Development and Validation of a SYBR Green Real Time PCR Protocol for Detection and Quantification of Nervous Necrosis Virus (NNV) Using Different Standards

Simple Summary In the present report, we designed a reverse transcription real-time quantitative polymerase chain reaction (RT-qPCR) for the detection and quantification of the nervous necrosis virus (NNV), a fish virus of great importance in Mediterranean aquaculture. The advantage of this procedure with respect to others is that it ensures the correct diagnosis of any viral type, at extremely low viral loads, thanks to a limit of detection lower than in previous methods, and demonstrates a better diagnostic sensitivity and specificity than PCR and cell culture isolation. Furthermore, this is the first time that an RT-qPCR procedure has been validated for the quantification of NNV, ensuring the reliability of the quantification, regardless of the calibration standard chosen. Therefore, its use would enable the comparison of data between different laboratories. Abstract The nervous necrosis virus (NNV) is a threat to fish aquaculture worldwide, especially in Mediterranean countries. Fast and accurate diagnosis is essential to control it, and viral quantification is required to predict the level of risk of new viral detections in field samples. For both, reverse transcription real-time quantitative polymerase chain reaction (RT-qPCR) is used by diagnostic laboratories. In the present study, we developed an RT-qPCR procedure for the diagnosis and simultaneous quantification of NNV isolates from any of the four genotypes. The method proved to be highly sensitive in terms of crude virus titer: 5.56–9.88 TCID50/mL (tissue culture infectious dose per mL), depending on the viral strain, and averaging 8.8 TCID50/mL or 0.08 TCID50/reaction. Other standards also yielded very low detection limits: 16.3 genome copies (cps) of purified virus per mL, 2.36 plasmid cps/mL, 7.86 in vitro synthetized RNA cps/mL, and 3.16 TCID50/mL of virus from infected tissues. The diagnostic parameters evaluated in fish samples were much higher in comparison to cell culture isolation and nested PCR. In addition, the high repeatability and reproducibility of the procedure, as well as the high coefficient of determination (R2) of all the calibration curves with any type of standard tested, ensure the high reliability of the quantification of NNV using this RT-qPCR procedure, regardless of the viral type detected and from the type of standard chosen.


Introduction
Viral encephalopathy and retinopathy (VER), caused by nervous necrosis virus (NNV), is one of the most threatening viral diseases affecting marine and, in a lesser extent, freshwater farmed fish worldwide. Furthermore, a high number of wild fish species has been reported to be susceptible to the virus and disease outbreaks have also been recorded mainly in the Mediterranean area [1]. The main target of NNV is the central nervous system, including the brain, spinal cord, and retina. Therefore, clinical signs of the disease

Virus Purification
The viruses were cultured in 10 flasks of 150 cm 2 with semiconfluent E-11 cells, and crude virus was collected when the monolayers displayed an extensive CPE. The viral suspensions were subjected to centrifugation at 4000× g for 20 min and the pellet was resuspended in 250 µL SSC 1X buffer (0.15 M NaCl, 0.015 M sodium citrate, pH 7.0), sonicated for 30 s (20 kHz, in 3 steps of 10 s), and subjected to 4000× g for 20 min. The virus was recovered by ultracentrifugation at 100,000× g for 90 min at 4 • C, loaded onto a layer of 30% sucrose (wt/vol, in SSC 1X buffer), and pelleted again at 120,000× g for 1 h at 4 • C. The virus pellet was resuspended in 200 µL of SSC 1X buffer and subjected to fractionating by equilibrium sucrose density-gradient centrifugation, using discontinuous gradients prepared by layering 2 mL volumes of 60, 50, 40, 30, and 20% (wt/vol) sucrose in SSC 1X buffer into 17 mL polyallomer tubes (Beckman). After ultracentrifugation at 100,000× g for 1 h at 4 • C, the visualized band was removed and washed by centrifugation at 100,000× g 90 min in 30% sucrose. The final pellet was resuspended in 200 µL of SSC 1X and stored at −20 • C until use.

RNA Extraction
Total RNA was extracted directly from viral suspensions using the RNeasy Mini kit (Qiagen, Madrid, Spain) according to the manufacturer's instructions. RNA samples were resuspended in nuclease-free water (Promega, Madrid, Spain), quantified by absorbance measurement at 260 nm using a NanoDrop ND 1000 Spectrophotometer (NanoDrop Technologies, Wilmington, NC, USA), and stored at −20 • C until use. The quality of the extracted RNA was evaluated from the ratio A260/A280 [20].

Primers Design
An in silico test, using the software AlleleID v7.74 (PREMIER Biosoft, San Francisco, CA, USA), was performed to select primers for the detection of strains from the four NNV genotypes. Multiple sequence alignment using Clustal X (2.0) included both genomic sequences of NNV strains representative of the 4 genotypes (SGWak97, Acc no. NC_008040 and NC_008041; JFIwa98, Acc no. NC_013458 and NC_013459; TPKag93, Acc no. NC_013460 and NC_013461; SJNag93, Acc no. AB056571 and AB056572) ( Table 1). Fifteen primer sets putatively amplifying amplicons of 78-178 bp were pre-selected (Supplementary Table S1) and tested by PCR using SYBR Green I detection chemistry, as described below, against the 4 NNV type strains, as well as an RGNNV/SJNNV reassortant strain (SpSs-IAusc160.03; [9]). Additionally, the sets providing the best results with the four strains were tested against a series of field samples belonging to our isolates library (Supplementary Table S2), and some obtained from other researchers. Finally, the following primer set-which provided the best results in the previous assays-was selected for thorough testing in the present study: NodR1_1619  and NodR1_1758 (5 -TGGCATGTACCACGGAAC-3 ) between nucleotide positions 1619 and 1758 (139 bp) of strain SGwak97 RNA 1.

cDNA Synthesis
The synthesis of cDNA from the extracted viral RNA was performed in a MyCycler thermal cycler (Bio-Rad, Madrid, Spain). An initial mixture of 2.5 ng/µL random primers (Promega, Madrid, Spain) and 9 µL (approximately 0.15-2 ng/µL, depending on the standard) of viral RNA in nuclease-free water (Bio-Rad, Madrid, Spain), was heated at 95 • C for 5 min and cooled down to 4 • C for at least 1 min. A reverse transcription mixture containing Superscript™ III reverse transcriptase (10 U/mL; Invitrogen, Madrid, Spain), RNaseOUT™ Recombinant RNase inhibitor (2 U/mL; Invitrogen), dNTPs (0.5 mM each), and dithiothreitol (0.05 M DTT) in First Strand buffer (Invitrogen) was added and the final solution (20 µL) was incubated at 25 • C for 10 min, followed by 50 min at 50 • C. Then the enzyme was inactivated by heating at 85 • C for 5 min.

SYBR Green qPCR
The PCR reactions were carried out in a final volume of 20 µL, containing 100 nM of each primer and 2 µL of cDNA (synthetized in the previous step) in iQ™ SYBR ® Green Supermix (Bio-Rad, Madrid, Spain). Following an initial 3 min denaturation/activation step at 95 • C, the mixture was subjected to 45 cycles of amplification (denaturation for 15 s at 95 • C; annealing and extension for 15 s at 58 • C) in a CFX Connect™ Thermal Cycler (Bio-Rad, Madrid, Spain). The generation of PCR products was monitored after each extension step at 58 • C by measuring the fluorescence of double-stranded DNA binding SYBR Green dye. In order to determine the melting temperatures of the amplified products after SYBR Green qPCR, the temperature was raised from 55 • C to 95 • C and the fluorescence was detected for 10 s after each 0.2 • C. From each reaction, the threshold cycle value (Ct) was established as the cycle number at which fluorescence was detectable over the threshold value calculated by the iCycler iQ™ software (Bio-Rad, Madrid, Spain) for cycles 2-10.

Analytical Sensitivity and Specificity
To assess the sensitivity of detection of the designed RT-qPCR diagnostic and quantification procedure, the limit of detection (LOD) was evaluated against several standards used as reference (prepared as described in Section 2.9): viral titer of crude virus, RNA copies from crude and purified virus, and copies of plasmid DNA (pDNA) and in vitro transcribed RNA (ivRNA). To this end, serial dilutions of each standard were prepared, RT-qPCR was applied to each dilution (in triplicate), and the LOD was deduced from the lowest dilution providing effective amplification with the three replicas ("The last dilution showing 100% response may be accepted as a conservative estimate of the lower limit of detection," as stated by the OIE for the validation of diagnostic assays for infectious diseases [12]). To assess the specificity, the 4 NNV strains described before were tested (intra-group specificity). In addition, the procedure was tested against infectious pancreatic necrosis virus (IPNV), Spajarup (Sp) strain, and viral hemorrhagic septicemia virus (VHSV), VHSV_Ssp11 strain (genogroup III) (inter-group specificity).

Repeatability and Reproducibility (R&R)
Repeatability is known as the precision of a measurement simultaneously applied on a minimum of 3 replicas of a sample, by the same operator using the same protocol, materials, and equipment. Reproducibility is the precision changing any or several of those parameters: day, operator, procedure, materials, and/or equipment. Repeatability and reproducibility were evaluated as the intra-assay and inter-assay (respectively) coefficient of variation (CV = standard deviation/average Ct × 100). The CV was calculated from the Ct values of the 3 replicas (intra-assay), or from the average values (from the 3 replicas) in three runs performed on non-consecutive days (inter-assay). Values of CV ≤ 10% were Animals 2021, 11, 1100 5 of 21 considered high repeatability and reproducibility, and CV ≤ 5% were indicative of very high R&R.

Reference Standards for Quantification
As indicated above, five types of templates were tested as standards of quantification: 1-Viral titers and RNA copies from crude virus.-The 4 reference strains were propagated in E-11 monolayers and, when CPE was extensive, the cell debris was removed by low speed centrifugation (2000× g, 10 • C, 15 min). Then the clarified virus in the supernatant was subjected to viral titration and RNA extraction as previously described. The number of copies corresponding to the quantity of RNA extracted was calculated from the formula γ = n/N × GL × NcMw, where γ is the amount (in grams) of viral RNA, n is the number of genomic RNA molecules (genome copies), N is the Avogadro number (6.022 × 10 23 ), GL is the genome length in nucleotides, and NcMw is the average molecular weight of a nucleotide (350. After purification of crude virus from 10 flasks of 150 cm 2 as described above, the RNA was extracted and quantified, and 20-fold dilutions (from 5.12 × 10 9 to 4.07 × 10 0 ) were subjected to RT-qPCR as described. 3-Plasmid DNA. pDNA was constructed with a 3067 pb nucleotide fragment from genomic RNA 1 of the SGWak97 strain (RGNNV). AccuPrime™ Taq DNA Polymerase High Fidelity (Life technologies, Thermo Fisher, Madrid, Spain) was used according to the manufacturer's specifications, employing specific primers (VNNR1_1F, 3 -CGCAAGGTTACCGTTTAGC-5 , and VNNR1_5L, 3 -GCCGAAGCGTAGGACAGCA TAAAG-5 ) designed to amplify that fragment. The PCR products were purified using the QIAquick gel extraction kit (Qiagen, Valencia, CA, USA) and ligated into the pGEM ® -T vector using the pGEM ® -T Vector System I (Promega, Madrid, Spain). Plasmids were transformed into DH5α competent cells (Life Technologies, Thermo Fisher, Madrid, Spain) and cultured under appropriate antibiotic selection, as indicated by the manufacturer. The cloned DNA plasmids were extracted and purified using the GeneJET™ Plasmid Miniprep Kit (Fermentas, Thermo Fisher, Madrid, Spain). Plasmid copies were estimated based on spectrophotometric analysis. The mass of a single pDNA molecule was calculated using the same formula for γ described above, where GL = 3067 nc, and NcMw (the average molecular weight of each pair of nucleotides) was estimated as 660 g/mol. The concentration of the resulting plasmid solution was around 50 ng/µL (corresponding to 7.50 × 10 9 copies/µL); 20-fold serial dilutions were prepared and subjected to qPCR amplification as described. 4-In vitro transcribed RNA. Approximately 400 ng of the previously constructed plasmid were linearized by digestion with Sal I in a 20 µL reaction according to the enzyme manufacturer's instructions (FastDigest ® SalI, Fermentas, Thermo Fisher, Madrid, Spain). The reaction mix was incubated 15 min at 37 • C, followed by a heat inactivation at 65 • C for 10 min. The linearized DNA was purified using the QIAquick PCR purification kit (Qiagen, Valencia, CA, USA) following the manufacturer's instructions. RNA transcripts were generated using the MEGAscript ® Kit (Ambion, Thermo Fisher, Madrid, Spain). Briefly, approximately 200 ng of linearized plasmid were added to 20 µL of transcription reaction containing 5 mM of each ribonucleotide and 2 µL of enzyme mix in 10X reaction buffer and incubated at 37 • C for 4 h. A DNase treatment was conducted using Ambion's TURBO™ DNase Kit (Ambion, Thermo Fisher, Madrid, Spain) for 30 min at 37 • C. The RNA was recovered using the MEGAclear™ Kit (Ambion, Thermo Fisher, Madrid, Spain) following the manufacturer's instructions. RNA concentration and copy number were estimated as above.
From the formula, 1 ng of RNA 1 ivRNA was estimated to contain 5.76 × 10 8 copies. Twenty-fold serial dilutions were prepared and subjected to RT-qPCR as above.

Reliability of the Viral Quantification in Infected Fish Tissues
A volume of 1 mL of titrated crude virus (SGWak97 strain) was mixed with 1 g of non-infected gilthead seabream brain (the fish were obtained from a fish market and checked by RT-qPCR before use). The mixture was homogenized using a Tissue Master 125 homogenizer (OMNI International, Kennesaw, USA) and incubated at 25 • C for 24 h. Afterwards, the tissue was pelleted at 3000× g for 20 min and both phases, supernatant and pelleted tissue, were titrated separately as described above. In parallel, non-infected seabream brain was similarly processed (suspended in one volume of L-15 medium) to prepare the serial w/w dilutions of the infected pelleted tissue, starting with a mixture of 400 µL of infected pellet + 400 µL of non-infected tissue, and continuing with dilutions of 1/5, 1/10, 1/100, and 1/50 using non-infected tissues). From each dilution, 200 µL were subjected to RNA extraction and RT-qPCR amplification as described above. The fish were handled according to the guidelines for good practices in laboratory animals of the European Union (directive 2010/63/EU). For the challenge, protocols approved by the Galician committee for experimental animal welfare (Xunta de Galicia, Permit ID 15010/2020/004) were followed to minimize fish suffering.
Sole (N = 22) and sea bass (N = 29) individuals were infected by immersion with a viral concentration of 10 5 TCID 50 / mL of RG/SJ (SpSs-IAusc160.03 strain) or RG/RG (SGWak97 strain) strains (respectively) ( Supplementary Table S2), as previously described [22]. A control group corresponding to each species (N = 5) was handled like the infected groups and mock infected with L-15 medium. The dead fish were immediately removed and frozen at −20 • C until subjected to viral detection by the RT-qPCR procedure described here, with cell culture isolation and nested PCR (nPCR) [22]. After 3 weeks, surviving fish were killed by an overdose of MS-222 and stored at −20 • C until viral detection. a-Assessment on samples from fish farms. A total of 406 fish from 3 species (9 turbot, Scophthalmus maximus; 99 gilt-head seabream, Sparus aurata; 298 Senegalese sole, Solea senegalensis) received from several European fish farms (mainly from the Iberian Peninsula) in 2019 and 2020 were subjected to viral detection applied with the three procedures cited above. Three types of tissues and organs were employed, depending on the age of the fish: blood, from breeders, or brain and/or a pool of kidney, spleen, and heart from fish of other ages.
b-Surveillance of fish from coastal populations. In an epidemiological campaign during 2019 and 2020, a total of 433 fish from 24 commercial species were bought in fresh fish markets in different coastal location of Galicia (NW Spain), and pools of heart, spleen, kidney, and brain were subjected to viral detection using the 3 methods indicated above.

Sensitivity and Specificity
Regarding the sensitivity (Table 2), the limit of detection of RNA from crude virus averaged 0.46 fg/µL of extracted RNA, corresponding to around 1.91 × 10 5 copies/mL. However, we must point out that although the lowest concentrations (26.5, 25.0, 23.5, and 23.0 ag/µL with SJ, RG, BF, and TPNNV types, respectively) provided no amplification (results not shown), due to the 20-fold serial dilutions applied (higher than the usual 10-fold dilutions), a LOD slightly lower than the observed cannot be ruled out. In terms of viral titer, the LOD of crude virus averaged 8.8 TCID 50 /mL, which is close to 10 4 times lower than the LOD as RNA copies per mL. Furthermore, using purified virus, the LOD value was 1.63 × 10 1 , which is close to the LOD using crude virus. The LOD values with pDNA (2.36 copies/mL) and ivRNA (7.86 copies/mL), as well as with infected fish tissues (3.16 TCID 50 /mL), were also closer to the value obtained with viral titer than with RNA copies from crude virus. Virus from tissue (Viral titer) 7 DR: 6 As TCID 50 /g c 3.16 × 10 0 ( 1 ) LOD: limit of detection using as standard crude virus RNA ( 2 ) or viral titer ( 3 ), purified virus ( 4 ), plasmid DNA ( 5 ), in vitro transcribed RNA ( 6 ), or virus from infected fish tissues ( 7 ). The data are expressed as RNA concentration (fg/µL) ( a ), RNA, pDNA (plasmid DNA), or ivRNA (in vitro transcribed RNA) copies/mL ( b ), or TCID 50 (per ml or per g) ( c ), as well as total TCID 50 per RT-qPCR reaction ( d ); the copies/mL were determined by the formula γ = n/N × GL × NcMw, where γ is the amount (in g) of nucleic acid, n is the number of copies, N is the Avogadro number (6.022 × 10 23 ), GL is the nucleic acid length in nucleotides, and NcMw is the average molecular weight of a nucleotide.

Validation of the Procedure for Quantification
Quantification using crude virus as standard. Two types of standard curves were obtained using crude virus: curves employing the viral concentration measured by a traditional titration procedure (TCID 50 / mL in this study) as standard, and those employing the number of RNA copies calculated from the RNA concentration of the same crude virus sample.
The data (as Ct values) obtained in all the assays (three repeats, three replicas each) are shown in Supplementary Table S3. To ensure that the viral concentrations tested were the same for the four viral types, the initial viral titers were adjusted to 3.16 × 10 7 TCID 50 /mL for three of the types (SJNNV, RGNNV, and TPNNV), and were slightly lower (1.78 × 10 7 TCID 50 /mL) with BFNNV. Those titers corresponded to initial RNA concentrations of 1.7, 1.6, 1.5, and 1.1 ng/µL (SJNNV, RGNNV, BFNNV, and TPNNV, respectively), and matched to 7.05, 6.62, 6.22, and 4.56 × 10 11 copies/mL (respectively) by the formula described above. These data reveal a difference of around 4 log 10 between both ways of measuring the concentration. Table S3 shows the results only for dilutions providing amplification (Ct < 40) with the three replicas and in all the repeats. Thus, the dynamic range (DR) with this procedure was 6 Log 10 ( Table 2). The coefficient of variation was in most cases below 5% (always below 10%) in all replicas, which denotes a very high repeatability. Analyzing the repeats, the CV values were again below 5%, also demonstrating a very high reproducibility using the same equipment. In addition, using the data from the four types, the CV values still were below 5% (Supplementary Table S3E), which demonstrates the high repeatability and reproducibility of the procedure, regardless of the viral type.
To simplify the interpretation of the results, the equations, reliability, and efficiency of viral titer and RNA copy standard curves are shown in Table 3 (all data and the corresponding curves are in Supplementary Table S3 and Supplementary Figures S1 and S2). Regardless of the strain, all the curves were reliable: R 2 > 0.95 in all replicas and repeats, and in most cases were higher than 0.99. In addition, the procedure showed high reliability across all four viral genotypes (average standard curve R 2 = 0.9993; Figure 1A,B). As expected, the differences between line equations using viral titer and RNA copies were just in the intersection value; this is because the slopes must necessarily be the same given that both share the same Ct values.  On the other hand, the efficiency of the RT-qPCR amplification, deduced from the slopes, were in all cases over 100%, with values closer to 100 in the case of SJNNV and RGNNV strains (Table 3).
Quantification using purified virus as standard.-Only purified virus from SJNNV and RGNNV strains was used (Supplementary Table S4). From the volume of purified virus subjected to RNA extraction (100 µL), concentrations of viral RNA of 56.9 and 34 ng/µL (RGNNV and SJNNV, respectively) were obtained. RNA samples from both viruses were adjusted to initial concentrations of around 13.75 ng/mL, which approximately corresponded to 5.21 × 10 9 copies/mL. Serial 20-fold dilutions were applied until a final concentration of 8.15 × 10 1 copies/mL was achieved. As observed, with the SJNNV type no positive amplification was obtained with the lowest RNA concentration (Ct ≥ 40, or-in most cases-no amplification at all), which gave a DR of 6 Log 10 . However, with the RGNNV strain the DR was 7, since positive amplification was attained at the lowest RNA concentration (Ct around 38 in the three replicas and repeats). Repeatability and reproducibility (R&R) were also very high since CV values below 5% were obtained in all cases (average CV = 1.7%; result not shown).
The reliability of each standard curve was always high (R 2 > 0.99) with any viral type (Table 4 and Supplementary Figure S3). The general standard curve with purified virus (averaging data from both strains) was also demonstrated to be reliable (R 2 = 0.9962; Figure 1C). Unlike those observed with crude virus, the RNA amplification efficiencies were below 100% in all cases, obtaining average E > 84% with SJNNV, and lower values (under 80%) with RGNNV. most cases-no amplification at all), which gave a DR of 6 Log10. How RGNNV strain the DR was 7, since positive amplification was attained at th concentration (Ct around 38 in the three replicas and repeats). Repeatabil ducibility (R&R) were also very high since CV values below 5% were obtain (average CV = 1.7%; result not shown). The reliability of each standard curve was always high (R 2 > 0.99) with (Table 4 and Supplementary Figure S3). The general standard curve with (averaging data from both strains) was also demonstrated to be reliable (R ure 1C). Unlike those observed with crude virus, the RNA amplification ef below 100% in all cases, obtaining average E > 84% with SJNNV, and lower 80%) with RGNNV.  ( 1 ) Regression lines (from 3 replicas) for RNA (copies/mL) from purified virus ( 2 ). The reliability of the lines is given by the coefficient of determination R 2 ( 3 ) and the amplification efficiency E ( 4 ). The results obtained with two VNNV (viral nervous necrosis virus) type are shown: striped jack nervous necrosis virus (SJNNV) and red-spotted grouper nervous necrosis virus (RGNNV).
Quantification using plasmid DNA as standard. A plasmid DNA carrying a 3067 bp cDNA fragment corresponding to the RGNNV strain was constructed and used as standard for NNV quantification. Amplification was reached at concentrations from 7.55 × 10 6 to 2.36 × 10 0 copies/mL, which meant a DR of 6 Log 10 (Supplementary Table S5); no amplification was obtained at the lowest concentration tested (1.18 × 10 −1 copies/mL). Again, R&R was revealed to be very high (CV values ranging from 0.18 to 3.28). Regression curves were demonstrated to be highly reliable: R 2 > 0.999 in the three repeats (Table 5; Supplementary Figure S4), and close to 1 in the general standard curve ( Figure 1D). As with purified virus, the efficiency of the amplification was below 100 (E = 88.4 for the general curve). ( 1 ) Regression lines from 3 replicas for plasmid DNA (pDNA; copies/mL) ( 2 ) values. The reliability of the lines is given by the coefficient of determination R 2 ( 3 ) and the amplification efficiency E ( 4 ).
Quantification using in vitro transcribed RNA as standard. In vitro transcription was applied on the previously described plasmid to obtain a transcribed RNA corresponding to the RNA 1 of the RGNNV type strain. As with the plasmid, only six concentrations (from 2.52 × 10 7 to 7.86 × 10 0 copies/mL) rendered positive amplification (DR = 6) (Supplementary Table S6 and Table 6). All the CV values were below 5% (ranging from 0.13 to 3.93), denoting high R&R. The high reliability of this standard was demonstrated by the high R 2 (>0.99) of all the regression lines (Table 6; Supplementary Figure S5, including the general curve for ivRNA (R 2 = 0.9974) ( Figure 1E). Regarding the amplification efficiency, this standard was the only one providing E values close to 100: ranging from 96.1 to 104.7 considering all replicas, and from 99.6 to 101.4 considering the repeats, and obtaining a general standard curve with E = 100.3. ( 1 ) Regression lines averaged from 3 replicas, determined for ivRNA (in vitro transcribed RNA) copies/mL ( 2 ) values. The reliability of the lines is given by the coefficient of determination R 2 ( 3 ) and the amplification efficiency E ( 4 ).
Quantification using virus from infected fish tissues as standard. Fish tissue infected in vitro with a known concentration of virus was diluted with non-infected tissue and subjected to RNA extraction and RT-qPCR as described. In a first assay, three sets of extraction were applied to the same tissue sample, and three replicas in three repeats of RT-qPCR amplification were performed within each one. It is worth mentioning the demonstrated repeatability and reproducibility of the assays, since in most cases the CV values were below 5% (and always below 10%) (Supplementary Table S7). More importantly, CV values were also below 10 even considering the Ct values from all the extractions, denoting the high R&R of this type of standard.
To analyze the reliability of such a standard, three types of curves were drawn. As shown in Supplementary Figure S6 and summarized in Table 7, the linear regression applied to the data yielded curves of low reliability (R 2 < 0.95) and extremely high E values (between 160 and 230). Suspecting that the failure could be due to the highest dilution, the regression analysis was repeated, eliminating the lowest viral titer, which made it possible to obtain highly reliable curves (R 2 < 0.99) with lower E values (around 115%). This result pointed towards a low efficiency of the procedure at the lowest viral concentrations in fish tissues. Therefore, a second degree polynomial regression analysis was applied, yielding reliable curves in all cases (Supplementary Figure S6), including the general curve ( Figure 1F). In a second assay, a wider range of viral concentrations was tested (from 3.16 × 10 0 to 2.51 × 10 7 TCID 50 /gr), and the same RT-qPCR procedure was applied as described, but without replicas or repetitions (Supplementary Table S7-Assay II). Similar to Assay I, the linear regression using all the concentrations was not reliable (R 2 = 9372), but adjusted to a second degree polynomial line (R 2 = 0.9975). The regression removing the lowest concentration (calculated between concentrations 6.31 × 10 1 and 2.51 × 10 7 ) was reliable (R 2 = 0.9611), and was not significantly different from that obtained in Assay I (results not shown).

Comparison Among All the Standards
The general curves (averaging data from all replicas, repeats, and strains tested) for each type of standard are compared in Figure 2. As observed, depending on the type of standard employed, a same Ct may correspond to different concentrations of viral units (TCID 50 or copies). The curves with purified virus and with crude virus RNA copies (green and black lines, respectively) are those that provided the highest concentration values: around 3-4 and 5-6 Log 10 (respectively), at the highest Ct values, and close to 10 and 12 at the lowest, respectively. Standards pDNA and ivRNA provided the lowest quantification values: less than 1 Log 10 at the highest Ct, and around 7 at the lowest. In between both groups, although closer to pDNA and ivRNA standards, was the quantification curve using viral titer standard as reference (blue line). The quantification curve corresponding to fish samples was also close to that of viral titer. It is important to remark that when drawing each standard curve averaged from all replicas and repeats, the reliable maximum Ct was reduced from around 40 to 36-37.   Table 8. Since all inoculated fish showed some clinical sign, we considered as the gold standard that all the inoculated fish should have been really infected. Regardless of the fish species, the only diagnostic procedure providing the maximum DSs, DSp, PPV, and PNV values was the RT-qPCR validated in this study. The diagnostic specificity and the negative predictive values were also maximum with isolation in cell culture, nPCR, or considering both in a parallel test. However, both the diagnostic sensitivity and negative predictive values were lower than with RT-qPCR.

Assessment of the Procedure for VNNV Detection and Quantification in Field Samples
In Table 8, the observed viral titers, calculated using the standard curve for crude virus titer, are shown to be higher in sea bass than sole. That type of standard curve was employed after demonstrating that no significant differences existed between the titer data extracted from the linear standard curves by crude virus and fish tissue viral titer, as well as with the fish tissue second degree polynomial curve in the Ct range of 11-5 (Supplementary  Table S8).  The reverse transcription real-time quantitative polymerase chain reaction (RT-qPCR) ( 2 ) procedure evaluated here was tested on tisue samples of sole (A) and sea bass (B) experimentally infected with strains of types ( 1 ) RG/SJ (red-spotted grouper nervous necrosis virus (RGNNV]/striped jack nervous necrosis virus (SJNNV)) and RG/RG, respectively. The threshold (Ct) values observed ( 3 ) and the viral titers (as Log 10 (TCID 50 /ml)) ( 4 ) deduced from the line y = −0.3438x + 13.161 (extracted from the viral titer standard curve; Table 2) are shown in comparison with isolation in cell culture ( 5 ) (+, isolated; -, not isolated), and with a previously reported nested PCR procedure [22] (+, detected; -, not detected) ( 6 ); ( 7 ), diagnostic sensitivity; ( 8 ), diagnostic specificity; ( 9 ), predictive positive value; ( 10 ), predictive negative value.

2-Evaluation of the procedure in farmed and wild fish
During 2019 and 2020, the RT-qPCR procedure was evaluated in fish samples from turbot, gilt-head seabream, and Senegalese sole farms, against the traditional cell culture isolation and the nPCR procedure previously used in our laboratory [22]. Among the more than 400 samples, 37 resulted positive by RT-qPCR, with Ct values between 25.02 and 37; among those, only 10 were positive with nPCR, those with Ct values by RT-qPCR below 33, and only in four of those samples could the virus be isolated (Table 9). Table 9. Assessment of the RT-qPCR procedure in tissues from cultured fish. Comparison with nPCR and cell culture isolation. In the same period, 433 fish from 24 species, obtained from coastal fish markets, were sampled and subjected to viral detection by the same three methods. The virus was not isolated in any case, and was detected in just 11 cases by nPCR (those yielding Ct 28.75-32.7 by RT-qPCR) and in 48 fish by RT-qPCR (with Ct values from 28.75 to 33.2) (data not shown). Positive fish were mainly from three species: European hake (Merlucius merlucius), sea bass (D. labrax), and megrim (Lepidorhombus boscii). These results are part of a further report on the presence of VNNV in the Northwest Spain.

Discussion
Molecular detection of viruses is of unquestionable importance in virology. This fact is demonstrated by the large number of publications on the design, optimization and validation of PCR-derived procedures, which in the last few years has been most focused on real-time PCR. This technique, in addition, has made it possible to simplify viral quantification, which is currently gaining relevance as a basic complement in research. Concerning fish nodaviruses, RT-PCR and real-time PCR procedures have been widely reported for detection and identification of the virus [1], but very few publications deal with the design and validation of RT-qPCR for quantification of this virus. Among the latest, three were designed with a similar strategy to that reported here, namely, absolute quantification, using external standards for constructing calibration curves [14,15,17]. A fourth one used a less common approach: competitive real-time RT-PCR [24]. Most of them focused the validation assays on the RGNNV type, probably because it is the most widely spread genotype [1], although some used different genotypes and a large number of strains for susceptibility analysis [15,17].
The detection and quantification procedure described here was initially thought to be tested against RGNNV, the most frequently isolated in the Iberian Peninsula [1]. However, the primers were designed to fit four NNV strains representing the four genotypes. The 14 primer pairs pre-selected in silico for sequences in RNA-1 and RNA-2 were tested in vitro with the four strain types. Many of them failed in repeatability and reproducibility and/or in specificity; among these, some were even more appropriate for typing (and will be part of a future report). Two sets, those providing the best results in specificity and R&R (NodR1 and NodR2), were also tested against field isolates. Among them, the one designed for RNA 2 was rejected for not providing the expected repeatability against some recombinant isolates (data not shown). Therefore, the NodR1_1619/NodR1_1758 set was selected to optimize and validate the procedure for detection and quantification. This first result does not agree with the strategy followed by Panzarin et al. [15], who used a primer set corresponding to RNA 2, or by Hick and Whittington [14], who obtained better results with sets also located in that segment. However, in a more recent report [17], Baud et al. opted for RNA 1, expecting to obtain better results with any of the viral genotypes. This was also our objective. Therefore, although in some specific assays of our study (using pDNA, ivRNA, and fish tissues) the reliability of the procedure was assessed only with one viral type (RGNNV), SJNNV and RGNNV were tested with purified virus as standard, and the four genotypes with crude virus.
Crude virus was precisely the first type of standard tested as a reference of quantification in this study, and the results demonstrated that procedure was highly sensitive. In fact, the limit of detection, in terms of viral titer, averaged 8.8 TCID 50 /mL (average from the four strain types). However, to compare with similar reported studies, we needed to use the result obtained with the RGNNV type (9.88 TCID 50 / mL ), which was similar to the 10 units reported by Panzarin et al. [15]. In addition, the level of sensitivity of our procedure exceeded by 30 or 60 times the one reported by Baud et al. [17] (LOD: 10 2.5 -10 2.8 TCID 50 /mL). On the other hand, the data reported by Hick and Whittington [14] seemed to be much lower (LOD = 0.4 TCID 50 ) than those obtained with our procedure. Nevertheless, it must be noted that the value reported by those authors was absolute, corresponding to units per reaction, and it was four times higher than the LOD per reaction showed by our procedure (LOD = 8.99 × 10 −2 TCID 50 ).
In terms of RNA concentration, the limit of detection averaged 0.46 fg/µL, which was much lower than that reported by Toubanaki et al. [25] for the RGNNV (10 fg/µL) and SJNNV (100 fg/µL) types. If the corresponding genome copy numbers were deduced, that value of LOD in weight would correspond to 1.91 × 10 5 copies/mL (1.71 × 10 2 cps/reaction), which seems to be a little higher with respect to the results obtained for other RNA viruses such as VHSV and infectious hematopoietic necrosis virus (IHNV) [26,27]. However, in terms of pDNA and ivRNA copies, the LODs obtained with this procedure (2.36 and 7.86 copies/mL, respectively) were clearly below those reported for both RNA [26,[28][29][30][31] and DNA [32][33][34][35][36] fish viruses. This discrepancy is related to the large difference observed between the LOD value as crude virus RNA copies and that of the rest of standards, which was over 4 Log 10 (from 1.17 × 10 4 to 8.09 × 10 4 times higher; data extracted from Table 2). This could be due to a high presence of cellular RNA, and partially to the quantity of RNA 2 that is present, which is not detected with this procedure. Moreover, this could also explain the significantly high efficiency values observed (average E = 121.2 ± 6.94).
Regarding the specificity of the procedure, both inter-group and intra-group specificity were evaluated. For the first, IPNV and VHSV were tested in this study, and no nonspecific reactions were observed at any time. The intra-group specificity, on the other hand, was demonstrated from the results obtained with crude virus as standard: similar LOD and DR, as well as reliable curves, regardless of the viral type. The reliability of the results in terms of repeatability and reproducibility was demonstrated by the very low CV values-in most cases below 5%-in all replicas and repeats (repeatability and reproducibility, respectively). Regarding reproducibility, we must say that although in this report we only present the results obtained in three different days with the same operator, other assays were performed by a second operator with lower experience, also providing CV values below 10%. These results are not shown here because three replicas were not always applied with all standards.
Several parameters can affect the efficiency of RT-qPCR reactions. One of them is the purity of the RNA template, including the presence of protein or DNA carry-over, and the template RNA/total RNA ratio. Although any of them could be involved in the high E values observed when cvRNA was used as standard, due to our experience with the method of extraction, we are extremely confident in its effectiveness. Therefore, we suspect that in the total RNA extracted there must be a large proportion of cellular RNA. Another factor is the pipet accuracy, but the low CV values obtained in our study make us reject this as a cause for efficiency failures. Finally, a poor optimization of the primer set could be an explanation to the low E values obtained with pvRNA and pDNA, and this is actually the reason for those low values. In fact, because one of our objectives was to design a universal method of quantification appropriate for any NNV type, in the previous primer pair selection assays we had to reject some sets that provided better efficiencies with some strains but that were not reliable with others. Therefore, in the end we had to condemn efficiency in favor of group specificity.
One of the main objectives of this study was to ensure the reliability of the quantification of this virus of any type, and to decide which would be the best choice of a standard to construct the calibration curve. Therefore, special emphasis was placed on evaluating the reliability of each curve for all replicas and repeats, and with all the viral types tested, and the results demonstrated the method to be reliable with any standard. Even when the results obtained with all the NNV strains tested were averaged, the standard curves were still reliable. This indicates that any standard can actually be employed.
However, it is important to point out that in some cases the comparison of quantitative results using different standards is not possible. These differences are barely shown by other authors. Considering NNV, Hick and Whittington [14] reported a difference of 2 Log 10 in the LOD between the use of cvRNA or pDNA as standard, and Nerland et al. [40] reported a difference of 3 Log 10 in the results using viral titer and ivRNA. However, there are no studies comparing quantification using different standard curves for NNV, and for other viruses they are very scarce. With VHSV, a certain parallelism between the different standard curves (using the same standards as in the present study) was observed [41]. This differs with the results obtained with NNV in the present study, since in some cases the curves intersected in low concentrations and in others in high concentrations. This is more similar to the results observed by Vázquez et al. [37], who applied RT-qPCR for the quantification of IPNV. In that case, they only tested two standards, cvRNA and ivRNA, and observed an intersection between curves at the lowest concentrations.
Regarding the viral quantification in infected fish tissues, and with respect to the way of evaluating this standard, other approaches than spiked tissues could have been used, such as using naturally infected fish tissues. However, a standard must be something that can be repeated in similar conditions, and using tissue from naturally infected fish would make the assay impossible to reproduce exactly in the same way. From the results obtained, it seems obvious that the dynamic range considering linear regression is shortened at the side corresponding to the lowest viral titers. To this regard, if low titers are expected, it must be considered that the quantification might be compromised if the data is extrapolated from a regression line; therefore, the use of a second degree polynomial regression curve is advised. However, as shown with the tissues from challenged fish, within the Ct range 11-35 the crude virus titer standard can be used with no significant differences in the deduced titers.
The RT-qPCR procedure was also tested in challenged fish, and all the diagnostic parameters (diagnostic sensitivity and specificity, and the predictive negative and positive values) showed the highest values possible, always higher that with virus isolation, nested PCR, or a combination of both. Furthermore, RT-qPCR was also assessed in tissues from cultured and wild fish, providing a higher number of positives than with the other detection methods.

Conclusions
The main conclusion of this study is that the RT-qPCR procedure proposed was validated for diagnosis of the disease and the simultaneous quantification of the virus. In addition, taking into account the reliability of the general standard curves (a representation of many replicas and repeats, maintaining a high level of repeatability and reproducibility), we consider that these results support the use of those general curves as reference for the quantification of any genotype of NNV.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/ani11041100/s1, Figure S1: Standard curves using crude virus titers, Figure S2: Standard curves using RNA copies from crude virus, Figure S3: Standard curves using RNA copies from purified virus, Figure S4: Standard curves with plasmid DNA corresponding to the RGNNV type, Figure S5: Standard curves with in vitro transcribed RNA corresponding to the RGNNV type, Figure S6: Detection and quantification of VNNV from infected fish tissues, Table S1: Primer sets tested, Table S2: List of viral strains tested, Table S3: RT-qPCR amplification of viral RNA from crude virus, Table S4: RT-qPCR amplification of viral RNA from purified virus, Table S5: qPCR amplification of pDNA, Table S6: qPCR amplification of ivRNA, Table S7: Detection of VNNV RNA from infected fish tissues. Table S8: Statistical differences between viral titers in experimentally infected fish tissues deduced from the Ct values observed by RT-qPCR and the different standard curves.  Institutional Review Board Statement: The fish were handled according to the guidelines for good practices in laboratory animals of the European Union (directive 2010/63/EU). For the challenge, protocols approved by the Galician committee for experimental animal welfare (Xunta de Galicia, Permit ID 15010/2020/004) were followed to minimize fish suffering.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.