Biostimulation to identify microbial communities involved in methane generation in shallow, kerogen-rich shales

Authors


Correspondance

Philippe M. Oger, PhD, Laboratoire de Géologie de Lyon, Ecole Normale Supérieure de Lyon, UMR, CNRS, 5276, 46, Allée d'Italie, F-69342 Lyon Cedex 07, France. E-mail: poger@ens-lyon.fr

Abstract

Aims

The aim of the present study was to design and test a method allowing the detection and quantification of methanogenic consortia in organic-rich rocks to determine the potential of methane biotransformation.

Methods and Results

Methanogen numbers in the rock are often below the detection levels of quantification methods. Biostimulation was tested as a means to specifically increase bacterial and archaeal numbers above the detection levels in microcosms. Biostimulation reveals the presence of active heterotrophic and syntrophic bacterial consortia, methane accumulation and methanogens in one of four rock samples. Syntrophs and heterotrophs were dominated by Firmicutes, whereas archaeal diversity was limited to methanogens. Methane-producing microcosms were characterized by a higher Firmicutes diversity.

Conclusions

Biostimulation is a reliable tool for detection of methanogenic consortia in organic-rich rocks. For routine and large scale experimentation, methane accumulation monitoring after biostimulation appears as the most time, work and cost efficient approach to detect the presence of active methanogenic consortia.

Significance and Impact of the Study

We report for the first time the presence of live methanogenic consortia in organic-rich shales and their ability to mineralize the rock into methane. This approach will be instrumental to quantify the potential of these rocks to produce methane as a novel energy source.

Introduction

Organic-rich layers of sedimentary basins are widespread environments in which the presence of methane has been known for a long time. In the shallow subsurface, e.g. at depth above 500 m below the surface, methane has a dual origin: a diffusive transfer of thermogenic methane from a deeper source, or a local microbial origin (Tissot et al. 1974). It is only recently that geochemical analyses have demonstrated the biological origin of deep-subsurface methane (Martini et al. 1996, 2003; Aitken et al. 2004; Jones et al. 2008; McIntosh et al. 2008). Methane produced by microbial activity can accumulate over geologic times in rocks with low permeability, e.g. tight sandstones, siltstones, shales, and coal beds. It is estimated that biogenic gas may amount to up to 20% of the total gas resource (Rice and Claypool 1981). The metabolic cascade of methanogenesis from oil or coal is now well understood (Anderson and Lovley 2000; Townsend et al. 2003), and involves the cooperation of different groups of micro-organisms (Schink 1997; Zengler et al. 1999; Aitken et al. 2004). First, the heterotrophic Bacteria hydrolyse the large, complex organic polymers into smaller and simpler substrates, such as sugars, lactate, volatile fatty acids and alcohols (acidogenesis). Then, these products are fermented by syntrophic Bacteria into acetate (acetogenesis), formate, H2 and CO2, which are substrates for methanogenic Archaea. Methanogenesis is performed by different Archaea, depending on the carbon source: acetoclastic, hydrogenotrophic, and methylotrophic Methanogens use acetate, CO2, and methyl-group containing compounds as electron acceptors respectively. Methanogens are present in a wide range of natural habitats; from rice-field soils, animal gastrointestinal tracts, marine and freshwater environments (Garcia et al. 2000; Conrad 2007), to oil reservoirs (Magot et al. 2000; Head et al. 2003; Roling et al. 2003; Aitken et al. 2004). Syntrophy, which is an interspecies hydrogen transfer between fermenting Bacteria and methanogenic Archaea, is essential for the production of methane under environmental conditions (Schink 1997; Stams and Plugge 2009).

Organic-rich shales represent one of the most abundant carbon-rich rocks in Europe and worldwide (Klemme and Ulmishek 1991). A significant part of those source-rocks have not undergone a sufficient burial to generate the pressure and temperature conditions necessary for the complete transformation of the organic matter into oil or coal. As such, their organic content is called immature, and is characterized by a higher proportion of biologically reactive molecules. These immature source-rocks represent a huge and yet untapped fossil carbon resource, which exceeds both the oil and coal resources (Rice and Claypool 1981; Schulz et al. 2010). Thus, there is today a considerable economical and geopolitical interest in these novel, nonconventional energy resources to try and generate fuel of interest for human consumption. Amongst those, microbiologically assisted methanization of the organic matter is one of the most promising technologies developed (Suflita et al. 2004; Gieg et al. 2008; Voordouw 2011). Methanization is known to occur in shales from field data showing the natural accumulation of biogenic methane in several sedimentary basins (Krumholz et al. 1997; Martini et al. 2008; McIntosh et al. 2008). Prior to any field exploitation, several major questions need to be addressed regarding (i) the structure of the microbial populations in the rock, (ii) the fraction of the organic matter effectively transformed into methane, or (iii) the kinetics of methanogenesis, to determine the economic feasibility and evaluate the potential of the source rock for in situ methanogenesis. If microbial consortia involved in the methanogenic degradation of the organic matter in the deep subsurface have been characterized for substrates such as coal formation waters (McIntosh et al. 2008; Warwick et al. 2008; Jones et al. 2010), or petroleum reservoirs production waters (Magot et al. 2000; Head et al. 2003; Roling et al. 2003; Aitken et al. 2004; Grabowski et al. 2005), there is to date almost no data available for organic-rich immature shales. Furthermore, in contrast to oil production waters and coal formation waters, restricted pore sizes in the shale (Fredrickson et al. 1997; Takai et al. 2003) and the nature of the rock allow only very low bacterial and archaeal numbers, which make the direct detection of Methanogens unrealistic (Green et al. 2008; Jones et al. 2010).

Developing an efficient method for the detection and monitoring of microbial consortia involved in the production of methane from organic-rich shales is a crucial step to evaluate the feasibility of the approach. Bacteria and Archaea can rarely be quantified directly, as a result of low initial numbers. Thus, culture-based approaches such as biostimulation, e.g. the specific enhancement of target populations via the amendment of specific nutrients, appear as a method of choice to increase the bacterial and archaeal counts above the detection levels, as well as increase the methanogenic activity of indigenous micro-organisms of the shales to allow for kinetic quantifications. This approach has been used to quantify and identify Methanogens in oil production waters and coal formation waters (Green et al. 2008; Jones et al. 2010; Penner et al. 2010; Wawrik et al. 2012). Biostimulation has been used with success for the detection and the enrichment of methanogenic consortia from coal samples (Jones et al. 2008, 2010), although this approach showed reproducibility issues. Indeed, only a fraction, between 1 in 4 and 1 in 2, of the microcosms evolved towards the production of methane (Jones et al. 2008). The low initial methanogen counts, the heterogeneity of their spatial distribution, as well as the fragility of the methanogenic consortia, which have been shown to be easily disrupted by manipulation (Hatamoto et al. 2007), may explain this variability, an issue that has not been addressed specifically to date.

In the present study, we have evaluated the use of biostimulation for the identification and quantification of Methanogens from immature source-rocks of the shallow subsurface. We show that biostimulation can be used with organic-rich shales from the Paris basin for the detection, identification and quantification of Methanogens. Methane accumulation monitoring was found to be the most time and cost effective method to demonstrate the presence of Methanogens in these rocks.

Materials and methods

Coring and field sampling

The target source rocks belong to the Lower Jurassic black shales of the eastern Paris basin, known as type II kerogen-rich shales (Tissot et al. 1974; Vandenbroucke and Largeau 2007). Specifically, the source-rock used is from the uppermost Pliensbachian – lowermost Toarcian stratigraphic interval, and is locally referred to as ‘Schistes-carton’ formation. It has been chosen for its kerogen-containing many aromatic and naphthenic rings, as well as some aliphatic chains (Vandenbroucke and Largeau 2007). Percussion drilling with a core sampler containing a plastic sampling tube (diameter of 89 mm) and without the use of drilling fluids was operated in the field to a depth of 12 m at Entrange (Moselle, North-East of France, 49°25′04·7″N, 06°06′14·17″E, Z = 261 m). The cutting head has been fitted with switchable core catcher to avoid sample contamination. Samples for geochemical analysis were collected on site every 10 cm along the core between 7·3 and 8·8 m and at 12 m. Cores were promptly stored in sealed plastic bags under anaerobiosis to avoid oxidation, and were subsampled for further experimentation in the laboratory under anaerobic conditions.

Geochemical analysis of core samples

Geochemical characterization of the organic matter was performed by Rock-Eval 6® (Behar et al. 2001) at the French Institute of Petroleum (IFP Energies Nouvelles, Rueil-Malmaison, France) using 100 mg of pulverized source rock.

Microcosm setup

Each rock sample was pulverized in the anaerobic growth chamber and used to set up at least four microcosms. Preliminary experiments were performed with various mineral bases of Methanogen media, water or diluted organic media (Data not shown and Table 1) tested for biostimulation. The most reproducible results were obtained with medium CP1 which consists of the mineral base of the Methanogenium medium (DSMZ, Germany) in which the total salt content had been reduced (Table 1). CP1 is composed of (l−1): 5 g NaHCO3, 4 g MgCl2 × 6H2O, 3·45 g MgSO4 × 7H2O, 2 g yeast extract, 2 g tryptone, 1 g NaCl, 1 g Na acetate, 335 mg KCl, 250 mg NH4Cl, 140 mg CaCl2 × 2H2O, 140 mg K2HPO4, 2 mg Fe(NH4)2(SO4)2 × 7H2O, 1 mg resazurin, 0·5 g Na2S, 0·5 g l−1, 0·5 g l-cysteine. The volume is adjusted to 1 l with water, sparged with N2/CO2 (80 : 20) and sterilized for 20 min at 121°C. Before use, 0·5 ml of Wolfe vitamin solution and 0·5 ml Wolfe mineral solution (Wolin et al. 1963) from anaerobic sterile stocks were added to CP1 medium. Putative enhancers of methanogenesis were tested to improve biostimulation in CP1 medium (Table 1). Although some were found to increase the production of methane, they were also found to interfere with the extraction of good quality gDNA from the microcosms. Thus, all microcosms were subsequently prepared with CP1 medium in sealed serum vials, under anaerobic conditions. Anoxic conditions in the microcosms were obtained by atmosphere substitution with N2/CO2 (80 : 20) and the addition of Na2S (0·5 g l−1) and l-cysteine (0·5 g l−1). The microcosms were incubated at a constant temperature of 25°C in the dark without shaking (conditions corresponding to a shallow burial) for 1 year. Routinely, microcosms were prepared from 4 g of pulverized shale and 25 ml of CP1 medium in a 50 ml vial. Microcosms for the monitoring of microbial populations by quantitative PCR and sequencing were prepared from 8 g of shale and 50 ml of CP1 medium to allow for frequent sampling. Microcosms are identified by their depth followed by the number 1 for the large microcosms and 2, 3 or 4 for the smaller ones, e.g. 12 m-1 for the large microcosm made out of sample from 12 m depth.

Table 1. Detection of Bacteria and Archaea as a function of medium composition
Sampling dateSampleCP1CP1 + MO4CP1 + NO2CP1 + MO4 + NO2CP1 + NO3CP1 + SO4
CH4aBact. 16SbArch 16SbCH4Bact. 16SArch 16SCH4Bact. 16SArch 16SCH4Bact. 16SArch 16SCH4Bact. 16SArch 16SCH4Bact. 16SArch 16S
J1ET1NANA+NANA+NANA++
ET2NANANANANANA++
ET3NA++NA+NA++NA++NA+NA++
J200ET1ND+++++++ND++±++ND++ND++
ET2++++++++++++++++ND+++
ET3++++++++++++++++++++++ND+++
J216ET3+++++++++++++++++++++++++++++
  CP1 + AcetateCP1 without tryptone and YELB/10LB/10 + NO3WaterSterilized rock
CH4Bact. 16SArch 16SCH4Bact. 16SArch 16SCH4Bact. 16SArch 16SCH4Bact. 16SArch 16SCH4Bact. 16SArch 16SCH4Bact. 16SArch 16S
  1. gDNA was prepared by the procedure modified from Porteous et al. (1997). The 16S rRNA gene from Bacteria and Archaea was amplified by PCR using the primers listed.

  2. NA, not analyzed; ND, not detected.

  3. a

    Relative methane quantification scale; <5 μmoles (+); 5–30 μmoles (++); >30 μmoles (+++).

  4. b

    Relative PCR efficiency scale: no band or inappropriate size (−); faint band (+); intense band (++).

J1ET1NA+NA+NANA++NA++NANA
ET2NA++NA+NANA++NA++NANANA
ET3NA+NANA++NA++NA+NANANA
J200ET1++++++ND++ND+ND++NDNANA
ET2ND+++++++ND+++ND++ND+NDNANA
ET3±+++++++++ND++ND++ND
J216ET3±++++++ND++++++ND

Monitoring of methane production

Methane production was monitored by gas chromatography by direct injection of 10 μl of headspace gas into a 7820A gas chromatograph (Agilent Technologies, Massy, France) equipped with a thermal conductivity detector and a flame ionization detector. Methane concentrations were obtained against a calibration curve prepared with pure CH4 (Air Liquide, France) and expressed in micromoles per gram rock (dry weight). Methane production was monitored twice a week for 4 months. Two additional punctual measures were made at days 262 and 340.

gDNA extraction

To monitor microbial populations, the large (8 g) microcosms were sampled every 2 weeks for the first 3 months and at days 262 and 340. Microcosms were homogenized by inversion, and 2 ml of culture was sampled using a 2 ml syringe. Samples were centrifuged for 5 min at 12 000 g, and the supernatant discarded. The pellet, corresponding to c. 200 mg of source rock (dry weight) was used for gDNA extraction. gDNA was prepared following the procedure of Porteous et al. (1997) with the following modifications. A solvent extraction step was added after the initial lysis to remove organic contaminations from the rock (Porteous et al. 1997). This step included extracting twice with phenol-chloroform-isoamyl alcohol (25 : 24 : 1) and once with chloroform-isoamyl alcohol (24 : 1). gDNA was further purified by chromatography on a column (Nucleospin® Tissue XS Macherey Nagel, Germany): the precipitated gDNA obtained above was resuspended in 80 μl of T1 lysis buffer T1 with 8 μl of proteinase K proceeding from step 2 according to the manufacturer. gDNA yield was quantified using the Quantit dsDNA HS Assay Kit and the Qubit fluorometer (Invitrogen, Eugene, OR, USA). gDNA Quality was evaluated by PCR and qPCR of the bacterial and archaeal 16S genes.

Cell enumeration

For total bacterial and archaeal 16S rRNA gene, and archaeal mcrA gene analyses, real-time PCR assays were performed with the Mx3000 QPCR system (Agilent Technologies, Santa Clara, CA, USA) using the nonspecific fluorophore SYBR green I of the Brilliant II Ultra-Fast SYBR® Green QPCR master mix (Stratagene, La Jolla, CA, USA). The specific primers used for each gene are given in Table 2. The 20 μl reactions contained 1 μl of target gDNA (1 ng μl−1), 0·3 μl of 1/500 diluted reference dye and 1 μ mol l−1 of each forward and reverse primers. The PCR program consisted of 10 min at 95°C and 40 cycles of 95°C for 30 s, annealing at the specific primer hybridization temperature for 30 s, 72°C for 30 s to 1 min, and a final cycle of 1 min at 95°C, 30 s at 53°C and 30 s at 95°C. Melt curve analysis to detect the presence of primer dimers was performed after the final cycle by increasing the temperature from the hybridization temperature to 95°C in 0·5°C increments every 10 s. Negative controls without template were included in each run. A specific cloned reference was used for each target gene. Gene copy numbers were expressed as copies per gram rock (dry weight) assuming one gene copy per cell for each target gene.

Table 2. List of the oligonucleotides used to amplify the 16S rRNA gene of Bacteria and Archaea, and the mcrA gene of Methanogens
 TargetNameSequence (5′→3′)Target and orientationAmplicon size (bp)Reference
  1. xxx denotes the barcode used for the 454-sequencing amplification.

qPCRBacteriaPAAGAGTTTGATCCTGGCTCAGUniversal eubacterial 16S rRNA gene, forward1600Edwards et al. (1989)
PHAAGGAGGTGATCCAGCCGCAUniversal eubacterial 16S rRNA gene, reverse
Archaea109FACKGCTCAGTAACACGTUniversal euryarchaeal 16S rRNA gene, forward867Grosskopf et al. (1998)
Arch 910RGCTCCCCCGCCAATTCUniversal euryarchaeal 16S rRNA gene, reverse
MethanogensMLfGGTGGTGTMGGATTCACACARTAYGCWACAmcrA gene, Forward470Luton et al. (2002)
MLrTTCATTGCRTAGTTWGGRTAGTTmcrA gene, reverse
454Bacteria1073FxxxACGAGCTGACGACARCCATGUniversal eubacterial 16S rRNA gene, forward286Uroz et al. (2010)
787RxxxATTAGATACCYTGTAGTCCUniversal eubacterial 16S rRNA gene, reverse
Archaea571FxxxGCYTAAAGSRICCGTAGCUniversal euryarchaeal 16S rRNA gene, forward339Grosskopf et al. (1998) and Baker et al. (2003)
Arch 910RxxxGCTCCCCCGCCAATTCUniversal euryarchaeal 16S rRNA gene, reverse

Phylogenomic studies

A cloning/sequencing approach has been applied to characterize the archaeal populations in all microcosms in which methane production was confirmed. The archaeal 16S rRNA gene was amplified with the primer pair 109F/910R (Table 2) and cloned into pGEM-T easy®. Ten individual clones per microcosm have been used for Sanger sequencing. Deep-sequencing was performed on one methane-positive and one methane-negative microcosms setup with the same source rock sample. Amplicon libraries were prepared as recommended for 454 pyrosequencing using a combination of two tagged primers. Bacteria were amplified with primers targeting the V5 and V6 hyper variable regions of the 16S rRNA gene, and Archaea with primers targeting the V3 and V5 hyper variable regions of the 16S rRNA gene (Table 2). Amplification generated PCR fragments of average length 250 and 320 bp for Bacteria and Archaea respectively. The PCR were performed in 100 μl reactions using 50 ng of gDNA as template. The PCR conditions used were 94°C for 4 min, 35 cycles of 94°C, 30 s denaturation 50°C, 1 min annealing and 72°C, 1 min 30 s extension; followed by 72°C, 10 min. The amplicons were purified as recommended using QIAquick PCR purification columns (Qiagen) and quantified with the Qubit™ dsDNA HS Assay kit (Invitrogen). Equimolar amplicons were pooled and submitted to pyrosequencing on a Genome Sequencer Junior Titanium Series (Roche Applied Science, Meylan, France) at Biofidal (Vaulx-en-Velin, France), according to the manufacturer's instructions for amplicons sequencing. Taxonomic analyses were performed with mothur 1·23 (Schloss et al. 2009) following the scheme described in Sogin et al. (2006) after sorting sequences using tools available on Galaxy (http://usegalaxy.org). Briefly, 454 sequence reads were trimmed and cleaned (trim.seqs) to conserve sequences ranging from 250 to 300 bp (Bacteria) or from 300 to 380 bp (Archaea). The dataset was reduced to unique sequences (unique.seqs), and chimeras were eliminated (chimera.uchime). The remaining sequences were aligned (align.seqs, dist.seqs) and clustered (cluster) using the same software. OTUs were defined at the 97% similarity level and used to calculate rarefaction curves and richness estimates (Chao1, Shannon indexes). Rarefaction curves were plotted using the R package (www.r-project.org). Representative sequences for each OTU were obtained (get.oturep) and used for phylogenomic affiliation at the 80% confidence threshold using the classification tools of the Ribosomal Database Project (http://rdp.cme.msu.edu).

Results

Geochemistry of the shale

Rock-Eval results are consistent with the expected profile of an immature, type II kerogen-rich rock (Tissot et al. 1974; Durand 2003). However, at the local scale, large differences in organic matter quality can be observed along the 5 m-long core (Fig. 1, Table 3). The mineral carbon content (MinC) varied from 2 to 9 wt %, the hydrogen index (HI) which reflects the saturation of hydrocarbons, varies from 200 to more than 600 mg HC g−1 total organic carbon content (TOC), whereas the oxygen index (OI) which reflects the level of oxidation, is fairly low and varies from 5 mg to more than 100 mg CO2 g−1 TOC. The levels selected for further experiments maximize organic matter divergence. Level 12 m has high HI and TOC values, as well as high oil potential (S2 value, Table 3). Level 8·2 m has low HI and TOC values, and low oil potential, whereas the OI is more than twice lower. Levels 7·3 and 7·8 m have intermediate values.

Figure 1.

Hydrogen index (HI) and Tmax crossed diagram of the core sample. The samples used for microcosm set ups are highlighted by closed diamonds.

Table 3. Geochemical characteristics of the four layers of the shale core chosen for microcosm setup
Depth (m)S1 (mg HC g−1 rock)S2 (mg HC g−1 rock)Tmax (°C)TOC (wt %)HI (mg HC g−1 TOC)OI (mg OI g−1 TOC)
  1. Tmax: calculated maximal temperature reached during diagenesis. TOC: total organic carbon content. HI: Hydrogen Index, e.g. the saturation level of hydrocarbons. OI: Oxygen Index, e.g. an indicator of the mean oxidation state of the organic matter.

7·30·1627·554225·7847746
7·80·3744·554169·1348830
8·20·1522·824175·442353
120·4941·644166·8660718

Quantification of methane production

Significant methane production was detected starting at day 64 in microcosm 12 m-1 and 12 m-2 (Figs 2 and 3). Methane concentration increased until day 96, where it reached a plateau c. 550 and 13 μmoles CH4 g−1 rock for microcosm 12 m-1 and 12 m-2 respectively. No methane accumulation was detected in any other microcosm.

Figure 2.

Monitoring of total bacterial (open triangle), total archaeal (open square) and Methanogens (open circle) population in the four large microcosms. Methane production which occurs only for microcosm 12 m-1 is also reported (dotted line).

Figure 3.

Monitoring of total bacterial (open triangle), total archaeal (open square), Methanogens (open circle) populations and methane accumulation (dotted line) in the four 12 m microcosms. The scale methane accumulation in the headspace of the microcosms is from 0 to 600 μmoles to the exception of that for microcosm 12 m-2 which has been adapted to show the lower methane accumulation.

Enumeration of Bacteria, Archaea and Methanogens

Bacterial populations followed a similar evolution in all microcosms regardless of whether they would evolve towards methane production (Figs 2 and 3). The numbers reached a plateau after c. 2 weeks of incubation, and remained stable at c. 109 cells g−1 of rock over the length of the experiment. Noticeably, the two microcosms showing methane production, e.g. microcosms 12 m-1 and 12 m-2, have the lowest initial bacterial counts, c. 3–4 logs lower than in the other microcosms. Archaeal populations followed a similar trend, but reached much lower numbers, c. 106 cells g−1 of rock to the exception of microcosm 12 m-1 and 12 m-2, in which methane production was evidenced, and where Archaea reached c. 109 and 107 cells g−1 of rock respectively. Methanogen numeration using the mcrA gene as target did not yield significant and reproducible numbers except in microcosm 12 m-1, in which Methanogens peaked at 108 cells g−1 of rock at day 120. The mcrA gene was detected significantly after a short lag period (around 28 days). Methane was detected significantly at day 64, thus 36 days after Methanogens began to accumulate in the microcosms. The peak of methane accumulation corresponds to the maximum counts of Archaea and mcrA genes at 120 days of incubation.

Sequences analysis

The archaeal 16S rRNA gene from a total of 13 methane-positive microcosms, obtained while optimizing the CP1 medium, was amplified and cloned. For each microcosm, 10 independent clones were sequenced. Phylogenomic analyses indicated a very low diversity amongst Archaea in the microcosms. Indeed, all sequences were associated with known Methanogens of the genus Methanoculleus (1 microcosm) and Methanosarcina (13 microcosms) irrelevant of the origin of the medium composition (data not shown). Methanosarcina sequences were most closely related to M. barkeri, which is a very common inhabitant of anoxic, methanogenic environments (lakes, marine sediments, the rhizosphere of plants). Methanosarcina is also commonly found in biodegraded oil and coal, and proposed to be responsible for the production of methane in situ (Grosskopf et al. 1998; Grabowski et al. 2005; Jones et al. 2010; De Vrieze et al. 2012; Wawrik et al. 2012). No sequence from nonmethanogenic Archaea were obtained although the primer pair used for amplification is not specific of Methanogens, which constitutes a strong indication that the vast majority, if not all, of the Archaea in the microcosms were Methanogens.

To decipher the basis for the divergent evolution of methane-positive (12 m-1, 5 samplings) and methane-negative (12 m-3, 3 samplings) microcosms prepared from the same rock sample, a more in-depth characterization of the microbial populations was undertaken using 454 pyrosequencing. The bacterial 16S rRNA gene has been amplified from all samples. The archaeal 16S rRNA has been amplified at day 262 and 344 for 12 m-1 but could be amplified only for day 344 in microcosm 12 m-3. A sequencing depth of c. 1000 sequences per sample was estimated in a preliminary analysis to be sufficient to monitor the diversity of the major microbial groups present in our microcosms. More than 9000 bacterial partial sequences were obtained for the 5 dates and 2 microcosms sampled, which defined 320 OTUs (Table 4). A large majority of the OTUs defined were present in both the methane-positive and the methane-negative microcosms, indicating a large overlap between the populations present in both setups. In fact, the sequences from the 20 largest shared OTUs represented a total between 42 to 65% of the total number of sequences (Table 5). For a number of sequences normalized to the sample least covered, the highest diversity was observed at day 1 in both microcosms (Table 4). Bacterial diversity does not show a strong reduction trend in the methane-positive microcosm over the 1-year of the experiment as the Chao or Shannon index remain stable (Table 4). However, bacterial diversity shows a shift in terms of taxonomy. Indeed, sequences related to the Proteobacteria, Actinobacteria and Bacteroidetes seen at day 1 are absent at later dates (Fig. 4, Table 4). This decrease is correlated with an increase in the proportion of Firmicutes from 82·6 to 98·8%. The bacterial diversity in the methane-negative microcosm decreases sharply from days 1 to 344 as seen in the diversity indexes (Table 3), in the taxonomy, and the average number of clone per OTU. Indeed, the 10 most abundant OTUs per treatment represented c. 63% of all sequences at day 1 for both microcosms (Table 6). This proportion remained similar throughout the experiment for the methane-positive microcosm. On the contrary, these 10 most frequent OTUs represented 73·1 and 80·1% of sequences at days 120 and 344 in the methane-negative microcosm. Sequences associated with the Firmicutes dominated the microcosms at all dates and for both microcosms. Amongst Firmicutes, most sequences were related to the genera Bacillus, Clostridium sensu stricto, Clostridium XlV, Clostridium XI, Trichococcus, Sedimentibacter, Desulfosporosinus and several yet unknown Clostridiales or Firmicutes (Tables 5 and 6). The remnant of Firmicutes-related sequences were associated to minor groups closely related to Anaerovorax, Acetobacterium, Lutispora, Coprococcus, Sporotomaculum, Sporacetigenium or Syntrophomonas (data not shown). The most abundant Firmicutes cluster were associated to Clostridium XI and a group of unidentified Clostridiales closely related to Sporacetigenium, representing 30 and 27% of all sequences and 50 and 33 different OTUs respectively. OTUs close to Sedimentibacter (28 OTUs) and Desulfosporosinus (36 OTUs) were also very frequent, but represented only 5–8% of the sequences each. The initial setup of the microcosms is characterized by very specific OTUs belonging to the Proteobacteria (Sulfurospirillum, Desulfomicrobium) and Bacteroidetes (Parabacteroides), which are not detected at other dates.

Figure 4.

Comparative bacterial diversity at days 1, 120 and 344 for the methane-positive and methane-negative 12 m microcosms. Actinobacteria: (image) unclassified Actinobacteria. Bacteroidetes: (image) Parabacteroides sp.; (image) Unclassified Parabacteroides. Chloroflexi: (image) unclassified Chloroflexi. Firmicutes: (image) Bacillus sp.; (image) other Bacilliales; (image) Clostridium (sensu stricto); (image) Sedimentibacter sp.; (image) Clostridium XIVa; (image) Desulfosporosinus sp.; (image) Clostridium XI; (image) Sporacetigenium-related unclassified Clostridiales; (image) other Clostridiales; (image) Unclassified Firmicutes. Proteobecteria: (image) Serratia sp.; (□) Other Proteobecteria.

Table 4. Bacterial diversity at the phylum level in the 12 m-1 methane-positive and the 12 m-3 methane-negative microcosms
 OTUsMethane-producingMethane-non-producing
Day 1Day 120Day 344Day 1Day 120Day 344
  1. Values reported are percentages of sequences representing each phylum for each date and microcosm.

  2. OTU: number of OTUs defining the phylum. Chao: Normalized diversity index of Chao. Shannon: Normalized diversity index of Shannon.

Actinobacteria20·980000·220
Bacteroidetes128·850·530·3522·370·110·56
Chloroflexi70·660·80·35002·53
Firmicutes28582·6297·698·7873·3398·3896·07
Proteobacteria136·881·070·524·301·300·84
Chao 264243249224194115
Shannon 3·813·933·853·723·292·71
Table 5. Proportion of the 20 most abundant and shared OTUs for comparable sampling dates
OTUs numberDay 1Day 120Day 344Identification
PNPPNPPNP
30·970·645·2924·0817·6640·62Unidentified Clostridiales
1778·7718·634·2313·9312·2410·78Clostridium_XI
1908·775·354·234·213·153·36Clostridium_XI
1549·091·9314·815·832·801·40 Bacillus
1591·950·432·120·320·870·98 Sedimentibacter
1651·950·430·260·860·520·56 Desulfosporosinus
1630·320·210·260·650·520·28Unidentified Enterobacteraceae
1722·927·921·590·761·220·14 Sedimentibacter
1880·970·641·060·221·220·14 Desulfosporosinus
1661·301·321·730·702·10 Desulfosporosinus
1555·522·140·260·430·170·42 Clostridium_sensu_stricto
112·601·082·452·66Unidentified Clostridiales
1930·210·320·170·70Clostridium_XI
1956·4916·060·350·42 Parabacteroides
2081·620·640·520·28Clostridium_XI
1740·320·260·110·14Unidentified Clostridiales
1530·320·860·790·76 Trichococcus
270·210·430·350·42Clostridium_XI
1805·821·570·14Unidentified Firmicutes
980·260·350·11Clostridium_XI
Total53·9056·3242·5955·7246·8565·65 
Table 6. Proportions of the ten most abundant OTUs for each sample
Day 1Day 120Day 344
PNPPNPPNP
9·27 Bacillus 18·63 Clostridium_XI 14·81 Bacillus 24·08Unidentified Clostridiales17·66Unidentified Clostridiales40·62Unidentified Clostridiales
8·94 Clostridium_XI 16·06 Parabacteroides 9·79Unidentified Firmicutes13·93 Clostridium_XI 12·24 Clostridium_XI 10·78 Clostridium_XI
8·94 Clostridium_XI 7·92 Sedimentibacter 5·82Unidentified Firmicutes9·94 Clostridium_XI 5·59 Clostridium_XI 9·24 Clostridium_XI
7·62Unidentified Firmicutes5·35 Clostridium_XI 5·29Unidentified Clostridiales5·83 Bacillus 5·42Unidentified Firmicutes6·02Unidentified Clostridiales
6·62Parabacteroides4·07 Clostridium_XlVb 4·23 Clostridium_XI 4·32Unidentified Clostridiales5·07Unidentified Clostridiales3·36 Clostridium_XI
5·96 Clostridium_sensu_stricto 3·00 Clostridium_XII 4·23Unidentified Firmicutes4·21 Clostridium_XI 3·15 Clostridium_XI 2·66Unidentified Clostridiales
5·63 Clostridium_sensu_stricto 2·57 Sedimentibacter 4·23 Clostridium_XI 4·10 Sedimentibacter 2·80 Bacillus 2·52 Sedimentibacter
5·30 Sulfurospirillum 2·14 Clostridium_sensu_stricto 3·70 Clostridium_XI 3·02 Bacillus 2·62Unidentified Firmicutes2·10 Desulfosporosinus
2·98 Sedimentibacter 1·93 Bacillus 3·17 Sedimentibacter 1·94Unidentified Firmicutes2·45Unidentified Clostridiales1·40 Bacillus
2·98 Clostridium_XI 1·93 Parabacteroides 2·91 Bacillus 1·73 Desulfosporosinus 1·57Unidentified Firmicutes1·40 Unidentified Clostridiales
64·24 63·60 58·20 73·11 58·57 80·11 

As expected from the cloning/sequencing data on other microcosms, the archaeal diversity was extremely reduced in our microcosms. Indeed, almost all the sequences, e.g. c. 99%, obtained at day 120 and 344 from the methane-positive microcosm were related to Methanogens, and more specifically to the genus Methanosarcina, confirming prior results. It is composed of one major OTU representing 96·84% of the sequences at day 120, and 100% at day 344. In the methane-negative microcosm, three different genera belonging to two distinct phyla were detected: the methanogen Methanoculleus, represented by one major OTU and 78% of clones was the dominant genus, Methanosarcina (1 OTU) accounted for c. 20% of the total archaeal diversity. The Methanosarcina OTU is identical to that of the methane-positive microcosm. The remaining sequences were associated with the genus Fervidicoccus (Crenarchaeota, Fervidicoccales, 1 OTU, c. 2%), which are organotrophs and not Methanogens.

Discussion

Estimating the possibility of microbial transformation of a source rock into methane requires a precise determination of its so-called ‘methanogenic potential’, which is linked to the presence of a suitable carbon resource and of a suitable microbial population. Therefore, it requires the numeration of the methanogenic and potentially syntrophic population, as well as the estimation of their efficiency to methanize the organic matter. In most instances, however, the numbers of Methanogens present in the rocks are too low to allow direct enumerations by the most widely accepted methods such as acridine orange tagging, self-fluorescence observations, or qPCR assays. Despite their numerous biases, culture-based approaches are the only way available to circumvent these limitations by allowing the enrichment of selected populations. The aim of the present work was to design a reliable, time-efficient approach to quantify the methanogenic potential of immature organic-rich source rocks.

Using the CP1 medium and lower Toarcian shale combination, we reached a reproducibility rate close to 1 in 2 microcosms (This study and unpublished data). Although far from perfect, this success rate is very similar to that reported for coal (Jones et al. 2008, 2010). It allows one to consider using as few as 4 to 6 microcosms per sample to determine the presence of Methanogens in a rock sample reliably. Whether this combination is specific to the rock matrix used in the present study remains to be investigated. Biostimulation was efficient as methane was produced in <3 months. Direct gDNA extraction from the rock using standard preparation procedures for soil were found unsatisfactory (data not shown), because of low yields, and difficulties to obtain PCR amplification of the DNA. This is most probably a result of the carbon-rich matrix of the source rocks which contains high concentrations of PCR inhibitors such as humic acids (Watson and Blackwell 2000). The best results were obtained following the protocol of Porteous et al. (1997), to which we have added additional solvent extraction steps to help remove PCR inhibitors, although gDNA extracted from the source rock never reached sufficient quality and yield to allow PCR or qPCR amplification of the bacterial or archaeal 16S genes. Similarly, the addition of methanogenesis enhancers such as molybdate or nitrite had an unexpected effect on our ability to obtain optimal quality gDNA from the microcosms, which prevented their use in microcosms setups, despite their positive impact on methane production. Microbial population variations over time are characterized by the loss of Proteobacteria and Bacteroidetes, and the correlated increase of Clostridia to represent more than 95% of the bacterial diversity, regardless of the ability of the microcosms to produce methane. The observed composition of the consortia is strikingly similar to those reported for experiments on the degradation of oil and coal (Green et al. 2008; Jones et al. 2008, 2010). Although very similar in evolution and diversity, the methane-positive and negative microcosms differs on two specific characteristics: (i) Initial bacterial counts in the two positive microcosms are 2 logs lower at day 1. Lower initial bacterial counts may reflect the local spatial heterogeneity in the source-rock, such as that observed in soils (Pallud et al. 2004); (ii) Over the 1 year experiment, the overall diversity is stable only in the methane-positive, whereas it is strongly reduced in the methane-negative microcosm (Table 4). The number of clones per OTU is also more evenly distributed in the methane-positive microcosm. It is conceivable that lower initial bacterial counts and a more evenly distributed diversity are the marks of the evolution towards a methane-producing microcosm.

Numerations obtained with the mcrA primer pair (Luton et al. 2002) in our microcosms have always been at least 1 log lower than that of the 16S rRNA gene primer pair 109F/910R (Grosskopf et al. 1998). This is extremely puzzling as almost all the sequences obtained for the archaeal 16S genes were associated to Methanosarcina sp, a genus on which the mcrA primers have been designed. The 16S rRNA gene primer pair 109F/910R used for the amplification of the 16S rRNA gene is universal for Euryarchaeota (Grosskopf et al. 1998; Baker et al. 2003). It has been used by us and several authors to amplify the 16S rRNA gene of a large diversity of Archaea. The 16S rRNA gene primer pair used for the 454 sequencing (571F/910R) is also universal for Euryarchaeota, but may result in different populations being targetted (Grosskopf et al. 1998; Baker et al. 2003). To evaluate whether the pool of clones sequenced represented bacterial target of different size in the microcosms, we performed comparative qPCRs with both 571F/910R and 109F/910R 16S rRNA primer pairs, as well as with the mcrA primer pair on the same gDNA set. The numeration obtained with the two primer sets were essentially identical (data not shown). Those obtained with the mcrA-specific primers ranged between one and two logs lower. Thus, it is clear that Methanogens, and especially Methanosarcina and Methanoculleus represent the only Archaea present in the methane-positive and methane-negative microcosms. The reasons for this discrepancy remain unclear. The mcrA-specific primer set has been designed for the amplification of known Methanogens including Methanosarcina isolates (Luton et al. 2002). It is a widely accepted marker for the detection and quantification of Methanogens. It is very unlikely that the Methanosarcina populations in our microcosms lack the mcrA gene, as it encodes the alpha-subunit of the Methyl coenzyme M reductase (MCR), a key enzyme of methanogenesis, which is ubiquitous amongst methanogenic Archaea (Reeve et al. 1997; Lueders et al. 2001). Our results clearly show that in our experimental systems, the use of mcrA as a marker for Methanogens results in a large underestimation of methanogen numerations. It is more likely that the Methanogens in our microcosms harbour copies of the mcrA gene than cannot be amplified with the mcrA universal primer set. Although it is tempting to imagine that they might harbour an alternate methanogenesis pathway. Isolation in pure cultures and characterization of Methanosarcina clones would be necessary to further address this question.

Deep-sequencing monitoring of one methane-positive and one methane-negative microcosms clearly showed the microbial populations to be dominated by Firmicutes and amongst those by several different OTUs related to Clostridium. This genus is largely dominant over all dates and treatments. Interestingly, species of this genus are known as fermenting organisms, accepting a wide range of substrates such as various carbohydrates (Schnurer et al. 1996), cellulose (Ng et al. 1977) or amino acids (Hoogerheide and Kocholaty 1938). They can oxidize anaerobically fatty acids, as well as propionate and acetate, to hydrogen and carbon dioxide in association with hydrogen-consuming organisms (Stieb and Schink 1985; Schnurer et al. 1996). The fermentation end-products of Clostridium species are various: acetate, propionate, isobutyrate, butyrate, iso-valerate, and valerate (Borsodi et al. 2003). Several other putative acetate producing fermenters are detected at significant levels inside the microcosms: Bacillus and Sedimentibacter, which are capable of fermenting fatty acids and other hydrocarbons to acetate (Borsodi et al. 2003); Parabacteroides which is detected only at day 1 is an obligate fermenter capable of degrading organic polymers such as cellobiose (Sakamoto and Benno 2006; Sakamoto et al. 2007) and Sporacetigenium, known to produce acetate, ethanol, H2 an CO2 from the fermentation of glucose (Chen et al. 2006). Methanosarcina is the only archaeal genus detected in the methane-positive microcosm. Known isolates from this genus are one of the most metabolically versatile methanogen group containing acetotrophs, hydrogenotrophs and methylotrophs. Thus, the methane-positive microcosm possesses a complete methanogenic consortium, composed of fermentative and acetogenic bacteria, e.g. syntrophs represented mostly by diverse Firmicutes, and methanogenic Archaea. The same phyla/consortia have been previously detected in coal bed methane formations of diverse sedimentary basins (Strapoc et al. 2008; Jones et al. 2010; Dawson et al. 2012; Wawrik et al. 2012), as well as in biodegraded oil reservoirs (Magot et al. 2000; Grabowski et al. 2005). Studies on coal revealed the ability of these fermenting and homoacetogenic Bacteria to hydrolyze macromolecules (Dawson et al. 2012; Zakrzewski et al. 2012). Thus, it is reasonable to assume that the consortium detected in the methane-positive microcosm is able to produce methane from the source rock. This hypothesis is consistent with our observation that methane accumulates in microcosms set up only with water (Table 1). The bacterial consortium characterized in the methane-negative microcosm is essentially identical to that of the methane-positive consortium, to the exception of a higher proportion of Clostridium. However, the major archaeal genus is Methanoculleus. In contrast to the Methanosarcina genus, species of Methanoculleus are known to produce methane from H2 + CO2 (Zakrzewski et al. 2012). Acetate is not used as a substrate (Dianou et al. 2001; Asakawa and Nagaoka 2003; Mikucki et al. 2003). The inadequacy between the bacterial consortia fermentation end-products and the substrates required for methanogenesis by Methanoculleus may explain in part our inability to detect methane in this microcosm. The method described here also provides a tool for further identification and characterization of methanogenic consortia isolated from organic-rich, deep-biosphere rocks. The organic matter present in these rocks shares numerous similarities with that of recalcitrant organic macropolymers, such as cellulose, found in several types of organic wastes, ranging from sludges to food industry relishes. Defining an efficient biofuel production method from these renewable resources is one of the challenges of biofuel research. Interestingly, the bacterial consortia enriched in our microcosms are strikingly similar to those isolated from thermophilic compost bioreactors (Izquierdo et al. 2010). The design of an effective selection and culture method for methanogenic consortia from the deep-biosphere opens the path to their metabolic engineering such as has been performed on cellulolytic micro-organisms (Deng and Fong 2011a,b) to improve their methane production rates, and improve microcosm reliability. This would constitute a first step towards the definition of a consolidated bioprocessing scheme for in situ methane production from organic-rich rocks.

In our assay, we tried to maximize the difference in quantity and quality of the organic matter present in the source rock to connect rock geochemistry and methane production. Although the set of conditions is limited, it appears that sample 12 m which was the only one found to support methanogenic growth is the sample with one of the highest TOC values (Table 3), the highest oil potential (S2) and the highest HI. The high HI value supposes a high level of aliphatic chains in the kerogen of the sample, potentially good substrates for fermentative Bacteria (Baskin 1997). The high HI value is also an indication of a better preservation of the organic material. On the contrary, the main Rock-Eval parameters characterizing samples from 8·2 m depth, such as the oil potential (S2), the free hydrocarbons (S1), the TOC as well as the HI are significantly lower. Thus, the high HI values for samples collected at 12 m depth may explain the presence of a methanogenic consortium of micro-organisms naturally present in the rock. In addition, the high OI values observed for samples 7·3, 7·8 and 8·2 m suggest that the rock above 8·2 m depth may have not always been in full anoxia, which could also explain the absence of Methanogens from these samples, as they are strict anaerobes. A more systematic approach on similar source-rocks is needed to check these preliminary correlations and establish a link between the geochemistry of the rock and the distribution and activity of Methanogens.

Our results confirm the strict correlation between methane production in the microcosms and Methanogens numeration via qPCR analyses of the 16S rRNA gene or mcrA. Deep-sequencing could confirm that fully functional syntroph-methanogen interactions took place inside the microcosms. In the present experiment, the increase in Methanogens can be detected before the increase in methane concentration. However, it is very difficult to perform meaningful qPCR numerations without a tedious gDNA preparation of the samples. Thus, on a routine basis, it seems easier, faster and less time-consuming to analyze CH4 accumulations by gas chromatography after enrichment/biostimulation, especially as this approach can be easily automated to process large numbers of samples. Once methane is detected, a molecular approach is the method of choice for confirmation and in-depth analysis.

Acknowledgements

The authors thank two anonymous reviewers for helpful comments on the draft. M.M. was supported by a fellowship of the Région Rhone-Alpes Programme CIBLE 2009. C.P. was supported by a grant from Lyon Sciences Transfert. This work was supported by two grants from the Région Rhone-Alpes to G.D. and P.O.

Ancillary