The active bacteria of a biofilm community grown directly on polychlorinated biphenyl (PCB) droplets were analyzed by 16S rRNA fingerprinting, identified by their 16S rRNA gene sequences and fatty acid profiling, and compared with isolates from the biofilm. Although, the multi-species biofilm degraded di- and trichlorinated PCB-congeners these substrates were not attacked by its individual isolated members, which suggests that a metabolic network is responsible for PCB degradation in the biofilm. The community metabolized (U- 13C]-2,2′-dichlorobiphenyl incorporating the label into certain phospholipid fatty acids matching those found in Burkholderia species. In contrast, abundant biofilm community members, like Methylobacterium species, did not incorporate the label. These results provide prima faciae evidence for Burkholderia species as the main degraders of PCBs in this type of aerobic soils.
Polychlorinated biphenyls (PCBs) are a class of organics consisting of 209 different compounds, called congeners, differing in the degree of chlorination and the position of the chlorine atoms on the aromatic rings. The remarkable physical properties of PCBs, and their extreme chemical stabilities were the reasons for their broad industrial application. The industrial production of PCB gave complex mixtures of congeners, which are usually grouped according to their content of chlorine atoms and had different trade names like Aroclor (USA), Clophen (Europe) or Kaneclor (Japan). All congeners are poorly soluble in water but well soluble in organic solvents including fat. These properties led to many industrial applications but make PCB a major environmental pollutant and allow its enrichment in food chains. As more attention was turned towards PCB, it became clearer that PCB was having a negative impact on many biological systems . It is estimated that until the final ban of PCB in the mid 1980s about 1.5 million tons of PCB were produced worldwide and a substantial fraction of it was finally released into the environment. PCB is now widely distributed in environments over the Earth and found even in remote parts of this planet .
Although many PCB-degrading bacteria are known, e.g., Pseudomonas pseudoalcaligenes, Sphingomonas aromaticivorans F199 , Comamonas testosteroni, Alcaligenes eutrophus H850 Rhodococcus globurulus P6  and Burkholderia xenovorans LB400 [8,9], this pollutant is recalcitrant in the environment and bioremediation has not been very successful. To assess the potential of intrinsic bioremediation and to understand the influence of this highly hydrophobic substrate on the autochthonous microbial community we characterized microbial communities from a highly PCB-contaminated moorland soil. This site has been studied intensively and a broad knowledge base exists on the bacterial community of the soil samples used in this study . The sandy soil was acidic with a pH of 4.0–4.5, low total organic carbon, and PCB concentrations reaching 24,100 ppm. The high PCB concentrations made this pollutant the main carbon source at this site. The questions we therefore posed were whether the microbial community can use PCB and, if so, which microorganisms are the main degraders.
To study the PCB-driven microbial community, we have developed a close-to-nature approach to study complex bacterial biofilms in microcosms . In this system, that is a microcosm filled with a soil slurry from the site under study, a biofilm develops on the lower surface of a floating plastic slide – Permanox®– as substratum, separated from the site soil by a water column. The formation of unusual microcolonies on the blank substratum, termed “clay hutches”, has been reported previously . To identify those bacteria that are directly able to tolerate and metabolize the hydrophobic and toxic substrate PCB, we dotted the Permanox® slide with microdroplets of Aroclor 1242 and analyzed the composition and activity of biofilms developing on the surface of the PCB microdroplets. These biofilms were characterized by a polyphasic approach comprising single-strand conformation polymorphism (SSCP) community fingerprinting, fatty acid methyl ester (FAME) profiling of polar lipids, and chemical analysis of PCBs. We also isolated microorganisms from the biofilm and analyzed their metabolic profiles. To identify the organisms responsible for the degradation of polychlorinated congeners in the biofilm community, (U-13C]-labelled 2,2′-dichlorobiphenyl was added to the PCB mixture and the flux of carbon into microbial biomarkers was monitored by isotope ratio mass spectrometry (IRMS).
2Materials and methods
2.1Characterization of the site
The site, situated north of Wittenberg, Germany, had been polluted more than a decade ago by PCB oil from condensers. The soil used in our experiments contained mainly sand with some clay and had a pH of 4.3, a low content of total organic carbon (TOC) of 1.27%, low total nitrogen (TN) of 32.5 ppm and a PCB concentration of 13,500 ppm. The PCB mixture was found to consist mostly of low-chlorinated congeners with small amounts of pentachloro-PCB, but no higher chlorinated congeners : 2.9% dichloro-, 44.7% trichloro-, 51.6% tetrachloro-, and 0.8% pentachloro-biphenyls. This congener pattern, and the relative amounts of the individual congeners, are similar to those reported for the Russian PCB mixture Trichlorophenyl, which resemble Arochlor 1242 . Because of this close similarity Aroclor 1242 was used as PCB mixture in our experiments. An obvious difference from commercial Trichlorobiphenyl was a depletion of less-chlorinated PCBs in the site.
Soil samples (300–500 g) were taken from the PCB-polluted site near Wittenberg, Germany and stored at 4 °C until use. Homogenized, freeze-dried soil (1 mg) was combusted in a Fisons EA 1108 element analyser with CHN packing for analysis of total organic carbon (TOC) and total nitrogen (TN). The analyses were run five times.
PCBs were extracted using a modified procedure described previously . To determine the effect of leaching, four Petri dishes loaded with PCB were exposed for 5 weeks to sterile water and the PCB remaining on the dishes was analyzed. The dishes were washed with acetone, the extracts were dried under a gentle stream of N2 and dissolved in 500 μl hexane. Capillary gas chromatographic analyses were performed on 1 μl aliquots of the extract on a Hewlett Packard 5890 Series II gas chromatograph equipped with a HP Ultra 2 capillary column (50 m by 0.2 mm; film thickness 0.11 mm) and FID detector. Hydrogen served as the carrier gas. Injector temperature was set to 250 °C and detector temperature was 300 °C. The oven program was: 80 °C for 3 min, 90–288 °C at 6 °C min−1 followed by an isothermal period of 20 min. The PCB congeners were identified by GC-MS  and by comparison with authentical standards. The 2,2′,5,5′-tetrachloro-PCB (PCB 52)  present in Aroclor 1242 is assumed not to be leached by water  and not aerobically degradable , so was used as an internal standard . The GC chromatograms were normalized to the area of PCB 52 from Aroclor 1242. The gas chromatographic–mass spectrometry analyses were performed with a similar gas chromatograph as described (same column and conditions but helium as carrier gas) connected to a HP 5989A quadrupole mass spectrometer. The electron impact ion source was maintained at 200 °C, while the quadrupole temperature was 100 °C. The electron energy was 70 eV.
Aroclor 1242 (4 μl) (Promochem, GB) was diluted in a mixture of acetone and iso-pentane (1:1.5, v:v) and slowly dropped on a sterilized substratum (Permanox® Petri dishes, d= 60 mm) (Lünsdorf, H., et al., in preparation). This procedure did not cover the walls of the Petri dishes with PCB and the biofilm grown there served as the PCB-free biofilm in the SSCP analysis. After evaporation of the solvent the disk was turned around and the other side was treated in the same manner; 18 of these dishes covered with PCB on both sides were fixed at the bottom of a reservoir filled with 4.5 l sterile tap water and 200 g of PCB-contaminated soil containing 13,500 ppm PCB (Fig. 1). The reservoir was gently shaken at 30 rpm. The structure and development of the biofilm on the PCB droplets were monitored by DAPI-staining and epifluorescence microscopy. Samples were stained by incubation of the biofilms in the dark for 5–10 min in a solution of DAPI, 1 μg ml−1, in sterile filtered PBS buffer, followed by several washing steps with buffer . After several weeks of incubation of the PCB-coated dishes in the microcosms, the walls of the Petri dishes (PCB-free) were cut off with scissors from the floor (PCB-coated) and the two parts were analyzed separately; biofilms were harvested with a sterile spatula. RNA was extracted from material representing 2–10 dishes, using the Fast-DNA-Spin-Kit for soil (Bio 101, CA, USA), and following the instructions of the manufacturer .
2.5SSCP fingerprint analysis of the rRNA amplicons
The RNA was then transcribed to cDNA and amplified by polymerase chain reaction (PCR) using the OneStep RT-PCR-Kit (Qiagen, Hilden, Gemany). The primers chosen for the amplification of bacterial 16S rRNA genes were forward primer Com1 and reverse primer Com2-Ph as published by Schwieger and Tebbe . The phosphorylated strand of the PCR products was digested by lambda exonuclease (New England Biolabs, Schwalbach, Germany), proteins removed by the Mini-elute-Kit (Qiagen, Hilden, Germany) as recommended by the manufacturer, and the remaining single stranded DNA was dried under vacuum. The DNA was then resuspended in denaturating SSCP loading buffer (47.5% formamide, 5 mM sodium hydroxide, 0.12% bromophenol blue and 0.12% xylene cyanol) and subjected to electrophoresis . Gels were run at 400 V for 18 h at 20 °C in a Macrophor electrophoresis unit (LKB Bromma, Sweden) and subsequently silver stained .
2.6Sequence determination of fingerprint bands
Single bands were excised from the gels and eluted in extraction buffer (10 mM Tris-buffer, 5 mM KCl, 1.5 mM MgCl2· 6H2O, 0.1% Triton X-100, pH 9.0) at 95 °C for 15 min. Extracts were centrifuged (14,000 rpm, 1 min) and the DNA in the supernatant fluid was used for a PCR with the primers described above. The PCR-product was cleaned (Mini-elute-Kit, Qiagen, Hilden, Germany) and sequenced with a sequencing kit (DYEnamic ET Terminator cycle sequencing kit by Amersham Biosciences, Freiburg, Germany) and both primers. The product was cleaned with the Dye Ex Spin Kit (Qiagen, Hilden, Germany) and the sequence analyzed on an ABI PRISMTM 337 DNA-Sequencer and 3100 Genetic Analyser. Comparisons of the sequences were performed using the N-FASTA program and the databases of EMBL and GenBank (Table 1).
Table 1. Phylogenetic assignments of community profile sequences
Closest sequence and Accession Nos.
aBands 1–16 were from biofilms grown on PCB, whereas bands 17–20 were from biofilms without direct PCB contact.
Uncultured bacterium (from a BTX-degrading community) AF312220
Burkholderia sp. AB101 AF219126
Planctomyces sp. X81950
Soil bacterium X64383
Uncultured bacterium (Antarctica) AF424411
Eubacterium halii L34621
7 + 8
Afipia massiliensis AY029562
Bradyrhizobium sp. AY430822
Methylobacterium sp. AF361189
Methylobacterium sp. Z23156
Methylobacterium sp. X89908
Methylobacterium sp. D32233
14 + 15
Aquabacterium sp. (drinking water biofilm) AF08958
Clone from Thailand AJ319572
Uncultured rhizosphere bacterium AJ431308
Uncultured bacterium AJ431298
Spirosoma-like sp. (from biofilm in air conditioner) X89919
Uncultured bacterium (from hot spring, Yellowstone Park) AF445698
2.7Isolation and characterization of biofilm community members and biphenyl degraders
Bacteria were isolated from PCB-contaminated soil (strains WAB584 and WAB586, described in ) using biphenyl as carbon source as described previously  and from several biofilm experiments on plates with R2A medium (WAB1197-WAB1237) and on biphenyl (WAB1246-1252). R2A was used to isolate as many biofilm community members as possible independent from their ability to use PCB. Bacteria scrabbed from the plates were suspended in 1 ml sterile 0.85% NaCl solution, the bacterial suspension was streaked out on the plates in a dilution series and well-separated colonies re-streaked to obtain pure clones. Clones were identified by sequencing their 16S rRNA genes  and comparison of these sequences with public and in-house databases. For a modified Bligh–Dyer extraction , 200 mg of wet biomass harvested in mid-logarithmic growth phase were used as described in detail previously . From all isolates the fatty acids of the phospholipids were determined. Individual fatty acid methyl esters were identified by their retention time compared to standards and control measurements with GC–MS .
Isolates were analyzed for substrate usage in liquid cultures with minimal medium (8 g NH4H2PO4, 0.2 g yeast extract, 2 g K2HPO4, 0.5 g MgSO4· 7H2O, 0.5 g Na2SO4, 0.5 g NaCl, 10 mg ZnCl2· 2H2O, 8 mg MnSO4· 7H2O, 10 mg FeSO4· 7H2O, and 50 mg CaCl2 in 1 liter of distilled water) and a single carbon source (Table 2). The concentrations of the carbon sources were either 100 mg l−1 for biphenyl compounds (added as solids or liquids to the medium), or 1 g l−1 for chlorobenzoates. Because some of the isolates showed growth on acetone the biphenyls were added directly although acetone was recommend as solvent . Measurements of D600 together with visual controls could reliably show any growth on biphenyls (e.g., WAB584). Minimal medium with and without carbon sources were inoculated with equal cell numbers and growth examined by determination of D600 after 0, 1, 2, 3, 5 and 7 days and compared to cultures without substrate.
Table 2. Metabolic profiles of bacterial isolates obtained from PCB-contaminated soil (WAB584, WAB586)  and PCB-biofilms (WAB1197-1252)
−: no growth; □: weak growth; : moderate growth; °: strong growth; BA: benzoic acid; xCl-BA: x-chloro-benzoic acid; BP: biphenyl; 4CL-BP: 4-chloro-biphenyl; Aro1221: Aroclor1221; Aro1242: Aroclor 1242; A: Arthrobacter, B: Burkholderia, M: Methylobacterium; W: isolate belongs to a novel genus preliminarily named “Wittenbergia”.
(U-13C]-2,2′-Dichlorobiphenyl (120 μg) (Cambridge Isotope Laboratories, MA, USA) was added to the PCB-mixture Aroclor 1242 in the proportion 13C:12C = 1:1650 (each dish: 5.6 mg Aroclor 1242 and 3.4 μg (U-13C]-2,2′-dichlorobiphenyl). The microcosm was harvested after 5 weeks and separated into two sets comprising six (sample A) and seven (sample B) dishes with the biofilms. From each set several pieces comprising about 10% of the total PCB area were cut and extracted with hexane to determine the composition of the remaining PCB. The biofilm community was analyzed with SSCP, and the lipids were extracted from the biomass to determine the composition and the isotope ratio of their fatty acids. The lipids were first separated into polarity classes and the fatty acid methyl esters (FAMEs) of the glycolipid (GL-FAMEs) and the phospholipid (PL-FAMEs) fractions were identified  and their isotope ratios determined.
To determine the carbon isotope ratios of the biomass or the substrate, samples of 0.02–0.5 mg of carbon were dried in a tin cup at about 70 °C, combusted in an EA Fisons 1108 with CHN packing, and analyzed with the isotope ratio mass spectrometer (IRMS, Mat 252, Finnigan, Bremen, Germany). The analyses were run five times. The ratios of 13C to 12C in the fatty acid methyl esters were measured with a GC (HP5-column) coupled via a combustion interface to the IRMS (GC-C-IRMS), using the δ-notation and PDB-standard (PeeDee Belemnite) as described previously . The analyses were run three times.
3.1Diversity of the biofilm community growing on the PCB droplets
A complex biofilm community originating from the soil inoculum was observed within a few weeks on the PCB droplets (Fig. 2). To identify the metabolically active members of the microbial community, the 16S rRNA was extracted from the biofilm and analyzed with RT-PCR amplification followed by SSCP community fingerprinting, which allowed a direct comparison with the microbial diversity reported previously for the site . The SSCP fingerprints showed a surprisingly complex community in the biofilms. Clear differences in the community profiles of biofilms grown on the substratum alone, and biofilms grown on the PCB droplets, were observed (Fig. 3). Extraction and sequencing of the main bands from the SSCP gel revealed members from the phyla Bacteroidetes, Firmicutes, Planctomycetes and the Alpha- and Betaproteobacteria, including members of the genera Methylobacterium and Burkholderia (Table 1). Sequences of organisms belonging to the Bradyrhizobiaceae, Betaproteobacteria and Firmicutes seem to be preferably found in the biofilms grown on the PCB droplets.
3.2Chemical analysis of the PCB droplets
The PCB on the substratum remaining at the end of the microcosm experiments was analyzed and compared with the composition of Aroclor 1242 and the PCB from a sterile control experiment (Fig. 4). The main difference was that mono- and dichlorinated biphenyls, 2,4,4′-/2,4′,5-trichloro-biphenyl (PCB28, PCB31) and 2′,3,4-trichloro-/2,2′,5,6′-tetrachloro-biphenyl (PCB33, PCB53) were significantly depleted from the biofilm droplets. A control experiment with a sterile soil inoculum revealed that possible differential leaching of congeners by water was small and not correlated with the observed decreases in di- and trichlorinated congeners.
3.3Metabolic diversity of isolates from the contaminated soil and the biofilms
To identify the degraders of the di- and tri-chlorinated PCB congeners bacteria from the soil were isolated using biphenyl as the sole source of carbon. Furthermore, bacteria from the biofilm grown on Aroclor 1242 were isolated on R2A agar. The different isolates were characterized by their ester-linked phospho- and glycolipid fatty acids and identified by their 16S rRNA gene sequences as belonging to the genera Burkholderia, Methylobacterium, Arthrobacter and a novel genus preliminary named “Wittenbergia”, closely related to Beijerinckia. Only the isolates of the genera Arthrobacter and Burkholderia could grow on several chlorinated benzoates and biphenyl, but these generally exhibited the ability to use a range of different substrates tested. None of the isolates obtained either from site soil or from the biofilms was able to grow on PCB congeners with two or more chlorine atoms including 2,2′-dichloro-biphenyl (data not shown) (Table 2).
3.4Incorporation of 13C-labelled 2,2′-dichlorobiphenyl into biomarkers
Since the biofilm, but not any of the isolates, degraded di- and trichlorinated-PCBs, these congeners are probably either degraded by uncultured members of the community, or as a result of metabolic interactions and networks in the community. Therefore, a tracer approach was applied to identify the degraders of polychlorinated congeners. The substrate Aroclor 1242 was amended with small amounts of (U-13C]-2,2′-dichloro-biphenyl in a ratio of 1:66, labelled to unlabelled congener, and used for the microcosm. The carbon isotope ratio of this congener thereby changed from δ13C =−15‰ to δ13C =+1900‰, while the other congeners showed an average isotope ratio of δ13C =−25‰.
Whereas the isotope ratios of the GL-FAMEs obtained from the biofilm grown on the 13C-enriched substrate were not much different from those of the unlabelled control, the PL-FAMEs all showed 13C-enrichments, some of which were highly significant (Table 3).
Table 3. Relative intensity and isotope ratios of GL- and PL-FAMEs obtained from two duplicate biofilms, A and B, grown for 5 weeks on Aroclor 1242 with (U-13C]-2,2′-dichlorobiphenyl
Rel. area (%) (n= 6)
δ13C (‰) (n= 3)
13.1 ± 4.3
296 ± 18
582 ± 7
33.0 ± 5.8
2384 ± 119
2677 ± 21
4.4 ± 1.2
5277 ± 332
9475 ± 104
1.0 ± 0.8
109 ± 72
251 ± 119
4.1 ± 0.6
3116 ± 744
4404 ± 183
35.5 ± 3.3
194 ± 20
235 ± 37
6.1 ± 1.9
324 ± 31
583 ± 61
1.5 ± 0.8
307 ± 33
420 ± 49
1.2 ± 0.2
217 ± 36
320 ± 32
58.5 ± 3.0
−28.5 ± 0.6
−29.7 ± 0.2
13.6 ± 0.6
−33.3 ± 0.1
−33.3 ± 0.7
0.8 ± 0.1
−18.9 ± 4.8
−19.9 ± 0.6
0.6 ± 0.2
−10.9 ± 4.0
−16.3 ± 5.1
9.1 ± 2.2
48.9 ± 2.7
28.0 ± 0.8
5.9 ± 2.4
−1.3 ± 0.3
16.7 ± 0.0
0.5 ± 0.1
−30.1 ± 7.7
−32.1 ± 7.2
1.2 ± 0.7
103.7 ± 5.9
44.9 ± 0.6
0.4 ± 0.0
75.5 ± 0.8
126.9 ± 0.4
9.5 ± 2.1
−22.1 ± 0.2
−24.7 ± 0.6
To identify the bacteria, which incorporated the 13C from the PCB substrate, the patterns of labelled PL-FAMEs were compared with those from the isolates. The Arthrobacter species isolated from soil and well growing on PCB (Table 2) were obviously missing in the biofilm because their characteristic PL-FAMEs (15:0i, 15:0a, 17:0i, 17:0a) could not be found among those of the biofilms in quantities sufficient for IRMS analyses. Although the characteristic PL-FAMEs of the Burkholderia (16:0, 17:0cycloω7c, 18:1ω9c, 19:0cycloω8c ), Methylobacterium and “Wittenbergia” isolates were observed in the biofilm community, only PL-FAMEs observed in the Burkholderia isolates had incorporated 13C from (U-13C]-2,2′-dichlorobiphenyl (Fig. 5). This pointed to species of the genus Burkholderia as the degraders of this congener. The PL-FAME 18:1ω7, which is characteristic for the Methylobacterium isolates of the biofilms, and which is the most abundant in the biofilm, showed only minor 13C enrichment.
Polychlorinated biphenyl congeners are superhydrophobic , hardly soluble in water, and therefore poorly bioavailable. This characteristic of PCBs is a key problem for bioremediation, and many attempts have been made to increase PCB bioavailability . Although biofilm growth on polycyclic aromatic hydrocarbons (PAHs) has been reported  the observation that biofilms can directly grow on the PCB oil is perhaps unexpected and demonstrates that some bacteria at least have adaptive mechanisms to deal with the hydrophobic stress PCBs generate for their membranes. The diversity of microorganisms detected in the biofilms grown on the PCB droplets was surprisingly high and included known organic pollutant degraders like Planctomycetes and Burkholderia. The relatedness of a SSCP band with a known sequence was in several cases not very helpful because the similarities were sometimes low and the closest relative was often a clone sequence of an uncultivated organism about which nothing is known. It is interesting to note that not a single sequence was found, which was identical to any of the isolates, clones of the PCR products of 16S rRNA genes or clones of the RT-PCR products of 16S rRNA from this site [10,33,43] Reasons for that are probably biases introduced by cloning of DNA, PCR in the SSCP analysis and selectivity in the different isolation approaches. Nevertheless, the same group of Burkholderia, Methylobacterium and “Wittenbergia” species has been found with all three techniques and therefore their detection was not compromised by the different approaches. Ambiguities resulting from the short length of sequences determined in SSCP analyses have been noted by others , but nevertheless, the SSCP analysis here pointed to the identity of those bacteria, preferentially found only in the biofilms grown on PCBs, and has provided the link with the FAMEs of organisms in the biofilm, and thereby led to the identification of the PCB degraders in the community labelled with 13C-enriched PCB.
From the PL-FAMEs extracted from the biofilm community 13 were sufficiently abundant to determine their δ13C data. Analysis of the labelling pattern of the PL-FAMEs revealed that not all FAMEs showed comparable enrichments but it can be assumed that the primary degrader of the labelled compounds shows higher enrichment of its biomass than a secondary consumer of its metabolites. Furthermore, PCB was not the only carbon source in the microcosm experiment because the soil used as inoculum contained 2.54 g TOC including 1.50 g carbon from the PCB contamination (=2.7 g PCB). Compared to the 100 mg Aroclor 1242 coating on the dishes, TOC is the major carbon input in the system. This is mirrored in the labelling pattern of the PL-FAMEs. Here 18:1ω7 is one of the main PL-FAMEs but showed only little 13C-enrichment. A comparison with the isolates revealed that 18:1ω7 is the main fatty acid of Methylobacterium and the low labelling together with the inability of the isolates to grow on any of the substrates tested (Table 2) makes it highly unlikely that strains of this genus are involved in the degradation of PCB. The same arguments exclude “Wittenbergia” from PCB degradation. Only the PL-FAMEs found in Burkholderia isolates were highly labelled and supported the notion that strains of this genus were the main degraders of the labelled congener. Burkholderia species show few and rather common PL-FAMEs, furthermore, not for all FAMEs reported from the isolates 13C data could be determined in the biofilm extract (e. g. the unknown FAME ECL 18.586 and 2-OH 16:0), however, the FAMEs of the biofilm with the highest 13C incorporation were all found in Burkholderia sp. as well. The findings from the labelling study were corroborated by the observation that only Burkholderia isolates from this site were able to use chlorinated benzoates or biphenyl as substrates for growth. It also is consistent with an earlier finding that species of the genus Burkholderia were abundant both in rRNA and in rDNA clone libraries generated from the polluted soil .
It is rather surprising that the GL-FAMEs did not show much incorporation of the 13C-label while the PL-FAMEs from the same biomass were highly 13C-enriched. A probable reason for this is a much slower turnover of glycolipids compared with phospholipids. It has been reported for bacteria that glycolipids are synthesized from phosphatidyl acids, which means that incorporation of 13C should be found first in the PL-FAMEs and only later in the GL-FAMEs [34,35]. Glycolipids seem to be much more static than the phospholipids because a measurable turnover could not be observed in the glycolipids of Acholeplasma laidlawii and their biosynthesis seems to be slower . A very slow incorporation of the 13C-label from the substrate into GL-FAMEs has also been found in experiments with pure cultures  indicating that this is a common phenomenon. The 13C isotope ratios of GL-FAMEs from a Mycobacterium sp. grown on different substrates were closer to those both of the biomass and the substrate than those of the PL-FAMEs  also indicating a slower turnover rate of the glycolipids compared to the phospholipids. Because of their slow 13C-incorporation, GL-FAMEs are of little use for pulse-labelling studies, even when they contain good biomarker fatty acids, but they may have merits for long-term experiments.
The chemical analysis of the PCB congeners remaining under the biofilms demonstrated that the bacteria in the biofilm preferably degraded the less chlorinated congeners, but degradation of 2,2′,5,6′-tetrachlorobiphenyl (PCB53) was also found (Fig. 4). This reflects previous findings both for PCB-degrading isolates, and for microbial degradation of PCBs in the environment . The degradation of 2,2′,5,6′-tetrachlorobiphenyl, however, has not so far been reported for a pure strain [41,42]. Although our Burkholderia isolates were able to use several chlorinated benzoates and biphenyls as carbon sources , they also could not grow on polychlorinated congeners. The growth of some isolates on Aroclor 1242 can be explained by the consumption of the monochlorinated congeners present in this mixture. A number of strains were found to oxidize 2,2′-dichloro-biphenyl in resting cell culture [8,40–42] but the only isolate reported so far to be able to grow on this congener is Alcaligenes sp. SK-4 . However, the detected labelling of the PL-FAMEs pointed to Burkholderia species being responsible for the metabolization of 2,2′-dichlorobiphenyl in the microbial community of the biofilm. It is clear from the SSCP fingerprints that, on one hand, Burkholderia species represent only a small fraction in the rather complex biofilm found on the PCB droplets but, on the other, that these bacteria are mainly responsible for the metabolism of PCBs as shown by the incorporation of the label. The microbial consortium of the biofilm probably achieves mineralization of at least this polychlorinated congener through metabolic interactions in the community network .
For PCBs it is known that biodegradation proceeds via an upper and a lower pathway, representing the sequential degradation of the two aromatic rings. Often bacteria express either the upper or the lower pathway, but not both [44,46]. This is also the case for many of the isolates we obtained from the PCB site (Table 2). These seem to represent cellular and metabolic modules of a community network exploiting a recalcitrant and complex substrate, in this case PCBs. Synthrophic interactions in biofilms degrading xenobiotics have been deduced on the basis of the ultrastructural organization observed in these biofilms . The isolates from the PCB biofilm belonging to the genera Methylobacterium and “Wittenbergia” could neither grow on biphenyl nor on chlorinated benzoates. However, they were always found as members of the mature PCB biofilm community. Their functional role in the PCB-degrading biofilm community remains unclear.
Metabolic networks certainly play an important role in biofilm activities. They have been elucidated for more simple bacterial communities  and also for biphenyl mineralizing consortia . Our results support the view that at least in PCB-polluted sites, diverse strains that individually have relatively limited catabolic activities are common, but that their collective activities and synergistic interactions in the biofilm constitutes a community metabolic network that enables utilization of a broad spectrum of compounds. Our experiments indicated that a number of bacteria that did not degrade PCB were always present in the mature biofilm community. We assume that these mediate important interactions, which provide the microbial biofilm consortium with increased stability, tolerance to stresses, and robustness over those possessed by individual strains, and are essential for the community to act as a functional unit. The obvious conclusion of this is that bioaugmentation with high-performance strains for in situ bioremediation must involve metabolic and functional compatibility between the added strains and the indigenous community, so that a functional metabolic network is obtained and beneficial interactions are achieved.
We thank Peter Wolff and René Huppmann for technical assistance in the chemical analyses, Dagmar Wenderoth and Jennifer Skerra for their help in the microbiological experiments and Esther Surges for the IRMS analysis. This work was supported by grants from the German Federal Ministry for Science, Education and Research (Project No. 0319433C) and the Helmholtz strategy funds “Soil function”.