(C) PLOS One
This story was originally published by PLOS One and is unaltered.
 . . . . . . . . . .



Detecting SARS-CoV-2 cryptic lineages using publicly available whole genome wastewater sequencing data [1]

['Reinier Suarez', 'Department Of Molecular Microbiology', 'Immunology', 'University Of Missouri-School Of Medicine', 'Columbia', 'Missouri', 'United States Of America', 'Devon A. Gregory', 'David A. Baker', 'Department Of Pathology']

Date: 2025-06 

Beginning in early 2021, unique and highly divergent lineages of SARS-CoV-2 were sporadically found in wastewater sewersheds using a sequencing strategy focused on amplifying the most rapidly evolving region of SARS-CoV-2, the receptor binding domain (RBD). Because these RBD sequences did not match known circulating strains and their source was not known, we termed them “cryptic lineages”. To date, more than 20 cryptic lineages have been identified using the RBD-focused sequencing strategy. Here, we identified and characterized additional cryptic lineages from SARS-CoV-2 wastewater sequences submitted to NCBI’s Sequence Read Archives (SRA). Wastewater sequence datasets were screened for individual sequence reads that contained combinations of mutations frequently found in cryptic lineages but not contemporary circulating lineages. Using this method, we identified 18 cryptic lineages that appeared in multiple (2–81) samples from the same sewershed, including 12 that were not previously reported. Partial consensus sequences were generated for each cryptic lineage by extracting and mapping sequences containing cryptic-specific mutations. Surprisingly, seven of the mutations that appeared convergently in cryptic lineages were reversions to sequences that were highly conserved in SARS-CoV-2-related enteric bat Sarbecoviruses. The apparent reversion to bat Sarbecovirus sequences is consistent with the notion that SARS-CoV-2 adaptation to replicate efficiently in respiratory tissues preceded the COVID-19 pandemic.

Wastewater surveillance has been used during the SARS-CoV-2 pandemic to monitor viral activity and the spread of viral lineages. Occasionally, SARS-CoV-2 sequences from wastewater reveal unique evolutionary advanced lineages of SARS-CoV-2 from an unknown source, which are termed cryptic lineages. Many groups nationwide also use wastewater surveillance to track the virus and upload that information to NCBI’s SRA database. That sequence data was screened to identify 18 cryptic lineages worldwide and identify convergent mutations throughout the genome of cryptic lineages that suggest reversion to residues common in SARS-CoV-2-related Sarbecoviruses.

Funding: This project has been funded in part by the National Institutes of Health (NIH) contract 75N93021C00045 for the Johns Hopkins Center of Excellence in Influenza Research to MCJ. This project has been funded by a gift from Heart of Racing and a grant from Inkfish (MCJ and DHO). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Copyright: © 2025 Suarez et al. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Around the world, many organizations use whole genome sequencing (WGS) to detect and identify SARS-CoV-2 variants in wastewater samples. Much of this data is uploaded to the National Center for Biotechnology Information’s (NCBI) Sequencing Read Archive (SRA) or one of its international equivalents, the International Nucleotide Sequence Database Collection (INSDC). In this report, we screen 135,672 samples from over 2000 sites across 45 countries and demonstrate the feasibility of screening the SRA database to detect SARS-CoV-2 cryptic lineages and analyze their mutations.

Cryptic lineages often forecast mutations that are eventually acquired by circulating lineages. For instance, Spike substitutions N440K, S477N, E484A, and Y505H had not been seen in any major circulating lineages prior to Omicron. Yet, these mutations had repeatedly appeared in cryptic lineages long before Omicron emerged [ 5 , 6 ]. Cryptic lineage mutations have also forecast numerous changes that were acquired by later Omicron lineages [ 13 ]. The convergence between mutations found in cryptic lineages and those eventually found in circulating lineages suggests that cryptic lineages and major circulating lineages share some selective pressures. However, many of the mutations seen repeatedly in cryptic lineages have yet to become prominent in any major circulating lineage [ 13 ]. It is unknown whether major circulating lineages will eventually acquire those mutations or whether those mutations account for selective pressures that differ from circulating lineages.

Wastewater surveillance has been widely used to identify chemicals and microbes [ 1 – 3 ]. During the SARS-CoV-2 pandemic, this technique gained prominence for its efficient tracking of various variants of concern [ 4 ]. Our group began tracking SARS-CoV-2 lineages from wastewater in early 2021, and in March 2021, we discovered the first instance of an evolutionarily advanced SARS-CoV-2 receptor binding domain (RBD) haplotype that appeared repeatedly in a single sewershed, which we later termed a “cryptic lineage” [ 5 ]. Examples of cryptic lineages have now been reported worldwide [ 5 – 11 ]. Similarities between genomes from persistent SARS-CoV-2 infections in immunocompromised patients and cryptic lineages suggest these may reside within immunocompromised individuals [ 8 , 12 , 13 ]. Furthermore, a single cryptic lineage derived from a lineage that stopped circulating in early 2021 was traced to a commercial building in late 2022, and 12S ribosomal RNA sequencing of the wastewater indicated that the only meaningful species contributing to the wastewater was human [ 13 ]. Therefore, cryptic lineages are believed to be derived from individuals with very long SARS-CoV-2 infections.

Results

Using conservative thresholds, our lab has identified over 20 cryptic lineages by amplifying the RBD sequence from SARS-CoV-2 RNA in wastewater samples [5,6,13]. From the previously discovered cryptic lineages, we compiled a list of mutations observed in cryptic lineages that were not commonly found in circulating lineages (S1 Fig). This list of 69 amino acid substitutions in the spike RBD was termed “cryptic lineage-defining amino acid substitutions.”

Using the search terms “SARS-CoV-2 wastewater”, we downloaded wastewater SARS-CoV-2 sequence reads from SRA that were available on February 18th, 2024 (S2 Data), that had sample collection dates on or before October 31 2023, mapped these reads to the SARS-CoV-2 genome (NC_045512), and processed them with the program SAM Refiner [14]. We identified individual sequencing reads in the SRA datasets that contained at least two of the cryptic lineage-defining amino acid substitutions (S1 Data). These were analyzed using automated scripts followed by manual validation to identify haplotypes that did not match any known sequence from a patient sample and appeared in at least two samples from the same sewershed. Using the subset of identified sequences, we found sequencing reads consistent with 18 independent cryptic lineages. Of the 18 identified lineages, three of the lineages we reported previously and three of the lineages had been reported by other groups [5–7,9,11,13]. The duration of detection varied widely among the cryptic lineages; the shortest time a cryptic lineage was detected was one month (CA-1 and NY-2), while two cryptic lineages were detected for over a year (UK-1 and WI-1) (Table 1).

After cryptic lineages were identified based on their RBD sequence, we retrospectively identified other datasets from the same sewershed that share cryptic-defining characteristics outside of the RBD to partially reconstruct lineage genomes. We compared the individual SARS-CoV-2 sequences present in wastewater samples from sewersheds containing cryptic lineages to the sequences from samples from neighboring (same state) sewersheds collected during the same time period and, when possible, sequenced by the same agency (S2 and S3 Data). Individual mutations that appeared in at least two samples from the cryptic sewershed and were at least 50x more prevalent in the cryptic-containing sewershed than in neighboring sewersheds were considered putatively cryptic-specific mutations (Fig 1a). Additionally, any mutation frequently appearing in the same sequence read as the cryptic-specific mutation was presumed to be present in the cryptic lineage (Fig 1b; see methods for specific criteria). This process was repeated with all 18 cryptic lineages to approximate the polymorphisms present in each lineage (S4 and S5 Data). A consensus sequence was generated for each cryptic lineage using its cryptic-specific mutations and sequences that appeared on the same read as the cryptic-specific mutations (Fig 1b, S6 Data). Generating a complete consensus sequence for each cryptic lineage was not possible in most cases because regions were excluded that did not have any cryptic-defining mutations. As a result, sequence coverage varied between cryptic lineages, with the highest coverage being 73.97% (MI-1) and the lowest 11.43% (CO-1). The consensus sequence was used as input for the phylogenetic software programs UShER [16] and Nextclade [15] to determine its predicted parent SARS-CoV-2 lineage (Table 1). All the cryptic lineages were predicted to be derived from lineages that stopped circulating months to years prior to their detection in wastewater (Table 1). A phylogenetic tree of the cryptic lineages illustrates the extreme diversity of these lineages (Fig 2). The use of a consensus sequence, which is derived from a mixture of diverse lineages with a shared common ancestor, could potentially influence the branch lengths in the phylogenetic tree, and may not fully capture the true diversity within each cryptic lineage.

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 1. Schematic of workflow. Samples from sewer shed facilities containing cryptic lineages (yellow) were compared against samples from neighboring sewer sheds that did not contain cryptic lineages (orange). A) Using the CH-1 cryptic lineage as an example, mutations found in at least two cryptic samples, with a prevalence of 50x more in the cryptic samples, are tentatively considered cryptic-specific (green). B) The sequence reads containing cryptic-specific mutations (red box) were mapped onto the SARS-CoV-2 genome, with varying coverage across the genome to create a consensus sequence (middle genome). To be mapped onto the genome, a cryptic-specific sequence must appear in two or more samples. https://doi.org/10.1371/journal.ppat.1012850.g001

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 2. The phylogenetic tree generated by NextClade illustrates the diversity of the cryptic lineages. The consensus sequences were uploaded onto Nextclade and compared against the Wuhan-Hu-1/2019 (MN908947). The phylogenetic tree highlights the diversity among the cryptic lineages detected. https://doi.org/10.1371/journal.ppat.1012850.g002

Interestingly, we observed the same mutations in the consensus sequence appearing in multiple independent cryptic lineages. Such convergent changes are unlikely to be sequencing artifacts and likely reflect adaptation to common selective pressures. Mutations that appeared in three or more cryptic lineages were mapped onto a diagram of the SARS-CoV-2 genome while excluding mutations found in the consensus sequence of the parent lineages (Fig 3a, 3b; S7 Data). We observed 85 nucleotide changes in at least three cryptic lineages. The most common changes in Spike were K417T (78%) and Q493K (56%), which have been shown to affect antibody escape [17,18] and may also have small effects on ACE2 binding, with effects ranging from slightly increasing to modestly decreasing binding depending on the genetic background [19]. Although K417T was present in the Gamma variant of concern and a few Omicron sub-lineages, such as BA.2.18, it has been present in less than 1% of circulating lineages found in people. By contrast, Q493K is extremely rare and has not been a lineage-defining change in any named PANGO lineage. The most common cryptic-specific mutations outside the Spike were in ORF1a (K1795Q) and ORF3a (H182D), each observed in 50% of the identified cryptic lineages.

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 3. Convergent cryptic changes found in ≥ 3 lineages. A) Convergent mutations that appeared in at least three cryptic lineages were mapped onto the spike protein based on their location and prevalence across all the cryptic lineages. B) Convergent non-Spike mutations mapped against the SARS-CoV-2 genome. Positions which contain multiple mutations in the same position are represented as stacked bars and color-coded. https://doi.org/10.1371/journal.ppat.1012850.g003

Among the 85 nucleotide changes that occurred convergently in at least three cryptic lineages, 79 changed a protein sequence through non-synonymous changes or deletions. Of the four changes that did not alter a protein sequence, two were silent (C25162A/Spike: L1200L, 22.22%), and three were in non-coding regions (T78A (16.67%), A178G (16.67%), and T29758G (33.33%)). Interestingly, we observed that the Spike change C25162A (L1200L) was always associated with the neighboring C25163A (Q1201K) change. These two mutations together create the sequence TCTAAAAGAACT, which is a near-perfect match to the consensus SARS-CoV-2 transcription regulatory sequence (TRS) TCTAAACGAACT [20]. Although C25162A and C25163A are relatively rare in patient sequences [21], the two changes usually occur together (>60% of the time). While the function of this additional TRS is not known, it is a possible explanation for the convergence of the silent C25162A change.

A particularly notable convergent non-coding change in the cryptic lineages is at the 3’ UTR of the SARS-CoV-2 genome, T29758G. This mutation is in the highly conserved region of the stem-loop two motif (s2m), which is found in many Coronaviruses and other RNA viruses [22–24]. Remarkably, the s2m in SARS-CoV-2 deviates from the consensus s2m found in other RNA viruses, including Sarbecoviruses, and the T29758G mutation restores SARS-CoV-2 to the consensus s2m sequence [24,25]. The s2m stem-loop is not essential for replication as the sequence was deleted in Omicron lineage BA.2 and all of its derivatives; thus, it has been nearly absent in circulating lineages for over two years [26]. However, in the case of the cryptic lineages, the sequence frequently reverts the SARS-CoV-2 s2m to the Sarbecovirus consensus sequence.

Several of the most common convergent changes in cryptic lineages, such as ORF1a: K1795Q and T29758G, were conversions to the sequence found in closely related bat Sarbecoviruses such as RaTG-13. Although SARS-CoV-2 is a human respiratory pathogen, the most closely related Sarbecoviruses primarily infect Horseshoe bats and are primarily believed to be enteric pathogens. To explore if other convergent changes in cryptic lineages represent reversions to the Sarbecovirus consensus sequence, the sequences of seven closely related Sarbecoviruses (RpYN06, RaTG-13, BANAL-52, BANAL-103, BANAL-116, BANAL-236, and BANAL-247) were compared to SARS-CoV-2 to identify amino acid positions that were conserved across all seven sarbecoviruses, but differed in the original SARS-CoV-2 A and B lineages.

A total of 26 amino acid substitutions were identified where the SARS-CoV-2 sequence differed from all seven of the bat sarbecoviruses. Of these 26 positions, 12 substitutions in cryptic lineages had reverted to the Sarbecovirus consensus sequence in at least one cryptic lineage, and seven of the reversions occurred in at least three cryptic lineages (Fig 4). As of October 31st, 2023, in these 26 positions, none had reverted to the Sarbecovirus sequence in over 1% of US patient SARS-CoV-2 sequences. However, low prevalence in patient sequences does not necessarily mean that mutations cannot provide a fitness advantage. A study by Bloom and Neher examined how frequently particular mutations were independently acquired by SARS-CoV-2 relative to how frequently they would be expected to be acquired by random chance [27]. From this, they assigned a fitness score to individual SARS-CoV-2 mutations. Of the seven Sarbecovirus reversions that were acquired by at least three cryptic lineages, only two were predicted to be advantageous. Interestingly, the change predicted to be the most deleterious is Spike: A372T. This position has been noted previously because all Sarbecoviruses except SARS-CoV-2 contain a predicted glycosylation site at position 370, and A372T restores this site. In addition to the five cryptic lineages that contained A372T, two other cryptic lineages also restore this glycosylation site through changes S371N and A372DEL. It has been experimentally demonstrated that restoring glycosylation to SARS-CoV-2 through the A372T mutation reduces viral replication in human lung cells [28]. In addition, it has been demonstrated that abolishing the glycosylation site at position 370 in bat Sarbecoviruses makes the virus highly sensitive to trypsin digestion at pH 5.5 [29]. Thus, the frequent reversion to the consensus bat sarbecovirus sequence in cryptic lineages is consistent with cryptic lineages being subject to some similar selective pressures as that of its enteric progenitors.

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 4. Chart of SARS-CoV-2 amino acids that deviate from the consensus Sarbecovirus amino acid sequence. Consensus amino acids found across seven bat Sarbecoviruses (orange) that are different in SARS-CoV-2 (yellow). The amino acid positions where a change is observed but differ from the Sarbecoviruses and SARS-CoV-2 are highlighted in blue. Frequency of patient sequences, reported by CoV-Spectrum [21], reverting to the Sarbecovirus consensus by October 2023 or November 2024 are shown. The independent occurrence calculated from Bloom and Neher’s calculator [26] for each mutation over the same time periods is shown, including its fitness score and effect. Mutations that did not appear in Bloom and Neher’s calculator were designated as not determined (ND). https://doi.org/10.1371/journal.ppat.1012850.g004

Five of the cryptic lineages were found to have small insertions (Fig 5). Three of the insertions occurred in the ectodomain of the structural proteins, specifically in the spike and M genes, as was previously noted for one cryptic lineage, and the other two insertions occurred in non-structural genes, ORF3a and ORF7a [13]. A closer observation of the inserted nucleotide sequence revealed that four of the five insertions were duplicated sequences from other parts of the SARS-CoV-2 genome.

One cryptic lineage was detected in SRA datasets from two different sewersheds separated by approximately 40 miles. Samples were independently obtained from both sewersheds and tested for the presence of the cryptic lineage. Samples from both sewersheds contained a cryptic lineage that closely matched the sequence observed in the SRA sequences (S2 and S3 Figs). Similar to our finding with the cryptic lineage found in Wisconsin (WI-1) [13], the sequence from the Ohio cryptic lineage did not remain static over a nine-month period (Fig 6). Both sewersheds from Ohio shared highly similar cryptic-specific mutation profiles throughout the dates detected in the SRA. Notably, mutations in the Spike protein, specifically N460K, F486P, Q493T, and P499T, were detected for the first time on the same date from both sewer sheds, strongly suggesting that this lineage was being deposited into wastewater from a single source, likely a person who commuted between the two locations. The Ohio cryptic lineage persisted until June 2023 before disappearing. The successful detection of the Ohio cryptic lineage from independently obtained samples serves as a valuable validation of the method’s ability to detect cryptic lineages from public repositories.

[END]
---
[1] Url: https://journals.plos.org/plospathogens/article?id=10.1371/journal.ppat.1012850

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/