Genomic wastewater surveillance of seasonal and zoonotic influenza A viruses in California during the 2024-2025 flu season

Technical Summary: Genomic Wastewater Surveillance of Human and Animal Influenza A Viruses in California (2024-2025)

Problem Statement
Wastewater-based surveillance offers a "One Health" approach to detect Influenza A viruses (IAV) from both human and animal hosts, potentially overcoming limitations of clinical surveillance such as resource constraints and sampling bias. However, effective genomic surveillance of IAV in wastewater faces significant hurdles. First, IAV exists at low concentrations relative to the total microbial community, necessitating enrichment steps that are often subtype-specific or lack whole-genome resolution. Second, existing bioinformatic deconvolution tools were primarily developed for SARS-CoV-2 (a single-genome virus) and struggle with the segmented nature of the IAV genome (eight segments) and the simultaneous presence of multiple subtypes. Current methods often fail to simultaneously deconvolute multiple subtypes to a subclade level or are biased toward specific segments (e.g., Matrix or Hemagglutinin), limiting the ability to track zoonotic threats like H5N1 alongside seasonal human strains (H1N1, H3N2).

Methodology
The study implemented a comprehensive framework for IAV genomic surveillance across three wastewater treatment plants (WWTPs) in California (East Bakersfield, Central Contra Costa, and Los Angeles County) from November 2024 to February 2025.

• Sample Processing: Wastewater samples underwent nucleic acid extraction using the Promega Wizard Enviro TNA Kit. Absolute IAV concentrations were quantified via RT-dPCR targeting the M-gene and H5 subtypes.
• Probe Enrichment: A custom IAV-specific probe-capture enrichment panel was applied, targeting 11 subtypes (including H1N1, H3N2, H5N1, and others) across all eight genomic segments. This approach was chosen for its tolerance to sequence mismatches and ability to capture novel variants compared to amplicon-based methods.
• Bioinformatics & Deconvolution:
  • Reads were quality-trimmed and mapped to a custom reference database of 100,003 unique IAV isolates (excluding HA segments due to laboratory contamination from prior experiments).
  • Two pipelines were compared for estimating clade relative abundances: the EsViritu pipeline and a custom Expectation-Maximization (EM) algorithm adapted from SARS-CoV-2 methods.
  • The EM algorithm was specifically modified to handle the segmented genome and infer relative abundances for multiple subtypes in parallel without relying on clustering that might obscure fine-scale genomic differences.
  • Phylogenetic analysis of H5N1 sequences was conducted using a whole-genome maximum-likelihood tree to characterize host origins (human, avian, dairy cow).
• Validation: The system was validated using a synthetic DNA whole-genome IAV mixture and compared against clinical sequencing data and test positivity rates from the California Department of Public Health (CDPH).

Key Results

• Enrichment and Detection: The probe-capture method successfully enriched all eight segments of circulating IAV subtypes. While the overall alignment rate was low (0.023%) due to low initial viral loads, enrichment yielded thousands of IAV reads compared to zero in unenriched samples.
• Deconvolution Accuracy: The custom EM algorithm outperformed EsViritu in estimating relative abundances. EsViritu exhibited a strong bias toward H5N1 when relying on single segments (like the M-gene), whereas the EM tool accurately reconstructed the expected proportions of mixed subtypes (H1N1, H3N2, H5N1) using the remaining seven segments.
• Viral Dynamics:
  • Seasonal Strains: Human seasonal H1N1pdm09 (subclades 6B.1A.5a.2a.1 and 6B.1A.5a.2a) and H3N2 (dominated by 3C.2a1b.2a.2a.3a.1) were detected consistently in both wastewater and clinical data.
  • H5N1 Detection: Wastewater surveillance consistently detected H5N1 clade 2.3.4.4b across all three sites. In contrast, clinical H5N1 detections were sporadic and rare.
  • Discrepancies: Wastewater peaks driven by H5N1 (e.g., in mid-November at LACSD and late December at BKERSFLD) did not correspond to peaks in clinical test positivity, suggesting wastewater captures signals (potentially from animal sources) that clinical surveillance misses.
• Phylogenetic Characterization: Wastewater H5N1 sequences clustered with reference sequences from dairy cow and avian infections. All human clinical H5N1 sequences clustered exclusively with dairy cow-associated strains. No mutations associated with mammalian adaptation (e.g., PB2 E627K) were observed in the wastewater samples.
• Coverage: Shorter internal segments (MP, NP, NS) showed higher coverage depth and breadth compared to longer segments (PB1, PB2, PA).

Significance and Claims
The paper claims to be the first study to apply an IAV-specific probe-capture enrichment panel targeting all eight segments and multiple hosts in wastewater, coupled with a probabilistic EM model for multi-subtype deconvolution.

• Methodological Contribution: The study demonstrates that the EM algorithm can accurately deconvolute clade-level relative abundances for multiple IAV subtypes simultaneously, even when the most informative segment (HA) is unavailable due to contamination. This fills a gap in bioinformatic tools tailored for the complex, segmented IAV genome in wastewater.
• Public Health Utility: The framework provides a "One Health" surveillance capability that complements clinical data. It strengthens human influenza monitoring by capturing a broader, less biased population sample and enhances pandemic preparedness by detecting animal-associated IAV (specifically H5N1) that may not yet be reflected in clinical case counts.
• Limitations: The authors modestly note that while wastewater sequencing provides clade-level information, it cannot definitively distinguish whether H5N1 detections originate from human or animal infections due to the high genetic similarity between dairy cow and avian H5N1 strains and the lack of specific host-distinguishing variants in the available data.

In summary, the study validates a robust genomic surveillance framework that improves the sensitivity and resolution of IAV monitoring in wastewater, offering a more comprehensive view of circulating viral subtypes than clinical surveillance alone.