The West-In-Pit (WIP) of Syncrude Canada, Ltd, established in 1997, was primarily used as a storage facility for fluid fine tailings and recycle water supply before its closure in December 2012 and repurposing as the first model End-Pit Lake for the oil sands industry in Canada (Syncrude, 2012). Before closure, the pond stored approximately 200 Mm³ of fluid fine tailings, capped with a minimum 5-m-deep layer (30 Mm³) of OSPW. Tailings porewater from this operation has previously been noted to contain 2,600 ng L⁻¹ of total PAH, including 101 ng L⁻¹ (780 nM) naphthalene (Madill et al., 1999). The OSPW used in this study was surface water (0–10 cm) sampled in August 2011. The chemical composition of the OSPW and the water sampling methods were described previously (Saidi-Mehrabad et al., 2013). Throughout this article, OSPW sampled from the WIP tailings pond is termed WIP-OSPW.

Potential rates of benzene and naphthalene oxidation (i.e., rates with excess substrate added) were measured in 20 mL (benzene) or 40 mL (naphthalene) amounts of OSPW in 120-mL serum vials, which were sealed with butyl rubber stoppers (20 mm). Vials containing MilliQ water were used as abiotic controls, and vials containing unamended OSPW were used as controls with no substrate addition. There is a small amount of hydrocarbon and other substrates in OSPW, therefore a slow CO₂ increase is expected. Hydrocarbon-amended samples, unfiltered OSPW controls, and MilliQ water controls were run in triplicates.

For benzene oxidation, 1 μl (= 11 μmol) of benzene (Sigma-Aldrich, St Louis, MO, USA) was added to 20 mL OSPW. Using a Henry's law constant of 0.18 M atm⁻¹, the initial benzene concentration was 0.30 mM in the liquid phase and 0.075 mmol L⁻¹ in the gas phase. For naphthalene degradation, a 4.0 mM naphthalene (Sigma-Aldrich) stock was prepared in a 20-mL inert non-degradable carrier 2, 2, 4, 4, 6, 8, 8-heptamethylnonane (HMN; Sigma-Aldrich) and 10-mL aliquots were added to 40-mL OSPW sample volumes. The concentration of naphthalene in HMN is well above the saturation point of naphthalene in water (about 0.24 mM), and therefore it is assumed that naphthalene is saturating, and that calculations of naphthalene degradation in the system need only account for the HMN phase.

Bottles were incubated at 20°C on a rotary shaker at 180 rpm to ensure aeration, for 23 days (benzene) or 14 days (naphthalene). At regular intervals, benzene and naphthalene were measured by gas chromatography as previously described (Berdugo-Clavijo et al., 2012; Fowler et al., 2012). Oxidation rates were calculated by linear regression of benzene and naphthalene depletion over time. Headspace CO₂ and O₂ were monitored by injection of gas samples into a Varian 450-GC gas chromatograph equipped with a thermal conductivity detector (150°C), after separation in a 2 mm × 0.5 m Hayesep N and a 2 mm × 1.2 m Molecular Sieve 16X column in series (70°C). CO₂ production rates were calculated by linear regression of CO₂ production vs. time.

To identify microbial communities actively assimilating the model aromatic compounds, isotopically fully-labeled benzene (¹³C₆H₆ 99 atom%, Sigma-Aldrich) and naphthalene (¹³C₁₀H₈ 99 atom%) were used in DNA-SIP. To concentrate bacteria, 150 mL of OSPW were filtered through a 0.2-μm polysulfone filter (Pall Life Sciences, East Hills, NY, USA). The filter and 20 mL of OSPW were added to 100-mL serum vials and supplemented with 5 μM of labeled benzene or naphthalene. Vials containing 20 mL of filtered OSPW without added substrate were used as controls. Headspace CO₂ was monitored for all SIP-enrichments and incubations were terminated when CO₂ levels in the microcosms increased by around 1.0 μmol of CO₂. After experimenting with different incubation times (3, 7, 9, and 14 days, data not shown), we chose the earliest times at which a shift toward heavy density fraction was visible in CsCl gradients, indicating ¹³C incorporation into microbial genomes. Shifts were visible after 7 days (naphthalene) and 9 days (benzene), of incubation. Cells were collected by centrifugation for 10 min at 10,000 × g. DNA was extracted, purified, separated by isopycnic centrifugation in CsCl, and fractionated as described previously (Sharp et al., 2012). DNA extracted from OSPW incubated for the same time without addition of substrates was used as a control to determine the expected position of unlabeled DNA in CsCl gradients. Heavy ¹³C-labeled DNA fractions of labeled samples were selected as shown in Supplementary Figure S1. The corresponding highest-density fractions of OSPW receiving the same treatments (OSPW-heavy), and unfractionated DNA from fresh OSPW (OSPW-control) were used as controls.

Amplification of 16S rRNA genes, sequencing on a Roche 454 GS FLX, and analysis using QIIME were all carried out as described previously (Sharp et al., 2014). Briefly, 16S rRNA genes were amplified from the gradient fractions using FLX Titanium amplicon primers 454T_RA_X and 454T_F containing the 16S rRNA gene targeted primers 926f and 1392r at their 3′- ends, along with adaptors necessary for the Roche Titanium chemistry (Sharp et al., 2014). The QIIME software platform was used to analyze the sequences (Caporaso et al., 2010). Low-quality sequences were removed based on a minimum quality score of 25, Operational Taxonomic Units (OTUs) clustered based on 97% sequence identity, chimeras removed via ChimeraSlayer, and sequences classified via BLAST against the Greengenes database. Taxonomic identifications were verified by manual BLAST of representative OTU sequences against the NCBI database.

Metagenomic DNA from OSPW was prepared and extracted as described previously (Saidi-Mehrabad et al., 2013). DNA concentration was determined with a Qubit Fluorometer using a Quant-iT dsDNA HS Assay Kit (Invitrogen, Carlsbad, CA, USA). The DNA (12 μg) was sequenced using a combination of Roche 454 GS FLX and paired-end Illumina HiSeq2000 platforms at the Genome Quebec and McGill University Innovation Centre, Montreal, Quebec. Quality control and assembly of metagenomic data was performed at the University of Calgary Visual Genomics Center, Calgary, Canada according to previous reports (Saidi-Mehrabad et al., 2013; Tan et al., 2013; Aguilar et al., 2016). Annotation was conducted by submission to the Joint Genome Institute IMG platform (Markowitz et al., 2012).

Unassembled read based analysis was performed on the Illumina HiSeq2000 output due to the higher sequencing depth, although similar results were obtained when 454 reads were used. Clipping of the Illumina adapters, quality-trimming and size-filtering for the raw reads was performed using a BBDuk tool (trimk = 27, trimq = 20; minlen = 80, http://jgi.doe.gov/data-and-tools/bbtools/). Curated databases for key genes encoding enzymes for aerobic benzene and naphthalene degradation were constructed in a two-step process. First, core datasets for each gene were created by retrieving DNA sequences from NCBI Gene database using Enzyme Commission numbers and gene names. Next, the core datasets were extended by querying their entries using BLASTN against the NCBI non-redundant and environmental samples databases (Coordinators, 2013). Sequences with nucleotide identities of 70% and above were added to the corresponding datasets. Processed metagenomic sequences were then queried against the reference datasets using BLASTN. Reads with minimum nucleotide sequence identities of 50% and alignment lengths of 75 bp were recruited. Custom python and bash scripts were created to run BLASTN, process the output, and quantify and recruit reads matching the specified criteria. Scripts are available at the GitHub repository (release v1.14.3, https://github.com/dunfieldlab/mg_wrapser). Recruited reads, corresponding to each dataset, were then queried by BLASTN against the NCBI non-redundant database to validate functional identity and processed with MEtaGenome ANalyzer (MEGAN v.5.11.3) for taxonomic assignment (Huson et al., 2007).

OSPW samples were initially enriched for 14 days with either 5 mM of benzene or naphthalene added (99.9%, Sigma-Aldrich). To cultivate bacterial strains from these enrichments, both complex media targeting versatile chemotrophs and basal mineral salts media containing the model hydrocarbons as sole carbon and energy sources were used. Enriched OSPW was diluted 1 : 10 with MilliQ water and 100-μl aliquots spread on 20% strength R2A (20R2A) plates adjusted to pH 8 with NaOH, as well as on a modification of the basal mineral salts medium M10 (Reasoner and Geldreich, 1985). M10 was prepared as described previously (Heyer et al., 2002), except that agar was substituted with 1.5% phytagel (Sigma-Aldrich), and MilliQ water was substituted with 0.2-μm-filtered OSPW to reproduce the natural conditions of tailings water. We named these modified media containing filtered tailings water 20R2A-T and M10-T. To specifically grow naphthalene-degrading microorganisms, a 10-fold serial dilution of the initial enrichment was applied (1 mL enrichment sample and 9 mL MilliQ water). Dilutions up to (10⁻¹⁰) were plated onto solid medium M10-T prepared in 125-mL wide-mouth jars (I-Chem, VWR, Radnor, PA, USA). These growth vessels were inverted and 3 g of naphthalene placed on the lids to diffuse naphthalene vapor as previously described (Jeon et al., 2004). For benzene-degrading microorganisms, OSPW samples enriched with 5 mM benzene were serially diluted as above, spread onto 20R2A-T in wide-mouth jars, and incubated with the addition of 0.1 mL of benzene per jar at 20°C for 60–70 days.

Colonies were picked and streaked onto new plates of the same medium until pure colonies were obtained. Colony PCR was performed to screen the isolates. Single colonies were picked with autoclaved toothpicks and frozen at –80°C overnight in 0.2-mL tubes to lyse cells. 16S rRNA genes were amplified using the universal primer set 9f and 1492b (Weisburg et al., 1991). PCR reaction conditions were: initial denaturation at 94°C for 10 min, followed by 35 cycles of 1 min at 94°C, 1 min at 56°C and 2 min at 72°C, and a 10-min final elongation at 72°C. PCR products were visualized on a 1% agarose gel and purified with an EZ-10 Spin Column PCR Purification Kit (BioBasic Inc., Markham, ON, Canada). DNA concentration was determined as described above. Sanger sequencing was performed at the University of Calgary Core DNA Services using Applied Biosystems 3730xl (96 capillary) genetic analyzer (Applied Biosystems, Foster City, CA, USA). Forward and reverse sequences were merged via EMBOSS 6.3.1 merger (Néron et al., 2009) and >1,300 bp sequences were identified via BLASTN (Altschul et al., 1997) against the EzTaxon server (Kim et al., 2012).

A number of isolates were tested for their benzene degradation. Exponentially growing bacteria in 20R2A-T medium (pH 8.5) were transferred to 20R2A liquid medium (20 mL in a 100-mL serum vial), and sealed with butyl rubber stoppers. Benzene (6 μM) was added to the cultures. Each sample was run in triplicate. Loss of benzene was measured in duplicate pure cultures as described above. Growth was monitored by measuring OD₆₀₀ on an Ultrospec 10 Cell Density Meter (Amersham Biosciences, Piscataway, NJ).

The benzene and naphthalene oxidation rates measured were roughly linear over time (Supplementary Figures S2, S3). Therefore we assume 0-order kinetics (i.e., the enzymes are saturated and the rates reported are V_max values not dependent on substrate concentration). The potential benzene oxidation rate was 4.3 μmol L⁻¹ OSPW d⁻¹ in the first 23 d, and the naphthalene oxidation rate was 21.4 μmol L⁻¹ OSPW d⁻¹ over 14 d (Supplementary Figures S2, S3). Conditions remained oxic throughout the incubations (the maximum O₂ decline was from 21% to 17% v/v). In the benzene enrichment 5.6 μmol of CO₂ L⁻¹ OSPW d⁻¹ was produced, 1.3 times of the expected CO₂ production based on the reaction: C₆H₆ + 2.5O₂ + NH₃ → C₅H₇O₂N + CO₂ + H₂O (Shuler and Kargi, 2002). In the naphthalene enriched samples, 144.1 μmol of CO₂ L⁻¹ OSPW d⁻¹ was produced, or 1.3 times of the expected CO₂ production based on the reaction: C₁₀H₈ + 7O₂ + NH₃ → C₅H₇O₂N + 5CO₂ + 2H₂O (Shuler and Kargi, 2002). The slightly higher than expected CO₂ production indicates that less C is assimilated from these compounds than predicted by simple models, or that added benzene and naphthalene enhance the degradation of other hydrocarbon substrates available in OSPW (Suthersan and McDonough, 1996).

Density gradient separation of extracted DNA after 9 days (benzene) or 7 days (naphthalene) of incubation indicated that ¹³C-substrate incubations led to a small shift in the DNA distribution toward heavier densities (Supplementary Figure S1). The positions of the “heavy” DNA fractions representing the active ¹³C-assimilating communities were determined by comparing density gradients of the DNA extracted from the SIP incubations to that of the control samples, as indicated in Supplementary Figure S1.

Community analysis of 16S rRNA genes amplified from OSPW DNA prior to the SIP incubations showed both methanogenic archaea and diverse bacteria. The combination of anaerobes, aerobes, and facultative anaerobes has been noted before (An et al., 2013; Saidi-Mehrabad et al., 2013) and is probably the result of active water circulation between anoxic and oxic water zones. The dominant classes/phyla of bacteria were Betaproteobacteria, Alphaproteobacteria, and Flavobacteria (Supplementary Figure S4). The predominant genera were Hydrogenophaga, Luteibacter, and Enhydrobacter (Figure 1). After fractionation of this DNA extract in a CsCl gradient, generally the highest density from which a PCR product could be obtained was 1.74–1.75 g/mL, which corresponded to the expected peak of the ¹³C-labeled DNA in the SIP incubations. The community detected in this heaviest fraction of unincubated OSPW was much simpler, and dominated by the genera Devosia, Methylomonas, Thauera, and Brevundimonas (Figure 1). This OSPW-heavy community is a simplified subset of the OSPW-control community (Figure 1, Supplementary Table S1), suggesting that a few species have high G+C contents and therefore high DNA densities. This is supported by reported G+C contents for isolates of these genera: which range from 47 to 58% for Methylomomas (Hoefman et al., 2014), 64–6% for Devosia (Nakagawa et al., 1996), and 66–69% for Thauera (Mechichi et al., 2002).

Communities detected in the heavy DNA fractions of the SIP incubations were compared to two controls: the complete native OSPW community and the native community only in the 1.74–1.75 g/mL fraction, to verify that the species detected by SIP were truly enriched with ¹³C. Only OTUs that were enriched at least 10-fold in the SIP heavy fractions compared to both controls, or OTUs that were enriched at least 2-fold and were predominant at >10% of the total DNA in the heavy fraction, were considered to be enriched in the added ¹³C-substrate.

The ¹³C-labeled, heavy DNA fraction from the benzene-SIP experiment was dominated by a close relative of Methyloversatilis universalis (Betaproteobacteria) that accounted for 86.9% of the community and was enriched 20-fold compared to the OSPW-heavy control (Figure 1, Supplementary Table S1). Zavarzinia (Alphaproteobacteria) was the second most predominant genus, enriched 26-fold compared to OSPW-heavy controls. This result was reproducible: in a second SIP experiment of ¹³C-benzene incubated for 14 d Methyloversatilis was also predominant in the heavy DNA fraction (Supplementary Table S2). A SIP with ¹³C-methanol also verified the methylotrophic phenotype of this OTU (data not shown).

The three most dominant taxa in the ¹³C-naphthalene SIP heavy fraction were identified as Chromatiaceae, Thauera, and Pseudomonas. Chromatiaceae and Pseudomonas OTUs were enriched >25-fold compared to the OSPW-heavy control (Supplementary Table S1). The predominant Thauera OTU was enriched only 2-fold, but was the second most abundant OTU in this fraction. Although this Thauera OTU probably has a high G+C content and is therefore also abundant in the OSPW-heavy control, we conclude that its 2x enrichment in the benzene-heavy fraction may indicate a labeling of this organism by ¹³C for 2 reasons: (i) the absolute amount of DNA in the ¹³C-naphthalene SIP heavy fraction is much greater than the amount of DNA in the control OSPW-heavy fraction (31.6 vs. 0.17 ng/μl), indicating that this fraction is mostly newly synthesized DNA; and (ii) considering only OTUs that are many times enriched is a good approach for less common OTUs but may overlook common species, since abundances are relative and therefore the higher the initial abundance, the less it can increase (i.e., a greater than 10-fold increase is impossible for an OTU already present at 10% of the control, as seen for Thauera). Therefore the 2x enrichment of this OTU in the heavy fraction, combined with its predominance and the obvious increase in total DNA in the 1.74–1.75 fraction of the naphthalene-SIP, likely indicate that Thauera also metabolized the naphthalene.

An overview of the OSPW metagenome is given in the Supplementary Table S3. A predicted 7.5% of the annotated genes belonged to the KEGG category “xenobiotics biodegradation and metabolism.” The metagenome shows the genetic machinery for aerobic degradation of diverse petroleum hydrocarbons. Aromatic and polyaromatic compounds are predicted to be broken down via three different aerobic pathways: dioxygenase and dehydrogenase reactions, ring removal from polycyclic aromatic ring structures, and two-monooxygenase reactions (Figure 2). Each process forms catechol as an intermediate, which is broken down through the catechol meta-cleavage pathway into Acetyl-CoA. Benzene is attacked in this scenario by either benzene/toluene dioxygenase, or by a monooxygenase then a phenol hydroxylase, while naphthalene is attacked by dihydroxynaphthalene dioxygenases (Figure 2).

A quantitative analysis of unassembled Illumina metagenome reads for the key genes shown in Figure 2 suggested that these were predominantly found in Betaproteobacteria (particularly Burkholderiales and Rhodocyclales) and Gammaproteobacteria (particularly Pseudomonadales and unidentified orders) and a lesser extent in Alphaproteobacteria and Actinobacteria (Figure 3; finer taxonomic resolution of these results is shown for each gene in the Supplementary Figures S5A–P). The most unusual patterns were obtained for dmpB encoding for catechol 2,3-dioxygenase as well as todC1 encoding benzene/toluene dioxygenase. Both showed a large number of hits to bacteria that could not be identified reliably, suggesting that at the enzyme level and possibly also the taxonomic level uncharacterized species are involved in the biodegradation of aromatics The unassembled Illumina read analysis was also used to calculate the most prevalent taxa within the set of key genes encoding aromatic-compound degradation (Figure 4). Prevalence of a taxon was determined as the proportion of the 14 raw read datasets (i.e., the 14 genes in Figure 2) a taxon was detected in. The taxa identified in Figure 4 are those with >60% of the key genes. They are shown to a higher taxonomic resolution than in Figure 3 because of the relative simplicity of the dataset. The most prevalent taxa were similar to the most abundant taxa, dominated by Burkholderiales, Rhodocyclales, Pseudomonadales.

Two of the bacterial taxa identified as important in the SIP studies, Pseudomonas (Pseudomonadales) and Thauera (Rhodocyclales), show high abundance of some genes (Figure 3 and Supplementary Figure S5) and high prevalence of the overall gene set (Figure 4). The three other taxa identified as important in the SIP studies, Methyloversatilis (Rhodocyclales), Chromatiaceae (Chromatiales), and Zavarzinia (Rhodospirilalles), were neither abundant nor prevalent (Figures 3,4). However, in the case of Methyloversatilis, a large number of genes could be mapped to closely related sister genera within the Rhodocyclales/Rhodocyclaceae (Azoarcus and Thauera) (Figure 4; Supplementary Figures S5). MEGAN is limited by the NCBI database and is unable to resolve taxonomy unambiguously (Huson et al., 2007), so some of these sequences may in fact represent gene content of the important Methyloversatilis OTU. One member of the Chromatiaceae (Chromatiales) was detected as the closest BLAST hit to a catechol 2,3-dioxygenase and a benzene/toluene dioxygenase (data not shown).

Pure cultures of Xanthobacter (strain OSPW1) and Zavarzinia (strain OSPW2) were isolated using 20R2A-T medium (Supplementary Table S4). These were able to perform complete degradation of 6 μM benzene in 14 and 8 days, respectively, when growing on complex 20R2A medium (Supplementary Figure S6). Thauera sp. strain OSPW4 was isolated on complex media incubated under naphthalene vapor, which shows that it tolerates naphthalene, although it may not grow on naphthalene as a sole substrate. Pseudomonas sp. strain OSPW3 was isolated on mineral salts medium M10-T with naphthalene vapor as a sole substrate, and therefore was capable of growth on naphthalene as the sole substrate. No relatives of the two other predominant bacteria detected in ¹³C-benzene and ¹³C-naphthalene SIPs were isolated (i.e., Methyloversatilis and Chromatiaceae).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.