Strains 16FHM2^T (=CCUG 74779^T=CECT 30571^T), 15FMM2 (=CCUG 74780) and 15FMM3 (=CCUG 74781) were isolated from gut contents of two specimens of wild Atlantic mackerel (Scomber scombrus) collected in the northern North Sea (International Council for the Exploration of the Sea, ICES region 4.a, in November 2018) (Marathe et al., 2022). The three strains were isolated on Mueller-Hinton (MH) agar containing meropenem (0.125 μg mL⁻¹), incubated at 30°C, for 36 h.

Initial identification was performed, using matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS), using Bruker MALDI Biotyper^® (Bruker Daltonics, Bremen, Germany), as previously described (Grevskott et al., 2024); the strains were identified as belonging to the genus Pseudomonas but could not be assigned to any described species. Total genomic DNA was extracted from strain 16FHM2^T, using the previously described “heat-shock” protocol (Welinder-Olsson et al., 2000). The nearly complete 16S rRNA gene was amplified by PCR, using modified versions of primers (16F28, 5’-AGAGTTTGATCKTGGCTCAG-3′ and 16R1494, 5’-TACGGYTA CCTTGTTACGAC-3′) (Lane, 1991; Hauben et al., 1997), as described previously (Carvalheira et al., 2020). The PCR products were purified and sequenced, as described previously (Jaén-Luchoro et al., 2020), using modified versions sequencing primers (16F530, 5’-TTCGTGCCAGCAG CCGCGG-3′ 16R806, 5’-GGACTACCAGGGTATCTAAT-3′; 16F1103, 5’-TGTTGGGTTAAGTCCCGCAAC-3′, and 16R1494 5’-TACGGYTA CCTTGTTACGAC-3′) (Lane, 1991; Hauben et al., 1997; Buchholz-Cleven et al., 1997) and an Applied Biosystems SeqStudio 8 Flex Genetic Analyzer system (Thermo Fisher Scientific, Waltham, MA, United States). The partial rpoD sequence was determined, using previously-described primers and protocols for amplification and sequencing (Mulet et al., 2009). The 16S rRNA and rpoD gene sequences were analyzed using EzBiocloud (Yoon et al., 2017), and NCBI BLAST (Altschul et al., 1990) against sequences of type material of the Nucleotide collection (nr/nt) of GenBank (Sayers et al., 2024), respectively.

The strains were grown overnight on MH Agar medium and were characterized by biochemical profiling, using the CCUG NFX worksheet,² including API ZYM and API 20NE commercial panels (bioMérieux, Marcy-l’Étoile, France). Bacterial growth was evaluated by streaking the strains on MH Agar plates and incubating at different temperatures ranging from 4 to 42°C for as long as 96 h.

Cell sizes and morphology were examined after 1 day incubation on Blood Agar medium, at 30°C, using a digital holotomographic microscope (DHM; HT-2, Tomocube Inc., Daejeon, South Korea), following previously described procedures (Fernández-Juárez et al., 2023). This enables non-invasive, label-free 3D imaging of bacterial cell morphology, based on obtaining tomographic refractive index (RI) data, which can subsequently be segmented to highlight particular cell structures based on their distinct RI signatures. For holotomographic imaging, bacterial colonies were recovered from the agar medium and re-suspended in phosphate-buffered saline (PBS). Then, 30 μL of the bacterial suspension was placed in a Tomodish (Tomocube Inc., Daejeon, South Korea) and covered with a coverslip in preparation for imaging. The resulting tomographic imaging data were visualized and analyzed, using TomoStudio and TomoAnalysis software (Tomocube Inc., Daejeon, South Korea). An average of 20 cells were imaged and analyzed, and the resulting images were exported to ImageJ software for additional analysis and figure preparation.

The strains were grown overnight on MH Agar medium with ampicillin (100 μg mL⁻¹) and analyzed for antimicrobial sensitivity, using broth dilution method on Sensititre™ plates (Thermo Fisher Scientific, Waltham, MA, United States), following the manufacturer’s instructions, as described previously (Grevskott et al., 2024). Briefly, suspensions were prepared from freshly grown cultures to an optical density of 0.5 McFarland. Ten microliters of suspensions were added to 10 mL of MH Broth and added to sensititre plates. The plates were incubated at 30°C (optimal temperature for the strains) for 24 h and read manually. Minimum inhibitory concentrations (MICs) were determined as the lowest tested antimicrobial concentrations with no observed growth, as recommended by the EUCAST reading guide for broth microdilution (Version 5.0, January 2024).

The strains were cultivated on Trypticase Soy Agar medium at 28°C for 24 h for cellular fatty acid-fatty acid methyl ester (CFA-FAME) analysis. The CFA-FAME profiles were determined, using a gas chromatograph (HP 5890; Hewlett-Packard, Palo Alto, CA, United States), following a standardized protocol similar to that of the MIDI Sherlock MIS system (Sasser, 2001), as described previously (Zamora et al., 2012).

Strains were cultivated on MH Agar medium with ampicillin 100 μg mL⁻¹ at 30°C for 24 h, and genomic DNA was prepared, using a DNeasy Blood & Tissue kit (Qiagen, Hilden, Germany), for Illumina short-read sequencing, and a modified version (Salvà-Serra et al., 2018) of a previously described protocol (Marmur, 1961) for Oxford Nanopore long-read sequencing. For Illumina sequencing, a paired-end library was prepared, using a Nextera DNA Flex library prep kit, and sequenced on an Illumina MiSeq platform (Illumina, Inc., San Diego, CA, United States) at the Norwegian Sequencing Centre in Oslo. For Oxford Nanopore sequencing, a library was prepared, using a Rapid Barcoding Sequencing kit (SQK-RBK004), and sequenced for 72 h on a MinION device (Oxford Nanopore Technologies, Ltd., Oxford, United Kingdom), at the Research Lab of the Culture Collection University of Gothenburg (CCUG), using a FLO-MIN106 (version R9.4.1) flow cell and analyzed, using the software MinKNOW version 3.6.5 (Oxford Nanopore Technologies), with default parameters. The raw Oxford Nanopore reads were base-called, using Guppy version 3.4.5 and evaluated, using NanoPlot version 1.26.3 (De Coster et al., 2018). The Oxford Nanopore sequence reads were assembled de novo following a previously described protocol (Wick et al., 2023). Briefly, four subsets of the Oxford Nanopore reads were created, using Trycycler v0.5.5 (Wick et al., 2021). Each subset was assembled de novo, using Flye v2.9.5 (Kolmogorov et al., 2019), Raven v1.8.3 (Vaser and Šikić, 2021) and Canu v2.2 (Koren et al., 2017). A consensus assembly was generated, using Trycycler v0.5.5, and polished using three different methods: Medaka v1.11.3 (i.e., using the Oxford Nanopore reads)³, with the mode r941_min_high_g344; Polypolish v0.6.0 (i.e., using the Illumina reads) (Wick and Holt, 2022); and POLCA, using Masurca v4.1.0 (Zimin and Salzberg, 2020; Zimin et al., 2013). Manual observations of read mappings were performed, using UGENE v48.1 (Okonechnikov et al., 2012). The Illumina-only assemblies were performed, using SPAdes version v3.13 (Bankevich et al., 2012), and evaluated, using QUAST version 5.2 (Gurevich et al., 2013). The genome assemblies were annotated, using PGAP v6.3 and v6.9 (Tatusova et al., 2016).

The amino acid sequences of possible β-lactamases were analyzed, using TBLASTN against the entire Core nucleotide database (core_nt) of NCBI GenBank (Sayers et al., 2024). A second search was conducted, against the Nucleotide collection (nr/nt), restricting the search space to the genus Pseudomonas (Taxonomy ID: 286). Sequences were also analyzed using BLASTP against the Beta-Lactamase DataBase (Naas et al., 2017). The genetic context was analyzed, using Unipro UGENE v48.1 (Okonechnikov et al., 2012).

Digital DNA–DNA hybridization (dDDH) (Auch et al., 2010) values were determined, using the Genome-to-Genome Distance Calculator (GGDC) v3.0 (Meier-Kolthoff et al., 2013). Average nucleotide identity values, based on BLAST (ANIb) (Goris et al., 2007; Altschul et al., 1990), were calculated, using the webserver JSpeciesWS (Richter et al., 2016). For each comparison, ANIb values were determined bi-directionally and the average calculated.

The matrix containing the average ANIb values was used to generate a dendrogram, using the software PermutMatrix v1.9.3 (Caraux and Pinloche, 2005). The dendrogram was constructed, using Pearson’s distance correlation and hierarchical clustering with an average linkage method (UPGMA). The dendrogram was displayed, using the Interactive Tree of Life (iTOL) v7.1 (Letunic and Bork, 2024).

A core genome-based phylogenomic tree was constructed, including type strains of species of the P. fluorescens and P. gessardii subgroups (Supplementary Table 1). Briefly, the genome sequences were annotated, using Prokka v1.14.6 (Seemann, 2014), and the annotated proteins sequences were compared, using the Software GET_HOMOLOGUES v17112020 (Contreras-Moreira and Vinuesa, 2013), with BLASTP (all vs. all) (Altschul et al., 1990). The sequences were clustered by applying a 70/70 criterion (i.e., 70% of identity over 70% of the length), using three clustering algorithms: BDBH; COGtriangles (Kristensen et al., 2010); and OrthoMCL (Li et al., 2003). The intersection of the three algorithms was used to determine a consensus core genome formed by single-copy orthologous sequences. The sequences were aligned, using Clustal Omega v1.2.0 (Sievers et al., 2011), and the alignment was trimmed, using Gblocks v0.91b (Castresana, 2000). Subsequently, the alignments were concatenated and a phylogenomic tree was constructed, using PhyML v20120412 (Guindon et al., 2010) and a Shimodaira-Hasegawa-like approximate likelihood-ratio test (SH-aLRT) for branching statistical support (Anisimova and Gascuel, 2006). The phylogenomic tree was visualized, using iTOL v7.1 (Letunic and Bork, 2024).

Additional related strain genome sequences were searched by analyzing the partial rpoD sequence, using BLASTN (Altschul et al., 1990), against the Nucleotide collection (nt) of NCBI. Metagenome-assembled genomes (MAGs) of the proposed novel species were searched using Protologger v2 (Hitch et al., 2021). The Branchwater Metagenome Query platform (Irber et al., 2022) was used to search in more than one million metagenomic datasets from the Sequence Read Archive (SRA) (Leinonen et al., 2011).

During a screening for β-lactam-resistant bacteria in gut samples of wild Atlantic mackerel (Scomber scombrus) from the northern North Sea, three Pseudomonas spp. strains were isolated. The rod-shaped cells were Gram-stain-negative, forming smooth, translucent (2–3 mm wide) colonies when grown for 36 h on MH agar medium with ampicillin.

MALDI-TOF MS typing analysis could not identify the strains to the species level. The 16S rRNA gene sequence showed relatedness to P. libanensis (99.9%), as well as to P. synxantha, P. gessardii and P. shahriarae (99.9%) while partial rpoD sequence indicated distant relationships to P. proteolytica (93.4%) and P. mucidolens (92.7%) and, thus, could not be assigned to any existing species of the genus Pseudomonas (Girard et al., 2020). The discrepancies observed in the identifications by 16S rRNA gene and rpoD sequence analyses indicated a high probability that the strains represented a novel species.

The strains are Gram-negative and motile. The strains grow between 10 and 35°C but not at 4 or 42°C on MH Agar medium. Only one of the strains was able to grow when cultivated on blood agar at 37°C, suggesting that persistence in humans might be possible. The strains demonstrated good growth at salinities up to 3% (w/v) NaCl, while variable growth was observed at 4.5 and 5% NaCl. The strains are catalase- and oxidase-positive and exhibit gelatine hydrolysis, nitrate reduction and esterase activities. Additional phenotypic traits, including the results of multiple growth assays and biochemical analyses, are presented in Supplementary Table 2. The cell morphology was studied by digital holotomographic microscopy and the cell size was determined to be 1.6 ± 0.3 μm by 0.8 ± 0.05 μm (Figure 1).

The strains exhibited high MIC values for ampicillin (>64 μg mL⁻¹), cephalosporins, such as cefazolin, cefuroxime and cefoxitin (>32 μg mL⁻¹), and carbapenems, such as ertapenem (8 μg mL⁻¹), as well as against azithromycin (8 μg mL⁻¹) and trimethoprim (>16 μg mL⁻¹), while the strains exhibited low MIC values for ciprofloxacin, gentamycin and tigecycline. The MICs for different antimicrobials tested are presented in Supplementary Table 3.

Strains 16FHM2^T, 15FMM2 and 15FMM3 presented CFA-FAME profiles characteristic of species of the genus Pseudomonas: palmitic acid (C_16:0), C_10:0 3-OH, C_12:0 2-OH and C_12:0 3-OH. The major CFAs of the strains were C_16:1 ω7c (36.0–37.2%), C_16:0 (21.6–23.4%), followed by the summed feature formed by C_18:1ω7c, 12 t and/or 9 t (11.7–12.9%) and C_17:0 cyclo (10.9–11.8%) (Table 1). Compared with P. mucidolens CCUG 1424^T, the three strains present similar CFA-FAME patterns overall, although they displayed higher proportions of C_17:0 cyclo and lower proportions of C_18:1ω7c/12t/9t.

The genomes of the three strains were sequenced, using an Illumina MiSeq platform. Additionally, strain 16FHM2^T was sequenced, using also an Oxford Nanopore MinION device. The assembly of strain 16FHM2^T resulted in a single complete sequence of 5,444,440 bp (Table 2). The assemblies of strains 15FMM2 and 15FMM3 resulted in two draft genome sequences of 5,406,388 and 5,405,281 bp, respectively. The number of coding sequences per genome ranged from 4,904 to 4,906 and the GC contents of the genomes were determined to be 58.9%.

ANIb and dDDH were calculated between the three strains of the proposed novel species, and between strain 16FHM2^T and the type strains of 55 closely-related Pseudomonas species belonging to the P. gessardii and P. fluorescens subgroups of the P. fluorescens group. The ANIb and the dDDH values between the three strains of the proposed novel species were 99.99 and 100%, respectively, confirming that the three strains are very similar and closely related to each other. The ANIb values between the genome sequence of strain 16FHM2^T and those of the type strains of species of the P. fluorescens and P. gessardii subgroups with validly published names ranged from 87.4 to 82.2%, while the dDDH values ranged from 35.3 to 26.8% (Table 3). The analyses confirmed that the most closely related species is P. mucidolens. These data indicate that strain 16FHM2^T represents a novel species within the P. gessardii subgroup of the genus Pseudomonas.

The ANIb values were determined between (all vs. all) genome sequences of type strains of the species of the P. gessardii and P. fluorescens subgroups. The values ranged from 82.02% (between P. mucidolens and P. kitaguniensis) to 96.44% (between P. panacis and P. marginalis). The ANIb values of strain 16FHM2^T compared with the type strains of other species ranged from 87.34% (P. mucidolens LMG 2223^T) and 82.16% (P. kitaguniensis MAFF 212408^T), which suggests that P. mucidolens is the most closely related species. Indeed, the dendrogram also shows the relationship of P. mucidolens to the proposed novel species, within the cluster formed by species of the P. gessardii subgroup (Figure 2).

A total of 381,851 amino acid positions, encoded by 1,361 single-copy shared genes, were used to construct the core genome-based phylogenomic tree, including the genome sequences of the type strains of species of the P. gessardii and P. fluorescens subgroups. The core genome confirms that the species are divided in two well-defined clusters, corresponding to the two subgroups included in the analysis, and that the proposed novel species is a member of the P. gessardii subgroup. Additionally, the analysis confirms that P. mucidolens is the most closely related species (Figure 3).

We detected a new variant of class C β-lactamase in our strains (NCBI Reference Sequence accession number: WP_410017811.1). This variant showed 85.3% amino acid identity (query coverage 100%) to a Class C β-lactamase from P. mucidolens. The search against the Beta-Lactamase DataBase revealed that it belongs to the family PFL and therefore it was designated PFL-7. The most closely related listed variant was PFL-5 (WP_017475175.1), from Pseudomonas sp. PAMC 26793, with 76% of amino acid identity. PFL-7 has a LysR family regulator (WP_410017810.1) encoded upstream of its gene, which shows 92.7% amino acid identity to the regulator from P. mucidolens. Thus, PFL-7 forms an AMP-C-AMP-R system detected in many bacteria (Balasubramanian et al., 2012).

The rpoD sequence-based search of the Nucleotide collection (nt) of NCBI did not detect any additional strain of the proposed novel species. No MAGs of the proposed novel species were found when screening thousands of MAGs, using Protologger. The Branchwater Metagenome Query search did not yield any match with a containment ANI (cANI) score larger than 0.97, which often represents a species-level match, but yielded 41 matches with a cANI score between 0.95 (33 matches) and 0.96 (8 matches) (Supplementary Table 4). These matches probably do not represent members of the same species but might represent related strains of related taxa within the analyzed metagenomic datasets. The samples originated from fresh water (n = 13), wastewater (n = 21) and food samples (leafy greens, chicken and a meat factory) (n = 7).

Pseudomonas imrae (im’rae. N.L. gen. n. imrae, formed from IMR, acronym for Institute of Marine Research, Norway, where the first strains were isolated and studied).

Cells are Gram-negative, rod shaped, non-spore forming, and strictly aerobic, positive for catalase and oxidase. Optimum temperature for growth is 25–30°C, with no growth observed at 4°C and 42°C. The strains are positive for catalase, oxidase, acid phosphatase, esterase, gelatin hydrolysis, l-Arginine dihydrolase activity, and nitrate reduction, while they are negative for urease, α-fucosidase activity, α-mannosidase activity, N-acetyl-β-glucosaminidase activity, β-glucosidase activity, α-glucosidase activity, β-glucuronidase activity, β-galactosidase, α-galactosidase, acetamide utilization, DNAse activity, indole production and esculin hydrolysis. The strains can grow in the presence of 3% NaCl and show variable growth at 4.5 and 5% NaCl with no growth observed above 6% NaCl. The predominant cell fatty acids are C_16:1 ω7c and C_16:0, followed by C_18:1ω7c/12 t/9 t and C_17:0 cyclo, which are present in lower levels.

The 16S rRNA gene sequence is highly similar to P. libanensis, P. synxantha, P. gessardii and P. shahriarae (99.9%), while the partial rpoD sequence shows highest similarity to P. proteolytica (93.4%). Members belong to class Gammaproteobacteria, order Pseudomonadales, family Pseudomonadaceae, genus Pseudomonas, P. fluorescens group, P. gessardii subgroup. The type strain of the species is Pseudomonas imrae strain 16FHM2^T (=CCUG 74779^T = CECT 30571^T); strains15FMM2 (=CCUG 74780) and 15FMM3 (= CCUG 74781) are other representatives. The strains were isolated from gut contents of two specimens of wild Atlantic mackerel (Scomber scombrus) collected in the northern North Sea, ICES region 4.a, in November 2018.