Correspondence: Wilfred F.M. Röling, Molecular Cell Physiology, Faculty of Earth and Life Sciences, VU University Amsterdam, De Boelelaan 1085, 1081 HV Amsterdam, The Netherlands. Tel.: +31 20 5987192; fax: +31 20 5987223; e-mail: email@example.com
The adsorption of nucleic acids to mineral matrixes can result in low extraction yields and negatively influences molecular microbial ecology studies, in particular for low-biomass environments on Earth and Mars. We determined the recovery of nucleic acids from a range of minerals relevant to Earth and Mars. Clay minerals, but also other silicates and nonsilicates, showed very low recovery (< 1%). Consequently, optimization of DNA extraction was directed towards clays. The high temperatures and acidic conditions used in some methods to dissolve mineral matrices proved to destruct DNA. The most efficient method comprised a high phosphate solution (P/EtOH; 1 M phosphate, 15% ethanol buffer at pH 8) introduced at the cell-lysing step in DNA extraction, to promote chemical competition with DNA for adsorption sites. This solution increased DNA yield from clay samples spiked with known quantities of cells up to nearly 100-fold. DNA recovery was also enhanced from several mineral samples retrieved from an aquifer, while maintaining reproducible DGGE profiles. DGGE profiles were obtained for a clay sample for which no profile could be generated with the standard DNA isolation protocol. Mineralogy influenced microbial community composition. The method also proved suitable for the recovery of low molecular weight DNA (< 1.5 kb).
All known life forms on Earth store hereditary information in nucleic acids like RNA and DNA (Gallori, 2011). These biomarkers can be extracted and used to provide information on the nature of organisms and their physiological characteristics. The field of microbial ecology has rapidly advanced in the last decades by the application of molecular, cultivation-independent approaches like PCR-based community characterization and metagenomics. However, extraction of nucleic acids from terrestrial samples can be challenging owing to the low abundance of life in certain environments, such as subsurface and extreme environments (e.g. Barton et al., 2006).
An extreme environment of particular interest is the planet Mars. Two Mars missions (Mars Science Laboratory launched in November 2011 and Exomars in 2018) will target molecules indicative of life in the near future (Ehrenfreund et al., 2011). Life may have developed on early Mars when it was warmer and wetter than it is today (Pollack et al., 1987). Currently, if life exists on Mars, it is likely scarce and localized in the subsurface because of the extreme conditions on its surface: high UV radiation, low temperatures, dryness, strong oxidizing conditions and a thin carbon-rich atmosphere (~ 95% CO2) (Tung et al., 2005; Grady, 2007).
Mars mineralogy comprises besides iron-rich minerals like iron oxides (magnetite and haematite) and iron oxyhydroxides (goethite and ferrihydrite) that may play role in biological iron reduction, also evaporites (sulphates, e.g. jarosite and gypsum), carbonates, silicates (olivine, pyroxene and plagioclase) and clay minerals (phyllosilicates such as montmorillonite and kaolinite) (Poulet et al., 2005; Chevrier & Mathé, 2007; Grady, 2007; Ehlmann et al., 2008; Mustard et al., 2008). These clay minerals originated from a combination of impacts, volcanism and a hydrological cycle early in Mars history (Poulet et al., 2005).
Both on Earth and on Mars, clays are of particular interest because of their implications in the origin of life (Bernal, 1951). Clays adsorb organic molecules (e.g. Saeki & Sakai, 2009), and this contributes to their ability to catalyse a diversity of organic reactions. RNA molecules bind efficiently to clays such as montmorillonite, which catalyse the formation of longer molecules (Ferris, 2005). Furthermore, DNA binding to clay minerals protects it against enzymatic digestion (Aardema et al., 1983), UV radiation (Scappini et al., 2004) and X-rays (Ciaravella et al., 2004). Clays are also evidence of past aqueous activity and have been considered a main target to future Mars missions (Ehrenfreund et al., 2011).
However, the adsorption to clays and other minerals interferes with extraction and characterization of nucleic acids (e.g. Novinscak & Filion, 2011) and provides an additional challenge for cultivation-independent characterization of low-biomass, Mars-like environments (Direito et al., 2011). Nucleic acid extraction methods for the detection of life on Mars, but also in low-biomass environments on Earth, should be robust and produce high quality nucleic acids, with minimal loss during extraction. Therefore, the aims of this research were the determination of the recovery of nucleic acids from relevant minerals on Earth and Mars, the testing of the tolerance of DNA to various physical and chemical conditions that are applied to dissolve mineral matrices or to extract organic compounds from solid matrices, and the optimization of sensitive DNA extraction methods.
Materials and methods
The terrestrial and Mars analogue minerals used for the determination of DNA recovery are described in Table 1. To obtain 1–2 mm size fractions, minerals were split, crushed and sieved at the Mineral Separation Laboratory of VU University Amsterdam, the Netherlands.
Table 1. Terrestrial and Mars analogue minerals used for the determination of DNA recovery
Clay mineral, phyllosilicate
Otay, San Diego County, California, USA; Clay Minerals Society, Colorado, USA
Clay mineral, phyllosilicate
Gonzales County, Texas, USA; Clay Minerals Society
Bacillus cereus, Escherichia coli CL4B and Shewanella putrefaciens 200R were grown in liquid Luria–Bertani broth medium and Deinococcus radiodurans in medium 53 as described by DSMZ (Deutsche Sammlung von Mikroorganismen und Zellkulturen). All strains were grown at 37 °C with agitation. Cell numbers were measured with a Multisizer 3 Coulter Counter (Beckman Coulter). pGEM-T plasmid containing a 1.5-kb 16S rRNA gene fragment was isolated using the GeneJET Plasmid Miniprep kit (Fermentas Life Sciences) from an E. coli JM 109 transformant obtained from a clone library described in Direito et al. (2011).
Minerals (5 g quantities) were individually incubated inside membrane pockets (polyamide membrane with 6-μm mesh width and 5% open surface, Solana NV, Schoten, Belgium) for 9 months at 8 m depth in a monitoring well in an iron-reducing aquifer (Banisveld, Boxtel, the Netherlands) (Direito et al., in preparation). Three clays (nontronite, montmorillonite and kaolinite) and two nonclay minerals (olivine and quartz) were examined in this study.
Quantification of DNA recovery from minerals
Mineral samples (0.25 g) were spiked with an 81-bp PCR fragment of the yeast Saccharomyces cerevisiae hexokinase 1 (YHXK1) gene (0.5 ng μL−1 final concentration). Generation of the spike, standard DNA extraction and qPCR-based quantification of recovered spike were performed according to Direito et al. (2011).
Determination of DNA tolerance to different physical and chemical conditions
Bacterial genomic DNA (16–60 ng μL−1) was subjected to a range of acidic treatments and high temperatures, listed in Table 2. Hydrofluoric acid (HF) treatments on DNA, and intact cells of the gram-negative S. putrefaciens 200R, gram-positive B. cereus and extremophile D. radiodurans, were performed according to a protocol to dissolve silicates (Boenigk, 2004). Cells were collected by centrifugation for 5 min, 15300 g and resuspended in 1 mL of 0.9% NaCl. DNA was extracted from 500 μL of cell suspension, and the remainder was used for microscopy and enumeration with a Multisizer 3 Coulter Counter (Beckman Coulter, CA). All experiments were performed in triplicate. Controls using DNase and RNase-free water (MP Biomedicals, Solon, OH) instead of acid were included. A standard DNA precipitation step was performed for all the acid treatments to remove acids. DNA was resuspended in TE buffer. To check DNA quality after treatments, agarose gel electrophoresis was performed. DNA recovery was quantified by qPCR and expressed relative to the control.
Table 2. Acidic and thermal treatments applied to isolated genomic DNA to establish its tolerance to these treatments. DNA recovery was determined by 16S rRNA gene-directed qPCR and expressed relative to the controls
Background on selecting the treatment
% DNA recovery
8 M HF at 30 °C
HF is used to remove silica in clay-rich sediments by dissolving silicates while leaving cells intact (Boenigk, 2004)
8 M HF at 80 °C
1 M HCl at 80 °C
The acid treatment removes lattice Al, Fe and Mg from clays and the alkaline digestion removes the silica (Mathers et al., 1954)
1 M HCl at 80 °C plus 5% Na2CO3 at 80 °C
0.5 mM HCl
Inorganic acids were tested to check their influence on DNA integrity
5 mM HCl
50 mM HCl
1 M formic acid
Organic acid with pKa (3.77) comparable to HF (pKa = 3.17). Organic acids excreted by, for example, fungi aid in dissolving minerals (Fomina et al., 2005)
5 min at 100 °C
Subcritical water extraction will be used to extract organic compounds like amino acids during a Mars mission. In the extractor the temperature is set between 100 and 300°C at 20 MPa for a number of minutes (Aubrey et al., 2008)
15 min at 100 °C
1 h at 100 °C
Standard and adapted DNA isolation protocols
The PowerSoil DNA isolation kit (MO BIO Laboratories, Solana Beach, CA) was used for DNA isolation from strains and mineral samples according to the manufacturer protocol, except for experiments in which the protocol was adapted to promote DNA recovery from clay after cell lysis.
In the later experiments, the PowerBead Tubes that contain beads and 750 μL of cell lysis solution were replaced by sterile 2-mL bead beat tubes containing 0.5 g of 2-mm diameter glass beads and 1 mL of one of the tested sterile extraction solution (Table 3). After the extraction solution, 60 μL of solution C1 from the PowerSoil kit (cell lysis solution) was added. Tubes were vortexed to mix all contents and cells disrupted. First, a large range of extraction solutions and incubation temperatures were tested, as indicated in Table 3, using 108E. coli CL4B cells added to calcium-rich montmorillonite clay. Incubation times were between 40 and 60 min. Tubes that were incubated at 30 °C were agitated at 200 r.p.m., while tubes incubated at 80 °C were not agitated. For the later, pressure was alleviated by slightly opening the cap (small torsion) from time to time. All phosphate buffers were pH 8 and prepared from NaH2PO4/Na2HPO4. After this incubation step, DNA isolation was continued according to the standard PowerSoil kit protocol, with an additional step for ATP containing extraction solutions. To avoid competition of ATP with DNA for binding to the silica spin filter, a standard DNA precipitation step (Miskin et al., 1999) was performed on the supernatant resulting from the addition of solution C2 (a reagent to precipitate non-DNA organic and inorganic material). It was expected that most of the much smaller molecule ATP would remain in solution and thus would be removed. The pellet was dissolved in 400 μL of phosphate buffer 1.2 M (pH 8) plus 200 μL of C2 PowerSoil solution and isolation continued according the PowerSoil kit protocol.
Table 3. Extraction solutions tested for enhancing DNA recovery by chemical replacement/competition, with the respective incubation temperature (°C) and a short description on how extraction solutions would enhance DNA recovery. The first column shows codes as used in the text. Incubation times were 40–60 min. All buffers were pH 8
Description of extraction solution
Hypothesized mechanism for enhancing recovery
Powersoil 750 μL (MO BIO Laboratories)
Standard kit (for comparison)
PowerSoil 1250 μL (MO BIO Laboratories)
Higher volume to achieve higher homogenization
1 M phosphate buffer/15% ethanol
Use of high concentrations of simple molecules with characteristics resembling those of DNA (negatively charged phosphate, pyrophosphate, ATP) to promote chemical replacement/competition for binding sites in clay, leaving more DNA in solution.
Ethanol at concentrations of 15% or higher changes the conformation of DNA forming loops and more complex structures (Fang et al., 1999) possibly promoting DNA recovery from minerals surfaces
0.1 M ATP/1 M phosphate buffer/15% ethanol
Other tested solutions
1 M phosphate buffer/15% ethanol
0.5 M phosphate buffer/15% ethanol
0.5 M phosphate buffer
1 M phosphate buffer
0.1 M ATP/0.5 M phosphate buffer/15% ethanol
0.1 M ATP/1 M phosphate buffer/15% ethanol
0.1 M ATP/0.5 M phosphate buffer
0.1 M ATP/1 M phosphate buffer
0.1 M ATP
0.1 M pyrophosphate
0.5 M EGTA
EGTA chelates divalent cations, such as found on clay, and therewith alleviates adsorption of DNA to clay
0.5 M EGTA/0.5 M phosphate buffer
0.5 mg mL−1 RNA (Boehringer Mannheim 109223)
RNA competes with DNA for binding sites on clay (Frostegård et al., 1999)
10 mg mL−1 BSA (New England BioLabs)
BSA is known to adsorb to clays (Servagent-Noinville et al., 2000) and therewith competes with DNA for binding sites, leaving more DNA in solution
For subsequent detailed statistical evaluation of the improvement in DNA recovery by the most promising extraction protocols and impact of cell quantities, extractions were performed in duplicate and started by adding known quantities of E. coli CL4B (105–108) to empty tubes for cells-only controls and to tubes containing 0.25 g of calcium-rich montmorillonite (Table 1). This clay mineral was selected because it revealed very low DNA recovery (Fig. 1). Previous extractions revealed that indigenous DNA was not present in this clay. To mimic a clay matrix in which cells are embedded, all tubes were briefly vortexed and placed at 4 °C with agitation (200 r.p.m.) during 20 min to homogenize. Cells-only controls were subjected to the same procedure. Centrifugation at 10 000 g for 10 min was performed and the supernatant removed, remaining the cells embedded in the clay or a pellet (for cells-only controls). DNA was quantified by qPCR and expressed in fg DNA per cell.
Primers targeted the bacterial 16S rRNA gene: primers F357 and R518 (Muyzer et al., 1993) were used to amplify a 0.2-kb fragment. A GC-clamp (Muyzer et al., 1993) was included in primer F357 for DGGE profiling. Primers 8F and 1512R were used to amplify 1.5-kb fragments (Felske et al., 1997). The PCR programme consisted of an initial denaturation at 94 °C for 5 min; 35 cycles of 94 °C for 30 s, 54 °C for 1 min and 72 °C for 1 min (90 s for 8F × 1512R); plus a final elongation step at 72 °C for 5 min. A total volume of 25 μL was used in each PCR reaction, containing 0.4 μM forward and reverse primers; 0.4 mg mL−1 bovine serum albumin (BSA; New England BioLabs, Leusden, the Netherlands); 12.5 μL Fidelitaq PCR Master Mix (2×) (USB Corporation, Cleveland, OH); 8.5 μL DNase and RNase-free water (MP Biomedicals) and 1 μL template.
Quantitative real-time PCR was performed with a 7300 Real Time PCR System (AB Applied Biosystems, CA) using primers targeting the bacterial 16S rRNA gene: F357 and R518 (Muyzer et al., 1993). The qPCR programme was: 50 °C for 2 min, 95 °C for 15 min, 40 cycles (95 °C for 15 s, 54 °C for 30 s, 72 °C for 30 s) plus 72 °C for 10 min, followed by a melting curve programme to determine the melting profile of obtained amplicons and check for primer dimers: 95 °C for 15 s, 60 °C for 1 min, 95 °C for 15 s and 60 °C for 15 s. qPCR reactions were performed in triplicate along with negative controls (DNase and RNase-free water; MP Biomedicals). A total volume of 22 μL was used in each qPCR reaction, containing: 1 μL of each primer (forward and reverse) 0.4 μM; 12.5 μL 2× DyNAmo HS SYBR Green qPCR master mix (Finnzymes, Espoo, Finland); 0.6 μL 50× ROX (Finnzymes); 6.9 μL DNase and RNase-free water (MP Biomedicals) and 3 μL template. Standard curves consisting of tenfold serial dilutions of genomic DNA in the range of 0.00001–100 ng μL−1 were included.
PCR products were analysed by denaturing gradient gel electrophoresis (DGGE; Bio-Rad DCode Universal Mutation Detection System). Markers (M12) consisting of a mixture of 12 different bacterial 16S rRNA gene fragments (prepared in our laboratory from cloned fragments) were used alongside the samples. Gels were 8% polyacrylamide (37.5 : 1 acrylamide/Bis) with a denaturing gradient of 30/55%. Electrophoresis was undertaken in 1× TAE buffer during 3.5 h at 200 V and 60 °C. Gels were stained with ethidium bromide, illuminated under a Vilber Lourmat (TCP-20-M) UV transilluminator and photographed.
Testing for PCR inhibitors
To verify that DNA extracts obtained with the standard PowerSoil kit did not contain compounds that could inhibit the PCR reaction, 1 μL of S. putrefaciens DNA (10 ng μL−1) was added to 1 μL template and subjected to a bacterial 16S rRNA gene PCR. The lack of amplification would indicate the presence of PCR inhibitors.
For the environmental samples and standards subjected to qPCR, a comparison of the individual PCR efficiencies was performed. Efficiency values were calculated using LinRegPCR version 12.17 (Ruijter et al., 2009; Tuomi et al., 2010). Normalized fluorescence data (Rn – normalized reporter signal values) from all samples were imported to LinRegPCR, and baseline correction and determination of individual PCR efficiencies were performed. Differences in amplification efficiencies would indicate the presence of inhibitors.
Two- and three-way analysis of variance (anova) and nonparametric rank-based test (Kruskal–Wallis) were performed with Systat version 7.0 (SPSS, Chicago, IL).
Quantification of DNA recovery from terrestrial and Mars analogue minerals
The recovery of a gene fragment of the yeast hexokinase gene, which was added as a spike to terrestrial and Mars analogue minerals, was determined after confirming that the gene target was not present in any of the unspiked minerals. Recovery depends strongly on the type of mineral (Fig. 1). All clays revealed very low spike recovery (< 0.5%), whereas the iron-bearing minerals haematite and magnetite showed the highest recovery values, with no significant loss. Goethite shared the same recovery as quartz (~ 50%). From the remaining minerals, jarosite presented the lowest recovery (0.07% recovery) followed by diopside (0.1%) and labradorite (0.25%) (Fig. 1).
Nucleic acids are known to adsorb strongly to clays (e.g. Cai et al., 2006) but also to other minerals (in more detail discussed in the discussion); thus, we suggest that low recoveries resulted from strong adsorption of DNA to minerals, especially to clay minerals. Therefore, the focus for the optimization of extraction was directed in particular to clays, as these occur widely and in relatively large quantities.
Tolerance of DNA to physical and chemical conditions
Acids have been used to dissolve minerals, in particular clay minerals, thus removing adsorbing minerals while leaving cells behind. An efficient disintegration method for direct counting of bacteria in clay-rich sediments is the dissolution of silicates with HF (Boenigk, 2004). Therefore, this method was tested on DNA and cells to check whether DNA would endure the treatment. HF showed a negative effect on DNA; qPCR revealed no significant recovery of amplifiable DNA (Table 2). For cells treated with HF, DNA recovery was also negligible: from 0.01% for gram-negative S. putrefaciens 200R and extremophile D. radiodurans to 0.05% for gram-positive B. cereus. These cells passed the treatment intact, with a reduction in the cell diameter by 35 ± 4% (n = 3). Other acid treatments can also remove lattice Al, Fe and Mg from clays, and subsequent alkaline digestion can remove silica (Mathers et al., 1954). This procedure, consisting of the use of 1 M HCl, was tested, but DNA was not recovered (Table 2). Subsequent tests with lower concentrations of HCl (0.5–50 mM) indicated that with increasing acid concentration, more degradation occurred, with only 5% remaining with 50 mM HCl (Table 2). An organic acid (1 M formic acid) with similar pKa (3.77) to HF (pKa 3.17) was tested, but the DNA recovery was very low (2%).
Subcritical water extraction is a method based on the decrease in the dielectric constant of liquid water by increasing the temperature and pressure making water with properties chemically similar to organic solvents. This method has been proposed for astrobiology missions to extract organic compounds (such as amino acids). It employs temperatures between 100 and 300 °C at 20 MPa for a number of minutes (Aubrey et al., 2008). To test whether DNA could withstand such high temperatures, DNA was subjected to 100°C for increasing periods of time. However, rapid degradation of DNA occurred (Table 2).
Optimization of DNA extraction methods
As acidic and high temperature approaches proved to be too destructive, a large range of solutions was tested to promote DNA recovery from clay (Table 3) by adjusting the cell-lysing step in the PowerSoil DNA isolation kit (MO BIO Laboratories). 108E. coli CL4B cells were added to the clay Ca-montmorillonite as it revealed low recovery (Fig. 1). Tested solutions included different high phosphate concentration buffers to promote chemical replacement/competition for binding sites (Table 3). Mixes of high concentration phosphate buffers with 15% ethanol were used and adapted from tested procedures for DNA desorption from aminosilane-modified magnetic nanoparticles (Tanaka et al., 2009). EGTA (ethylene glycol tetraacetic acid) was also tested because it is known to chelate divalent ions. By chelating ions in the mineral matrix, it would be expected to prevent DNA adsorption. BSA was used in a similar fashion because it is known to adsorb to clays (Servagent-Noinville et al., 2000).
Initial tests, checked by agarose electrophoresis, suggested that the most efficient extraction solutions were PowerSoil 1250, P/EtOH and ATP/EtOH (see Table 3 for details). These treatments were subsequently compared with the standard PowerSoil kit (Fig. 2). DNA was extracted from 105, 106, 107 and 108 cells, and for each treatment, also a control, without the clay, was included. The presence of clay had a negative effect on the amount of DNA extracted per cell (3-way anova, F1,32 = 16.208, P < 0.001). In the absence of clay, no significant effect of extraction solution (F3,16 = 2.315, P = 0.11), cell numbers (F3,16 = 2.867, P = 0.07) or their interaction (F9,16 = 0.525, P = 0.84) on DNA recovery was observed. However, solution ATP/EtOH generally showed lower recoveries compared with P/EtOH, possibly due to competition of ATP with DNA in binding to the membrane of the spin filter. The introduced DNA precipitation step was not effective because ATP was not completely removed probably due to co-precipitation with DNA (data not shown).
In the presence of clay, the buffer solution had a significant effect on DNA recovery (F3,16 = 140.125, P < 0.001), with P/EtOH having a significant higher recovery than the other solutions (Fig. 2), while also ATP/EtOH had a significant (P < 0.05) higher recovery than the standard kit PowerSoil. The PowerSoil 1250 treatment was not significantly better than the standard kit. A significant effect of added cell numbers was observed (Fig. 3). There is a decreasing trend of DNA recovery with higher cell numbers, except for P/EtOH. P/EtOH and ATP/EtOH have higher DNA recoveries than PowerSoil Standard, and 1250. P/EtOH always shows better results than the standard kit, while buffer ATP/EtOH worked relatively better at the lower cell numbers (105). Variation between replicates was determined by calculating the coefficient of variation and was on average 32 ± 30% (n = 16) and not influenced by the type of extraction solution.
Extraction and characterization of DNA from microorganisms on minerals incubated in an aquifer
The P/EtOH solution resulted in the best recovery of DNA from cells added to clay. To test the efficiency and reproducibility of the P/EtOH solution with environmental samples differing in mineralogical composition, this solution and the standard PowerSoil kit were applied to minerals that were incubated for 9 months at 8 m depth in an anaerobic, iron-reducing aquifer. Figure 4 shows the number of 16S rRNA gene copies retrieved per gram of montmorillonite, nontronite, olivine and quartz. These minerals were selected as they represented minerals with low recovery (the clays montmorillonite and nontronite), intermediate (olivine) and high recovery (quartz; Fig. 1). Significantly higher numbers of 16S rRNA gene copies were extracted with the optimized protocol from the minerals that previously revealed low recovery [montmorillonite and nontronite (Mann–Whitney U-test, P = 0.021)] or intermediate recovery [olivine (P = 0.050)] but not for quartz, where more gene copies were obtained with the standard kit, and the difference between the two extraction approaches was not significant (P = 0.127). In addition, the comparison of the individual qPCR amplification efficiencies of these samples and standards revealed similar amplification efficiencies (average = 87.7%, standard deviation = 2.6%, n = 120) thus were not indicative of PCR inhibition.
Figure 5 shows the DGGE profiles for olivine, quartz and kaolinite after triplicate extractions with the standard and optimized protocols. For olivine and quartz, the DGGE profiles were highly reproducible and not influenced by the type of extraction protocol employed. For the clay kaolinite, no DGGE profile could be generated when the standard PowerSoil protocol was used, while employing the P/EtOH solution resulted in a complex, reproducible profile. The lack of amplification in the kaolinite extracts obtained with the standard PowerSoil was not because of the presence of PCR inhibitors because PCR products were obtained when Shewanella DNA was added to the PCR reaction.
For each mineral, clearly different community profiles were obtained.
Recovery of DNA molecules differing in length
To check the recovery of DNA molecules differing in length with the optimized protocol using P/EtOH, defined quantities of intact genomic DNA, a 4.5-kb plasmid and PCR fragments with sizes 0.2 and 1.5 kb, all containing 16S rRNA gene fragments, as well as intact cells, were added to calcium-rich montmorillonite and extracted according to the adapted protocol. Two-way anova with the type of spike and presence of clays as factors showed that the type of spike (F4,10 = 2.085, P = 0.16) did not significantly affect recovery, while the presence of clay lowered the recovery (F1,10=9.741, P < 0.05). Significant interaction between the two factors was observed (F4,10 = 4.290, P < 0.05: Fig. 6). In the absence of clay, the type of spikes did not affect recovery (F4,5 = 1.300, P = 0.38). Whereas in the presence of clay, there is a significant difference in extraction (F4,5 = 9.978, P < 0.05: Fig. 6). Plasmids and DNA fragments (0.2 and 1.5 kb) revealed similar recovery, while for cells and genomic DNA, the recoveries were lower (Fig. 6). As a result, extraction appears more efficient for lower molecular weight DNA.
DNA released during DNA extraction can bind to minerals, having a detrimental impact on the cultivation-independent detection and characterization of microbial life (Direito et al., 2011). Metagenomic studies provide essential information on microbial ecology, but their results strongly depend on the applied extraction method (İnceoğlu et al., 2010). Extraction bias is known to affect metagenomic studies in such a way that current DNA extraction methods only provide access to hardly over 60% of the total metagenome in soil (Lombard et al., 2011). Therefore, it is of utmost significance to optimize DNA extraction methods to achieve high, unbaised recovery of nucleic acids for use in molecular characterization of microbial communities. Numerous studies have addressed methodological issues important for DNA extraction from soils and sediments such as unbiased lysis and high yield of high molecular weight DNA. For instance, bead-beating methods have been considered superior for cell lysis and DNA yield and these are currently included in commercial DNA extraction kits (Takada-Hoshino & Matsumoto, 2004). Commercial kits accelerate and standardize sample processing, being widely used and considered suitable for obtaining good quality DNA (e.g. Dineen et al., 2010). However, adequate kit selection is dependent on the characteristics of the soil samples (Dineen et al., 2010). In addition, protocols for removal of PCR inhibitors such as humic acids have been developed (e.g. Kallmeyer & Smith, 2009). So far, problems have been encountered in developing a generic DNA extraction protocol that can be used on all soil types (Lombard et al., 2011) and with different mineralogies. One major remaining issue is the loss of DNA during extraction, which is in particular a significant problem for low-biomass environments. While whole genome amplification methods have been used to obtain more DNA for metagenomic studies on low-biomass environments (Abulencia et al., 2006), whole genome amplification methods also have their biases (Abulencia et al., 2006; Kim & Bae, 2011). DNA is lost during extraction, because of adsorption in particular. Adsorption problems have been addressed using competitor molecules like skim milk (Takada-Hoshino & Matsumoto, 2004) or RNA (Frostegård et al., 1999) to saturate adsorption sites in clay-rich samples; however, DNA yields increased only marginally (Frostegård et al., 1999). The use of these biological competitors may also interfere with the analysis of other biomarkers and pose a contamination risk. Here, we have investigated systematically which minerals negatively affect DNA recovery. We then improved DNA recovery by a simple adaptation of a commercial extraction kit and showed that the optimized procedure did not affect the retrieved community fingerprint.
We quantified recovered nucleic acids by amplification, as in principle, a single molecule can be detected and also the presence of amplifiable DNA allows for subsequent characterization of community composition, for example, by DGGE fingerprinting. PCR can be subject to inhibition (Wintzingerode et al., 1997); however, our analyses did not reveal inhibition: spiking extracts with genomic DNA resulted always in good amplification, while amplification efficiencies in qPCR were not affected by the type of mineral or extraction protocol from which the DNA extract was derived. Other DNA quantification methods also have their drawbacks: UV spectrophotometry is relatively insensitive and can overestimate the amount of DNA because of the presence of micron-size clay particles, phosphate or other non-DNA UV absorbing materials (Yankson & Steck, 2009). Fluorescent dyes like PicoGreen are very sensitive for dsDNA, but also subject to interference by humic acids (Bachoon et al., 2001) and quantification with PicoGreen does not indicate whether DNA is amplifiable.
Among the minerals systematically tested in this study, iron-rich minerals like haematite and magnetite revealed highest recovery, which is an important observation considering that these minerals are appealing for the search of life on Mars. Anaerobic microbial iron oxidation and reduction have been indicated as the most primordial forms of microbial metabolisms on early Earth and possibly also on other iron-mineral-rich planets like Mars (Weber et al., 2006). Other tested minerals revealed low recovery (< 10%), including minerals implicated in supporting life. For example, apatite is an important source of leachable phosphate for microbial growth under phosphate-limited conditions (Bennett et al., 2001). DNA may strongly adsorb to apatite because of the interaction between positively charged calcium ions in apatite and the negatively charged phosphate groups of DNA (Okazaki et al., 2001). Quartz (SiO2) revealed higher DNA recovery than most of the other silicates (e.g. olivine, diopside and labradorite), which all contained cations. Saeki & Kunito (2010) indicated that DNA adsorption to silica (SiOx) is enhanced by the presence of cations. Monovalent cations (e.g. Na+) reduce the electrostatic barrier between DNA and silica while divalent cations (e.g. Ca2+ and Mg2+) neutralize charges by binding to both the DNA phosphate backbone and to Si-OH groups in the silica, forming bridges between the two (Nguyen & Elimelech, 2007) and enhancing adsorption (Garko & Stewart, 1994). Among the silicates, clays, or phylosilicates (montmorillonite, nontronite and kaolinite), showed lowest DNA recovery but are also considered to be a important target to future Mars missions (Ehrenfreund et al., 2011). Clays indicate past aqueous activity, protect DNA against degradation for long periods of time (Trevors, 1996) and have been implicated in the origin of life as catalysers of reactions (Ferris, 2005). Consequently, in particular, DNA extraction methods for clay-rich samples required optimization.
Acidic treatments have been used to dissolve minerals (Mathers et al., 1954; Boenigk, 2004) and therewith aid in separating cells from minerals. However, the harsh, acidic conditions led to the destruction of DNA, even though methods such as HF dissolution leave cell forms intact (Boenigk, 2004; this study) and enable the recovery of other biomarkers, such as proteins (Schulze et al., 2005). Therefore, milder approaches were needed to improve the recovery of DNA from samples rich in DNA-binding minerals. Adsorption of molecules on clays is thought to occur at the edges of clay mineral sheets, where aluminium groups coordinate with water molecules. This generates acidic binding sites (Al-OH2+) to which negatively charged molecules such as phosphate groups in DNA, but also other polyanionic molecules, can bind (Ferris, 2005; Saeki & Sakai, 2009). Frostegård et al. (1999) showed that the addition of excess RNA prior to cell lysis enhanced DNA recovery, but even then, the recovery rate never exceeded 3% (Frostegård et al., 1999). Milder approaches to enhance DNA recovery were therefore in particular based on promoting chemical competition for adsorption sites in minerals using (nearly) saturated polyanionic solutions, such as phosphates. These solutions were included in a direct cell lysis protocol where the microbial cells are first broken in the soil/mineral matrix, and subsequently, the released DNA is purified from the cellular debris and soil particles. Current cell lysis buffers for DNA isolation can be divided in two groups: Tris–HCl/EDTA and phosphate buffers with pH 7.0–8.0 (He et al., 2005). Previous research had already revealed the positive effect of the use of 0.12 M phosphate buffers in DNA extraction on DNA recovery, compared with phosphate-free extraction buffers (Ogram et al., 1987). However, employing 0.12 M phosphate buffer in extraction still leads to large losses of DNA for clay samples (data not shown). Different forms of phosphate, such as inositol phosphate (Goring & Bartholomew, 1952) and metaphosphate (Pietramellara et al., 2001), were previously examined in relation to the adsorption of DNA to clays, but not in relation to subsequent PCR-based quantification and characterization of recovered nucleic acids. Yankson & Steck (2009) tested the DNA desorption from kaolinite with different buffers, including 100 mM phosphate that showed the best results. Consequently, a large range of phosphate-based buffer solutions were tested, and DNA recoveries were quantified via qPCR.
The most effective buffer was P/EtOH (1 M phosphate buffer, 15% ethanol (pH 8.0), with incubation at 80 °C for 40 min). This method was based on a procedure to release DNA from aminosilane-modified magnetic nanoparticles (Tanaka et al., 2009). Ethanol at concentrations of 15% or higher changes the conformation of DNA, forming loops and more complex structures (Fang et al., 1999). It probably promotes DNA desorption from minerals surfaces (like phyllosilicates) because linear DNA is thought to bind more strongly than supercoiled DNA by having a continuous line of contact with the surface (Melzak et al., 1996). At pH 8, adsorption to clay is less than at lower pH (Khanna & Stotzky, 1992; Cai et al., 2006; Saeki & Kunito, 2010).
The introduction of the P/EtOH buffer into the MO BIO PowerSoil DNA extraction protocol resulted in higher DNA yields, up to 100-fold compared with the standard MO BIO PowerSoil buffer. This was observed for both minerals spiked with known quantities of cells or DNA (Figs 2, 3 and 6) as well as for individual minerals incubated under in situ aquifer conditions (Fig. 4). Equally important, the inclusion of the P/EtOH buffer did not have an influence on the microbial community composition in the retrieved DNA compared with the standard protocol and despite up to 100-fold higher extraction efficiencies. Highly reproducible DGGE profiles were obtained (Fig. 5), and in one case, a DGGE profile was obtained for a clay sample for which no DGGE profile could be generated when DNA was isolated according to the standard isolation protocol. Biases in the DGGE profiles because of higher extraction efficiencies were also not expected as nucleic bases are not known to be involved in the adsorption process (Pietramellara et al., 2001) and thus differences between microorganisms in the GC content of their DNA should not affect adsorption. The application of the optimized protocol to minerals incubated under in situ conditions in an anaerobic aquifer revealed that mineralogy affects microbial community composition. This is currently subject of a more detailed study (Direito et al., in preparation).
Some studies have shown that there is preferential adsorption of small DNA fragments to soils and clays (Ogram et al., 1994; Pietramellara et al., 2001) probably due to size exclusion or mechanisms related to the kinetics of diffusion (Ogram et al., 1994). As previously mentioned, linear DNA is thought to bind more strongly than supercoiled DNA by having a continuous line of contact with the surface (Melzak et al., 1996). Consequently, smaller-sized linear molecules would likely to be more adsorbed. However, for clay-rich samples, extraction seems more efficient for lower molecular weight DNA (plasmids and DNA fragments) when the P/EtOH solution is employed. A possible explanation could be the conformation changes in DNA molecules because of the inclusion of 15% ethanol in the extraction solution (Fang et al., 1999), in particular enhancing desorption for smaller DNA fragments. Short dsDNA usually adopts a linear form (Cleaves et al., 2011), while larger DNA molecules present more complex conformations. The higher recovery of DNA molecules with this solution, irrespective of the molecular weight and conformation of DNA, with the P/EtOH, indicates its suitability for astrobiology studies on Martian samples, where if any DNA is present, it might be not well preserved. The optimized protocol appears also suitable for the extraction of fragmented and degraded ancient DNA. A drawback for studies on active microbial communities might be that extracellular DNA, not associated with living cells, will be relatively well recovered.
The following modification of the PowerSoil DNA isolation protocol (MO BIO Laboratories) is suggested for the extraction of DNA from environmental samples, in particular samples with low biomass and rich in clay minerals:
Add 0.25 g of sample to sterile bead-beating tubes containing glass beads (0.5 g, 2 mm).
Add 1 mL of 1 M phosphate buffer/15% ethanol (pH 8) and 60 μL of solution C1 (if these solutions are precipitated, heat to dissolve).
Vortex to mix the contents.
Break cells, for example, using a FastPrep FP120 cell disrupter machine (BIO101 Thermo Savant; Qbiogene, Cedex, France) at speed 5.5, for two times 30 s, with cooling down in between.
Incubate tubes in a block heater at 80 °C for 40 min. Alleviate pressure from time to time by slightly opening the cap (small torsion only).
Continue DNA isolation according to standard PowerSoil DNA isolation protocol.
We expect that these steps can also be easily incorporated in other (commercial) DNA extraction protocols. The PowerSoil kit was selected here because it is widely used for DNA isolation from soils (e.g. Alcántara-Hernández et al., 2009) and provides good results when compared with other available kits for soils (e.g. Mahmoudi et al., 2011). Despite the strongly improved recovery of DNA from clay samples, still a considerable part of DNA present in cells will be lost during extraction (up to ~ 90%). Therefore, spiking samples with a defined quantity of DNA prior to extraction followed by targeting a gene, only present in the spike, by qPCR on the isolated DNA, is advised to determine the recovery of DNA from any sample. The experiment on the recovery of DNA molecules of various sizes suggests that genomic DNA or cells might best be used as spike, as shorter fragments are relatively better extracted and might lead to bias in determining recovery. Other studies have also considered the use of genomic DNA, cells and plasmids as spikes (Lee et al., 1996; Mumy & Findlay, 2004). The advantage of using intact cells over genomic DNA is that most DNA in environmental samples probably is found inside cells. Care should be taken that the spike does not contain genes that one would want to characterize for the environmental samples. This may complicate the use of spikes based on genomic DNA or cells, for example, in case 16S rRNA gene-based community profiling is planned. Further optimizing DNA recovery is therefore still warranted, as in particular for clay calcium-rich montmorillonite, the DNA recoveries remained below 10%. Also, we addressed single minerals while in nature sediments will contain mixtures of minerals, and it is currently difficult to predict how DNA extraction and community fingerprinting will be affected by mixtures of mineral particles, where each mineral may harbour its own community (Fig. 5).
The authors would like to thank Wim Lustenhouwer for his help in mineral selection and access to some minerals; Roel van Elsas for his assistance in the Mineral Separation Laboratory (FALW, VU University Amsterdam, the Netherlands); Richard J. Smeets for the use of hydrofluoric acid and the respective laboratory facilities and Martin Braster for helping in the hydrofluoric acid experiment. The research was funded by a grant from the Netherlands Organization for Scientific Research (NWO/SRON User Support Programme Space Research, project ‘Molecular detection of life on Mars’ ALW-GO- PL/07-11).