During January 19–May 19, 2024, the microbiology laboratory collected 7,257 respiratory specimens, of which 6,766 were lower respiratory tract specimens: 5,922 expectorated sputum, 388 tracheal aspirates, and 456 bronchial specimens, including BALF and endobronchial biopsy specimens. The other 491 respiratory specimens were upper respiratory tract specimens: 113 nasal or nasopharyngeal swab samples, 283 oral or throat swab samples, and 95 gastric aspirates.

Among collected specimens, 66 (64 sputum and 2 throat swab) specimens from 65 patients grew thermotolerant bacteria that could not be identified by MALDI-TOF spectrometry (Table 1). Among the 66 thermotolerant bacterial specimens, we identified Tepidimonas spp. in 42 (63.6%, 0.6% of all respiratory specimens, 41 sputum and 1 throat swab) specimens from 41 patients by 16S rRNA sequencing; 37 were T. aquatica, 3 were undetermined Tepidimonas spp., 1 was T. taiwanensis, and 1 was T. fonticaldi.

Patients with Tepidimonas spp. bacterial isolates were from all age groups; 56.1% of patients were female and 43.9% were male (Table 2). The most common underlying conditions in the cohort were solid organ malignancy (41.5%), hypertension (36.6%), chronic lung disease (24.4%), and hyperlipidemia (24.4%). Among Tepidimonas spp. isolates, 58.5% were from sputum saved in the outpatient setting or collected <48 hours of hospital admission, suggesting a possible community source. Common clinical signs and symptoms included cough, fever, and shortness of breath. Some of the patients had abnormal chest radiograph findings, including consolidation or pulmonary infiltrates (26.8%), pleural effusion (19.5%), and chronic lung changes (24.4%). However, 34.1% of patients with Tepidimonas spp. isolated had unremarkable findings on chest radiographs, representing possible bacterial colonization of the respiratory tract.

Among laboratory characteristics of the 41 sputum specimens, 51.2% had trace leukocytes, and 46.3% had <10 epithelial cells per field (Table 3). In addition, 36.6% had growth of other pathogenic organisms, the most common of which were Haemophilus influenzae and Flavobacterium species.

Novel Species Identification

Our study identified 3 clinical isolates of a novel Tepidimonas sp. bacterium. Here, we report phenotypic characteristics and genomic features of those bacteria, including detection of virulence, antimicrobial resistance (AMR), and stress-related genes. To investigate potential sources of this bacterium, we also performed environmental sampling in the hospital.

Phenotypic Analysis and Biochemical Characteristics

The 3 undetermined sequences detected by 16S rRNA sequencing displayed 96.68%–97.59% sequence identity to known Tepidimonas spp. available in the BLAST database, suggesting that the taxonomic position of those isolates remained uncertain. The sequences were from 3 isolates (HKU77, HKU78, and HKU79) from 3 patients: 1 with pulmonary metastasis, 1 with community-acquired pneumonia, and 1 with underlying Graves’ disease and schizophrenia who had fever during hospitalization.

Gram stain of isolates HKU77, HKU78, and HKU79 revealed gram-negative, non–spore-forming, motile rods with occasional spherical enlargement (Appendix Figure, panel A). We observed small translucent and nonpigmented colonies with no hemolysis on blood agar after 24 hours of incubation. The size of colonies gradually increased with prolonged incubation. We noted no pigment production at 35°C or 42°C, but colonies displayed a slight brown pigmentation after prolonged incubation at 50°C. Optimal growth of those bacteria occurred at 50°C, and we noted no visible growth at 25°C (Appendix Figure, panels B–D). The 3 isolates did not grow in anaerobic conditions. Growth on chocolate agar and Haemophilus test medium at 50°C was similar to that for blood agar; we observed slower growth on Brucella agar. However, we noted no growth on buffered charcoal yeast extract agar or brain heart infusion agar supplemented with X and V factor. We observed no fluorescence under ultraviolet light (Table 4).

Biochemically, all 3 isolates were oxidase-positive and weakly catalase-positive (with 15% hydrogen peroxide) and tested negative for lipase, lecithinase, DNAase, and gelatinase at 50°C. The isolates were positive for nitrate reductase, glutamyl arylamidase, and tyrosine arylamidase and weakly positive for urease. Sugar fermentation (glucose, lactose, sucrose, citrate), indole, H₂S production, and ortho-nitrophenyl-β-galactoside were all negative after incubation at 50°C. We evaluated enzymatic activity by using the VITEK AutoMicrobic GNI+ card (bioMérieux, https://www.biomerieux.com), which was positive for glutamyl arylamidase and tyrosine arylamidase, and 1 strain (HKU78) showed additional enzymatic activity of L-proline arylamidase and a positive Ellman’s test result.

The 3 isolates were susceptible to piperacillin, piperacillin/tazobactam, ceftazidime, amikacin, ciprofloxacin, tobramycin, and cefepime, according to the Clinical Laboratory Standards Institute Pseudomonas breakpoints (5). The MICs of penicillin for the 3 isolates were <0.016 μg/mL, and MICs of vancomycin were 48–64 μg/mL. Comparison with phenotypic characteristics of T. aquatica and T. taiwanensis showed that the Tepidimonas species from our 3 strains did not grow on buffered charcoal yeast extract agar, but both T. aquatica and T. taiwanensis did. Furthermore, T. taiwanensis is gelatinase- and nitrate-positive, but T. aquatica and the novel Tepidimonas species from our study are not (Table 4).

Comparative Genomic Characterizations

Figure 1

Figure 1. Circular genome map of novel thermotolerant Tepidimonas spp. bacteria detected in human respiratory specimens, Hong Kong, China, 2024. Using patient strain HKU77^Tas the reference genome forming...

Figure 2

Figure 2. Comparison of feature counts of functional annotations from detection of novel thermotolerant Tepidimonasspp. bacteria in human respiratory specimens, Hong Kong, China, 2024. Annotations were determined by using Rapid...

The de novo assembly, using Illumina and Oxford Nanopore Technology reads, yielded complete genomes for the strains HKU77, HKU78, and HKU79 (Figure 1). The genome sizes of the 3 strains ranged from 2,842,968 to 2,891,522 bp, and GC content spanned 67.1%–67.2% (Table 5). We submitted the assembled genome sequences to PROKKA (https://github.com/tseemann/prokka) for annotation, which resulted in 2,695–2,786 protein-coding sequences, 3 rRNA operons, and 50–52 tRNA-coding genes (Table 5). Annotation via the RAST pipeline (Rapid Annotations using Subsystems Technology; Argonne National Laboratory, https://www.anl.gov/mcs/rast-rapid-annotation-using-subsystem-technology) indicated that the 3 strains possess a similar number of genes and subsystems, except for strain HKU78, which harbored a phage genome containing an additional integron integrase IntI1 gene related to the phages, prophages, transposable elements, and plasmids subsystem in RAST (Figure 2).

Figure 3

Figure 3. Phylogeny of novel thermotolerant Tepidimonasspp. bacteria isolated from human respiratory specimens, Hong Kong, China, 2024. Phylogenetic trees show the relationship between 3 patient strains identified in this study...

To determine the phylogenetic positions of the 3 strains within the genus Tepidimonas, we used Type (Strain) Gene Server (Leibniz Institute, https://tygs.dsmz.de) results to construct a phylogenetic tree incorporating genome data from the 3 strains and complete genomes of other Tepidimonas spp. type strains. That analysis revealed that HKU77, HKU78, and HKU79 cluster together, forming a distinct and well-supported phylogenetic clade, separate from other Tepidimonas species, and are most closely related to T. fonticaldi (Figure 3, panel A). Similarly, the phylogenetic tree constructed from core-genome sequences displayed a consistent topology (Figure 3, panel B). In silico genome-to-genome comparison demonstrated that the 3 strains from our study shared pairwise digital DNA-DNA hybridization (dDDH) values ranging from 95.3% to 97.8% among each other. Those 3 strains were most closely related to T. fonticaldi AT-A2 (dDDH value 26.1%–26.2%), followed by T. taiwanensis I1–1 (dDDH value 24.7%–24.8%), and T. charontis SPSP-6T (dDDH value 24.0%–24.3%), but the dDDH values were below the 70.0% threshold, indicating the 3 strains from our study are a different species (Table 6). On the basis of those results, we propose that strains HKU77, HKU78, and HKU79 represent a newly identified Tepidimonas species. We suggest the name Tepidimonas hongkongensis sp. nov. for the location of isolation and that strain HKU77^T (GenBank accession no. CP187300) be designated as the type strain.

Virulence, AMR, and Stress-Related Genes

We characterized virulence factors of T. hongkongensis by using the Virulence Factors of Pathogenic Bacteria database (https://www.mgc.ac.cn/VFs/main.htm), on the basis of available sequences (Appendix Table 1). We identified several genes that contribute to the function of motility and adherence, some of which are similar to genes of type IV pili of Pseudomonas spp., including pilB, pilC, and pilT, contributing to the adhesion of T. hongkongensis to other cells. We also detected different genes related to immune evasion and several genes that contribute to counteracting phagocytosis in humans. Those genes included uge, which is observed to generate the capsule in Klebsiella spp. Bacteria, and algC, algU, and algW, which are responsible for alginate regulation and biosynthesis in Pseudomonas spp. bacteria. In addition, we observed genes in T. hongkongensis that are associated with secretion systems in other bacteria, including clpV (type VI secretory system), epsE, and lspG (type II secretory system). T. hongkongensis strains also contained hemE and hemL genes, which are observed in heme biosynthesis in Haemophilus spp. bacteria.

Another species, T. taiwanensis, has previously been reported to produce alkaline protease and polyhydroxyalkanoates (PHA), renewable and biodegradable polymers that can replace conventional plastic (6). Comparison of T. hongkongensis and type strain T. taiwanensis LMG22826T (GenBank accession no. CP083911) showed that T. hongkongensis contained alkaline proteases, except lysyl endopeptidase. Furthermore, T. taiwanensis possesses phaC and phaR genes that are necessary for PHA production, as well as phaZ coding PHA depolymerase. Similarly, all 3 T. hongkongensis strains possessed the phaC, phaR, and phaZ, suggesting that T. hongkongensis is associated with PHA production. Oxidation of thiosulfate to sulfate has been reported in all Tepidimonas species through the sox pathway (7), and all 3 T. hongkongensis strains contained the genes encoding the sulfur-oxidizing protein.

Using the National Center for Biotechnology Information AMRfinder (https://www.ncbi.nlm.nih.gov/pathogens/antimicrobial-resistance/AMRFinder), we found that all 3 T. hongkongensis harbored aminoglycoside resistance genes (aadA2) and a β-lactamase gene (bla_OXA-2). We also identified stress-related genes, including multiple heat-shock proteins and efflux pumps, including qacL (Appendix Table 2).

Characteristics for Patients with T. hongkongensis Strains

The isolate of strain HKU77^T was from a patient with pulmonary metastases. The DNA GC content was 67.2% mol; we considered this the type strain. The isolate of the second strain, HKU78, was from a patient with community-acquired pneumonia and had DNA GC content of 67.1% mol. The isolate of the third strain, HKU79, was from a patient with underlying Graves’ disease and paranoid schizophrenia and had DNA GC content of 67.2% mol.

Thermotolerant bacteria can thrive under elevated temperatures, typically 45°C–60°C. Examples of thermotolerant bacteria causing human infection include Campylobacter fetus (8) and Mycobacterium xenopi (2). Various thermotolerant bacteria, such as Tepidimonas spp., have been isolated or detected in environmental hot water samples, including hot springs (9–11). However, the clinical consequences of most thermotolerant bacteria are uncertain.

In addition to Tepidimonas spp., our study identified different thermotolerant bacteria in clinical specimens, including S. aquatica, which belongs to the Comamonadaceae family and was first isolated from a hot spring in Taiwan (9) but was not known to cause human infection. In addition, we identified Vulcaniibacterium thermophilum in 4 patient samples; it was first isolated from a geothermal soil sample in Tengchong, Yunnan Province, in southwest China in 2012 (10) and was the reported cause of prosthetic joint infection in 1 case (13). In addition, Thermomonas spp. bacteria, which we detected in 2 specimens, frequently have been isolated from hot water (11,14), marine sediment (15), and soil (16–18). One study reported its detection in the naso-oropharyngeal microbiome from breast cancer patients with severe COVID-19 (19). Lactobacillus delbrueckii bacteria, which we detected in 1 specimen, commonly are used for probiotics and food fermentation (20); however, that bacterium has been reported to cause urinary tract infection and bacteremia (21–24). Finally, we detected 3 Bacillus spp. bacteria, which are known to form heat-resistant spores; although non–B. anthracis Bacillus are usually considered environmental bacteria, they can cause infection in patients with prostheses and central catheters (25,26).

The genus Tepidimonas was established in 2000 (27), and its members are gram-negative bacilli with positive catalase and oxidase activity that are motile due to a single polar flagellum. Tepidimonas spp. bacteria typically exhibit optimal growth at temperatures >45°C; therefore, most have been isolated from relatively high-temperature environments, such as hot springs and spas worldwide (9–11). As of January 25, 2026, only 8 Tepidimonas species have been validly published in the List of Prokaryotic names with Standing in Nomenclature: T. ignava, T. aquatica, T. taiwanensis, T. thermarum, T. fonticaldi, T. sediminis, T. alkaliphila, and T. charontis (https://lpsn.dsmz.de/genus/tepidimonas). Within the genus, T. taiwanensis is the only strain that uses glucose and fructose for growth, due to the presence of putative ABC glucose/mannose (gtsABCD) and fructose (frcABC) transporters (7), whereas other species are frequently referred to as asaccharolytic. Similar to a previous report (7), the novel T. hongkongensis bacterium does not use glucose and fructose for growth. Of note, T. taiwanensis produces alkaline protease and PHA, renewable and biodegradable polymers that can replace conventional plastic (6). Genome analysis revealed that T. hongkongensis contains most of the alkaline protease genes found in T. taiwanensis, as well as genes associated with PHA production (12). Analysis of stress-related genes in T. hongkongensis revealed the presence of heat-resistant protein and ATPase genes, which explain its thermotolerant characteristic.

In our study, 0.6% of respiratory specimens grew Tepidimonas spp. bacteria after prolonged incubation at 50°C, and most patients from whom specimens were collected had solid organ malignancy and chronic lung disease, although ages were wide ranging. Other Tepidimonas spp. bacteria have been detected in clinical samples. For instance, a proposed new Tepidimonas species, T. arfidensis, was cultured from a bone marrow aspirate sample from a patient with leukemia with neutropenic fever (28). Other Tepidimonas spp. bacteria have been implicated in different human microbiome studies, including the urine microbiome for urinary incontinence (29,30), the endometrial microbiome for endometriosis (31), the gut microbiome in primary sclerosing cholangitis (32), and the sinus microbiome in chronic rhinosinusitis (33). In our study, only 36.6% of the patients had growth of other pathogenic bacteria in the sputum (Table 3), suggesting that Tepidimonas spp. are pathogenic and can cause lower respiratory tract infection.

The tap water samples collected from the hospital revealed the presence of T. hongkongensis sp. nov. and a high percentage of S. aquatica. Those results align with previous findings that Tepidimonas spp. and S. aquatica bacteria are commonly found in the environment (9,34). Of note, we detected Tepidimonas spp. bacteria more frequently than S. aquatica in the water samples, but we detected more Tepidimonas spp. than S. aquatica in respiratory specimens (Tables 1, 7), suggesting that Tepidimonas spp. bacteria could have a greater potential for colonizing the human respiratory tract than other environmental bacteria.

Comparative genomic analysis revealed that all 3 T. hongkongensis strains contained various virulence genes spanning critical functional classes, including adherence, iron uptake, lipid metabolism, nutritional virulence, serum resistance, and stress adaptation, but those genes are absent in S. aquatica LMG 23380 (Appendix Table 1). Those genes support colonization of host tissue, nutrient scavenging, immune evasion, and persistence under stress, further underscoring the potential pathogenicity of T. hongkongensis. Although resistome profiling did not demonstrate substantial antimicrobial resistance in T. hongkongensis (Appendix Table 2), we did detect it in environmental samples, and its ability to acquire other resistance, such as colistin resistance, through horizontal gene transfer from other gram-negative bacteria, is not known (35,36). Therefore, further surveillance is warranted.

Whether detection of T. hongkongensis in the hospital water systems warrants interventions requires further risk assessments and studies. First, the infectious threshold of T. hongkongensis should be established to assess the risk that patients could acquire it from the water system. Furthermore, whether usual water disinfection protocols, such as chlorine or ultraviolet light, are effective against this bacterium is unknown. Ongoing vigilance and documentation will help define the baseline incidence and clinical significance for this bacterium and guide future recommendations.

The first limitation of our study is that it was conducted in a hospital network in Hong Kong; whether similar observations occur in other hospitals and countries requires further investigation. Second, we only used 50°C for agar incubation to isolate thermotolerant bacteria, which might have failed to recover other clinically relevant pathogens that do not grow at such high temperatures. Third, although isolation of Tepidimonas spp. bacteria from the clinical specimens in our study might be related to a hot water source, other studies have isolated various Tepidimonas spp. bacteria in sterile clinical specimens, suggesting this genus might not solely be an environmental contaminant. Finally, isolation of Tepidimonas spp. and other thermotolerant bacteria in respiratory specimens could represent colonization instead of respiratory pathogens. However, many patients in our study had underlying solid organ malignancy, suggesting pathogenicity is still possible in severely immunocompromised patients, such as bone marrow transplant recipients. Further studies are required to investigate the pathogenic potential of T. hongkongensis and the range of illnesses the bacteria can cause.

In conclusion, we successfully isolated 42 thermotolerant Tepidimonas spp. bacterial strains from clinical specimens, including 3 novel strains confirmed by whole-genome sequencing. We propose the name T. hongkongensis sp. nov. to describe the 3 novel strains and designation of HKU77^T as the type strain of this novel species. The combination of high incubation temperature for selection, prolonged incubation for growth, and MALDI-TOF mass spectrometery and 16S rRNA sequencing enabled discovery of this novel species. Analysis of water samples from the hospital further confirmed the presence of T. hongkongensis in the environment. Although clinical implications of T. hongkongensis, especially in immunocompromised patients, are not yet known, its potential spread in water systems is concerning. Ongoing vigilance and documentation will help define the baseline incidence and clinical significance of T. hongkongensis and guide future recommendations.