The novel genus, ‘Candidatus Phosphoribacter’, previously identified as *Tetrasphaera*, is the dominant polyphosphate accumulating lineage in EBPR wastewater treatment plants worldwide

Phosphorus is a critical and limited resource, driving interest in cost-effective and sustainable removal and recovery methods from wastewater. EBPR relies on polyphosphate-accumulating organisms (PAOs). While ‘Candidatus Accumulibacter’ has been considered the key PAO, members classified as Tetrasphaera are frequently dominant in EBPR systems worldwide. However, 16S rRNA gene-based taxonomy has been inadequate for resolving the true phylogeny of Tetrasphaera, and prior isolates belong to multiple genera within Dermatophilaceae. The research addresses whether the abundant, uncultured Tetrasphaera clade 3 comprises distinct, novel genera and clarifies their roles as PAOs in EBPR. The study aims to resolve taxonomy, diversity, distribution, and metabolic potential using high-quality MAGs, global 16S rRNA datasets, and in situ validation, improving models for phosphorus removal and resource recovery in WWTPs.

• PAO physiology has been studied for decades; the classical PAO model (based on Ca. Accumulibacter) involves PHA and glycogen cycling across anaerobic/oxic phases. Tetrasphaera differ, generally lacking PHA storage and showing more versatile metabolisms.
• Eight Tetrasphaera species have been described from isolates, but genome-based phylogeny reassigns them to several genera in Dermatophilaceae (Phycicoccus, Knoellia, and Tetrasphaera sensu stricto), indicating the former genus is paraphyletic.
• Tetrasphaera clade structure based on 16S rRNA genes divides into clades 1–3, with clade 3 comprising environmental sequences only and being dominant in EBPR but lacking MAGs/isolates.
• Prior work suggests T. elongata stores polyP oxically and releases anoxically, using amino acids and glucose; glycogen’s role is uncertain due to lack of detection in situ.
• Resolving clade 3 phylogeny and physiology is essential because it dominates Danish and global EBPR systems and underpins phosphorus removal performance.

• Genome recovery and phylogenomics: Screened 1083 high-quality Danish WWTP MAGs to identify Tetrasphaera-related MAGs (n=14) using GTDB-Tk. Built a maximum-likelihood genome tree from concatenated 120 single-copy marker proteins (WAG+G, 100 bootstraps) with IQ-TREE; visualised in ARB/ITOL. Outgroup: Kineococcus spp. ANI (pyani ANIb) calculated among MAGs and isolates to define genus/species boundaries; completeness/contamination assessed with CheckM.
• rRNA gene phylogenies: Extracted 16S and 23S rRNA genes via Prokka/Fxtract; aligned with MAFFT; built ML trees with IQ-TREE (16S: TIM3+F+I+G4; 23S: TIM3+F+R4; 100 bootstraps). Outgroup: Kineococcus spp. Low-quality/fragmented sequences removed.
• Community profiling: Used MIDAS3 (Danish; 712 samples, ≥13,500 reads/sample) and MIDAS4 (global; 847 samples from 438 WWTPs, ≥10,000 reads/sample) V1–V3 16S rRNA amplicon datasets to quantify species/ASV abundances across process configurations and countries. Metagenome read-mapping from 69 Danish WWTP metagenomes to species-representative MAGs to resolve co-occurrence and diversity.
• Comparative genomics: Annotated MAGs and isolates with EnrichM; assessed KEGG module completeness (≥80%), KO enrichment for clade comparisons, and Ca. Phosphoribacter species differences. Validated annotations, synteny, and pangenome using MicroScope (core vs species-specific genes at 50% aa identity and 80% coverage). OrthoFinder (DIAMOND backend) identified orthogroups unique to Ca. P. baldrii vs Ca. P. hodrii; functional annotation via EggNOG-mapper.
• Secreted peptidases: Predicted subcellular localization with PSORTb; searched MEROPS peptidases via DIAMOND BLASTP (e≤1e−20); signal peptides predicted with PRED-TAT; filtered out COG category M (cell wall/membrane/envelope) via EggNOG-mapper to retain extracellular peptidases.
• Experimental validation: Full-scale activated sludge batch P-cycling experiments (Aalborg West WWTP). Aerated sludge to deplete intracellular carbon; anoxic incubation (N2 headspace) with acetate/glucose/casamino acids (500/250/250 mg/L) for 3 h; periodic ortho-P sampling. FISH-Raman microspectroscopy (species-specific FISH probes) quantified intracellular polyP, PHA, and glycogen at start and end; design and in silico validation of new 16S/23S rRNA-targeted FISH probes (MathFISH, SILVA TestProbe), plus use of existing probes; microscopy to assess morphology and hybridization.

• Taxonomic redefinition: Former Tetrasphaera are paraphyletic within Dermatophilaceae and represent at least 6–7 genera. Among 14 high-quality WWTP MAGs, 11 belong to 16S clade 3; genome ANI delineates two novel genera within clade 3: ‘Candidatus Phosphoribacter’ (midas_s_5) and ‘Candidatus Lutibacillus’ (midas_s_45), matching a genus ANI boundary ~75–77%.
• Species diversity: Ca. Phosphoribacter comprises at least six genomically distinct species (Pbr1–Pbr6; <95% ANI), despite highly similar or identical 16S rRNA sequences (97.5–100% identity), showing 16S rRNA gene lacks species/genus resolution for this lineage.
• Dominant PAOs: Ca. Phosphoribacter and Ca. Lutibacillus are the most abundant former Tetrasphaera PAOs in EBPR plants in Denmark and globally. Tetrasphaera relative abundances in EBPR (C,N,DN,P) plants are highest in Denmark (14.9% average), Canada (8.4%), and Malaysia (7%), and are enriched in EBPR configurations vs non-EBPR (C or C,N) processes.
• Two co-dominant species: The two most abundant Ca. Phosphoribacter species, Ca. P. baldrii (Pbr1) and Ca. P. hodrii (Pbr2), share an identical V1–V3 16S ASV (ASV1) but differ markedly genomically (~82% ANI), and often co-occur at high abundance, implying niche partitioning.
• Metabolic potential: Clade 3 encodes complete central carbon pathways (glycolysis, PPP, TCA), diverse transporters for sugars and amino acids, and widespread capacity for lactate use and β-oxidation-associated pathways. Polyphosphate metabolism genes (ppk, ppx-gppA, pstSCAB, pit, phoU/phoRB) are nearly ubiquitous across TRC, consistent with PAO potential.
• Storage compounds: None of the TRC genomes encode the full PHA synthesis suite (PhaABC), and Raman-FISH did not detect PHA or glycogen in situ in clade 3, contrasting with classical PAOs. Genes for glycogen-like α-glucans may relate to capsular material rather than energy storage. Cyanophycin synthesis/degradation genes (cphA/cphB) are common in TRC, suggesting alternative nitrogen/carbon storage.
• Secreted peptidases and proteasomes: Clade 3 MAGs encode multiple predicted secreted peptidases (2–7 per MAG), indicating extracellular protein/peptide scavenging; most TRC genomes encode proteasomes (63/69), potentially aiding resource recycling under fluctuating WWTP conditions. Ca. Accumulibacter MAGs lacked predicted secreted peptidases.
• Denitrification steps: Many TRC genomes encode respiratory nitrate reductase (NarGHI), but downstream nitrite, nitric oxide, and nitrous oxide reduction genes are less widespread—suggesting participation in partial denitrification/nitrogen cycling rather than complete denitrification.
• Species-level differences: Ortho-group analyses identified 371 OGs unique to Ca. P. baldrii (e.g., transposases, toxin/antitoxin systems, biotin/folate transport/modification) and 355 OGs unique to Ca. P. hodrii (e.g., sugar kinases, sucrose utilization, cytochrome P450s). Substrate distinctions include: Ca. P. baldrii potentially uses ethanolamine and glycerol (from phosphatidylethanolamine turnover) and lacks pta/ackA; Ca. P. hodrii encodes N-acetylglucosamine utilization and the pta/ackA pathway for acetate metabolism/fermentation.
• Experimental validation: Species-specific FISH probes visualized Ca. P. baldrii and Ca. P. hodrii as short rods (0.6–0.9 × 1–1.5 µm) in microcolonies. Raman-FISH detected substantial intracellular polyphosphate in both Ca. Phosphoribacter and Ca. Lutibacillus, with dynamic P cycling across anaerobic/oxic phases in batch experiments; no PHA or glycogen detected.

This study resolves the taxonomy of the dominant Tetrasphaera clade 3 in EBPR, demonstrating it comprises two novel and globally abundant genera, Ca. Phosphoribacter and Ca. Lutibacillus. The findings show that 16S rRNA genes lack sufficient resolution for Dermatophilaceae lineages and that genome-resolved approaches are essential. The two co-dominant Ca. Phosphoribacter species, despite identical 16S ASVs, are genomically and metabolically distinct and regularly co-occur, suggesting complementary niches, likely driven by differences in carbon/nitrogen substrate use (e.g., ethanolamine vs N-acetylglucosamine; presence/absence of pta/ackA). Widespread polyphosphate metabolism genes and in situ Raman-FISH/P-cycling experiments confirm that both Ca. Phosphoribacter and Ca. Lutibacillus are active PAOs, implicating them as major contributors to P removal. Their metabolic profiles (absence of PHA/glycogen storage, potential use of amino acids/sugars, partial denitrification) indicate ecological strategies distinct from Ca. Accumulibacter and Dechloromonas, with practical implications for EBPR process design, monitoring (probe selection), and P recovery strategies. The reclassification renders model isolate T. elongata a poor proxy for clade 3 physiology, underscoring the need to tailor EBPR models to the newly defined taxa.

• The former Tetrasphaera lineage is paraphyletic and includes multiple genera; clade 3 comprises two novel, globally abundant PAO genera: ‘Candidatus Phosphoribacter’ and ‘Candidatus Lutibacillus’.
• Ca. Phosphoribacter contains at least six species, including two co-dominant species (Ca. P. baldrii and Ca. P. hodrii) that are indistinguishable by common 16S V1–V3 amplicons but differ genomically and metabolically, enabling coexistence.
• Genomic evidence and in situ FISH-Raman and P-cycling experiments establish these lineages as key PAOs in EBPR worldwide, often equalling or surpassing Ca. Accumulibacter in abundance.
• The physiology and storage strategies of these PAOs differ from classical models, lacking PHA/glycogen storage and potentially utilizing alternative substrates and cyanophycin. Future work: Isolate/enrich representatives to resolve growth requirements; experimentally verify predicted substrate ranges (e.g., ethanolamine, N-acetylglucosamine), electron acceptor usage, and storage compounds; integrate species-resolved dynamics into EBPR process models; develop diagnostic probes/assays for routine monitoring and control.

• No cultured isolates of clade 3 genera were available; conclusions rely on MAG-based inference and in situ validation, which may miss low-abundance genes or functions.
• 16S rRNA gene data lack resolution for species delineation; species assessments depend on genome ANI and metagenomic mapping, which can be influenced by MAG completeness and assembly biases.
• Experimental P-cycling used mixed-substrate additions and community sludge, not pure cultures; substrate preferences and metabolic pathways remain to be confirmed per species.
• Denitrification capabilities beyond nitrate reduction were variably encoded; actual in situ nitrogen transformation fluxes were not quantified.
• Global abundance estimates are based on amplicon datasets with primer bias (V1–V3) and variable sampling depth, potentially under/over-representing lineages in some regions.