Acetate Production from Oil under Sulfate-Reducing Conditions in Bioreactors Injected with Sulfate and Nitrate

Abstract

Oil production by water injection can cause souring in which sulfate in the injection water is reduced to sulfide by resident sulfate-reducing bacteria (SRB). Sulfate (2 mM) in medium injected at a rate of 1 pore volume per day into upflow bioreactors containing residual heavy oil from the Medicine Hat Glauconitic C field was nearly completely reduced to sulfide, and this was associated with the generation of 3 to 4 mM acetate. Inclusion of 4 mM nitrate inhibited souring for 60 days, after which complete sulfate reduction and associated acetate production were once again observed. Sulfate reduction was permanently inhibited when 100 mM nitrate was injected by the nitrite formed under these conditions. Pulsed injection of 4 or 100 mM nitrate inhibited sulfate reduction temporarily. Sulfate reduction resumed once nitrate injection was stopped and was associated with the production of acetate in all cases. The stoichiometry of acetate formation (3 to 4 mM formed per 2 mM sulfate reduced) is consistent with a mechanism in which oil alkanes and water are metabolized to acetate and hydrogen by fermentative and syntrophic bacteria (K. Zengler et al., Nature 401:266–269, 1999), with the hydrogen being used by SRB to reduce sulfate to sulfide. In support of this model, microbial community analyses by pyrosequencing indicated SRB of the genus Desulfovibrio, which use hydrogen but not acetate as an electron donor for sulfate reduction, to be a major community component. The model explains the high concentrations of acetate that are sometimes found in waters produced from water-injected oil fields.

INTRODUCTION

Oil can be converted under anaerobic conditions in the absence of electron acceptors by consortia of fermentative and syntrophic bacteria and methanogens to methane and CO₂ (1, 2, 3). Water is a major reagent in this reaction. For the conversion of hexadecane, the following reactions have been proposed with acetate and dihydrogen (H₂) as potential intermediates (1, 4):

$$4C_{16}{H}_{34}+64H_{2}O\to 32{CH}_3{COO}^{-}+32H^{+}+68H_2$$ (1)

$$32{CH}_3{COO}^{-}+32H^{+}\to 32{CH}_4+32{CO}_2$$ (2)

$$68H_2+17{CO}_2\to 17{CH}_4+34H_{2}O$$ (3)

$$4C_{16}{H}_{34}+30H_{2}O\to 49{CH}_4+15{CO}_2$$ (4)

These reactions have been proposed to be catalyzed by (i) fermentative and syntrophic bacteria, including deltaproteobacterial syntrophs of the genus Smithella (3), (ii) acetotrophic methanogens of the genus Methanosaeta, and (iii) hydrogenotrophic methanogens. Because water is either available in subsurface oil fields or being injected to produce oil, these reactions can, in principle, continue indefinitely. Also, because the fermentative and syntrophic bacteria prefer low-molecular-weight components, these reactions decrease this fraction, gradually converting light oil into heavy oil (5).

When the injection water contains sulfate, reduction of sulfate to sulfide (souring) using oil organics as an electron donor is catalyzed by sulfate-reducing bacteria (SRB; 2, 6, 7, 8). Souring is widespread in inland and offshore oilfields with low and high down-hole temperatures (8) with SRB growing in a biofilm near the injection well. SRB using toluene and other aromatics, alkanes, or other oil components directly as electron donors for sulfate reduction have been described (9–13). These SRB, for example, Desulfobacula toluolica and Desulfatibacillum alkenivorans, combine hydrocarbon degradation and sulfate reduction activities in a single cell. However, oil degradation under sulfate-reducing conditions catalyzed by fermenting and syntrophic bacteria (reaction 1) and SRB using hydrogen or acetate as an electron donor for sulfate reduction (e.g., the genera Desulfovibrio and Desulfobacter) can also be envisaged (14) but has so far not been demonstrated.

Souring of oil fields is unwanted and can be prevented or reversed by the injection of nitrate, stimulating heterotrophic nitrate-reducing bacteria (hNRB) and sulfide-oxidizing nitrate-reducing bacteria (6, 7, 15–21). Both NRB activities produce nitrite as an intermediate, inhibiting dissimilatory sulfite reductase Dsr, the highly conserved enzyme that produces the sulfide (22–24). In applying this biotechnology (25–28), volatile fatty acids (VFA; e.g., acetate, propionate, and butyrate) are often considered to be the predominant electron donors for nitrate reduction and decisions with respect to the nitrate dose to be used take VFA concentrations in produced waters into account (29). Direct use of oil components, especially toluene, as electron donors for nitrate reduction has also been demonstrated (30, 31, 32). When VFA were used as electron donors in microcosm or bioreactor studies, production of acetate from propionate and butyrate was seen under sulfate-reducing, but not under nitrate-reducing, conditions (33, 34). The aim of the present study was to evaluate sulfide production in bioreactors containing residual heavy oil as the electron donor and carbon source to more closely mimic souring and its control with nitrate in an actual oil field. Specifically, we compared limiting (4 mM) and excess (100 mM) concentrations of nitrate injected either as pulses or continuously. Interestingly, the data indicate that acetate is also a major intermediate in the sulfidogenic metabolism of oil.

MATERIALS AND METHODS

Media and enrichment cultures.

Coleville synthetic brine medium (CSBK) contained, per liter, 1.5 g NaCl, 0.05 g KH₂PO₄, 0.32 g NH₄Cl, 0.21 g CaCl₂ · 2H₂O, 0.54 g MgCl₂ · 5H₂O, and 0.1 g KCl. Addition of 2 mM Na₂SO₄ gave CSBK-S. After autoclaving, 30 ml of 1 M NaHCO₃ and 1 ml of trace elements (35), autoclaved separately, were added and the pH was adjusted to 7.2 to 7.5. An SRB enrichment culture was made by inoculating 5 ml of produced water from the Medicine Hat Glauconitic C (MHGC) field (6) into 100 ml of CSBK with 5 mM sulfate and 5 ml of MHGC oil.

Bioreactor start-up.

Five 60-ml syringe columns (2.5 by 16 cm) were fitted with glass wool at the bottom and then packed with sand with an average grain size of 225 μm (Sigma-Aldrich). The top of each bioreactor contained a layer of polymeric mesh and a perforated rubber stopper through which spent medium flowed through Tygon tubing into an effluent container (Fig. 1A). Dry columns were assembled inside an anaerobic glove bag with an atmosphere of 90% N₂ and 10% CO₂ (Coy Laboratory Products Inc.) after the materials were autoclaved. The columns were flooded with anaerobic CSBK by a multichannel peristaltic pump (Gilson Inc., Minipuls-3, 8-channel head). The average pore volume (PV), determined from the wet weight minus the dry weight of the columns, was 28.30 ± 0.55 ml. Medium-saturated columns were flooded with 1 PV of heavy oil collected from production wells 11-PW, 13-PW, and 15-PW of the MHGC field. An American Petroleum Institute gravity of 16, a density of 0.959 g/ml, and a viscosity of 1,900 cP at 23°C are typical values for this oil. Because of the high oil viscosity, the majority of the medium (1 cP) was displaced by oil. The amount of water recovered was 26.8 ± 0.4 ml (0.947 PV). Oil-saturated columns were then again flooded with anaerobic medium for 8 h at 0.5 to 1.0 ml/min, representing 8.5 to 16 PV. The low medium-to-oil viscosity ratio caused viscous “fingering”: on average, 16.3 ml (0.608 PV) of oil was recovered, indicating that 8.0 ml (0.371 PV) of oil remained in place. Sampling ports were located near the influent and effluent regions (Fig. 1). The columns were inoculated via the influent port with 0.5 PV of the aqueous phase of the SRB enrichment culture. Pumps were turned off, and the culture was allowed to grow batchwise for 1 month prior to the injection of CSBK-S at a flow rate of 0.65 ml/h, giving a retention time of 27 h. The bioreactors were run at ambient temperature (22°C).

Fig 1.

(A) Diagram of the bioreactor setup used in this study. A layer of glass wool (GW) was placed at the bottom, and a polymeric mesh (PM) was placed at the top. Three-way valve ports are indicated above and below the bioreactor columns designated influent (IN) and effluent (EF). (B) Outline of the bioreactor fractions obtained when the bioreactor was dismantled. Fractions F2 to F9 represent the sand pack, whereas F1 and F10 represent the glass wool and polymeric mesh, respectively. The influent port (IN) was also obtained as a fraction. Dist., distance.

Nitrate injection.

Once partial (50%) reduction of sulfate to sulfide in the bioreactor effluent was observed, CSBK-S was amended with 4 or 100 mM nitrate, referred to as CSBK-LN (low nitrate) and CSBK-HN (high nitrate), respectively. Four bioreactors were injected with nitrate, and one was injected without nitrate, as indicated in Table 1. Bioreactors were injected continuously with nitrate for 120 days. Another set of bioreactors was subjected to nitrate pulses by alternately injecting CSBK-S and CSBK-LN or CSBK-S and CSBK-HN for a total of 200 days. The injections of nitrate were made over a period of 3 to 10 days and in some cases as long as 20 days, the flow rate remained the same during these pulses. Samples were drawn from the influent (IN) and effluent (EF) ports every 1 to 5 days for the determination of sulfide, sulfate, nitrate, nitrite, and acetate concentrations. This caused some of the residual oil to be pulled down from the column, staining the end of the inlet tubing. Aqueous sulfide concentrations were determined colorimetrically with N,N-dimethyl-p-phenylenediamine (36), while dissolved nitrate, nitrite, and sulfate were determined from frozen samples (−20°C) by high-pressure liquid chromatography as described elsewhere (33, 37).

Table 1.

Bioreactors used in this study

Bioreactor	Electron donor	Injection strategy
Sa	Heavy oil	Continuous 2 mM SO42−
S-LNa	Heavy oil	Continuous 2 mM SO42−, 4 mM NO3−
S-HNa	Heavy oil	Continuous 2 mM SO42−, 100 mM NO3−
S-LNP	Heavy oil	Continuous 2 mM SO42−, pulsed 4 mM NO3−
S-HNP	Heavy oil	Continuous 2 mM SO42−, pulsed 100 mM NO3−

^aDissected after 117 days of continuous injection.

Dissection of the bioreactor, biomass isolation, and nucleic acid extraction.

After completion of the continuous injections, bioreactors S, S-LN, and S-HN were frozen at −80°C overnight. A hacksaw was used to cut the plastic syringe column and the sand pack core into discrete fractions (Fig. 1B), which were placed in sterile Falcon tubes (Becton, Dickinson Co.). For biomass isolation and nucleic acid and oil extraction, fractions were washed twice with 10 ml of pyrophosphate buffer (100 mM, pH 7.2) on a shaker at 4°C overnight. The biomass-containing fractions were transferred to centrifuge tubes and concentrated by centrifugation at 14,000 rpm for 25 min at 4°C and resuspension in 2 ml buffer. Duplicate 10-μl samples were used for protein determination with a Bio-Rad DC protein assay kit, as outlined by the manufacturer. Genomic DNA (gDNA) was extracted with the FastDNA Spin Kit for Soil (MP Biomedicals, Santa Ana, CA) by using procedures outlined by the manufacturer, including two bead-beating steps. The DNA was resuspended in binding matrix, transferred to SPIN columns, and centrifuged at 14,000 rpm for 1 min, leaving the DNA attached to the binding matrix while the flowthrough contained unpurified RNA. The DNA was eluted with DNase-free water. The total RNA from the flowthrough was purified separately with an AllPrep DNA/RNA minikit (Qiagen). The total RNA was treated with DNase I Amplification Grade (Invitrogen) and then synthesized into cDNA with a qScript Flex cDNA kit (Quanta BioSciences, Inc.).

Functional gene and microbial community analysis.

Quantitative PCR (qPCR) with primer sets V17m/napA-4r and p2060F/4R targeted the napA (periplasmic nitrate reductase) and dsrB (dissimilatory sulfite reductase) genes, respectively (38, 39), with a Corbett Rotor Gene (34). These markers are commonly used to assess nitrate- and sulfate-reducing activities. Standards were prepared from purified PCR products amplified from gDNA extracted from nitrate- and sulfate-reducing enrichment cultures. PCR conditions were as described elsewhere (34). PCR mixtures were prepared in a total volume of 12.5 μl containing 10 pmol of the forward and reverse primers, 6.25 μl of qPCR master mix (Qiagen) with SYBR green chemistry, 4.25 μl of DNase-free water, and 1 to 11 ng of DNA or cDNA template.

Quantification was for triplicate amplifications from software-derived standard lines generated for napA (R², 0.994; efficiency, 1.15; slope, −3.00) and dsrB (R², 0.992; efficiency, 1.09; slope, −3.12) for dilutions of 10² to 10⁸ copies/ml.

For pyrosequencing, the extracted DNA was amplified for 30 cycles with FLX titanium primers 454T_RA_X and 454T_FwB as described elsewhere (31, 40). These have the sequences for 16S primers 926Fw and 1392R as their 3′ ends. Purified 16S amplicons (∼125 ng) were sequenced at the Genome Quebec and McGill University Innovation Centre, Montreal, Quebec, Canada, with a Genome Sequencer FLX instrument and GS FLX Titanium series kit XLR70 (Roche Diagnostics Corporation). Data analysis was conducted with Phoenix 2, a 16S rRNA data analysis pipeline (40).

Community analysis of field samples.

Five samples of MHGC oil field-produced water were used for DNA isolation, PCR, and pyrosequencing as described above. Two of these were from producing wells showing nitrate breakthrough (2-PW and 7-PW), whereas three were from producing wells not showing nitrate breakthrough (3-PW, 4-PW, and 10-PW) at the time of sample collection (31).

Estimation and analysis of residual oil.

Dichloromethane (DCM) was used to extract residual oil from collected fractions. DCM-extracted oil was transferred to preweighed glass vials. The DCM was evaporated under a fume hood for 48 h. The weight of concentrated heavy oil was then determined for each bioreactor fraction. The oil-occupied PV was calculated as the volume of oil collected (milliliters) divided by the total PV (milliliters) of the fraction.

A 1-μl sample of 0.1 g of oil/ml of DCM was analyzed by gas chromatography-mass spectrometry (GC-MS) as described elsewhere (31). The sample was injected with an autoinjector 7683B series (Agilent Technologies, Santa Clara, CA) into a gas chromatograph (7890N series; Agilent) that was connected to an mass spectrometer (5975C inert XL MSD series; Agilent). The gas chromatograph was equipped with an HP-1 fused silica capillary column (length, 50 m; inner diameter, 0.32 mm; film thickness, 0.52 μm; J&W Scientific) with helium as the carrier gas. Standards run for identification purposes were toluene; ethylbenzene; p/m-xylene; o-xylene; 2-ethyltoluene; C₃, C₄, C₅, C₆, C₇, and C₈ n-alkylbenzenes; and C₁₀, C₁₂, and C₁₆ n-alkanes. Peak areas were integrated, and the percent oil degradation of each component was calculated from peak areas of similarly treated control MHGC oil from production wells 11-PW, 13-PW, and 15-PW. The ratio of oil components to the internal standard pristane, which is not biodegraded by anaerobic oil enrichments, was calculated for n-alkanes (C₉ to C₂₉) and alkylbenzenes, respectively (41–43).

Nucleotide sequence accession numbers.

The entire set of raw reads obtained in this study is available from the Sequence Read Archive at NCBI under accession numbers SRR631902 to SRR631908.

RESULTS

Acetate formation during injection with sulfate or sulfate and nitrate.

Use of oil as an electron donor in bioreactor S, injected with CSBK-S containing 2 mM sulfate, gave partial reduction of sulfate at the influent port from day 1 to day 65 (Fig. 2A). From day 66 onward, sulfate reduction was nearly complete, forming 1.66 ± 0.07 mM sulfide with 0.21 ± 0.06 mM sulfate remaining (Fig. 2A). At the effluent port, partial sulfate reduction was observed from day 1 to day 33, whereas 1.79 ± 0.04 mM sulfide was formed with 0.03 ± 0.02 mM sulfate remaining for days 34 to 120 (Fig. 2B). On average, 60% of the injected sulfate was reduced at the bioreactor inlet, with 30% being reduced past this region. Acetate was consistently present from day 66 onward (Fig. 2A and B). Following a peak concentration of 5.0 to 6.5 mM, the concentration plateaued at 3.5 mM and then declined to 0.5 to 1.0 mM.

Fig 2.

Time course analysis of influent (IN) and effluent (EF) concentrations of bioreactor S injected with CSBK-S (2 mM sulfate) (A, B), bioreactor S-LN injected with CSBK-SLN (2 mM sulfate and 4 mM nitrate) (C, D), and bioreactor S-HN injected with CSBK-SHN (2 mM sulfate and 100 mM nitrate) (E, F). Concentrations of sulfate (●), sulfide (○), nitrate (▲), nitrite (△), and acetate (×) are indicated as a function of time. Note that the values for acetate and sulfide are both zero in panels E and F.

Once sulfate reduction was established (between days 1 and 8), 2 mM sulfate and 4 mM nitrate were injected continuously into bioreactor S-LN, causing the sulfide concentration to drop to zero and the sulfate concentration to rise to 2 mM (Fig. 2C and D). Some nitrate was observed at the influent port from day 8 to day 28, whereas at the effluent port, nitrate was observed on days 8 and 12 only (Fig. 2C and D). Nitrite was observed at the influent port from day 8 to day 49 (average of 1.52 mM) and at the effluent port from day 12 to day 49 (average of 0.81 mM). Considering that conversion of nitrate to nitrite and of nitrate to N₂ represent 25 and 100% reduction, respectively, and assuming that the missing N had been reduced to N₂ (37), we deduce that for days 1 to 49, on average, 66% of the injected nitrate was reduced at the influent port, 17% was reduced in the remainder of the bioreactor, and 17% remained as nitrate. From day 50 to day 122, no nitrate or nitrite was observed at the influent or effluent port, indicating that 100% of the injected nitrate was reduced at the influent port. Sulfate reduction restarted under these conditions, with 68% of the sulfate being reduced near the bioreactor inlet and 28% being reduced in the remainder of the bioreactor from day 69 to day 122. Net acetate production was observed only under sulfate-reducing conditions; i.e., no acetate was observed from day 1 to day 49 when nitrate and/or nitrite were present. The similar acetate concentration profiles at the influent and effluent ports (Fig. 2C and D) indicated that most was generated near the influent port, with little acetate production or use in the remainder of the bioreactor. As in the absence of nitrate (Fig. 2A and B), acetate peaked at 5.0 to 5.5 mM, plateaued at 3.0 to 3.5 mM, and then declined to 0 to 0.5 mM (Fig. 2C and D).

Continuous injection of 2 mM sulfate and 100 mM nitrate into bioreactor S-HN completely stopped sulfate reduction over the duration of the injection period (Fig. 2E and F). On average, 1.75 mM sulfate, 85.3 mM nitrate, and 9.3 mM nitrite were present in the bioreactor effluent. Hence, 63% of the 14.7 mM nitrate that was reduced remained as nitrite whereas 37% appeared to be further reduced to nitrogen. No acetate was observed under these conditions. The very similar concentration profiles of nitrate and nitrite at the influent and effluent ports (Fig. 2E and F) indicate that hNRB were active mainly near the inlet, not in the remainder of the bioreactor.

The injection of either low- or high-concentration pulses of nitrate indicates the rate at which communities can switch from sulfate to nitrate reduction and vice versa. Low (4 mM) nitrate pulses in bioreactor S-LNP led to an increase in the sulfate concentration and a decrease in the sulfide concentration, the latter often reaching zero. Effects were most pronounced at the influent port (Fig. 3A). However, even while nitrate injection was in progress, recovery of sulfate reduction occurred, restoring the sulfide production within 18 and 14 days at the influent and effluent ports, respectively. Sulfate reduction was associated with the production of up to 5 mM acetate. When nitrate injection was resumed on day 202 following a long period of injection of sulfate only (days 122 to 201), both nitrate and nitrite were observed at the influent and effluent ports. Sulfide dropped to 0, and sulfate rose to 2 mM (Fig. 3A and B), after which recovery of sulfate reduction was observed. Increased sensitivity to nitrate following prolonged injection of sulfate has been documented and explained before (34).

Fig 3.

Metabolite concentrations at the influent (IN) and effluent (EF) ports of bioreactors subjected to pulsed nitrate injection. The enclosed boxes mark when CSBK-S (2 mM sulfate) was switched to CSBK-SLN with 2 mM sulfate and 4 mM nitrate in bioreactor S-LNP (A, B) or to CSBK-SHN with 2 mM sulfate and 100 mM nitrate in bioreactor S-HNP (C, D). Concentrations of sulfate (●), sulfide (○), nitrate (▲), nitrite (△), and acetate (×) are indicated as a function of time.

Injection of pulses of 100 mM nitrate into bioreactor S-HNP caused breakthrough of nitrate and nitrite and complete inhibition of sulfate reduction at both the influent and effluent ports (Fig. 3C and D). Recovery of sulfate reduction was observed when nitrate injection was stopped and was associated with the production of up to 5 mM acetate. The time required for recovery depended on the length of the nitrate pulse. For nitrate pulses of 2 to 8 days, recovery to 1 mM sulfide in the effluent took 2 to 7 days, whereas for nitrate pulses of 23 and 37 days, recovery took 18 and 16 days, respectively. Higher nitrite concentrations during prolonged injections (on average, 24 and 11 mM, respectively), compared to shorter injections (on average, 6.5 mM) could be responsible for these differences. The formation of similar acetate concentrations of up to 5 mM in bioreactors S-LNP and S-HNP when no nitrate was injected and sulfate-reducing conditions were re-established (Fig. 3) indicated that pulsed injection of 100 mM nitrate inhibited but did not eradicate the sulfidogenic consortia in the bioreactors.

Distribution of biomass, residual oil, and residual oil components.

The distribution of attached biomass, determined as the protein concentration (milligrams per milliliter) of extracted fractions along the flow path, indicated the presence of 0.16, 0.34, and 1.28 mg/ml protein at the inlets of bioreactors S, S-LN, and S-HN, respectively (see Fig. S1A to C in the supplemental material). For bioreactors S-LN and S-HN, this represented 31 and 51% of the total extracted biomass, with lower protein concentrations being observed further along the bioreactor flow path. In the case of bioreactor S, somewhat higher biomass concentrations were observed in fractions F4 and F5 (see Fig. S1A).

Fraction F1 (Fig. 1B, 0 to 1.5 cm), containing glass wool, had more oil than any of sand-containing fractions F2 to F10 (see Fig. S1D in the supplemental material). Relative to an MHGC heavy-oil control, the average percent loss of C₁₉-to-C₂₉ alkanes for fractions F1 to F10 under sulfate-reducing conditions in bioreactor S was 33.8% ± 3.0%, compared to 11.8% ± 1.5% and 13.3% ± 2.5% in bioreactors S-LN and S-HN, respectively (see Fig. S2). Of the C₁₃-to-C₁₈ alkanes, 12 to 15% was lost in all three bioreactors (see Fig. S2). Determination of the alkane-to-pristane ratios in individual fractions indicated that loss of alkanes was not confined to a particular region but was similar along the length of bioreactors S-LN and S-HN. Increased loss of C₁₉-to-C₂₉ alkanes was also seen in most fractions of bioreactor S (results not shown). Our experimental design did not allow reliable determination of toluene and other alkyl-benzenes, which were depleted in all three columns. However, microcosm studies of MHGC oil incubated with produced water only (31) or with a produced water inoculum and medium (30) had previously indicated that hNRB prefer toluene and other alkyl benzenes as electron donors for nitrate reduction, whereas SRB used both toluene and other alkylbenzenes, as well as alkanes (30, 31, 44). Partial or complete inhibition of sulfate reduction in bioreactors S-LN and S-HN was likely responsible for the decreased use of alkanes under nitrate-reducing conditions.

Gene distribution by qPCR.

The dsrB and napA gene distributions for fractions from bioreactors S, S-LN, and S-HN are shown in Fig. 4 for gDNA and cDNA templates. The former indicates the potential for nitrate and sulfate reduction, while the latter represents the active expression of these genes under bioreactor conditions. The data were corrected for differences in the volumes of extracted fractions and are expressed as log gene copies per milliliter. Most of the dsrB copies were found in the influent region (IN-F2) of bioreactors S and S-LN, both for cDNA and for gDNA (Fig. 4A). The presence and expression of the dsrB gene in the influent region (IN only) of bioreactor S-HN were 1 to 2 orders of magnitude lower than in bioreactors S and S-LN, in agreement with the fact that sulfate reduction was inhibited (Fig. 2E and 4A). However, higher numbers of genomic dsrB copies were present deeper in bioreactor S-HN. Active expression of napA was highest at the inlet of bioreactors S-HN and S-LN (Fig. 4B). In bioreactor S-LN, this region overlapped with the active expression of dsrB. Genomic napA amplicons in the S-HN bioreactor were more evenly distributed. Interestingly, genomic napA was high but active expression was low in the inlet region of bioreactor S. Hence, the inlet of bioreactor S has potential NRB, which may grow fermentatively in the absence of nitrate.

Fig 4.

Quantification of the dsrB (A) and napA (B) genes in the combined fractions indicated. gDNA and cDNA data are represented by black and gray bars, respectively. Bars are averages of triplicate determinations; the standard deviations are also shown.

Microbial community analysis by 454 pyrosequencing.

For bioreactors S and S-LN, analyses were done with DNA from fractions IN-F2 (influent) and F5 to F10 (effluent). For bioreactor S-HN, analyses were done with DNA from fractions IN (influent), F1 to F3, and F7 to F10 (effluent). The latter two had similar community compositions (Fig. 5) and are therefore both referred to as effluent fractions. Comparison of microbial community compositions in a dendrogram indicated distinct bioreactor communities (bioreactor S influent, bioreactor S effluent, bioreactor S-LN influent, bioreactor S-LN effluent, bioreactor S-HN influent, and bioreactor S-HN effluent), which all differed significantly from communities in produced waters from the MHGC field (Fig. 5). Rarefaction curves and bioinformatic analyses indicated that the community in bioreactor S-HN had fewer operational taxonomic units (OTUs) than the communities in bioreactors S and S-LN (see Fig. S3 in the supplemental material) (Table 2). Numbers of OTUs and diversity indices for bioreactors S and S-LN were comparable to those from produced waters (Table 2). Predominant taxa in the 12 amplicon libraries are indicated in Table 2; for a more detailed survey, see Table S1 in the supplemental material.

Fig 5.

Relational tree for microbial community compositions derived by pyrosequencing. Compositions of influent and effluent samples from the bioreactors and of produced waters from the MHGC field formed distinct clusters. The composition of a tailing pond surface water sample was used as the outgroup to root the tree. The scale indicates the fraction of sequence divergence.

Table 2.

Survey of sequences and derived numbers of OTUs and taxa obtained for the 12 amplicon libraries indicated in Fig. 5^a

Sample	No. of reads	No. of OTUs (95% confidence level)	Estimated total no. of OTUs (Chao)	No. of taxa	Shannon evenness index	Shannon index	Simpson index	Predominant taxa (%)
S IN	6,054	151	208	63	0.6	3.2	0.1	Desulfovibrio (23.0), Acetobacterium (19), Desulfuromonas (18.5), Methanosaeta (13.7), Desulfobulbus (3.3), Bacteroidetes (2), Acidaminobacter (1.9), Desulfuromonadales (0.7), Acholeplasma (0.6), Desulfobacter (0.5), Desulfomicrobium (0.5)
S EF	5,764	192	264	125	0.5	3.0	0.2	Propionibacteriaceae (43.8), Acetobacterium (8.7), Desulfovibrio (6.6), Thermanaerovibrio (4.5), Kosmotoga (3.9), Methanosaeta (3.1), Methanomethylovorans (1.9), Bacteroidetes (1.5), Anaerolineaceae (1.5), Acidaminobacter (1.3)
S-LN IN	7,428	196	330	69	0.4	2.6	0.3	Acetobacterium (54.6), Acidaminobacter (17.8), Fusibacter (4.7), Pseudomonas (4.2), Proteiniphilum (3), Desulfovibrio (1.62), Lachnospiraceae (1.2), Clostridiales (1.0), Desulfuromonas (0.9), Thermanaerovibrio (0.9)
S-LN EF	3,372	162	221	124	0.6	3.0	0.1	Aquabacterium (28.3), Pelomonas (18.4), Propionibacteriaceae (13.1), Clostridiales (3.7), Sediminibacterium (3.2), Caulobacteraceae (3.1), Senobacteraceae (2.5), Sphingomonas (2.2), Rhodococcus (1.5), Methylobacterium (1.5)
S-HN IN	12,392	11	17	12	0.1	0.3	0.9	Thauera (95.6), Pseudomonas (3.4)
S-HN EF (F1-F3)	9,041	104	195	65	0.3	1.8	0.4	Thauera (69.7), Rhodobacteraceae (9.4), Limnobacter (8.7), Proteiniphilum (2.6), Pseudomonas (2), Burkholderiales (1.5), Rhizobium (1.05), Phenylobacterium (0.6), Tessaracoccus (0.6), Anaerolineaceae (0.4)
S-HN EF (F7-F10)	9,960	123	192	86	0.4	2.1	0.3	Thauera (56.9), Limnobacter (10.2), Rhodobacteraceae (8.8), Pseudomonas (4.5), Rhizobium (4.3), Anaerolineaceae (3.9), Proteiniphilum (2.1), Rhizobiales (0.9), Phenylobacterium (0.9), Alkalibacter (0.7), Tessaracoccus (0.6)
2-PW	11,244	377	661	206	0.6	3.6	0.1	Thauera (28.3), Methanoculleus (16.8), Pseudomonas (6.9), Bacteriovorax (5.95), Chryseobacterium (4.4), Flavobacterium (3.2), Methanolinea (2.8), Methanosaeta (2.5), candidate division OP3 (1.9), Rhizobium (1.6)
7-PW	10,509	285	523	149	0.6	3.6	0.1	Methanoculleus (27.5), Thauera (12.7), Methanolinea (10.1), Methanosaeta (4.6), Magnetospirillum (4.5), Methanocalculus (4.3), candidate division OP3 (3), Deferribacteraceae (3), candidate division OP9 (2.4), Smithella (1.75), Peptococcaceae (1.5), Petrobacter (1.4), Pseudomonas (1.3), Desulfotomaculum (1.2), Kosmotoga (1), Methanobacterium (0.9)
3-PW	12,810	193	340	103	0.5	2.8	0.2	Methanoculleus (41.6), candidate division OP3 (10.7), Peptococcaceae (10.4), Methanosaeta (6.8), Smithella (6.5), Methanolinea (3.7), Spirochaetaceae (1.7), Pseudomonas (1.5), Kosmotoga (1.1), candidate division OP8 (1), candidate division OP9 (0.7), Syntrophus (0.7)
4-PW	10,278	236	362	108	0.5	3.1	0.1	Methanoculleus (28.7), candidate division OP3 (26.2), Methanolinea (8.8), Methanosaeta (5.3), candidate division OP9 (4.2), Kosmotoga (1.5), Smithella (1.1), Chloroflexi (1), Spirochaetaceae (1), Anaerolineaceae (0.9), Peptococcaceae (0.6), Thermanaerovibrio (0.6)
10-PW	13,261	171	230	111	0.5	2.7	0.2	Methanoculleus (44.4), candidate division OP3 (11.6), Methanocalculus (6.1), Methanofollis (5.5), candidate division OP9 (4), Smithella (3.4), Methanosaeta (2.5), Chloroflexi (2.3), Spirochaetaceae (2.3), Methanolinea (2.3), Kosmotoga (1.2), Peptococcaceae (1.1), Anaerolineaceae (0.9), Thermanaerovibrio (0.9), Syntrophus (0.8)

^aThe estimated maximum number of OTUs (Chao) and diversity indices are also indicated. Predominant taxa (percentage of all reads), identified mostly to the genus level, are indicated for each amplicon library. For additional information on all of these, see Table S1 in the supplemental material.

The microbial community in bioreactor S influent was dominated by the SRB (class/genus) Deltaproteobacteria/Desulfovibrio and Deltaproteobacteria/Desulfobulbus, as well as by Deltaproteobacteria/Desulfuromonas. Other prominent taxa were the fermentative Clostridia/Acetobacterium and the acetotrophic methanogens Methanomicrobia/Methanosaeta. Bioreactor S effluent had a different community composition, dominated by fermentative Actinobacteria/Propionibacteriaceae with lower proportions of Desulfovibrio, Acetobacterium, Thermoanaerovibrio, Kosmotoga, and Methanosaeta being present.

Bioreactor S-LN influent had high proportions of Acetobacterium and of fermentative Clostridia/Acidaminobacter and Clostridia/Fusibacter, as well as of the hNRB Gammaproteobacteria/Pseudomonas. The hNRB Betaproteobacteria/Thauera, as well as Desulfovibrio, were also present (see Table S1 in the supplemental material). Bioreactor S-LN effluent had high proportions of Betaproteobacteria/Aquabacterium, Betaproteobacteria/Pelomonas, and Propionibacteriaceae and lower proportions of Clostridia/Clostridiales, Sphingobacteria/Sediminibacterium, and Alphaproteobacteria/Caulobacteraceae. Reads for Acetobacterium were 1,000-fold less frequent than in the influent fraction.

Bioreactor S-HN influent was completely dominated by Thauera with smaller proportions of Pseudomonas, Betaproteobacteria/Limnobacter, and Alphaproteobacteria/Rhodobacteraceae being present. Bioreactor S-HN effluent (F1 to F3 and F7 to F10) was also dominated by Thauera, with Limnobacter, Rhodobacteraceae, Pseudomonas, and Bacteroidetes/Proteiniphilum also present.

Produced waters from the MHGC field with nitrate breakthrough (Table 2, 2-PW and 7-PW; see Table S1 in the supplemental material) had high proportions of the hNRB Thauera, which was much less strongly represented in produced waters without nitrate breakthrough (Table 2, 3-PW, 4-PW, and 10-PW; see Table S1). All produced water samples had high proportions of the hydrogenotrophic methanogen Methanomicrobia/Methanoculleus, which was only a minor component of amplicon libraries from the S and S-LN bioreactors (see Table S1). All produced waters also had high proportions of candidate division OP3, which was not found in any of the bioreactor samples (see Table S1). The acetotrophic methanogen Methanosaeta was observed in all field samples, as well as in bioreactors S and S-LN.

DISCUSSION

Studies of souring and its control with nitrate or nitrite have mostly used VFA or lactate as the electron donor for nitrate and sulfate reduction (15, 19, 21, 33, 34, 45, 46). Only Myhr et al. (18) also used an oil-flooded bioreactor. Although the microbial community in our bioreactors used residual heavy oil effectively as an electron donor for nitrate and sulfate reduction, it took approximately 2 months for maximal rates of these processes to be established (Fig. 2A to D). Complete reduction of injected nitrate (4 mM) and sulfate (2 mM) under these conditions indicated that the residual oil in the sand pack represented an excess of electron donor. Nitrate was completely reduced near the influent port, whereas sulfate was also reduced deeper in bioreactor S-LN. Expression profiles of the napA and dsrB genes confirmed this distribution of nitrate and sulfate reduction activity and indicated the presence of up to 10⁷ gene copies/ml (Fig. 4).

The finding that acetate was produced from oil whenever sulfate was used as an electron acceptor provides new information on the mechanism of oil degradation under sulfate-reducing conditions. Continuing with equation 1 and assuming that H₂ is used as the electron donor for sulfate reduction, we find:

$$68H_2+17{{SO}_4}^{2-}+17H^{+}\to 17H{S}^{-}+68H_{2}O$$ (5)

$$4C_{16}{H}_{34}+17{{SO}_4}^{2-}\to 32{CH}_3{COO}^{-}+17H{S}^{-}+15H^{+}+4H_{2}O$$ (6)

Overall, equation 6 indicates that if H₂ formed in equation 1 serves as the electron donor for the reduction of sulfate in the bioreactors, then the formation of 3.8 mM acetate is expected per 2 mM sulfate reduced to sulfide. This is in the range of the values shown in Fig. 2 and 3. The decline of acetate concentrations under conditions where all sulfate has already been reduced is likely caused by acetotrophic methanogenesis. Acetate is expected to be used by acetotrophic SRB in the presence of excess sulfate, whereas H₂ is used by hydrogenotrophic methanogens in the absence of sulfate (1, 4, 14). In the presence of nitrate, hNRB catalyze the complete oxidation of hydrocarbon to CO₂ without excretion of acetate. These hNRB do not depend on the fermentative bacteria that degrade oil under sulfate-reducing or methanogenic conditions. Although hydrocarbon-oxidizing, sulfate-reducing bacteria are known (9–13), they were not identified in the samples analyzed in the present study. A common denominator of all hydrocarbon-oxidizing SRB appears to be that they were isolated from marine sediments with hydrocarbon pollution or from a seawater-containing oil storage tank. These marine environments may have excess sulfate (typically 25 mM) and limiting hydrocarbon, favoring SRB that completely oxidize hydrocarbon substrates. In contrast, the MHGC field and the bioreactors described here have limiting sulfate (1 to 2 mM) and excess hydrocarbon. We propose that in bioreactors S and S-LN, reaction of oil with limiting sulfate is through equations 1, 5, and 6, as indicated by the presence of significant proportions of hydrogenotrophic SRB and by the production of acetate in accordance with equation 6.

Injection of low concentrations of sulfate (1 mM) and nitrate (2 mM) into the MHGC field is thought to give rise to zones of nitrate reduction, sulfate reduction, and methanogenesis along the flow path from injection to production wells (6, 31, 34). These migrate from injectors to producers, depending on the availability of toluene, the preferred electron donor of hNRB of the genus Thauera (30, 31, 44). In the absence of nitrate breakthrough, produced waters contain sulfide but no sulfate and are dominated by the presence of hydrogenotrophic methanogens (Methanoculleus, Methanolinea, and Methanocalculus) and the acetotrophic methanogen Methanosaeta (Table 2, 3-PW, 4-PW, and 10-PW; see Table S1 in the supplemental material). hNRB and SRB are present in small proportions (see Table S1). With nitrate and/or nitrite breakthrough, the proportion of the hNRB Thauera increases considerably (see 2-PW and 7-PW in Table S1). Increases in other potential hNRB, e.g., Pseudomonas (32) and Deferribacteraceae (7), are also seen. The latter were not observed in bioreactors, but Thauera clearly increases in bioreactors when nitrate is included.

Fermentative bacteria degrading oil in syntrophy with methanogens or SRB are expected to be present in bioreactors S and S-LN and absent from bioreactor S-HN. Deltaproteobacteria/Syntrophus and Deltaproteobacteria/Smithella, which have been proposed to degrade especially alkanes in syntrophy with methanogens (1, 3, 4), are present in low proportions in bioreactors S and S-LN and much more significantly in MHGC produced waters (see Table S1 in the supplemental material). Other bacteria that satisfy this criterion are Acetobacterium, Acidaminobacter, Propionibacteriaceae, Synergistetes/Thermanaerovibrio, Kosmotoga, Fusibacter, Aquabacterium, Sediminibacterium, and many others. Several of these have been found in communities degrading xenobiotic compounds or in communities derived from oil, gas, or coal fields (47–51). Our knowledge of anaerobic, syntrophic hydrocarbon degradation is still rudimentary, making it impossible to implicate taxa on the basis of 16S rRNA pyrosequencing surveys. Like the genus Acetobacterium, the genus Desulfuromonas may be involved in fermentative acetate metabolism in ways that we do not yet understand (52). Syntrophic acetate oxidation (CH₃COO⁻ + H⁺ + 2H₂O → 2 CO₂ + 4H₂) coupled to hydrogenotrophic methanogenesis has been described as a significant anaerobic pathway in high-temperature oil fields (53). It is not thermodynamically feasible at lower temperatures, as in the MHGC field.

Acetate has been found in waters produced from many oil fields, with a reported range of 0.3 to 20 mM (54). Pham et al. (55) found up to 3.6 mM acetate in produced waters from a mesothermic Alaskan reservoir, and Kuijvenhoven et al. (29) reported up to 15 mM acetate in waters produced from seawater-injected oil fields off the coast of Nigeria and in the North Sea. Produced waters from the high-temperature Kaparuk field had acetate concentrations of 8.3 to 14.7 mM (56). Our results indicate a mechanism through which this acetate may form and are of both fundamental and practical interest.