To prepare the inoculum for this study, produced water was obtained from five production wells (PW; 4-PW, 7-PW, 18-PW, 32-PW, 33-PW; Figure S1) in the Medicine Hat Glauconitic C (MHGC) field (Voordouw et al., 2009). Although this oilfield has been used as a study site for nitrate treatment of souring for 10 years (Suri et al., 2017), previous investigations of this oilfield have reported chemical and functional evidence of nitrate-free “zones” harboring active methanogenic archaea (Agrawal et al., 2012), from which all PW samples were collected for this study. Berdugo-Clavijo and Gieg (2014) also demonstrated that methanogenic crude oil biodegradation could be established from these produced waters. No detectable amounts of nitrate or sulfate, and only low concentrations of sulfide (0.05 mM), nitrite (0.01 mM), and ammonium (0.39 mM) were measured in the produced water samples (methods described in Voordouw et al., 2009). Therefore, we determined that the overall risk of nitrate/nitrite inhibition of methanogenesis was low and that PW samples could be used to study methanogenic crude oil biodegradation.

Samples were collected on May 20th, 2015 in 1-L Nalgene bottles filled to the brim to minimize oxygen ingress during transportation. Upon arrival in the lab, samples were stored in an anaerobic chamber (10% CO₂/90% N₂) at room temperature. Most samples contained 5–10% heavy oil (v/v), which was separated from the produced water by centrifugation (25,000 × g_av for 20 min). Recovered oil samples were combined and stored anaerobically at 4°C prior to use. The crude oil was not sterilized before use, but this appeared to have no effect on control (sterile) incubations established in this study as no methane was produced from sterile controls.

Oil-free produced water from all five PW samples was combined in equal ratios and 500 mL aliquots were dispensed into five sterile, custom-made glass vessels. Each bottle was outfitted with two Balch tube ports; one near the neck of a 1-L Schott flask and the second near its base (in the aqueous phase); the neck of the bottle was also sealed in glass to create an air-tight container (Figure S2). This design allowed for routine sampling of the headspace and culture fluids, respectively, without disturbing the oil-water transition zone. Cultures were amended with either a light oil (°API = 32) or MHGC heavy oil (°API = 16) in excess (20 mL) before sealing microcosms with butyl rubber stoppers and aluminum crimps. The intent of our experiment was not to completely biodegrade all susceptible crude oil components, but rather to observe the microorganisms and processes responding to amendment with either oil source. Cultures were incubated at MHGC reservoir temperatures (30°C) under dark and static conditions for 17 months. Note here that no supplemental growth medium, reducing agents or other culturing agents were added to cultures so as best to simulate the minimal nutrient availability of this particular reservoir. Due to the limited volumes of PW fluids collected during the sampling trip, only one live replicate was prepared per oil type so as to ensure that adequate sterile (autoclaved) replicates (one for each oil type) and an oil-free control (incubated for 21 months) could also be established.

During the incubation period, microcosm headspaces were routinely monitored for methane production by gas chromatography (GC; Fowler et al., 2012). Additionally, cultures were subsampled in duplicate (2 × 25 mL) at designated time points (after 1, 2, 4, 8, 12, and 17 months) for chemical (metabolites) and DNA analyses. Following a centrifugation step (30,000 × g_av), supernatants were acidified with 6 M HCl (pH < 2) and extracted and analyzed for putative hydrocarbon metabolites as silylated compounds by gas chromatography-mass spectrometry (GC-MS) following the procedure outlined by Berdugo-Clavijo and Gieg (2014). Putative hydrocarbon metabolites from silylated organic extracts were positively identified using MSD ChemStation software (version E.02.02.1431; Agilent Technologies) and by matching GC retention times and MS profiles to authentic standards purchased from Sigma Aldrich (97 – ≥99.5% purity) or that were chemically synthesized. Alkyl-substituted benzylsuccinic acids were made using the reflux reaction procedure described by Bickford et al. (1948), while an authentic standard of n-octylsuccinic acid was prepared by base hydrolysis (Kropp et al., 2000). Calibration curves of representative TMS-derivatized standards were used to quantify detectable (≥10 nM) hydrocarbon metabolites of interest (Berdugo-Clavijo and Gieg, 2014).

End point crude oil subsamples (1 mL) were collected from oil-amended cultures and sterile controls for hydrocarbon analysis. Though quantifying the extent of crude oil biodegradation was not a goal of our experiment, we were still interested in screening samples for any measurable hydrocarbon losses. Crude oil samples were diluted 1:10 in dichloromethane and analyzed in triplicate by GC-MS as described by Berdugo-Clavijo and Gieg (2014). Hydrocarbon loss was determined as a function of alkane or aromatic hydrocarbon to pristane or phenanthrene peak area ratios, respectively, as these components naturally present in both crude oils were deemed to be recalcitrant to degradation.

Pelleted cells from subsampled culture fluids were extracted for genomic DNA using the FastDNA SPIN Kit for Soils (MP Biomedicals) and normalized to a concentration of 0.5 ng/μL. Each DNA extract was then probed for the presence of anaerobic hydrocarbon activation genes using a series of established primer sets and thermocycling conditions (Table 1). All PCR reactions (25 μL) were prepared with 12.5 μL 2x Master Mix (Fermentas; Thermo Fisher Scientific), 0.5 μL of each forward primer and corresponding reverse primer (10 μM), and 1 μL of template DNA. Purified amplicons of expected size were cloned using a commercial kit (TOPO TA; Thermo Fisher Scientific) and sent to Eurofins Genomics (Eurofins MWG Operon LLC, Huntsville, AL, USA) for Sanger sequencing. Trimmed consensus sequences of correct identity were queried against the NCBI non-redundant nucleotide database using BLASTn to identify homology to known sequences, and used to build bootstrapped maximum likelihood trees (500 replicates) in MEGA7 (Kumar et al., 2016). Retrieved hydrocarbon activation gene sequences were deposited in GenBank and are available under the accession numbers MG460804–MG460820.

To capture and quantify as many fumarate addition genes (in the produced water cultures) as possible in a single assay, non-degenerate primers MHGC_bssAf (GACGACGGCTGCATGGA) and MHGC_bssAr (GCCTTCCCAGTTGGCGTA) targeting bssA were designed from aligned gene sequences retrieved from this study. Primer specificity was verified in Primer-BLAST (Ye et al., 2012), and PCR/qPCR products obtained using the non-degenerate primers (~708 bp) were cloned and sequenced to confirm that they targeted the correct gene. bssA gene abundance in technical duplicate DNA samples was determined in relation to calibration standards from a 10-fold dilution series (10⁹-10¹ gene copies per μL) of PCR amplified bssA clones obtained in this study. qPCR reactions comprised SsoFast Evagreen Supermix (5 μL), PCR primers (1 μL of 10 pmoles/μL each), RNAse-free water (3 μL), and DNA template (1 μL), and were carried out using a BioRad CFX96 thermocycler as followed; an initial denaturation step (5 min at 94°C), up to 40 cycles of 1 min at 94°C and 1 min at 59.5 °C, and melt curve analysis (65–95°C with an increase of 0.5°C every 5 s). The efficiency in qPCR reactions of primer set MHGC bssA ranged between 95 and 105%, with R²-values for calibration curves >0.99. Reactions yielded a single amplification product, and log gene-abundance values for samples all fell within the linear range of the standard calibration. Gene abundances were also queried for the oil-free and sterile control DNA extracts. A similar procedure was attempted to design qPCR primers for assA, but they failed to amplify the target gene above threshold (quantifiable) levels; only in positive control tests was the expected quantification observed. We experienced the same problem when using established primer sets assA2Fq/assA2Rq described by Aitken et al. (2013).

Amplification and Illumina sequencing of extracted DNA was carried out by a two-step method targeting the V6-V8 regions of the 16S rRNA gene using universal primers 926Fi5 (TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGAAACTYAAKGAATTGACGG) and 1392Ri7 (GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGACGGGCGGTGTGTRC). In the first round of PCR, reactions (25 μL) contained 12.5 μL 2x PCR Master Mix (Fermentas), 5 μL of each primer (1 μM) and 2.5 μL of gDNA template. PCR assays were performed using a three-step thermoprofile previously shown to obtain high amplicon yields from low concentrations of DNA (Klindworth et al., 2013): initial denaturation at 95°C for 5 min; 25 cycles of 95°C (40 s), 55.0°C (2 min), 72.0°C (1 min); final extension at 72°C for 7 min. In the second round of PCR, indices were added to amplicon ends using Nextera XT Index Kit primers (P5-S50X-OHAF and P7-N7XX-OHAF; Illumina). Reaction volumes were increased to 50 μL and contained 25 μL 2x PCR Master Mix, 5 μL of each primer (1 μM) and 10 μL of purified PCR I amplicon. Reaction conditions were modified from round I PCR: 95°C 3 min; 8 cycles of 95°C (30 s), 55.0°C (30 s), 72.0°C (30 s); 72°C 5 min. Amplicons of expected length were confirmed on a 1% agarose gel, purified using the Agencourt AMPure XP magnetic bead system (Beckman Coulter), and sequenced using the 300PE (paired-end) MiSeq protocol at the Department of Biological Sciences, University of Calgary.

Read assembly and 16S rRNA gene sequencing analysis were performed in QIIME (version 1.9.1; Caporaso et al., 2010). Prior to assembly, reads were visually inspected in DADA2 (Callahan et al., 2016) and trimmed of low-quality ends. Reads were merged with a 50 bp overlap with ≤10% allowed mismatches, and subjected to additional quality control steps such as the removal of chimeras, ambiguities and sequences with an average quality score of less than 20. The final reads were clustered into operational taxonomic units (OTUs) at a 97% species cutoff and classified against the SSU SILVA 119 database (Quast et al., 2013). The average read abundance of technical duplicates was used to profile microbial community compositions over time. Merged sequence reads were deposited to the National Center for Biotechnology Information Short Read Archive (SRA) under BioProject PRJNA417121 and accession numbers SAMN07977243–SAMN07977267.

Methane production was monitored following the establishment of two crude oil-amended incubations prepared from heavy oil reservoir produced water (Figure 1). The incubation amended with light oil produced near identical amounts of CH₄ (1307 μmol) as the heavy oil-amended incubation (1352 μmol) in uniform rates across the 525-day incubation period (0.14 μmol CH₄/day/g of light oil vs. 0.13 μmol CH₄/day/g of heavy oil). The only deviation in CH₄ production was in the first 60 days of incubation, where an apparent lag phase was observed in the light oil incubation.

Over the course of the 17-month incubation period, we identified and quantified various compounds in both oil-amended cultures that were not present in the oil-free and sterile controls (Figure 2; Table S1). The average concentration of hydrocarbon metabolites in the oil-amended cultures peaked between 2 and 4 months of incubation, and included compounds with mass spectral profiles indicative of C₁-C₉ alkane fumarate addition (alkylsuccinates) and aromatic compound biodegradation (e.g., benzoate, toluates, 2-methylnaphthoate, cyclohexane carboxylate). No fumarate addition products for aromatic hydrocarbons could be detected in either culture at any time point. Generally, the concentration of aromatic acids detected in the light oil-amended culture was greater than in the heavy oil-amended culture (Figure 2). Trace amounts of C₁-C₄ alkylsuccinates (0.07–0.44 μM) were also predominantly detected in the light oil culture (Figures 2, 3). Interestingly, we identified a total of four peaks in the light oil-amended culture with MS fragment ions corresponding to propane or butane fumarate addition products (two each), suggesting that hydrocarbon activation was occurring at both the primary and secondary carbon atom (Figure 3; Kniemeyer et al., 2007). The mass spectral pattern of the putative n-propylsuccinate aligns with a previously published reference standard (Savage et al., 2010), but does require verification with an authentic standard in our laboratory. At T₈ and T₁₂, the concentration of most alkylsuccinates was below detectable limits (≤10 nM), and cumulative aromatic/other hydrocarbon metabolites also decreased to 2.6–7.6 μM; up to 52 times less than concentrations observed at T₂ and T₄ (Figure 2). Select metabolites were again detected by the end of the incubation period.

Triplicate oil samples from each oil-amended microcosm were recovered and assessed by GC-MS analysis after 17 months of incubation. Measurable losses of C₇-C₈ n-alkanes (47–79%), cyclohexane (22–71%), and some alkyl-substituted aromatic hydrocarbons were observed in oil-amended cultures as compared to sterile controls (Figure S3). The loss of these hydrocarbons corresponded well with hydrocarbon metabolites detected in the organic extracts (Figure 2). Other hydrocarbons detected in the incubations did not decrease in abundance relative to their corresponding sterile control (Figure S3), which we attribute to amending microcosms with an excess of oil.

Twenty established primer sets targeting anaerobic hydrocarbon activation genes were used to probe genomic DNA from the light- and heavy oil-amended cultures. Of these, 11 sets of primers could amplify assA or bssA gene fragments present in one or both cultures in at least one time point sample (Table 1), with the most amplification observed using primer sets 7772f/8546r (Winderl et al., 2007) and FAE-B (von Netzer et al., 2013). Purified and cloned amplicons of target genes were sequenced to determine the taxonomic affiliation of each PCR product and to assess the diversity of putative hydrocarbon degraders in each culture. In all, 3 unique assA and 14 bssA gene fragments were retrieved across both oil-degrading microcosms; PCR products for polycyclic aromatic hydrocarbon (PAH) activation genes (nms encoding for naphthylmethylsuccinate synthase and ncr encoding for naphthyl-coenzyme A reductase) were not obtained at any time point.

Maximum likelihood trees of the recovered fumarate addition gene fragments revealed that all assA gene sequences clustered within a Smithella subclade predominantly enriched from alkane and/or crude oil substrates, whereas bssA sequences were distributed within a largely uncharacterized clade (Figure 4). The bssA clade was phylogenetically distinct from published fumarate addition gene sequences belonging to cultured aromatic hydrocarbon degraders (<77% sequence similarity), thus we assessed their taxonomic affiliations to previously characterized enrichment cultures and environmental strains. We found that the bssA gene fragments retrieved from the oil-degrading cultures shared 90–91% sequence similarity to a presumed Desulfotomaculum sp. previously recovered from a methanogenic toluene-degrading enrichment culture (Bacterium bssA-1; Edwards and Grbić-Galić, 1994; Washer and Edwards, 2007). Our bssA gene fragments also clustered closely (92 – 99% sequence similarity) to several uncultured prokaryotic clones recovered from the Mildred Lake Settling Basin, a tailings pond located in Northern Alberta, Canada, from which materials were used to establish a series of hydrocarbon-degrading enrichment cultures (e.g., Siddique et al., 2006, 2007, 2011; Tan et al., 2013, 2015a).

To gain a better understanding of assA and bssA activity in the oil-degrading cultures, fumarate addition gene abundances were estimated over time by qPCR analysis. Though we were able to detect assA genes in the previously mentioned PCR assays, we were unable to quantify them using published qPCR primers (Aitken et al., 2013) and experimental primers; only positive control tests yielded a quantifiable amplicon product. Consequently, changes in assA gene abundance over time could not be evaluated. It is not known whether the abundance of assA is indeed below detectable limits (<10¹ gene copies/μL), or if the primers selected failed to capture all phylotypes present in the DNA extracts.

In contrast, bssA gene fragments were successfully quantified over the 17-month incubation period. Gene abundances were below detectable limits at T₀, but became enriched after only 1 month of incubation (Figure 5). In the light oil culture, bssA gene abundances increased > 25-fold between T₄ and T₈, and continued to increase up to a maximum of 6.92 × 10⁵ copies/μL (Figure 5A). While bssA gene abundances in the heavy oil culture also increased during the first 8 months of incubation, values decreased by 70% over the remaining incubation period (Figure 5B).

Time-resolved sequencing of 16S rRNA gene fragments was carried out to compare microbial community dynamics in response to amendment with light and heavy oil (Figure 6, Table 2). Both the light and heavy oil cultures saw substantial shifts in microbial community composition during the 17-month incubation under simulated reservoir conditions: many of these shifts were similar across each culture, but distinctions were also evident. The most apparent increase in abundance over time was seen within the Firmicutes and Euryarchaeota phyla, comprising 60 and 74% of quality sequence reads in both oil-amended cultures by T₁₇, respectively (Figure 6). This represented a 20- to 25-fold increase in abundance from reported T₀ values, which collectively had comprised less than 3% of total reads (Figure 6). Members of the Firmicutes were dominated by the enrichment of a single OTU affiliated with Desulfotomaculum after T₄, making up to 30.5% of reads by the end of the 17-month incubation period (Table 2). This OTU shared >97% sequence similarity to an uncultured Peptococcaceae bacterium clone recovered from the Mildred Lake Settling Basin (accession no. EU22655; Siddique et al., 2012); a bacterium thought to participate in methanogenic hydrocarbon biodegradation. This trend also closely mirrored the increase in bssA gene abundances over time, particularly in the light oil-amended culture (Figure 5). Thus, it is plausible that (at least some of) the bssA gene sequences retrieved from our cultures belong to Desulfotomaculum and/or the Peptococcaceae family. Other members of the Firmicutes were also enriched (though to a lesser extent) during the first 4 months of incubation, including Dethiosulfatibacter (up to 8.8% of light oil reads) and Moorella (up to 10.0% of heavy oil reads), but their relative abundance decreased to <0.1% immediately afterwards (Table 2). Methanogenic Euryarchaeota became enriched after the first month of incubation and proliferated up to 39.1–43.3% of total reads by T₁₇ (Figure 6), including hydrogenotrophic (Methanocalculus, Methanoculleus, and Methanolinea) and acetotrophic (Methanosaeta) representatives (Table 2). Though the ratios of methanogens are similar between both cultures, a greater proportion of reads belonging to Methanocalculus were enriched over time in light oil-amended samples, whereas Methanolinea was up to three times as prevalent in the presence of heavy oil at T₁₇. Other putative hydrocarbon fermenters and/or hydrocarbon degradation-associated bacteria were detected during the incubation period, including microorganisms affiliated with the Peptococcaecae, Anaerolineaceae, or Syntrophaceae families (averaging 2.5% of reads; Table 2). We also detected an OTU belonging to Smithella (up to 6.1% of reads), but unlike Desulfotomaculum, its read abundance declined after 4 months of incubation. Several of the T₀ reservoir-associated OTUs, consisting primarily of Proteobacteria, were reduced to <0.5% of reads after just the first month of incubation (includes members of the Alpha-, Beta-, and Gammaproteobacteria), while others saw a more gradual decrease in abundance over time (e.g., Deltaproteobacteria, Deferribacteres; Figure 6; Table 2).

Other OTUs predominantly affiliated with known protein/or organic acid scavengers (e.g., Pelobacter, Bacteroidales) saw a rapid increase in abundance (up to 39.4% of reads) at T₁, and immediately declining afterwards (Table 2). Interestingly, three candidate phyla were enriched at different times during the incubation period. Most notably, the proliferation of “Atribacteria” (formerly OP9/JS1) paralleled that of Desulfotomaculum, increasing by up to 7.0–9.1% over 17 months (Figure 6; Table 2). In contrast, the enrichment of “Cloacimonetes” (WWE1) peaked near T₈, and “Marinimicrobia” (SAR406) remained between 1.1 and 5.1% abundance after T₀ (Figure 6; Table 2).

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.