We describe a protocol for filtration of water samples with a filter cartridge and extraction of environmental DNA (eDNA) without having to cut open the housing to remove the filter. This protocol is developed for metabarcoding eDNA from fishes, but is also applicable to eDNA from other organisms.
Recent studies demonstrated the use of environmental DNA (eDNA) from fishes to be appropriate as a non-invasive monitoring tool. Most of these studies employed disk fiber filters to collect eDNA from water samples, although a number of microbial studies in aquatic environments have employed filter cartridges, because the cartridge has the advantage of accommodating large water volumes and of overall ease of use. Here we provide a protocol for filtration of water samples using the filter cartridge and extraction of eDNA from the filter without having to cut open the housing. The main portions of this protocol consists of 1) filtration of water samples (water volumes ≤4 L or >4 L); (2) extraction of DNA on the filter using a roller shaker placed in a preheated incubator; and (3) purification of DNA using a commercial kit. With the use of this and previously-used protocols, we perform metabarcoding analysis of eDNA taken from a huge aquarium tank (7,500 m3) with known species composition, and show the number of detected species per library from the two protocols as the representative results. This protocol has been developed for metabarcoding eDNA from fishes, but is also applicable to eDNA from other organisms.
Environmental DNA (eDNA) in aquatic environments refers to genetic material found in the water column. Recent studies demonstrated the utility of eDNA for detecting fishes from various aquatic environments, including ponds1-3, rivers4-8, streams9, and seawater10-14. Most of these studies focused on detection of a single or a few invasive1,4-6,8,14 and rare or threatened species3,9, while some recent studies attempted simultaneous detection of multiple species in local fish communities7,9,12,13,15 and mesocosms11,12.
The latter approach is called "metabarcoding" and eDNA metabarcoding uses one or multiple sets of PCR primers to coamplify a gene region across taxonomically diverse samples. This is followed by library preparation with indexing and adapter addition, and the indexed libraries are analyzed by a high-throughput parallel sequencing platform. Recently Miya et al.12 developed universal PCR primers for metabarcoding eDNA from fishes (called "MiFish"). The MiFish primers target a hypervariable region of the mitochondrial 12S rRNA gene (163-185 bp), which contains sufficient information to identify fishes to taxonomic family, genus and species except for some closely related congeners. With the use of those primers in eDNA metabarcoding, Miya et al.12 detected more than 230 subtropical marine species from aquarium tanks with known species composition and coral reefs near the aquarium.
While optimizing the metabarcoding protocol to accommodate natural seawater with varying levels of eDNA concentration from fishes, we have noticed that the MiFish primers occasionally failed to amplify the target region for subsequent library preparation. One of the more likely reasons for this unsuccessful PCR amplification is lack of adequate amounts of the template DNA contained in small volumes of water filtered (i.e. 1-2 L). Although eDNA concentration from a specific taxonomic group is unknowable before the amplification, filtration of large water volumes (>1-2 L) would be a simple and effective means to collect more eDNA from the aquatic environments with scarce fish abundance and biomass, such as open-ocean and deep-sea ecosystems.
Relative to disk fiber filters conventionally used in a number of fish eDNA research16, filter cartridges have the advantage of accommodating larger water volumes before clogging17. Actually, a recent study showed large volume (>20 L) filtration of coastal seawater samples using filter cartridges18. In addition, they are individually packaged and sterile, and several steps of the experimental workflow can be performed in the filter housing, thus reducing the probability of contamination from the laboratory19. The latter feature is critical for eDNA metabarcoding, in which the risk of contamination remains among the greatest experimental challenges20,21. Despite these technical advantages of filter cartridges, it has not been used in eDNA studies of fishes with two exceptions8,15.
Here we provide a protocol for filtration of water samples with the filter cartridge and extraction of eDNA from its filter without having to cut open the housing. We also provide two alternative water filtration systems depending on the water volumes (≤4 L or >4 L). To compare the performance of the newly-developed protocol and a previously-used protocol using a glass-fiber filter in our research group12,14,22,23, we perform eDNA metabarcoding analysis of seawater from a huge aquarium tank (7,500 m3) with known species composition, and show the number of detected species derived from the two protocols as representative results. This protocol has been developed for metabarcoding eDNA from fishes, but is also applicable to eDNA from other organisms.
NOTE: This protocol does not deal with water sampling and metabarcoding methods. Water may be sampled in different manners depending on study purposes16 and see Miya et al.12 for details of the metabarcoding methods using MiFish primers. Note that the sampled water should be kept very cold and filtered within a few hours to avoid degradation of eDNA. Also note that this protocol involves the use of a rotary shaker and an incubator, and the latter must be large enough to accommodate the former. In addition, a centrifuge that can accommodate both 15 ml and 50 ml conical tubes is indispensable to remove the remaining liquid from the post-filtration filter and to collect extracted DNA within the cartridge, respectively.
1. Processing a Screw Cap and a 1 L Plastic Bag
NOTE: Skip this step if the filtration volume is >4 L.
2. Assembly of the Filtration System
3. Filtration of Water Samples (≤4 L) using the Filter Cartridge
NOTE: Skip this step if the filtration volume is >4 L. This filtration system requires a self-standing panel for hanging the plastic bag filled with 1 L of water. A mesh panel, multiple prongs, and a stand for the panel, all available from online stores, would be useful for assembling this unit. Autoclave the inlet and outlet luer caps for the filter cartridge before use.
4. Filtration of Water (>4 L) Samples using the Filter Cartridge
Note: Skip this step if the water filtration volume is ≤4 L. This filtration system requires a 10 L book bottle equipped with a valve and a disposable 10-ml pipette tip. An inner diameter of the 10 ml pipette tip (15.0 mm) and a taper of the tip end should fit to an outer diameter of the the valve (15.0 mm) and inlet port of the filter cartridge, respectively. Both connections are retained securely during filtration in a friction fit. Sterilize the pipette tip with 10% commercial bleach (ca. 0.6% sodium hypochlorite) before use.
5. Extraction of eDNA from the Filter
NOTE: In steps 4 and 5, we use a commercial kit, largely following a protocol for "nucleated blood" provided by the kit. For simplicity, we describe the procedure for processing an individual cartridge. In practice, we recommend processing 8 (or the maximum number of the centrifuge) or fewer filters at a time.
6. Purification of Extracted DNA
NOTE: We elute eDNA with 100 µl buffer AE instead of 200 µl specified in the manual of the commercial kit.
It is technically difficult to isolate and quantify only fish eDNA from the extracted bulk eDNA, because the MiFish primers coamplify the target region from some non-fish vertebrates, such as birds and mammals, with PCR products of the same size (ca. 170 bp)12. Instead of quantifying fish eDNA, we perform MiFish metabarcoding analysis of eDNA from an aquarium tank with known species composition using the two different methods of filtration and DNA extraction, and compare the numbers of detected species per library from the two protocols. This simple experiment was designed to show feasibility of the present protocol as the "representative results" and not to rigorously demonstrate its superiority over others.
We sampled a total of 30 L of seawater from a Kuroshio tank in the Okinawa Churaumi Aquarium, Okinawa, Japan (26˚41'39"N, 127˚52'41"E). The Kuroshio tank is designed for exhibiting marine megafauna, with dimensions (L x W x D) of 35 m x 27 m x 10 m and a total water volume of 7,500 m3. It houses approximately 60 large-sized marine fish species characteristic to areas around the Kuroshio, one of the western boundary currents flowing northeastward along the entire length of Japan. In a previous MiFish metabarcoding study, Miya et al.12 detected 61 of the 63 species (97%) contained in the tank from five 2 L seawater samples.
For filtration of the seawater, we used filter cartridges (pore size = 0.45 µm) and glass-fiber disk filters (pore size = 0.70 µm). We concurrently filtered the seawater on the same manifold using the two protocols for four water volumes (1, 2, 3, and 4 L; total, eight filters). After the filtration of seawater, we also filtered 1 L of PCR-grade water with the two filters (filter cartridge and glass-fiber filter) as the negative controls (two filter blanks). From those types of filters, we extracted eDNA using the new and previously-used protocols12, both of which employ the same commercial kit. A total of 10 eDNA samples (including the two negative controls) were subjected to MiFish metabarcoding analysis as described in Miya et al.12
Briefly, we performed 1st-round PCR to amplify the target region (mitochondrial 12S rRNA gene; ca. 170 bp) and to append adapters for paired-end sequencing. For eight eDNA samples (excluding the two negative controls), we conducted 10 PCR replicates to see variations in the number of detected species within each sample (total 80 PCRs). We also added the filter and PCR blanks to each eight eDNA samples (total 16 PCRs). In total, we obtained 96 PCR products from the 1st-round PCR and they were used as templates for 2nd-round PCR for dual indexing and appending additional adapters for library preparation following Miya et al.12
The paired-end sequencing (2 x 150 bp) of the 96 libraries yielded 4,127,546 reads, with an average of 42,995 reads per library (40,539 reads in the filter cartridge and 43,121 reads in the glass-fiber filter). After demultiplexing and subsequent pre-processing of the raw data, 1,068,266 reads were retained for the BLAST searches for taxonomic assignment. Of these reads, 830,788 reads were identified as those species contained in the Kuroshio tank (Table 1). We detected 60 tank species from 830,788 reads and the average numbers of detected species for the 10 PCR replicates from the eight eDNA samples ranged from 29 for 1 L (glass-fiber filter) to 55 for 4 L (filter cartridge) (Table 1). Read numbers of the two blanks from the eight eDNA samples were minor, ranging from 0 to 12 for the tank species.
We found that the number of detected species by the filter cartridge was significantly higher than those of the glass-fiber filters (ANCOVA, F = 381.8, P <0.001; Figure 1) across 1-4 L seawater volumes. Number of detected species by the filter cartridge were approximately 1.5 times higher than those using the glass-fiber filters and in both methods, the number of detected species increased with the volume of water filtered (ANCOVA, F = 164.2, P <0.001).
Higher numbers of detected species by the filter cartridge may result from one or more of the following steps in the experimental workflows: filtration of the sampled water through (1) different materials (hydrophilic PVDF [polyvinylidine difluoride] vs. glass microfiber) and/or (2) different nominal pore sizes (0.45 µm vs. 0.70 µm) in the filter cartridges and glass-fiber filters could retain more eDNA from fishes on the former filter; (3) use of a rotary shaker in the preheated incubator produces more DNA yield from the filter cartridges than from the folded glass-fiber filters in a spin column placed on a motionless heat block; (4) collection of the extracted DNA by centrifugation from the filter cartridge is more efficient than that from the glass-fiber filter owing to different filter materials and/or different centrifugation methods. Apparently, a more rigorously designed study is required to pinpoint the factors responsible for the consistently higher number of detected species by the filter cartridge in the present study.
Figure 1: The Number of Detected Species Plotted against Filtered Water Volumes in MiFish Metabarcoding Analysis of eDNA from the Kuroshio Tank, Okinawa Churaumi Aquarium. The number of detected species is restricted to the species contained in the tank. From 30 L of seawater, we filtered 1, 2, 3, and 4 L samples using the filter cartridges and glass-fiber filters and extracted eDNA from those filters using the present and previously described protocols12, respectively. We performed MiFish metabarcoding analysis (simultaneous detection of multiple species) for 10 PCR replicates from the four water volumes using two protocols. The horizontal bar in the box represents the median; the upper and lower edges of box correspond to inter-quartiles, the vertical bars correspond to 1.5 x quartiles, and the dots represent outliers. Please click here to view a larger version of this figure.
Table 1: Taxonomic Composition and Read Numbers of the 60 Species Detected in MiSeq Analyses of eDNA Samples from the Kuroshio Tank. Only those species contained in the tank with reference sequences in the custom database are shown. Please click here to view a larger version of this table.
In many metabarcoding studies using environmental samples such as water and soil, post-filtration treatment of the filter cartridge is generally as follows24,25: 1) cutting open or cracking the housing with hand tools (tubing cutter or pliers); 2) removal of the filter from the cartridge; and 3) cutting the filter into small pieces with a razor blade for DNA extraction. To avoid such cumbersome and time-consuming procedures that are prone to contamination in the laboratory, we have attempted several DNA extraction methods within the housing of the filter cartridge using a widely-used, inexpensive commercial kit, and successfully developed the present protocol with favorable results from eDNA metabarcoding analysis (Figure 1).
One of the more critical steps in the present protocol is the use of a rotary shaker in a preheated incubator, that is capable of providing good DNA yield for subsequent library preparation in the eDNA metabarcoding analysis. Constant stirring of the proteinase-K solution within the housing at optimal temperature probably facilitates DNA extraction from particles trapped on the filter. Note, however, that this protocol involves the use of both a rotary shaker and an incubator, and the latter must be able to accommodate the former. In addition, a centrifuge that can accommodate both 15 ml and 50 ml conical tubes is indispensable for removing the remaining liquid from the post-filtration filter and collecting extracted DNA within the cartridge, respectively.
There is another option to achieve such DNA extraction without having to cut open the cartridge (see Table of Materials/Reagents). The kit does not use any enzymes for DNA extraction; instead, it uses a special lysis buffer in combination with additional mechanical lysis with bead beating. In preliminary experiments based on natural seawater from the temperate coastal Pacific, we attempted DNA extractions using this kit along with a prototype of the present protocol using the commercial kit. With the use of the latter prototype protocol, we consistently recognized distinct amplification products in gel electrophoresis for the 1st-round PCR. In contrast, we failed to obtain any PCR products through the use of a different kit (see Materials/Reagents Table) despite following two alternative protocols provided by the manufacturer. Considering that PCR inhibitor removal steps are included in the two protocols of the second kit (see Materials/Reagents Table), this observation suggests a lack of adequate amounts of DNA in the template. Such relatively low DNA yields from environmental samples using the second kit are reported in recent studies26,27.
In the present protocol, we have chosen the filter cartridge with large nominal pore size (0.45 µm), while aquatic microbial studies conventionally use the smaller one (0.22 µm) after prefiltration through larger pore-size filters (1 to 3 µm)28. The reasons of our choice are simple and straightforward, and are not based on theoretical arguments: 1) the pore size is much closer to that of the glass-fiber filters (0.70 µm) we have conventionally used in previous eDNA studies12,14,22,23; 2) it is expected to allow larger volumes of water to be filtered than the smaller one. The latter feature is particularly significant for our current research projects undergoing in the open-ocean and deep-sea ecosystems with scarce fish abundance and biomass. In fact, we confirmed successful filtration of an independently sampled 10-L of seawater from the Kuroshio tank with no noticeable clogging (results not shown). Recently, however, Turner et al.29 demonstrated that 0.2 µm filtration maximized carp eDNA capture (85% ± 6%) while minimizing total (i.e. non-target) eDNA capture (48% ± 3%), but filter clogging limited this pore size to a sample volume of less than 250 ml. Additional studies are needed to determine the optimal filter pore size and water filtration volume of the filter cartridge for eDNA metabarcoding in various types of aquatic environments with varying levels of fish abundance and biomass.
One of the limitations of this protocol is that the water filtration should be performed in the laboratory where a power supply is available. Because of limited chemical stability of DNA, eDNA begins to degrade as soon as it is shed from an organism and is detectable only for a short period of time in aquatic environments (hours to days)30. Therefore it is ideal to filter the water sample immediately after collection to avoid significant degradation of eDNA. In this regard, developments of the on-site filtration methods using the filter cartridge would be a promising next step. The portability and sterility of the filter cartridge enable development of the on-site filtration methods, which will provide more intact eDNA that can reflect species composition of the fish community more accurately in eDNA metabarcoding. The collected eDNA on the filter cartridges may be stabilized by the injection of a commercial reagent31 (see Materials/Reagents Table) or Longmire buffer32 into the cartridge before bringing it back to the laboratory.
The authors have nothing to disclose.
This study was supported as basic research by CREST from the Japan Science and Technology Agency (JST) and by grants from JSPS/MEXT KAKENHI (Number 26291083) and the Canon Foundation to M.M. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Mesh panel | Iris Ohyama | MPP-3060-BE | |
Metal prong | Iris Ohyama | MR12F | |
Stand for the mesh panel | No brand | 4184-9507 | available from Amazon Japan |
1-L plastic bag with screw cap | Yanagi | DP16-TN1000 | |
Male luer-lock connector | ISIS | 11620 | |
10-mL pipette tip | Eppendorf | 0030 000.765 | |
10-L book bottle with valve | As One | 1-2169-01 | |
Sterivex-HV filter | Millipore | SVHVL10RC | denoted as "filter cartridge" throughout the ms and used in the protocol |
Male luer fitting | As One | 1-7379-04 | |
Female luer fitting | As One | 5-1043-14 | |
Inlet luer cap | ISIS | VRMP6 | |
Outlet luer cap | ISIS | VRFP6 | |
High vacuum tubing | As One | 6-590-01 | |
Vacuum connector | As One | 6-663-02 | |
Silicone stopper | As One | 1-7650-07 | |
Manifold | As One | 2-258-01 | |
Aspirator-GAS-1 | As One | 1-7483-21 | |
DNeasy Blood & Tissue Kit (250) | Qiagen | 69506 | |
PowerWater Sterivex DNA Isolation Kit | MO BIO | 14600-50-NF | denoted as "optional kit" in the ms |
Tabletop Centrifuge | Kubota | Model 4000 | Maximum speed 6,000 rpm |
Fixed-angle rotor | Kubota | AT-508C | |
Adaptor for a 15 mL conical tube | Kubota | 055-1280 | |
RNAlater Stabilization Solution | Thermo Fisher Scientific | AM7020 | |
Parafilm | PM992 | denoted as "self-sealing film" |