Use of MALDI-TOF mass spectrometry for identification of Weissella tructae

TL;DR

This study shows that MALDI-TOF mass spectrometry can quickly and accurately identify the fish pathogen Weissella tructae, which is important for aquaculture diagnostics.

Contribution

The study demonstrates that MALDI-TOF MS can distinguish W. tructae from W. ceti, which traditional methods like 16S rRNA sequencing cannot.

Findings

MALDI-TOF MS and duplex PCR perfectly identified W. tructae and W. ceti.

16S rRNA sequencing failed to differentiate between W. tructae and W. ceti due to high sequence similarity.

PCA analysis revealed species-specific peaks for W. tructae at 3700.11, 3720.75, and 7406.47 m/z.

Abstract

Weissella ceti, the etiological agent of hemorrhagic septicemia in rainbow trout (Oncorhynchus mykiss), was recently reclassified as a new species, Weissella tructae. This taxonomic update underscores the need for diagnostic methods capable of accurately identifying W. tructae and distinguishing it from W. ceti in a rapid, reliable, and cost-effective manner. In this study, we evaluated matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) for the identification of W. tructae. A total of 60 isolates, including 58 Weissella spp. from rainbow trout, the W. tructae type strain WS08ᵀ, and the W. ceti type strain CECT 7719ᵀ, were identified using MALDI-TOF MS, duplex PCR, and 16S rRNA gene sequencing, and the results were compared. A MALDI-TOF dendrogram and a 16S rRNA phylogenetic tree were constructed, and representative spectra of W. tructae and W. ceti were analyzed using principal component analysis (PCA). MALDI-TOF MS and duplex PCR showed perfect agreement, whereas 16S rRNA gene sequencing failed to discriminate between the two species due to their high sequence similarity. PCA further supported the discriminatory capability of MALDI-TOF MS, revealing species-specific peaks for W. tructae (3700.11, 3720.75, and 7406.47 m/z). Collectively, our findings demonstrate that MALDI-TOF MS is a reliable and rapid method for identifying W. tructae and distinguishing it from W. ceti, representing a valuable diagnostic tool for aquaculture applications.

Figures

Phylogenetic tree of *Weissella* spp. based on partial 16S rRNA gene sequences. The tree was constructed using Neighbor Joining method and Kimura 2-parameter model. Bootstrap values (expressed as percentages of 1000 replicates) are shown at each node. The scale bar represents 0.01 substitutions per nucleotide position

MALDI Biotyper dendrogram of *Weissella* spp. based on MSP data generated by MALDI-TOF. *Enterococcus hermanniensis* was included as an outgroup

PCA of *W. ceti* CECT 7719^T^ and *W. tructae* WS08^T^, WS74, WS105, and W-1. Each dot represents one of the 24 individual spectra used to construct each MSPFig. 4Influence of individual m/z peaks on PC1 (Load1) and PC2 (Load2). Load1 represents the contribution of each peak to PC1, and Load2 represents the contribution to PC2. Peaks with higher negative Load1 values were strongly associated with *W. tructae*, while peaks with higher positive Load1 values were more strongly associated with *W. ceti*

Full text

Introduction

Weissellosis is an emergent disease affecting cultured rainbow trout (Oncorhynchus mykiss) worldwide, with reported outbreaks in China [1], Brazil [2], USA [3], Japan [4], Mexico [5], Colombia [6], Peru [7] and Vietnam [8]. The disease manifests as hemorrhagic septicemia, characterized by clinical signs such as anorexia, lethargy, exophthalmia, ascites, and hemorrhages throughout the body, particularly on the mouth, oral cavity, tongue, and eyes, leading to moderate to high mortality during outbreaks [1–7]. However, these clinical signs are not pathogen-specific and may also be triggered by other bacteria such as Aeromonas hydrophila, Pseudomonas fluorescens, Yersinia ruckeri, and Lactococcus garvieae, highlighting the need for laboratory-based diagnosis [9–13] [13].

Weissella tructae is the species name recently assigned to the bacterium responsible for weissellosis in rainbow trout (Oncorhynchus mykiss) [14]. This microorganism was previously classified as Weissella ceti and was thought to be the same species isolated from beaked whale (Mesoplodon bidens) [15]. Currently, molecular approaches like duplex PCR and quantitative PCR are the primary methods used to diagnose weissellosis [13].A polyphasic characterization approach helped elucidate the differences between W. ceti and W. tructae, thereby confirming that the trout-associated bacterium represents a distinct species [14]. However, polyphasic characterization is not suitable for routine diagnostics due to its high cost, labor-intensive procedures, and time-consuming nature.

Matrix Assisted Laser Desorption/Ionization – Time of Flight (MALDI-TOF) mass spectrometry has transformed clinical microbiology by providing rapid and accurate bacterial identification [16]. Despite its advantages, the application of MALDI-TOF in aquaculture remains limited, primarily due to the absence of Main Spectra Profiles (MSPs) for key fish pathogens in commercial databases [17]. Nonetheless, several studies have demonstrated the successful use of MALDI-TOF for identification of aquaculture-relevant microorganisms including Flavobacterium psychrophilum [18], Vibrio spp [19]., Aeromonas [20], Lactococcus garvieae, and Streptococcus spp [21]. through the incorporation of MSPs into local databases.

Given the growing impact of weissellosis in aquaculture, this study aimed to evaluate the application of MALDI-TOF mass spectrometry for the rapid, reliable, and cost-effective identification of W. tructae, the bacterium responsible for this disease.

Materials and methods

Isolates and culture conditions

A total of 60 isolates recovered from cultured rainbow trout, and one isolate recovered from beaked-whale were thawed from storage temperature (−80 °C) to room temperature and streaked on Man, Rogosa & Sharpe (MRS) agar plates, which were then incubated at 28 °C for 48 h to grow the bacteria.

The 60 rainbow trout isolates consisted of 59 Brazilian isolates, obtained from the culture collection of the Laboratory of Aquatic Animal Diseases (AQUAVET), including the type strain of W. tructae WS08^T^ (CBMAI 2730), and one Mexican isolate (W-1) [5] obtained from the Centro de Investigación y Estudios Avanzados en Salud Animal at the Universidad Autónoma del Estado de México. The outbreaks from which these isolates were recovered are comprehensively detailed in previous articles [2, 5, 22]; however, key metadata are summarized in Table 1. The Weissella ceti type strain CECT 7719^T^, originally isolated from a beaked whale, was obtained from the Laboratorio de Patología de Organismos Acuáticos y Biotecnología Acuícola at the Universidad Andrés Bello, Chile, and was used as a reference strain for the W. ceti species.Table 1Weissella spp. Isolates used in this studyNumberIsolatesStateCountryYear of isolationReference1WS01Rio de JaneiroBrazil2008 Figueiredo et al., 2012 [2]2WS02Rio de JaneiroBrazil2008 Figueiredo et al., 2012 [2]3WS03Rio de JaneiroBrazil2008 Figueiredo et al., 2012 [2]4WS05Rio de JaneiroBrazil2008 Figueiredo et al., 2012 [2]5WS07Rio de JaneiroBrazil2008 Figueiredo et al., 2012 [2]6WS11Rio de JaneiroBrazil2008 Figueiredo et al., 2012 [2]7WS12Rio de JaneiroBrazil2008 Figueiredo et al., 2012 [2]8WS15Rio de JaneiroBrazil2008 Figueiredo et al., 2012 [2]9WS20Rio de JaneiroBrazil2008 Figueiredo et al., 2012 [2]10WS23Rio de JaneiroBrazil2008 Figueiredo et al., 2012 [2]11WS32São PauloBrazil2009 Figueiredo et al., 2012 [2]12WS35São PauloBrazil2009 Figueiredo et al., 2012 [2]13WS38São PauloBrazil2009 Figueiredo et al., 2012 [2]14WS41São PauloBrazil2009 Figueiredo et al., 2012 [2]15WS42São PauloBrazil2009 Figueiredo et al., 2012 [2]16WS44São PauloBrazil2009 Figueiredo et al., 2012 [2]17WS46São PauloBrazil2009 Figueiredo et al., 2012 [2]18WS50São PauloBrazil2009 Figueiredo et al., 2012 [2]19WS53São PauloBrazil2009 Figueiredo et al., 2012 [2]20WS54São PauloBrazil2009 Figueiredo et al., 2012 [2]21WS56São PauloBrazil2009 Figueiredo et al., 2012 [2]22WS58São PauloBrazil2009 Figueiredo et al., 2012 [2]23WS60São PauloBrazil2009 Figueiredo et al., 2012 [2]24WS65São PauloBrazil2009 Figueiredo et al., 2012 [2]25WS66São PauloBrazil2009 Figueiredo et al., 2012 [2]26WS71São PauloBrazil2009 Figueiredo et al., 2012 [2]27WS72São PauloBrazil2009 Figueiredo et al., 2012 [2]28WS74Minas GeraisBrazil2009 Figueiredo et al., 2012 [2]29WS75Minas GeraisBrazil2009 Figueiredo et al., 2012 [2]30WS76Minas GeraisBrazil2009 Figueiredo et al., 2012 [2]31WS78Minas GeraisBrazil2010 Costa et al., 2014 [22]32WS79Minas GeraisBrazil2010This study33WS81Minas GeraisBrazil2010This study34WS83Minas GeraisBrazil2010This study35WS85Minas GeraisBrazil2010This study36WS86Minas GeraisBrazil2010This study37WS87Minas GeraisBrazil2010This study38WS88Minas GeraisBrazil2010 Costa et al., 2014 [22]39WS89Minas GeraisBrazil2010This study40WS91Minas GeraisBrazil2010 Costa et al., 2014 [22]41WS92Minas GeraisBrazil2010This study42WS93Minas GeraisBrazil2010This study43WS94Minas GeraisBrazil2010This study44WS98Minas GeraisBrazil2010 Costa et al., 2014 [22]45WS99Minas GeraisBrazil2010This study46WS100Minas GeraisBrazil2010This study47WS101Minas GeraisBrazil2010 Costa et al., 2014 [22]48WS102Minas GeraisBrazil2010 Costa et al., 2014 [22]49WS103Minas GeraisBrazil2010This study50WS104Minas GeraisBrazil2010This study51WS105Minas GeraisBrazil2010 Costa et al., 2014 [22]52WS106Minas GeraisBrazil2010This study53WS107Minas GeraisBrazil2010This study54WS109Minas GeraisBrazil2012This study55WS110Minas GeraisBrazil2012This study56WS111Minas GeraisBrazil2012 Costa et al., 2014 [22]57WS112Minas GeraisBrazil2012This study58WS113Minas GeraisBrazil2012This study59W-1-Mexico2015 Castrejón-Nájera et al., 2018 [5]60WS08^T^Rio de JaneiroBrazil2008 Figueiredo et al., 2012 [2]61CECT 7719^T^-Spain2011 Vela et al., 2011 [15]

MALDI-TOF real time identification (RTC)

The MSPs from W. ceti CECT 7719^T^, W. tructae WS08^T^, and other three W. tructae (WS74, WS105 and W-1) are not available in Bruker Daltonics’ public database. However, they have been previously created by Pereira et al. (2022) and included in AQUAVET’s in-house database for comparison in order to obtain the RTC score for each strain.

The RTC of the 60 isolates was prepared according to Assis et al. (2017). Briefly, after 48 h of growth on MRS agar plates at 28 °C, a single colony from each bacterial strain was spotted onto a steel target plate using a sterilized toothpick. Then, 1 µL of 70% formic acid was added to each spot and allowed air-dry. Subsequently, 1 µL of α-cyano-4-hydroxycinnamic acid (HCCA) matrix solution (Bruker Daltonics, Bremen, Germany) was applied, and the spots were again air-dried. The steel target plate was then inserted into the MicroFlex LT instrument (Bruker Daltonics), and spectra were acquired using FlexControl software (Bruker Daltonics), which detects ionized bacterial peptides based on their m/z ratio. Prior to measurements, instrument calibration was performed using the Escherichia coli DH5 alpha bacterial standard (Bruker Daltonics).

The scores were calculated using MALDI Biotyper version 3.1 (Bruker Daltonics). According to the manufacturer, an RTC score < 1.700 indicates no reliable identification, a score ≥ 1.700 and < 2.000 indicates genus-level identification, and a score ≥ 2.000 indicates species-level identification.

Identification by 16S rRNA sequencing and phylogeny

The colonies were transferred into Eppendorf tubes containing 20 µL of elution buffer and subjected to thermal lysis at 95 °C for 15 min. The cell extract was used as template for amplification of the 16S rRNA gene via PCR using a Veriti thermal cycler (Applied Biosystems, USA). Universal primers C70 (Forward: 5’AGAGTTTGATYMTGGC-3’) and B37 (Reverse: 5’-TACGGYTACCTTGTTACGA-3’) were used, following the method described by Fox et al. [39].) The PCR conditions consisted of three stages: (i) an initial denaturation at 95 °C for 15 min, (ii) followed by 35 amplification cycles, each comprising denaturation at 95 °C for 1 min, annealing at 58 °C for 45 s, and extension at 72 °C for 1 min, (iii) final elongation at 72 °C for 15 min. PCR amplification was verified by agarose gel electrophoresis. PCR products were purified using AMPure XP (Beckman Coulter, USA).

Sequencing began with a second amplification step using the BigDye ™ Terminator Cycle Sequencing Kit (Applied Biosystems, USA) on the Veriti thermal cycler. The sequencing PCR conditions were as follow: initial denaturation at 96 °C for 1 min, followed by 30 cycles of 96 °C for 15 s, 50 °C for 15 s, and 60 °C for 4 min. Following this PCR, samples were precipitated using 65% isopropanol and 60% ethanol. Sequencing was then performed on an ABI 3500 genetic analyzer (Applied Biosystems, USA). Forward and reverse reads were obtained separately, manually trimmed, and assembled into contigs using BioEdit software. These contigs were compared against sequences in the National Center for Biotechnology Information (NCBI) database using the Basic Local Alignment Search Tool (BLAST) algorithm. Isolate were classified based on the highest identity percentage to known sequences. Species-level identification was assigned according to the best match. In cases where the highest identity was equally shared among different Weissella species, the isolate was classified as Weissella spp. Additionally, since some sequences in the NCBI database submitted prior to formal designation of W. tructae may not have been updated, those labeled as W. ceti but isolated from O. mykiss were considered W. tructae, following the findings of Pereira et al. (2022). Unidentified of ambiguous sequences in the database were disregarded. Moreover, similarity to the type strains W. tructae WS08 ^T^ and W. ceti CECT 7719^T^ was manually checked in the ranking list of BLAST results.

A 16S rRNA phylogenetic analysis was performed to provide a broader perspective of Weissella species differentiation. The 16S rRNA gene sequences of W. tructae WS08^T^ and W. ceti CECT 7719^T^ were included to represent their respective nodes. Eight Weissella sequences obtained from diverse rainbow trout outbreaks around the world (NC36/NZ ANCA01000001.1; IMP-BG-B054/MN860274.2; MHW1608-01/KY697255.1; W-1/MH091070.1; HLB2_K_Columbia/MK968274.1; TB02/CP194562.1; WS74/CP009223.1; WS105/CP009224.1) as well as representative 16S rRNA gene sequences from 23 species of Weissella (W. bombi/NR 136437.1, W. ceti/FN813251.2, W. cibaria/MG982483.1, W. coleopterorum/NR 180654.1, W. confusa/MG786556.1, W. diestrammenae/NR 118386.1, W. halotolerans/NR 040812.1, W. hanii/AY040669.1, W. hellenica/NR 113775.1, W. jogaejeotgali/NR 145896.1, W. kandleri/NR 044659.2, W. koreensis/NR 029041.1, W. minor/NR 040809.1, W. muntiaci/NR 170492.1, W. oryzae/NR 114312.1, W. paramesenteroides/NR 104568.1, W. sagaensis/NR 175448.1, W. salipiscis/AB257595.1, W. soli/NR 025642.1, W. thailandensis/NR 040822.1, W. tructae/NZ_CP007588.1, W. uvarum/NR 134229.1 and W. viridescens/NR 040813.1) were retrieved from NCBI database and used for the phylogenetic analysis based on 16S rRNA gene sequencing data. An Enterococcus plantarum sequence (NR 118050.1) was used as outgroup.

Identification by NC36 strain-specific duplex PCR

Isolates identification was also performed using NC36 strain-specific duplex PCR [13]. This method was originally developed to identify trout weissellosis occurring in the United States by amplification a 500 bp fragment of the putative platelet-associated adhesion locus of NC36 strain, as well as a 725 bp fragment of a genus‐specific 16S rRNA sequence. Duplex PCR was carried out using the primers WeisF (5’-TCTAGGAGCGAATAAGAACG-3’), WeisR (5’-CTGTTGATGCAGAAATAGCA-3’), WeigF (5′‐CGTGGGAAACCTACCTCTTA‐3′) and WeigR (5′‐CCCTCAAACATCTAGCAC‐3′) as recommended by Snyder [13]. Amplicons were visualized by capillary gel electrophoresis using a standard capillaries on the QIAxcel (QIAGEN, Venlo, Netherlands). Isolates exhibiting amplification bands at approximately 500 bp and 725 bp were classified as W. tructae, while those displaying only the 725 bp band were identified as Weissella spp.

MALDI-TOF dendrogram and principal component analysis (PCA)

MALDI-TOF MS data was used to create a dendrogram using MALDI Biotyper software version 3.1.66 (Bruker Daltonics). The analysis included previously generated MSPs of W. tructae WS08^T^, W-1, WS74 and WS105 and W. ceti CECT 7719^T^ [14], along with all available Weissella spp. MSPs in the Bruker database (W. viridescens,* W. minor*,* W. confusa*,* W. halotolerans* and W. hellenica). Enterococcus hermanniensis was used as an outgroup.

To further assess discriminatory capacity of MALDI-TOF MS spectra and identify species-specific peaks, a PCA was performed using ClinProTools software version 3.0 (Bruker Daltonics). The 24 individual spectra used to generate MSPs for W. ceti CECT 7719^T^ and W. tructae WS08^T^, WS74, WS105 and W-1 were imported into ClinProTools software version 3.0 (Bruker Daltonics), where PCA plot were generated using default parameters.

Statistical analysis

To assess the level of agreement between MALDI-TOF, 16S rRNA gene sequencing, and duplex PCR for the identification of W. tructae, a statistical congruence analysis was performed using Cohen’s Kappa coefficient (κ). Calculations were performed in R software, (version 4.5.1) with irr package (version 0.84.1) [23, 24].

Results

The complete identification results obtained by MALDI-TOF, 16S rRNA gene sequencing, and duplex PCR are presented in Table 2. All 59 isolates recovered from O. mykiss were identified as W. tructae by MALDI-TOF, with identification scores ranging from 2001 to 2580. In costrast, the W. ceti CECT 7719^T^ was correctly identified as W. ceti with a score of 2.223. When the 59 isolates were compared to the MSP of the W. ceti CECT 7719^T^, identification scores ranged from 664 to 1703.Table 2. Identification of Weissella spp. Isolates by 16S rRNA gene sequencing, MALDI-TOF, and duplex PCRIsolate16S rRNA gene sequencingMALDI-TOFDuplex PCRBest identificationSimilarity to CECT 7719^T^ (%)Similarity to WS08^T^Best identificationScore to CECT 7719^T^SpeciesSimilarity (%)SpeciesScoreWS01Weissella tructae98.9898.4398.7W. tructae24511349*+/+WS02Weissella tructae99.4598.999.18W. tructae25801454+/+WS03Weissella tructae99.7999.2599.52W. tructae23891498+/+WS05Weissella tructae98.2997.7498.02W. tructae20291408+/+WS07Weissella spp.99.5899.5899.38W. tructae25731501+/+WS11Weissella spp.99.0298.7899.02W. tructae23131479+/+WS12Weissella spp.98.598.598.34W. tructae23931502+/+WS15Weissella tructae98.1297.8898.12W. tructae25671373+/+WS20Weissella tructae98.5398.1698.53W. tructae20051552+/+WS23Weissella spp.10010099.75W. tructae23491485+/+WS32Weissella tructae99.7599.599.75W. tructae25261219+/+WS35Weissella tructae98.0997.6897.82W. tructae23281575+/+WS38Weissella tructae99.0498.3299.04W. tructae20011196+/+WS41Weissella tructae10099.75100W. tructae24941387+/+WS42Weissella tructae10099.76100W. tructae25031485+/+WS44Weissella tructae10099.76100W. tructae24701387+/+WS46Weissella tructae10099.74100W. tructae24581182+/+WS50Weissella tructae10099.29100W. tructae20601503+/+WS53Weissella spp.98.7798.7798.59W. tructae20681372+/+WS54Weissella tructae10099.75100W. tructae21321481+/+WS56Weissella tructae99.0898.3999.08W. tructae23711311+/+WS58Weissella tructae99.3198.3999.31W. tructae21741118+/+WS60Weissella tructae98.2898.0398.28W. tructae2454908+/+WS65Weissella tructae97.7497.0397.74W. tructae23181516+/+WS66Weissella spp.99.8399.8399.66W. tructae25401158+/+WS71Weissella spp.97.7497.7497.49W. tructae2522942+/+WS72Weissella spp.98.7298.7298.5W. tructae24851370+/+WS74Weissella tructae98.1497.6697.87W. tructae24111438+/+WS75Weissella tructae99.7699.5299.76W. tructae25521127+/+WS76Weissella tructae97.8697.6297.86W. tructae24321380+/+WS78Weissella tructae97.1896.9297.18W. tructae25361266+/+WS79Weissella spp.97.6897.6897.5W. tructae24081336+/+WS81Weissella tructae98.4897.9298.20W. tructae23951095+/+WS83Weissella tructae97.6197.3797.61W. tructae25121210+/+WS85Weissella tructae97.3397.0697.33W. tructae21851383+/+WS86Weissella tructae97.3296.3497.18W. tructae2536973+/+WS87Weissella tructae98.0297.7798.02W. tructae22071355+/+WS88Weissella tructae98.2898.0498.28W. tructae24711070+/+WS89Weissella tructae97.39797.3W. tructae2514837+/+WS91Weissella tructae97.2997.0297.29W. tructae23991354+/+WS92Weissella tructae98.1397.6698.13W. tructae22071368+/+WS93Weissella tructae99.8398.699.83W. tructae24601485+/+WS94Weissella spp.98.8798.8798.59W. tructae2517973+/+WS98Weissella tructae97.5696.2397.56W. tructae24701264+/+WS99Weissella tructae97.8797.3397.60W. tructae25401071+/+WS100Weissella tructae99.898.6399.8W. tructae25331099+/+WS101Weissella tructae97.5797.2897.55W. tructae24241255+/+WS102Weissella tructae98.7698.598.75W. tructae25921091+/+WS103Weissella tructae97.3797.1497.37W. tructae24611121+/+WS104Weissella spp.99.3199.3199.14W. tructae24481319+/+WS105Weissella tructae97.8697.3997.63W. tructae2483744+/+WS106Weissella tructae97.3697.1297.36W. tructae24701402+/+WS107Weissella tructae98.3998.0698.39W. tructae2557952+/+WS109Weissella tructae99.5898.3499.58W. tructae22821487+/+WS110Weissella spp.99.5899.5899.37W. tructae22881540+/+WS111Weissella tructae97.4596.8397.45W. tructae21691665+/+WS112Weissella spp.98.398.398.09W. tructae21361532+/+WS113Weissella tructae97.897.5697.8W. tructae2415664+/+W-1Weissella tructae99.7299.1599.43W. tructae23631703+/+CECT 7719^T^Weissella ceti99.9399.9399.08W. ceti22232223+/-*

+/+ indicates amplification of both the Weissella genus-specific and Weissella tructae NC36-specific sequences; +/- indicates amplification of the Weissella genus-specific sequence, while the Weissella tructae NC36-specific primers did not yield a product.

The 16S rRNA gene sequences had an average length of 609 bp and 100% query coverage. Among the 59 isolates recovered from rainbow trout, 43 were best identified as W. tructae, with sequence similarities ranging from 97.18% to 100%. The remaining 16 isolates could not be assigned to the species level due to identical highest similarities scores to both W. tructae and W. ceti. Similarities to the W. tructae type strain ranged from 97.18% to 100%, while similarities to the W. ceti type strain ranged from 96.23% to 100%. The isolate recovered from M. bidens was best identified as W. ceti, showing 99.93% similarity.

The duplex PCR method successfully amplified the Weissella genus-specific target in all 60 isolates. The W. tructae NC36-specific PCR amplicon (approximately 525 bp) was obtained only from isolates recovered from O. mykiss. No amplification with the W. tructae NC36-specific primers was observed for the isolate identified as W. ceti from M. bidens.

The Cohen’s kappa analysis revealed no agreement between MALDI-TOF and 16S rRNA gene sequencing (κ = 0) for identification of W. tructae, as well as between 16S rRNA gene sequencing and duplex PCR (κ = 0). In contrast, a perfect agreement was observed between MALDI-TOF and duplex PCR results (κ = 1).

The phylogenetic analysis based on partial 16S rRNA gene sequences (Fig. 1) showed that all eight representative isolates recovered from rainbow trout outbreaks clustered within the same branch as the W. tructae WS08ᵀ. In contrast, the W. ceti CECT 7719ᵀ was placed outside the W. tructae cluster, grouping instead with other Weissella spp. representatives.Fig. 1. Phylogenetic tree of Weissella spp. based on partial 16S rRNA gene sequences. The tree was constructed using Neighbor Joining method and Kimura 2-parameter model. Bootstrap values (expressed as percentages of 1000 replicates) are shown at each node. The scale bar represents 0.01 substitutions per nucleotide position

The MALDI-TOF dendrogram, shown in Fig. 2, demonstrates a clustering pattern of the W. tructae isolates that closely mirrors the topology observed in the 16S rRNA gene phylogenetic tree. The W. tructae isolates formed a highly homogeneous group, whereas the W. ceti isolate appears as the closest related taxon among the analyzed samples.Fig. 2MALDI Biotyper dendrogram of Weissella spp. based on MSP data generated by MALDI-TOF. Enterococcus hermanniensis was included as an outgroup

PCA showed that PC1 accounted for approximately 70% of the total variation in the MALDI-TOF spectra of W. tructae and W. ceti, while PC2 explained around 13%. The remaining components each contributed 5% or less to the total variation. PC1 was strongly associated with species differentiation, as W. tructae spectra were tightly clustered at the opposite end of the PC1 axis from W. ceti spectra (Fig. 3). The peak at 2334.38 m/z, which was exclusively present in W. ceti, was the most influential variable contributing to PC1 separation. Combined PC1 and PC2 further revealed that the peaks at 3700.11, 3720.75, and 7406.47 m/z, found only in W. tructae, also played a role in species discrimination (Fig. 4).Fig. 3PCA of W. ceti CECT 7719^T^ and W. tructae WS08^T^, WS74, WS105, and W-1. Each dot represents one of the 24 individual spectra used to construct each MSPFig. 4Influence of individual m/z peaks on PC1 (Load1) and PC2 (Load2). Load1 represents the contribution of each peak to PC1, and Load2 represents the contribution to PC2. Peaks with higher negative Load1 values were strongly associated with W. tructae, while peaks with higher positive Load1 values were more strongly associated with W. ceti

Discussion

Rapid and accurate identification tools are essential for detecting pathogenic microorganisms, including W. tructae, as early diagnosis is critical for guiding treatment strategies and limiting the spread of emerging pathogens. Notably, W. tructae was only recently described in our previous study, where a comparative polyphasic approach was applied to characterize Weissella isolates from rainbow trout and distinguish them from W. ceti, originally isolated from beaked whales [14]. Given the recent recognition of W. tructae and the urgent need for rapid diagnostic methods, MALDI-TOF MS emerges as a promising solution. This technology has revolutionized microbial diagnostics in both human and veterinary medicine by providing rapid, reliable, and cost-effective identification [25]. However, its use for identifying W. tructae had not yet been validated, until now.

In this study, MALDI-TOF and duplex PCR showed perfect agreement, while 16S rRNA gene sequencing demonstrated no congruence with both methods. This discrepancy can be attributed to fundamental differences in the principles behind each approach. The duplex PCR assay target a fragment of the putative platelet-associated adhesion locus from strain NC36, which is absent in W. ceti [13]. This target has shown strong discriminatory power and enables reliable identification when analyzing W. tructae isolates from Brazil, Mexico, and the United States [13]. In contrast, MALDI-TOF was highly effective, classifying bacteria by comparing their protein mass spectra, primarily composed of ribosomal proteins, to a reference database [26, 27]. By capturing unique protein expression profiles, MALDI-TOF can resolve closely related species with high accuracy [28].

On the other hand, 16S rRNA gene sequencing, although widely used in bacterial taxonomy, relies on highly conserved gene. This limits its resolution when distinguishing between closely related taxa such W. tructae and W. ceti [14]. Our finding, based on a more representative set of isolates, are consistent with previous results using only WS08ᵀ, W-1, and CECT 7719ᵀ [14]. Indeed, several studies have shown that protein-based profiling can outperform 16S rRNA gene sequencing in differentiating closely related species [28–31]. Moreover, the widely accepted ≥ 97% sequence similarity threshold for species-level identification [32–34] proved insufficient in this case, as all isolates shared sequence similarities above this cut-off with both W. tructae and W. ceti type strains.

Isolate CECT 7719^T^ was excluded from the Cohen’s kappa analysis to ensure the assessment of diagnostic agreement focused exclusively on W. tructae. To perform a similarly rigorous analysis for W. ceti, a more extensive and diverse set of isolates would be required to enhance the reliability of conclusions and better reflect the species’ diversity. The inclusion of isolates from multiple sources or geographic origins would help capture natural intraspecific variation, leading to more robust findings. However, to our knowledge, W. ceti has only been isolated from M. bidens in a single reported study [15].

Both the 16S rRNA gene phylogenetic tree and the MALDI-TOF phyloproteomic dendrogram showed that Weissella isolates from rainbow trout clustered closely with the W. tructae WS08ᵀ. A similar correspondence between phylogenetic and proteomic clustering was previously reported for Leuconostoc mesenteroides [35]. Interestingly, no intraspecific variation was observed within the W. tructae cluster in MALDI-TOF dendrogram, highlighting the consistency of protein expression profiles among these strains. Several strains originally deposited in GenBank as W. ceti, including WS74, WS105 [36], NC36 [37], IMP-BG-B054 [7], MHW1608-01 [4], and HLB2_K_Colombia [6] were isolated from diseased rainbow trout and now group within the same clade as WS08ᵀ and W-1. Based on these findings, we recommend their reclassification as W. tructae [14].

The PCA analysis further demonstrated the power of MALDI-TOF mass spectra to discriminating between W. tructae and W. ceti. PC1 alone explained ~ 70% of the total variance and effectively separated the species, with W. tructae and W. ceti spectra clustering at opposite ends of the PC1 axis. Similar application of PCA using MALDI-TOF data have successfully distinguished other Weissella species, such as W. cibaria and W. confusa [38]. In our study, the discriminatory capacity of PC1 was mainly driven by the peak at 2334.38 m/z, present exclusively in W. ceti, as well as peaks at 3700.11, 3720.75, and 7406.47 m/z, which were unique to W. tructae. These consistent, species-specific peaks reinforce the reproducibility of proteomic differences and support MALDI-TOF not only as a diagnostic tool but also as a platform for identifying reliable biomarkers with epidemiological value in aquaculture.

Conclusion

MALDI-TOF MS was demonstrated to be a reliable method for the identification of W. tructae, with the capacity to differentiate it from W. ceti. In addition, the detection of species-specific peaks 3700.11 m/z, 3720.75 m/z and 7406.47 m/z highlights their potential use as biomarkers for W. tructae, facilitating its rapid and accurate identification in diagnostic and epidemiological contexts.