Genotypic and Phenotypic Characterization of Corynebacterium ulcerans Strains Isolated from Domestic Animals in Brazil and Structural Implications of Mutations Conferring Quinolone-Resistance

1. Introduction

The genus Corynebacterium includes 166 valid species, some of them pathogenic to humans and animals [1]. Corynebacterium diphtheriae, the type species, is the leading cause of diphtheria, a potentially fatal disease characterized by local and systemic effects due to diphtheria toxin (DT), an exotoxin produced by the microorganism when lysogenized by phages carrying the tox gene [2,3]. Despite the global decline in cases following widespread vaccination, diphtheria remains endemic in some regions and is still reported among vaccinated individuals [4,5,6], with concerns heightened by decreased vaccination coverage during the COVID-19 pandemic [7]. Consequently, World Health Organization and the Pan American Health Organization (WHO/PAHO) have issued alerts stressing the need for preventive actions.

Beyond C. diphtheriae, other species in the C. diphtheriae complex, including Corynebacterium belfantii, Corynebacterium pseudotuberculosis, Corynebacterium ramonii, Coryneba-cterium rouxii, Corynebacterium silvaticum, and Corynebacterium ulcerans, may also carry the tox gene and cause diphtheria-like infections in humans and animals [8,9,10,11,12,13,14,15].

C. ulcerans is an emerging zoonotic pathogen that colonizes domestic [16,17] and wild animals [18,19], with human infections primarily linked to close contact with pets [20]. Both toxigenic and atoxigenic strains are increasingly reported worldwide, with the former often causing diphtheria-like disease [21]. While early treatment with erythromycin or penicillin is recommended [22], resistance is emerging [4], prompting consideration of alternatives such as quinolones.

However, limited genomic data constrain understanding of C. ulcerans transmission and the emergence of toxigenic clones. A recent genomic study of 582 isolates from diverse hosts and regions applied a novel core genome genotyping approach, revealing major sublineages and diverse mobile elements, including five tox prophage families and a novel tox-carrying element, indicating interspecies prophage transfer within the C. diphtheriae complex [23].

As potentially toxigenic corynebacteria from animals are not subject to mandatory reporting in most countries, epidemiological data remain scarce, limiting public health responses and knowledge on zoonotic transmission risks. This study aimed to phenoty-pically and molecularly characterize three atoxigenic C. ulcerans strains isolated from domestic animals in Brazil, identifying potential virulence factors and antimicrobial resistance genes, with a focus on the structural implications of mutations conferring quinolone resistance.

2. Results

2.1. Phenotypic Identification and Antimicrobial Susceptibility Profile

After a 48 h incubation period, the aspect of all grown colonies was dry, white and opaque on the surface of blood agar. The Gram stain from these colonies revealed Gram-positive bacillary forms arranged in pallid shapes with angular formations between cells. In all cultures, MALDI-TOF MS (Matrix-Assisted Laser Desorption Ionization Time-of-Flight) analysis identified isolates as C. ulcerans (99% probability), which were named as C. ulcerans IHP37393, IHP103889 and IHP106492, respectively. In the first culture of ear injury in a dog from the State of Paraná, another colony was also identified as Staphylo-coccus aureus (99% probability). In the second culture of neck abscess in a cat from the State of Pernambuco, Enterococcus spp. was also identified with 99% probability. Finally, in the third culture of ear injury in a dog from the State of São Paulo, Morganella morganii was also identified with 99% probability. There is no information about monitoring or researching careers at home.

According to the halos obtained by disk diffusion method, the clinical isolates were categorized as sensitive to the following antimicrobials tested: clindamycin (IHP106492 = 15mm; cut-off ≥ 15mm), erythromycin (IHP37393 = 38mm; IHP103889 = 38mm; IHP106492 = 34mm; cut-off ≥ 24mm), linezolid (IHP37393 = 38mm; IHP103889 = 36mm; IHP106492 = 32mm; cut-off ≥ 25mm), rifampicin (IHP37393 = 36mm; IHP103889 = 36mm; IHP106492 = 36mm; cut-off ≥ 24mm), trimethoprim-sulfamethoxazole (IHP37393 = 31mm; IHP103889 = 35mm; IHP106492 = 33mm; cut-off ≥ 23mm ) and tetracycline (IHP37393 = 33mm; IHP103889 = 36mm; IHP106492 = 35mm; cut-off ≥ 24mm). Susceptibility with increased exposure was observed to: benzylpenicillin (IHP37393 = 24mm; IHP103889 = 20mm; IHP106492 = 19mm; cut-off 12-49mm) and ciprofloxacin (IHP103889 = 35mm; IHP106492 = 36mm; cut-off 24-49mm). Resistance to clindamycin (IHP37393 = 14mm; IHP103889 = 14mm; cut-off < 15mm) and ciprofloxacin (IHP37393 = 0mm; cut-off < 24mm ) were also observed.

2.2. Genome Features and Taxonomy

The average genomes size is approximately 2.5 Mb. The genomes had predicted GC content of 53.3%. Relevant information about the assemblies of IHP37393, IHP103889 and IHP106492 strains, as well as the median coverage, N50, number of CDS, rRNA and tRNA are shown in Table 1. No chimerism was found.

The taxonomy position of the strains was confirmed as C. ulcerans by TYGS and GTDB-Tk. ANI values obtained from the comparison between our strains and close refe-rence genomes identified by TYGS are shown in a heatmap (Supplementary Materials Figure S1). The results showed ANI values of 99.9% for our strains in comparison to C. ulcerans NCTC 7910T. Moreover, as expected, the results of DDH in silico comparing our genomes with the C. ulcerans NCTC 7910T revealed values of 88.5%, 88.1% and 88.4% to IHP37393, IHP103889 and IHP106492, respectively (Supplementary Materials Table S1).

2.3. Multilocus Sequence Typing (MLST) Characterization and Phylogenetic Analysis

Analysis of the 7 housekeeping genes showed that all strains included in this study belong to ST-339. The allelic profile found was 41 (atpA), 35 (dnaE), 79 (dnaK), 49 (fusA), 52 (leuA), 48 (odhA) and 39 (rpoB).

In the phylogenetic analysis, it was possible to observe that all strains belong to ST-339 formed a well-defined clade with strong bootstrap support (100%), highlighted in Figure 1 by a yellow square. This clade is divided into two subclades which are highlighted in Figure 1 by orange and green squares. The first one contained one of our strains (IHP103889) and two Austrian strains (04-15 and 06-19). The last ones had more phylogenetic proximity to each other than to IHP103889, which was evidenced by another subclade with good bootstrap support (100%) between them, and probably because these strains were isolated from humans, while IHP103889 was isolated from a cat. The other subclade was composed of two strains from this study (IHP37393 and IHP106492) and another Brazilian strain (BR-AD 22), all of them isolated from dogs. Moreover, it was possible to observe that ST-339 clade was separated from C. ulcerans 4724 strain (ST-327) and isolated from a dog in Switzerland by a distinct clade having a common ancestor among them and having a strong bootstrap support (100%).

2.4. Prediction of Mobile Genetic Elements and CRISPR-Cas Systems

No plasmids were detected in the genomes of our strains using PlasmidFinder. Besides, IntegronFinder did not identify any integrons in the genomes. Detailed analysis of the ISs annotated by ISEScan identified three different IS families (IS110, IS21, IS256) and the same number of insertion sequences in the genome of each strain. In all of them, IS110 (n = 2 in each strain) was the most abundant IS, followed by IS21 and IS256 (n = 1 in each strain). The copy number of the annotated ISs was stable in all strains. All information about the IS copies in each strain, besides the transposase gene into the complete IS are in Supplementary Materials Table S2.

Using PHASTEST, we predicted only one intact prophage in the genome of IHP37389 strain (region length: 47.9 Kb; completeness: 130; total number of CDS: 28; region position: 201090-249006; most common phage: Rhodococcus phage Jace - GenBank accession number: NC047974; GC content: 55.58%). In the genome of IHP103889 strain, two prophages were predicted, one intact prophage (region length: 38.2 Kb; completeness: 130; total number of CDS: 29; region position: 322879-361082; most common phage: Gordonia phage Nyceirae - GenBank accession number: NC031004; GC content: 55.29%) and one questionable prophage (region length: 14.3 Kb; completeness: 80; total number of CDS: 14; region position: 61225-75524; most common phage: Gordonia phage GMA5 - GenBank accession number: NC030907; GC content: 57.32%). Lastly, in the genome of IHP106492 strain, two prophages were also detected, one intact prophage (region length: 42.5 Kb; completeness: 100; total number of CDS: 20; region position: 205082-247623; most common phage: Corynebacterium phage Adelaide - GenBank accession number: NC048791; GC content: 54.21%) and one questionable prophage (region length: 14.1 Kb; completeness: 90; total number of CDS: 13; region position: 59635-73774; most common phage: Gordonia phage GMA5 - GenBank accession number: NC030907; GC content: 55.11%). All predicted phage genes predicted in intact prophages of each strains are shown in Figure 2.

CRISPRCasFinder identified the type I-E CRISPR-Cas system, lacking the cse1 gene in all strains. IHP106492 strain harbored the addition type IU CRISPR-Cas system, which contained cas2, cas1, cas3, csb2 and csb1genes. Three CRISPR arrays with evidence levels equal to 4 were found in each one of the strains. Through the analysis of the spacer diversity among all CRISPR arrays, we found a total of 192 spacer sequences, in which IHP103889 strain carried the largest number of them, 82 spacers, followed by IHP37393 and IHP106492 with 55 spacers each one. The CRISPRTarget database identified 86 spa-cers, 24 spacers in IHP37393, 37 spacers in IHP103889 and 25 spacers in IHP106492. The identity of spacers with the highest matches values is shown in Supplementary Materials Table S3. Besides, linear genome maps including the location of the cas genes and some CRISPR arrays are shown in Supplementary Materials Figure S2.

2.5. Identification of Genes Encoding Antimicrobial Resistance and Virulence Factors

As shown in Figure 3, compared to reference C. ulcerans NCTC 7910T, incomplete pilus clusters spaABC (srtA and spaC), spaDEF (srtB and srtC) and spaGHI (spaI) were predicted in all strains, encoding the pilus SpaA-, SpaD- and SpaH-type, respectively. Moreover, all strains harbored one gene that encodes surface-anchored pili proteins (sapD). Genes involved in ABC transporter (fagABC operon and fagD gene), ABC-type heme transporter (hmuTUV cluster), Ciu iron uptake and siderophore biosynthesis system (ciuABCDE cluster) and iron-dependent regulator of diphtheria toxin production (dtxR) were also identified in all strains.

The embC, mptC and aftB genes associated with the biosynthesis of lipoarabinomannan-like lipoglycan (CdiLAM) were also found in all isolates, as well as some urease- encoding genes, ureB and ureG, and ctpV gene, a putative copper exporter. As expected, the pld gene encoding the sphingomyelin-degrading phospholipase D was found in all strains. Other genes were predicted by PanViTa, including leuD, mprA, ndk, regX3, sigE, sigH e tufA, however they presented low percentage of identity. The tox gene was not found in any of our strains. The antimicrobial resistance genes including rpoB2, rbpA, gyrA and gyrB were predicted in all strains.

2.6. Mutation Analysis

The DNA gyrase subunit A (GyrA) of the ciprofloxacin resistant strain IHP37393 exhibited two mutations compared to susceptible strains (Figure 4; Supplementary Materials Figure S3). The mutant protein has a leucine and a glycine at positions 89 and 93, respectively, whereas susceptible strains possess a serine and an aspartate at these positions [24]. In contrast, the DNA gyrase subunit B (GyrB) of the resistant strain showed no mutations relative to susceptible strains.

The gyr89L-93G model, after undergoing a 5000-step energy minimization using the steepest descent algorithm, exhibited 95.2% of amino acid residues in favored regions, while the remaining 4.8% were in allowed regions (Supplementary Materials Figure S4A). Additionally, a 100-nanosecond simulation of the model in its apo conformation confirmed that the protein maintains a stable structure (Supplementary Figure S4B, S4C). These results suggest that the model is structurally stable and functionally viable, making it suitable for subsequent in silico analyses.

Molecular docking analysis revealed that the 5BTC-ciprofloxacin complex had a binding score of -6.204, which is comparable to the gyr89L-93G-ciprofloxacin complex, with a score of -6.246. However, in the 5BTC-ciprofloxacin complex, ciprofloxacin inte-racts with Ser91 and Asp94 from chain A via a water/magnesium ion bridge and forms hydrophobic interactions with arginine and glycine at positions 60 and 61 of chain B (Figures 5A, 5B). In contrast, in the gyr89L-93G-ciprofloxacin complex, the interaction at position 89 is weakened due to the serine-to-leucine substitution, resulting in a hydrophobic interaction in the region (Figures 5C, 5D). Additionally, the aspartate-to-glycine substitution at position 93 leads to a loss of interaction of the mutated protein with the water/magnesium ion bridge complex. It is worth noting that serine 91 and aspartate 94 in the 5BTC protein occupies the same positions as serine 89 and aspartate 93 in the gyr89L-93G model. Thus, despite the similar binding affinity energies obtained from molecular docking, our findings indicate a significant loss of key interactions in the cipro-floxacin-bound mutated DNA gyrase, which impacts its binding stability and function.

The 100-nanosecond molecular dynamics simulation of the 5BTC-ciprofloxacin complex demonstrated that ciprofloxacin remained bound to the DNA gyrase protein throughout the entire simulation (Figure 6).

Like the other ciprofloxacin-susceptible C. ulcerans strains, the 5BTC DNA gyrase contains serine and aspartate at positions 89 and 93 in the GyrA subunit, highlighting these residues as key determinants of ciprofloxacin inhibitory activity (Figure 7).

In contrast, the molecular simulation of the gyr89L-93G-ciprofloxacin complex revealed a highly unstable interaction, with the ligand dissociating from the enzyme's active site within the first 10 nanoseconds of the simulation (Figure 6). The final frame of the 5BTC-ciprofloxacin simulation (A,B) showed that, despite some positional flu-ctuations, ciprofloxacin maintained interactions with the key amino acid residues throughout the simulation. Conversely, in the gyr89L-93G-ciprofloxacin complex, cipro-floxacin completely lost interaction with the mutated protein almost immediately, with no interaction observed beyond 0.04 nanoseconds (Figures 8C, 8D).

3. Discussion

In the present study, we characterized three C. ulcerans strains (IHP37393, IHP103889, IHP106492) isolated from domestic animals in Brazil. Identified by MALDI-TOF, all showed ANI values above the boundary proposed to species definition (95–96%) [25] and dDDH values above the recommended cut-off point of 70% [26], confirming their classification. Given the limited genomic data available for this species, we conducted molecular typing and phylogenetic analysis.

Classical MLST remains a valuable tool for epidemiological studies, supporting the surveillance of C. diphtheriae complex species and contributing to outbreak investigations [27]. In the present study, all isolates belonged to sequence type ST-339, a lineage pre-viously reported in atoxigenic C. ulcerans strains isolated from Austrian patients with skin infections [28], in a nasal sample from an asymptomatic dog in Brazil [29], and in a dog with an ulcerative lesion in Italy [30]. Additionally, ST-339 strains have been identified in Canada, France, and Germany according to Institut Pasteur MLST database.

The detection of ST-339 across such geographically distant regions and diverse hosts raises the hypothesis of a potential international dissemination of this lineage, possibly facilitated by animal-human-environment interactions. The repeated identification of ST-339 in both human and animal samples suggests that certain clones may possess adaptive advantages favoring their persistence and spread across different ecological niches. This finding underscores the importance of integrating molecular typing data with global surveillance systems to better understand transmission dynamics and the potential for transboundary circulation of C. ulcerans strains.

In our study, the phylogenetic analysis using the core genome, although it showed that our strains grouped with those that had the same ST, also evidenced the formation of distinct clades among them, characterizing genetic diversity among the isolates. Corro-borating our findings, genetic diversity among C. ulcerans species has already been observed in a previously published study, in which the formation of two distinct clades was evidenced [31]. Interestingly, it was also observed that regardless of the formation of the clades, the genetic variability within both clades allows isolates from humans and animals to be distributed heterogeneously in the study carried out, once again showing the tremendous genetic variation among the isolates, thus suggesting a significant adaptive evolution across the different hosts. The observation of our results also allows us to reaffirm, mainly from an epidemiological point of view, that C. ulcerans is geographically distributed, including the isolates that present significant genetic variations.

Horizontal gene transfer is key to bacterial genome evolution, promoting survival, adaptability, and acquisition of virulence and antimicrobial resistance genes, with mobile genetic elements (MGEs) playing a central role [32]. Thus, we analyzed our bacterial genomes for MGEs, including plasmids, integrons, insertion sequences, and prophages.

Plasmids and integrons contribute to the development of antimicrobial resistance and to the spread of virulence genes conferring selective advantage to bacteria [33,34]. None were detected in our strains, consistent with other C. ulcerans studies. In contrast, C. diphtheriae harbors plasmids such as pNG2 (erythromycin resistance) and pLRPD (multidrug resistance) [35,36], as well as integrons carrying resistance genes [37].

Insertion sequences (IS), the smallest and most common MGEs, significantly in- fluence bacterial genome structure, plasticity and function [38,39]. They vary in copy number and can mobilize within genomes and through other MGEs like plasmids and phages [40]. We identified three IS families (IS110, IS21, IS256) with identical copy numbers in our strains, a finding previously reported in C. silvaticum [9]. These IS families were originally isolated from Streptomyces coelicolor, Pseudomonas aeruginosa, and S. aureus, respectively [41], and have been associated with antimicrobial resistance in Corynebacterium striatum [42] and Corynebacterium jeikeium [43].

Prophages, common in C. diphtheriae complex genomes, contribute to their genomic plasticity [44]. We identified one intact prophage per strain, all from the Siphoviridae family, with GC contents exceeding the typical 53.3% of C. ulcerans genomes. In strain IHP37393, Rhodococcus phage Jace carries integrase and repressor genes, suggesting it is a temperate, circularly permuted phage [45]. In IHP103889, Gordonia phage Nyceirae is linked to prophage-mediated viral defense [46], while for IHP106492, Corynebacterium phage Adelaide's lifestyle remains uncertain (https://phagesdb.org/phages/).

CRISPR-Cas is an adaptive defense system widely found in many bacteria and archaea that helps them to recognize and destroy invading genetic elements, including viruses, phages and plasmids [47]. This system is composed of a repeated short array separated by spacers, a leader sequence containing the promoter and located upstream of the CRISPR array and, a set of CRISPR-associated genes (cas) which encode Cas proteins with endonuclease activity [48]. In this study, we found the type I-E CRISPR-Cas system in all strains, which was also detected in other C. diphtheriae complex species, including C. diphtheriae [49] and C. rouxii [13]. However, in our strains it was observed a lack of cse1 gene, which is required for Cas3 to recognize the DNA target site and position itself adjacently to the protospacer adjacent motif to ensure cleavage [50]. This lack might indicate an incomplete or modified system, although no study with the absence of this gene in C. diphtheriae complex species has been published. The high diversity of targets for the spacer sequences found in this study, like corynebacterial (C. ulcerans and C. diphtheriae), Phage Rhodococcus and Phage Streptococcus phiSASD1, highlights the effectiveness of the CRISPR-Cas in providing protection against phage infections. Moreover, IHP106492 strain harbored an additional type IU CRISPR-Cas system, containing cas2, cas1, cas3, csb2 and csb1 genes. This type of CRISPR-Cas system has also been found in Bifidobacterium spp. and other components of the intestinal microbiota of humans and animals [51,52,53]. However, even in these species, type IU CRISPR-Cas system has not yet been well cha-racterized. Additional studies should be performed to characterize this type of CRISPR-Cas system in C. ulcerans.

Several virulence factors related to adhesion, invasion, and iron acquisition were identified, supporting the pathogenic potential of this species. All strains harbored three incomplete pilus clusters - spaABC (srtA, spaC), spaDEF (srtB, srtC), and spaGHI (spaI) - indicating a pilus gene variation possibly linked to altered adherence mechanisms. These clusters are essential for epithelial adherence and colonization [54,55]. The spaD gene, encoding a surface-anchored pilus protein, was also present in all strains. Iron acquisition systems, critical for C. diphtheriae complex survival, were consistently identified. All strains carried the putative fagABC operon with fagD. An in vitro study with C. pseudotuberculosis fagB(C) mutant demonstrated that the fag genes expression in the host contribute to virulence when compared to wild type in a goat model of caseous lymphadenitis [56]. Similarly, the ciu cluster (ciuABCDE), important for survival under iron limitation, was detected in all isolates [57]. While C. diphtheriae possesses both the hmuTUV hemin transporter and HtaA-C proteins for hemin uptake [58,59], only the hmuTUV cluster was fully present in our C. ulcerans strains, suggesting limited hemin utilization, as also observed in C. rouxii [14].

The diphtheria toxin (DT) production, encoded by tox, has been described in C. diphtheriae, C. ulcerans, and C. pseudotuberculosis [60]. In this study, tox gene was absent in all C. ulcerans strains, but the dtxR gene, encoding the DtxR regulator of DT and side-rophore synthesis, oxidative stress response, and other promoters, was detected [61]. Additionally, all strains possessed the pld gene, encoding phospholipase D (PLD), a key virulence factor shared with C. pseudotuberculosis, involved in host invasion, persistence, and lesion formation [62,63,64,65]. All strains also carried embC, mptC, and aftB genes linked to CdiLAM, contributing to epithelial adherence and cell wall biosynthesis [66,67]. As expected, urease-related genes, including ureB and ureG, were present in all isolates, distinguishing C. ulcerans and C. pseudotuberculosis from C. diphtheriae [68,69].

As shown in Figure 3, the rpoB2 and rbpA genes, previously associated with rifampin resistance [36,70], were predicted in all strains. Nevertheless, further studies are required to demonstrate how this resistance occurs in Corynebacterium species. Moreover, gyrA and gyrB genes, which encode DNA gyrase, were also detected in all strains.

Fluoroquinolones, like ciprofloxacin, inhibit bacterial growth by targeting DNA gyrase. They prevent the resealing of double-stranded DNA breaks that normally occur after DNA strand passage, resulting in the formation of stable covalent enzymes–DNA adducts, known as cleaved complexes [71]. At higher concentrations, fluoroquinolones cause the release of these DNA breaks from the complexes, leading to chromosome fragmentation and, ultimately, cell death [72]. Ciprofloxacin primarily interacts with the α4 helix of the DNA gyrase A subunit, with additional non-specific contacts involving amino acid residues near the ciprofloxacin C7 group in the GyrB subunit. A key residue implicated in fluoroquinolone resistance is the second position of the α4 helix in GyrA, which is typically serine in fluoroquinolone-sensitive DNA gyrases [73]. This serine residue has been observed to participate in a water/magnesium ion bridge that stabilizes enzyme–drug interactions [74]. Also, acquired fluoroquinolone resistance is commonly associated with mutations in the α4 helix region, particularly at residues 87–94 in GyrA Mycobacterium tuberculosis and Corynebacterium spp. [75,76,77].

Previous studies on Corynebacterium spp. have shown that amino acid substitutions within the QRDR, particularly the replacement of a polar amino acid with a hydrophobic one at positions 87 and 91, correlate with increased minimum inhibitory concentration (MIC) values and resistance to quinolones, including ciprofloxacin [75,77]. In C. ulcerans susceptible to fluoroquinolones, these positions correspond to serine and aspartate at residues 89 and 93, respectively. Both are polar amino acids capable of forming hydrogen bonds with water molecules via their hydroxyl and carboxyl side chains [78], making them crucial for ciprofloxacin binding and overall quinolone efficacy.

Here, we provide significant insights into the antimicrobial resistance and pathogenic potencial of C. ulcerans, particularly concerning fluoroquinolone resistance. In this study, the mutated gyrA sequence (gyr89L-93G) contains leucine and glycine substitutions at positions 89 and 93, respectively. As aliphatic and nonpolar residues, they lack functional groups necessary for hydrogen bonding with water molecules [79], disrupting the interaction between the ciprofloxacin–water/magnesium ion bridge complex and the mutated enzyme. These observations underscore the critical role of this bridge in fluoroquinolone action and may contribute to the development of novel antibacterial agents to combat resistance.

Zoonotically acquired C. ulcerans infections are a matter of concern worldwide [23] and highlight the need for surveillance and expansion of knowledge about this pathogen. The present study provided a comprehensive genomic analysis of three C. ulcerans strains isolates from domestic animals in Brazil. Although additional studies are still necessary, the data obtained in the present work can contribute to understanding the dissemination and evolution of virulent and antimicrobial resistant strains of this zoonotic pathogen and, also to the establishment of monitoring, prevention and treatment measures for C. ulcerans infections in humans and animals.

4. Materials and Methods

4.1. Origin of Bacterial Strains

Three animals from different Brazilian states with signs of discomfort were treated at veterinary clinics. The first one, with an ear injury, was a 14-year-old dog from the State of Paraná. The second one, with a neck abscess, was an 8-year-old cat from the State of Pernambuco. The last one, with an ear injury, was a 3-year-old dog from the State of São Paulo. For analysis, three swabs were collected and sent to Hermes Pardini Institute (Fleury Group), Minas Gerais, Brazil. Bacterial cultures were performed on 5% sheep’s blood agar (Plastlabor®, Brazil) and incubated at 37°C for 48 h. The semi-automated system VITEK® MS (bioMérieux®, France) was used to identify isolated strains through MALDI-TOF MS analysis. Bacterial spots of 1 to 3 colonies were placed on the target slide. Then, 1 μL α-cyano-4-hydroxycyanic acid matrix - VITEK MS-CHCA (bioMérieux®, Brazil) was applied over the samples and air dried until the matrix and samples co- crystallized. The slide was loaded into the VITEK® MS system to acquire protein mass spectra, mainly composed of ribosomal protein. The obtained mass spectra were compared with the MYLA® software version 4.7.1 database (bioMérieux®, France).

4.2. Antimicrobial Susceptibility Testing In Vitro

Antimicrobial susceptibility profiles of all isolates were performed using disk diffusion method according to the guideline provided by the Brazilian Committee on Antimicrobial Susceptibility Testing (BrCAST) version valid from April 2024 (https://brcast.org.br/). Bacterial suspensions were prepared in saline to the density of a 0.5 McFarland turbidity standard and seeded on a Mueller-Hinton agar supplemented with 5% defibrinated horse's blood and 20 mg/L β-NAD (Plastlabor®️, Brazil). Then, the following antimicrobials (Oxoid®, Brazil) were placed on the surface of the seeded plates: benzylpenicillin (1 U), ciprofloxacin (5 μg), clindamycin (2 µg), erythromycin (15 µg), linezolid (10 µg), rifampicin (5 µg), trimethoprim-sulfamethoxazole (23,75-1,25 µg) and tetracycline (30 μg). Test plates were incubated at 35 ± 1°C in a 5% CO2 atmosphere for 40-44 h. The interpretation of susceptibility was performed according to the breakpoints established by the BrCAST guideline as well as the quality control, which was carried out using Streptococcus pneumoniae ATCC 49619.

4.3. Genome Sequencing, Assembling and Annotation

Bacterial genomic DNA was extracted using the QIAamp® DNA Blood Mini Kit (QIAGEN®, Germany) according to the manufacturer’s instructions. Whole-genome sequencing (WGS) was performed using Illumina® NextSeq 550 platform (California, USA). Library was constructed with the DNA Prep Library Preparation Kit (Illumina Inc.). Sequences quality analysis was performed using FastQC v.0.12.1 (https://github.com/s-andrews/FastQC). All genomes were assembled de novo using Unicycler v.0.5.1 [80] and contigs with less than 200 bp were trimmed. To evaluate the guanine and cytosine (GC) content, size, and fragmentation of the genomes, we used QUAST v.5.2.0 [81]. Completeness and contamination levels were estimated using CheckM2 v.1.0.2 [82] and the location of ribosomal RNA genes in genomes was predicted using Barrnap v.0.9 (https://github.com/tseemann/barrnap). Chimerism was checked using GUNC v.1.0.6 [83]. Genomes were annotated using Prokka v.1.14.6 [84] and deposited in GenBank.

4.4. Genomic Taxonomy

Taxonomy classification of the strains was performed using Type Strain Genome Server (TYGS) [85] and Genome Taxonomy Database Toolkit (GTDB-Tk) v.2.4.0 [86]. Average Nucleotide Identity (ANI) values among our strains and close reference genomes identified by TYGS were calculated using PyANI v.0.2.12 [87]. DNA-DNA hybridization (DDH) was determined in silico comparing our genomes with the type strains of the closely related species: C. diphtheriae NCTC 11397T, C. belfanti FRC0043T, C. ramonii FRC0011T, C. rouxii FRC0190T, C. pseudotuberculosis ATCC 19410T, C. silvaticum KL0182T, and C. ulcerans NCTC 7910T using the Genome-to-Genome Distance Calculator (GGDC) v.3.0 with BLAST+ [88]. The results were based on the recommended formula 2 (identities/HSP length) [89].

4.5. Multilocus Sequence Typing (MLST) Characterization and Phylogenetic Analysis

Considering the seven housekeeping genes, atpA, dnaE, dnaK, fusA, leuA, odhA, and rpoB obtained from whole genome, sequence type (ST) was determined in silico using the Institute Pauster MLST database (https://bigsdb.pasteur.fr/diphtheria/). The sequences of these seven housekeeping genes were also deposited in this database which provides access to genotyping data for C. diphtheriae complex isolates worldwide.

We used the Genbank database from National Center of Biotechnology Information (NCBI - https://www.ncbi.nlm.nih.gov) to retrieve some genomic sequences of C. ulce-rans that we consider important and related to our study. Before downloading the genomes in nucleotide FASTA format, we checked completeness and contamination levels using CheckM2 v.1.0.2 [82]. The genome of C. silvaticum KL0182T was added as an outgroup. For multiple sequence alignment of the core genome, we used PPanGGOLiN v.2.2.0 [90] with MAFFT with default option to perform the alignment [91]. SNP-sites v.2.5.1 [92] was chosen to extract single nucleotide polymorphisms (SNPs). The evolutio-nary model and phylogenetic inference were estimated by the IQ-TREE2 v.2.0.7, with the maximum likelihood method. The support values were calculated using 1000 bootstrap replications. The tree was visualized using iTOL v.6 [93].

4.6. Prediction of Mobile Genetic Elements and CRISPR-Cas Systems

PlasmidFinder v.2.1.6 was used for in silico detection of plasmids [94], and IntegronFinder v.2.0 was used for identifying and analyzing integrons across the genomes [95]. Insertion sequences (IS) were identified using ISEScan v.1.7.2.3 [96]. Prophage sequences were identified and annotated with Phage Search Tool with Enhanced Sequence Translation (PHASTEST) [97]. Intact prophage regions and their predicted phage genes were visualized using Circular Genome View Server in Proksee [98].

CRISPRCasFinder v.4.2.30 was used to analyze the presence of Clustered Regularly Interspaced Short Palindromic Repeats (CRISPRs) and Cas proteins [99]. We only included CRISPR arrays with evidence levels equal to 3 or 4 [99], and the type of CRISPR-Cas cassette was determined according to the previously described nomenclature and classification [100]. Spacer sequences were analyzed for their identity in the CRISPRTarget database [101]. Spacer hits were selected from the CRISPRTarget with a cut-off identity cover (IC) score above 0.80 [102].

4.7. Identification of Genes Encoding Antimicrobial Resistance and Virulence Factors

VFanalyzer was implemented to screen potential virulence factors using the VFDB (Virulence Factor Database) database [103]. Furthermore, PanViTa v.1.1.3 [104] to search for antimicrobial resistance genes and virulence genes using CARD (Comprehensive Antibiotic Resistance Database) and VFDB databases, respectively. For more accurate results, we also used the BlastKOALA server [105,106]. Circular genome map comparisons were built with C. ulcerans NCTC 7910T as reference using BRIG (Blast Ring Image Generator) software v.0.95 [106], to show positions for virulence factors and antimicrobial resistance genes.

4.8. Mutation Analysis

For the identification of mutations in the quinolone resistance-determining region (QRDR) in our strains, we used the susceptible DNA gyrase sequence obtained from C. ulcerans strain 809 (GenBank accession number: GCA_000215645.1) [24]. This sequence was aligned against IHP37393 strain (resistant to ciprofloxacin), and with IHP103889 and IHP103492 strains (susceptible to increased exposure to ciprofloxacin). The alignment was performed using MUSCLE in MEGA software v.11.0.13 with default parameters [89] and analyzed using Jalview v.2.4.11.1 [107].

The 3D structure of IHP37393 DNA gyrase protein (gyr89L-93G) was modeled using Swiss-Model [108], based on the DNA gyrase complex structure deposited in the Protein Data Bank (https://doi.org/10.2210/pdb5bta/pdb) resolved by X-ray diffraction (resolution 2.55 Angstroms), which shares 78.11% sequence identity and coverage of 0.48. The gene-rated model was protonated in pH 7.0 in Charmm-gui web server [109] and then, su-bjected to energy minimization with 5000 steps using the steepest descent algorithm in GROMACS – 2023 [110] and it was submitted for quality analysis using the MolProbity web server [111].

Molecular docking was performed using DockThor [112] with the M. tuberculosis DNA gyrase complexed with ciprofloxacin (redocking) (https://doi.org/10.2210/pdb5btc/pdb), which shares 76.0% sequence identity with gyr89L-93G. During the docking process we included the magnesium ion which is described as involved in protein-ligand interaction along with its associated water molecules [73]. Si-milarly, molecular docking was performed with ciprofloxacin and the gyr89L-93G stru-cture, also accounting for the magnesium ions and their coordinating water molecules.

The molecular dynamics simulations were conducted using GROMACS - 2023, employing the CHARMM36 force field [113]. Initially, the system was solvated with TIP3P water within the simulation box and then neutralized. The system underwent energy mi-nimization using the steepest descent method until the maximum force was reduced to less than 1000 kJ/mol/nm. Following minimization, the system was gradually relaxed and heated to 310 K while maintaining the protein in a fixed position. The system was equilibrated at 310 K before performing 100 nanoseconds of molecular dynamics simulations under standard in vitro conditions, including physiological temperature (25ºC), pH 7.0, and 1 atm pressure. GROMACS routines were used to calculate root mean square deviation (RMSD), root mean square fluctuation (RMSF) for the protein and ligand backbone, and the radius of gyration (Rg) of the protein.

Ciprofloxacin parameterization was performed using the CGenFF web server [114], following the GROMACS protein-ligand complex simulation tutorial [115]. All simulation plots were generated using Xmgrace (Grace-5.1.25). The 2D-dimensional interaction maps were produced with LigPlot+ v.2.2.9 [116], while 3D-structural representations of the protein-ligand complexes and docking analyses were visualized using PyMOL v.2.5.0 (The PyMOL Molecular Graphics System, v.1.2r3pre, Schrödinger, LLC).