Comparison of four DNA extraction kits efficiency for 16SrDNA microbiota profiling of diverse human samples

Background

A paradigm shift in the origin of many human diseases occurred 15 years ago when the role of the different microbiotas inhabiting the host (intestinal, vaginal, skin, etc.) on its own physiology and in the pathophysiology of numerous diseases has emerge [1]. The term ‘human microbiota’ has gained enormous attention in research since then, reaching more than 32,000 articles in Pubmed in the last 5 years. Indeed, the drop in sequencing cost and the development of bioinformatical tools simplified the analysis of microbiota for many laboratories. Several protocols for sampling, DNA extraction, sequencing, and bioinformatic analysis are now available. However, this multitude of protocols does not allow easy comparisons between studies nor the pooling of data in an open-science strategy.

One significant cause of variability between studies analyzing microbiota through amplicon sequencing is DNA extraction process, next after sample type and origin and storage methods [2,3]. Several kits are available on the market, whose protocols share similar pipelines: lysis of bacteria through mechanical, chemical or enzymatic methods, binding of DNA on columns, several wash steps to remove proteins and other contaminants and final elution. The type of bacterial wall disruption technique can affect DNA yield, quality of DNA as mechanical lysis can disrupt all types of bacteria but likely fragment genomic DNA [4]. On the other hand, enzymatic and chemical lysis methods are less efficient to lyse bacteria and might result in under-estimation of tough-to-disrupt bacteria, yet preserving DNA quality [5]. DNA binding column characteristics and washing solution composition also differ between kits, resulting in different DNA yield and purity.

In recent years, more attention has been given to the standardization of the workflow of microbiome research [2,6,7]. However, different internal extraction kits and methods are used in different laboratories. Recently, Costea et al. evaluated 21 methods for extracting DNA from three continents and suggested a protocol, called the Q protocol, as the gold standard for human stool samples [6]. They reported that it was not known if this method was optimal for samples other than fecal matter, such as low biomass samples.

Our objective was therefore to assess the efficiency of 4 commercially DNA extraction kits on bacterial DNA quantity, quality and to analyze the impact on bacterial composition and diversity of different samples. We checked DNA extracted with those methods quantity and quality and performed 16SrDNA sequencing using Illumina technology followed by taxonomic assignment to analyze differential abundance and alpha and beta diversity.

Methods

Sample collection

The outline of the study design is presented (Figure 1). Samples were divided in five groups from different collection sites which differed in term of microbiota density: chyme (mix of gastric juice and partly digested food passing from the stomach to the small intestine [n = 4]), feces (n = 5 from infants; n = 5 from adults), bronchoalveolar lavage fluid (BAL) (n = 3), sputum (n = 3). Samples were collected following the recommendations of the International Human Microbiome Standards (IHMS) (www.microbiome-standards.org). When received at the laboratory, BAL and sputum samples were centrifuged for up to 10 minutes at 8,000 × g to concentrate the bacterial cells, discard most of the supernatant, leaving only leaving only 200 μl. Fecal samples were manually homogenized and weighed into separate aliquots. Then all the samples were stored at 80°C until DNA extraction.

DNA extraction

Samples were processed using four commercial DNA isolation kits. QIAamp PowerFecal Pro DNA kit (QPFPD, QIAGEN), Macherey Nucleospin Soil (MNS, MACHEREY-NAGEL) Macherey Nucleospin Tissue (MNT, MACHEREY-NAGEL) and MagnaPure LC DNA isolation kit III (MPLCD, ROCHE) were used following manufacturers' instructions for each kit. A peanut-sized stool samples (∼200 mg) suspended in 100 μl of PBS, 200 μl of liquid samples (BAL and sputum) and 500 μl of chyme were used for each extraction.

QPFPD

Each sample as described before was introduced in bead-containing PowerBeads Pro Tubes with lysis buffer. These mixtures were subjected to mechanical lysis for 5 minutes. Different buffers were added before transferring the mixture to MB Spin Columns and centrifuged. After different washing steps of the column, DNA was eluted with 100 μl C6 buffer and stored at -80°C.

MNS

In this method, each sample were introduced in tubes containing type A beads. To lyse each sample and precipitate contaminants, different buffers were added to samples. These mixtures were subjected to mechanical lysis for 5 minutes. Obtained lysate was then filtered and DNA bind to a NucleoSpin Soil column. After a few washes of the column, DNA was eluted with 100 μl SE buffer and stored at -80°C.

MNT

This extraction kit uses proteinase K for lysis. Proteinase K solution was added to each sample which was then thoroughly mixed and incubated at 56°C during 1 h then at 70°C during 10 min. To precipitate DNA, ethanol 100% was added and mixture were subjected to mechanical lysis for 5 minutes. To bind DNA, lysate was load onto a NucleoSpin Tissue column. After a few washes of the column, DNA was eluted with 100 μl BE buffer and stored at -80°C.

MPLCD

In this DNA extraction kit proteinase K was used, associated with lysis buffer. This buffer was added to each sample and samples were incubated for 10 min at 65°C and another 10 min at 95°C. To homogenize these lysates, they were subjected to mechanical lysis for 30 s. Lysate were then transfer in the wells of the Sample Cartridge, each well contains magnetic-bead. DNA bind to these Magnetic Glass Particles (MGPs). Wash Buffer is then used to wash MGPs with bound DNA to remove unbound substances. Purified DNA was eluted with a 100 μl of low salt elution buffer and stored at -80°C.

Each kit had different types of cell lysis and DNA binding methods as summarized in Table 1.

Assessment of quantity & quality of DNA

DNA concentration

The concentration of DNA obtained after the different extraction procedures for each sample was quantified using a Qubit 3.0 fluorometer with the Quant-iT dsDNA HS Assay (assay range between 0.2 and100 ng; sample concentration between 10 pg/μl and 100 ng/μl) and the Quant-iT dsDNA BR Assay (assay range between 2 and 1000 ng; sample concentration between 100 pg/μl and 1000 μg/μl), according to manufacturer's instructions (ThermoFischer Scientific).

DNA quality

DNA quality was determined using a NanoDrop ND-1000 [8] after calibration according to the NanoDrop 2000–2000c & 1000 Calibration Check procedure (ThermoFisher Scientific). The optical density (OD) ratio between 260 nm and 280 nm was used to asses acidic pH or contamination (proteins, phenol, other impurities) for samples with a ratio below 1.5–2.0. The OD 260 nm/230 nm was also used to check contamination by polysaccharides, phenol or low pH [8,9] for samples with a ratio below 1.75.

DNA quality was also determined on 1 μl of total DNA using Agilent 4200 Tape Station (Agilent), which separates DNA fragments by size on a chip. Laser induced fluorescence is then used to translate bands in peaks to evaluate the distribution and relative amounts of DNA of different sizes.

DNA sequencing

The V3–V4 region of the 16S rRNA gene was amplified by PCR using F343 and R784 primers (F343: CTTTCCC TACACGACGCTCTTCCGATCTTACGGRAGGCAGCAG, R784: GGAGTTCAGACGTGTGCTCTTCCGATCTTACCAGGGTATCTAATCCT) as previously described [10]. Briefly, purified genomic DNA (gDNA) was amplified with the primers F343 and R784 using 30 amplification cycles with an annealing temperature of 65°, producing amplicons of 510 bp, although the exact length varies depending on the organisms. Extremely high-quality, full-length reads of the entire V3 and V4 region were generated thanks to the overlapping of the MiSeq paired 250-bp reads, in a single run. Home made 6 bp index added to R784 during a second PCR (12 cycles, forward primer (AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGAC) and reverse primer (CAAGCAGAAGACGGC ATACGAGAT-index-GTGACTGGAGTTCAGACGTGT) was used to perform single multiplexing After purification of the PCR products, they wereloaded onto the Illumina MiSeq cartridge according to the manufacturer instructions. PhiX was used to check the quality of the run internally., Each pair-end sequences were assigned to its sample thanks to the previously integrated index. Flash software [11] was used to assemble each pair-end sequences using at least a 10 bp-overlap between the forward and reverse sequences, allowing 10% of mismatch. These sequencing and cleaning steps were performed at the GeT-PlaGE platform (Toulouse, France) as described by Zemb et al., 2020 [12].

The resulting sequences were processed using QIIME2 pipeline (v. 2018.4, https://qiime2.org/) after importation of reads as “PairedEndFastqManifestPhred33” format, read denoising with DADA2 and truncation of the forward and reverse sequences at 240 bases and all other parameters set to default. Chimeric sequence were detected, non-16S rDNA gene identified n, and amplicon sequence variants (ASVs) with a similarity threshold of 99% clustered using the SILVA (release 132) reference database [13,14]. ASVs with a total frequency <10 in the entire table and present in only 1 sample were removed, as well as reads classified as Archaea or Unassigned.

After cleaning and filtering of the reads, only the samples corresponding to the stools could be analyzed since too few reads passed the quality controls for other samples.

The core diversity analysis was performed with QIIME2 diversity core-metrics-phylogenetic plugin, with a specific sampling depth, 5900 reads for the children stool samples and 6800 reads for the adult stool samples.

Statistical analysis

To assess the influence of DNA extraction kits on DNA quantity, quality (i.e., A260/280 and A260/230), Chi2 or Kruskall-Wallis tests were perfomed. Species richness (Chao1) and evenness (Shannon index) were calculated for alpha diversity estimations and compared using Kruskal-Wallis tests. Permutational multivariate analysis of variance (PERMANOVA with the Adonis function) on beta diversity distance matrices (Bray Curtis, Unweighted Unifrac) was used to evaluate the impact of the 4 different extraction kits on microbiota composition variability between samples. A P-values <0.05 was considered significant. To compare the taxonomy composition between different extraction kits Kruskall-Wallis or Wilcoxon tests were performed.

Results

Influence of DNA extraction kits on the quantity & quality of DNA

Pairwise comparisons indicated that the MPLCD kit yielded the lowest amount of DNA (p < 0.01, Figure 2) for stool samples (children and adult) compared with the other kits. No significant difference between extraction kits was observed for the other samples. Noteworthy, DNA concentration was very low for non-stool samples, especially for chyme ones (<10 ng/μl), irrespective of the extraction kit (Figure 3).

The impact of extraction kits on DNA quality was then assessed via the measurement of A260/A280 and A260/A230 ratios using a Nanodrop or via fragment size distribution using a Bioanalyzer. Nearly all the samples exhibited an A260/A280 ratio greater than 1.5, which is indicative of relatively pure DNA, except for chyme samples extracted with the MNS, MNT and QPFPD kits and BAL samples extracted with the QPFPD kit (Table 2).

MNT was the kit that produced the highest percentage of DNA extracts with a A260/A230 ratio greater than 1.75, indicative of less residual carryover compared with the other kits (Table 3). Noteworthy, all the DNA extracts produced with the MPLCD kit exhibited a A260/A230 ratio below 1.75, irrespective of the type of samples. Whatever the DNA extraction kit used, all the A260/230 ratio of DNA extracts from chyme samples were under 1.75.

Stool samples were analyzed with a Bioanalyzer DNA high-sensitive chip (Agilent), which provides information on DNA fragment size distribution. Bioanalyzer images of the libraries showed a homogenous smear of DNA from 1500 to 10000 bp for the QPFPD, MNS and MNT extraction kits (data not shown). For MPLCD, all samples showed a similarly average size <500 pb (data not shown). Other samples (chyme, BAL and sputum) were not analyzed due to the poor quantity of the DNA extracted.

Influence of DNA Extraction Kits on Microbial Diversity & Composition

Diversity Analysis

A total number of 609,773 raw sequences was obtained, varying from 0 to 23,136 sequences by samples (Table 4). The number of reads for chyme, sputum and BAL samples was generally very low. Therefore, no core metrics analyze was possible and no data are presented for these samples. To determine if DNA extraction kits impact beta diversity, PCoA plots were constructed based on Bray–Curtis and unweighted UniFrac distance matrices to assess if samples clustered together or not. As shown in Figure 3A & B, samples clustered according to the individual and not the DNA extraction kits. Beta-diversity analyses did not show any significant sample clustering based on extracted method (p > 0.05) (Table 5).

Furthermore, we assessed the impact of the different extraction kits on alpha-diversity using the Chao and Shannon indices. No difference was found among the different DNA extraction kits for both alpha-diversity metrics (p > 0.05, Figure 4).

Taxonomic Identification

Taxonomic compositions of the microbiota were compared with evaluate if the different DNA extraction kits affected community composition. When annotated at the phylum, family or genus levels, the microbial profile detected in children and adult stools demonstrated no significant differences between QPFPD, MNS, MNT and MPLCD (p > 0.05, Figure 5).

Discussion

Our objective was to asses if the type of DNA extraction kits used to extract DNA before 16S rDNA sequencing can impact microbiota diversity and composition analysis of diverse human samples. We used 4 commercially available kits to extract DNA from low microbiota concentration samples (sputum, BAL, chyme) or high microbiota concentrations ones (children and adult stools) to compare their efficiency in terms of DNA yield, DNA quality and microbiota diversity after sequencing. The 4 commercially available kits used in this study extracted bacterial DNA with some variabilities. Among the 4 kits, MPLCD was the less efficient in terms of amount and quality of DNA, whereas the other 3 kits resulted in a better quantity and quality of DNA.

The DNA extraction kits tested differed in terms of lysis technique (mechanical lysis and/or chemical lysis and/or heating) and DNA binding technique (columns vs magnetic beads). QPFPD and MNS both involves bead beating while MNT and MPLCD relies on proteinase K and mechanical (vortex) lysis and heating. Among factors contributing to DNA quantity and quality variations when extracting microbial nucleic acids, lysis technique seems a key step [15,16]. However, the literature is still controversial. Indeed, while only few significant differences between DNA extraction methods when assessing the role of lysis technique were observed by Claassen et al., Yuan et al. reported that bead beating and/or mutanolysin was the most effective cell lysis and DNA recovery technique [17,18]. A combination of lysing procedures seems a good option to capture the most accurate community composition since hot phenol and bead beating were described as the most effective lysing procedure in another study [19]. A bead-beating step during DNA extraction has been shown to be efficient to break down the cell wall of Gram-positive bacteria [20]. QPFPD and MNS uses bead beating as mechanical lysis. By contrast, MPLCD and MNT combine proteinase K buffer and no bead-beating. In our study, the additional bead beating step in QPFPD and MNS did not influence the relative abundance of phyla in these samples, as expected by Gryp et al. [21]. Indeed, Gryp et al. showed that the Qiagen PowerMicrobiome kit combining bead beating with proteinase K and mutanolysin was more effective than without pre-treatment. The QIAmp DNA Stool Minikit, with a beat-beating step and the QIAmp PowerFecal Pro DNA Kit provided comparable results in terms of DNA quantity and microbiome profiles [22].

The second factor that could have affected DNA yield and quantity is the type of technique used to trap DNA: columns are used in the QPFPD, MNS and MNT kits while magnetic beads are used in the MPLCD kit. Spin Column technology relies on the affinity between negatively charged DNA and positively charged silica material. This results in selective binding of DNA to the silica matrices and wash-out of the rest of the cell components and other chemicals. These methods include steps of lysis then centrifugations for the purpose of DNA binding. DNA is finally eluted from the silica matrix by any hyposmotic solution, such as nuclease-free water in the QPFPD, MNS and MNT kits. The different washing and elution steps are achieved to purify and elute a high amount of nucleic acids as well as to recover small fragments DNA efficiently. This technology has been shown to result in high-purity DNA with great quantitative and qualitative reproducibility [23].

Magnetic beads technology is an alternative way for DNA extraction. This technology is widely used in automated extraction procedures for a large number of samples [24,25]. The technique is based on magnetic particles, whose surface is positively charged and binds DNA at pH ≤6.5. A change in pH to pH ≥8.5 allows the release of the DNA bound to the particles [26]. Functionalized magnetic particle or beads coupled to adequate buffers have been developed to result inefficient DNA extraction. One positive aspect of this technology is the absence of centrifugation steps possibily producing shear forces, causing breaking of nucleic acids.

Even if, MPLCD was the only kit using magnetic beads and yielded the lowest DNA quantity, magnetic bead DNA extraction method is not less efficient in DNA extraction. In accordance with Knudsen and al [27]. DNA concentration is independent of DNA separation types. In their study, the highest DNA concentrations for human feces were obtained using the Easy-DNA (phenol-chloroform precipitation), MagNA Pure (magnetic beads), QIAamp DNA stool minikit (silica membrane- based columns) procedures. On average across the types of specimen (Human, pig Sewage), the highest DNA concentrations were obtained using Easy-DNA (Phenol-chloroform precipitation) and QIAStool (Silica membrane- based columns), and the lowest were obtained using the PowerSoil.HMP (Silica membrane-based columns) and InnuPure (Magnetic beads) methods.

In our study, the extraction kit used did not result in any difference in in microbial community diversity, considering the Shannon or Chao diversity indices. Thus, probably because of the amplification step before 16S rDNA sequencing, the quantity of DNA yielded by the extraction method does not constitute an attention point when choosing a DNA extraction kit. In accordance with Knudsen and al, there was no significant correlation between the amount of DNA obtained and the community diversity or richness measured [27]. Using synthetic consortia, Ducarmon et al. showed that the different DNA extraction methods they used allowed them to retrieve the theoretical relative bacterial abundance but with, differences according to the methods they used, while little variation were seen with the type of library preparation and sequencing they tested. Moreover, the bioinformatic pipelines they used resulted in different results for observed richness, but diversity and compositional profiles were comparable [28].

Our results were also in contrast with Kennedy and al study [29], where significant differences in DNA yield and bacterial composition for kits used to isolate bacterial DNA from stool were observed. However, the PCR products were sequenced by the Roche 454 and not by Illumina. The lower error rates, higher throughput [30] and higher read quality [31] obtained with Illumina technology, results in higher quality data and could explain these differences.

Nucleic acid quality and quantity differed significantly not only between the DNA extraction kit method [27,32,33], but also by sample types differing in bacterial biomass. Chymes, BAL and sputum can be considered low biomass samples [34–36]. The complexities of 16S rDNA sequencing from low biomass samples are increasingly recognized and have broad applicability [34,37–40]. In our hand, none of the extraction kits tested with manufacturer conditions was efficient to extract DNA with sufficient quality and quantity for further analysis DNA extraction protocols should probably be adapted to this type of very demanding samples, by adapting the storage, the mode of lysis, the buffers used, etc. Moreover, it seems important to introduce Mock standards in these methods, to prevent possible contamination of these low biomass samples with high biomass samples (stools) that can be extracted in parallel [41].

Conclusion

Although MPLCD was the less efficient kit in terms of DNA quantity and quality, it was as efficient as QPFD, MNS and MNT kits to capture microbial richness and diversity for stool samples. These results will prove useful for researchers or practitioners investigating the microbiota in selecting an alternative kit to an other kit or a discontinued kit. These highlight the potential comparable results when cross-comparing studies that use these different kits.