Wastewater-based genomic surveillance of SARS-CoV-2 in vulnerable communities in Mumbai

Wastewater sample collection and RNA isolation

Wastewater samples were systematically collected twice weekly from eight open drain sites in Mumbai’s slums [Dharavi: (D1, D2, D3), Kherwadi: (K4, K5), Behrampada: (B6, B7), Siddharth Nagar: (S8)] between August 17, 2022 and June 19, 2023. This was done with permission from the Municipal Corporation of Greater Mumbai (MCGM) and following guidance from the National Institute of Virology’s (NIV) polio surveillance programme. The collection sites were selected based on major slum locations in consultation with MCGM. To expand the analysis citywide, two new collection sites at sewage treatment plants (STPs) – Bandra STP and Worli STP – were introduced in March 2023. The Institutional Research Ethics Committee granted a waiver from ethics approval, as the study did not directly involve human participants. Additionally, the Institutional Biosafety Committee approved handling the live virus.

Sampling was conducted using 500 ml polypropylene bottles, with 400 ml collected between 8 AM and 11 AM to ensure efficient recovery. After sealing, bottles were wiped with 70% ethanol, placed in ziplock bags, and stored in an ice-packed cooler. Samples were processed within 24 h at a Biosafety Level 2 facility to obtain near real-time information on viral concentrations in sewage. Following pasteurization at 60°C for 1 h, 45 ml of wastewater (in duplicates) was centrifuged at 5,000 x g for 10 min at 4°C. The 40 ml supernatant was filtered first through two Whatman Grade 1 filter papers and then through a 0.22 μm PES membrane filter. The filtrate solution was combined with 3.2 g of PEG (8,000 MW) and 0.68 g of NaCl (1.7%) in a sterile 50 ml falcon bottle and incubated for 16 h at 48°C until PEG completely dissolved (SOP provided by Science and Engineering Research Board-SERB, https://serb.gov.in/ ).

After PEG dissolution, the solution was centrifuged at 10,000 rpm for 30 min at 4°C. Following supernatant removal, 140 μl of elution (TE) buffer was added to resuspend the pellets; one set of resuspended pellets was stored at -80°C for future use. The solution was vortexed to dissolve the pellets, and viral RNA was isolated using the QiaAmp viral RNA mini kit (Qiagen GmBH, Hilden, Germany) following the manufacturer’s protocol.

SARS-CoV-2 RT-qPCR detection

Real-time quantitative PCR (RT-qPCR) was conducted using the Bio-Rad CFX96 system (Bio-Rad, California, USA) and GenePath Dx CoViDx One v2.1.1TK kit (GenePath Diagnostics), targeting SARS-CoV-2 specific genes (N, RdRp, E). The kit provided qualitative outcomes, and viral load quantification was done using the COVID-19 viral load Calculation tool (RUO). Each sample was tested in duplicate to reduce false negatives, with any replicate positive for at least two genes considered positive. Tap water served as the negative control.

SARS-CoV-2 genome amplification and sequencing

Positive samples with Ct values lower than 35-40 (before February 2023 <35, after was <40) were sequenced either by Illumina (n=38) or OxfordNanopore (n=85) sequencing. Pooled libraries were prepared using the Illumina COVIDSeq protocol for Illumina sequencing. cDNA was amplified using ARTIC primers in two pools, followed by sequencing on the NextSeq™ 2000 platform (Illumina, California) with 116 bp read length.

Nanopore sequencing used LunaScript RT SuperMix Kit (New England Biolabs) for cDNA synthesis and MIDNIGHT primers for amplification. Barcoding and adaptor ligation followed the rapid MIDNIGHT Protocol, and sequencing was performed on the MinION MK1B with the SpotON flow cell.

Early warning SARS-CoV-2 detection strategy

A correlation analysis between viral RNA concentrations in wastewater and positive cases in the city was done to generate an early warning for impending SARS-CoV-2 waves.

Bioinformatics pipeline and analyses

Wastewater samples comprised a mix of actively circulating (sub-)variants. The raw reads were aligned to the SARS-CoV-2 reference genome (NC_045512.2) using the Burrows-Wheeler Alignment tool (BWA) MEM¹⁷. Coverage statistics were obtained using SAM tools¹⁸ and primer trimming for Illumina sequenced samples was performed with iVar¹⁹. Aligned reads were used for single nucleotide variants (SNVs) and INDELs calling using BCFtools¹⁸.

For lineage abundance prediction, a deconvolution matrix was generated using Freyja²⁰– a bioinformatics pipeline for estimating lineage abundance from wastewater. Freyja utilizes SNV frequency and sequencing depth to estimate accurate lineage abundance within samples.

For the comparison of viral load across different seasons, the specific months were selected based on the weather patterns of the current research year. The monsoon periods were defined as August 17 to October 27, 2022, and June 19 to June 28, 2023. The winter period extended from November 2022 to February 2023, while the summer period was from March 14 to June 14, 2023. Statistical analyses and graph plotting were performed using GraphPad Prism version 9.1.2 ( https://www.graphpad.com/features ).

Sample collection overview and positivity rate

We analysed SARS-CoV-2 RNA variation in wastewater samples over 11 months (August 2022 to June 2023) to assess the pandemic’s status in vulnerable settings of Mumbai. Among the 728 samples processed (672 from open drains and 56 from STPs), 292 tested positive for at least two out of three targeted genes (E-gene, N-gene and RdRp). No SARS-CoV-2 genes were detected in negative control samples. Positive samples’ Ct values ranged between 30 and 40. The average positivity rate among all the samples collected during the study period was determined to be 36.5 per cent.

The calculation of the average positivity rate (%) for the wastewater samples collected is depicted below: $\begin{array}{c}\text{Average Positivity Rate }\left(\%\right)\\ =\frac{\text{total number of positive samples per time point *}100}{\text{total number of collection sites per time point}}\end{array}$

A gradual decrease in positivity rate occurred from Oct 2022, with no (very low in November 2022) positivity between November 2022 and January 2023. The positivity rate in the samples increased from February 2023 to April 2023, followed by a decline from May 2023 to June 2023 (Fig. 1).

Close

Fig. 1.

Monthly average positivity rate (%) among collected wastewater samples.

Export to PPT

A bar plot analysis (Supplementary Fig. 1) illustrated the distribution of viral load (number of copies) across different sites. Siddharth Nagar open drains exhibited the highest average viral load (mean: 4,585 copies/ml), followed by Kherwadi (mean: 2,440 copies/ml). Among the STPs, Worli STP showed a high viral load (mean: 3,358 copies/ml). A paired t-test revealed no significant difference (P=0.53) in viral load between STPs and open drains from the 57th time point (Supplementary Fig. 2).

Temporal variation in SARS-CoV-2 RNA concentration and early warning detection

We compared the average positivity rate (%) of samples to Mumbai’s COVID-19 case load during surveillance. A temporal trend analysis (Fig. 2) correlated SARS-CoV-2 RNA in wastewater with Mumbai’s positive cases. Viral load peaked twice during the SARS-CoV-2 waves in August 2022 and end of February 2023. The peak viral load coincided with the rise and decline of positive cases in Mumbai. From November 2022 to the second week of February 2023, no positive samples were found, aligning with Mumbai’s trendline of positive cases. A rise in viral load from February 24, 2023, preceded the surge in Mumbai’s cases by around 2-3 wk. Mumbai’s positive cases increased until May 24, 2023, and then declined in June 2023. Hence, the detection of SARS-CoV-2 RNA in wastewater signalled the early stages of local outbreaks.

Close

Fig. 2.

Trend of COVID-19 case load in Mumbai and positivity rate (%) in wastewater samples processed at The Foundation for Medical Research.

Export to PPT

Sequencing statistics

Of the collected samples, 123 (17%) were sequenced by either Illumina or Nanopore sequencing. Statistics and lineage abundance were determined for 103 samples (Illumina: n=38; Nanopore: n=65). Sequenced reads ranged from 0.001 to 1.8 million, with a median sequence coverage of 209x. Genome coverage varied from 6 per cent to 98 per cent (Supplementary Table I).

Detection of variants

In August 2022, BA.2.75 and its sub-variants predominated (70%) among other Omicron sub-variants. BA.2.75 dominance continued in September 2022 (80%), alongside BA.2.10 (5%) and the recombinant variant XBB.1* and XBB.2* (8%). From October 2022, fewer samples were sequenced, revealing XBB.2* and other Omicron sub-variants (Table I and Fig. 3).

Close

Fig. 3.

SARS-CoV-2 lineage distribution in wastewater samples from August 2022 – June 2023 was predicted using Freyja ( https://github.com/andersen-lab/Freyja ).

Export to PPT

During the second peak observed in this study between February and May 2023, XBB.1.16.1 (25%) and XBB.1.16 (15%) were dominant, followed by various sub-variants. These trends continued through May 2023, showing fluctuations in dominant variants (Supplementary Table II). All the variants for each sample are provided in the supplementary table II. Comparing variants from sewage drains and STPs revealed no differences (Supplementary Table III).

Impact of seasonal change on viral load

We conducted a comparative analysis of viral load concentrations across various seasons (monsoon, winter and summer) to understand the role of seasonal changes in facilitating transmission. The average viral load in monsoon, winter and summer was determined to be 1,695 copies/ml, 72.5 copies/ml and 3,657 copies/ml, respectively. Summer exhibited a notably high average viral load (after excluding one outlier data point). One-way ANOVA test for viral load variations among different seasons showed statistically significant differences between viral loads of winter and summer (P=0.0007) (Fig. 4). Further, seasonal viral load was compared for each open drain site using one-way ANOVA, and for STPs, a t-test was used as the viral load was only available for the monsoon and summer seasons (Table II). All the open drains showed significant differences (P≤0.0001) in seasonal viral loads except Siddharth Nagar (S8) which had zero viral load in winter (P=0.1). The viral load difference between monsoon and summer for both the STPs was non-significant (P= 0.29, 0.15). Winter and summer showed significant differences in viral load at all sites except S8. The viral load between monsoon and summer was significant for all sites except Dharavi (D2) and S8 (Table II).

Close

Fig. 4.

Comparision of seasonal change on mean viral load (copies/ml).

Export to PPT

Data availability

The genomic data produced in this research is accessible and downloadable through GISAID accession numbers (EPI_ISL_18216418 - EPI_ISL_18216453 and EPI_ISL_18226307 - EPI_ISL_18226373) on the GISAID repository ( https://gisaid.org/ ).