Unlike other environmental samples, activated sludge is composed of nearly all microbial cells and their products (Frølund et al. 1996
). The cells cluster together and are enclosed with EPS, which can protect them from shear forces and chemical reagents including sodium dodecyl sulfate (Davies et al. 1998
). In terms of productivity and diversity, results from this study showed that the mechanical homogenization (the bead-beating step) is obviously necessary for DNA extraction from sludge samples.
The factors that affect DNA yield for a kit are mainly the efficiency of the cell lysis step and the subsequent losses during purification. The five kits with the bead-beating step have minor differences in the lysis process except that the FA-SS kit contains glass beads with different sizes (0.1–1 mm in diameter). The big glass beads may be efficient for dispersing cells from clusters, and the small ones are dedicated to crush the cells. However, the FA-SS kit also contains a unique matrix that specifically binds DNA, whereas all of the other six kits just adopt a spin column to bind DNA. The two unique designs in the FA-SS kit may promote the quantity and quality of the extract DNA from AS, as indicated by the result.
The quantity of DNA is usually not of great concern for PCR-based community analysis because even as low as 10–100 ng DNA (equals to about 106–107 cells) is already enough for amplification and then sequencing. However, for the current metagenomic study sequenced by the Illumina platform, 3–10 μg virginal, highly pure DNA is needed. This makes the FA-SS and MB-PS the only two candidate kits. Dramatically, our result showed that the low quantity and even the low quality of the extracted DNA could also provide a fair profile of the bacterial community. For example, the ZR-SM kit produced very low concentrations of DNA with low quality (ratio of OD260/OD280 around 1.0), but the OTU-based and taxonomic analysis indicated that the results reflected reasonably the major bacterial community profile, with only a slight difference from the three MoBio kits and FA-SS kit. However, the QG-ST and EP-SM kits that also extracted low quality and quantity of DNA showed much higher biases on the community profile compared with the other five kits. This suggests that the ZR-SM kit may be efficient in cell lysis, but loses much DNA during the subsequent purification steps, which is a random event. Thus, it does not affect the community structure. It is noteworthy, however, that all the kits could be utilized efficiently by changing some of the operations. For example, the EP-SM kit performs the centrifugation at 1,000–2,000 g in some cases, which may be fair for soils (the density is much higher than activated sludge), but unsuitable for sludge samples, and could cause loss of sample. Increasing the strength of the centrifugation may increase the yield for this situation.
Other than yield and fidelity, a co-existing problem is that the DNA extract from commercial kits are usually small in segment size. This may be the result of the high shear force during the bead-beating or vortex processing. Small pieces of DNA are not suitable for construction of the fosmid, cosmid, and BAC libraries that prefer genomic DNA fragments over 25 kb, which are usually extracted by lab-developing methods (Robe et al 2003
). Moreover, if the extracted DNA is used in full-length 16S rRNA gene (~1.5 kb) amplification, ~10 kb-sized fragments theoretically lose about 15 % of the genes. However, the current high throughput sequencing will be little affected because of the short amplified regions (mostly <400 bp).
For environmental samples, especially for those containing bulking water, fixation is needed before long-term transportation and storage. DNA may be altered in two different ways without fixation: (1) The bacterial community may change rapidly during transportation and storage because of the change of environment, and (2) DNA may leak out from cells that die during transportation and storage into the bulking water and then be washed away. For sludge samples, fixation in 50 % ethanol (final concentration) was recommended, which is the same as sample fixation for fluorescence in situ hybridization (FISH) (Xia et al. 2007
). Another advantage is that 50 % ethanol would not be frozen at –20 °C. The results in this study showed that the fixation could improve DNA yield, although the reason is unclear. In addition, most of the slight variation of the bacterial community between the fixed and nonfixed treatments could not be attributed to the fixation. It could arise from the biases of PCR or sequencing.
The total OTU number and the diversity indexes based on the total OTUs could not be the key criteria for the evaluation of the efficiencies of DNA extract kits, especially under the conditions that not all species were detected by sequencing at insufficient depth, considering the extremely high diverse bacteria in activated sludge. In fact, under 46,734 sequencing depth, the ~4,000 OTUs in each treatment usually had about 50–70 % singletons and >
90 % OTUs containing <10 tags (data not shown), which were obviously rare groups with little significance, having abundances of 0.002–0.02 %). The more OTUs and higher diversity indexes may represent more bacterial species at the price of biases on the abundances of the dominant or subdominant groups if the kits could not extract DNA effectively from certain such groups. On the contrary, the top 500 and 50 OTUs usually accounted for more than 80 and 50 % of total tags, respectively. Thus, they are more suitable to evaluate the efficiency of the kits. Under these conditions, the five kits with the bead-beating step are significantly better than the two kits that only used lysis buffer. However, the differences among the five kits need to be determined by taxonomic analysis.
The Gram-positive bacteria are resistant to both detergents and mechanical resistance because of their thick cell wall (Bollet et al. 1991
) or because some of them can form spores (Kuske et al. 1998
). Therefore, it could be simply considered that the more Gram-positive bacteria are detected, the more efficient the DNA extraction kits are. In terms of this, the most efficient two kits were FA-SS and MB-US because more Gram-positive Actinobacteria
were detected in the two treatments at both the phylum and genus levels than the others. Actinobacteria
is an ubiquitously dominant phylum in AS and plays key roles in polymer degradation, glycogen accumulation, and polyphosphate accumulation (Seviour and Nielsen 2010
). A study that used untreated sludge to perform PCR and cloning could not detect Actinobacteria
, although about 13 % of the cells belonged to this phylum, as determined by FISH (Snaidr et al. 1997
). Another study that adopted denaturing gradient gel electrophoresis as the DNA extraction evaluation method treated this high G
C phylum as a key indicator for DNA extraction methods (Niemi et al. 2001
). All these suggested the abundance of Actinobacteria
could be a key sign for efficiency of DNA extraction, especially for cell lysis. Recently, a high-throughput sequencing, metagenomic study of AS found biases when comparing the sequencing data with the results from the FISH method (Albertsen et al. 2011
). Very interestingly, the study also found that the Actinobacteria
are seriously underestimated in the metagonomic data comparing with the FISH results, although the FA-SS kit was adopted. A flaw emerged that bead-beating was performed only for 3
5 s. This observation indicates that the operational time for bead-beating should also be concerned. By contrast, the detection of the other Gram-positive phylum, Firmicutes
, was minimally affected by different kits, even for the two inefficient kits. Also interestingly, a genus that belongs to the Firmicutes
, was more abundant in treatments of the two inefficient kits. This observation indicates the different efficiencies of the kits in detecting various subgroups of Gram-positive bacteria. The other underestimated phylum, Nitrospirae
, has a special wide periplasmic space, which is near twice that in other Gram-negative bacteria (Watson et al. 1986
). This structure may hinder the release of DNA following inefficient cell lysis treatment. Similarly, DNA from Chloroflexi
was hard to extract, possibly also because of the layered cell envelopes (Sutcliffe 2011
). However, the reason for the underestimation of Alphaproteobacteria
Moreover, the results from the inefficient kits are also valuable because they imply which groups tend to be overestimated. In this study, the Gammaproteobacteria, Deltaproteobacteria, Bacteroidetes, and many rare phyla were overestimated in the two kits with low efficiency. This implies that bacteria within these groups are more likely to be destroyed and, thus, overestimated if inefficient methods are adopted to disrupt all the bacterial cells. Since the usual high richness of Gammaproteobacteria and Bacteroidetes in many environmental samples was found, it is noteworthy whether they were overestimated by the inefficient DNA extraction to some extent.
In summary, in the light of our results, the FastDNA@ SPIN kit for Soil is recommended for DNA isolation of activated sludge samples because of its high yield, purity, and excellent cell-breaking capability. Although the number of total OTUs from this kit was not high, the major groups and Gram-positive bacteria that were identified indicated its reliability compared to other kits. The three MoBio kits and ZR-SM kit were also fair, but insufficient in yield and/or purity, which are essential for current metagenomic studies. In addition, the results also proved that the bead-beating step is necessary for activated sludge samples because some phyla, such as Actinobacteria and Nitrospirae, are significantly resistant to the simply chemical cell lysis treatment. Careful selection of extraction kits or methods should be considered if these phyla would exist dominantly in certain environmental samples.