All procedures involving bacterial handling, culture, and MALDI-TOF MS sample preparation were conducted in a Biosafety Level 2 (BSL-2) laboratory in accordance with the institutional biosafety requirements of Chulalongkorn University. The laboratory procedures were reviewed and approved by the Institutional Biosafety Committee of Chulalongkorn University (Approval No. 2031043). Moribund fish used for diagnostic sampling were euthanized at the sampling sites by immersion in clove oil (1 g/L) for approximately 10 min in accordance with the previously described procedure [20]. Internal organs were subsequently collected aseptically for bacteriological examination.

The study was conducted over a three-month period, from August to November 2025. The bacterial isolates investigated in this study originated from diseased freshwater fish sampled from July 2019 to July 2020 in Thailand and Vietnam. Thai isolates were obtained from Samut Sakhon and Nakhon Pathom Provinces, whereas Vietnamese isolates originated from Tra Cu District, Tra Vinh Province; Chau Phu District, An Giang Province; and Thanh Binh District, Dong Thap Province. The isolates were submitted to the Center of Excellence in Fish Infectious Diseases, Faculty of Veterinary Science, Chulalongkorn University, Thailand, for diagnostic examination and further characterization. The six isolates selected for MSP database construction were recovered from diseased snakehead fish collected in Nakhon Pathom Province, Thailand, in 2020.

This laboratory-based diagnostic method development and validation study aimed to construct and evaluate a species-specific MALDI-TOF MS MSP database for A. schubertii. Among 81 bacterial isolates obtained from diseased freshwater fish, 19 were identified as A. schubertii and included in the study. Six representative isolates (CUVETASC03–CUVETASC08) were used for MSP library construction, whereas the complete set of 19 isolates, comprising the six MSP development isolates and 13 additional isolates, was used to assess identification performance. Four closely related non-target Aeromonas species were included as outgroup controls for assessment of spectral discrimination and clustering. Molecular confirmation of the six MSP development isolates was performed using 16S rRNA and gyrB sequence analyses.

Among 81 bacterial isolates obtained from diseased freshwater fish during 2019–2020, 19 were identified as A. schubertii (Table 1). Of these, 17/19 isolates were recovered from striped snakehead (Channa striata), one from Asian seabass (Lates calcarifer), and one from hybrid red tilapia (Oreochromis sp.).

Moribund fish were euthanized at the sampling sites by immersion in clove oil (1 g/L) for approximately 10 min, as previously described [20]. Samples from internal organs, including the liver, spleen, and kidney, were collected aseptically using sterile inoculating loops and streaked onto tryptic soy agar (TSA; Bacto™, Becton, Dickinson and Company [BD], Sparks, MD, USA) supplemented with 5% sheep blood. The inoculated plates were incubated at 28°C for 24 h. Representative colonies with distinct morphologies were selected for preliminary identification using conventional microbiological procedures, including Gram staining and catalase and oxidase tests. The resulting phenotypic characteristics were consistent with classification within the genus Aeromonas. Confirmed isolates were preserved in tryptic soy broth (TSB; Bacto™) supplemented with 20% glycerol and stored at −80°C until further analysis.

Species-level identification of the six isolates selected for MSP development was confirmed by PCR amplification and sequencing of the 16S rRNA and gyrB genes. The 16S rRNA gene was amplified using primers 27F (5′-AGAGTTGATCATGGCTCAG-3′) and 1492R (5′-GGTTCACTTGTTACGACTT-3′) [21], whereas gyrB was amplified using primers gyrB3-F (5′-TCCGGCGGTCTGCACGGCGT-3′) and gyrB14-R (5′-TTGTCCGGGTTGTACTCGTC-3′) [9, 22]. The expected amplicon sizes were approximately 1500 bp for the 16S rRNA gene and 1100 bp for gyrB.

Genomic DNA was extracted from the tested isolates using the boiling method [23]. Amplification of the 16S rRNA gene was performed with an initial denaturation at 94°C for 5 min, followed by 35 cycles of denaturation at 94°C for 1 min, annealing at 56°C for 30 s, and extension at 72°C for 90 s, followed by a final extension at 72°C for 10 min. Amplification of gyrB was performed with an initial denaturation at 94°C for 2 min, followed by 30 cycles of denaturation at 94°C for 30 s, annealing at 56°C for 30 s, and extension at 72°C for 2 min, followed by a final extension at 72°C for 5 min.

PCR products were visualized on 1.5% agarose gels stained with RedSafe™ and purified using the NucleoSpin® Gel and PCR Clean-up Kit (Macherey-Nagel, Germany). Purified amplicons were subjected to Sanger sequencing, and forward and reverse sequences were assembled using BioEdit version 7.1.1. Species identification was confirmed using Basic Local Alignment Search Tool (BLAST) against the National Center for Biotechnology Information (NCBI) database, applying sequence-identity thresholds of ≥99.5% for 16S rRNA and ≥99.0% for gyrB. Sequences were aligned using ClustalW, and maximum-likelihood phylogenetic trees were constructed using MEGA X version 10.2.6 with the Tamura–Nei model and 1,000 bootstrap replicates. Gaps and missing data were handled using the partial-deletion option with a site-coverage cutoff of 95% [24–26].

All A. schubertii isolates were recovered from frozen stocks and subcultured on TSA (Bacto™) at 28°C for 24 h. Well-isolated colonies were subjected to protein extraction using the standard ethanol–formic acid–acetonitrile protocol for MALDI-TOF MS analysis. Briefly, 1 µL of extracted protein was spotted onto a 96-position polished-steel MALDI target plate, overlaid with α-cyano-4-hydroxycinnamic acid (HCCA) matrix, and air-dried at room temperature. Samples were analyzed using a microflex® LRF MALDI-TOF MS instrument (Bruker Daltonics, Bremen, Germany), and the resulting spectra were compared with the MALDI Biotyper® (Bruker Daltonics) reference library [27].

Six representative A. schubertii isolates (CUVETASC03–CUVETASC08) were selected for MSP library cons-truction to incorporate multiple independent isolates and capture intraspecies spectral variability. These isolates originated from diseased snakehead fish collected in Nakhon Pathom Province, Thailand, in 2020. The isolates were recovered from glycerol stocks, subcultured on TSA, and subjected to ethanol–formic acid–acetonitrile protein extraction.

Briefly, bacterial biomass was suspended in 300 µL of high-performance liquid chromatography-grade water and 900 µL of absolute ethanol, centrifuged, air-dried, and subsequently extracted with 30 µL of formic acid and 30 µL of acetonitrile. A 1 µL aliquot of the resulting supernatant was spotted onto a polished-steel MALDI target plate and overlaid with HCCA matrix. Following mass calibration, smoothing, normalization, baseline correction, peak detection, and outlier removal, spectra were analyzed using FlexAnalysis software version 3.0 (Bruker Daltonics) in positive linear mode over a mass range of 2–20 kDa.

For MSP generation, each isolate was analyzed using eight independent spots, with three spectra acquired from each spot, yielding 24 raw spectra per isolate. Spectra showing weak signal intensity, irregular baselines, peak shifting, or poor reproducibility were excluded. A minimum of 20 high-quality spectra per isolate were retained for MSP construction using MBT Compass Explorer® (Bruker Daltonics) [28, 29]. MSPs were accepted when reproducible peak patterns and high intrastrain spectral similarity were consistently observed across technical replicates.

PCA, spectral stacking, and dendrogram analyses were performed using ClinProTools® version 3.0 (Bruker Daltonics), and the resulting MSPs were incorporated into an in-house A. schubertii database [27, 30]. Four closely related non-target species, Aeromonas hydrophila, Aeromonas veronii, Aeromonas jandaei, and Aeromonas caviae, were included as outgroup reference species for discriminatory and clustering analyses. These species were selected because of their phylogenetic relatedness to A. schubertii and their frequent association with freshwater fish diseases [27, 31].

The identification performance of the developed MALDI-TOF MS database was evaluated using 19 A. schubertii isolates, with particular emphasis on identification reliability and repeatability. Discriminatory performance was further assessed using the four closely related non-target Aeromonas species as outgroup controls.

MALDI-TOF MS identification scores were interpreted according to the manufacturer's recommendations. Scores <1.70 were classified as unreliable; scores of 1.70–1.99 indicated reliable genus-level identification; scores of 2.00–2.29 indicated reliable species-level identification; and scores ≥2.30 indicated highly reliable species-level identification. Repeatability was assessed by analyzing each isolate in triplicate and calculating the coefficient of variation (CV) of the corresponding log score values. A CV ≤10% was considered indicative of acceptable repeatability. Discriminatory performance was assessed by comparing protein spectral profiles and evaluating clustering patterns between A. schubertii and the four non-target Aeromonas species.

Log score values were summarized as mean ± standard deviation from triplicate technical replicates. Repeatability was evaluated using the CV of log score values, with values ≤10% considered indicative of acceptable repeatability. Pearson correlation coefficients were calculated between corresponding processed spectral-intensity profiles of technical replicates to evaluate spectral reproducibility. PCA, spectral stacking, and dendrogram analyses were performed using ClinProTools® version 3.0 (Bruker Daltonics).

In addition to the six A. schubertii isolates, four closely related non-target species, Aeromonas hydrophila, Aeromonas veronii, Aeromonas jandaei, and Aeromonas caviae, were included as outgroup controls. All isolates generated analyzable MALDI-TOF MS spectra following spectral preprocessing, including baseline correction, smoothing, normalization, peak detection, and outlier removal. Spectral stacking demonstrated distinct protein mass spectral patterns among A. schubertii and the non-target Aeromonas species (Figure 2).

The A. schubertii isolates CUVETASC03–08 showed a distinct and relatively broad distribution in PCA space, particularly along PC1 and PC2 (Figure 3), indicating considerable intra-species variation among the isolates. Their separation from the reference Aeromonas species suggests that CUVETASC03–08 possess a distinct overall multivariate profile.

Dendrogram analysis further demonstrated clear differentiation between A. schubertii and the non-target Aeromonas species. The A. schubertii isolates clustered tightly within a distinct group, whereas A. hydrophila, A. veronii, A. jandaei, and A. caviae were separated from the A. schubertii cluster. The two major clusters merged at a linkage distance of approximately 0.75, whereas most within-cluster distances were <0.40, indicating comparatively high intraspecies spectral similarity and greater interspecies spectral divergence (Figure 4).

The clear separation of the four non-target Aeromonas species from A. schubertii in both PCA and dendrogram analyses supported the discriminatory capability of the custom MSP database.

PCR amplification of the 16S rRNA and gyrB genes was successfully performed for all six isolates selected for MSP construction (CUVETASC03–CUVETASC08). Clear single amplicons corresponding to the expected sizes of approximately 1500 bp for the 16S rRNA gene and 1100 bp for gyrB were observed, consistent with the positive controls (Figure 5). Subsequent Sanger sequencing followed by BLASTn analysis against the NCBI nucleotide database demonstrated high sequence similarity with reference A. schubertii sequences.

Sequence analysis demonstrated 99.68%–99.80% identity with reference A. schubertii strains across the evaluated 16S rRNA and gyrB sequences (Table 3). These findings supported the species-level taxonomic assignment of the six isolates used for MSP construction. Maximum-likelihood phylogenetic analyses were subsequently performed using approximately 1,500 bp of the 16S rRNA gene and 1100 bp of gyrB.

The maximum-likelihood phylogenetic trees incorporated reference A. schubertii sequences retrieved from GenBank together with selected non-A. schubertii taxa. In both the 16S rRNA- and gyrB-based trees, isolates CUVETASC03–CUVETASC08 clustered with established A. schubertii reference strains and formed a distinct clade, further supporting their taxonomic identification as A. schubertii (Figure 6).

The application of MALDI-TOF MS enabled characterization of A. schubertii isolates based on their protein mass spectral profiles. The generated spectra were highly reproducible among technical replicates, with Pearson correlation coefficients of 0.95–0.97, indicating consistent protein fingerprint profiles. Although gel stack visualization showed similarities among members of the genus Aeromonas, multivariate analyses provided greater discriminatory resolution. PCA demonstrated that A. schubertii isolates formed a distinct cluster separated from the evaluated non-target Aeromonas species, whereas dendrogram analysis similarly showed tight clustering of A. schubertii and separation from Aeromonas hydrophila, Aeromonas veronii, Aeromonas jandaei, and Aeromonas caviae.

The observed differentiation is consistent with previous findings indicating that Aeromonas species can exhibit characteristic protein mass fingerprints that facilitate species-level discrimination by MALDI-TOF MS [29]. The clustering pattern observed for A. schubertii is also consistent with its placement within Aeromonas group 501 reported by Benagli et al. [32], a group characterized by distinct phenotypic and genotypic features [33]. Collectively, the present spectral and clustering results support the capacity of MALDI-TOF MS to discriminate A. schubertii from the closely related species included in this study when appropriate reference spectra are available.

Despite the advantages of MALDI-TOF MS, the commercial Bruker reference database provided incomplete species-level resolution for A. schubertii. Of the 19 isolates examined, 15 (78.95%) were reliably identified at the species level, whereas four (21.05%) achieved only genus-level identification. This finding illustrates an important limitation of reference database-dependent identification: diagnostic performance is influenced by the representation and quality of spectra available for the target organism. Underrepresentation of environmental and aquatic bacterial species in reference libraries may consequently reduce identification confidence for less frequently encountered pathogens such as A. schubertii.

Identification performance may also be affected by methodological differences in sample preparation. Variations in extraction solvents and acid concentrations can influence protein recovery and spectral quality, thereby affecting the resulting identification profiles [34–36]. Organic acid-based extraction procedures preferentially recover abundant low-molecular-weight proteins, including ribosomal and nucleic acid-binding proteins; therefore, differences in extraction chemistry can alter the detectable protein profile [37]. Standardization of both sample preparation procedures and reference spectral libraries is consequently important when MALDI-TOF MS is applied to less well-represented aquatic bacterial species.

Development of the custom A. schubertii MSP database substantially improved identification performance. Following incorporation of species-specific reference spectra, all 19 isolates (100%) were correctly identified at the species level, with log scores ranging from 2.49 to 2.68. In comparison, the commercial database yielded species-level identification for only 78.95% of the isolates, with log scores of 2.01–2.22 among those identified at the species level. Thus, targeted addition of high-quality A. schubertii reference spectra improved both the proportion of isolates identified at the species level and their associated identification confidence.

The custom database was evaluated using 19 A. schubertii isolates recovered from multiple freshwater fish hosts, including snakehead fish, Asian seabass, and hybrid red tilapia. This host representation provided an opportunity to evaluate the database against isolates originating from different fish hosts. However, the MSP library itself was constructed from six isolates obtained from diseased snakehead fish in Nakhon Pathom Province, Thailand. Therefore, the present findings should not be interpreted as demonstrating that the MSP library captures the full host-associated or geographic diversity of A. schubertii. Broader validation using isolates from additional hosts and geographic regions remains necessary.

The improvement achieved after database customization is consistent with previous studies demonstrating that MALDI-TOF MS identification of Aeromonas spp. is strongly dependent on reference database completeness [20, 29]. Difficulties in discriminating fish-associated Aeromonas species using commercial databases have previously been reported [29], whereas customized MSP libraries containing additional reference spectra have improved identification performance [20]. The increase in species-level identification from 78.95% to 100% observed in the present study further supports targeted spectral enrichment as a practical strategy for improving MALDI-TOF MS identification of underrepresented aquatic bacterial pathogens.

The capacity to expand MALDI-TOF MS databases also offers opportunities to incorporate spectra representing broader host, geographic, and strain diversity, potentially increasing their utility for epidemiological surveillance and outbreak investigations [38, 39]. Nevertheless, these potential applications require validation using appropriately diverse isolate collections before broader epidemiological utility can be established.

Molecular analyses provided independent support for the taxonomic identification of the six isolates used for MSP construction. Sequence analysis of the 16S rRNA and gyrB genes demonstrated 99.68%–99.80% identity with reference A. schubertii strains, and maximum-likelihood phylogenetic analyses showed that all six isolates clustered with established A. schubertii reference strains. These findings supported the species-level assignment of the isolates used to generate the custom MSP database.

Although 16S rRNA gene sequencing is widely used for bacterial identification and phylogenetic characterization, its discriminatory capacity may be limited among closely related Aeromonas species, including A. veronii, Aeromonas ichthiosmia, and Aeromonas simiae. In contrast, housekeeping genes such as gyrB can provide greater discriminatory resolution because of their higher sequence variability while retaining conserved functional roles [40]. Furthermore, multilocus sequence analysis based on concatenated housekeeping genes has been reported to provide improved taxonomic and phylogenetic resolution within the genus Aeromonas [41–43]. The combined use of 16S rRNA and gyrB analyses in the present study therefore provided complementary molecular evidence supporting the taxonomic identity of the isolates used for MSP construction.

The findings indicate that MALDI-TOF MS, when supported by an appropriately curated species-specific MSP database, can provide rapid and reliable identification of A. schubertii following bacterial isolation. This capability may be particularly relevant to surveillance of white nodular disease, which has been associated with A. schubertii infection in freshwater fish. Because A. schubertii has also been reported in other aquatic hosts and opportunistic human infections, improved species-level identification may have relevance beyond freshwater fish diagnostics.

Combining MALDI-TOF MS protein fingerprinting with molecular confirmation also provides complementary approaches to bacterial characterization. Molecular methods can support taxonomic confirmation, whereas MALDI-TOF MS can facilitate rapid processing of cultured bacterial isolates once an appropriate reference database has been established. This combination may therefore strengthen laboratory identification and epidemiological surveillance, particularly for aquatic pathogens that are poorly represented in commercial spectral databases.

The customized MALDI-TOF MS approach may also offer practical workflow advantages compared with sequencing-based identification because MALDI-TOF MS enables rapid analysis of cultured colonies and supports high-throughput processing. However, this study did not formally compare turnaround time, workload, and cost. Consequently, the current data cannot quantify the relative economic and operational advantages of the custom MALDI-TOF MS approach over molecular identification methods. Future studies should directly compare diagnostic turnaround time, laboratory workload, and cost-effectiveness between these approaches.

Beyond its immediate application to A. schubertii, the developed MSP database may provide a foundation for expansion of reference spectra for aquatic Aeromonas species. Continued addition of well-characterized isolates could support the development of a broader aquatic Aeromonas MSP reference network. Future technological advances may also permit integration of such databases with portable or field-oriented MALDI-TOF MS platforms; however, the feasibility and diagnostic performance of such applications require dedicated validation.

Several limitations of the present study should be considered when interpreting the findings. First, the MSP database was constructed using only six A. schubertii isolates, all originating from diseased snakehead fish collected in Nakhon Pathom Province, Thailand. Although the database was subsequently evaluated using 19 isolates originating from Thailand and Vietnam and from multiple fish hosts, this collection may not adequately represent the broader genetic and proteomic diversity of A. schubertii across different geographic regions, hosts, and production systems. Second, MALDI-TOF MS performance can be influenced by sample preparation procedures, instrument settings, calibration, and laboratory-specific workflows, which may affect the transferability and reproducibility of the database across laboratories.

Third, validation was performed within the study setting and did not include independent external or multicenter laboratory validation. Although the discriminatory analyses included four closely related non-target Aeromonas species, a broader panel of phylogenetically and diagnostically relevant bacterial species would provide a more comprehensive assessment of specificity. Fourth, whole-genome sequencing was not performed; therefore, direct comparison of MALDI-TOF MS identification with genome-based taxonomic approaches was beyond the scope of the study. Finally, formal analyses of diagnostic turnaround time and cost-effectiveness were not undertaken.

Future studies should expand the MSP database using a larger collection of molecularly and genomically characterized A. schubertii isolates representing diverse hosts, geographic regions, and epidemiological settings. Multicenter validation should assess interlaboratory reproducibility and database transferability across different MALDI-TOF MS workflows. Inclusion of a broader range of non-target Aeromonas species and other phenotypically similar aquatic bacteria would further strengthen assessment of diagnostic specificity. Integration with whole-genome sequencing and multilocus sequence approaches may also help characterize strain diversity and determine whether spectral variation corresponds to genetically distinct lineages. In addition, prospective diagnostic studies comparing MALDI-TOF MS with conventional and sequencing-based methods in terms of accuracy, turnaround time, workload, and cost would help establish its practical value for routine aquaculture diagnostics.