRNA genes which formed distinct bands. Several bands were assigned to the same cluster, e.g., bands 2 to 4 (RBF 2), bands 7 and 11 (RBF 4), and bands 8, 12, and 13 (RBF 3). Clusters of related rRNA gene sequences are typically found in clone libraries (for an example, see reference 10). PCR artifacts are suspected to be responsible for some of this microdiversity (37). Another reason might be operon heterogeneity (32). However, there are also data which suggest that this microdiversity reflects clusters of related species which form ecologically distinct populations (1, 8). In the river biofilms analyzed here, bands assigned to the same cluster or even species did not always show identical seasonal variability. For example, bands 12 and 13 were identified as H. palleroni. However, only band 13 could be observed in the Elbe, where it was particularly strong in winter, while band 12 was never detected there. In the Spittelwasser biofilms, these two bands were not always equally strong, but at some times band 12, and at others band 13, dominated. The data suggest that these bands corresponded to distinct populations of highly related microorganisms with different ecological distributions. Conversely, bands 2 to 4 (RBF 2) did not appear independently but appeared to be strictly correlated and might be an of sequence heterogeneity among rRNA operons.

On the other hand, electrophoretic mobilities of only distantly related strains were in one case indistinguishable (band 7). This was also observed recently by other authors (16). In addition, some organisms built a fuzzy band in TGGE analysis (44) and thus were not recognized in the community fingerprints. In spite of these and intrinsic limitations of PCR-based methods (reviewed in reference 43), gradient gel electrophoresis combined with sequence analysis was a relatively fast and comprehensive approach for identifying those members of the bacterial communities which were present in higher concentrations. Since all samples were treated equally, changes in band intensity could be interpreted as changes in the relative abundance of this particular 16S rRNA gene. In such a way, seasonal variation of the community composition could be observed. Casamayor et al. (8) demonstrated by a combination of denaturing gradient gel electrophoresis and FISH in meromictic Lake Vilar, Spain, that shifts in the relative intensity of a band in the community profile correlated with changes in the abundance of the respective microorganism, determined by FISH counts with specific probes.

Cosmopolitan freshwater clusters?

The vast majority of analyzed clones fell into the Comamonadaceae family (13) of the β-proteobacteria. Half of these clones were affiliated within the recently suggested globally distributed freshwater cluster βI (14), once more indicating that these bacteria are indeed predominant in aquatic habitats worldwide. Within the βI cluster, three subclusters (RBFs 1 to 3) emerged in our study which were characteristic for the investigated river biofilms.

River biofilm clones of subcluster RBF 1, related to R. antarcticus, had remarkable similarities with sequences retrieved from aquatic habitats in different regions of the world. RFB 2 contained the largest number of sequences and represented a not-yet-cultivated species which was mainly found in the Elbe River. Cluster RBF 3 had a high similarity with the described species H. taeniospiralis and H. palleronii, respectively. The nearest relatives of biofilm clones of all three clusters were environmental sequences from lotic habitats or estuaries, possibly indicating a preference of these groups for lotic habitats.

No clones were found which belonged to the other described cosmopolitan freshwater clusters (14, 45, 46). Most of the sequences analyzed in these investigations were from lake and river bacterioplankton, whereas the river biofilms investigated here represent a combination of planktonic and sessile communities undergoing changes during growth. For example, sequences falling into freshwater cluster βII, affiliating with Polynucleobacter necessarius, possibly belong to free-living organisms. They were so far only retrieved from protozoa and particle-free water samples, e.g., groundwater or prefiltered water of the Columbia River and its estuary (10, 27).

The remaining biofilm clones formed six clusters, of which RBFs 5 to 7 in the Leptothrix subgroup represent typical cultivated aquatic bacteria. The closest cultivated relative of biofilm clones of RBF 5 was Aqu. commune, which was shown to be ubiquitous in freshwater with specific probes (26). Biofilm clones in RBF 7 affiliated with the versatile S. natans, which is known to form biofilms on solid surfaces and to occur in natural freshwater habitats as well as in manmade habitats like activated sludge (reviewed in reference 28). A broad distribution of these aquatic organisms may be indicated by this and other published culture-independent analyses (36, 46). Environmental clones in the uncultivated cluster RBF 9 were also exclusively from freshwater environments. Remarkable was the high similarity of the Spittelwasser sequence to one clone retrieved from Lake Sapgyo, South Korea (99.6%), demonstrating intercontinental occurrence of this group.

Most river biofilm clusters included cloned sequences from aquifers, groundwater, activated sludge, or soil. Since rivers are open systems, these bacteria may have been transported into the rivers and were able to survive well in the biofilms. The occurrence of soil bacteria in aquatic habitats was also found by Crump et al. (10). Biofilm cluster RBF 4, which contained Elbe and Spittelwasser clones as well as diverse soil clones, might be a particularly good example for that phenomenon.

Effect of pollution on the β-proteobacterial community.

River biofilm clusters were often comprised of clones from both sites, and 8 of 13 TGGE bands were shared by both rivers. Thus, in the extremely polluted Spittelwasser River and the less polluted Elbe River, the biofilm communities differed mainly with respect to the relative abundance of their predominant β-proteobacteria. Since a broad spectrum of xenobiotics and metals comprised the pollution cocktail of the Spittelwasser River, and none of these was the dominant carbon source or electron acceptor for the microbial community, specialists for each of these compounds possibly constituted only a small fraction of the microbial community and thus were not detected by the 16S rDNA approach. Similarly, Wikstrom et al. (41) found no effect on the community composition in lake microcosms after amendment with trinitrotoluene by using random amplified polymorphic DNA analysis. Accordingly, from our river biofilms only one Nitrosomonas oligotropha-like sequence was retrieved (Spb253), and it came from the Spittelwasser, which had a very high concentration of ammonia compared to the Elbe. Nitrosomonas-like organisms usually represent only a small subpopulation of the total community (19).

The dominant β-proteobacteria described here must be assumed to be tolerant to the concentrations of pollutants and metals present, especially in the Spittelwasser River. Biofilms with their thick layers of extracellular polysaccharide are known to protect bacteria from toxic compounds, including antibiotics. Besides, many catabolic plasmids are widespread in nature and their frequency increases in communities as a response to pollutant stress (21). In addition, close cell-to-cell contact in biofilms promotes plasmid transfer (42).

We observed three significant differences in the community structures of the two river biofilm communities. First, the uncultivated cluster RBF 2 consisted almost exclusively of Elbe clones; there were three corresponding bands in the community fingerprint, of which only one could be detected in the Spittelwasser River. Second, sequences of RBF 6 were only found in the Elbe clone library and the respective TGGE band was only present in Elbe community profiles. Third, RBF 3 sequences (closely related to H. taeniospiralis and H. palleronii) were only found in the Spittelwasser clone library. The TGGE bands corresponding to RBF 3 (bands 8, 12, and 13) were very strong in the Spittelwasser community fingerprint throughout the year while in the Elbe only two relatively weak bands could be detected. The predominance of these clones in the strongly polluted Spittelwasser River and the detection of several TGGE bands within this cluster, which showed different seasonal patterns in Elbe and Spittelwasser, suggests the presence of several closely related populations (species, subspecies, or strains) which may be specially adapted to the pollutants present in these rivers. It has to be noted, however, that several pollutants (e.g., DDT, tributyltin) had increased concentrations in wintertime. This is well documented in the Elbe (Table 3), whereas there are only occasional measurements from the Spittelwasser available. Thus, since temperature and pollutant load were to a certain extent correlated, no definite conclusion with respect to the underlying causal relationship was possible.

Seasonal variation.

Seasonality was especially observed for band 2 (RBF 2, representing uncultivated organisms mainly found in the Elbe) and band 5 (RBF 1, representing relatives of R. antarcticus), which were strongest in winter. The nearest described species R. antarcticus (25) and related clones (38, 40) were retrieved from habitats with low temperatures (Antarctica, Crater Lake, and Siberian reservoir). Interestingly, cloned ribosomal sequences from winter bacterioplankton in eutrophic Siberian reservoirs (38) often clustered with sequences from cold habitats. In addition, band 8 (RBF 3, H. taeniospiralis, 98.4%) and band 10 (RBF 8) also appeared most intense in wintertime in both habitats. Psychrotolerance has already been described for some members of the genus Hydrogenophaga (30). The predominance of these bands in both rivers in winter suggests that temperature- rather than river-specific parameters (e.g., water flow and pollution) caused this pattern, especially since the appearance and disappearance of bands 2 and 5 coincided exactly with seasonal temperature changes (May 1997, Oct 1997, and May 1998). This phenomenon was less pronounced in the Spittelwasser River, where the temperature did not drop as dramatically as in the Elbe. For the river biofilms investigated here, an increase in relative abundance in winter has also been observed for the Cytophaga-Flexibacterium cluster by using FISH with specific probes (6).

We can only speculate on the reasons for the seasonal patterns of some of the minor bands on the TGGE profiles, for which corresponding clones could not be found because they represented rare components of the pooled clone library. Since these bands were typically present in one river only, and only for a short time, the corresponding organisms may have been adapted to single events like algal blooms or particular pollutant loads.

In addition to the parameters listed in Tables 2 and 3, a large set of other river parameters monitored by the water quality authorities was checked for a possible correlation with changes in biofilm community fingerprints, but except for temperature and the pollutants listed in Table 3, no coincidence was found. This might indicate that the biofilm microbial community was well buffered and not affected by these parameters. On the other hand, the large number of potentially relevant parameters made a simple correlation in the frame of this study impossible. Besides, many other parameters which might also influence community composition had to remain unconsidered, especially biotic factors like the interaction between bacteria or bacteria and algae, grazing by protozoa and macroinvertebrates, and lysis caused by viruses (5, 15).

Given the complexity of rivers, the large input of allochthonous material, and the variability of physicochemical parameters during the seasons, the composition of the dominant representatives of the β-proteobacterial biofilm communities as revealed in this study appeared to be amazingly stable. The data suggest the presence of clusters of related organisms, some of them are well-described aquatic species and others are not-yet-cultivated taxa, whose relative abundances changed in response to seasonal variation in physicochemical conditions, most likely temperature and certain pollutants, and which may be distributed globally.