We provide optimised Mu-DNA protocols for soil, tissue and water samples (Detailed protocols can be found at: https://doi.org/10.17504/protocols.io.qn9dvh6). Each protocol consists of five stages for DNA extraction: lysis, inhibitor removal, silica binding, wash and elution (Figure 1). Mu-DNA uses a lysis buffer modified from Brolaski et al. (2008). The buffer incorporates guanidine thiocyanate to denature proteins (Pitcher et al. 1989), trisodium phosphate to release adsorbed DNA (Ogram et al. 1987) and ethylenediaminetetraacetic acid (EDTA) to reduce DNA oxidation from metal ions (Lloyd and Phillips 1999). A sodium dodecyl sulphate (SDS) solution is added to the lysis buffer in all protocols to disrupt lipid membranes and degrade proteins. The presence of both EDTA and SDS at the lysis stage inhibits nuclease activity (Williams et al. 1980), greatly reducing the degradation of DNA. For soil extractions, the SDS additive includes aluminium ammonium sulphate to reduce humic substances (Braid et al. 2003). For soil and water filter extractions, bead milling is performed for unbiased high yield DNA liberation (Robe et al. 2003). Tissue extractions have a Proteinase K incubation period for enzymatic lysis and protein digestion. Soil and water lysates are purified with a contaminant and inhibitor removal solution. This contains ammonium acetate to precipitate proteins (Crouse and Amorese 1987), aluminium ammonium sulphate (Braid et al. 2003) and calcium chloride (Wechter et al. 2003, Singh et al. 2014) to remove contaminants and inhibitors, in particular remaining humic substances and fine sediment particles. DNA is subsequently bound to a spin column silica membrane under chaotropic conditions with guanidine hydrochloride (Davis et al. 1997). Ethanol washes then remove remaining contaminants prior to elution.

Figure 1.

Simplified Mu-DNA extraction protocols for soil, tissue and water samples. All extractions use stock and working solutions and are divided into five interchangeable stages: lysis, inhibitor removal, silica binding, wash and elution.

The stages of Mu-DNA are designed to be modular and interchangeable between protocols to facilitate optimisation of extraction methods for a given sample type. For example, a bead milling or inhibitor removal stage can be incorporated in a tissue extraction protocol and a tissue wash stage added to a soil or water extraction protocol. All processes are scalable based upon initial sample amount or transferred supernatant volumes.

To determine the performance of Mu-DNA, isolated DNA yield and purity were compared to that from the relevant commercial kits across soil, stool, tissue and water samples (Table 1). The molecular weight of extracted DNA from soil, tissue and water samples was compared between respective methods. Three to five biological replicates were performed per extraction method for each sample.

For each sample type, three different samples (A, B and C) were selected for comparison (Table 1). Sample A represented a commonly encountered sample of its type whereas B and C were representative of more challenging samples.

Table 1.Download as CSV 

Samples used for comparison of methods in this study. Shown are the amounts of each sample processed per extraction method: either Mu-DNA or the relevant commercial kit (Qiagen DNeasy).

Sample	Description	Area sampled	Sample amount	Extraction methods	Lysis apparatus	Replicates
Soil A	Garden soil	Topsoil – surface 5 cm	0.25 g	PowerSoil Mu-DNA: Soil	Tissuelyser II	5
Soil B	Ephemeral pool sediment	Topsoil – surface 5 cm	0.25 g	PowerSoil Mu-DNA: Soil	Tissuelyser II	5
Soil C	Diesel polluted soil	All available	0.25 g	PowerSoil Mu-DNA: Soil	Tissuelyser II	3
Stool A	Erinaceus europaeus	All available	0.25 g	PowerSoil Mu-DNA: Soil	Tissuelyser II	5
Stool B	Anser anser	All available	0.25 g	PowerSoil Mu-DNA: Soil	Tissuelyser II	5
Stool C	Lutra lutra	All available	0.25 g	PowerSoil Mu-DNA: Soil	Tissuelyser II	5
Tissue A	Nimbochromis livingstonii	Flank muscle	25 mg	Blood and Tissue Mu-DNA: Tissue	NA	5
Tissue B	Oniscus asellus	Lateral half	25 mg	Blood and Tissue Mu-DNA: Tissue	NA	3
Tissue C	Lumbricus terrestris	Central segments	25 mg	Blood and Tissue Mu-DNA: Tissue	NA	3
Water A	Shallow eutrophic lake	Shoreline surface	150 mL	PowerWater Mu-DNA: Water	Tissuelyser II	5
Water B	Ephemeral pool mesocosm	Surface	50 mL	PowerWater Mu-DNA: Water	Vortex Adapter	3
Water C	Deep oligotrophic lake	Shoreline surface	1 L	PowerWater Mu-DNA: Water	Vortex Adapter	5

Soil samples were collected from three soil types: A (garden soil; high organic content), B (ephemeral pool sediment; high clay content) and C (diesel polluted soil; high contaminant levels). All samples were loosely mixed at collection. In sterile laboratory conditions, 5 g of each sample was put through a 2 mm mesh sieve to remove large particulate debris before being thoroughly homogenised with a pestle and mortar. The homogenate was separated into multiple 0.25 g (wet weight) subsamples and stored at -20 °C until required for extraction.

Stool samples were collected from three species with different diets: A (European hedgehog, Erinaceus europaeus; omnivore), B (Greylag goose, Anser anser; grazer) and C (Otter, Lutra lutra; carnivore, high number of volatile organic compounds). In sterile laboratory conditions, each sample was thoroughly homogenised with a pestle and mortar. The homogenate was separated into multiple 0.25 g (wet weight) subsamples and stored at -20 °C until required for extraction.

Tissue samples were taken from ethanol preserved specimens of three species: A (Cichlid, Nimbochromis livingstonii; muscle tissue), B (Woodlouse, Oniscus asellus; high chitin content) and C (Earthworm, Lumbricus terrestris; mucus rich with soil gut contents). Multiple 25 mg (dry weight) subsamples of specimens were removed and stored at -20 °C until required for extraction.

Three water samples types were collected: A (shallow eutrophic lake; high sediment load and faecal matter), B (ephemeral pool mesocosm; turbid, high algal content) and C (deep oligotrophic lake; low particulate matter). After collection, samples were transported on ice and stored at 4 °C until filtered. Filtering took place less than 16 hours after collection in sterile laboratory conditions. Each water sample was thoroughly mixed by pouring and then split into two subsamples of equal volume. Subsamples were vacuum-filtered through sterile 47 mm diameter 0.45 μm Whatman cellulose nitrate membrane filters (GE Healthcare), labelled and stored at -20 °C until required for extraction.

DNA extractions of replicate samples followed the protocol of Mu-DNA for the sample type or the relevant DNeasy kit (Table 1). Protocols were modified as follows: Lysis and DNA purification for all protocols were carried out using identical lysis apparatus and spin columns. Soil and stool samples were lysed in 2 ml microcentrifuge tubes (Starlab) on a TissueLyser II (Qiagen) at 30 Hz for 10 minutes. Water samples were lysed in 7 ml Bijou tubes (Sigma-Aldrich) on either a TissueLyser II at 30 Hz for five minutes or Vortex Genie (Scientific Industries) with Vortex Adapter (MoBio) at maximum speed for five minutes. The DNeasy PowerSoil and DNeasy PowerWater Bead Tube contents were transferred to the new tube type prior to lysis. Tissue samples were lysed overnight for identical time periods and incubated at the temperatures specified per protocol. Where required, all available supernatant was transferred and reagent volumes were adjusted accordingly. EZ-10 DNA Mini Spin Columns (NBS Biologicals) were used for DNA purification in all protocols. Elution buffers used in each protocol were added to spin column membranes and left to incubate at room temperature for one minute before final collection. A single elution of the specified volume was performed for each protocol.

dsDNA yield from all extractions was measured with a Qubit 3.0 fluorometer high-sensitivity (HS) dsDNA assay (Invitrogen). Isolated DNA purity was measured with a Nanodrop 1000 spectrophotometer (Thermo Fisher Scientific) recording A₂₆₀/A₂₈₀ and A₂₆₀/A₂₃₀ ratios for all extractions (see Olson and Morrow 2012). To test for the presence of any inhibiting factors in downstream applications, PCRs were run on all extractions. No PCR additives, such as BSA, were used to enhance PCR amplification. DNA extractions were amplified using the broad range DNA barcoding primers LCO1490 and HCO2198 (Folmer et al. 1994). PCRs were 25 μl final reaction volumes composed of 1 μl template DNA, 12.5 μl MyTaq Red Mix (Bioline), 9.5 μl ddH₂O and 1 μl of each 10 μM primer. All PCRs were performed on Veriti 96-Well Thermal Cyclers (Applied Biosystems) under the cycling conditions: 180 s at 94 °C, 37 × (30 s at 94 °C, 60 s at 52 °C, 90 s at 72 °C), 600 s at 72 °C, 600 s at 4 °C. PCR products were visualised on 1.5% agarose gels. All amplifications were given a PCR index score in comparison to a strong positive as follows: no amplification (0), weak amplification (1), moderate amplification (2) and strong amplification (3). Inhibition was considered present in an extraction if its index was ‘0’. To determine the extent of inhibition exhibited in samples, those with a PCR index of ‘0’ underwent further PCRs at 1:10, 1:100 and 1:1000 dilutions.

To assess the integrity and molecular weight of DNA from the Mu-DNA protocols for soil, tissue and water, extractions were compared to those of their commercial counterparts. The highest yielding sample extractions per method were chosen from the highest yielding sample type. Selected extractions (5 μl) were visualised on a 0.5% agarose gel against a GeneRuler 1 kb Plus DNA Ladder (Thermo Fisher Scientific).

To demonstrate its adaptability, Mu-DNA was optimised for samples where inhibition (PCR indices of ‘0’) was evident. Optimised protocols were then compared to the relevant commercial kit in fresh extractions from sample remnants.

Sample collection and preparation

A minimum of 2 l of water was collected from 13 shore sample sites around Windermere (Lake District, Cumbria, UK). Samples were transported on ice. Under sterile laboratory conditions, samples were thoroughly mixed by pouring and split into paired 1 l subsamples to be filtered. Filtering took place in less than 16 hours and filters were stored as above. DNA extractions followed the protocol of Mu-DNA: Water described above or DNeasy PowerWater. Identical lysis and purification conditions for both protocols were maintained: all filters were lysed in DNeasy PowerWater Bead Tubes and MB Spin Columns (Qiagen) were used for purification of all subsamples. Lysis was performed on a Vortex Genie (Scientific Industries) with Vortex Adapter (MoBio) at maximum speed for five minutes.

Library preparation

A double-indexed library was prepared following a 2-step PCR based protocol (Kitson et al. 2018) using primers for the vertebrate 12S mitochondrial gene region (Riaz et al. 2011, Kelly et al. 2014). In short, an initial PCR reaction amplified the target region using individually indexed 12S primers for each sample. To minimise PCR and sequencing bias, three sets of three PCR replicates per sample were performed to create three technical replicates with individual indices. Collection blanks were included in PCRs along with positive and negative controls. Replicate PCR products were pooled and purified using double-size selection with Mag-Bind RNXPure Plus beads (Omega Bio-tek) to remove nonspecific products and primer dimers. Final library concentration was assessed via qPCR assay using the NEBNext library quantification kit (New England Biolabs) and diluted as required to a final concentration of 4 nM. The final library was run at 15 pM concentration with 10% PhiX on an Illumina MiSeq using 600 bp V3 chemistry. A detailed protocol can be found in Supplementary file 1.

Bioinformatics and data analyses

metaBEAT, a custom bioinformatics pipeline (https://github.com/HullUni-bioinformatics/metaBEAT), was used to process sequencing outputs. The workflow consisted of the following steps: (i) demultiplexing; (ii) trimming, merging and quality filtering; (iii) chimera detection; (iv) clustering; (v) taxonomic assignment against a curated database. A low-frequency noise threshold approach was used to remove potential false positives from the metaBEAT data (Hänfling et al. 2016), only records exceeding a minimum proportion (0.001) of read counts in a sample were accepted as “true” positive records. Remaining reads were converted to relative species abundance (%) of assigned reads.

All statistical analysis was performed in R 3.2.5 (R Core Team 2016) using the VEGAN package (Oksanen et al. 2017). DNA yield and purity measures for extractions were compared with a linear model using planned contrasts between methods per sample. Metabarcoding of lake water DNA extractions were analysed using an analysis of similarity (ANOSIM) of relative species abundance across all replicates between methods. Non-metric multidimensional scaling (NMDS) ordination was used to visualise differences in extraction methods across all replicates grouped by site and extraction.

A cost per extraction was calculated for Mu-DNA: Soil, Tissue and Water. Costs per extraction were compared to those of DNeasy PowerSoil, DNeasy Blood and Tissue and DNeasy PowerWater, respectively. All costs used for comparisons were based on undiscounted list prices (GBP excluding VAT and shipping) for chemicals, plastics (excluding pipette tips) and Qiagen kits.

Raw data and scripts are available on Open Science Framework (https://doi.org/10.17605/osf.io/vrb4a). Sequencing data are available from NCBI Sequence Read Archive (Bioproject: PRJNA473636, SRA accession numbers: SRR7234627–SRR7234708).

Our Mu-DNA method exhibited similar, if not significantly higher, dsDNA yields than the DNeasy kits for most extractions (Figure 2). The DNeasy kit achieved significantly higher dsDNA yields than Mu-DNA only for Tissue B (woodlouse). A₂₆₀/A₂₃₀ ratios for all extractions were similar or significantly higher for Mu-DNA except Stool B, which was significantly lower than the DNeasy kit. Soil B had higher A₂₆₀/A₂₃₀ ratios from the DNeasy kit whereas Soil C had a significantly higher value from Mu-DNA. Stool A had higher A₂₆₀/A₂₃₀ values from Mu-DNA whereas Stools B and C had higher values from the DNeasy kit. All A₂₆₀/A₂₃₀ purity measures for tissue extractions were similar. All water sample extractions had higher A₂₆₀/A₂₃₀ measures from Mu-DNA. PCR inhibition (PCR index of ‘0’ in some or all extractions) was only detected in commercial kit extractions for Soil C and Mu-DNA extractions for Stools B and C. All other samples exhibited complete PCR success from both methods. Mu-DNA had a higher PCR index than the DNeasy kit for Tissue B despite having lower dsDNA yield. PCR inhibition was overcome by extraction dilution (1:10) for DNeasy kit extractions for Soil C and Mu-DNA extractions for Stool C. However, Mu-DNA extractions for Stool B failed to amplify across all extractions at any dilution tested, indicative of high level inhibition.

The DNeasy kits reliably extracted inhibition-free DNA from all sample types except Soil C (diesel polluted soil). Compared to this baseline of extraction success, our Mu-DNA protocols, with the exception of two samples (Stools B and C), performed similarly. Therefore the three basic Mu-DNA protocols we provide for soil, tissue and water are highly suitable for many sample types. Our unmodified protocols successfully extracted inhibition-free DNA from 10 out of 12 of the samples tested in this study. Modification of our protocols for the more challenging samples is described later (see Adaptability of Mu-DNA).

We used A₂₆₀/A₂₈₀ and A₂₆₀/A₂₃₀ UV absorbance measures via spectrophotometry to determine the quality of DNA extractions as suggested by Olson and Morrow (2012). The ideal measures for pure DNA are shown in Figure 2, yet in some cases, they are exceeded. These measures can be influenced by many aspects, such as invertebrate chitin (Athanasio et al. 2016) and RNA. Spectrophotometry of extracted DNA can be affected by the presence of co-extracted RNA by inflating A₂₆₀ values, therefore the ratios used for purity evaluations are skewed upwards. In our study, we refrained from the use of RNase so as to give a true representation of the method in an unmodified state. Should RNA-free DNA be required for any sample type, we suggest an RNase A treatment for a short incubation period (<1 hour) post-lysis. As purity measures can be affected by many factors, extracted DNA quantity and quality can therefore only reliably be ascertained by a combination of high-sensitivity dsDNA assays, gel electrophoresis visualisation of extracted DNA and intensity of PCR amplification success.

Figure 2.

Isolated dsDNA yield, purity and PCR index of samples used in the comparison of methods. Total dsDNA yield, A₂₆₀/A₂₈₀, A₂₆₀/A₂₃₀ ratios and PCR indices are shown for soil, stool, tissue and water samples per method. Horizontal dashed lines indicate ideal measures of A₂₆₀/A₂₈₀ and A₂₆₀/A₂₃₀ ratios for pure DNA. Asterisks indicate significant differences between methods (planned contrast linear model, p<0.05).

The highest yielding extractions per method (Qiagen DNeasy or Mu-DNA) from Soil C, Tissue A and Water B were selected for DNA integrity and molecular weight visualisation (Figure 3). All extractions had a molecular weight of approximately 10 kbp or higher. Similar integrity of extracted DNA was observed in the soil and water samples, however DNeasy Blood and Tissue had poor integrity compared to Mu-DNA: Tissue.

High molecular weight DNA extraction is desirable for many next generation sequencing (NGS) studies. It also allows for long range PCR amplification of whole mitochondrial genomes from eDNA samples (Deiner et al. 2017). Bead milling lysis has been shown to cause shearing of nucleic acids, resulting in low molecular weight of extracted DNA (Bürgmann et al. 2001). Our method yielded DNA of ≥10 kbp (Figure 3) in bead milled extractions but shearing is still present, evident in an extended smear. However, Mu-DNA protocols exhibited increased concentrations of higher molecular weight DNA than their commercial counterparts. Reducing bead milling times or enzyme digestion temperatures are both possible with Mu-DNA to reduce DNA shearing depending upon user end requirements. Additional measures can be taken to reduce the effects of physical and enzymatic shearing of DNA during sample preparation, extraction and even handling (see Klingstrom et al. 2018), yet these could become time-consuming for very large sample numbers.

Figure 3.

Integrity and molecular weight of soil, tissue and water sample extractions from the methods compared in this study. Shown are the highest yielding extractions per method from Soil C, Tissue A and Water B. Extractions are indicated by the relevant method for sample type; DNeasy (Q) or Mu-DNA (M).

PCR inhibition was present in DNA extractions of two samples for the Mu-DNA protocol: Stools B and C. The modular aspect of the Mu-DNA method was employed to optimise extractions for each of these samples to achieve complete initial PCR success. For Stool B, a tissue lysis stage that incorporated bead milling was used. A 0.25 g sample was added to 0.5 g of 1 – 1.4 mm garnet beads. A 2.5 x volume tissue lysis mixture was added. Soil protocol bead milling was performed followed by overnight tissue protocol incubation. The extraction then followed the soil protocol with a tissue protocol wash stage. For Stool C, the soil protocol was modified with a tissue protocol wash stage. These modifications improved DNA purity for both sample types with successful PCR amplification (Figure 4).

Figure 4.

Optimised Mu-DNA protocols for stool samples that previously failed to achieve inhibition-free DNA. Optimised protocols are compared to DNeasy PowerSoil. Total dsDNA yield, A₂₆₀/A₂₈₀, A₂₆₀/A₂₃₀ ratios and PCR indices are shown for stool samples B and C. Horizontal dashed lines indicate ideal measures of A₂₆₀/A₂₈₀ and A₂₆₀/A₂₃₀ ratios for pure DNA. Asterisks indicate significant differences between methods (planned contrast linear model, p<0.05).

The modular adaptability of Mu-DNA allows for its application across different sample types or integration into existing protocols. For example, Spens et al. (2016) made use of DNeasy Blood and Tissue in their protocols for water filters, including Sterivex filters. We adapted and tested these protocols (data not shown) and found them to be easily changed to use a Mu-DNA: Tissue/Water protocol. For this, we recommend beginning with a tissue protocol lysis, adjusting the volumes as required, then following the water protocol from inhibitor removal through to elution. In this way, contaminants are greatly reduced and there is no need for extra purification of extractions (unlike with Niemiller et al. 2017). However we found that neither DNeasy Blood and Tissue nor the adapted Mu-DNA: Tissue/Water protocol could achieve inhibition-free DNA from turbid, algal rich waters (Water B) as effectively as DNeasy PowerWater or Mu-DNA: Water (data not shown). Solid phase reversible immobilisation (SPRI) DNA purification, based on Rohland and Reich (2012), can achieve higher DNA yield and purity than spin column based protocols (Vo and Jedlicka 2014). The Mu-DNA method can be easily converted to SPRI purification by replacing the silica binding step with an SPRI protocol. However Vo and Jedlicka (2014) found SPRI to only have improved performance with less contaminated samples, such as avian oral and cloacal swab extractions. SPRI DNA purification is therefore best reserved for relatively clean environmental sample types, in particular clear lake and stream waters or tissue samples (see Mayjonade et al. 2016).

Our modular approach to DNA extractions is not a new concept. Lever et al. (2015) developed a modular extraction method for multiple environmental samples. Although a more complex protocol, it is nonetheless highly efficient and many aspects of the study can be applied to Mu-DNA. For example, fine tuning of pH and phosphate concentration for lysis of specific sample types could lead to increased DNA yields. Our method uses chemical flocculation of inhibitors from extracted DNA and is pH sensitive (see Dong et al. 2006). For this reason, we did not explore the higher pH lysis of Lever et al. (2015) and it remains an aspect open for future investigation.

After the application of noise filtering thresholds to read count data, both methods detected the same 15 fish species previously recorded in Windermere (Hänfling et al. 2016). Broadly, individually sequenced samples cluster by site when visualised with NMDS ordination with some variance between replicates (Supplementary file 2). Although species detected varied between method replicates per site (Supplementary file 3), there was no significant difference between methods in overall species relative abundance (ANOSIM: R = -0.02, p = 0.93) and both methods produced high similarity species profiles for the lake as a whole (Figure 5). This shows that Mu-DNA produces DNA of sufficient quality for metabarcoding approaches even when target DNA concentration is low and that no bias is introduced through the choice of extraction method.

Figure 5.

Species profiles of Windermere from metabarcoding of extractions using the compared methods of this study. Relative species abundance (%) of assigned reads is given per method; DNeasy PowerWater or Mu-DNA: Water. Positioning of species is arbitrary and arranged alphabetically. Diamonds indicate the position of low abundance species in the profiles for both methods.

Mu-DNA protocols cost less per extraction than the commercial kits to which they were compared (Table 2). Initial consumable costs for our method are higher than purchasing a single commercial kit yet the number of extractions covered by this cost is considerable (see Supplementary file 4). For the cost comparison, institutional discounts were not considered. Were they to have been taken into account, the cost of Mu-DNA would be appreciably lower. Lower costs, combined with a modular application across multiple sample types, makes the method an attractive alternative to commercial kits.

Table 2.Download as CSV 

Cost per extraction for Mu-DNA protocols and the commercial kits compared in this study.

	Cost per extraction (GBP)
DNeasy PowerSoil (100)	5.24
Mu-DNA: Soil	0.71
DNeasy Blood and Tissue (250)	2.92
Mu-DNA: Tissue	0.67
DNeasy PowerWater (100)	7.03
Mu-DNA: Water	0.83