Amplicon sequencing with the ONT platform represents a culture-free approach offering numerous advantages. Key benefits include the generation of long-read data, the ability to accurately classify organisms at the species level, and a rapid processing workflow11. Furthermore, this method is cost-effective, attributed to the reusability of flow cells and its fast turnaround time. This study demonstrates the application of ONT-based clinical metagenomics for detecting bacterial and fungal microbiota through full-length 16S rDNA and ITS sequencing. Of the 56 MTB samples, 22 (39.29%) did not yield detectable M. tuberculosis DNA using ONT sequencing. While 16S rDNA sequencing is a well-established and robust method for microbiome analysis, one major limitation is its inability to differentiate between closely related species, particularly within the Mycobacterium genus, where genetic similarity is high13. Factors such as lower bacterial load or sample degradation could theoretically influence the results. The retrospective design of the study resulted in incomplete clinical data, including antibiotic usage, smoking history, HIV status, and comorbidities, which may have confounded microbiome analysis.
Alpha diversity analysis of bacterial communities revealed significant differences between MTB and negative groups, suggesting complex interactions within the existing bacterial community. This enrichment in MTB group could result from multiple factors. M. tuberculosis may alter the environment, thereby promoting the growth of certain bacterial taxa. Moreover, the host immune response to tuberculosis infection might indirectly influence the bacterial community structure14,15. Although an increase in richness was observed, the absence of a significant difference in the Shannon index suggests that while the number of bacterial species is impacted, the relative abundance distribution (evenness) remains similar between the groups. This indicates that tuberculosis infection predominantly affects species richness without disrupting the overall community balance. Beta diversity analysis of bacterial communities highlights the impact of M. tuberculosis infection on community composition. The distinct clustering of MTB and negative samples demonstrates compositional changes in bacterial communities caused by the infection. These changes may involve variations in the relative abundance of key bacterial taxa, the presence or absence of specific indicator species, or both. The analysis of bacterial microbiota identified four dominant phyla: Firmicutes, Proteobacteria, Actinobacteria, and Bacteroidetes, consistent with previous reports on the respiratory microbiome16,17. At the species level, Streptococcus mitis, Streptococcus salivarius, and Veillonella parvula were among the most abundant in both groups, reflecting their roles as core members of the respiratory microbiota and normal oral flora. However, these bacteria can act as opportunistic pathogens under certain conditions18,19,20. M. tuberculosis was only detected in the MTB group, validating the culture results. Stenotrophomonas maltophilia is a multidrug-resistant organism and one of the most common strains associated with intensive care unit-acquired and nosocomial infections21. In the present study, it exhibited a significantly higher relative abundance in the MTB group. A previous report also highlighted that the culture-based isolation rate of Stenotrophomonas maltophilia in TB patients indicates it is a major co-occurring species with M. tuberculosis22. Differential abundance analysis revealed distinct bacterial taxa associated with each group. Stenotrophomonas maltophilia and Mycobacterium tuberculosis were significantly enriched in MTB samples, whereas Prevotella melaninogenica, Veillonella parvula, Corynebacterium striatum and Pseudomonas aeruginosa were more abundant in negative samples. This M. tuberculosis-associated dysbiosis, characterized by depletion of commensals and enrichment of opportunistic pathogens, could have implications for respiratory tract susceptibility to secondary infections.
Fungal communities responded differently to tuberculosis infection. Similar fungal richness and evenness indices suggest that tuberculosis infection does not affect fungal community structure. However, significant differences in fungal beta diversity indicate compositional shifts, with distinct fungal taxa differentiating MTB from negative samples. Analysis of fungal communities revealed that Ascomycota and Basidiomycota were the predominant phyla in both groups, consistent with prior studies of the respiratory mycobiome23. Analysis revealed that Candida orthopsilosis was only detected in the MTB group. Candida orthopsilosis, a human fungal pathogen within the Candida parapsilosis species complex24, may have a specific association with tuberculosis infection, as evidenced by its enrichment in MTB samples. Previous studies have also reported co-infections involving Candida species in tuberculosis patients25,26. In contrast, Wallemia muriae and Aureobasidium leucospermi were significantly enriched in the negative group, highlighting distinct fungal community structures.
The co-occurrence network analysis highlights intricate interactions within the respiratory microbiota influenced by tuberculosis infection. M. tuberculosis showed positive correlations with Stenotrophomonas maltophilia, Prevotella copri and Klebsiella pneumoniae, suggesting that synergistic interactions may be driven by shared metabolic pathways, immune evasion, or resource availability27,28. This co-occurrence may promote infection persistence or worsen disease progression. Conversely, M. tuberculosis displayed negative correlations with Actinomyces odontolyticus and Prevotella melaninogenica, indicating competitive interactions. While Actinomyces odontolyticus is a rare respiratory pathogen29, co-infections with M. tuberculosis have been scarcely reported30. Previous study reported an increase in Prevotella melaninogenica abundance following anti-TB treatment31. Similarly, Hu et al. observed elevated Prevotella levels after successful TB treatment32. The reduced abundance of Prevotella melaninogenica in untreated TB patients may impair lung function and exacerbate inflammation33. In negative samples, fungal taxa like Candida tropicalis and Apiotrichum montevidense were more abundant. Apiotrichum montevidense correlated positively with Haglerozyma chiarellii, Candida tropicalis and Stenotrophomonas maltophilia. Interestingly, Haglerozyma chiarellii was high abundant in MTB samples. These findings suggest that specific fungal species may either co-exist with or compete against MTB-associated bacteria, potentially influencing the composition of the respiratory microbiome in tuberculosis infection. The fungal enrichment in culture-negative samples may reflect opportunistic growth in patients presenting with symptoms that often resemble mycobacterial infections, as fungal infections are known to mimic tuberculosis and other clinical conditions34.
Our study demonstrates the effectiveness of ONT-based amplicon sequencing in analyzing bacterial and fungal microbiota at the species level in tuberculosis infection. MTB group exhibited higher bacterial diversity and distinct community structures, while fungal diversity remained stable, showing some compositional changes. These results reveal complex microbial interactions influenced by M. tuberculosis, highlighting their potential impact on infection and disease progression.
Methods
Clinical experiment statement
This study was approved by the Institutional Review Board of the Faculty of Medicine, Chulalongkorn University (COA no. 1094/2024; IRB no. 0451/67; Bangkok, Thailand). All methods were carried out in accordance with relevant guidelines and regulations. Informed consent was obtained from all participants included in the study.
Study cohort
Leftover sputum samples from patients with suspected TB infection at King Chulalongkorn Memorial Hospital (Bangkok, Thailand) between 2019 and 2024 were used in this study. A total of 161 samples were divided into two cohorts, including MTB and negative groups based on sputum culture and the GenoType® Mycobacterium CM/AS/NTM-DR line probe assay (Hain Lifescience GmbH, Germany) to identify the mycobacterial species. The inclusion criteria included participants aged ≥ 18 years, with sputum samples from patients suspected of tuberculosis infection based on clinical presentations or findings from additional investigations (e.g., chest X-rays) as determined by physicians. Only sputum samples with bacterial culture results were included. Exclusion criteria included patients aged < 18 years, patients infected with non-tuberculous mycobacteria (NTM), sputum samples without culture results, and samples with a volume of less than 200 µL. The final number of samples was 97, consisting of 41 negative samples and 56 MTB samples.
Mycobacterial culture
Clinical specimens were processed using the sodium hydroxide-N-acetyl-L-cysteine-sodium citrate (NaOH/NALC-Na citrate) method35. Briefly, specimens were decontaminated and digested by mixing them with an equal volume of NaOH/NALC-Na citrate solution (2% NaOH, 0.5% NALC, and 1.45% Na-citrate) for 15 min at room temperature. Phosphate buffer (pH 6.8) was then added to adjust the final volume to 40 mL, followed by mixing through inversion and centrifugation at 3,000 ×g for 15 min at 12 °C. After centrifugation, the supernatant was carefully discarded, and the remaining sediment was resuspended in 2.5 mL of fresh phosphate buffer. A 500 µL aliquot of the sediment was inoculated into mycobacterial growth indicator tubes and incubated in the BD BACTEC MGIT 960 System (Becton Dickinson, USA) for up to six weeks or until a positive signal was detected. The GenoType® Mycobacterium CM/AS/NTM-DR line probe assay (Hain Lifescience GmbH, Germany) was performed according to the manufacturer’s instructions to identify mycobacterial species.
DNA extraction
Genomic DNA was extracted from sputum samples using the ZymoBIOMICS™ DNA Miniprep Kit (Zymo Research, USA) and a FastPrep-24™ 5G homogenizer (MP Biomedicals, USA). Each 200 µL sputum sample was first treated with 0.1% dithiothreitol (final concentration of 0.01% or 0.65 mM) and vortexed for 20 s to ensure thorough mixing. The specimen was then incubated on a shaker at 200 rpm and 37ºC for 20 min, following the protocol to achieve a final elution volume of 50 µL per sample The concentration of genomic dsDNA was measured using NanoPhotometer® C40 (Implen, Germany). The extracted DNA was stored at − 20 °C for future analysis.
Bacterial amplification
The full-length bacterial 16S rDNA (V1-V9 regions) was amplified using primers modified from a previous study36, incorporating 3’ specific target sequences (italized) and 5’ adaptors. The primers used were 16S_27 F: 5’-TTTCTGTTGGTGCTGATATTGCAGRGTTYGATYMTGGCTCAG−3’ and 16S_1492R: 5’-ACTTGCCTGTCGCTCTATCTTCCGGY TACCTTGTTACGACTT−3’. PCR amplification was performed using the Ultra HiFidelity PCR Kit (Tiangen®, China). The 10 µL reaction mixture consisted of 5 µL of 2×Ultra HiFi mix, 2 µL of PCR Enhancer, 0.25 µM of each forward and reverse primer, 1.5 µL of DEPC-treated water, and 10 ng of nucleic acid template. The amplification conditions were as follows: initial denaturation at 98 °C for 30 s, followed by 30 cycles of 98 °C for 10 s, 60 °C for 10 s, and 72 °C for 45 s, with a final extension at 72 °C for 5 min. Amplicon barcoding was performed using a five-cycle PCR with barcode primers from the PCR Barcoding Expansion 1–96 kit (EXP-PBC096; ONT, UK). The barcoding PCR conditions included an initial denaturation at 98 °C for 30 s, followed by five cycles of 98 °C for 10 s, 60 °C for 10 s, 72 °C for 50 s, and a final extension at 72 °C for 5 min. The PCR products were separated by 1% agarose gel electrophoresis using RedSafe™ Nucleic Acid Staining Solution, then purified with a QIAquick Gel Extraction Kit (QIAGEN, Germany) according to the manufacturer’s guidelines.
Fungal amplification
The full-length fungal ITS region (ITS1 and ITS2) was amplified using primers adapted from a previous study37,38, incorporating 3’ specific target sequences (italized) and 5’ adaptors as follows: ITS_1 F: 5’-TTTCTGTTGGTGCTGATATTGCTCCGTAGGTGAACCTGCGG−3’ and ITS_4R 5’-ACTTGCCTGTCGCTCTATCTTCTCCTCCGCTTATTGATATGC−3’. PCR amplification was carried out using the KOD One™ PCR Master Mix (Toyobo, Japan). The total 10 µL PCR reaction contained 5 µL of KOD One™ PCR Master Mix, 0.3 µL of each 10 µM forward and reverse primer, 3.4 µL of DEPC-treated water, and 10 ng of genomic DNA. The PCR cycling conditions consisted of an initial denaturation at 98 °C for 2 min, followed by 35 cycles of 98 °C for 10 s, 60 °C for 10 s, and 68 °C for 10 s, with a final extension at 68 °C for 5 min. For amplicon barcoding, a five-cycle PCR was performed using the PCR Barcoding Expansion 1–96 kit (ONT, UK). The barcoding PCR conditions included an initial denaturation at 98 °C for 2 min, followed by five cycles of 98 °C for 10 s, 60 °C for 10 s, and 68 °C for 10 s, with a final extension at 68 °C for 5 min. After amplification, the PCR products were analyzed by 1% agarose gel electrophoresis and subsequently purified using a QIAquick Gel Extraction Kit (QIAGEN, Germany).
Library preparation for nanopore sequencing
The DNA library was then purified using the QIAquick PCR Purification Kit (Qiagen, Germany) and measured for quantity by Qubit® dsDNA High Sensitivity Assay Kit (Thermo Fisher Scientific, USA) and measured on a Qubit® 4.0 Fluorometer (Thermo Fisher Scientific, USA). The library was pooled and size-selected using 0.5 × (for 16S) and 0.8 × (for ITS) Agencourt AMPure XP beads (Beckman Coulter, USA). A total of 1 µg of the barcoded library was ligated to nanopore adaptors using the SQK-LSK114 Ligation Sequencing Kit (ONT, UK) and subsequently sequenced on a flow cell R10.4.1 with the MinION Mk1C sequencer (ONT, UK).
Bioinformatic analysis
The FAST5 files from Nanopore sequencing were basecalled using Guppy basecaller v6.5.7 (ONT, UK) in super-accuracy mode with a minimum Q-score of 15. The quality of the obtained FASTQ sequences was evaluated with MinIONQC v1.4.139. Demultiplexing and adaptor trimming of the FASTQ reads were conducted with Porechop v0.2.4 (available at https://github.com/rrwick/Porechop). Each sample’s filtered reads were subsequently used for reads clustering, polishing, and taxonomy identification via the NanoCLUST pipeline40. Taxonomic categorization was performed using the Blastn algorithm, with reference sequences obtained from the RDP version 2.1.3 (16S rDNA gene sequences) and SILVA version 138.1 (ITS region sequences) databases. Total Sum Scaling (TSS) normalization was applied to account for sequencing depth variations across samples, ensuring that the relative abundance of taxa was comparable for downstream analyses. For filtering, all taxa were retained without exclusion of low-abundance taxa to preserve potentially biologically relevant rare taxa. The bacterial and fungal abundance outputs were further examined using QIIME2 (https://qiime2.org/) v2021.241, which generated a collapsed taxonomy table. In addition, MicrobiomeAnalyst v2.0 was used to assess the microbiota’s alpha and beta diversity42. Differential abundance analysis was performed using the LEfSe method, with an LDA score threshold of ± 4 and a P-value of 0.05 for bacteria and ± 3 with a P-value of 0.05 for fungi43. The co-occurrence network analysis of bacterial and fungal microbiota within sputum samples was carried out using the Sparse Correlations for Compositional Data (SparCC)44 method (correlation threshold 0.3, P-value threshold 0.05) with the plugin in MicrobiomeAnalyst42.
