Technical framework for wastewater-based epidemiology of SARS-CoV-2.

Wastewater-based epidemiology (WBE) is expected to become a powerful tool to monitor the dissemination of SARS-CoV-2 at the community level, which has attracted the attention of scholars all over the world. However, there is not yet a standard protocol to guide its implementation. In this paper, we proposed a comprehensive technical and theoretical framework of relative quantification via qPCR for determining the virus abundance in wastewater and estimating the infection ratio in corresponding communities, which is expected to achieve horizontal and vertical comparability of the data using a human-specific biomarker as the internal reference. Critical factors affecting the virus detectability and the estimation of infection ratio include virus concentration methods, lag-period, per capita virus shedding amount, sewage generation rate, temperature-related decay kinetics of virus/biomarker in wastewater, and hydraulic retention time (HRT), etc. Theoretical simulation shows that the main factors affecting the detectability of virus in sewage are per capita virus shedding amount and sewage generation rate. While the decay of SARS-CoV-2 RNA in sewage is a relatively slow process, which may have limited impact on its detection. Under the ideal condition of high per capita virus shedding amount and low sewage generation rate, it is expected to detect a single infected person within 400,000 people.

Introduction

The ongoing global pandemic of coronavirus disease 2019 (COVID-19) is caused by a novel severe acute respiratory syndrome coronavirus (SARS-CoV-2) (Zhou et al., 2020). It has been a Public Health Emergency of International Concern, which has caused confirmed infection of 146,807,900 and death of 3,105,958 (Johns Hopkins Coronavirus Resource Center, accessed 26th Apr. 2021, https://coronavirus.jhu.edu/map.html). In recent studies, the RNA of SARS-CoV-2 has been detected in the stool of symptomatic, asymptomatic, pre-symptomatic and even post-symptomatic infected individuals (Bogler et al., 2020; Cai et al., 2020; Gao et al., 2020; Holshue et al., 2020; A. Tang et al., 2020; Wolfel et al., 2020; Y. Wu et al., 2020; J. Zhang et al., 2020; N. Zhang et al., 2020; W. Zhang et al., 2020). The fecal SARS-CoV-2 RNA detection rate ranged from 15 to 83% among infected patients. The load of SARS-CoV-2 in feces is in the range 5 × 103–107.6 RNA copies/mL (Foladori et al., 2020). Meanwhile, researchers all over the world have also detected the RNA of SARS-CoV-2 from the local sewage (Ahmed et al., 2020a; Haramoto et al., 2020; Kumar et al., 2020; La Rosa et al., 2020; Medema et al., 2020; Peccia et al., 2020; Randazzo et al., 2020; Rimoldi et al., 2020; Sherchan et al., 2020; Westhaus et al., 2020; D. Zhang et al., 2020) (see Supplementary Table S1 for more details). Thus, wastewater-based epidemiology (WBE) of SARS-CoV-2 which surveys sewage for virus RNA to inform epidemiological monitoring of COVID-19 has been proposed as the most promising method to determine the burden of undiagnosed infections at the population level (Bivins et al., 2020; Daughton, 2020; Hart and Halden, 2020; Kitajima et al., 2020). WBE is defined as the normalization of analyte influent concentration to per capita mass loads using the daily flow and wastewater treatment plant (WWTP) population, which provides population-scale information on human activity within catchment boundaries (Choi et al., 2018). It has been theorized for two decades (Daughton, 2001) and was initially implemented for estimating the community drug abuse (Zuccato et al., 2008). Thereafter its application field expands to community-level assessments on per capita consumption, use, exposure, or release of various chemical or biological agents (Supplementary Fig. S1) (Choi et al., 2018).

At present, reverse transcription quantitative polymerase chain reaction (RT-qPCR) is commonly used to detect the RNA of SARS-CoV-2 in sewage. Most of them adopt absolute quantification, and virus gene standard (normally a plasmid with SARS-CoV-2 genes) is needed to construct standard curve (Medema et al., 2020). Absolute quantification gives the real concentration of virus RNA in sewage, which can be correlated with local clinical observations (D'Aoust et al., 2021a; Medema et al., 2020; Nemudryi et al., 2020; Westhaus et al., 2020). While other studies only reported the value of cycle threshold (C T) (Kumar et al., 2020) or even just gave qualitative results (La Rosa et al., 2020). Such data provide limited information for monitoring actual infection ratio in the community. Various factors affect the accuracy of absolute quantitative estimation of community infection ratio or hinder the lateral comparison between different regions. For example, in the combined sewage system, the diluting effect of rainfall will affect the concentration of virion in sewage. In addition, the sewage generation rate is in the range of 50–500 L/person/day, which fluctuates with seasons and affects the concentration of virus in the sewage. Furthermore, in order to estimate the infection ratio in a community, it is necessary to accurately measure the community population, i.e. the population normalization for the data (Polo et al., 2020). This is a challenging task for areas with high population mobility. Several studies have tried to quantify some population-related biomarkers, such as human ribonuclease P (Peccia et al., 2020), pepper mild mottle virus (PMMoV) (D'Aoust et al., 2021a; D'Aoust et al., 2021b; Graham et al., 2021; Jafferali et al., 2021; F. Wu et al., 2020), creatinine, urea, benzotriazole, caffeine etc. (Rimoldi et al., 2020; Westhaus et al., 2020) to achieve this goal. Surprisingly, so far, no study has directly adopted relative quantification strategy with a human-specific biomarker as internal reference.

In this paper, we try to provide a better alternative for monitoring the SARS-CoV-2 in sewage, adopting the relative quantification with qPCR. Based on this concept, a self-contained technical framework is to be proposed by incorporating a human-specific fecal biomarker (human-specific genes or genes of human gut-specific bacteria and bacteriophage) as internal reference during the quantification. The nucleic acid property of the chosen biomarker confers themselves the possibility of being quantifiable with the same processes and platforms as the viral nucleic acid. The promising biomarkers that can be used as internal reference of qPCR in the are recommended. The critical stages and key points involved in the WBE of SARS-CoV-2 are evaluated. The decay kinetics of virus/biomarker in sewage, which is one of important factors affecting the detection sensitivity, is studied in a mimic mode in virtue of qPCR.

Theory and methodology

Relative quantification via qPCR

Two popular operation modes of qPCR are available for the quantification of genetic RNA or DNA: SYBR green and TaqMan. In this study, we prefer using the TaqMan mode due to its improving specificity. During qPCR amplification, the fluorescence intensity at the Nth amplification cycle (∆R ) can be expressed as:where ϕ is the luminescent intensity of unit fluorescence molecule; c 0 is the initial copy number of the gene to be amplified; ξ is the apparent amplification coefficient, which is 2 under ideal conditions.

By selecting the characteristic genes of biomarkers related to population (such as characteristic genes of human feces-associated bacteria or bacteriophage (Stachler et al., 2017)) as reference genes, the relative quantification of the target virus genes can be realized. The same threshold of fluorescence intensity (∆R ) is set for both reference and target genes.

In the case of using the same fluorescent reporter molecule (i.e. ϕ  = ϕ ), the relative abundance of the target viral gene (R ) can be expressed as:where C T is the cycle threshold when the fluorescence of the PCR product is detected crossing the threshold.; the subscript s indicates the reference gene; and the subscript t refers to the target virus gene.

Estimation of infection ratio

For the population in a certain sewage catchment, the relative abundance of virus gene in the sewage is proportional to the infection ratio in the catchment. The concentration of virus/biomarker in sewage is not only closely related to their per capita shedding amounts, but also affected by their decay kinetics in sewage. In addition, their recovery rate during the concentration of virus should also be considered. Then, the detected relative abundance of virus in the sewage samples (R ) can be correlated with the infection ratio by the following equation.where the infection ratio , is the ratio of the number of infected persons (P ) to the total population (P ) in the sewage catchment; S is the per capita shedding amount of virus/biomarker (copies/person/day); η is the residual fraction of virus/biomarker after decay in sewage, which is related to temperature and hydraulic retention time (HRT); and r is the recovery rate of virus/biomarker. Similarly, the subscript s indicates the reference gene (biomarker); and the subscript t refers to the target virus gene. Then the relative abundance of target virus gene in the feces sample of infected person (R ) can be expressed as:

Then, the infection ratio can be calculated with the following equation.

In silico performance evaluation of primer-probe sets

SARS-CoV-2 is an enveloped positive-sense single-stranded RNA virus, showing a higher propensity for mutation (Rambaut et al., 2020; X. Tang et al., 2020). The inclusivity of the primer-probe sets currently used in RT-qPCR needs to be evaluated. Over 23,000 SARS-CoV-2 genome sequences were downloaded from GISAID's EpiCoV™ database in late December 2020. There have been more than 180,000 sequences in the database then. To reduce the data size, we chose sequences uploaded in three time periods to download. They are sequences uploaded from Oct. 24th 2019 to Apr. 20th 2020 (8,156 sequences), from Aug. 10th 2020 to Aug. 25th 2020 (7,594 sequences) and from Dec. 3th 2020 to Dec. 9th 2020 (7,967 sequences), respectively. After filtering out the sequences with poor quality, the remaining sequences were used for in silico analysis of present primer-probe sets. The genome sequence of bat coronavirus RaTG13, which has an identity of 96.2% with the genome of SARS-CoV-2, was also included in the assay (Zhou et al., 2020). All the downloaded genome sequences were clustered with CD-HIT-EST using a sequence identity cut-off of 1.0 to group genomes with identical sequence into one cluster (Huang et al., 2010). Representative genome sequences (n = 12,769) of all clusters were aligned via the multiple sequence alignment program MAFFT (Katoh et al., 2019). The mutation rate of each site in the target binding region of every primer/probe was recorded, respectively. The primer-probe sets with less mutation rate in the targeting region are considered to be more inclusive. The exclusivity of primer-probe sets was also determined by assessing their ability to distinguish SARS-CoV-2 from RaTG13. The primer pairs with high inclusivity and high exclusivity were further evaluated via Primer-BLAST at the website of national center for biotechnology information (NCBI).

Virus detectability in sewage samples

The concentration of the biomarker (population-related bacteria or bacteriophages) corresponding to the reference gene is relatively high in sewage. Therefore, it is expected that there will not be a situation where the reference genes cannot be detected. The detectability of the relative quantification is the same as that of the absolute one, which is determined by the concentration of virus RNA in the sewage.

In this work, the sensitivity of the whole method is expressed in terms of the scale of a Population containing a Single Infected Person (PSIP), among which the SARS-CoV-2 RNA shedding off the infected person is detectable in the sewage samples collected from the catchment of the same population. The higher the value is, the more sensitive the detection is. The SARS-CoV-2 loading to sewage was estimated using recently reported its excretion specific load (E r) in human stool (copies/g feces), and assuming a fecal load of M f g feces/day/person. Then, the PSIP can be expressed as:where LOD means the limit of detection, which is set to be 1 copy/mL sewage for the RT-qPCR method according to the reported even lower LOD values (Ahmed et al., 2020a; Haramoto et al., 2020); Vs is the sewage generation rate (L/person/day).

Decay kinetics of virus/biomarker in sewage

It worth noting that the decay kinetics of virus/biomarker in wastewater is an important factor to be considered whether it is relative or absolute quantification. The virus/biomarker will decay exponentially with time in sewage of given composition (Ahmed et al., 2020b; Hart and Halden, 2020).where n 0 is the initial amount of virus/biomarker discharged into the sewage system; n is the amount after a hydraulic retention time (HRT) of t hours in the sewage system before being sampled; k is the first order decay rate constant; and t 1/2 the half-life (i.e. T 50 the time for half decay).

The decay residual fraction η can be expressed as:

According to Eq. (8) where T 90 is the time required for 1 log10 decay, and T 99 the time for 2 log10 decay.

The decay of virus/biomarker in sewage is closely related to the temperature, which is expressed with the Arrhenius equation.where E is the activation energy of the decay reaction (J/mol); R is the ideal gas constant (8.314 J/mol/K); T the absolute temperature (K); and C the constant related to standard state setting. Therefore, the decay rate constant and half-life at different sewage temperatures can be calculated using Eqs. (14), (15), respectively. where k 1 is the decay rate constant at temperature T 1; k 2 the decay rate constant at temperature T 2; t 1/2,1 is half-time at temperature T 1; t 1/2,2 is half-time at temperature T 2.

The decay kinetics of the virus in the sewage also affects the detection sensitivity. According to the linear correlation between lnk and 1/T [Eq. (13)], the decay activation energy (E ) of SARS-CoV-2 RNA in sewage can be deduced using the reported decay rate constant at different temperatures (Ahmed et al., 2020b). The decay rate constant at each temperature is calculated by introducing E into Eq. (14). The impact of temperature and HRT on detection sensitivity is evaluated.

Results and discussion

Critical stages and key points

The critical stages and key points involved in the wastewater-base epidemiology of SARS-CoV-2 are summarized in Supplementary Table S2. Many factors need to be considered in both relative and absolute quantification, while relative quantification exhibits more advantages over absolute quantification. The ultimate purpose of determining the virus content in sewage is to estimate the infection ratio at population level. The results from absolute quantification are sensitive to the fluctuation of sewage generation rate (Vs) and the dilution effect of rainfall events. As the reference biomarker and the target virus will be diluted to the same extent when sewage generation rate increases or the rainfall event occurs, the relative quantification will not be affected too much. In our previous work, qPCR was used to quantify the prevalence of enteropathogenic E. coli (EPEC) in sewage successfully. The fluctuation of absolute quantification data spanned two orders of magnitude over a period of one year, while relative quantification data changed by no more than an order of magnitude (Yang et al., 2014). The relative quantitative data more accurately reflected the changes of EPEC in the sewage. In the relative quantification process, it is not necessary to accurately record the volume of samples used for virus recovery and nucleic acid extraction, nor to accurately determine the volume of the final nucleic acid extract. In addition, there is no need of the target gene standard to construct the standard curve. Even so, in order to ensure the reproducibility of the results and the standardization of the methods, essential information such as sample processing procedure, qPCR protocol and data analysis method etc. should be provided when reporting, as recommended in previously published guidance (Bustin et al., 2009).

One of the major challenges in the practice of WBE on the community monitoring of COVID-19 is the establishment of standardized methods and procedures. Among them, the recovery of virus from sewage samples is the first and the most important step. Supplementary Table S1 summarizes present studies reporting the detection of SARS-CoV-2 in wastewater. As shown in the table, at present, the main virus concentration methods for SARS-CoV-2 from sewage samples are PEG precipitation, electronegative membrane adsorption and ultrafiltration etc. These methods of concentrating nonenveloped virus from sewage are more targeted at the recovery of virus particles from the sewage supernatant which has been pretreated to remove suspended solids and bacteria. However, as an enveloped virus, SARS-CoV-2 is different from nonenveloped virus in nature, which will affect its partition behavior in sewage. Therefore, the concentration method should be adjusted accordingly (Ye et al., 2016). According to our knowledge and reported studies, the virus adsorbed on the sewage particles cannot be neglected and may even account for the major portion of the SARS-CoV-2 in the sewage (Balboa et al., 2020; D'Aoust et al., 2021b; Graham et al., 2021; Petala et al., 2021; Westhaus et al., 2020). By far the highest value of virus content ever detected in sewage-related samples is from the primary sewage sludge, and the value is as high as 4.6 × 108 RNA copies/L (Peccia et al., 2020). However, we do not recommend directly using the virus RNA abundance in the primary sewage sludge to estimate the community infection ratio. The HRT of sludge is longer than that of wastewater, resulting in the accumulation of virus in sludge. If one wants to use the abundance of sludge virus RNA to estimate the community infection ratio, the retention time of sludge or the residence time distribution need to be considered. Accordingly, we suggest that virus RNA of both solid and liquid phase of sewage samples should be considered in wastewater surveillance. Simultaneous recovery of virions from both liquid and solid phases of sewage also gave higher detection sensitivity (higher PSIP values) as shown in Supplementary Table S1 (La Rosa et al., 2020; Wurtzer et al., 2020). But some studies have pointed out that retaining more sewage solid likely introduces more matrix inhibition (Gonzalez et al., 2020). So, it is important to minimize matrix inhibition by optimizing the process of viral nucleic acid extraction and adding inhibitor neutralizing reagent such as bovine serum albumin (BSA) solution in the PCR reaction system. Effective recovery of virus or viral nucleic acid from sewage sample will largely prompt the detectability of the SARS-CoV-2 RNA.

According to Eq. (6), the infection ratio of communities in the sewage catchment can be calculated from the relative abundance of the target virus gene. The premise is to determine the average relative abundance of the target virus gene in the feces of the infected person. It has been reported that not all infected people are positive for RNA of SARS-CoV-2 in their feces, virus shedding amount in the feces varies from person to person, which also fluctuates throughout the infection period (Wolfel et al., 2020; N. Zhang et al., 2020). In order to incorporate these factors into consideration, the range of infection ratio is ought to be estimated via such as Monte Carlo model (Ahmed et al., 2020a; Hasan et al., 2021). In addition, the decay kinetics and recovery efficiency of virus and reference biomarker from sewage are also factors that need to be considered, which will be elaborated in the following sections. Under the relatively loose experimental operation standard, the relative abundance data obtained by different laboratories can be compared horizontally. In addition, the relative abundance data obtained from sewage catchments of different population sizes can be directly used to compare the severity of epidemics among communities. The results can be used to guide regional policy adjustment for either restart of economies or continuous lockdown. For a specific community within a sewage catchment, the parameters (η , η , r and r ) in Eq. (6) can be considered as constants when using fixed sampling mode, virus concentration method, nucleic acid extraction method and standardized qPCR relative quantification operation. The equation can be simplified towhere according to Eq. (6). If accurate clinical diagnosis data are available, the constant K can be accurately estimated by correlating the actual infection ratios with the relative abundances of virus RNA in sewage. With sufficient basic data, the method of WBE can even be combined with machine learning to further improve the prediction accuracy of community infection ratio.

Besides the above key points which may affect the performance of the quantification, biosafety related factors also need to be considered. Researchers in Italy and Germany have proved that the infectivity of SARS-CoV-2 in sewage is limited (Rimoldi et al., 2020; Westhaus et al., 2020). However, the possibility of fecal-oral transmission cannot be completely ruled out. Furthermore, it seems that SARS-CoV-2 can be effectively removed in the sewage treatment process. Researchers from China, Spain, India, and Italy have detected the RNA of SARS-CoV-2 in the influent of the sewage treatment plant, but not in the effluent (Kumar et al., 2020; Randazzo et al., 2020; Rimoldi et al., 2020; Wang et al., 2020). That is to say, the location of sampling may affect the detectability of the virus. Personal protection (wearing N95 mask, gloves, and goggles etc.) during sampling and pasteurization of samples (60 °C, 30–90 min) before virus concentration are essential for biosafety reasons.

Performance of primer-probe sets

A total of 18 currently used primer-probe sets were evaluated in this work (Supplementary Table S3). Their locations in the SARS-CoV-2 genome are illustrated in Supplementary Fig. S2. The mutation rate of SARS-CoV-2 genome at the probe/primer targeting regions is shown as a heatmap in Fig. 1A . The original data are available as online supplementary datasets (Supplementary Dataset S1 and S2). Four primer-probe sets targeting different open read frames in the virus genome with high inclusivity (low mutation rate at their targeting regions, marked in green in Fig. 1A) and high exclusivity (more mismatches with the sequence of RaTG13, Fig. 1B) were recommended (Table 1 ). Sequences of some primers were amended according to the sequence alignment and/or Primer-BLAST analysis, and the primers were renamed accordingly. Although we have carefully evaluated these primer-probe sets, it is still found that some mutant virus sequences in the database cannot perfectly match the primer/probe sequences. Compared with the virus genome sequences uploaded before the end of April 2020 (Supplementary Fig. S3), the number and mutation rate of point mutations have increased significantly, which implies that the rapid and successive mutation in virus genome has affected the accuracy of the nucleic acid detection. Especially, those point mutations near 3′ ends of primers may cause false negative. For example, the mutation in SARS-CoV-2 lineage B.1.1.7, a variant first detected in the UK in September 2020, prevents PCR amplification of the spike gene target (Challen et al., 2021; Davies et al., 2021). Therefore, we suggest that two to three primer-probe sets should be used simultaneously in practice to ensure the accuracy of detection. By adopting proper primer pairs, PCR products can even be used for virus typing and epidemiological tracking.

Fig. 1

Heatmap indicating the mutation rate of SARS-CoV-2 genome at the probe/primer targeting regions (A) and the specificity (exclusivity) evaluation of recommended primer/probe sets by matching their sequences with the genome of the bat coronavirus RaTG13 (B). The color of each square in the heatmap (A) represents the number of sequences (among the total 12,769 sequences) that mutate at the site, which is displayed in an exponential gradient. The sequence of each primer/probe is exhibited accordingly. Magenta texts indicate mismatched sites of the primer/probe to the PCR template and the expected matched base is listed in the bracket. The primers/probes indicated with asterisk in both panels (A) and (B) are exhibited the reverse complementary sequences. Most of them are reverse primers. Those primer/probe sets recommended in this work are marked in green in panel (A), among which the sequences of Charité RdRP-SARSr primer/probe set have been amended (indicated with red texts in panel (B)). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Table 1

Recommended primer/probe sets with high specificity and inclusivity according to the sequence alignment analysis.

Institutes	Target	Name	Sequence (5′-3′)	Tm (°C)	Product size (bp)	Reference
France Pasteur	RdRP-IP4	RdRp-IP4-KY_Fa	GGT ARC TGG TAT GAT TTC G	58	107	(WHO, 2020)
		RdRp-IP4_P	FAM–TCA TAC AAA CCA CGC CAG G–BHQ1
		RdRp-IP4_R	CTG GTC AAG GTT AAT ATA GG
Germany Charité	RdRP	RdRp_SARSr-KY-Fa	GTG AAA TGG TCA TGT GTG GCG G	58	100	(Corman et al., 2020)
		RdRp_SARSr-P2	FAM–CAG GTG GAA CCT CAT CAG GAG ATG C–BHQ1
		RdRp_SARSr-KY-Ra	CAAATG TTA AAA ACA CTA TTA GCA TA
HKU SKL of EID	S	Chan-S_F	CCT ACT AAA TTA AAT GAT CTC TGC TTT ACT	55	158	(Chan et al., 2020)
		Chan-S_P	FAM–CGC TCC AGG GCA AAC TGG AAA G–BHQ1
		Chan-S-KY_Ra	CAA GCT ATA ACR CAG CCT GTA
US CDC	N	2019-nCoV_N2-F	TTA CAA ACA TTG GCC GCA AA	55	67	(CDC, 2020)
		2019-nCoV_N2-P	FAM–ACA ATT TGC CCC CAG CGC TTC AG–BHQ1
		2019-nCoV_N2-R	GCG CGA CAT TCC GAA GAA

The sequences of the primers marked in bold have been amended at the underlined sites according to the sequence alignment and/or Primer-BLAST analysis, and the primers have been renamed accordingly.

Reference biomarkers

To realize the relative quantification of SARS-CoV-2, a population-related reference biomarker needs to be chosen. There are numerous assays available for the quantitative assessment of human fecal pollution (Harwood et al., 2014), which can meet this demand. Those qPCR-based microbial source tracking (MST) methods targeting mitochondrial DNA, rRNA and functional genes involved in microorganism-human interaction are favorable candidates. In Table 2 , we recommend some promising qPCR assays that have been proved to be effective for the detection of human-associated MST markers. The genes from the three main sources can be used as reference genes for the relative quantification: genes from human hosts (mitochondria) (Schill and Mathes, 2008), genes from human-associated bacteria (Bacteroidales) (Shanks et al., 2010) and bacteriophages (crAssphage) (Dutilh et al., 2014; Stachler et al., 2017). We also list the most commonly used PMMoV assay in recent WBE of SARS-CoV-2 therein for reference. Since it targets the plant virus, which may be not human-specific. Therefore, it should be applied with caution. As shown in Table 2, the MST method targeting the reference gene should be of high enough sensitivity (true-positive rate) and specificity (true-negative rate). Furthermore, the reference gene must also be of a high enough content in feces or waste, so as to ensure that it can still be detected in the sewage even at a high dilution level.

Table 2

Human-associate biomarkers in MST assays as promising candidates of the references for the relative quantification of SARS-CoV-2.

Organism	Assays	Primers and Probes	Tm (oC)	Product Size (bp)	Sensitivity (n)	Specificity (n)	References
Bacteroidales	HF183	HF183-1: ATCATGAGTTCACATGTCCGBFDRev: CGTAGGAGTTTGGACCGTGTBFDFAM: 6-FAM-CTGAGAGGAAGGTCCCCCACATTGGA-TAMRA	60	167	100% (16)	90.9% (174)	(Haugland et al., 2010; Shanks et al., 2010)
	HumM2	Hum2F: CGTCAGGTTTGTTTCGGTATTGHum2R: TCATCACGTAACTTATTTATATGCATTAGCHumP: FAM-TATCGAAAATCTCACGGATTAACTCTTGTGTACGC-TAMRA	60	101	100% (16 feces, 20 SEW)	99.2% (285)	(Shanks et al., 2009)
	HumM3	Hum3F: GTAATTCGCGTTCTTCCTCACATHum3R: GGAGGAAACAAGTATGAAGATAGAAGAATTAAHum3P: FAM-AGGTCTGTCCTTCGAAATAGCGGT-TAMRA	60	83	100% (16 feces, 20 SEW)	97.2% (285)	(Shanks et al., 2009)
	BacH	BacH-f: CTTGGCCAGCCTTCTGAAAGBacH-r: CCCCATCGTCTACCGAAAATACBacH-pC: FAM-TCATGATCCCATCCTG-NFQ-MGBBacH-pT: FAM-TCATGATGCCATCTTG-NFQ-MGB	61	93	95% feces (21), 100% wastewater (20)	99% (302)	(Reischer et al., 2007)
	BtH (B. thetaiotomicron a-mannanase)	BtH-F: CATCGTTCGTCAGCAGTAACABtH-R: CCAAGAAAAAGGGACAGTGGBtH-P: FAM-ACCTGCTG-NFQ	60	63	100% (10 feces, 20 SEW)	100% (160)	(Yampara-Iquise et al., 2008)
crAssphage	CPQ_056	056F1: CAGAAGTACAAACTCCTAAAAAACGTAGAG056R1: GATGACCAATAAACAAGCCATTAGC056P1: FAM-AATAACGATTTACGTGATGTAAC-MGB	60	126	100% (9 SEW)	98.6% (222)	(Stachler et al., 2017)
	CPQ_064	064F1: TGTATAGATGCTGCTGCAACTGTACTC064R1: CGTTGTTTTCATCTTTATCTTGTCCAT064P1: FAM-CTGAAATTGTTCATAAGCAA-MGB	60	148	100% (9 SEW)	98.6% (222)	(Stachler et al., 2017)
Mitochondria	mtCytb (Cytochrome b)	Hum_mtCytb-F: AGTCCCACCCTCACACGATTCTTTHum_mtCytb-R: AGTAAGCCGAGGGCGTCTTTGATTHum_mtCytb-P: FAM-ACCCTTCATTATTGCAGCCCTAGCAGCACT-TAMRA	60	185	100% (2 feces, 10 SEW)	100% (18)	(Schill and Mathes, 2008)
Plant virus	PMMoV	PMMoV-F: GAGTGGTTTGACCTTAACGTTImage 1GA⁎PMMoV-R: TTGTCGGTTGCAATGCAAGTPMMoV-P: FAM-CCTACCGAAGCAAATG-TAMRA	60	68	66.7% (12/18 feces)	ND	(Haramoto et al., 2013; Zhang et al., 2006)

The original reference (Zhang et al., 2006) for PMMoV detection missed a “T” in the forward primer which is indicated in red letter.

The validity of reference biomarkers (reference genes) can be verified by comparing the abundance of different reference biomarkers. The premise of this self-consistent mode of the validation is that all reference biomarkers are all population related. Dilution will not affect the relative abundance between different reference biomarkers. Therefore, the relative abundance will not fluctuate dramatically over time. The result can also support the feasibility of relative quantification of SARS-CoV-2. A recent attempt has provided us with a good paradigm to evaluate the robustness of the selected reference biomarkers (Ahmed et al., 2020c). In a steady sewage system, the biomarker with less abundance fluctuation among different dates should be more suitable to be used as the internal reference, which indicates that crAssphage is more suitable than PMMoV. In addition, reasonably selected reference gene itself is a good internal control of the whole method, which can exclude the influence of concentration efficiency, inhibiting substance of PCR and other factors to a certain extent (Jafferali et al., 2021). A recent study has shown that not only the DNA of crAssphage, but also its transcription product RNA can be detected in sewage (Wilder et al., 2021). Therefore, crAssphage has the potential as a dual reference biomarker for simultaneously detecting RNA and DNA viruses.

Factors affecting the detectability

First, according to Eq. (7), per capita virus shedding amount (S  = E  ∙ M ) and sewage generation rate (V ) will directly affect the virus concentration in the sewage, thus affecting the detectability. The virus shedding varies from person to person. Even for the same infected person, the virus emission also changes during the whole infection period. According to previous reports, the excretion specific load of SARS-CoV-2 (E ) is in the range of 103–108 RNA copies/g feces (Foladori et al., 2020; Kitajima et al., 2020; Wolfel et al., 2020). We assume the fecal load (M ) in the range of 100–400 g feces/person/day and the sewage generation rate (V ) in the range of 100–400 L/person/day (Hart and Halden, 2020). Without considering the decay of virus in sewage (i.e. η  = 1), the detection sensitivity (PSIP) is calculated with Eq. (7) as high as 4 × 105 (Fig. 2A ), which means that one single infected person is detectable among a population of 4×105 at ideal conditions with the highest load of virus in feces (108 RNA copies/g), the highest fecal load (400 g feces/day/person) and the lowest sewage generation rate (100 L/person/day). With the decrease of per capita virus shedding amount and the increase of sewage generation rate, the detection sensitivity gradually decreased to PSIP = 0.25, which means that in the case of extremely low virus shedding amount (1 × 105 RNA copies/day/person) and high sewage generation rate (400 L/person/day), the detection of sewage from one single infected person may also be negative.

Fig. 2

Factors affecting the detectability of SARS-CoV-2 in sewage. The detection sensitivity (PSIP) is affected not only by per capita virus shedding amount S and sewage generation rate Vs (A), but also by the factors (temperature and HRT) affecting the virus decay kinetics in sewage (B). The decay rate constant (k) at different temperatures (T) are calculated according to the reported experimental data (C). The calculated values of T50 and T90 are compared with experimental ones (D). The experimental data of the decay kinetics of SARS-CoV-2 RNA are available in reference (Ahmed et al., 2020b).

On the other hand, the detectability is also affected by the decay kinetics of virus and the HRT. The decay of virus may be due to biological and chemical activities in sewage (Mohamed Hamouda et al., 2021). Pasteurization can delay the decay kinetics of virus in sewage via eliminating bacterial extracellular enzyme activity and protozoan or metazoan predation (Ye et al., 2016). Recently, Ahmed et al. reported the decay kinetics of SARS-CoV-2 RNA in pasteurized and unpasteurized sewage (Ahmed et al., 2020b). We extracted the decay kinetics data of SARS-CoV-2 RNA in unpasteurized sewage from Table 3 in their paper. According to Eq. (11), plotting lnk versus 1/T gave a straight line with a slope of −E /R. Thus, the value of E was determined to be 27,110.3 (J/mol). As shown in Fig. 2C and D, the values of k, t 1/2 and T 90 in the temperature range of 0 to 40 °C were also calculated with Eqs. (14), (15), (11), respectively. With the change of HRT and sewage temperature, the decay residual fraction of virus η can be calculated with Eq. (9). If the virus shedding amount of 400 × 108 copies/person/day and sewage generation rate of 100 L/person/day are set, the detection sensitivity (PSIP) varies between 2.94 × 105 and 4 × 105 with changing η (calculated with Eq. (7) and shown in Fig. 2B). The calculated values of T 90 at different temperatures are in good agreement with the experimental ones (T 90 in Fig. 2D). The HRT (~24 h) of the virus in sewage is relatively short in comparison to its half-life (2.25–10.36 days). Because this is the only case of kinetics study for the decay of SARS-CoV-2 RNA in wastewater, we can only conservatively say that the decay of the virus may have limited impact on its detectability (Fig. 2B). Compared with this recent report, previous studies on the decay kinetics of coronavirus in sewage have given a much shorter half-life (1.00–1.35 days) (Casanova et al., 2009; Gundy et al., 2009; Rabenau et al., 2005). It is worth noting that the previous studies mostly investigated the infectivity of the virus (i.e., viral survival) and this report investigated the nucleic acid substance of the virus, which is fundamentally different. As an important part of the technical system of WBE, the decay kinetics of pathogenic microorganisms (or their RNA/DNA) in sewage should be systematically evaluated. If necessary, the main chemical and biological factors affecting their persistence in the sewage should be determined.

The above simulation results show that the most important factor affecting the detectability of SARS-CoV-2 RNA in sewage is the virus shedding amount off the infected person (S ). In this work, we only discussed the virus discharged into sewage through the feces of infected persons. But in fact, other virus-containing secretions of patients may also be discharged into the sewage system, such as sputum. The virus content in sputum is as high as 2.35 × 109 RNA copies/mL (Wolfel et al., 2020). Therefore, the shedding amount of virus in the sewage from infected persons should be higher than what we have estimated. The previous simulation study has shown that the detection of one infected case among a population of 2,000,000 is theoretically feasible (Hart and Halden, 2020). However, according to our estimation, this population should be in the order of 100,000 among which the detectability of single infected person is more promising. The experimental study also has proved that sewage sample started to give positive RT-qPCR signal of virus genes when the observed COVID-19 prevalence was around or even below one case in 100,000 people (Medema et al., 2020). According to the calculations, under optimal conditions, one infected person can be detected in a population of 400,000 by the means of WBE. For Chengdu, a city with 16 million residents, a reasonable layout of 40 to 160 sampling points will be sufficient to realize the epidemic surveillance of the whole city. Such a small-scale of detection (40–160 samples) can even be routinely completed in our laboratory every day to implement the real-time monitoring of the epidemic situation for the entire city. It is conservatively estimated that the whole process from sample pretreatment, virus concentration, nucleic acid extraction to the final RT-qPCR will take about 8–10 h.

Mimic study on the decay kinetics of virus/biomarker

The decay kinetics of virus/biomarker (target or reference genes) in sewage can be studied via microcosm experiments in the laboratory (Ahmed et al., 2020b). For studying the decay kinetics of biomarkers (bacteria or bacteriophages), freshly collected sewage samples can be incubated at certain temperature and be tracked the change of biomarker content over time. For the virus, certain number of virus (inactivated SARS-CoV-2 virions) will be spiked in the sewage before incubation. QPCR will be still used to track the changes of virus/biomarker in the sewage. where c is the concentration of virus/biomarker i in the sewage at time t (RNA or DNA copies/mL sewage), c is the initial concentration of virus/biomarker i (RNA or DNA copies/mL sewage), C the threshold cycle number for the initial sewage sample, C the threshold cycle number for the sewage sample at time t. Plotting C versus t will give a straight line with a slope of 1/t 1/2 (Eq. (17)). Thus, one can obtain the half-life of virus/biomarker at different temperature. Then plotting lnk versus 1/T will also give a straight line with a slope of −E /R (Eq. (12)). Introducing E into Eq. (15) can calculate the half-life (t 1/2) at any temperature.

Considering the pathogenicity of SARS-CoV-2, we can alternatively use the enveloped virus surrogates to study the decay kinetics in sewage and mimic its recovery from sewage samples. Typical surrogates are murine hepatitis virus (MHV) and Pseudomonas phage ϕ6 (Ye et al., 2016). According to the detected temperature of sewage samples, introducing corresponding t 1/2 and the estimated HRT (t) into Eqs. (9), (6) will give a reasonable estimation of the real infection ratio in the sewage catchment with the experimentally determined relative abundance of SARS-CoV-2. In addition, if the bacteriophage is chosen as the reference biomarker, its decay behavior in sewage may be similar with SARS-CoV-2, due to their similar properties (McMinn et al., 2017).

Summary of the technical framework

The technical framework of WBE for SARS-CoV-2 is illustrated in Fig. 3 . Five main steps are involved: sampling, virus concentration, RNA/DNA extraction, RT-qPCR/qPCR, and data/results reporting. According to the different research purposes and local epidemic situations, different sampling strategies may be adopted. In areas where the COVID-19 is still exploding, we can not only track the change of infection ratio in a community over time, but also compare the epidemic burden of different communities. For the former, the strategy of repeated sampling at the same sampling point on different dates (temporal) will be adopted, which can be used to track the changing trend of community epidemic. While for the latter, samples will be collected in different communities (spatial) on the same dates, which can realize the horizontal comparison of epidemic burden among different communities and provide regional guidance for keeping isolation or restarting the economy. If accurate community infection data are available, both strategies can eventually establish a quantitative relationship between the abundance of sewage virus RNA and community infection ratio. In a low prevalence area where the epidemic has been under control, reasonable layout of sampling points can realize real-time monitoring of the whole area and prevent the resurgence of the epidemic caused by imported asymptomatic infected persons or virus contaminated cold chain foods. Once a positive signal is detected at a certain sampling point, we can add secondary sampling points in the corresponding area or conduct regional full staff screening to track down the infected persons. Regardless of the sampling strategy, we recommend using 24-hour flow-weighted-composite samples for testing (Ahmed et al., 2020c). For the steps of virus concentration and RNA/DNA extraction, we recommend recovering virus from both solid and liquid phase and extracting both RNA and DNA from concentrated samples, because of the adsorption tendency of enveloped virus on solid particles and the DNA property of the reference gene. In the detection step, the RT-qPCR and qPCR will be applied for the quantification of multiple viral genes and the reference gene to give the relative abundance of the virus relative to the reference biomarker. When reporting the data, it is necessary to clarify some important matters that affect the interpretation of the results, such as the lag-period between WBE and clinic data (D'Aoust et al., 2021a; Peccia et al., 2020), per capita virus shedding amount, sewage generation rate, and the decay kinetics and HRT of virus/biomarker (virus RNA and reference gene) etc.

Fig. 3

Illustration of the technical framework of the WBE for SARS-CoV-2 based on relative quantification via qPCR. Different strategies can be adopted at the sampling step: temporal, spatial and combination of spatial and temporal. The 24-hour flow-weighted-composite samples are recommended for testing. It is recommended to recover virus from both solid and liquid phase and to extract both RNA and DNA from concentrated samples. When estimating the community infection ratio, the relative abundance of virus RNA (against the reference gene) in the sewage is adopted. When reporting the results, some important matters need to be announced, such as: lag-period between clinic and WBE data, per capita virus shedding amount, sewage generation rate, decay kinetics of virus/reference biomarker, HRT etc.

Conclusion

By incorporating a human-specific fecal biomarker (reference gene) as internal reference of qPCR, a technical framework of WBE based on relative quantification has been established for monitoring the dissemination of SARS-CoV-2.

Many factors that may affect virus detectability and estimation of community infection ratio need to be considered. Such as lag-period, per capita virus shedding amount, sewage generation rate, decay kinetics of virus/biomarker, temperature, and HRT, which should be announced at the reporting of the results.

Inclusivity of present primer-probe sets was evaluated. The primer-probe sets with good specificity and inclusivity are recommended.

Recovery and concentration methods for the enveloped virus SARS-CoV-2 need to be optimized. Virions in both the liquid and solid phases of sewage should be recovered simultaneously.

The technical framework established in this paper can not only be used to monitor the current COVID-19 pandemic, but also monitor other viruses in sewage, and carry out community epidemiological monitoring of corresponding diseases or provide early warning for the next human pandemic.

Nomenclature

apparent activation energy of the decay reaction

average relative abundance of the viral gene in the feces of infected persons

relative abundance of the target viral gene

threshold of fluorescence intensity

fluorescence intensity at the Nth amplification cycle

initial copy number of the target gene during qPCR amplification

the concentration of virus/biomarker i in the sewage

the constant in Eq. (13)

cycle threshold

excretion specific load of virus in human stool (copies/g feces)

community infection ratio

first order decay rate constant (h−1)

proportional constant between virus relative abundance and community infection ratio

limit of detection, in our case 1 copy/mL sewage for the RT-qPCR method

initial amount of virus/biomarker

amount of virus/biomarker after hydraulic retention time (HRT) of t

arbitrary cycle number

per capita daily excretion amount (g feces/person/day)

number of infected persons in the community within a sewage catchment

Population with Single Infected Person, the parameter for measuring detection sensitivity (detectability)

total population within a sewage catchment

recovery rate of virus/biomarker

the ideal gas constant (8.314 J/mol/K)

per capita shedding amount of virus/biomarker (copies/person/day)

hydraulic retention time (HRT) of virus/biomarker in sewage (h)

absolute temperature (K)

half-life of virus in the sewage (h)

time for 1 log decay (h)

time for 2 log decay (h)

sewage generation rate (L/person/day)

Greek letters

residual fraction of virus/biomarker after decay in sewage

apparent amplification coefficient, equaling 2 under ideal condition

luminescent intensity of unit fluorescence molecule

Subscripts

initial state

parameters in related to the reference gene

parameters in related to the target viral gene, indicating the retention time of virus/biomarker in sewage

indicating temperature T 1

indicating temperature T 2

CRediT authorship contribution statement

Jinyong Wu: Conceptualization, Methodology, Data curation, Visualization, Writing – original draft, Writing – review & editing. Zizheng Wang: Methodology, Data curation, Visualization, Writing – original draft, Writing – review & editing. Yufei Lin: Formal analysis, Writing – original draft, Writing – review & editing. Lihua Zhang: Formal analysis, Writing – original draft, Writing – review & editing. Jing Chen: Conceptualization, Writing – review & editing. Panyu Li: Writing – review & editing. Wenbin Liu: Writing – review & editing. Yabo Wang: Methodology, Writing – review & editing. Changhong Yao: Writing – review & editing. Kun Yang: Conceptualization, Funding acquisition, Methodology, Data curation, Visualization, Writing – original draft, Writing – review & editing.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.