Prevalence and genetic distribution of Legionella spp. in public bath facilities in Kobe City, Japan

Legionella species are ubiquitous bacteria found in soil and freshwater that cause the disease, legionellosis, in humans. Infection can cause two types of disease: severe pneumonia (Legionnaires' disease) and mild febrile illness (Pontiac's disease) (Fields et al. 2002). More than 66 distinct Legionella species have been identified, at least 30 of which are known to cause diseases in humans (Cunha et al. 2016; LPSN 2023). More than 90% of cases of legionellosis are caused by Legionella pneumophila. L. pneumophila comprises at least 15 serogroups (SGs), and SG1 accounts for a majority of clinical cases (Graham et al. 2020).

The most common mechanism of infection in humans is the inhalation of Legionella-contaminated aerosols. Water systems of artificial facilities, including potable water systems, spa water, and cooling towers in large buildings, hotels, hospitals, and public baths contaminated with Legionella, have been recognized as sources of outbreaks and sporadic cases of Legionnaires' disease. Outbreaks and sporadic cases associated with Legionella-contaminated hot water systems and cooling towers have occurred worldwide (Maisa et al. 2015; Weiss et al. 2017).

In Japan, although there are many cases where the source of infection is unknown, public baths are recognized as a major source of infection (Amemura-Maekawa et al. 2018). Recently, we reported a case of Legionella infection occurring in a bath facility in Kobe City, Japan (Nakanishi et al. 2022). Moreover, several outbreaks associated with public baths have been reported (Infectious Disease Surveillance Center 2000); therefore, monitoring the contamination of Legionella in public baths is important to prevent legionellosis.

Amemura-Maekawa et al. have reported the genotype of L. pneumophila SG1 isolated from bathwater in Japan (Amemura-Maekawa et al. 2012). However, public bath facilities in Japan receive water from a variety of sources, including hot springs. Kanatani et al. elucidated the differences in the genetic characteristics of L. pneumophila isolated from shower water and bath water in public bath facilities in Toyama prefecture in Japan (Kanatani et al. 2017). This indicates that there are differences in the distribution of Legionella spp. among water sources in public bath facilities.

In our previous studies, we demonstrated that certain clonal groups have been established in cooling towers in Japan through 10 years of annual periodic monitoring (Nakanishi et al. 2019). We have also conducted regular annual monitoring for the presence of Legionella in the bath water of public baths in Kobe, Japan and collected Legionella strains between 2016 and 2021.

This study aimed to characterize the genetic distribution and pathogenic characteristics of Legionella spp. in bath facilities to aid epidemiological investigations and help prevent outbreaks. In this study, we characterized the distribution of species and SGs among bath facilities that are categorized into two groups, and determined the genotype and the presence of lag-1, a pathogenicity-related genetic marker, in L. pneumophila SG1 strains. We also compared the genotypes of clinical isolates from unknown infectious sources isolated in Kobe City during the same period as the bathwater isolates.

A total of 338 environmental isolates from July 2016 to November 2021 were analyzed in this study. The bath facilities were divided into two groups: group A had 35 bath facilities that sourced their water from the particular iron and salty hot springs, while the source of water in group B consisted of 46 other bath facilities is tap water, groundwater, or other hot springs. We collected 196 strains from bath facilities in group A and 142 strains from bath facilities in group B. In addition, nine strains were collected from the sputum of patients with Legionnaires' disease as surveillance during the same period in Kobe City. The environmental isolates were collected using the following method: water samples (500 mL) were filtered using a 0.2 μm pore-size polycarbonate membrane (catalog no. GTTP04700; Millipore, Billerica, MA, USA), and the membrane was resuspended in distilled water (5 mL) and vortexed for 5 min. After the concentrated samples were heated at 50 °C for 20 min, they were spread onto glycine–vancomycin–polymyxin B cycloheximide agar plates (Kanto chemical Co., Inc., Tokyo, Japan). The agar plates were incubated at 36 °C for 7 days in a moist chamber. The mosaic-cut, glass-like, and cysteine-requiring colonies were presumed to be Legionella. The isolates were subcultured in buffered charcoal yeast extract agar plates (Kanto chemical Co., Inc., Tokyo, Japan), dissolved in 100 μL of Tris-EDTA buffer, and incubated at 95 °C for 10 min. After centrifugation to remove cell debris, the supernatant was used as a DNA template for polymerase chain reaction (PCR).

Species of Legionella isolates were determined using the Oxoid Legionella latex test (Kanto Chemical Co., Inc., Japan) and slide agglutination tests with commercial antisera (DENKA Corporation, Tokyo, Japan). Furthermore, 16S rRNA genes or mip genes of some isolates were sequenced as described previously (Ratcliff et al. 1998). SGs of L. pneumophila strains were determined using slide agglutination tests with commercial antisera (DENKA Corporation, Tokyo, Japan) and a multiplex PCR (M-PCR) serotyping assay (Nakaue et al. 2021; Komatsu et al. 2023). The serotype determined by M-PCR serotyping was denoted as SGg (SG-genotype) (Komatsu et al. 2023).

Genotyping was conducted using multi-locus variable number tandem repeat analysis (MLVA) and sequence-based typing (SBT). MLVA-12 molecular genotyping analysis by modified multiplex PCR and capillary electrophoresis were carried out for 78 L. pneumophila SG1 strains, as described previously by Sobral et al. (2011). Briefly, 12 MLVA regions were divided into three PCRs: A set (Lpms01-NED, Lpms31-FAM, Lpms33-VIC, Lpms35-PET), B set (Lpms03-VIC, Lpms13-NED, Lpms19-PET, Lpms34-FAM), and C set (Lpms38-NED, Lpms39-PET, Lpms40-FAM, Lpms44-VIC). Each PCR product was analyzed with an ABI 3500 Genetic Analyzer (Applied Biosystems, Foster City, CA, USA). A GeneScan 1200 and 600 LIZ Size Standard (Applied Biosystems) was used to provide internal size markers in A and B sets, and C set, respectively. Fragment sizes were measured using GeneMapper Ver. 4 (Applied Biosystems). The repeat number was calculated from the size and assigned according to the number of repeats for each locus using L. pneumophila Philadelphila-1 as reference strain. For the 54 L. pneumophila SG1 strains identified as major genotypes by MLVA-12 molecular genotyping analysis, SBT was performed according to the protocol of the European Working Group for Legionella infection (EWGLI) using seven genes (flaA, pilE, asd, mip, mompS, proA, and neuA), as described previously (Gaia et al. 2005; Ratzow et al. 2007). Seven allele profiles and sequence types (STs) were identified based on the EWGLI SBT database. A minimum spanning tree (MST) based on MLVA was constructed using the Bionumerics software (Bionumerics ver.7.5; Applied Math., Sint-Martens-Latem, Belgium). The MST was used to indicate differences in the number of loci between operational taxonomic units, with categorical coefficients of similarity and a priority rule for the highest number of single-locus variants as parameters. MLVA clonal complexes (CCs) were defined as single- and double-locus variants according to a previous report (Sobral et al. 2011; Komatsu et al. 2023). The PCR was performed to detect the lag-1 gene as previously described (Kozak et al. 2009).

Fisher's exact test was used to evaluate differences in the frequency of L. pneumophila isolation and the prevalence of pathogenic genetic markers between the groups. A p-value of <0.001 was considered statistically significant.

We investigated the distribution of Legionella spp. in public bath facilities that sourced their water from the particular hot springs (group A) and other bath facilities (group B) in Kobe City, Hyogo, Japan. L. pneumophila was the most frequently isolated species, accounting for 58.7% (115/196) and 82.4% (117/142) of isolates in groups A and B, respectively. While SG1 was the most abundant SG in both groups, different bath facilities had different distributions of other L. pneumophila strains; SG5/SGg5 (21/142, 14.8%), SG6 (22/142, 15.5%), and SG9/SGg9 (15/142, 10.6%) prevailed in group B, and SGg4/10 (16/196, 8.2%) in group A (Table 1).

In group A, L. israelensis (29/196, 14.8%), L. londiniensis (26/196, 13.3%), and L. micdadei (20/196, 10.2%) were the dominant species present with L. pneumophila. Ten Legionella species other than L. pneumophila were detected in group B. These results indicate that there are differences in the distribution of Legionella between hot spring facilities and other general public bath facilities.

PCR amplification showed that 38.5% (30/78) of L. pneumophila SG1 isolates were positive for the lag-1 gene. Significantly more isolates from bath facilities in group A harbored the lag-1 gene than those from group B (52.2% (24/46) vs. 18.8% (6/32), p < 0.001 based on Fisher's exact test). Among the 14 STs in the 4 CCs, strains belonging to ST-KB1, ST-KB2, and ST1428 harbored lag-1 (Table 2). These results suggest that there are differences in lag-1 distribution among the bath facilities.

We compared the genotypes of nine clinical isolates from sporadic cases of legionellosis for which the source of infection had not been identified. They were collected through surveillance during the same period as the isolates from bath facilities. All strains were assigned to different MLVA types and STs and harbored the lag-1 gene. Eight strains, excluding patient 3 (P3), were included in the soil group (S1, S2, and S3), according to a previously reported grouping (Amemura-Maekawa et al. 2012). Nine clinical strains were dispersed in the MST, whereas there was a bathwater isolate that differed only by one allele from ST481 of patient 3 (P3, KL1805) (Figure 1 and Table 3). No strains from bath water in our samples exhibited the same MLVA genotypes or STs as the clinical strains.

This study revealed differences in the distribution of Legionella spp. between hot springs and other public bath facilities. While L. pneumophila was the most predominant species in both groups, the percentage of L. pneumophila was higher in group B than in group A (82.4%, 117/142 in group B vs. 58.7%, 115/196 in group A, p < 0.001 based on Fisher's exact test). While SG5 and SG6 were predominant in Japan's public baths, as previously reported (Amemura-Maekawa et al. 2008), SGg4/10 was predominant in hot springs in our area. Therefore, it was suggested that the distribution of SGs other than L. pneumophila SG1 differed according to the bath facility type and the area.

In bath facilities in hot springs, the highest percentages of the three species, L. israelensis (14.8%), L. londiniensis (13.3%), and L. micdadei (10.2%), were detected, followed by L. pneumophila. These results indicated that these species, along with L. pneumophila, are the dominant species in hot springs. Notably, L. israelensis, which was not detected in group B, appeared to be a species characteristic of group A. Although L. londiniensis and L. micdadei have been isolated from hot springs (Karasudani et al. 2009), L. israelensis was also found to be a species that colonized in hot springs in Japan. L. israelensis has been isolated from oxidation ponds and fishponds in Israel, but very little detailed information on its ecological characteristics is available (Bercovier et al. 1986). The unique spring characteristics of the area (e.g., pH, temperature, and composition) may be suitable for the survival of L. israelensis. In public facilities other than hot springs, 10 species were sparsely isolated. Among the detected species, L. micdadei, L. feeleii, L. londiniensis, L. nagasakiensis, L. jordanis, L. maceachernii, L. rubrilucens, and L. oakridgensis have been clinically reported in patients with legionellosis (McDade 2008; Yang et al. 2012; Amemura-Maekawa et al. 2018). Although legionellosis caused by non-L. pneumophila species are rare in Japan (Amemura-Maekawa et al. 2018), there is a risk of infection with non-L. pneumophila in bath facilities. Kanatani et al. also reported the prevalence and differences in Legionella spp. between shower water and bath water sources in public bath facilities in Toyama prefecture, Japan (Kanatani et al. 2017). Thus, the distribution of serotypes and Legionella species between hot springs and other bath facilities may differ because of differences in water sources and local areas in Japan.

We identified four major CCs of L. pneumophila SG1 in the study area. Additionally, 10 STs (ST-KB1–KB10) identified as unique types in this study had novel allelic profiles. CC1 and CC3 are representative genotypes found in hot springs and other bath facilities, respectively. ST1, belonging to CC4, is also the most predominant ST in cooling towers in Japan (Nakanishi et al. 2019) and is known to be environmentally and clinically predominant worldwide (Kozak-Muiznieks et al. 2014; Lévesque et al. 2016). Therefore, it is suggested that ST1 is widely present in artificial environmental waters, regardless of region. Amemura-Maekawa et al. reported that MST analysis based on the SBT of SG1 environmental isolates can be divided into eight major CCs, including three B groups (B1, B2, and B3), two C groups (C1 and C2), and three S groups (S1, S2, and S3), which include major environmental isolates derived from bath water, cooling towers, and soil and puddles, respectively (Amemura-Maekawa et al. 2012). In our sample set, CC1 and CC2 were included in the B1 group, while CC3 and CC4 were included in groups B2 and C1, respectively. However, a correlation with the grouping of Amemura-Maekawa et al. suggested that the genotype distribution varied by region and water source. Through this surveillance, the same MLVA genotypes and STs as those of the clinical strains were not present among the strains isolated from bath water. This suggested that they were infected from other sources or in other areas. STs of the clinical strains were included in the soil groups (Table 3), suggesting soil as another potential source for L. pneumophila infection. It would be necessary to also survey the distribution of STs in the soils of local garden and potted plants from a public health perspective (Zhan et al. 2022). Significantly more isolates from bath facilities in group A harbored the lag-1 gene than those from group B, which suggests that these hot spring isolates have the potential to cause human legionellosis. Therefore, improved control and prevention strategies are urgently needed.

In conclusion, we elucidated differences in the distribution of Legionella species, SGs, and genotypes between hot springs and other public baths. We also identified four major CCs in L. pneumophila SG1 and found that CC1 is a specific genotype with the lag-1 gene in hot springs. Our study will be useful for estimating the sources of infection in Japan and developing prevention strategies.

This research was supported by MHLW Health and Labour Sciences Research (grant no. JPMH22LA1008) and the Japan Society for the Promotion of Science KAKENHI (grant nos.19K12376 and 23K11469).

S.K. and N.N. designed the study methods and wrote the first draft of the manuscript. N.N., S.K., and S.T. carried out bacterial isolation, identification typing, and PCR; and S.K. and N.N. analyzed the data. All the authors have read and agreed to the published version of the manuscript.

This study was approved by the Kobe City Review Board; Ref. SenR3-1.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.