This study was conducted in accordance with the Animal Care guidelines of Kasetsart University and was approved by the Institutional Biosafety Committee (Approval No. IBC-66-V12). Permission for sample collection was obtained from all participating poultry farms before study initiation. All experimental procedures involving Salmonella spp. were performed under Biosafety Level 2 containment in compliance with applicable biosafety and laboratory safety regulations.

The study was conducted from May 2023 to February 2024 at the Kamphaeng Saen Veterinary Diagnostic Center, Faculty of Veterinary Medicine, Kasetsart University, Kamphaeng Saen, Nakhon Pathom, Thailand, and the Department of Veterinary Technology, Faculty of Veterinary Technology, Kasetsart University, Bangkok, Thailand. Samples were collected from broiler production systems in the central and eastern regions of Thailand.

This study was designed as a cross-sectional observational study. A total of 63 samples from two broiler farms during the study period were selected by convenience sampling based on sample availability. The study had two main objectives. Objective 1 was to evaluate the performance of MALDI-TOF MS for the identification and differentiation of Salmonella serotypes compared with the conventional subtyping method. Objective 2 was to determine the MIC and MBC of TCA against the Salmonella isolates obtained in this study.

A total of 63 samples were obtained from the Kamphaeng Saen Veterinary Diagnostic Center and collected from two chicken farms, with 30 samples from Farm 1 and 33 samples from Farm 2. Samples included surface swabs, rectal swabs, boot swabs, rice hulls, box liners, paper padders, ceca, mouse feces, chicken breast without skin salted (SBB), and boneless leg (BL).

Surface swabs were collected from environmental surfaces in poultry houses (floors, walls, and equipment) using two sterile cotton swabs, covering approximately 100 cm², and the swabs were added to 5 mL of buffered peptone water (BPW). Rectal swabs were collected using sterile cotton swabs, and five swabs were added into 10 mL of BPW. Boot swabs were collected from disposable boot covers worn while walking through poultry houses to sample litter-associated contamination. Solid samples, including rice hulls, box liners, paper padders, and meat samples, were collected aseptically, and 25 g of each sample was added to 225 mL of BPW. All samples were transported to the laboratory at 4°C.

For pre-enrichment, samples were mixed with 5–225 mL of BPW, depending on the sample type, using a stomacher at maximum speed for 2 min and then incubated at 37°C for 16–20 h. After pre-enrichment, 100 µL of each culture was inoculated onto modified semisolid Rappaport–Vassiliadis agar and incubated at 41.5°C for 18–24 h. Growth from modified semisolid Rappaport–Vassiliadis agar was streaked onto xylose lysine deoxycholate agar and brilliant green agar, followed by incubation at 37°C for 18–24 h. Presumptive isolates were confirmed using biochemical tests, including triple sugar iron agar, urease test, and L-lysine decarboxylation test.

A total of 63 field Salmonella isolates obtained from poultry-related samples were analyzed. The isolates comprised six serotypes: S. Enteritidis, S. Typhimurium, S. Kentucky, Salmonella Newjersey, Salmonella Fresno, and Salmonella Weltevreden (Table 1). These field isolates were used for MALDI-TOF MS-based identification and serotype-specific peak analysis, as well as for TCA antimicrobial susceptibility testing.

In addition to field isolates, reference strains were included to support method validation and comparative analyses. The reference strains used in this study were S. Typhimurium American Type Culture Collection (ATCC) 13311 and ATCC 14028 (ATCC, Manassas, VA, USA), S. Enteritidis DMST 15676, and S. Kentucky DMST 62216 (Department of Medical Sciences, Nonthaburi, Thailand). These reference strains were used as controls for MALDI-TOF MS spectral profiling and for comparison of MIC and MBC determinations.

Pure cultures of confirmed Salmonella isolates were preserved for long-term storage. In brief, isolates were grown overnight in tryptic soy broth (TSB) at 37°C to reach the late exponential to early stationary phase. For cryopreservation, 500 µL of the TSB culture was mixed with TSB containing 30% glycerol, and the mixture was stored at −20°C until further analysis. The storage duration ranged up to approximately 4 years prior to MALDI-TOF MS analysis and antimicrobial susceptibility testing.

Salmonella spp. isolates preserved in 30% glycerol were streaked onto tryptic soy agar (TSA), and reference strains (S. Typhimurium ATCC 13311, S. Typhimurium ATCC 14028, S. Enteritidis DMST 15676, and S. Kentucky DMST 62216) were streaked onto blood agar. All cultures were incubated at 37°C for 18–24 h. A single colony from each culture was selected and smeared onto a VITEK® MS 48-well target slide (bioMérieux, Marcy-l’Étoile, France). The direct smear method was used.

To ensure analytical reproducibility, each isolate was spotted in duplicate. The mass spectra were acquired in linear positive ion mode using the VITEK® MS system. The laser intensity was adjusted to obtain processed spectra with a minimum acceptable signal intensity of 30 mV. Each spectrum consisted of accumulated profiles, with one profile corresponding to five laser shots. A target of 100 acceptable profiles per spot was applied, with a minimum of 30 good profiles. Subsequently, 1 µL of α-cyano-4-hydroxycinnamic acid matrix solution was overlaid and allowed to completely dry at room temperature. The matrix solution consisted of 28% acetonitrile, with 3.10 g of α-cyano-4-hydroxycinnamic acid dissolved in 100 mL of solvent. Once dried, the target slide was inserted into the instrument.

Instrument calibration was performed prior to sample analysis using a fresh culture of Escherichia coli ATCC 8739 as the calibration standard. Calibration was based on comparison of measured time-of-flight values with a predefined reference mass list of ribosomal protein masses and was accepted only when the measured masses matched the reference values within the manufacturer-defined tolerance limits. Species-level identification was accepted only when the system generated a single, consistent identification with high confidence across replicate spots. Isolates yielding multiple identifications, low-confidence results, or inconsistent outcomes between replicates were considered unreliable and excluded from further analysis.

The mass spectral profile was analyzed using Spectral Archive and Microbial Identification System (SARAMIS) software version 4.1.0 with database KB 4.17, operated in research use only mode. Identification confidence levels were interpreted according to SARAMIS classification criteria. A “Good” identification was defined as a single taxonomic choice with a probability ranging from 60% to 99.9%. Results classified as “Low discrimination” were defined by two to four possible choices with a cumulative probability of 100%, and further confirmatory testing was required. Isolates yielding no significant choice or more than four possible choices with a cumulative probability of 100% were classified as “No identification” or “Inconclusive identification” and were excluded from downstream analysis.

Mass spectral profile analysis was confined to the 2,000–20,000 Da range because this range predominantly detects ribosomal proteins, which are highly stable and minimally affected by culture conditions [20]. These characteristics make them reliable biomarkers for accurate microbial identification. The mass spectra of the reference strains were compared with those of the isolated samples. Automated peak picking was performed using default SARAMIS parameters, including baseline subtraction, normalization, and selection of reproducible peaks based on signal-to-noise ratio and relative peak intensity. Peak data were analyzed as relative abundance rather than absolute intensity.

Specific peak analysis was performed using the SARAMIS database to identify serotype-associated mass peaks through comparative evaluation of mass spectral profiles between field isolates and their corresponding reference strains. For each serotype, the presence of peaks was assessed across multiple field isolates and technical replicates to evaluate reproducibility and consistency.

A serotype-specific mass peak was defined when it was consistently detected in the majority of isolates (≥70%) belonging to the same serotype, reproducible across replicate spectra, and absent or inconsistently detected in isolates of other serotypes. Peaks that fulfilled these criteria were retained for further analysis.

To ensure analytical robustness, peak comparisons were restricted to spectra that met predefined quality acceptance criteria. Outlier values were identified during peak comparison when detected m/z values fell outside the expected serotype-associated peak range or showed inconsistent detection across technical replicates. Such outliers were excluded from peak calculations before comparative analysis.

TCA (CAS No. 140-10-3; 97% purity) was purchased from Fluka (Buchs, Switzerland). A total of 0.148 g of TCA was weighed and dissolved in 1,000 µL of dimethyl sulfoxide (DMSO; CAS No. 67-68-5; 99.9% purity), with complete solubility ensured by vortex mixing at room temperature, to prepare a stock solution. Subsequently, the stock solution was subjected to two-fold serial dilutions to obtain concentrations of 20, 10, 5, 2.5, 1.25, 0.625, 0.312, and 0.156 mM. The prepared solutions were stored at 4°C, protected from light, and used within 2 weeks of preparation. The final DMSO concentration in each well was 0.0156% (v/v). All other chemicals used in this study were of analytical grade.

Gram-negative S. Enteritidis, S. Typhimurium, S. Kentucky, S. Newjersey, S. Fresno, and S. Weltevreden were used in this study. All strains were prepared by inoculating a single colony into 30 mL of TSB and incubated at 37°C for 22 h under shaking at 250 rpm. The absorbance of each bacterial strain was measured at 595 nm using a spectrophotometer, adjusted to 0.1–0.2 with TSB, and diluted 1:100 to achieve a final concentration of 0.001–0.002.

The MIC of TCA was determined using the broth dilution method with modifications in accordance with the Clinical and Laboratory Standards Institute M07™ guideline. Different concentrations of TCA were used. Two microliters of TCA dissolved in DMSO were added to 98 µL of each bacterial suspension into polystyrene microtiter plates to achieve concentrations of 20, 10, 5, 2.5, 1.25, 0.625, 0.312, and 0.156 mM, respectively. The lowest concentration of TCA required to visibly inhibit growth was considered the MIC.

A mixture of 2 µL of DMSO and 98 µL of bacterial suspension served as the positive control, while a mixture of 2 µL of DMSO and 98 µL of TSB served as the negative control. The microtiter plates were incubated at 37°C for 22 h. The absorbance of each plate was measured at 595 nm using a microplate reader. Then, 50 µL of p-iodonitrotetrazolium violet (INT) was added to indicate bacterial growth at all concentrations of TCA. Samples were incubated at 37°C for 30 min. All MIC assays were performed using three technical replicates and were repeated across ten independent biological replicates.

The overnight culture of bacteria from MIC determination was used. A volume of 5 µL was removed from each well, streaked onto TSA plates, and incubated at 37°C for 22 h. The MBC was defined as the lowest concentration of TCA at which bacterial growth was not observed. All MBC determinations were conducted using three technical replicates and ten biological replicates, consistent with the MIC assay design.

Quality control in MIC and MBC assays was ensured by including appropriate growth and sterility controls. The positive control consisted of 2 µL of DMSO mixed with 98 µL of bacterial suspension in TSB, adjusted to an initial inoculum of approximately optical density 0.001–0.002. This control consistently showed visible turbidity after incubation, confirming normal bacterial growth in the absence of TCA.

The negative control consisted of 2 µL of DMSO mixed with 98 µL of sterile TSB without bacterial inoculation, which remained clear throughout incubation, confirming medium sterility and absence of contamination.

A solvent control containing a bacterial suspension with DMSO at the same final concentration used in test wells but without TCA was included to evaluate potential solvent effects. No inhibitory effect of DMSO on bacterial growth was observed, indicating that any growth inhibition detected was attributable to TCA rather than the solvent. Visible turbidity was interpreted as bacterial growth, whereas a clear appearance indicated growth inhibition.

The performance of MALDI-TOF MS was evaluated using overall percentage agreement (OPA), positive percentage agreement (PPA), negative percentage agreement (NPA), and Cohen’s kappa statistic, with corresponding 95% confidence intervals. MIC and MBC values were summarized as mean ± standard deviation. Differences in MIC and MBC values among Salmonella serotypes were evaluated using the non-parametric Kruskal–Wallis test due to non-normal data distribution. All statistical analyses were two-tailed, and p < 0.05 was considered statistically significant. Statistical analyses were performed using Statistical Package for the Social Sciences software version 26.0 (IBM Corp., Armonk, NY, USA). MIC and MBC data processing and descriptive statistical analyses were performed using Microsoft Excel 365.

Salmonella strains were isolated from various sources along the poultry production chain, including surface swab (n = 1), rectal swab (n = 2), boot swab (n = 36), rice hull (n = 2), box liner (n = 7), paper padder (n = 2), ceca (n = 3), mouse feces collected from the poultry house (n = 1), SBB (n = 8), and BL (n = 1) (Table 1).

The highest detection frequency of Salmonella was observed in boot swab samples, whereas the lowest detections were recorded in surface swab, mouse feces, and BL samples. Boot swab samples accounted for the majority of isolates (36/63, 57.1%), while surface swab, mouse feces, and BL samples each represented 1.6% of the total isolates.

MIC represents the lowest concentration of an antimicrobial agent that completely inhibits visible bacterial growth. It was determined by measuring absorbance using a microplate reader, with growth confirmation performed using INT.

The MIC values for S. Enteritidis, S. Typhimurium, S. Kentucky, S. Newjersey, S. Fresno, and S. Weltevreden demonstrated complete growth inhibition at a concentration of 10 mM, as indicated by the absence of color change.

MBC represents the lowest concentration of an antimicrobial agent that completely eliminates bacterial viability. It was determined by culturing samples from 96-well plates containing TCA concentrations of 20, 10, 5, 2.5, 1.25, 0.625, 0.312, and 0.156 mM onto TSA plates, including appropriate controls.

The MBC values for all tested Salmonella serotypes showed complete bacterial killing at 20 mM, with no growth observed on agar plates.

All serotypes exhibited identical MIC and MBC values for TCA, with no statistically significant differences among serotypes (p > 0.05) (Table 4).

Kruskal–Wallis analysis further confirmed that the distribution of MIC and MBC values did not differ significantly among serotype categories (p > 0.05), indicating comparable antimicrobial susceptibility to TCA across the tested Salmonella serotypes.

The Salmonella serotypes identified in this study are consistent with previous reports from poultry production systems in Southeast Asia, where S. Enteritidis and S. Typhimurium are commonly detected, and S. Kentucky is frequently reported as a dominant serotype. The detection of less common serotypes, such as S. Newjersey, S. Fresno, and S. Weltevreden, further reflects regional diversity and highlights the importance of surveillance approaches capable of detecting both prevalent and emerging serotypes within the poultry chain.

This study demonstrates the strong potential of MALDI-TOF MS for identifying Salmonella serotypes in the poultry production chain [12, 20, 22]. The high levels of agreement between MALDI-TOF MS and conventional subtyping methods, with OPA, PPA, and NPA values ranging from 97.3% to 100%, indicate that MALDI-TOF MS can be effectively integrated into routine diagnostic procedures for epidemiological surveillance and outbreak investigations [12, 23]. However, the slightly lower agreement observed for S. Enteritidis (κ = 0.65) compared with S. Typhimurium and S. Kentucky (κ = 0.93) suggests that certain serotypes may share similar mass spectral patterns, which can occasionally lead to misidentification [5, 21].

Misclassification observed in one S. Enteritidis isolate may be attributed to biological and technical factors inherent to MALDI-TOF MS-based serotyping. Variations in protein expression, particularly ribosomal proteins that dominate MALDI-TOF MS spectra, may alter peak intensities or result in missing diagnostic peaks at the strain level. In addition, S. Enteritidis shares highly conserved protein profiles with closely related serotypes, which can produce overlapping spectral features and reduce discriminatory power. Furthermore, limited strain diversity in reference spectral libraries may contribute to misclassification when field isolates exhibit atypical or underrepresented protein patterns. These factors emphasize the need for expanded reference databases and complementary subtyping approaches when interpreting MALDI-TOF MS serotype assignments. Similar overlaps related to shared ribosomal proteins or biomarker peaks have been reported previously [11, 24]. Expanding spectral databases with additional reference strains, particularly epidemiologically important serotypes, would help reduce this variability [12, 25].

The detection of specific peaks at 6094 ± 2 m/z for S. Enteritidis, 7156 ± 2 m/z for S. Typhimurium, and 5370 ± 2 m/z for S. Kentucky demonstrates that peak-based discrimination is feasible. These findings are consistent with those reported by Jirapanth et al. [21] and Yang et al. [24], who showed that incorporating serotype-specific peaks into MALDI-TOF MS workflows can improve subtyping accuracy. In contrast, the absence of identifiable peaks for S. Newjersey, S. Fresno, and S. Weltevreden highlights serotype-specific biomarker gaps for less common serovars, providing insight into serotype-level spectral variation and the current limitations of MALDI-TOF MS peak detection algorithms.

These gaps are most likely attributable to limited representation of uncommon serotypes in commercial MALDI-TOF MS reference databases, which are often biased toward prevalent serotypes from Europe and North America. This bias may reduce discriminatory accuracy for field isolates originating from Southeast Asia. Expansion of reference spectral libraries and integration of machine learning-based approaches that analyze full-spectrum patterns may improve discrimination among closely related serotypes and support the development of region-specific diagnostic tools [12, 19, 26].

This study also evaluated the antimicrobial activity of TCA against all Salmonella serotypes. The observed MIC (10 mM) and MBC (20 mM) values indicate consistent inhibitory and bactericidal effects across all isolates. Although these MIC and MBC values are relatively high compared with conventional antimicrobials, higher localized concentrations may be feasible in certain applications, supporting the potential role of TCA as a complementary control measure in poultry production and food safety systems. Such applications may include surface treatments or processing steps where exposure is spatially and temporally limited, allowing higher effective concentrations without systemic application.

Future studies may explore formulation strategies, including encapsulation, stabilization, or nano-formulations, to enhance the antimicrobial efficacy of TCA at lower effective concentrations. These findings align with previous reports demonstrating that cinnamic acid derivatives exert antimicrobial effects by disrupting cell membrane integrity, altering permeability, and inhibiting essential biosynthetic pathways [15, 17, 18, 27]. Although effective concentrations were relatively high compared with synthetic antimicrobials, the natural origin and broad-spectrum activity of TCA make it an attractive candidate for use as a food preservative or alternative antimicrobial agent, particularly for AMR mitigation [13, 28, 29]. This study demonstrates the consistent inhibitory effect of TCA across all tested serotypes, supporting its potential application as a natural preservative to reduce reliance on conventional antibiotics in poultry production.

The integration of MALDI-TOF MS for rapid serotyping with TCA as a natural antimicrobial offers a combined approach for controlling Salmonella in poultry production systems. Rapid and accurate serotype identification enables timely epidemiological interventions, while plant-based antimicrobials may reduce dependence on conventional antibiotics and limit the emergence of resistant strains. Within a hurdle-based control framework, TCA may also be used synergistically with other non-antibiotic interventions, such as organic acids, sanitation measures, or physical treatments, thereby enhancing overall efficacy while minimizing reliance on high concentrations of a single agent.

Several limitations should be acknowledged. No standard AQC strain was included in this study. In addition, MALDI-TOF MS performance depends on database completeness; therefore, rare or emerging serotypes may not be accurately detected without regular updates [12, 22]. Moreover, the in vitro nature of antimicrobial assays may not fully represent in vivo conditions, where food matrix composition, pH, and storage conditions may influence antimicrobial effectiveness [19, 30]. Future research should focus on field validation of MALDI-TOF MS subtyping protocols, expansion of spectral libraries, optimization of TCA formulations to achieve effective doses, and evaluation of synergistic effects with other natural antimicrobials or physical treatments [18, 31, 32]. We propose a dual-control strategy in which rapid MALDI-TOF MS-based serotyping is paired with MIC/MBC screening of plant-based antimicrobials, establishing an integrated and rapid response framework for poultry chain surveillance and foodborne pathogen control.