Aquaculture is becoming a major contributor to the world food supply and is expected to provide a sustainable source of animal protein to meet the world population demands of 9.9 billion people by 2050 (Garlock et al., 2022; Grace et al., 2019). However, the global aquacultural industry has been increasingly impacted by episodes of aquatic animal diseases in the last half-decade, incurring economic losses of at least US$ 6 billion annually (Shinn et al., 2018; Stentiford et al., 2020). Therefore, this sparks urgency for the development of efficient pathogen detection strategies to effect timely disease management measures, which are imperative for safeguarding aquaculture productivity and global food biosecurity (Bondad-Reantaso et al., 2018; Mugimba et al., 2021; Pingali, 2021).
Viral infections have periodically ravaged food production levels and threaten sustainability by inducing high mortality rates in fish livestock (Kawato et al., 2021; Mondal et al., 2021; Mugimba et al., 2021; Taha et al., 2020). In the last three years, Singapore's commercial fish farms have been impacted by infectious diseases and mass mortalities of diseased fish were periodically reported (Chen et al., 2019; Kiat et al., 2023). Red seabream iridovirus virus (RSIV) and nervous necrosis virus (NNV) are some of the common etiological agents that cause Singapore fish farm production diseases (Hegde et al., 2002; Kurita & Nakajima, 2012). RSIV, which causes red seabream iridoviral disease (RSIVD), is a member of the genus Megalocytivirus within the Iridoviridae family (Kurita & Nakajima, 2012; Mohr et al., 2015). This virus is an icosahedral-shaped, 200–240 nm particle-sized, double-stranded DNA virus (~112 kbp genome size) (Oh & Nishizawa, 2016; Tamaru et al., 2006) and is notable for mass mortalities in aquacultural livestock (Kawato et al., 2021; Kurita & Nakajima, 2012). Infected fish exhibit signs of lethargy and severe anemia, along with splenic enlargement (Kawato et al., 2021; Liu et al., 2018; Sumithra et al., 2022). RSIVD is a World Organization for Animal Health (WOAH) notifiable disease and poses a grave threat to RSIVD-free regions. More than 30 farmed marine fish species are known to be susceptible and it is difficult to eradicate the virus once it establishes endemism (Kawato et al., 2021; Sumithra et al., 2022). On the other hand, NNV is a small-sized icosahedral-shaped (~30 nm), positive-sense single-stranded RNA virus (~4 kbp genome size) which causes viral nervous necrosis (VNN) or viral encephalopathy and retinopathy (VER). Fish populations infected with NNV can have mortality rates of almost 100% (Afsharipour et al., 2021; Taha et al., 2020; Zorriehzahra, 2020). Infected fish are characterized by behavioral abnormalities due to viral impacts on the nervous system (Qin et al., 2020). NNV affects more than 40 different ray-finned fish species and is observed to be expanding in geographic ranges over the last 3 years (Arimoto et al., 1992; Toffan et al., 2019).
The monitoring of disease in aquaculture is challenging as most infections are only apparent at stages where the impact is irreversible, such as when fish display conspicuous clinical signs during advanced infection phases, or when mass mortality is observed (Gomes et al., 2017; Kawato et al., 2021). Conventional diagnostic methods involving morphological identifications are laborious and invasive, which may result in the sacrifice of the infected animal (Farrell, Yetsko, et al., 2021; Kawato et al., 2021). Moreover, many organ samples may have to be processed, which is time-consuming and delays response time in disease investigation and outbreak management efforts (Sieber et al., 2020). Hence, this necessitates the development of a more effective health monitoring and early detection system to curb disease spread in fish farms. Considerations of non-invasive detection alternatives have been hampered by the low abundance of pathogens shed from infected hosts into the surrounding environment, rendering indirect sampling with environmental detection to be impractical (El Boujnouni et al., 2022). These challenges have been overcome with the advent of environmental DNA (eDNA) and RNA (eRNA) techniques in recent years, which can be used as a non-invasive alternative for pathogen detection and biosurveillance applications.
The two key components of the eDNA/eRNA workflow to enhance sensitivity for pathogen detection regardless of the shedding rate or infection stages involves concentrating genetic material from the environmental samples, coupled with the use of specific primers designed to enrich the trace signals of target taxa from bulk samples (Sengupta et al., 2019). Various concentration methods have been utilized for aquatic eDNA/eRNA experiments, such as salt precipitation (Deiner et al., 2015), vacuum pump filtration (Ip, et al., 2022; Ip, Chang, et al., 2021; Ip, Tay, et al., 2021; Kawato et al., 2021), peristaltic pump filtration (Peixoto et al., 2021), centrifugal ultrafiltration (Wong et al., 2021), and syringe filtration (Doi et al., 2021; Wong et al., 2020). While different methods vary in nucleic acid recovery rates from water samples, each can be applied to optimally address challenges from different experimental scenarios. Furthermore, viral particle sizes, morphology, and genetic makeup (DNA or RNA virus) could drive the main challenges in the application of different methods. Therefore, our study aims to establish eDNA and eRNA protocols for the simultaneous detection of both DNA and RNA fish viruses, RSIV and NNV respectively. As the amount of virus in seawater is known to be low, we investigated the recovery efficiencies of three different concentration methods, namely, (i) vacuum pump filtration with iron flocculation, (ii) Sterivex syringe filtration, and (iii) centrifugal ultrafiltration. The incorporation of eDNA/eRNA protocols to existing fish disease investigation workflows will support management strategies for future virus outbreaks in commercial fish farms.
MATERIALS AND METHODSAn overview of the experimental workflow is illustrated in Figure 1. This consisted of two main experiments: (i) the optimization and evaluation of automated and spin column (manual) kit extraction methods of viruses concentrated from vacuum pump filtration with iron flocculation, Sterivex syringe filtration, and centrifugal ultrafiltration, and (ii) the optimization and evaluation of virus eDNA/eRNA recovery efficiency of serial dilutions of spiked seawater concentrated by the three filtration methods.
Enlarge this image.FIGURE 1. Workflow overview of virus recovery by concentrating eDNA and eRNA from artificially spiked seawater in this study, comparing recovery rates of three different water filtration and two different nucleic acid extraction methods. Sankey diagram was created with sankeymatic (https://sankeymatic.com/build/) and icons were adapted from Biorender (https://biorender.com).
An achieved RSIV isolate (laboratory reference number A14/07/02), which was later being described as TGA14 in Kurita and Nakajima (2012), was propagated in a cell line (SKF-9) derived from spotted knifejaw (Oplegnathus punctatus) (Kawato et al., 2017). The propagated virus (A14/07/02, 3p) was stored at −80°C until further use. DNA was extracted from the RSIV isolate using the QIAamp® DNA Mini Extraction Kit (Qiagen) following manufacturer's recommendations.
For the detection of RSIV, the extracted DNA was subjected to a Megalocytivirus (MCV) qPCR assay as described by Mohr et al. (2015). The primer and probes for MCV qPCR consisted of the forward primer 5′- TGACCAGCGAGTTCCTTGACTT-3′, reverse primer 5′- CATAGTCTGACCGTTGGTGATACC -3′ and hydrolysis probe 5′-6-carboxyfluorescein [FAM]-AACGCCTGCATGATGCCTGGC-Carboxytetramethylrhodamine [TAMRA]-3′. Each qPCR reaction was a total volume of 25 μL consisting of 12.5 μL of Maxima probe qPCR Master Mix (2X) (Thermo Fisher Scientific), with a final concentration of 900 nM for each primer and 250 nM for probe, and 2 μL of DNA template. The MCV qPCR was done on the Applied BiosystemsTM 7900HT Fast Real-Time PCR system, with thermal cycling conditions as follows: 50°C for 2 min, 95°C for 10 min and 45 cycles of 95°C for 15 s and 60°C for 1 min. To determine virus concentration of the RSIV isolate, the extracted DNA was run concurrently with 10-fold serially diluted plasmid (107–100 copies), containing the following RSIV DNA sequence:
TGACCAGCGAGTTCCTTGACTTTTGGAACGCCTGCATGATGCCTGGCAGCAAACAATCTGGCTACAACAAGATGATTGGCATGCGCAGCGACCTGGTGGGCGGTATCACCAACGGTCAGACTATG.
The qPCR assays were conducted in triplicate reactions and standard curves were obtained by plotting the log10 plasmid serial dilutions against the average qPCR quantification cycle values (Cq) (Appendix S1). Copy numbers of the RSIV isolate was determined by calculating from the Cq values against the standard curve using the equation (Taengphu et al., 2022):[Image Omitted. See PDF]
Lastly, the RSIV stock was used to spike in seawater to get a concentration of approximately 1000 copies of RSIV per μL of seawater (Table 1).
TABLE 1 Determination of average virus copy numbers from triplicate dilution assays of RSIV and NNV stock.
| Virus | Average virus copies per 2 μL of extracted RSIV stock DNA/2.5 μL of extracted NNV stock RNA | Average virus copies in elution of RSIV stock DNA (200 μL)/NNV stock RNA (60 μL) | Estimated copies of virus per 1 μL of RSIV/NNV stock | Volume of RSIV/NNV stock to spike for 1e3 virus copies/μL in 5 L seawater |
| RSIV | 17,067,443.99 | 1,706,744,399.00 | 8,533,722.00 | 0.586 |
| NNV | 27,989,535.72 | 671,748,857.28 | 3,358,744.29 | 1.489 |
An achieved NNV isolate (laboratory reference number M45/12/91) was propagated in a cell line derived from Asian sea bass larvae (Chong et al., 1987) following the protocol described in Hegde et al. (2002). The propagated virus (M45/12/91, 3p, 10-3 dilution) was stored at −80°C until further use. RNA was extracted from the NNV isolate using QIAamp® viral RNA mini kit (Qiagen) following manufacturer's recommendations.
For the detection of NNV, extracted RNA was subjected to NNV RT-qPCR assays as described by Hick and Whittington (2010). The primer and probes for NNV RT-qPCR consisted of forward primer 5′-CTTCCTGCCTGATCCAACTG-3′, reverse primer 5′-GTTCTGCTTTCCCACCATTTG-3′ and probe 5′-6-carboxyfluorescein [FAM]-CAACGACTGCACCACGAGTTG-Carboxytetramethylrhodamine [TAMRA]-3′. Each NNV RT-qPCR was carried out in a 20 μL reaction volume containing 10 μL of 2X RT-PCR buffer, 0.8 μL of 25X RT-PCR Enzyme Mix from AgPath-ID™ One-Step RT-PCR (Thermo Fisher Scientific), with final concentration of 250 nM for each primer and 200 nM for the probe, and 2.5 μL of RNA template. The RT-qPCR was performed on the Applied Biosystems™ 7900HT Fast Real-Time PCR system, with thermal cycling conditions as follows: 45°C for 10 min, 95°C for 10 min and 45 cycles of 95°C for 15 s and 60°C for 45 s.
To determine virus concentration of the NNV isolate, extracted RNA was run concurrently with 10-fold serially diluted plasmid (107–100 copies), containing the following NNV cDNA sequence:
CTTCCTGCCTGATCCAACTGACAACGATCACACCTTCGACGCGCTTCAAGCAACTCGTGGTGCAGTCGTTGCCAAATGGTGGGAAAGCAGAAC.
The assay was conducted in triplicate reactions and standard curves were obtained by plotting the log10 plasmid serial dilutions against the average qPCR Cq values (Appendix S1). Similarly, the copy numbers of the NNV isolate were determined with the same equation used for RSIV calculations and the NNV stock was used to spike in seawater to get a concentration of approximately 1,000 copies of NNV per μl of seawater (Table 1).
Collection of seawater for spiking RSIV and NNVUsing a 5-L Van Dorn horizontal water sampler, natural seawater was collected off Platform 2 of the Sungei Buloh Wetland Reserve (GPS coordinates: 1°26′57.6″ N, 103°43′36.5″ E). The water sampling was conducted under clear weather conditions during the high tide time of the day, at 2.0–2.4 m above the chart datum. Seawater samples were dispensed into clean 1- and 2-L Nalgene bottles and were transported back to the laboratory within 20 min. The bottles of seawater were progressively cooled for ~16 h to 10°C, followed by storing at −80°C until further use. For spiking experiments, the frozen seawater samples were either thawed for 5–6 h in a water bath set at room temperature, or 2–8°C for 16–18 h.
We applied stringent contamination control measures for all experiments. All collection bottles and filtration equipment (filter units and membranes) were submerged in 10,000 PPM sodium hypochlorite for ~30 min and dried thoroughly before use. Seawater filtration was conducted in a biological safety cabinet (BSC), and all post-filtration work was performed in a separate BSC. The PCR master mixes were prepared in a designated room.
Optimizing filtration methods for concentrating RSIV and NNV from spiked seawaterThree methods for concentrating eDNA/eRNA from water samples were optimized for the co-recovery of RSIV and NNV from seawater samples (Figure 1). The thawed seawater was spiked with RSIV and NNV stock with the proportions: 0.586 mL of RSIV and 1.489 mL of NNV stock into a final volume of 5 L spiked seawater, thereby amounting to an approximate 1000 copies of both RSIV and NNV per 1 μL of seawater (Table 1). Subsequently, 10-fold serial dilutions of the spiked seawater were prepared. 1 mL of the thawed natural seawater was collected and tested as a field negative control to confirm that there was no virus detection from the natural environment.
Iron flocculation and vacuum pump filtrationBefore proceeding with filtration, 1 mL of spiked seawater was aliquoted separately as pre-filtration control. Each set of 650 mL spiked seawater was pre-filtered through two to three 2.0 μm isopore polycarbonate (PC) membrane filters (Merck millipore) with a vacuum pump (Boeco Vacuum Pump R-300). The volume of seawater from the 2.0 μm flow through was measured with an autoclaved measuring cylinder and iron flocculation was done in accordance with Kawato et al. (2021). Briefly, iron chloride solution (4.83 g FeCl3.6H2O in 100 mL distilled water) was prepared and added at 1:10000 volume ratio to the pre-filtered seawater (i.e., 20 μL iron chloride solution in 200 mL of seawater). An autoclaved sterile magnetic stirrer was added to the seawater-iron chloride suspension mix and stirred at 200 rpm for 1 h at room temperature. Thereafter, the suspension mix was separated into three 200 mL aliquots and vacuum filtered through three 0.4 μm isopore PC membrane filters respectively (three technical replicates). The triplicate 0.4 μm filters were separately transferred into 2 mL Eppendorf tubes and stored at 4°C until nucleic acid extraction on the same day, or frozen at −80°C until nucleic acid extraction.
Sterivex syringe filtrationSimilar to the vacuum pump filtration, each set of 200 mL of seawater was drawn up with a 20 mL luer-lock syringe and pushed through a Sterivex syringe filter (GP pressure filter unit, Merck Millipore) containing a polyethersulfone (PES) membrane with 0.22 μm pore size. A new Sterivex cartridge and syringe was used for each filtration of 200 mL of spiked seawater, generating three technical replicates as well. After each 200 mL seawater filtration, the syringe was filled with air and pushed through the Sterivex cartridge to eject any residual seawater that remained. This step was repeated three times with each of the triplicate samples. Lastly, the Sterivex syringe filters were stored at 4°C until nucleic acid extraction on the same day, or frozen at −80°C until nucleic acid extraction.
Centrifugal ultrafiltrationDue to this method's technical limitations in processing small liquid volumes (maximum loading volume is 15 mL each time), a reduced volume of 45 mL spiked seawater was used with centrifugal ultrafiltration. Three sets of 45 mL of spiked seawater were used for filtration with three separate centrifugal ultrafiltration tubes (three technical replicates). We placed 15 mL of spiked seawater into each Amicon Ultra-15 (molecular weight cut-off 50 kDa) centrifugal ultrafiltration filter unit (Merck Milipore), which was then centrifuged at 2000 g, 4°C for 10 mins. This process was repeated three times with the progressive addition of spiked seawater until 45 mL of seawater had been filtered through each tube. However, centrifugation time was adjusted accordingly to ensure that the volume of captured virus concentrate in the upper reservoir reaches approximately 200 μL. The virus concentrate that remained in the upper reservoir was then transferred into a 1.5 mL tube and stored at 4°C until nucleic acid extraction on the same day, or frozen at −80°C until nucleic acid extraction.
Nucleic acid extraction method Automated extraction protocolThe original stock of spiked seawater was concentrated as described in Methods 2.3.1, 2.3.2 and 2.3.3. Thereafter, the filters were extracted using IndiMag Pathogen Kit (Indical Bioscience) on the ABI MagMax™ Express-96 Magnetic Particle Processor in accordance with manufacturer's instructions, but with slight modifications as decribed below.
For the vacuum pump filtration, we tested two ways of processing the 0.4 μm vacuum filter membranes prior to nucleic acid extraction. For the first processing method, the filter membranes were cut into small pieces using autoclaved scissors before adding 695 μL of Buffer VXL (MagAttract Suspension G was excluded to prevent loss of magnetic particles) with 20 μL of Proteinase K. The small pieces of filter membranes and the suspension were vortexed for 2 min before proceeding with downstream nucleic acid extraction.
For the second processing method, 100 μL of Phosphate Buffered Saline (PBS) was added to the 2 mL tube containing a 0.4 μm membrane filter. Next, 1 mL of chloroform was added to disintegrate the filter. The mixture was vortexed briefly and centrifuged for 5 min at 5000 g and at 4°C, followed by removing the chloroform layer carefully. The tubes were once again centrifuged briefly and 200 μL of the mixture was directly loaded onto a 96-well S-block deep well plate, following the extraction kit's protocol for the purification of pathogen nucleic acids from fluid samples.
For samples concentrated by centrifugal ultrafiltration, 200 μL of virus concentrate was used directly for nucleic acid extraction following the kit's protocol for purification of pathogen nucleic acids from fluid samples. As for the viruses concentrated using Sterivex syringe filters, the concentrate was manually eluted with a 5 mL syringe containing 975 μL buffer VXL (MagAttract Suspension G was excluded to prevent loss of magnetic particles) and 20 μL proteinase K. The buffer solution was repeatedly passed through the Sterivex filter for a total of five times, before proceeding with the downstream extraction steps.
Manual extraction protocolThe original stock of spiked seawater was concentrated as described in Methods 2.3.1, 2.3.2 and 2.3.3. Thereafter, the filters were extracted using QIAamp® viral RNA mini kit (Qiagen) according to the manufacturer's recommendations. For viruses concentrated using vacuum pump filtration with iron flocculation (chloroform treatment) and centrifugal ultrafiltration, we used 200 μL of sample for extraction. We added 800 μL of Buffer AVL containing carrier RNA into the 1.5 mL micro-centrifuge tube containing the 200 μL of sample. For viruses concentrated using the Sterivex syringe filters, 1.2 mL of Buffer AVL was drawn up with a sterile 5 mL luer-lock sterile syringe and pushed through the Sterivex syringe filter to obtain at least 1 mL elute from each Sterivex cartridge. The elute was then repeatedly flushed through the Sterivex cartridge five times to obtain an approximate 200 μL of final elute before proceeding downstream with the RNA extraction.
Recovery rate calculations and statistical analysesThe relationships between virus copy numbers and average Cq values were investigated using linear regression models to determine the copy numbers for both viruses. The average recovery rate of each filtration method for each of the viruses was derived from triplicate assays (See Methods 2.1; Data S1). The recovery rate calculations were done with the following formula (Kim et al., 2022), where 200 mL seawater was used for vacuum pump and Sterivex syringe filtration, while 45 mL seawater was used for centrifugal ultrafiltration:[Image Omitted. See PDF]
The virus recovery rates were graphically represented with boxplots and the differences in recovery rates were analyzed with a one-tailed t test. All standard curves and boxplots were generated using ggplot2 v3.3.6 in R version 4.0.5.
RESULTS Optimization of automated and manual nucleic acid extraction methodsWe explored both the spin column extraction and the automated extraction methods for co-extracting RSIV (DNA virus) and NNV (RNA virus). To assess whether both methods are comparable, we analyzed the viral recovery rates from spiked seawater, of spin column extraction with automated extraction across the three filtration methods. No significant differences in the recovery rates for RSIV and NNV between both nucleic acid extraction methods were found, despite observing that the automated extraction method had a lower recovery rate than the spin column extraction method (Figure 2). Critically, the largest discrepancy in recovery rates between extraction methods was with vacuum filtration for RSIV and centrifugal ultrafiltration for NNV (Figure 2; Appendix S1).
Enlarge this image.FIGURE 2. Boxplots with points jittered, showing differences in virus recovery rates between automated and spin column extraction methods for all three filtration methods (color), for the eDNA/eRNA from both viruses (a) RSIV and (b) NNV.
Next, we assessed the eDNA/eRNA recovery rates for two different ways of processing filter membranes obtained from vacuum pump filtration, (i) cutting with scissors and (ii) disintegration in chloroform. There were significant differences when comparing the cutting method recovery rate of RSIV and NNV with the chloroform treatment, where the latter attained up to 6–10-fold higher eDNA/eRNA recovery rate (Appendix S1).
Virus eDNA/eRNA recovery efficiency and detection limit from filtration of spiked seawaterSeawater was spiked with approximately 1000 copies of RSIV and NNV per 1 μL of seawater and was serially diluted 10-fold to determine the detection limit and virus nucleic acid recovery rate. Subsequently, the spiked seawater samples were concentrated and extracted with the spin column extraction method. We found that centrifugal ultrafiltration had the highest NNV eRNA average recovery rate of ~60%, followed by vacuum pump filtration with ~6% and Sterivex syringe filtration with ~8% (Figure 3). Overall, there was a higher recovery rate at higher serial dilutions across all three filtration methods.
Enlarge this image.FIGURE 3. Boxplots with points jittered, showing differences in virus recovery rates between three filtration methods, Sterivex syringe filter, centrifugal ultrafiltration, and vacuum pump filtration with iron flocculation, from serial dilutions of seawater (color) that was artificially spiked with eDNA/eRNA from (a) RSIV and (b) NNV. Significant differences in virus recovery rates are indicated by “*”. Only the spin column extraction method was used.
For vacuum pump filtration with iron flocculation of each 200 mL of the spiked seawater sample, RSIV and NNV could be detected up to 10−4 and 10−3 dilutions, respectively (Figure 3, Table 2). The average recovery rate of RSIV was 18.71% for 10−2 dilutions and 32.61% for 10−4 dilutions. This was higher and significantly different from NNV which had an average recovery rate of 4.05% for 10−3 and 5.92% for 10−2 dilutions (Figure 3, Table 2).
TABLE 2 Average RSIV and NNV eDNA/eRNA recovery rates (%) from triplicate serial dilutions of virus-spiked seawater (SW) using three filtration methods with the spin column nucleic acid extraction method.
| Filtration | Virus | Spiked SW dilution | Mean virus copies per 2/2.5 μL of RSIV/NNV concentrate | Mean virus copies in elution (100 μL) | Estimated copies of virus/μL of spiked SW | Total estimated virus copies in 200 mL spiked SW for V & S/45 mL for C | Recovery rate (%) |
| Vacuum (V) | RSIV | 10−4 | 130.43 | 6522 | 0.1 | 20,000 | 32.61 |
| 10−3 | 952.44 | 47,622 | 1 | 200,000 | 23.81 | ||
| 10−2 | 7483.65 | 374,183 | 10 | 2,000,000 | 18.71 | ||
| NNV | 10−3 | 295.92 | 11,837 | 1 | 200,000 | 5.92 | |
| 10−2 | 2024.62 | 80,985 | 10 | 2,000,000 | 4.05 | ||
| Sterivex (S) | RSIV | 10−3 | 74.15 | 3708 | 1 | 200,000 | 1.85 |
| 10−2 | 721.21 | 36,061 | 10 | 2,000,000 | 1.80 | ||
| NNV | 10−2 | 4005.87 | 160,235 | 10 | 2,000,000 | 8.01 | |
| Centrifugal (C) | RSIV | 10−3 | 63.11 | 3650.39 | 1 | 45,000 | 8.11 |
| NNV | 10−3 | 593.62 | 28,454.82 | 1 | 45,000 | 63.23 | |
Sterivex syringe filtration of 200 mL spiked seawater could recover RSIV and NNV up to the 10−3 and 10−2 dilutions (Figure 3, Table 2). The average recovery rate of RSIV for 10−3 and 10−2 dilutions was similar at 1.85% and 1.80% respectively. Although NNV could only be detected up to the 10−2 dilution, the recovery rate of NNV was 8.01% higher than the recovery rate of RSIV.
Lastly, centrifugal ultrafiltration of 45 mL spiked seawater could recover both viruses of up to 10−3 dilutions (Figure 3, Table 2). Centrifugal ultrafiltration's recovery rate for both RSIV and NNV at 10−3 dilutions was 8.11% and 63.23% respectively. Comparisons of each filtration method's attributes and differences in assay costs have been summarized in Tables 3 and 4 respectively.
TABLE 3 Comparison of attributes across the three seawater filtration methods based on this study's findings.
| Attributes | Vacuum (V) | Sterivex (S) | Centrifugal (C) |
| Batch sample processing | No | Yes | Yes |
| On-site filtration | No | Yes | No |
| Require power supply | Yes | No | Yes |
| Water processing volume | Flexible (~50 mL to more than 2 L) (Ip, Chang, et al., 2021) | Up to 2 L | Up to 45 mL |
| Set-up reusability | Yes, filter units must be disinfected prior to use | No, one time usage | No, one time usage |
| Optimal application | Detection of lower amounts of DNA viruses, e.g., RSIV | Rapid field application | Detection of smaller-sized viruses, e.g., NNV, efficiently (~30 nm) |
| Limitations | Membrane clogging, single sample filtration | Lower recovery rate | Filtering low volumes of water |
TABLE 4 Cost comparison across the seawater filtration and nucleic acid extraction methods based on main consumables and reagents used in this study.
| Filtration method | Extraction method | Main consumables and reagents per sample | Estimated cost per sample (US$) |
| Vacuum filtration (V) | Manual |
1 × Isopore PC membrane 2.0 μm 1 × Isopore PC membrane 0.4 μm 1 reaction of manual RNA mini kit |
$10 |
| Automated |
1 × Isopore PC membrane 2.0 μm 1 × Isopore PC membrane 0.4 μm 1 reaction of automated extraction kit |
$7 | |
| Sterivex (S) | Manual |
1 × Sterivex cartridge (GP-PES) 1 × 20 mL luer lock syringe, 1 × 5 ml luer lock syringe 1 reaction of manual RNA mini kit |
$13 |
| Automated |
1 × Sterivex cartridge (GP-PES) 1 × 20 mL luer lock syringe, 1 × 5 ml luer lock syringe 1 reaction of automated extraction kit |
$10 | |
| Centrifugal (C) | Manual |
1 × Amicon ultra-15 tube 1 reaction of manual RNA mini kit |
$25 |
| Automated |
1 × Amicon ultra-15 tube 1 reaction of automated extraction kit |
$22 |
We examined the relationship between virus recovery rates and each method's filter pore size. The vacuum filter had the largest pore size of 0.4 μm (400 nm), Sterivex's pore size was 0.22 μm (220 nm), and the pore size of the molecular weight cut-off 50 kDa centrifugal ultrafiltration filter unit was converted to approximately 2.4 nm (
Enlarge this image.FIGURE 4. Linear regression models fitted for recovery rates (y-axis) of RSIV (a, c, e) and NNV (b, d, f) against filter pore size (x-axis), from spiked seawater experiments (a, b), vacuum extraction with chloroform (c, d) and without chloroform treatment (e, f). R and p-values are indicated in the plot titles. Boxplots for each method are presented with points jittered and the 95% confidence interval of each linear regression model is represented by the gray-shaded area.
Water is the key medium for the transport and transmission of waterborne pathogens (Wang et al., 2022), and rapid environmental changes in the water can quickly impact animal health and livestock productivity. Therefore, the development of efficient pathogen detection strategies is critical for providing timely disease management measures. In this study, we systematically compared virus eDNA/eRNA recovery rates and demonstrated the feasibility of eDNA/eRNA approaches for the simultaneous detection of two major aquatic DNA and RNA viruses, RSIV and NNV respectively. By spiking natural seawater with RSIV and NNV, we successfully tested three filtration and two extraction methods to recover the viruses. Despite limitations in sample processing throughput and potential membrane clogging from filtering turbid seawater (Table 3; Coutant et al., 2021; Kumar et al., 2022; Peixoto et al., 2021), virus concentration from water samples by vacuum pump filtration is cost-effective and commonly applied as it yields substantial recovery rates (Tables 3 and 4; El Boujnouni et al., 2022; Jeunen et al., 2022; Kumar et al., 2022). Centrifugal ultrafiltration can be used to concentrate smaller-sized virus particles but only allow the processing of small volumes of water and is the costliest (Tables 3 and 4; Wong et al., 2021). However, the advantage of centrifugal ultrafiltration is the ability to perform batch filtration of multiple samples, which is advantageous for virus surveillance and disease investigation laboratories' operations that require high sample processing throughput. Although enclosed filter systems like the Sterivex syringe filtration is costlier than vacuum pump filtration (Table 4), it offer opportunities for automated and large-scale field sample processing. As such, this reduces assay contamination, logistics, and manpower involvement, as compared to transporting water samples back to the laboratory for processing (Wong et al., 2020).
Streamlining the nucleic acid extraction workflows can improve efficiency and reduce the time needed to deduce infection statuses with environmental DNA/RNA tools. Although both RSIV and NNV were recoverable by all the tested methods, the recovery rates varied between different filtration and nucleic acid extraction methods. We initially chose to apply spin column (manual) extraction as it is one of the most popular eDNA extraction methods (Brannelly et al., 2020). However, we noted that the spin column extraction workflow can become laborious with an increase in sample numbers. This initiated our evaluation of the automated extraction method as a potentially less laborious substitute by comparing recovery rates between the extraction methods. Generally, the recovery rates of spin column extraction are comparable with automated extraction methods across the filtration methods. Moreover, we found that the viral RNA spin column extraction kit is also suitable for the processing and recovery of DNA viruses like RSIV. As such, laboratories may assess and choose the most suitable extraction method depending on the cost of extractions kits and the availability of appropriate equipment with sufficient manpower (Tables 3 and 4). Relatedly, most vacuum pump filtration studies typically employ mechanical treatment of the filter membranes before extraction (shaking, vortex resuspension, membrane cutting) (Kawato et al., 2021; Tsang et al., 2021; Wang et al., 2022). We found that cutting filter membranes was a laborious process and prone to contamination as the smaller membrane pieces may be occasionally scattered on the BSC working surfaces during the cutting process (Ip et al., 2022; Weli et al., 2021).Fortuitously, we observed an instantaneous disintegration of the cut filter membrane pieces when performing Trizol and chloroform RNA extraction. This near-complete disintegration of the filter membranes in chloroform is both manpower efficient and more effective than cutting up the filters for nucleic acid extraction, which was recognized as a viable alternative to streamline our nucleic acid extraction workflow. Coincidentally, a similar method with chloroform treatment was also previously applied in a study of NNV from seawater (Nishi et al., 2016). Given that disease outbreaks can strike aquacultural farms without warning and livestock survival rates can deteriorate rapidly, our efforts in streamlining the nucleic acid extraction workflows of eDNA/eRNA virus detection protocols can help to reduce sample processing time and improve disease investigation response time, thus, boosting eDNA/eRNA tools' utility in future animal health management strategies.
The optimization of eDNA/eRNA concentration protocols showed varying virus recovery rates from spiked seawater among the three filtration methods. The average recovery rates for RSIV were the highest when using vacuum pump filtration with iron flocculation, where RSIV could be detected in up to 10−4 dilutions or with approximately 10,000 viral particles per 100 mL of seawater. Centrifugal ultrafiltration shared similar average recovery rates with vacuum pump filtration at 10−3 dilution for the detection of NNV, which translates to approximately 100,000 viral particles per 100 mL seawater. The Sterivex syringe filtration was the least sensitive as it required approximately 100,000 to 1000,000 viral particles in 100 mL spiked seawater to return a positive detection result of both RSIV and NNV (10−2 dilution of viruses). The increased recovery rate of vacuum pump filtration for RSIV is likely due to the use of iron flocculation, which helped to coagulate the viral particles and increase the likelihood of detection, as compared to the other two methods which had no iron flocculation done (Kawato et al., 2021; Kim et al., 2022; Nishi et al., 2016; Wang et al., 2022).
Using the vacuum pump filtration with the iron flocculation method, RSIV recovery rate was previously reported to be ~80% (Kawato et al., 2016) and ~50% for NNV (Nishi et al., 2016), while we found lower recovery rates for both RSIV (~30%) and NNV (~6%). These past studies filtered different volumes of seawater with different filter membrane pore sizes. Nevertheless, our results showed that the centrifugal ultrafiltration was highly effective for recovering NNV and exceeded that of Nishi et al. (2016)'s results with a ~ 60% recovery rate, but unexpectedly dismal for RSIV recovery with only ~8%. Similar trends were observed with the Sterivex syringe filters that had higher recovery rates of NNV than RSIV, where the latter only had <2% recovery success regardless of 10−2 or 10−3 dilutions. At 10−2 spiked seawater dilutions, the Sterivex syringe filters recovered ~8% of NNV RNA, outperforming the vacuum pump filtration with iron flocculation which recovered only ~6% of NNV. The discrepancy in recovery rates between both viruses using the same filtration method could be due to the inherent nature of DNA and RNA viruses, where the RNA is expected to be less stable, have a shorter half-life, and reduced ability to persist in the environment than DNA (Cristescu, 2019; Taengphu et al., 2022). However, this was inconclusive from our results and seemed somewhat relevant only to vacuum pump filtration. Interestingly, differences in nucleic acid stability between the DNA and RNA viruses were not observed in the recovery rate trends of the Sterivex syringe filtration (0.22 μm) and centrifugal ultrafiltration (molecular weight cut-off 50 kDa or 2.4 nm) methods. Therefore, we hypothesized that differences in recovery rates could be due to the differences in viral particle sizes; RSIV (~240 nm) particle size is up to eight-fold larger than NNV (~30 nm). The Sterivex syringe filtration and centrifugal ultrafiltration methods have smaller pore sizes than vacuum pump filtration and are more suited for capturing much smaller viral particles, thus, expectedly more efficient in recovering NNV from seawater. This is supported by the significant negative relationship where the smaller pore-sized filtration methods had the highest recovery rates of NNV. On the other hand, RSIV is much larger in size and when captured by small pore-sized membranes, could face challenges from inadequate release from the fipores for downstream lysis and nucleic acid extraction. As such, optimal recovery of RSIV would require mechanical or chemical disintegration of the filter membranes for the virus to be released into the liquid buffers during the lysis step. However, the membrane disintegration approach was not employed in Sterivex syringe filters and centrifugal ultrafilters because of the technical incompatibility with both filtration protocols. Relatedly, the vacuum pump filtration method had the highest recovery rates of RSIV because of the disintegration of the filter with chloroform, which is supported by a significant positive relationship. However, the RSIV recovery rates shared no significant relationship sacross different filtration methods when filters were cut into smaller pieces rather than dissolved. This reiterates that the chloroform-filter-disintegration treatment is crucial for ensuring the proper release of RSIV into the lysis buffer for downstream extraction steps. As NNV did not have the highest recovery rates with vacuum pump filtration and iron flocculation, it could be due to the smaller-sized virus not being efficiently captured on the larger pore-sized vacuum filters. This remains to be tested with smaller pore-sized vacuum pump filter membranes and whether the overall iron flocculation efficiency was lower with NNV than RSIV. All in all, increased eDNA/eRNA recovery rates of viruses between filtration methods was likely attributed to the use of iron flocculation to enhance viral particle capture onto filter membranes, while each filtration method's viral recovery rate could be explained by the size of the target virus particle in relation to the pore size of the filters.
Depending on the stage of infection or extent of a disease outbreak at aquacultural farms, animal health managers looking to integrate eDNA methods must adjust their investigative approaches to balance the viral recovery rate trade-offs with the technical advantages of each filtration method of choice (Tables 3 and 4; Qiao et al., 2016). For passive surveillance or early detection of pathogens before the onset of clinical signs, vacuum pump filtration with iron flocculation is likely the most suitable method since it can filter large volumes of seawater to increase the likelihood of detecting low amounts of viruses (Brannelly et al., 2020). When the situation at the aquacultural farms has progressed to a stage where the viral load in the seawater is expectedly higher with observed widespread disease outbreaks and mass mortalities, it would be more appropriate to employ Sterivex syringe and centrifugal ultrafiltration methods. Despite lower sensitivities and/or smaller volumes of water filtration throughput, both methods can batch process samples for the rapid generation of detection results (Table 3), allowing quick assessments to promptly limit the extent of viral spread. Although filtering more seawater is recommended for increasing the likelihood of pathogen detection (Brannelly et al., 2020), it is noteworthy that prefiltration with a larger pore-sized filter (2.0 μm and above) is needed for the removal of suspended material (sediment, plankton, etc.) from natural seawater that would otherwise promote membrane clogging and inhibit the PCR process (Brannelly et al., 2020; Honjo et al., 2010). In this study, we visually observed that the seawater was clear and did not observe substantial membrane clogging. It is noteworthy that turbidity of actual seawater samples may affect the iron flocculation and the selection of filter pore sizes must adjusted accordingly to minimize membrane clogging. Future studies can compare the impacts of various seawater turbidity levels on the filtration times, efficacy of iron flocculation, and volume of seawater that can be filtered without membrane clogging.
Nevertheless, considerations on sample transportation time can influence the type of filtration method to be used. It is impractical to transport large volumes of seawater over long distances to a diagnostic laboratory for testing, especially when fish farms are typically situated offshore in remote areas (Chu et al., 2020). As such, Sterivex syringe filters can allow on-site filtration by farm staff and the batch transportation of filter cartridges would also be more logistically efficient. Apart from understanding the technical details needed for applying the most suitable method to test seawater for pathogens, a sound understanding of the target virus's morphological structure and biology is as important for making well-informed decisions on the filtration method of choice. To ensure an optimal viral recovery rate from seawater, smaller pore-sized methods like centrifugal ultrafiltration are more suited for capturing smaller-sized viral particles. This study sets the baseline for viral size comparison concerning the efficiency of virus capture between filtration methods, despite only using two viruses that were both icosahedral-shaped and limited to two sizes (~30 nm and ~ 240 nm). We managed to detect another aquatic animal virus from seawater samples with this study's optimized virus-eDNA workflows (Ip et al., 2023). Future work could examine a complete range of viral particle sizes and shapes with other filtration or concentration methods, such as polyethylene glycol (PEG) precipitation, activated carbon adsorption, and membrane adsorption (Wang et al., 2022).
Lastly, it is imperative to acknowledge that eDNA/eRNA, being environmental samples and not tissue sampled directly from the host, can only allow indirect inference of active infections as the pathogen-eDNA signal may have stemmed from dead viral particles (Greer & Collins, 2007; Kaganer et al., 2021; Vilaça et al., 2020). Following WOAH procedures on notifiable disease-causing pathogens, a qPCR positive detection of pathogen DNA/RNA from environmental samples is inadequate for activating full-scale disease management measures (Amarasiri et al., 2021; Farrell et al., 2021). Pathogen detection from the environment also renders the loss of ability in the quantitative inference of infection intensity as compared to the direct processing of organ samples (Gomes et al., 2017; Vilaça et al., 2020). Consequently, eDNA/eRNA protocols must be rigorously optimized and validated to ensure sufficient assay sensitivity and specificity, thus reducing false positive or negative detections, and upholding the reliability of the results.
Nevertheless, emerging studies on analyzing environmental microRNAs (miRNA) to quantify animal host immune response signals hold tremendous potential in elevating the disease investigation utility of eDNA/eRNA-based pathogen detection methods (Cristescu, 2019; Ikert et al., 2021). Analyzing host miRNA signals in conjunction with pathogen detection from environmental samples will boost the reliability of detection results and complement existing diagnostic methods in inferring active viral infections (Cristescu, 2019; Ikert et al., 2021). Finally, with the increasing number of studies demonstrating eDNA/eRNA's utility in correlating genetic material concentration with species abundance estimates (Ip et al., 2022; Rourke et al., 2022; Spear et al., 2021; Yates et al., 2021), it may also be feasible to quantify pathogen abundances and host response signals for determining prevalence and intensity of the infection. Future studies can explore the implementation of eDNA/eRNA tools to identify and pre-empt aquacultural farms that are at potential infection risk, since eDNA/eRNA has shown capabilities in predicting pathogen occupancy from the natural environment (Kaganer et al., 2022; Miaud et al., 2019; Sieber et al., 2020; Vilaça et al., 2020). This allows the implementation of timely animal health management measures and may improve the productivity of aquacultural farms worldwide.
CONCLUSIONImplementation of virus eDNA/eRNA detection tools for pathogen surveillance and disease investigation at aquacultural fish farms requires careful consideration of the choice of seawater filtration methods, which is dependent on the stage of the outbreak, the distance of transporting samples from the farm to a laboratory, turbidity of the seawater, the laboratory resources at hand, and the morphological structure of the suspected disease-causing pathogen. Apart from being a non-invasive and animal-welfare-friendly alternative for disease investigation, the environmental sample testing methods optimized in this study have advantages in consistent recovery rates, allowing batch processing of water samples to reduce sample processing time and enhance throughput, and offer on-site filtration possibilities. Our results suggest that vacuum pump filtration with iron flocculation is effective for early disease stages where viral shedding rates into the water are low. The centrifugal ultrafiltration method is efficient for recovering smaller-sized viral particles, while larger viral particles require filter membrane disintegration during DNA/RNA extraction for improved recovery rates. Automated DNA extraction methods can reduce laboratory processing time as these methods have comparable yields with manual extraction methods. This study's findings have contributed to an improved understanding of applying appropriate seawater filtration methods for optimal virus eDNA/eRNA recovery, as the variations in recovery rates can influence detection successes and the feasibility of these early disease detection tools for supporting management strategies in future virus outbreaks at commercial fish farms.
AUTHOR CONTRIBUTIONSThe data analyses were performed by Y.C.A.I., with assistance from J.C., Y.H.C., S.R., and L.Y.T. The figures and the manuscript draft were prepared by Y.C.A.I. with inputs from J.C., L.Y.T, C.L., and Z.Y.B.T. The data collection was done by Y.H.C., S.R., K.N., W.Y.J.W., L.Y.T., C.L., J.C., and Y.C.A.I. Study initiation and guidance were by S.F.C, C.J.F., J.C., and Y.C.A.I. The study was supervised by C.J.F. and Z.Y.B.T. All authors contributed to the manuscript editing process, approved the final version for publication, and declared that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
ACKNOWLEDGEMENTSWe are grateful to Dr. Yasuhiko Kawato for providing the SKF9 cell line and advice on the recovery of RSIV from seawater using vacuum pump filtration and iron flocculation. We thank Dr. Huangfu Taoqi for his advice on the study design during the initial phase of the study. We acknowledge the invaluable laboratory work assistance from Ding Xinru and Gui Jia Qing during their internship. We would also like to thank Sungei Buloh Wetland Reserve colleagues for facilitating the natural seawater collection.
[Correction added on 27 May 2023, after first online publication: the Acknowledgements section has been included in this version.]
FUNDING INFORMATIONThis study was funded by the National Parks Board of Singapore.
DATA AVAILABILITY STATEMENTAll data are available in the main text and Data S1.
[Correction added on 27 May 2023, after first online publication: the Supplementary Data S1 has been updated in this version.]
© 2024. This work is published under http://creativecommons.org/licenses/by/4.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.
Details
; Chen, Jing 1 ; Tan, Li Ying 1 ; Lau, Clara 1 ; Ying Hui Chan 1 ; Balasubramaniam, Ravendrakumar Shanmugavelu 1 ; Wan Yen Jovinc Wong 1 ; Ng, Kaitlyn 1 ; Zi Yan Brian Tan 1 ; Fernandez, Charlene Judith 1 ; Siow Foong Chang 1 ; Him Hoo Yap 1 1 Animal & Veterinary Service, National Parks Board, Singapore City, Singapore