Enrichment cultures inoculated with Arctic soil yielded a biofilm that grew on pyrene and phenanthrene. In a 60-day period, the biofilm degraded 20 μg ml−1 pyrene or 39 μg ml−1 phenanthrene. Single colonized pyrene crystals (approximately 1.5×0.75×0.35 mm) yielded 1011 culturable heterotrophs and 105 biofilm propagules. Analysis of ribosomal intergenic spacers identified six phylotypes in a clone library from the pyrene biofilm. Analysis of 16S rRNA gene sequences indicated that the phylotypes, in order of decreasing abundance, are most closely related to members of the genera Polaromonas, Sphingomonas, Alcaligenes, Caulobacter and Variovorax. Two isolates capable of growth on pyrene, both Pseudomonas spp., were obtained from the pyrene enrichment culture. Growth of microbial biofilms on polycyclic aromatic hydrocarbons has not been reported previously, and this mode of growth may have important effects on substrate uptake.
Biological degradation of polyaromatic hydrocarbons (PAHs) has gained attention over the last decades, since it has been shown to be a cheaper method to clean up contaminated soils than incineration or storage of the soil . During treatment of PAH-contaminated soil, it is important to optimize degradation parameters such as oxygen, nitrogen, phosphorus, temperature and PAH bioavailability [2,3]. In order to obtain an effective treatment it is also useful to identify the microbial populations, and their degradation capabilities in the soil. With the purpose of producing both pure isolates and mixed populations, for bioaugmentation, several attempts have been made with various techniques to isolate and describe bacterial communities in soil and water systems . Studies of pure cultures have shown possible degradation pathways and allowed the study of the enzyme systems involved [2,4,5]. What is less well understood is the interaction between the bacteria and the substrates. An important consideration is the low water solubility of PAHs. Also, toxicity of PAHs and their metabolites to PAH-degrading bacteria has been reported . Depending on the way the PAH degraders are exposed to PAHs (i.e. solid or dissolved PAHs), different strains with different degradation capabilities will be selected [7–9]. Degradation can be enhanced by sorbing PAHs or other organic contaminants to a matrix prior to treatment [7,10]. Close proximity of cells to a poorly soluble substrate may enhance degradation kinetics . If microorganisms can utilize a substrate directly, without the need for the compounds to be dissolved into the aqueous phase, degradation may be further enhanced . Thus, attachment of microorganisms directly to compounds may facilitate degradation of those compounds and be a useful strategy for treatment of compounds with low water solubility, such as PAHs. Growth of microorganisms attached to naphthalene, phenanthrene and anthracene has been reported [12–14].
In this study, a novel mode of bacterial growth on pyrene is reported. A dense bacterial biofilm grew directly on pyrene crystals in liquid medium without increasing the turbidity of the liquid. The community in the biofilm was studied by microscopy, isolation of pure cultures and analysis of cloned PCR amplicons from the bacterial ribosomal operon (rrn). The degradation of various PAHs by the biofilm was measured by gas chromatography with a mass spectrometer detector.
2Materials and methods
Bushnell–Haas medium, tryptic soy broth (TSB) and Davies minimal medium were purchased from Difco, CA, USA. Naphthalene (99%), phenanthrene (99%), pyrene (99%), 9,10-dimethylanthracene (99%), 1,2-benzanthracene (99%), chrysene (99%) and benzo[a]pyrene (98%) were purchased from Sigma Aldrich, CA, USA.
2.2Enrichment of biofilm on pyrene crystals
Arctic soil contaminated with greater than 50 mg kg−1 PCBs plus associated oil (possibly containing PAHs) was collected from Saglek, Labrador, Canada (58°N, 64°W). Enrichment cultures were in 240-ml Teflon-septum-sealed dark bottles (Supelco, CA, USA). Cultures contained 50 ml Bushnell–Haas medium plus approximately 300 mg l−1 of pyrene as the sole organic substrate, added in crystal form after autoclaving. The pyrene crystals were approximately 1–2 mm by 0.5–1 mm by 0.2–0.5 mm. Cultures were inoculated with 1.0 g of the soil and incubated at 7°C or 22°C on a shaker at 150 rpm or stationary. After 3 months, the pyrene crystals at 22°C were colonized by a fluffy biofilm. One colonized crystal each was transferred to fresh medium with pyrene, phenanthrene, 9,10-dimethylanthracene, 1,2-benzanthracene, chrysene or benzo[a]pyrene. All PAHs were added in crystal form. After 2 months of incubation, the pyrene and phenanthrene crystals in the secondary cultures had dense biofilms. DNA was then extracted from the pyrene biofilm (see below). Uninoculated controls with pyrene did not produce any biofilm or become turbid due to growth.
2.3Isolation of pyrene degraders
Pyrene degraders were isolated from the biofilm by plating crystals with biofilm (from the secondary culture above) on TSB (3 g l−1) or Bushnell–Haas medium with agar. Pyrene was added as pure crystals on top of the latter plates with and without streaking the crystals on the plates. The plates were incubated at 22°C. Colonies growing near the crystals were picked and streaked on fresh plates until pure colonies were obtained.
2.4Quantitative analysis of pyrene and phenanthrene degradation
Liquid cultures were prepared in amber 240-ml Teflon-septum-sealed bottles with 10.0 ml sterile Bushnell–Haas medium plus 40 μg ml−1 phenanthrene or pyrene added after autoclaving. Each culture was inoculated with a pyrene crystal colonized with biofilm. Uninoculated controls were also prepared. All treatments were prepared in triplicate and incubated on a shaker at 150 rpm, in darkness, at 22°C for 60 days before PAHs were analyzed.
Crystals with visible biofilm production were removed from the medium with sterile Pasteur pipettes and placed on microscope slides. The preparations were examined with an Olympus IX70 light microscope with a 3CCD camera connected to a PC computer. Files were obtained in tif format using Imagepro (Olympus). Pictures were cropped and adjusted for only contrast and brightness.
Total DNA was extracted from 0.2 g wet weight of pyrene crystals using a FastDNA Spin kit (Bio 101, Quantum Technologies, CA, USA). The bead beating time was extended to two times for 2.5 min at 5000 cycles min−1. The DNA obtained was stored in TE buffer at −20°C before use. RIS analysis was done using primers S926f (5′-CTYAAAKGAATTGACGG-3′), targeting bacterial 16S rDNA, and L189r (5′-TACTGAGATGYTTMARTTC-3′), targeting bacterial 23S rDNA, as described previously . The 50-μl PCR reaction contained 25 pmol of each primer, 200 μM of each dNTP, 1.5 mM MgCl2, 1×PCR buffer (20 mM Tris–HCl pH 8.4, 50 mM KCl), 670 μg ml−1 bovine serum albumin and 1.25 U of Taq DNA polymerase (Gibco BRL) and approximately 300 ng DNA template. The PCR program was a hot start at 95°C for 2 min; 30 cycles of: 94°C for 0.5 min, 47°C for 0.5 min, 72°C for 2 min; and a final extension time of 5 min at 72°C. Resulting amplicons included RISs plus flanking fragments of the 16S and 23S rDNA (RIS-rDNA). The amplicons were approximately 800–1500 kb and were separated on a 2.0% agarose gel. A 100-bp ladder (Gibco BRL) was used to calibrate the gels. The gels were stained with GelStar (MFC Bioproducts, ME, USA).
2.7Cloning, restriction fragment length polymorphism (RFLP) analysis and sequencing
The RIS-rDNA amplicons were cloned using the TOPO TA cloning kit (Invitrogen, CA, USA). The cloned plasmids were purified by alkaline lysis miniprep . The clones were separately digested with MspI, MboI and TaqI (Life Technologies, Rockville, MD, USA) for RFLP analysis. For each amplicon with a unique RFLP pattern, a fragment (approximately 500 bp) of the 16S rDNA gene included in the amplicon was sequenced. The orientation of each cloned insert was determined by two PCR reactions with each clone using the primer pairs M13f plus L189r or M13f plus S926f. Then, the appropriate primer, M13f or M13r, was used to sequence the rDNA fragment. Sequencing was performed by the Nucleic acid and Protein Sequencing Unit (University of British Colombia, Vancouver, BC, Canada). A phylogenetic tree was constructed using the partial 16S rDNA sequences and reference strain sequences aligned using Sequence Aligner implemented in the Ribosomal Database Project  and then manually aligned using Gene Doc . The reference strain sequences were from GenBank and include the closest match for each sequence resulting from a BLAST search. Evolutionary distances were calculated according to Jukes and Cantor , and an unrooted neighbor-joining phylogenetic tree was constructed using the Phylip package . The phylogenetic tree was verified using 100 randomly produced bootstraps.
2.8Enumeration of total heterotrophs and pyrene biofilm propagules
One crystal with biofilm was transferred to a 10-ml vial with either 2.0 ml TSB or 2.0 ml Bushnell–Haas medium. The colonized crystal was vortexed and then crushed with a sterile spatula then vortexed again, which made the solution turbid from breaking up the biofilm. A 10-fold dilution series was done in each medium in triplicate. Control dilution series were made with spent culture medium without crystals. To each vial with Bushnell–Haas medium, approximately 300 mg l−1 of pyrene in crystal form was added. Cultures were incubated at 22°C in darkness without shaking. After 5 days of incubation, a positive score for culturable heterotrophs was based on visible turbidity in cultures on TSB. After 60 days of incubation, a positive score for biofilm propagules (the minimum inoculum capable of forming a biofilm) was based on visible biofilm formation on any pyrene crystals. Microscopy was used to verify that the biofilm was composed of cells.
2.9Analysis of PAHs
Whole cultures (10.0 ml) were acidified with 1.0 ml of 3.0 M H2SO4 and extracted with 10.0 ml ethyl acetate on a shaker at 150 rpm overnight at 22°C. Extracts were dried over anhydrous sodium sulfate before analysis on a Varian 3400Cx Gas Chromatograph with a Saturn 4D ion trap MS detector. The column was a DB5-MS (J&W Scientific) 30 m, i.d. 0.25 mm, film thickness 0.25 μm. The temperature program was 40°C for 5 min, 10°C min−1 to 245°C and hold for 30 min. The injector (Varian 1078) was operated at 240°C, and the transfer line at 250°C. Splitless injection was for 30 s on a 0.8-mm i.d. liner. The Ion trap was operated at 70 eV and scan range m/z 90–400. The carrier gas was helium at 10 Psi. The sample volume was 1.0 μl.
2.10Nucleotide sequence accession numbers
The partial 16S rDNA sequences of the clones and the isolates were deposited in GenBank under the following accession numbers: M6, AF312669; M7, AF312670; M8, AF312671; M15, AF312672; M16, AF312673; M20, AF312674; K319, AF312675; PK4, AF312676.
3.1Biofilm growth on pyrene crystals
Microbial growth on the pyrene crystals was visible to the eye after 3 months of incubation of the primary enrichment culture at 22°C. Shaken and stationary cultures yielded indistinguishable biofilms on the crystals. Shaken cultures were propagated for further study. Parallel cultures at 7°C yielded no visible growth. In the secondary culture, fresh crystals were visibly colonized within 1 month (Fig. 1). The biofilm was initially orange but darkened to brown or black after several months. During the colonization of the crystals, the medium remained clear. After 4 months, the optical density at 600 nm of the medium of the primary culture was 0.025. When colonized pyrene crystals were used as inocula, phenanthrene was the only other PAH that was colonized. Biofilm did not form on 9,10-dimethylanthracene, 1,2-benzanthracene, chrysene or benzo[a]pyrene after more than 5 months of incubation. However, the culture containing 1,2-benzanthracene turned slightly yellow after 5 months of incubation, indicating that transformation of the compound occurred without biofilm production. The biofilm grown on phenanthrene was yellow–white rather than the orange of the biofilm on pyrene.
3.2Quantitative analysis of pyrene and phenanthrene degradation
Secondary cultures inoculated with one colonized crystal from the primary culture removed 50% of the pyrene (40 mg l−1 added) or 98% of the phenanthrene (40 mg l−1 added) in 60 days. Both compounds were removed during biofilm production on the crystal surfaces. No PAH removal occurred in the uninoculated controls. Therefore, PAH removal was presumably due to biodegradation by the biofilm members. No metabolites from either pyrene or phenanthrene were detected at the end of the experiment. Therefore, the PAHs appear to have been mineralized. During the extraction of the PAHs, the biofilms were broken up which suggests that the loss of PAHs was not due to strong sorption to the biofilm.
The disrupted biofilm from a single pyrene crystal yielded 1011 culturable heterotrophs and 105 biofilm propagules. Spent medium with no colonized crystal yielded 106 culturable heterotrophs per ml and no biofilm propagules.
3.4RFLP and phylogenetic analysis of the biofilm
The RIS-rDNA amplicons from pyrene and phenanthrene biofilms had similar but not identical banding patterns (Fig. 2). RFLP analysis of 22 cloned RIS-rDNA amplicons from the pyrene biofilm identified six distinct phylotypes (Fig. 3). The RIS-length polymorphism (RIS-LP) bands from the six phylotypes corresponded to bands in the pyrene biofilm community RIS-LP pattern (Fig. 2). The phylotypes were phylogenetically identified by their partial rDNA sequences (Fig. 4). Five of the six phylotypes are closely related to described strains by partial 16S rDNA sequence with a sequence similarity of >93% for M15 and of >98% for M6, M7, M8 and M16. The sixth phylotype, M20, is affiliated with the α-subdivision of the Proteobacteria. The three most abundant phylotypes in the library, in order of decreasing abundance, are affiliated with the genera Polaromonas, Sphingomonas and Alcaligenes.
3.5Isolation of pyrene-degrading bacteria
From 25 colonies on plates with pyrene crystals, two morphologically distinct pyrene-degrading strains, K319 and PK4, were isolated. The two strains had a very similar major RIS-rDNA band, but each had a distinct secondary band (Fig. 2). The two strains had RIS-rDNA bands distinct from those in the banding patterns from the pyrene and phenanthrene biofilms and distinct from those of the clones from the pyrene biofilm. The two strains were clearly distinguished from one another by RIS-rDNA RFLP analysis. The two strains also differed in substrate use. Strain PK4 grew on dodecane and hexadecane as sole carbon sources, while strain K319 did not. Both strains grew slowly on pyrene but did not grow on naphthalene, fluorene or phenanthrene. They grew fast on dextrose, pyruvate or arabinose at 22°C. The two strains are associated with the fluorescent Pseudomonads on the basis of their 16S rDNA sequences (Fig. 4). On TSB plates, both isolates formed mucoid, gray–white, smooth, circular colonies. On Davies minimal medium with pyrene, both isolates grew well from 7°C to 30°C. They were Gram-negative, motile rods 1.5–3 μm long by 0.8–1.1 μm wide. Neither isolate formed a biofilm on pyrene crystals during 3 months of incubation.
The production of a dense biofilm directly on the pyrene crystals suggests that the pyrene is not toxic to the responsible organisms. Microorganisms attached to the solid crystals may have an advantage over unattached microorganisms. The solubility of pyrene in water is low, 0.13 mg l−1. The attachment may permit a relatively high rate of growth. The dense biofilm was capable of removing 20 mg of pyrene per liter in 60 days, which is a relatively high rate for such low nutrient conditions. In another study, Pseudomonas putida KBM-1 in mineral medium at 20°C was shown to aerobically degrade 0.13 mg of pyrene per liter in 50 days . Sepic et al.  reported that Mycobacterium sp. PYR-1 in mineral medium at 22°C on a shaker removed 4 mg of pyrene per liter in 28 days. Higher rates of pyrene removal were reported by Boonchan et al. ; their bacterial consortium in mineral medium at 30°C on a shaker degraded up to 225 mg of pyrene per liter in 12 days. The rate of pyrene degradation by the biofilm is difficult to compare with other studies, since biomass and growth are problematic to monitor. Also, the size of pyrene crystals used in various studies could vary and affect growth kinetics. Because no samples were taken for pyrene analysis during the incubation, no data are available for a kinetic analysis of the pyrene degradation in the biofilm. Thus, the maximum pyrene removal rate by the biofilm is likely greater than the above value.
Pyrene degradation by both pure and mixed cultures has been reported before [6,24,25]. However, to our knowledge, no previous report shows biofilm growth on pyrene crystals in liquid media. In this study, no carbon source other than pyrene was available, which may have favored biofilm formation. A high pyrene removal rate was observed in some cases where the substrate or microorganisms were sorbed onto carrier material such as beads [7,10] or in a non-aqueous phase liquid . The reason for increased degradation in these cases is likely efficient contact between cell surfaces and substrates . Some microorganisms are also known to produce surfactants [2,28], which in turn may help to solubilize hydrophobic substrates and enhance their transport into the cell. The latter mechanism is presumably even more efficient if the cells have direct contact with the compounds, instead of using only the dissolved phase.
4.2Bacterial community in the biofilm
Most of the clones in the library are related to strains with physiological characteristics that would be adaptive in the biofilm. The most abundant clone in the library (M7) is most closely related to a cold-adapted organism, Polaromonas vacuolata 34-P from Antarctic sea ice . The second most abundant phylotype is closely related to a Sphingomonas sp. S213 isolated from soil (unpublished, GenBank entry dbj:AB018439) and a Sphingomonas sp. PC5.28 from a cooling coil fins in an air-handling system (unpublished, GenBank entry emb:SS16SPC58). Two other clones from the biofilm (M15 and M16) are most closely related to hydrocarbon degraders, Alcaligenes sp. NKNTAU that uses alkane sulfonates under nitrate reducing conditions  and Variovorax paradoxus E4C that uses aromatic compounds (unpublished, GenBank entry gb:AF209469.1). Another clone from the biofilm (M6) is most closely related to Caulobacter sp. type IV CB35 . Caulobacter has been shown to form biofilms during growth before .
Some or all of the major members of the biofilm community appear to be diverse members of the α- and β-Proteobacteria. Because of potential biases in DNA extraction, amplification and cloning, one cannot assume that relative abundance of clones in the library reflects the relative abundance of the corresponding organisms in the biofilm. It is also possible that some biofilm members were not detected by RIS-rDNA analysis. However, it is likely that the clones represent relatively abundant members of the biofilm community. The RIS-rDNA clones appear to represent all of the major bands in the biofilm community banding pattern (Fig. 2), suggesting that abundant phylotypes in the library were not lost during the cloning step.
It is possible that multiple rrn operons can cause one organism to yield multiple RIS-rDNA bands. This appeared to be the case when RIS-rDNA was amplified from the isolates (Fig. 2). However, each of the RIS-rDNA RFLP phylotypes appeared to represent a distinct organism, since each of the RIS-rDNA RFLP phylotypes had a distinct partial rDNA sequence.
When biofilm from pyrene was used to inoculate phenanthrene crystals, the resulting biofilm yielded a RIS-rDNA banding pattern slightly different than that of the pyrene biofilm (Fig. 2). It appears that the two biofilm communities shared some of the major populations, but the phenanthrene biofilm appears to have included at least one major population not present in the pyrene biofilm.
The isolates from the pyrene biofilm were distinct from the biofilm community members identified by RIS-rDNA analysis (Fig. 2). The isolates were members of the fluorescent Pseudomonas group within the γ-Proteobacteria, which are commonly isolated by plating techniques using sole carbon sources, including PAHs. The RIS-rDNA amplicons of the pure isolates did not match any of those detected in the biofilm. This suggests that the α- and β-Proteobacteria in the pyrene biofilm are more difficult to cultivate than the γ-Proteobacteria that were isolated. Since RIS-rDNA amplicons were readily obtained from the isolates, it appears that the isolates are present in the biofilm or surrounding medium at relatively low abundance.
This work was supported by a Strategic Grant from the Natural Science and Engineering Council of Canada.