Characterization of Lactococcus isolated from diseased yellowtail in Japan using whole-genome sequencing

Abstract

Lactococcus is the causative agent of lactococcosis in fish; however, its genetic characteristics in yellowtail from Japan have not been fully elucidated. In this study, we used whole-genome sequencing (WGS) to investigate virulence genes, antimicrobial resistance genes, and genetic relationships among 40 Lactococcus isolates from diseased yellowtail in 2021. Serotype II accounted for 82.5% of the isolates, while the remaining 17.5% comprised serotype I. No novel serotype (serotype III) was identified. Compared with serotype III Lactococcus strains isolated from other fish species, virulence gene patterns varied among serotypes, and the nucleotide sequences were distinct from those of serotype III. The isolates showed resistance rates of 87.5% to lincomycin and 12.5% for erythromycin, and no resistance to tetracycline, ampicillin, or florfenicol was detected. All erythromycin-resistant isolates carried the erm(B) gene on a plasmid. The core-genome single-nucleotide polymorphism-based phylogenetic tree showed clear clustering by serotype. Although no novel serotype was found in the yellowtail, isolates from other fish species exhibited distinct gene sequences encoding surface proteins, suggesting caution regarding vaccine efficacy. Because high lincomycin resistance was observed, susceptibility testing is recommended to guide appropriate antimicrobial selection. Overall, the elucidation of the WGS-based characteristics of field isolates of Lactococcus provides useful information for vaccine use and antimicrobial therapy for the prevention and treatment of lactococcosis.

INTRODUCTION

Lactococcus garvieae is the etiological agent of lactococcosis, a highly fatal, hemorrhagic septicemia that affects yellowtail and other cultured fish species in both freshwater and saltwater environments [42]. L. garvieae has been isolated from milk of cows [9] and buffaloes with mastitis [39] and from various human blood, urine, and skin specimens [44, 45, 47]. In humans, L. garvieae is considered an opportunistic pathogen responsible for a wide range of infections [33, 37]. The virulence of L. garvieae in fish is partly dependent on its capsule-forming ability [42]. In addition, L. garvieae possesses multiple pathogenic genes [41].

L. garvieae isolated from diseased fish is classified into two groups: KG⁺ (non-capsulated) and KG⁻ (capsulated) [20]. Both KG⁻ and KG⁺ strains agglutinate with anti-KG⁻ phenotype serum. KG⁻ strains can lose their capsules and transition to the KG⁺ type during subculture, indicating that both types belong to the same serotype [3, 48]. However, L. garvieae isolates that did not agglutinate with antiserum against capsulated L. garvieae cells (KG⁻ phenotype cells) were designated as serotype II in 2012 [15, 31]. Recently, serotype II isolates have been reclassified as Lactococcus formosensis [21]. However, L. formosensis continues to be commonly designated as serotype II. Therefore, in this study, L. formosensis is referred to as serotype II. Some evidence suggests that serotype II is more pathogenic to fish than serotype I [21].

Vaccination is considered the most effective strategy for controlling lactococcosis [24, 42]. Autogenous formalin-inactivated vaccines have been developed using L. garvieae isolates from fish farms experiencing outbreaks [12, 13]. In Japan, lactococcosis caused by L. garvieae was first confirmed in yellowtail farms in 1974 [15]. Economic losses due to L. garvieae (serotype I) have been substantial, and an oral vaccine was approved in 1997. The approval of an injectable vaccine in 2000 significantly reduced economic losses due to lactococcosis [15, 46]. However, in 2012, serotype II was isolated, raising concerns that existing vaccines may not provide sufficient protection against all L. garvieae strains [15, 30]. Consequently, a vaccine against serotype II was approved in 2016, followed by an oil-adjuvanted vaccine in 2019 [28]. However, in 2021, L. garvieae isolates that did not agglutinate with serotype I or II antisera were recovered from diseased greater amberjacks [1]. Minami et al. [25] proposed classifying these isolates as serotype III, as they likely represent a new serotype. Their study also indicated that a combined inactivated vaccine for serotypes I and II exhibited only limited protective efficacy against this newly identified serotype. Subsequently, a vaccine also effective against serotype III was introduced to the Japanese market in May 2025.

Antimicrobials such as macrolides, tetracyclines, and phenicols have been used to treat lactococcosis [5, 16, 19]. In 2002, Kawanishi et al. [19] reported antimicrobial resistance in L. garvieae isolates from cultured yellowtail, amberjack, and kingfish in Japan. Their findings showed that 44% of the isolates were concurrently resistant to erythromycin, lincomycin, and oxytetracycline, and all resistant isolates possessed the erm(B) and tet(S) resistance genes. Another study examining L. garvieae isolates from yellowtail, amberjack, and kingfish collected between 2003 and 2009 found resistance rates of 16.3%, 81%, and 15.9% for erythromycin, lincomycin, and oxytetracycline, respectively; multiple resistance against all 3 drugs was observed in 11.3% of these isolates [16].

The Ministry of Agriculture, Forestry, and Fisheries of Japan monitors antimicrobial resistance in Lactoccocus isolated from diseased fish [40]. Between 2011 and 2019, the resistance rates to lincomycin, erythromycin, and oxytetracycline ranged from 31.5% to 92.6%, 0% to 10.3%, and 0% to 12.8%, respectively. Tetracycline resistance in L. garvieae is believed to be conferred by tet(S), whereas erythromycin resistance is attributed to erm(B) [19, 22]. Recently, a novel gene, lsa(D), which confers resistance to lincosamides, including lincomycin, was identified in L. garvieae [36]. In Japan, erythromycin is the most frequently used antimicrobial for treating lactococcosis [40]. Resistance to florfenicol has not been reported in diseased fish-derived Lactococcus in Japan.

The distribution of resistance and virulence genes in L. garvieae serotypes I and II isolated from yellowtail in Japan remains unclear. Furthermore, no study has investigated the genetic differences between these serotypes in diseased yellowtail using whole-genome sequencing (WGS). This study aimed to elucidate the genetic characteristics, including antimicrobial resistance and virulence genes, of serotype I and serotype II L. garvieae through WGS. In addition, we compared the genetic characteristics of serotypes I and II with those of the recently identified serotype (serotype III). The findings of this study are expected to provide valuable insights for the prevention of lactococcosis using vaccine and the selection of effective antimicrobial agents in aquaculture.

MATERIALS AND METHODS

Bacterial isolates

In 2021, a total of 76 Lactococcus isolates were obtained from diseased fish (Seriola quinqueradiata, yellowtail) across 6 prefectures in western Japan. The isolates were collected under the Japanese Veterinary Antimicrobial Resistance Monitoring System (JVARM) [40]. Among the collected isolates, 40 were selected for this study after excluding those collected during the same period from the same farm. The numbers of isolates collected from prefectures A, B, C, D, E, and F were 6, 8, 9, 10, 6, and 1, respectively. Identification and serotyping of the Lactococcus isolates were performed using PCR [30] at each prefectural fisheries research station. Three L. garvieae isolates (MS210819A, MS210922A, and MS211013A3) that did not agglutinate with serotype I or serotype II antisera were also included. MS210819A and MS210922A, suspected to represent serotype III, were obtained from diseased greater amberjacks between August and October 2021 [25]. MS211013A3 was isolated from diseased amberjacks at the same time as the other two isolates. These three isolates were serotyped by PCR according to Araki et al. [7] and were confirmed to be serotype III [6]. The genome sequence of L. garvieae MS210922A was deposited in GenBank under accession number AP026069 [1].

Antimicrobial susceptibility testing

The minimum inhibitory concentrations (MICs) were determined through the broth microdilution method according to the standards of the Clinical and Laboratory Standards Institute [8] using Frozen Plate “Eiken” (Eiken Chemical Co., Ltd., Tokyo, Japan). The following antimicrobials were tested: ampicillin (ABPC), oxytetracycline (OTC), erythromycin (EM), lincomycin (LCM), oxolinic acid, sulfadimethoxine, and florfenicol (FF). Escherichia coli ATCC 25922 and Aeromonas salmonicida ATCC 33658 were used for quality control in the MIC determinations. The breakpoints set in the JVARM were used [40].

Genome sequencing and analysis

Genomic DNA was prepared for WGS using a Wizard HMW DNA Extraction Kit (Promega, Madison, WI, USA). DNA samples were subjected to tagmentation and PCR amplification using a Nextera XT DNA Sample Preparation Kit (Illumina, Inc., San Diego, CA, USA). The MiSeq Illumina platform (Illumina, Inc.) in paired-end mode (2 × 300) was used for WGS. De novo assembly was performed using Shovill software (version 1.0.4) (https://github.com/tseemann/shovill). Contig sequences were analyzed using ResFinder 4.0 (https://cge.cbs.dtu.dk/services/ResFinder/) to infer antimicrobial resistance genes except for lsa(D). For resistance gene detection using ResFinder, the detection criteria were an identity of 90% or more and a coverage of 60% or more. The lincosamide/streptogramin A/pleuromutilin resistance gene, lsa(D), was detected using ABRicate (https://github.com/tseemann/abricate). A database of lsa(D) was created using the previously reported gene sequence (accession No. NG_071178) [36]. The presence of virulence factor-encoding genes was detected using ABRicate. A database of target genes previously predicted to be involved in pathogenicity was created (Table 1) [26, 27]. The database also included the glyceraldehyde-3-phosphate dehydrogenase (GAPDH) gene (accession No. FJ524849) which encodes a surface-exposed protein frequently targeted in subunit vaccine development [18]. The identity and coverage thresholds used in ABRicate were 75% and 0%, respectively. The short-read Illumina and long-read MinION (Oxford Nanopore Technologies, Oxford, UK) platforms were used to obtain complete circular plasmid sequences. Long reads were generated using a Rapid Barcoding Kit (SQK-RBK004; Oxford Nanopore Technologies) and flowcell R9.4 (Oxford Nanopore Technologies) in a MinION sequencer, and hybrid de novo assembly was performed using LoRDEC v0.9 (https://academic.oup.com/bioinformatics/article/30/24/3506/2422175) and Flye v2.9 (https://pubmed.ncbi.nlm.nih.gov/30936562/). Snippy (version 4.6.0) (https://github.com/tseemann/snippy) was used to generate a core genome alignment based on single-nucleotide polymorphisms (SNPs) and insertion/deletions (indels). The genome sequences of L. garvieae ATCC 49156 (accession no. AP009332) and Lactococcus formosensis MS200408A (AP027275) were used as references. Gubbins (version 2.3.4) was used to remove recombination from the resulting alignment [10]. The core alignment obtained after the removal of the polymorphic sites was used to generate a phylogenetic tree using the IQ-TREE web server (http://iqtree.cibiv.univie.ac.at/) with 1,000 bootstrap replicates and ascertainment bias correction. The association between the phylogenetic tree and the isolates’ metadata was visualized using ggtree (version 3.6.2) [49]. The assembled plasmid genome was used for BLASTn (https://blast.ncbi.nlm.nih.gov/Blast.cgi) analysis of all the available bacterial genomes. Pairwise genomic nucleotide sequence comparison was performed using EasyFig (version 2.2.5) [38].

Table 1. Putative virulence genes.

Gene product	Gene	Locus in Lg2 (AP009333)
Capsule gene cluster (CGC1-CGC18)	-	LCGL_0431-0448
Haemolysin 1	hly1	LCGL 0323
Haemolysin 2	hly2	LCGL 0374
Haemolysin 3	hly3	LCGL 0597
NADH oxidase	-	LCGL 0664
Superoxide dismutase	sod	LCGL 0285
Phosphoglucomutase	pgm	LCGL 0521
Adhesin Pav	adhPav	LCGL 1330
Adhesin PsaA	adhPsaA	LCGL 1533
Enolase	eno	LCGL 1514
LPxTG-1	-	LCGL 1005
LPxTG-2	-	LCGL 1410
LPxTG-3	-	LCGL 1585
LPxTG-4	-	LCGL 1672
Adhesin	adh	LCGL 0196
Adhesin cluster 1	adhCI	LCGL 0842
Adhesin cluster 2	adhCII	LCGL 0843
Adhesin cluster 3	adhCIII	LCGL 0844
Adhesin cluster 4	adhIV	LCGL 0845

Data availability

All raw short-read sequence data have been deposited in the DNA Data Bank of Japan (BioProject PRJDB20419).

RESULTS

Serotypes

Seven of the isolates (17.5%) belonged to serotype I, while 33 of the isolates (82.5%) belonged to serotype II.

Virulence genes

All isolates carried the hemolysin-1, -2, and -3; NADH oxidase; superoxide dismutase; phosphoglucomutase; adhesin Pav; adhesin PsaA; enolase; LPxTG-2 and -3; and adhesin cluster-1, -2, -3, and -4 genes. However, the L. garvieae serotype I isolates and III isolates (MS210819A, MS210922A, and MS211013A3) also possessed the adhesin and LPxTG-1 and -4 genes (Supplementary Tables 1, 2). The coverage and identity rates were 100% for most genes, but 3 out of 7 serotype I isolates exhibited less than 90% coverage (80.08%) for the LPxTG-2 gene (Supplementary Table 1). These three isolates were closely positioned on the phylogenetic tree (Fig. 1). L. garvieae MS210819A, MS210922A, and MS211013A3 contained all 18 genes; however, their identity rates were lower than those of the other serotype I isolates except for the hly1 gene. In particular, the identity rate for LPxTG-2 was low (86.66% to 98.72%). These results suggest that the surface protein structure of serotype III may be different from that of the other serotypes. Regarding capsule gene cluster (CGC) genes, serotype I isolates carried all 18 genes, except for one isolate (R3-Lg-Aq-63), which carried only CGC1. Serotype II and L. garvieae MS210819A, MS210922A, and MS211013A4 lacked CGC genes. All isolates possessed the GAPDH gene with high sequence identity (>99.21%) to the reference sequence.

Fig. 1.

Genetic relatedness, epidemiology, resistance phenotypes, and resistance gene profiles of Lactococcus garvieae isolates (n=40). Reference: L. garvieae ATCC 49156 (AP009332). EM_R, erythromycin (EM) resistance; LCM_R, lincomycin (LCM) resistance; lsa(D) _R, reference type of lsa(D) (NG_071178); lsa(D) _V, variant type of lsa(D); erm(B), erythromycin resistance gene.

Antimicrobial susceptibility and resistance genes

The resistance rates to LCM and EM were 87.5% and 12.5%, respectively (Table 2). Six isolates (15.0%) were susceptible to all 3 antimicrobials with defined breakpoints, while 5 isolates (12.5%) were resistant to both EM and LCM. All EM-resistant isolates carried the erm(B) gene. Resistance to OTC was not detected.

Table 2. Distribution of minimum inhibitory concentrations (MICs) and resistance (%) in Lactococcus garvieae (n=40).

All isolates carried the lsa(D) gene, with 4 distinct variants identified compared with the resistant-type reference sequence (accession No. NG_071178) (Fig. 1, Supplementary Table 3). Four LCM-susceptible L. garvieae serotype I isolates exhibited 90.89% sequence identity (lsa(D) _V1) with the reference sequence. In contrast, 3 serotype I isolates were resistant to LCM despite having the same identity as the reference sequence (lsa(D) _V2). The amino acid, Lsa(D), at position 36 was substituted from G to D in lsa(D) _V1 and from G to A in lsa(D) _V2. One serotype II isolate exhibited 99.93% identity with the reference sequence, yet exhibited LCM sensitivity (lsa(D) _V3). The amino acid, Lsa(D), at position 233 was substituted from W to premature termination codon (PTC) in lsa(D) _V3. Isolates MS210819A, MS210922A, and MS211013A4 showed 91.23% identity (lsa(D) _V4) with the reference sequence. In these strains, the amino acids at positions 36 and 233, which were both in Lsa(D), were the same as those in the reference strain, but the amino acids at position 322 and 364 were replaced by E and V, respectively. All isolates that exhibited 100% identity with the reference sequence were resistant to LCM.

Hybrid assembly was performed for two of the five EM-resistant isolates, and circular plasmid nucleotide sequences were obtained (pLg85 and pLg133). These 2 plasmids were 100% identical, with a genome size of 27,704 bp. BLASTn analysis of pLg85 and pLg133 revealed high similarity (98% query coverage, 100% identity) with the L. formosensis strain KH2101 plasmid pkh2101 (26,850 bp) (ON321837) [2]. The query coverage and identity of pLg85 and pLg133 compared to the L. garvieae serotype I plasmid pKL0018 (20,034 bp) (accession no. AB290882) [23] were 44% and 96.43%, respectively. Pairwise genomic nucleotide sequence comparison of plasmid pLg85 with pkh2101 and pKL0018 revealed that pLg85 and pkh2101 were nearly identical, whereas pLg85 and pKL0018 exhibited slight differences (Fig. 2). pKL0018 contained the tetracycline resistance gene, tet(S), and the macrolide resistance gene, erm(B), while the other three plasmids contained only erm(B) as drug resistance gene.

Fig. 2.

Multiple sequence alignment of plasmids: pKL0018 (accession no. AB290882) [23]; pkh2101 (ON321837) [2]; pLg85 and pLg133 (this study). Red arrows indicate erm(B) genes.

Phylogenetic analysis

An SNP-based phylogenetic tree showed that the isolates were distinctly clustered by serotype (Fig. 1). Serotype III was not included in the serotype I or II cluster. Within the serotype I cluster, the three lincomycin-resistant isolates were closely grouped. In the serotype II cluster, genetically related isolates were found within the same prefecture. Isolates harboring erm(B) were also closely located within the same cluster; however, their origin spanned three prefectures (Fig. 3).

Fig. 3.

Genetic relatedness, epidemiology, resistance phenotypes, and resistance gene profiles of Lactococcus garvieae isolates (serotype II) (n=33). Reference: Lactococcus formosensis MS200408A (AP027275). EM_R, erythromycin (EM) resistance; LCM_R, lincomycin (LCM) resistance; lsa(D) _R, reference type of lsa(D) (NG_071178); lsa(D) _V, variant type of lsa(D); erm(B), erythromycin resistance gene..

DISCUSSION

Lactococcosis is an economically significant disease, and comprehensive information on serotyping, virulence profiles, and antimicrobial resistance in Lactococcus isolates is critically valuable for vaccination strategies and evidence-based antimicrobial therapy. WGS provides detailed and integrated insights into these factors. In addition, molecular epidemiological analysis using WGS has become useful for evaluating the transmission routes of Lactococcus isolates in aquaculture.

Since its first outbreak in 2012, serotype II infection has been a major concern in Japanese aquaculture [35]. More recently, a potentially novel serotype (serotype III) that does not agglutinate with serotype I or serotype II antisera was isolated from greater amberjack and striped jack in August 2021 [1, 21, 25]. Serotype III isolates were additionally obtained from yellowtail in 2022 [7]. However, serotype III was not detected in this study. As the isolates examined were collected between April and November 2021, we speculated that serotype III was absent because this represented an early sampling period. Nevertheless, continuous monitoring of its emergence is necessary. The results of this study showed that the sequence of the gene encoding the surface protein in serotype III isolated from fish species other than yellowtail differed from that of other serotypes. It has been reported that the effectiveness of existing vaccines against serotype III isolates is low [25]. In Japan, a vaccine effective against serotype III has already been released, but it will likely take some time for its use to become widespread. Therefore, it is necessary to strengthen aquaculture hygiene to prevent infection with serotype III.

In the phylogenetic tree, Lactococcus isolates were distinctly classified by serotype. This result is consistent with previous phylogenetic analyses based on MLSA [25] and core genome sequences [21]. Serotype III clustered closely with serotype I isolates and exhibited a similar pattern of virulence gene possession. However, a previous study suggested that serotype II is the most pathogenic serotype, followed by serotype III, with serotype I being the least pathogenic of the three [21]. Among the 18 genes examined in this study, serotype II harbored just 15 genes, and the degree of gene identity was lower than in other serotypes. Previous reports have shown that some virulence genes are unique to serotype II, and that a prophage pattern distinct from other serotypes plays a role in the pathogenicity of serotype II [21]. These factors may contribute to the elevated pathogenicity of serotype II.

Our findings showed that only serotype I possessed CGC genes. Regarding the role of capsulization in the pathogenicity of Lactococcus, some reports suggest a role, while others do not [17, 24, 32], indicating that the relationship between the capsule and pathogenicity remains unclear. Although serotype II lacks CGC genes, genes involved in capsule structure (e.g., glycosyltransferase and peptidoglycan N-acetylglucosamine deacetylase), which may contribute to immune evasion, have been identified in this serotype [21]. These enzymes are likely involved in the pathogenicity of serotype II.

LPxTG is a surface protein that plays a crucial role in bacterial virulence [41]. Serotype I and serotype III were shown to carry LPxTG-1 to -4, whereas serotype II only harbored LPxTG-2 and -3. This gene is not considered a major factor in the pathogenicity of serotype II. However, the LPxTG-2 gene in serotype III exhibited low homology (86.66%) compared with the reference gene, which may contribute to pathogenic differences between serotype I and serotype III.

GAPDH is an outer membrane protein that is often targeted in subunit vaccines [18]. Because GAPDH expresses epitopes on its surface, it is readily recognized as a foreign substance by the host immune system, triggering a protective immune response. The results of this study revealed minimal diversity in the GAPDH gene among serotypes I, II, and III, suggesting that this gene is unlikely to significantly affect vaccine-induced protection.

Antimicrobials have been widely used to control lactococcosis [5, 19]. In this study, the resistance rates to EM and LCM were identified as 12.5% (5/40) and 85.0% (34/40), respectively. All EM-resistant isolates carried erm(B) and were also resistant to LCM. The breakpoints for EM and LCM were set at 16 μg/mL, referring to the values used in JVARM [40]. Isolates classified as resistant using these breakpoints showed strong concordance with the presence of erm(B) in EM resistance and mutations in lsa(D) gene in LCM resistance. In 2002, Kawanishi et al. [19] reported an LCM resistance rate of 44.1% for L. garvieae isolates, while Furushita et al. [16] documented a resistance rate of 81.0% between 2003 and 2009. Although direct comparisons are limited due to differences in testing methods and breakpoints, these findings suggest an increasing trend in LCM resistance. In this study, 97.7% of serotype II isolates and 28.6% of serotype I isolates showed LCM resistance. Previously, LCM resistance in serotype II isolates was reported to be approximately 90% between 2018 and 2021 [2]. The increasing prevalence of LCM-resistant serotype II may contribute to a gradual rise in overall LCM resistance for serotype I and II isolates combined. It has been reported that the amino acid of residue in Lsa(D) at position 36 in serotype I and at position 233 in serotype II is involved in LCM resistance in Lactococcus [34, 36]. Our results showed that there are two types of Lsa(D) amino acid sequences in serotypes I and II, and the relationship between these types and LCM resistance was consistent with previous reports. The amino acid sequence of Lsa(D) in serotype III differed from that of serotypes I and II. Although the susceptibility of serotype III to LCM was not confirmed in this study, it has been reported previously that serotype III is resistant to LCM [6, 21]. The relationship between lsa(D) gene mutations and LCM resistance in serotype III requires further investigation.

Kawanishi et al. [19] reported that the EM resistance rate for L. garvieae isolates was 44.1% in 2002, whereas Furushita et al. [16] documented an EM resistance rate of 16.3% between 2003 and 2009. In addition, Kawanishi et al. reported an OTC resistance rate of 44.1% for L. garvieae isolates in 2002, whereas Furushita et al. found no OTC resistance (0%) between 2003 and 2009. Although direct comparisons are limited due to differences in test methods and breakpoints, these findings suggest that EM and OTC resistance rates decreased between 2002 and 2009. Transmissible plasmids carried by multidrug-resistant isolates resistant to EM, LCM, and OTC harbor the respective resistance genes erm(B) and tet(S) [22]. Kawanishi et al. [19] found that isolates resistant to all three antimicrobials were predominant during the early 2000s. However, such multidrug-resistant isolates have not been detected among L. garvieae serotype I and serotype II isolates collected after 2009 [2, 16, 19]. The plasmid identified in this study carried only erm(B) and had a different structure from previously reported plasmids that harbored both erm(B) and tet(S) [22]. Based on our findings and previous reports, only serotype II possesses a plasmid containing erm(B) [2]. In contrast, serotype I is suggested to harbor plasmids carrying both erm(B) and tet(S). It is possible that serotype I isolates resistant to EM and OTC have been eliminated over time, contributing to the observed decrease in EM and OTC resistance rates. The decline in serotype I isolation is presumed to be due to vaccine use [16].

In the phylogenetic tree of serotype II isolates, five isolates carrying erm(B) belonged to the same cluster, which also included isolates that lacked erm(B). This suggests that erm(B) may be horizontally transferred between isolates within fish farms. Farmed yellowtail are sometimes may be moved from one farm to another during the breeding process [4]; therefore, it is possible that isolates carrying erm(B) were brought in from other farms. A previous study reported that a plasmid carrying erm(B) was transferred from L. garvieae serotype II to Enterococcus faecalis in vitro [2]. Macrolides, including EM, account for 43.5% of the antimicrobials used in aquatic animals in Japan [40]. Despite the selective pressure of EM use, erm(B) has not spread widely among Lactococcus, possibly because isolates carrying a plasmid with erm(B) experience reduced fitness. It is therefore assumed that Lactococcus acquires erm(B)-containing plasmids from marine bacteria. Previous reports have indicated that enterococci in marine environments harbor erm(B) [11, 43], suggesting that erm(B) may be maintained by these bacteria or other marine microbes. Since erm(B) from enterococci or other marine bacteria has the potential to be transferred to Lactococcus, the use of EM should be carefully managed to prevent further dissemination of erm(B) among Lactococcus.

To date, there have been limited reports of diseased yellowtail derived Lactococcus isolates collected from a wide area with their analysis using core-genome SNPs in Japan. In this study, the phylogenetic tree showed that isolates were distinctly clustered by serotype. Previous pulsed-field gel electrophoresis (PFGE) analyses indicated low band pattern diversity among L. garivieae isolates, making differentiation between individual isolates challenging [14, 19]. In contrast, serotype I isolates were classified into 17 groups by biased sinusoidal field gel electrophoresis (BSFGE), a modified PFGE technique, but serotype II isolates were classified into only two groups [29, 35]. However, core-genome SNP analysis revealed genotypic diversity among isolates. Within the serotype I cluster, two lincomycin-resistant isolates and one moderately resistant isolate were grouped in the same cluster, despite being collected from different prefectures. In Japan, yellowtail fry are sourced from the ocean and then farmed in different prefectures. Although the fry are collected from specific prefectures, if they originate from the same sea area, there is a possibility that Lactococcus isolates with similar genetic characteristics may be present in farms across multiple prefectures.

Although serotype III was not isolated from yellowtail in this study, the genetic sequences of surface proteins in serotype III isolated from other fish species differ from those in the established serotypes, warranting caution regarding the efficacy of existing vaccines. While the resistance rate to LCM was high, no resistance was observed to OTC, ABPC, or FF, suggesting that antimicrobial susceptibility testing could aid in selecting effective treatment options. WGS provided valuable insights into the characterization of Lactococcus, supporting the prevention of lactococcosis through targeted vaccines and appropriate antimicrobial selection.

CONFLICTS OF INTEREST

The authors declare no conflicts of interest.