Occurrence and reduction of F-specific RNA bacteriophage genotypes as indicators of human norovirus at a wastewater treatment plant

Noroviruses (NoVs) are a commonly reported cause of nonbacterial gastroenteritis and have caused numerous outbreaks of gastroenteritis in Japan (NIID 2018). The illness:infection ratio of NoVs is very high, and approximately 80% of individuals who come in contact with NoVs will exhibit signs of infection, such as diarrhea and vomiting (Moe 2009). NoVs have been detected at comparatively high concentrations and frequencies in raw and treated sewage in Japan, and these concentrations increase in winter due to the increase in the number of patients with NoV gastroenteritis during this season (Katayama et al. 2008; Hata et al. 2013; NIID 2018). Therefore, it is important to reduce the concentrations of NoVs through wastewater treatment to lessen human health risk from waterborne diseases.

F-specific RNA bacteriophages (FRNAPHs) infect Escherichia coli cells that possess F pili. FRNAPHs consist of a single-stranded RNA molecule in an icosahedral capsid 20–30 nm in diameter. They have been suggested to be viral indicators because their size, structure, behavior, abundance and survival rate in both the environment and water treatment facilities are similar to those of NoVs (IAWPRC 1991).

FRNAPH is classified into four different genotypes (GI–GIV), with each having different occurrence and resistance in a water environment and stability against water treatment (Schaper et al. 2002a; Haramoto et al. 2012, 2015; Hata et al. 2016; Lee et al. 2018). Previous studies have reported that FRNAPH GI and GIV are the dominant genotypes in the feces and wastewater of animals, whereas FRNAPH GII and GIII are the dominant genotypes in human feces and raw sewage from municipal wastewater treatment plants (WWTPs) (Schaper et al. 2002b; Cole et al. 2003). FRNAPH GI has the highest resistance in natural environments and to water treatments, such as disinfection processes, compared with the other genotypes (Schaper et al. 2002a; Haramoto et al. 2015; Hata et al. 2016; Lee et al. 2018). Conversely, FRNAPH GIV has the lowest resistance to various inactivation processes (Schaper et al. 2002a), but is rare in environmental water and wastewater samples (Schaper et al. 2002b; Haramoto et al. 2012, 2015; Hata et al. 2016).

GII and GIII have been reported to be more predominant than GI and GIV in influents of WWTPs; this is believed to be due to the dominance of GII and GIII in human feces, as described above (Cole et al. 2003; Haramoto et al. 2012, 2015; Hata et al. 2013; Hartard et al. 2015). However, FRNAPH GI showed the lowest reduction during wastewater treatment, followed by FRNAPH GII and GIII (Hata et al. 2013; Haramoto et al. 2015). Thus, FRNAPH GI has been suggested to be an appropriate indicator for virus reduction during wastewater treatment. In addition, other studies have proposed FRNAPH GII as a potential indicator of NoVs because of the relationship between the presence of NoVs and FRNAPH GII in shellfish and environmental samples (Vergara et al. 2015; Hartard et al. 2016, 2018). According to the results of these studies, the occurrence and reduction of FRNAPH GII could also be measured and used as an indicator of NoVs in WWTPs. Although the occurrence and reduction of NoV and FRNAPH genotypes in WWTPs have been well studied (Haramoto et al. 2012, 2015; Hata et al. 2013), these studies did not investigate the relationship between NoV and FRNAPH genotypes in WWTPs. Evaluating this relationship is important to determine whether FRNAPH genotypes can be used as indicators of NoV in wastewater treatment scenarios. Therefore, this is an area for novel study, which we aimed to address in our research.

Although the occurrence and reduction of FRNAPH GI–GIII in WWTPs has been well studied, only a limited number of studies have investigated FRNAPH GIV because of its low detection rate (Haramoto et al. 2012). FRNAPH GIV has been reported to be the dominant genotype in feces and wastewater of cattle, poultry and pigs in France, Spain, South Africa and the USA (Schaper et al. 2002b; Cole et al. 2003; Hartard et al. 2015). Therefore, it seems possible that FRNAPH GIV is predominant in WWTPs, where livestock waste is mixed into WWTP influent. Furthermore, the relationship of FRNAPH GIV with NoVs despite animal-specific FRNAPH genotypes would make it a useful indicator of NoVs in WWTPs where the livestock waste is mixed into WWTP influent.

This study aims to clarify the occurrence and reduction of indigenous FRNAPH genotypes in WWTPs. Furthermore, we also aim to determine the occurrences and reductions of NoVs and compare them with the results of FRNAPH genotypes to identify NoV indicators in WWTPs.

Influent and effluent samples were collected from a pilot-scale WWTP (6.2 m³/d), which uses the conventional activated sludge treatment process, with 1,700–2,100 mg/L of mixed liquor suspended solids. This pilot-scale WWTP takes its feed water from the influent of a full-scale WWTP located in Ibaraki Prefecture, Japan. This full-scale WWTP (capacity 165,000 m³/day, supporting the equivalent of 2.9 million inhabitants) was designed to eliminate classical pollutants (carbon, nitrogenous compounds and phosphorus). According to a report about livestock waste treatment published by the Ministry of Agriculture, Forestry and Fisheries in Japan (MAFF 2011), approximately 5% of livestock waste is treated by the public sewage system in this area. A total of 17 WWTP influent and effluent samples were collected once a month from May 2017 to April 2018, with extra samples collected in May and June 2017, November 2017 (two samples) and January 2018. The WWTP receives rainwater along with wastewater, and our survey period included one incidence of rainfall in July 2017. The characteristics of the WWTP influent and effluent samples are summarized in Table 1.

FRNAPH concentrations in each sample were measured using three different quantification assays. The first was a conventional plaque assay, which detects total infectious FRNAPHs. The second was a reverse transcription-quantitative polymerase chain reaction assay (RT-qPCR), which detects FRNAPH genomes. However, RT-qPCR cannot differentiate between infective and inactive FRNAPH genotypes. The third method was an integrated culture–RT-PCR combined with a most probable number (MPN) assay (IC–RT-PCR–MPN). This can detect infectious FRNAPH genotypes (Hata et al. 2016; Lee et al. 2018).

The concentration of total infectious FRNAPHs was measured by conventional plaque assay using Salmonella enterica serovar Typhimurium WG49 as the host strain according to the ISO standard 10705-1 and the previous study (ISO 10705-1 1995; Mooijman et al. 2002). For the WWTP effluent sample, 1 ml or 5 mL of sample was mixed with agar medium. Ten-fold serial dilutions of the WWTP influent sample were performed to generate dilutions ranging from undiluted to 10⁻², and 1 mL of each dilution was mixed with agar medium. Duplicate plating was performed for each water sample and negative control. The detection limits of the WWTP influent and effluent samples were 2.7 and 2.0 log₁₀ PFU/L, respectively.

The collected water samples were concentrated by a polyethylene glycol sedimentation method, using polyethylene glycol #6000 (Nacalai Tesque, Japan) (Yasui et al. 2016). Polyethylene glycol #6000 at a final concentration of 8% and sodium chloride (Wako, Japan) at a final concentration of 0.4 M were added to the water samples (10 mL of the WWTP influent sample and 200 mL of the WWTP effluent sample). The water samples were stored overnight at 4 °C and then centrifuged at 12,000 × g for 30 min. The sediment was resuspended in DEPC-treated water (Nippon Gene, Japan) and the sediment was collected. A concentrated sample was obtained with a final volume of approximately 1 mL. Viral RNA in the concentrated samples was extracted using a QIAamp Viral RNA Mini QIAcube Kit and QIAcube (Qiagen, Japan) according to the manufacturer's protocol. This step was followed by RT for cDNA synthesis with a High Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific, Japan). Quantitative detection of FRNAPH and NoV genomes from the obtained cDNA was performed using a QuantStudio™ 12 K Flex Real-Time PCR System (Thermo Fisher Scientific, Japan). The reaction mixture consisted of 5.0 μL of cDNA, 12.5 μL of TaqMan gene expression master mix (Thermo Fisher Scientific, Japan), 1.0 μL each of 10 μM forward and reverse primers and 0.5 μL of 5.0 μM TaqMan probe. Sequences of primers and TaqMan probes for the detection of FRNAPH and NoV genotypes were derived from previous studies by Wolf et al. (2008) and Kageyama et al. (2003), respectively. The genome copy numbers (copies) were determined from previously prepared standard curves. Separate curves were created using plasmid DNA that contained various virus gene sequences targeted for amplification and 10-fold serial dilutions (from 10⁵ to 10⁰ copies per reaction). The qPCRs were performed in duplicate and were considered positive only when both PCR tubes fluoresced with sufficient intensity and the average cycle threshold value was <40. The quantification limits, corresponding to the detection of 10⁰ copies per reaction in both PCR tubes, were 4.2 log₁₀ copies/L for the WWTP influent sample and 2.9 log₁₀ copies/L for the WWTP effluent sample for quantifying of FRNAPH and NoV genomes by RT-qPCR. When no increase in fluorescence was observed within 40 cycles, the sample concentration was classified as ‘Not detected’ (ND).

Murine NoV (MNV, S7-PP3 strain) was used as a sample process control to determine the efficiency of RNA extraction-RT-qPCR. MNV was provided by Dr Y. Tohya (Nihon University, Japan) and was propagated in RAW 264.7 (ATCC TIB-71) cells (American Type Culture Collection, USA). Prior to extracting the viral RNA, 5.0 μL of a known concentration of MNV stock (10⁴–10⁵ copies/5.0 μL) was spiked into 140 μL of concentrated sample as well as into pure water for use as a control (i.e., with no inhibition). MNV-RNA was co-extracted with other indigenous viral RNAs from the samples and the MNV-RNA yield was determined by RT-qPCR (Kitajima et al. 2010). da Silva et al. (2007) reported that the recovery rate of the process control can be classified as poor (<1%), acceptable (1–10%) or good (>10%). Recovery rates of spiked MNV in the WWTP influent and effluent samples were 39–117% and 53–113%, respectively, which indicated no significant inhibitory effect on the RNA extraction or RT-qPCR procedure. Thus, they were not used to adjust the concentrations of FRNAPH and NoV genomes.

To quantify infectious FRNAPH genotypes, IC–RT-PCR–MPN was performed as described in our previous study (Lee et al. 2018). Infectious FRNAPH genotypes in the samples were primarily propagated by mixing with a liquid medium containing WG49 as the host strain. Genotyping based on RT-PCR was subsequently applied to these propagated infectious FRNAPH genotypes, followed by quantification using the MPN method. The sample volumes (n = 3 for each volume) of the WWTP influent and effluent samples were 1, 0.01, 0.001 and 0.0001 mL and 100, 10, 1 and 0.1 mL, respectively. The detection limits of the WWTP influent and effluent samples were 2.48 and 0.48 log₁₀ MPN/L, respectively.

All statistical analyses were performed using IBM SPSS PASW Statistics 18. Initially, all concentrations (log₁₀ copies/L) and log₁₀ reductions data were assessed for normality using the Shapiro–Wilk test. All the distributions were found to be normal; parametric tests were used in the statistical analysis. The significant differences in the concentrations and log₁₀ reductions among FRNAPHs and NoVs were determined by one-way analysis of variance (ANOVA) and Tukey's multiple-comparison test. The significant differences in the reductions of FRNAPH genotypes by IC–RT-PCR–MPN and RT-qPCR were determined by Student's t-test. Pearson's correlations were used to test the relationships among concentrations and log₁₀ reductions of FRNAPHs and NoVs. If the resultant P value was <0.05, the difference and correlation were considered to be significant.

Using IC–RT-PCR–MPN, infectious FRNAPH GII and GIII were detected in all influent samples (17 out of 17) and GI and GIV were detected in 16 (94%) out of 17 influent samples (Figure 1(a) and 1(b) and Table 2). The mean concentrations of the infectious FRNAPH genotypes (5.1–5.4 log₁₀ MPN/L), with the exception of GI, were not significantly different (P > 0.05). The mean concentration of GI (3.6 log₁₀ MPN/L) was the lowest among the infectious FRNAPH genotypes (P < 0.05). The concentration of GIV showed seasonal variation during 1 year, whereas this was not observed in the other infectious FRNAPH genotypes (Figure 1(a) and 1(b)). During the one incidence of rainfall (July 2017), only the concentration of infectious FRNAPH GIV was increased in the WWTP influent samples (Figure 1(a)).

The total infectious FRNAPHs were detected by plaque assay in all influent samples (17 out of 17) at a concentration similar to the highest concentration among the infectious FRNAPH genotypes (Figure 1(c) and Table 2). The mean concentration of total infectious FRNAPHs (5.6 log₁₀ PFU/L) was not significantly different from those of infectious FRNAPH GII, GIII and GIV (P > 0.05).

Using RT-qPCR, FRNAPH GI and GIV genomes were detected in 12 (71%) and 15 (88%) out of 17 influent samples, respectively (Figure 1(d) and Table 2). FRNAPH GII and GIII genomes were detected by RT-qPCR in 17 (100%) and 15 (88%) out of 17 influent samples, respectively (Figure 1(e) and Table 2). Among the mean concentrations of FRNAPH genomes determined by RT-qPCR in the influent samples (Table 2), the concentration of GII (8.5 log₁₀ copies/L) was the highest (P < 0.05), followed by that of GIII (6.3 log₁₀ copies/L) and GIV (6.0 log₁₀ copies/L), which were not significantly different (P > 0.05). The mean concentration of GI (5.0 log₁₀ copies/L) was the lowest among the FRNAPH genomes, as determined by RT-qPCR (P < 0.05). Seasonal variations during 1 year were not observed in the concentrations of FRNAPH genomes (Figure 1(d) and 1(e)).

NoV GI and GII genomes were detected by RT-qPCR in 16 (94%) and 17 (100%) out of 17 influent samples, respectively (Figure 1(f) and Table 2). The mean concentration of NoV GI (5.9 log₁₀ copies/L) was not significantly different from those of FRNAPH GI, GIII and GIV, as detected by RT-qPCR (P > 0.05). The mean concentration of NoV GII (7.8 log₁₀ copies/L) was higher than that of NoV GI (P < 0.05), but the difference was not significant (P > 0.05). Seasonal variations of NoV concentration were not observed during our survey period.

The detection rates obtained using IC–RT-PCR–MPN and plaque assay (84–100%) were higher than those obtained using RT-qPCR (71–100%) in the results of FRNAPH genotypes.

In the WWTP influent samples, we investigated the correlation among the concentrations of FRNAPH genotypes by IC–RT-PCR–MPN and RT-qPCR, total infectious FRNAPHs by plaque assay and NoVs by RT-qPCR. Table 3 shows the significant relationships (P < 0.05) and all the results are presented in Table S1 (available with the online version of this paper).

The concentration of FRNAPH GII correlated with that of FRNAPH GIII by both IC–RT-PCR–MPN (r = 0.51, P < 0.05) and PCR (r = 0.88, P < 0.01) in the influent samples (Figure 1(b) and 1(e) and Table 3). Moreover, a correlation between the concentrations of NoV GI and GII by PCR (r = 0.69, P < 0.01) was observed (Figure 1(f) and Table 3). However, no FRNAPH genotype concentration was correlated with NoV concentration (P > 0.05; Table S1).

Infectious FRNAPH genotypes, excluding GIV (14 out of 17, 82%), were detected by IC–RT-PCR–MPN in all effluent samples (Figure 2(a) and 2(b) and Table 2). The mean concentrations of the infectious FRNAPH genotypes (2.3–3.0 log₁₀ MPN/L), except GIV, were not significantly different (P > 0.05). The mean concentration of GIV (2.0 log₁₀ MPN/L) was the lowest among the infectious FRNAPH genotypes, but did not significantly differ from that of GIII (P > 0.05). Seasonal variation in the concentration of infectious FRNAPH genotypes was not observed (Figure 2(a) and 2(b)).

Total infectious FRNAPHs were determined by plaque assay in all effluent samples, with a mean concentration of 3.5 log₁₀ PFU/L (Figure 2(c) and Table 2), which was not significantly different from those of infectious FRNAPH GI and GII (P > 0.05).

Among the mean concentrations of FRNAPH genomes determined by RT-qPCR in the effluent samples, the concentration of GII (6.9 log₁₀ copies/L) was the highest (P < 0.05), followed by those of GIV (4.1 log₁₀ copies/L), GI (3.9 log₁₀ copies/L) and GIII (3.8 log₁₀ copies/L), but there were no significant differences between GI, GIII and GIV concentrations (P > 0.05; Figure 2(d) and 2(e) and Table 2). Seasonal variations in FRNAPH genomes by RT-qPCR were not observed.

The mean concentration of NoV GI (4.5 log₁₀ copies/L) determined by RT-qPCR in the effluent samples was not significantly different from those of FRNAPH GI, GIII and GIV genomes determined by RT-qPCR (P > 0.05; Figure 2(f) and Table 2). The mean concentration of NoV GII (6.2 log₁₀ copies/L), which was not significantly different from that of FRNAPH GII (P > 0.05), was higher than that of NoV GI (P < 0.05). Seasonal variations of NoVs were not found in WWTP effluent samples (Figure 2(f) and Table 2).

The detection rates obtained using IC–RT-PCR–MPN and plaque assay (82–100%) were higher than those obtained using RT-qPCR (47–100%) in the results of FRNAPH genotypes.

The log₁₀ reductions of FRNAPHs and NoVs were calculated from sample sets that were positive for both WWTP influent and effluent samples (Figure 3).

The lowest mean log₁₀ reduction of infectious FRNAPH genotypes determined by IC–RT-PCR–MPN was that of FRNAPH GI; this showed a reduction of 0.7 ± 0.5 log₁₀ (P < 0.05), which was 2.5 log₁₀ higher than that of GIV (3.2 ± 0.9 log₁₀). The mean log₁₀ reduction of infectious FRNAPH GII (2.1 ± 0.7 log₁₀) by IC–RT-PCR–MPN was lower than that of GIII (3.1 ± 0.6 log₁₀) (P < 0.05) and was similar to that of total infectious FRNAPHs by plaque assay (2.1 ± 0.5 log₁₀) (P > 0.05). Among the mean log₁₀ reductions of FRNAPH genomes determined by RT-qPCR, GI (1.4 ± 0.9 log₁₀) showed the lowest log₁₀ reduction, followed by GII (1.6 ± 0.7 log₁₀), GIV (2.1 ± 1.1 log₁₀) and GIII (2.5 ± 0.7 log₁₀). Comparing the log₁₀ reductions among FRNAPH genotypes by IC–RT-PCR–MPN and RT-qPCR, only GIV was observed to be significantly different (P < 0.05). The mean log₁₀ reductions of NoV GI (1.4 ± 0.6 log₁₀) and GII (1.6 ± 0.6 log₁₀) were not significantly different from those of FRNAPH GI, GII and GIV genomes by RT-qPCR (P > 0.05).

Next, we investigated the correlation among the log₁₀ reductions of FRNAPH genotypes during wastewater treatment by IC–RT-PCR–MPN and RT-qPCR, total infectious FRNAPHs by plaque assay and NoVs by RT-qPCR. The significant relationships (P < 0.05) are shown in Table 4 and all the results are presented in Table S2 (available with the online version of this paper).

The log₁₀ reductions of total infectious FRNAPHs by plaque assay correlated with those of infectious FRNAPH GI (r = 0.52, P < 0.05) and GII (r = 0.49, P < 0.05) by IC–RT-PCR–MPN. A correlation between FRNAPH GII and GIII genomes determined by RT-qPCR was observed (r = 0.87, P < 0.01). In addition, a correlation between NoV GI and GII, as determined by RT-qPCR, was observed (r = 0.82, P < 0.01). The log₁₀ reduction of NoV GII correlated with those of FRNAPH GII (r = 0.69, P < 0.01) and GIII (r = 0.67, P < 0.05) genomes by RT-qPCR, which are human-specific FRNAPHs. In contrast, animal-specific FRNAPH genotypes (GI and GIV) did not correlate with NoV GI and GII or with human-specific FRNAPHs (GII and GIII) (Table S2).

This study aimed to investigate the occurrence and reduction of FRNAPH genotypes in an attempt to identify indicators of NoV in a WWTP.

FRNAPH GI and GIV are generally associated with animal pollution because they have been detected in the feces and wastewater of cattle, swine, sheep, horses, chickens, gulls, ducks and geese in France, Spain, South Africa and the USA (Schaper et al. 2002b; Cole et al. 2003; Hartard et al. 2015). FRNAPH GI and GIV have also been reported to appear in domestic sewage, but at lower frequency and numbers than GII and GIII (Schaper et al. 2002b; Haramoto et al. 2012, 2015; Hata et al. 2013), which are generally associated with human pollution (Cole et al. 2003). In particular, many previous studies have reported that FRNAPH GIV was either not detected or detected at low concentrations in raw sewage of WWTPs in Japan and many other countries (Ogorzaly & Gantzer 2006; Haramoto et al. 2012, 2015; Hata et al. 2013; Hartard et al. 2015). Furthermore, a previous study that investigated the concentrations of infectious FRNAPH genotypes in primary effluents of WWTPs in Japan reported that GIV was found in 50% of these plants and in concentrations lower by 1.7 log₁₀ than the predominant GII and GIII (Lee et al. 2018). In our present study, however, FRNAPH GIV was detected in the influent samples with a generally high detection rate (IC-PCR, 94%; PCR, 88%) and concentration (Figure 1(a)), which was at the same level as the human-specific FRNAPH genotypes (GII and GIII). The numbers of cattle, poultry and pigs in the region (Ibaraki Prefecture) targeted in this study were higher than in the regions (Yamanashi, Shiga and Okinawa Prefecture) of previous studies (Haramoto et al. 2012, 2015; Lee et al. 2018) according to the Statistical Yearbook of Ministry of Agriculture, Forestry and Fisheries in Japan (MAFF 2016). Schaper et al. (2002b) showed that FRNAPH GIV was most predominant in the feces and wastewater of cattle, poultry and pigs. Furthermore, approximately 5% of the livestock waste was treated by the public sewage system in our targeted research area. Therefore, we proposed that FRNAPH GIV is predominant in WWTPs where the livestock waste is mixed into WWTP influent. Conversely, Hartard et al. (2015) recently reported detecting FRNAPH GII and GIII in feces and wastewater of duck and cattle origin, respectively, using the RT-PCR assays designed by Wolf et al. (2008). The RT-PCR assays designed by Ogorzaly & Gantzer (2006) were more specific for wastewater from human origin. Therefore, further investigation is needed for microbial source tracking using FRNAPH genotypes by Ogorzaly and Gantzer's RT-PCR systems.

Furthermore, the seasonality of infectious FRNAPH GIV by IC–RT-PCR–MPN was observed in influents with higher concentrations in winter than in summer (Figure 1(a)), with the exception of an incident of rainfall (July 2017). Previous studies have shown that infectious FRNAPH GIV in raw sewage was detected only in winter (Haramoto et al. 2012, 2015). Hata et al. (2016) also showed that infectious FRNAPH GIV was found only in river samples in winter. Schaper et al. (2002a) reported that, among the FRNAPH genotypes, FRNAPH GIV demonstrated the lowest resistance to natural inactivation processes such as temperature, ammonia and extreme pH. Lee et al. (2018) also showed that, among the FRNAPH genotypes, FRNAPH GIV had the highest ultraviolet sensitivity. Thus, the seasonal trends in infectious FRNAPH GIV in WWTP may be attributed to the inactivation that occurs due to higher temperatures and the effect of the sun's ultraviolet radiation in summer compared with winter. Furthermore, it should be noted that the concentration of FRNAPH GIV by RT-qPCR did not show seasonality because RT-qPCR detects genomes regardless of infectivity (Figure 1(d)). On the other hand, infectious FRNAPH GI showed no seasonal variation over 1 year in the WWTP influents (Figure 1(a)). FRNAPH GI has been reported to show the highest resistance to natural inactivation processes (temperature, ammonia and extreme pH) and disinfection processes (chlorination and ultraviolet) among the FRNAPH genotypes (Schaper et al. 2002a; Muniesa et al. 2009; Lee et al. 2018). Thus, in contrast to FRNAPH GIV, the concentration of infectious FRNAPH GI was stable over 1 year in the WWTP influents because of its resistance to inactivation. Furthermore, the concentration of infectious FRNAPH GI in the WWTP effluent samples was higher than that of infectious FRNAPH GIV, whereas GIV was more predominant than GI in the influents (Figure 2(a)) because among the FRNAPH genotypes, GI had the lowest mean log₁₀ reduction by wastewater treatment (Figure 3). Previous studies have also shown that FRNAPH GI has the most resistance to wastewater treatment among the indicator bacteria, enteric viruses and other FRNAPH genotypes (Hata et al. 2013; Haramoto et al. 2015). Thus, FRNAPH GI was suggested as an appropriate viral indicator during wastewater treatment as it represents the worst-case scenario. On the other hand, FRNAPH GIV was efficiently reduced by wastewater treatment. As described previously, the reason for this might be that FRNAPH GIV was easily inactivated by various processes. The mean log₁₀ reduction of FRNAPH GIV shown by RT-qPCR was significantly lower than that shown by IC–RT-PCR–MPN (P < 0.01), whereas the mean log₁₀ reductions shown by IC–RT-PCR–MPN and RT-qPCR of the other FRNAPH genotypes were not significantly different (Figure 3). Thus, this result indicates that FRNAPH GIV was inactivated during wastewater treatment. To our knowledge, this is the first study showing seasonality and reduction, including inactivation, of the infectious FRNAPH GIV.

Generally, FRNAPH GII and GIII showed high detection rates and concentrations in influent samples (Figure 1(b) and 1(e)). Furthermore, the concentration of FRNAPH GII correlated with that of FRNAPH GIII, as determined by both IC–RT-PCR–MPN (r = 0.51, P < 0.05) and RT-qPCR (r = 0.88, P < 0.01) in the influent samples (Table 3). These results confirm that FRNAPH GII and GIII flow into WWTP from the same source (i.e., human feces). The log₁₀ reductions of FRNAPH GIII by IC–RT-PCR–MPN and RT-qPCR were significantly higher than those of GII (P < 0.05). Hata et al. (2013) also showed that FRNAPH GIII was more effectively removed than GII during wastewater treatment. On the other hand, a correlation between the log₁₀ reductions of FRNAPH GII and GIII by RT-qPCR (r = 0.87, P < 0.01) was observed (Table 4). The removal mechanisms of FRNAPH GII and GIII may appear to be similar, but GIII was more strongly adsorbed on activated sludge than GII because of the difference in surface properties, such as hydrophobicity (Langlet et al. 2008). As a result, FRNAPH GIII could be more effectively removed than GII during wastewater treatment. However, further investigations are needed to reveal the surface properties of FRNAPH genotypes to activated sludge during wastewater treatment.

Occurrences of NoVs in wastewater have been well studied. Previous studies have reported that NoVs are more prevalent in winter than in summer because of an epidemic period of NoV gastroenteritis in Japan (Katayama et al. 2008; Hata et al. 2013). In our results, however, their concentrations were stable in the WWTP influent samples during our survey period and were not higher during winter (Figure 1(f)). According to the Infectious Agents Surveillance Report during 2013/14–2017/18 from the National Institutes of Infectious Diseases in Japan (NIID 2018), the incidence of NoV gastroenteritis during the 2017/18 season was less than that during other seasons (∼120 vs ∼170 to ∼ 370 on a peak number of agents during the 2017/18 season vs the 2013/14–2016/17 season, respectively). Thus, this could be the reason for stable concentrations of NoVs in the influent samples during our survey period (2017/18 season). The relationship between the presence of NoVs and FRNAPH genotypes in shellfish and tropical urban freshwater catchment has been investigated in previous studies (Vergara et al. 2015; Hartard et al. 2016, 2018). These studies have reported that the concentration and presence of FRNAPH GII correlated with those of NoVs. However, in our study, we did not observe a relationship between the occurrence of NoVs and FRNAPH genotypes in WWTP influents (Table 3). On the other hand, the log₁₀ reduction of NoV GII by RT-qPCR correlated with those of FRNAPH GII (r = 0.69, P < 0.01) and GIII (r = 0.67, P < 0.05) by RT-qPCR during wastewater treatment (Table 4), whereas animal-specific FRNAPH genotypes (GI and GIV) did not correlate. It might be proposed that NoV GII, FRNAPH GII and FRNAPH GIII were similarly removed by wastewater treatment. Furthermore, the mean log₁₀ reduction of NoV GII by RT-qPCR was not significantly different from that of FRNAPH GII by RT-qPCR (P > 0.05). These results suggest that FRNAPH GII could be used as an appropriate indicator of NoV GII during wastewater treatment.

We detected FRNAPHs in WWTP influent and effluent using three quantification assays. The detection rates obtained using culture-based methods (plaque assay and IC–RT-PCR–MPN) were higher than those obtained using RT-qPCR (Table 2) because of their lower detection limits. We also evaluated the log₁₀ reduction of FRNAPHs during wastewater treatment using the three quantification assays. The log₁₀ reductions of FRNAPH genotypes, except GI, by RT-qPCR were lower than those by IC–RT-PCR–MPN because RT-qPCR cannot evaluate inactivation (Figure 3). It is apparent that RT-qPCR underestimates the reduction of infectious FRNAPHs during wastewater treatment. It is therefore important that culture-based methods, such as plaque assay and IC–RT-PCR–MPN, are employed to accurately determine viral reduction, including inactivation. On the other hand, the log₁₀ reductions of total infectious FRNAPHs by plaque assay correlated with those of infectious FRNAPH GI and GII by IC–RT-PCR–MPN (Table 4). FRNAPH GII has the lowest mean log₁₀ reduction among the predominant infectious FRNAPH genotypes (GII, GIII and GIV) in WWTP influents. Despite its lower concentration in WWTP influents than those of other FRNAPH genotypes (GII, GIII and GIV), FRNAPH GI has the lowest mean log₁₀ reduction during wastewater treatment. Thus, the log₁₀ reductions of total infectious FRNAPHs are reflected in those of FRNAPH GI and GII. Based on the above findings, we suggest FRNAPH GI and GII as potential indicators for viral reduction during wastewater treatment because of the worst-case scenario and correlation with NoV, respectively. Therefore, the reduction of total infectious FRNAPHs, which is easily determined by conventional plaque assay, also holds potential as an indicator for viral reduction.

In this study, we aimed to investigate the occurrence and reduction of FRNAPH genotypes in an attempt to identify NoV indicators in wastewater. All FRNAPH and NoV genotypes were detected in WWTP influents at high rates (71–100%). Importantly, the concentration and detection rate of infectious FRNAPH GIV in influent samples were at the same level as those of the human-specific FRNAPH genotypes (GII and GIII) in WWTPs where the livestock waste is mixed into WWTP influent with seasonality. Furthermore, infectious FRNAPH GIV was the most efficiently reduced genotype in the WWTP among the infectious FRNAPH genotypes because of inactivation during wastewater treatment. These results demonstrated the seasonality of WWTP influents and the inactivation during wastewater treatment of infectious FRNAPH GIV. The log₁₀ reductions of FRNAPH GII and NoV GII during wastewater treatment showed a relationship (r = 0.69, P < 0.01) and no significant difference on the mean value (P > 0.05). These results suggest that FRNAPH GII could be used as an appropriate indicator of NoV GII during wastewater treatment. FRNAPH GI was also suggested as an appropriate viral indicator as it represents the worst-case scenario because of its highest resistance to wastewater treatment compared with the other FRNAPH and NoV genotypes. Further studies may evaluate the inactivation of NoVs during disinfection processes, such as chlorination and ultraviolet radiation, using the infectious FRNAPH genotypes determined by both IC–RT-PCR–MPN and RT-qPCR.

This work was supported by a KAKENHI grant (18K13863) from the Japan Society for the Promotion of Science (JSPS), Japan.