Extended direct lysis method for virus detection on berries including droplet digital RT-PCR or real time RT-PCR with reduced influence from inhibitors

Detection of viruses on berries is a challenging task, often hampered by the presence of RT-qPCR inhibiting substances from berry juice. A direct extraction method for virus detection (murine norovirus and GA phage) on frozen raspberries was previously published. We expanded (different types of berries and viruses) and improved the method using MobiSpin S400 columns that filter nucleic acids based on size-exclusion chromatography. While no inhibition was detected in filtered RNA, unfiltered RNA needed from 1:2 to more than 1:8 dilution in order to remove inhibition. The modified method gave recoveries of bovine norovirus around 40.8 ± 4.5% (40.0 ± 7.0%), 48.0 ± 26.0% (50.5 ± 7.8%), 28.3 ± 2.6% (45.8 ± 6.6%) from frozen (fresh) raspberries, strawberries and blueberries, respectively. For the same samples, recoveries of hepatitis A virus were 34.0 ± 5.9% (34.0 ± 6.0%), 40.0 ± 13.3% (34.2 ± 10.5%) and 23.0 ± 6.8% (31.5 ± 7.9%). For adenovirus40 (DNA virus), recoveries were 21.2 ± 8.6%, 16.0 ± 3.2% and 5.7 ± 0.2% from fresh raspberries, strawberries and blueberries respectively and column filtration did not add any improved effect. The modified method is effective and timesaving for detection of viral RNA from both fresh and frozen berries. As an emerging detection and direct quantification method, droplet digital RT-PCR was compared to RT-qPCR and was much less influenced by inhibitors when detecting mengovirus in unfiltered RNA from berries. However, for low levels of pure RNA, RT-qPCR showed slightly higher sensitivity and more stable results.


Introduction
During production of berries (pre harvest, harvest and post harvest), human viral pathogens such as norovirus (NoV) and hepatitis A virus (HAV) may contaminate products through irrigation water and workers' hands. As NoV and HAV are stable from days to weeks at room temperature, environmental contamination and poor hygiene routines may eventually result in outbreaks of disease. A total of 1589 cases of hepatitis A were linked to the consumption of frozen berries in European countries from January 2013 to August 2014 (Severi et al., 2015). Regarding NoV, contaminated frozen strawberries caused gastroenteritis in more than 10 000 people in Germany in 2012 and frozen raspberries caused multiple outbreaks in Finland in 2009 (Sarvikivi et al., 2012;Mäde et al., 2013).
Detection of viruses in food matrixes is important in the study of food borne outbreaks and in surveillance of virus contamination in foods. However, virus detection on berries has been a difficult task, mainly due to the fragile texture of strawberries and raspberries and various substances in berry juice that inhibit detection by (RT)-qPCR.
In order to concentrate virus and remove inhibitors, several methods have been developed, including polyethylene glycol (PEG) precipitation (Butot et al., 2007;Scherer et al., 2010;Mäde et al., 2013), capture of NoV using magnetic beads coated with porcine gastric mucin (Tian et al., 2005), receptor-binding capture and magnetic sequestration (RBCMS) (Pan et al., 2012), and direct RNA extraction using Trizol (Baert et al., 2008). The International Organization for Standardization (ISO) has adopted the PEG protocol for detection of HAV and NoV in fresh fruits and vegetables (ISO, 2017). This method is time consuming and extracted RNA may still contain RT-PCR inhibitors (Summa et al., 2012;Bartsch et al., 2016). A more simple and rapid direct lysis method for detecting viral genomes (murine NoV and bacteriophage GA) on frozen raspberries has been published (Perrin et al., 2015). This method was later compared to the ISO method regarding detection of human NoV in strawberries (Bartsch et al., 2016), showing comparable low recoveries (0.5 ± 0.54 versus 1.7 ± 2.3%). However, using the ISO method and inhibitor removal from RNA with MobiSpin S400 columns, a recovery of 5.6 ± 1.6 to 15.3 ± 9.7% was found (Bartsch et al., 2016). In order to increase the usefulness of the rapid method of Perrin et al, the present work developed it further, combined it with MobiSpin columns and tested it on several types of berries (strawberries, blueberries and raspberries), using different viruses. Furthermore, the original direct lysis method requires a freezing step before RNA extraction to reduce leakage of raspberry juice inhibitors into the washing buffer. As the freezing step reduces the convenience for onsite field investigation, the method was also optimized for virus detection in fresh berries. In addition, droplet digital RT-PCR (RT-ddPCR) was compared with RT-qPCR regarding tolerance to inhibitors during detection of the viral RNA.

Artificial contamination of berries
Fresh berries originating from Morocco, Spain and Norway were purchased from local food stores. Samples of fresh berries (each of 25 g), consisting of 1-2 strawberries, 8 raspberries or 16-20 blueberries, were spiked with 40 μl mixed virus suspension in PBS which contained 3.8 × 10 5 copies of HAV, 3 × 10 4 copies of BNoV and 9.6 × 10 5 copies of AdV40 by evenly spreading drops on the surface. Berries spiked with 40 μl PBS, and no virus, were included as negative controls. The berries were air dried in a fume hood at room temperature (RT) for 1 h. "Fresh berries" were immediately used for further processing, while "frozen berries" were kept at −20°C overnight before use.

Processing of berries for virus detection using direct lysis
Each sample of fresh or frozen raspberries and blueberries was divided between two 50 ml Falcon tubes, while strawberries were put in a plastic bag with zipper. A total of 4 ml lysis buffer (RT) (NucliSENS, Biomerieux, France) was added directly to each berry sample. Mengovirus was spiked into the samples in order to study its applicability as a process control. For fresh berries, the tubes were gently rotated for around 5 min, while for frozen berries the incubating time could be flexible (5-10 min, depending on berry size). As chemicals (mainly guanidium isothiocyanate) in the lysis buffer precipitate after contacting with frozen berries, it is important to wait for complete solubilizing and the buffer to regain the clear look. The lysis buffer was removed with a pipette, centrifuged at 12 000 g for 5 min at RT, and used for RNA and DNA isolation.

Extraction and purification of RNA and DNA
Nucleic acids were isolated from the total buffer volume using the NucliSENS MiniMag and 75 μl magnetic silica particles. The particles were washed twice with 400 μl buffer 1, twice with 500 μl buffer 2 and the NA from each sample were eluted in 100 μl buffer. In order to Fig. 1. Flowchart of the original (left) and modified method (right) for virus detection in berries. As chemicals in the lysis buffer precipitate on contact with the frozen berries, the incubating time should be flexible until the chemicals solubilize and the lysis buffer becomes clear. More juice will be released into the buffer during prolonged incubation and with ruptured berries. The volume ratio of lysis buffer and juice should be kept at 4: 1 and extra buffer may be added before the centrifugation step. A sediment of jelly-like substances will appear after centrifugation, while viral RNA/ DNA remains in the supernatant. evaluate MobiSpin S400 columns (Mobitec, Germany) for the capacity of reducing inhibitors, NA from two parallel samples (2 × 25 g of berries) were pooled and mixed carefully. Half of the volume was purified according to the MobiSpin manual (filtered sample) and the other half served as an unfiltered control. Briefly, the column resin was resuspended by vortexing and the column placed in a 2 ml micro centrifuge tube (1.5 ml tubes should be avoided in this step) and pre-spun for 1 min at 800 g. The tube was discarded, and the column placed in a new 2 ml tube. One hundred μl NA was applied to the matrix surface, and the column spun for 2 min at 800 g. Filtered and unfiltered NA were aliquoted and kept at −80°C until (RT)-qPCR analysis. The steps in the original method published by Perrin and the present modified method are presented in Fig. 1.

(RT)-qPCR
Primers and probe for HAV, mengo, AdV40 and BNoV were as previously described (Pintó et al., 2009;ISO, 2017;Jothikumar et al., 2005;Costafreda et al., 2006;Christensen et al., 2017). The (RT)-qPCR was carried out using the AriaMx Realtime PCR System (Agilent Genomics, USA). RNA UltraSense™ One-Step Quantitative RT-PCR System (Invitrogen, USA) was used to detect viral RNA (HAV, mengo and BNoV), while TaqMan™ Environmental Master Mix 2.0 (Applied Biosystems, USA) was used for AdV40 DNA. Two μl of NA was used in a total volume of 20 μl, and two parallels were analyzed for each sample. The RT-qPCRs were run at 50°C for 30 min, 95°C for 3 min, and finally 40 cycles of 95°C for 15 s and 60°C for 1 min. The AdV40 qPCR was run at 95°C for 10 min followed by 40 cycles of 95°C for 15 s, 58°C for 15 s and 60°C for 45 s.

Droplet digital RT-PCR (RT-ddPCR) for mengovirus
The RT-ddPCR was carried out using the QX200TM Droplet digital ™ PCR system (Bio-Rad, CA, USA) and the One-step RT-ddPCR advanced kit (Bio-Rad). Final concentrations were 900 nM of primers and 250 nM probe as recommended by the kit manufacturer. For RT-ddPCR, two μl RNA was used in the initial 20 μl reaction mixture which, together with the droplet generating oil, resulted in a final volume of 40 μl. Thermal cycling was performed on a Bio-Rad C1000 using the following protocol: 50°C for 1 h, 95°C for10 min, 40 cycles of 95°C for 15 s, 60°C for 1 min, and a final step of 98°C for 10 min. The ramp rate was set to 2°C/second and results were analyzed by using the Quantasoft 1.7.4.0917 software (Bio-Rad).

Testing of (RT)-qPCR inhibition and the effect of MobiSpin columns on NA from berries processed by direct lysis
Each berry sample was spiked with 3.8 × 10 5 copies of HAV, 3 × 10 4 copies of BNoV and 9.6 × 10 5 copies of AdV40 and processed as described. In order to find any inhibitor-reducing effects of the Mobispin column, two μl filtered and non-filtered NA from frozen and fresh berries were tested in RT-qPCR, using dilutions 1:1, 1:2, 1:4 and 1:8 (NA: DEPC water). Inhibition was evaluated by comparing the Ct values from the dilution series. A difference of 1.0 between the Ct values for a given sample and its 1:2 dilution indicates absence of inhibition when the amplification efficiency is 100%. The experiment was performed three times and two parallels of each sample (type of berry and fresh/frozen) were included in each setup. In order to increase the sensitivity of detection, increased volumes of filtered NA from fresh and frozen berries were also tested for HAV (four and eight μl) and AdV40 (four μl) as models for RNA and DNA viruses, respectively.

Estimation of virus recovery rate (RR)
Samples from the inhibition study were further used to estimate virus recovery. Based on the results in that study, filtered NA was used for detection of viral RNA and filtered/unfiltered for detection of AdV DNA. The RR was calculated by comparing the Ct value of the control (virus mixture used for spiking) to the Ct value of samples (spiked berries). The following formula was used in which E denotes the efficiency of amplification (Livak and Schmittgen, 2001): RR%=(1 + E) (Ct control−Ct sample) ×100% E was calculated from standard curves based on serial dilutions of homologous viral NA using software version 1.1 of the AriaMx Realtime PCR System (Agilent Genomics, USA). The efficiency was 0.91 for HAV, 0.95 for mengovirus, 0.99 for AdV40, and 1.03 for BNoV.

Limit of detection (LOD) of viruses on strawberries
In order to estimate the LOD for HAV, BNoV and AdV40, samples of strawberries (25 g) were spiked as described in 2.2. Two-fold serial dilutions of mixed virus were used for spiking and the amount of each virus was in the range 25 to 400 copies per sample. Then berry samples were frozen before processing by direct lysis and Mobispin filtration of RNA/DNA. The study was performed three times. Four μl of NA was used in each (RT)-qPCR. Absolute quantification of viral NA was performed using standard curves generated from 10-fold serial dilutions of linearized plasmids containing target sequences of HAV, BoNV and AdV40, respectively (GenScript, NJ, USA).

Comparison of RT-ddPCR and RT-qPCR inhibitor tolerance and sensitivity
Mengovirus was used as a model virus in order to study RT-ddPCR and RT-qPCR for detection of viral RNA in samples with inhibitors (nonfiltered RNA from spiked strawberries, three replicates) and without inhibitors (pure RNA from cultivated virus, two replicates). Thereafter, RNA was serially diluted 2-fold and two μl used in RT-ddPCR and RT-qPCR. To find any difference in sensitivity, endpoint dilution was used for pure RNA.

Statistical analysis
Any difference in virus recoveries was analyzed by Student's t-test. Differences were considered statistical significant with p < 0.05.

(RT)-qPCR inhibition and the effect of MobiSpin columns
The original direct lysis method for simple detection of viruses in raspberries includes a freezing step and 5 min incubation in lysis buffer in order to reduce the release of raspberry juice (Fig. 1). Our initial experiments with this method, gave inconsistent results on strawberries, but more stable results on blueberries and raspberries. RNA extracted from these berries needed at least a 1:4 dilution in order to remove RT-qPCR inhibitors ( Table 1).
As a possible alternative to dilution of RNA, MobiSpin S400 column filtration was included in the direct extraction protocol for frozen and fresh berries (Fig.1). Table 1 shows the difference in detection of HAV and BNoV RNA and AdV40 DNA with or without column filtration. In general, processing of fresh berries without filtering the RNA resulted in higher Ct-values compared to frozen berries. The degree of inhibition from the different types of berries was strawberries > raspberries > blueberries. However, the MobiSpin S400-Column proved highly efficient in removing inhibitors from all three types of berries. For AdV40, there was no significant qPCR inhibition in DNA isolated from any of the fresh berries (Table 1).
In order to increase the sensitivity of virus detection, larger volumes of filtered NA were tested in (RT)-qPCR for HAV and AdV40. Four μl could be used without any obvious negative effect and generally gave lower Ct values than two μl (Table 2).

Recovery rates of viruses from berries
Recovery rates for BNoV, HAV, mengovirus and AdV40 using MobiSpin column filtered NA from frozen and fresh strawberries, raspberries and blueberries are shown in Fig. 2. For AdV40, RR was also estimated from non-filtered DNA.
Comparing recoveries of viruses from fresh and frozen berries, there was no statistical significant difference, except that higher BNoV recovery could be achieved in fresh blueberries (p < 0.05).
As recoveries for RNA viruses were relatively even for fresh and frozen berries, analysis for AdV40 DNA was only performed on fresh berries. In general, the recovery of AdV40 DNA was lower compared to viral RNA, and especially from blueberries (p < 0.05). For raspberries and strawberries, a lower recovery was found for AdV40 compared to BNoV and mengovirus (p < 0.05), but there was no significant difference between AdV40 and HAV (p 0.10-0.17). Recovery of AdV40 from blueberries was also lower compared to raspberries and strawberries (p < 0.05).

Table 1
MobiSpin column filtration of RNA from berries reduces (RT)-qPCR inhibition. RNA/DNA was isolated from virus-spiked strawberries, raspberries and blueberries. The berries were processed fresh or after a freezing step and RNA/DNA was isolated with or without column filtration after extraction by the direct lysis method. The RNA/DNA was twofold serial diluted, and two μl was analyzed in a 20 μl (RT)-qPCR reaction. Ct values are presented as mean ± standard deviation from three individual replicates, two parallel samples and two parallel (RT)-qPCR reactions. Samples showing inhibition are in bold. BNoV: bovine norovirus; HAV: hepatitis A virus; AdV40: adenovirus 40; F: filtered; N-F: non-filtered. An increase of the Ct value of 1 corresponds to ca 50% reduction in virus copy number.

Table 2
The volume of filtered RNA could be increased to 20% of an (RT)-qPCR reaction volume. Two, four and eight μl MobiSpin column-filtered RNA/DNA were tested in a 20 μl total volume. Samples that show inhibition are in bold. ND; not done.

(RT)-qPCR detection limits of viruses on strawberries
In order to find the LOD for the different viruses, strawberries were spiked with serial dilutions of the mixed virus suspension, frozen and RNA extracted with the optimized direct lysis method and column filtration. Detection limits for HAV, BNoV and AdV40 on artificially contaminated strawberries are listed in Table 3. The LOD could repeatedly reach 100 copies/25 g for HAV and BNoV, and 200 copies/ 25 g for AdV40.

Inhibitor tolerance and sensitivity of RT-ddPCR and RT-qPCR
RT-ddPCR showed a tolerance to substances in unfiltered RNA from strawberries that inhibited RT-qPCR (Table 4). For RT-qPCR, at least a 1:8 dilution of unfiltered RNA was needed to get positive results in all samples. However, when comparing sensitivity of detection using pure RNA, more consistent results on low copy numbers were achieved by RT-qPCR (Table 4).

Discussion
Berries are prone to contamination by viruses through sewagecontaminated irrigation water or by infected farm workers and food handlers. The current method for molecular detection of virus in berries adopted by ISO includes steps of elution and PEG precipitation, which originate from historical protocols used to recover infectious viruses from solid matrices (Lowther et al., 2019). The procedure includes a step of pectinase digestion, as pectin is regarded as a potential RT-qPCR inhibitor. However, there are still other types of inhibitors in strawberries (Bartsch et al., 2016) and raspberries (Summa et al., 2012). In order to improve virus detection, Perrin et al. developed a direct extraction method for raspberries, which achieved a significantly higher viral extraction efficiency (46.2%) of murine NoV and GA phage than the standard ISO method (20.3%) (Perrin et al., 2015). However, this method could not be successfully applied on viruses in strawberries as the RR was 0.52% (Bartsch et al., 2016). Bartsch et al. also found that RNA from different batches of strawberries contained various levels of inhibitors, giving no obvious effect or false negative results if samples were not properly diluted (Bartsch et al., 2016). In a study by Summa et al., 28% of NoV was recovered from raspberries by the standard ISO method, but the RNA had to be diluted 10 times for use in RT-qPCR due to inhibitors (Summa et al., 2012). As it is difficult to estimate the level of inhibitors simply by the appearance of berries, which may pose a risk to consumers even with low levels of virus, a convenient and effective method to remove inhibitors, other than dilution, is vital to make the analysis applicable.
The present paper describes an improved method for virus detection in strawberries, raspberries and blueberries using direct extraction and removal of inhibitors. HAV, AdV40 and BNoV were included, as HAV and human NoV are considered the most important foodborne viral pathogens and AdV40 and 41 have been suggested as indicators of fecal contamination in water (Jothikumar et al., 2005). Bovine NoV was used as a surrogate for human NoV due to the overall similarity of the viruses (Vashist et al., 2009) and because it is non-infectious to humans and Fig. 2. Recovery rates of viruses in spiked frozen and fresh strawberries, raspberries and blueberries. RNA/DNA was extracted with the direct lysis method combined with or without filtration using MobiSpin columns. Non-filtered controls (viruses in PBS) were defined as 100%. Recovery of viruses in filtered controls and berries was presented as percentage of the nonfiltered control. Filtered controls (F-control) were included in order to demonstrate the loss of RNA through filtering. Two μl RNA was analyzed in a 20 μl (RT)-qPCR volume. BNoV: bovine norovirus; HAV: hepatitis A virus; AdV40: adenovirus 40; F: filtered; N-F: nonfiltered.

Table 3
Detection limits for hepatitis A virus (HAV), bovine norovirus (BNoV) and adenovirus 40 (AdV40) on artificially contaminated strawberries (25 g per sample) that were frozen before direct extraction and filtration with Mobispin S400 columns. Four μl RNA/DNA was analyzed in a 20 μl (RT)-qPCR. Spiking  therefore safe to handle in the lab. The use of a model virus could influence the results, however, a BNoV with a fecally origin is probably as fit as cultivated murine NoV as a model for human NoV, in berry studies. Initially, strong inhibition could be observed when ruptured, overripe, dark-red strawberries were processed by the original direct lysis method. Also for blueberries and raspberries, different levels of inhibition were due to various ripe conditions, berry intactness and plant type. When including the MobiSpin column to remove inhibitors, detection of all the RNA viruses in all berry types, fresh and frozen, was improved (Table 1) although some RNA was lost (Table 1 and Fig. 2). Purification using the MobiSpin S400 column is based on size-exclusion chromatograph. The column is pre-packed with a Sephacryl HR matrix and designed for a wide variety of separation tasks. It requires the target to be at least 20 times larger than the impurities and has been applied to remove dNTPs, oligos and salt from NA, and dye terminators or unincorporated labelled nucleotides from DNA labelling reactions (Schneider et al., 2004;Sachsenröder et al., 2012). Other chromatography-based columns have been used in removing salts, small proteins and polysaccharides from various sample matrices, such as seminal fluid or stool (Schrader et al., 2012). Bartsch et al. had used Mobispin S400 columns to purify RNA isolated from strawberries by the standard ISO method (Bartsch et al., 2016). For NoV GII in frozen berries, they found an increase in recovery from 2.83 ± 2.92% to 15.28 ± 9.73% (batch 1), and from 0.59 ± 0.49% to 5.60 ± 1.58% (batch 2). In the present study, the recovery of BNoV from strawberries was 48.0 ± 26.0% (frozen) and 50.5 ± 7.8% (fresh), while the recovery of HAV from strawberries was 40.0 ± 13.3% (frozen) and 34.2 ± 10.5% (fresh) (Fig. 2). The recovery of viral RNA in raspberries (34-63%) was comparable to the results from the original research (46.2%) (Perrin et al., 2015), but the present method reduced inhibition by RNA purification instead of dilution (1:10), which is favorable as berries may contain low amounts of virus. It is difficult to compare recoveries from different studies as berries may contain different levels of RT-qPCR inhibitors or protocols be executed with minor differences. The frozen berries used in the present study may be different to commercial berries, which are sometimes washed before freezing. The intactness of commercial frozen berries is not as good as for fresh berries, and release of juice is often observed after defrosting. In the present study, release of juice was more often observed in fresh berries during shaking with lysis buffer for 5 min. Although 4 ml lysis buffer is suggested for 25 g of berries (Fig. 1), more buffer should be added after shaking to keep the 4:1 ratio of buffer to released juice. Guanidium isothiocyanate is the component in lysis buffer that denatures proteins, lyses viral particles and releases viral NA, and a certain concentration is needed to obtain efficient release of NA. A low viral RR (0.52%) using the direct lysis method may partly be due to only 2 ml lysis buffer being used for 25 g of strawberries (Bartsch et al., 2016).
The LOD for viral RNA from strawberries achieved by the presented method (100 copies /25 g) was lower than the LOD achieved when combining the standard ISO method and column filtration (216 copies of NoV/25 g) (Bartsch et al., 2016). It was also similar to the LOD achieved by another modified direct extraction method on raspberries in which the RNA was diluted to avoid inhibitors (Summa and Maunula, 2018). Summa and Maunula combined direct NA extraction with minor supplement of PEG and got four positive results out of nine with 100 copies of NoV per 25 g raspberries. In the present study viral LOD was estimated for strawberries only, as initial testing showed similar recoveries of viral RNA for the three berry types, and strawberries contained the highest amounts of inhibitors. The relevance of comparing LODs between publications is, however, limited since no reference material has been used in the studies.
The sensitivity of virus detection depends on the RR, purity of viral RNA and the sample volume used in the RT-qPCR. As RNA filtered by the MobiSpin S400 column showed no inhibition (Table 1), the RNA volume could be doubled to four μl (20% of the total volume) in the RT-qPCR (Table 2). Although column filtering reduced the final amount of RNA with an average of 37% (Fig. 2), doubling the volume used for analysis is beneficial when analyzing berries that might contain low level of virus (giving pure RNA in an amount of 126% compared to using two μl of unfiltered RNA). In comparison, unfiltered RNA from fresh strawberries needed at least 1:8 dilutions to remove the influence of inhibitors, reducing the RNA amount to 12.5%.
The column filtration step was not necessary when viral DNA was the target, which indicates that the inhibition mainly lies in the reverse transcription (RT) of RNA (Table 1 and 2). The RR of AdV DNA was low compared with viral RNA, especially from blueberries (Fig. 2), despite a relatively low interference from inhibitors. The low recovery of AdV DNA from blueberries may be due to substances that bind to viral DNA or silica particles and thereby interfere with isolation of NA, or due to viral DNA being more easily adsorbed on blueberry matrix after lysis of the virus particles. Although the RR of AdV DNA was low compared to the RNA viruses, screening of adenoviruses could be used as a parameter for fecal contamination of strawberries and raspberries. However, further studies should be performed in order to optimize elution of viral DNA from berries.
Virus detection in food matrices should include a process control in order to monitor the quality of the total operation. As the standard ISO method includes mengovirus, the applicability of this virus as a process control for the optimized direct lysis method, was tested. Recoveries of mengovirus were higher than for HAV in blueberries and raspberries (p < 0.05), but not in strawberries (Fig. 2). Recoveries of mengovirus Table 4 Comparison of RT-qPCR and RT-ddPCR regarding inhibitor tolerance (left) and sensitivity (right), using mengovirus RNA. Unfiltered RNA isolated from spiked strawberries was used in the tolerance test, while pure RNA from cell culture supernatant was used in the sensitivity study. The RNA was two-fold serially diluted and two μl used in (RT)-qPCR and RT-ddPCR. Results are presented individually. Ct: Ct-value; copy: RNA copy number; N: negative. were not significantly different to BNoV in raspberries (p 0.06-0.24).
The results indicate that the recovery of mengovirus is relatively stable and that this virus could be used as a process control for the optimized protocol. The recovery of mengovirus was higher from raspberries and blueberries than from the control sample after filtration (Fig. 2). This can partly be explained by the variation of the results, but also by the fact that mengovirus was added directly to the berries soaked in lysis buffer and not onto the berries before drying, as for the other viruses.
Recently ddPCR has been confirmed to be a sensitive method enabling accurate quantification of a target NA sequence (Pinheiro et al., 2012). As ddPCR is based on the "most probable number" approach, it was interesting to compare RT-ddPCR with RT-qPCR regarding inhibitor tolerance and sensitivity of virus detection. Detection of unfiltered mengovirus RNA from strawberries was used to test the "proof of principle" as these extracts were highly inhibitory (Fig. 2). The results show that RT-ddPCR had a tolerance to substances that inhibited RT-qPCR (Table 4). For RT-qPCR, at least 1:8 dilutions of the RNA were necessary to remove the influence of inhibitors. However, when comparing the methods for the analysis of low levels of pure RNA, RT-qPCR seemed to give results that were more consistent compared to the RT-ddPCR (Table 4). The phenomenon of qPCR showing higher sensitivity and less variability than ddPCR with low viral load has also been noticed for the DNA viruses cytomegalovirus (Hayden et al., 2013(Hayden et al., , 2016 and hepatitis B virus (Boizeau et al., 2014). Digital droplet PCR includes automatic partitioning of samples (20 μl into 20,000 droplets) and three steps of sample transfer, which may result in lower repeatability for detection of low copy numbers (< 10 copies) of pure NA. However, for detection of RNA in environmental or food samples that contain inhibitors, as demonstrated in the present study, RT-ddPCR has the advantages of direct absolute quantification and higher tolerance to RTinhibitors compared to RT-qPCR. As RT-qPCR inhibition was found for all three RNA viruses used in the present study, RT-ddPCR diagnostic is most likely advantageous for a range of RNA viruses in berries.
The conclusion based on the present study is that the optimized direct lysis method, which included column purification of RNA, was time-saving, efficient, and could be used in (RT)-qPCR analysis of different viruses from several types of berries under different conditions (fresh/frozen, intact/ruptured). As column purification also reduces the amount of RNA, RT-ddPCR, which showed resistance to inhibitors, is a very suitable alternative for detection and direct quantification of viral RNA in the berries.