Fish culture
Fertilized eggs of chum salmon [Oncorhynchus keta (Walbaum, 1792)] were obtained from Unosumai hatchery, Iwate, Japan. The eggs were fertilized artificially on 4th December 2018 and eye-stage embryos were transferred to the Atmosphere and Ocean Research Institute, Chiba, Japan. Hatched salmon fries were fed with commercial diet after emergence in freshwater. Natural seawater (SW, salinity 35 ‰) was obtained from the Kuroshio Current at Hachijō-jima. The SW was stored in underground facilities and salmon eDNA was not detectable from the SW stock from in-house experiment. Salmon juveniles were transferred to seawater tanks after 2-months culture in freshwater. The experimental tank contained 500 L seawater with recirculation and temperature control at 14 °C. Twenty-five individuals (fork-length: ca. 20 cm; weight: ca. 50 g) were kept in the tank and the water served as a source of eDNA in the present study. All animal studies were performed according to the Guideline for Care and Use of Animals approved by the Animal Experiment Committee of The University of Tokyo.
Filtration of water
Two concentrations of salmon water were prepared. “Neat salmon water” was obtained directly from the stock tank. “Diluted salmon water” was prepared by diluting the tank water 10 times with seawater that had not been used for any previous fish culture. To standardize filtration conditions, 1 L of salmon water at either concentration was filtered through a Sterivex cartridge (0.45 μm, Millipore SVHV010RS, Merck, Tokyo, Japan) using a vacuum manifold (Manifold KMP-3, Advantec, Tokyo, Japan) attached to an aspirator (Eyela A-1000S, Tokyo Rikakikai, Tokyo, Japan) (Fig. 7). After filtration, the outlet of the Sterivex cartridge was sealed by Parafilm and 1.6 mL of RNAlater was introduced into the cartridge through the inlet. The inlet was subsequently sealed by Parafilm and the cartridge was stored at − 20 °C until further processing. All the Sterivex cartridges were then randomly assigned to different extraction protocols after the filtration step to reduce bias on filtration order.
Representative photos of Sterivex cartridges with different extraction protocols. (A) A cartridge extracted by 0.44 mL lysis mix protocol. Note the water mark that is present on the filter surface (white arrow), suggesting insufficient contact with the lysis buffer mix. (B) A cartridge extracted by 2.0 mL lysis buffer mix with backwash. Note that no water mark is found as in (A). (C) A cartridge centrifuged by angled rotor. Note that remaining lysate is present (black arrow).
DNA extraction and purification
All the connectors, stoppers, and silicon tubes were decontaminated by bleaching before use5. The RNAlater in the Sterivex cartridges was thawed on ice and removed by aspiration through the outlet. A silicon tube (1.5 cm; i.d. 3 mm; o.d. 9 mm) was used to connect a Luer-Lock adaptor (VRF306, AS ONE, Osaka, Japan) to the outlet such that both inlet and outlet can be sealed by Luer-Lock stoppers (VRMP6, AS ONE, Osaka, Japan). We tested three protocols that were modified from a published method that was widely used in the eDNA field13. The following reagents are based on the Qiagen DNeasy Blood and Tissue Kit (ThermoFisher Scientific, Waltham, MA, USA), with additional reagents (e.g. Buffer AL and proteinase K) purchased from the same sources when necessary. The major modifications are summarized in Fig. 7. For Protocol 1, we followed the original extraction protocol13. A lysis buffer mix (PBS 220 μL, Buffer AL 200 μL, Proteinase K 20 μL; 440 μL total volume) was introduced to the cartridge through the inlet. Both ends of the cartridge were sealed by Luer-Lock stoppers and the cartridge was incubated at 56 °C for 30 min with mild rotation. In Protocol 2, approximately four-times volume (2 mL) of lysis buffer mix (PBS 990 μL, Buffer AL 910 μL, Proteinase K 100 μL) was introduced through the inlet. After sealing the ends, the cartridge was incubated at 56 °C for 30 min without rotation. The rotation was omitted because the filter surface was completely covered by the lysis buffer mix. In Protocol 3, 2 mL lysis buffer mix (PBS 990 μL, Buffer AL 910 μL, Proteinase K 100 μL) was introduced into the cartridge through the outlet using a 2.5 mL syringe (Terumo Corporation, Tokyo, Japan) to flush the cartridge in a reverse direction. After sealing the ends, the cartridge was incubated at 56 °C for 30 min without rotation.
After the incubation, each Sterivex cartridge was placed in a spin column (maxi spin, flat bottom, Ciro Manufacturing Corporation, Deerfield Beach, FL, USA) attached to a 50 mL centrifuge tube, with the cartridge inlet facing downward. The unit was centrifuged with a swing-type rotor at 2500g for 10 min at 25 °C to elute the content. The cartridge and spin column were removed and molecular grade ethanol (99.5%, Fujifilm Wako Pure Chemical Corporation, Osaka, Japan) was added to one-third of the final volume (200 μL for Protocol 1 and 1 mL for Protocols 2 and 3). The mixture was loaded on a DNeasy Blood and Tissue Kit column attached to a vacuum manifold and the column was washed by 0.8 mL AW1 buffer and 0.8 mL AW2 buffer sequentially. The DNeasy column was dried by centrifugation at 17,700g for 2 min. Adsorbed DNA was eluted from the column with 75 μL AE buffer twice to maximize recovery. A total of 150 μL sample was collected from each extraction. Six replicates of Sterivex cartridges were used for each protocol. To test the extraction efficiency, each Sterivex cartridge was extracted 3 times and the eluted samples were analyzed separately.
An aliquot of DNA sample (100 μL) was further purified by Qiagen DNeasy PowerClean Pro Cleanup Kit (ThermoFisher Scientific, Waltham, MA, USA) according to the manufacturer’s protocol.
To test the binding capacity of the DNeasy column, we performed an additional experiment to determine whether the quantity of sample input will have a saturating effect on the column binding. We prepared a high sample input group by filtering 2 L of neat salmon seawater per Sterivex cartridge, while 1 L of 10×-diluted salmon seawater per cartridge was considered as a low sample input group. The extraction was performed using Protocol 3. Since we hypothesized that the binding capacity of one DNeasy column could be saturated by high input, we compared the eDNA extracted by a single column or double columns connected in series. In the case of double columns, the elution was performed individually for the upper and lower columns, and the eluates were analyzed separately.
Chum salmon eDNA quantification
The eDNA was measured by quantitative PCR (qPCR) as described in a previous study5, with slight modification. Instead of using ABI TaqMan Environmental Master Mix 2.0 (Applied Biosystems, Foster City, USA), Takara Probe qPCR Mix (Takara Bio Incorporation, Kusatsu, Japan) was used as we showed that this reagent has a higher capacity to resist inhibition by environmental contaminants in the PCR reaction (see Results). We compared the resistance to contaminants by Environmental Master Mix and Takara Probe qPCR Mix according to the spiking/dilution methods described previously5, using the environmental samples collected from Murohama Bay of Otsuchi, which contained high levels of PCR inhibitors.
Method validations in field samples
Thirty-four random seawater samples (1 L each) were collected from the Otsuchi River mouth (141′26.917 E, 39′55.021 N), Iwate Prefecture between March 15th–19th, 2020. Negative control was prepared by filtering distilled water (1 L). The water samples were filtered and treated with RNAlater as described in previous section. The cartridges were extracted according to the Protocol 3 described in previous section. To test the extraction and purification efficiencies, 1010 copies of a control plasmid (self-ligated pGEMT-easy, Promega, Madison, WI, USA) were added to each 2 mL lysis buffer mix (PBS 990 μL, Buffer AL 910 μL, Proteinase K 100 μL) at the beginning of the extraction. The extraction efficiency was calculated from ratio of control plasmid quantified from the DNeasy-extracted samples relative to input quantity. Real-time PCR for control plasmid was designed at the M13 regions: forward primer (pGEMT-F), 5′-TTTCCCAGTCACGACGTT-3′; reverse primer (pGEMT-R), 5′-TTCACACAGGAAACAGCTATGA-3′; probe (pGEMT-probe), 56/FAM/ACGCGTTGG/ZEN/ATGCATAGCTTGAGTA/3lABkFQ (Integrated DNA Technologies Inc., Coralville, IO, USA). Chum salmon eDNA concentrations of both DNeasy extracted samples and samples further purified by PowerClean Pro Cleanup Kit were measured to estimate the recovery rate of PowerClean purification. To estimate the PCR inhibitor effects from samples without purification by PowerClean Pro Cleanup Kit, an additional assay was performed with 105 copies of chum salmon standard spiked into each PCR reaction. The chum salmon standard used in spiking is the same plasmid DNA5 used for constructing the standard curve. A reduction in spiked values calculated by subtracting the spiked copy number to original copy number indicates that PCR reaction was not optimal.
$$ {text{Detection}},{text{rate}},{text{of}},{text{spiked}},{text{DNA}} = frac{{left( {{text{copy}},{text{in}},{text{spiked}},{text{sample}} – {text{copy}},{text{in}},{text{original}},{text{sample}}} right)}}{{left( {10^{5} ,{text{copy}},{text{of}},{text{spiked}},{text{DNA}}} right)}} times 100% $$
Detection rates lower than 100% indicate that the PCR reaction could be affected by PCR inhibitors in environmental samples.
To test whether addition of bovine serum albumin (BSA) can further protect the real-time PCR reaction from non-specific inhibition, a final concentration of 1 μg/μL BSA (A4161, Sigma-Aldrich, St. Louis, MO, USA) was added to each Takara Probe qPCR Mix reaction. Real-time PCR was performed on the serially-diluted inhibitor-rich sample and the slope of the dilution curve was determined. PCR efficiencies were calculated according to the following equation (Efficiency = 10−1/slope)27. As we found improved quantification from BSA experiment, we further tested chemical additives including 2% DMSO (D2650-5X Sigma-Aldrich, St. Louis, MO, USA), 0.5% Tween 20 (P9416 Sigma-Aldrich, St. Louis, MO, USA), 0.01% Formamide (F9037 Sigma-Aldrich, St. Louis, MO, USA), and a protein-based additive GP-32 protein (Nippon Gene, Toyama, Japan) at 0.1 μg/μL final concentration28. A pooled eDNA sample was made from mixing some inhibitor-rich samples from the 34 random environmental samples, and subsequently used in this preliminary test. The PCR amplification plots were compared between chum salmon standard and the pooled sample. Slope of the PCR amplification plot was obtained by linear regression on the steepest linear portion of the curve. PCR reaction was compromised by the inhibitor when we observed a reduction in saturated fluorescence and a flatten slope of the linear portion of the amplification plot of the pooled sample in comparison to those of standard. We considered that the rescue from reduction in saturated fluorescence and slope change are indicating an increase in resistance to PCR inhibitors by the additive.
Statistical analysis
The concentrations of salmon eDNA extracted by various protocols were analyzed by two-way ANOVA followed by Tukey’s multiple comparisons (GraphPad Prism Ver. 6 for Windows, San Diego, CA, USA). Statistical significance (p < 0.05) among various groups was denoted by alphabet letters. Purification efficiency of PowerClean Pro Cleanup Kit was calculated from paired samples quantified before and after the purification. Pearson’s correlation was performed to test whether the pre-purified quantity has a correlation to the purification efficiency in field water samples.
Source: Ecology - nature.com