A total of 16 TB patients (10 drug-sensitive and 6 drug-resistant TB patients) and 16 healthy individuals were recruited between January to March 2019 in Guangzhou Chest Hospital.

To be eligible for the study, TB patients were required to meet following criteria: (1) age > 18 years old; (2) sputum culture identified as MTB; (3) without respiratory infection induced by non-TB pathogen based on the infection indicators in blood (monocyte-to-lyphocyte ratio, procalcitonin, C-reactive protein) [12, 13], imaging examination (chest CT) [14], and pathogen detection (sputum culture for common non-TB bacterial and fungal, antibody detection for common virus and chlamydia, and polymerase chain reaction for virus if necessary) [15]; (4) without severe systemic diseases (malignant hypertension, myocardial infarction, etc.), diabetes, or malignant tumors.

During the same period, 16 healthy medical staff working in the same hospital and frequency-matched to the cases (± 5 years), were recruited. For healthy medical workers, detailed inclusion criteria were as follow: (1) aged > 18 years old; (2) physical examination and chest X-ray showing no abnormalities; (3) without respiratory infection based on the pneumonia-related symptoms (including fever, cough, expectoration, or chest pain), the infection indicators in blood (procalcitonin, C-reactive protein), imaging examination (chest CT), and pathogen detection (sputum culture for TB and non-TB pathogens, antibody detection for common virus and chlamydia, and polymerase chain reaction for virus if necessary) [15, 16]; (4) without history of chronic respiratory diseases, coronary heart disease, malignant tumors, diabetes, hyperthyroidism, malnutrition, and obesity; (5) without history of smoking.

The study was approved by the ethics committee of Guangzhou Chest Hospital and each participant provided their informed consent before inclusion in the study.

We adopted the method of ultrasonic atomization-induced deep sputum collection to collect the samples for 16 S rRNA sequencing. This method was non-invasive and barely collected the samples from upper respiratory tract [17]. Furthermore, researches have demonstrated that there was no statistically significant difference in the sensitivity and specificity of pathogen detection between deep sputum samples and bronchoalveolar lavage (BAL) samples [18]. For the TB patients who could provide sputum samples spontaneously expectorated, we also performed ultrasonic atomization-induced deep sputum collection on them to ensure the samples were comparable by exclude the bias caused by different sputum collection procedures.

Briefly, sputum were induced using a hypertonic saline solution with an ultrasonic nebulizer. For healthy individuals who should not have any sputum at all, bronchial secretion was collected. The minimal volume of each collection was 2-5 ml. In cases where this volume was not achievable after the induction procedure, we will repeated the induction and collected the best sample for analysis. The ultrasonic nebulizer was disinfected after each use according to standard hospital protocols to prevent cross-contamination. Saline solutions were changed for each patient to maintain sterility.

In vitro experiments using either molecular or genotypic techniques to detect resistance-conferring mutations, or phenotypic methods (by BACTEC MGIT 960 System) were used to determine the drug susceptibility of MTB, as previously described [19]. MTB isolates susceptible to all four of the first-line drugs were identified as drug-sensitive MTB (DS-MTB). MTB isolates resistant to any anti-TB medicine were identified as drug-resistant MTB (DR-MTB).

The TB patients receive anti-TB treatments according to the principles outlined in the WHO consolidated guidelines on tuberculosis [11, 20]. The treatment outcomes were defined based on the sputum culture results and chest X-rays/CT evaluation [21]. Patients with markedly effective treatment outcome were defined as the ones who had two consecutive sputum culture negative results in the sixth month of treatment at leats 30 days apart, significant absorption of pulmonary lesions (absorbed ≥ 50%), and cavities closure. Patients with effective treatment outcome were defined as the ones who had two consecutive sputum culture negative results in the month 6 of treatment at least 30 days apart, but the chest X-rays/CT evaluation results did not meet the definition of markedly effective. Patients with treatment failure were defined as the ones whose sputum culture was positive in the fifth month or later during treatment.

Sputum samples were liquefied with 4% NaOH and buffered in pH 6.8 phosphate buffer. Centrifugation was performed and the sediment was retained for DNA extraction. The NEB next microbiome DNA enrichment Kit (New England Biolabs, Ipswich, MA, US) was used to extract microbial community DNA according to the manufacturer’s instructions. The extracted DNA was quantified using the Qubit® dsDNA BR Assay Kit (Invitrogen, USA) and its quality was checked by running equal aliquots of the sample on 1% agarose gel.

The variable region V1-V3 of bacterial 16 S rRNA gene was amplified using degenerate PCR primers 8 F (5’-AGAGTTTTGATYMTGGCTCAG-3’) and 518R (5’-ATTACCGCGGCTGCTCG-3’). Both forward and reverse primers were tagged with Illumina adapter, pad and linker sequences. PCR enrichment was performed in a 50 μL reaction containing 30ng template, polymerase and PCR master mix. PCR cycling conditions were: 94℃ for 3 min, 30 cycles of 94℃ for 30s, 50℃ for 45s, 72℃ for 45s and final extension at 72℃ for 10 min. The PCR products were purified with AmpureXP beads and eluted in Elution buffer. Libraries were qualified by the Agilent 2100 bioanalyzer (Agilent, USA). The validated libraries were used for sequencing on Illumina HiSeq platform (BGI, Shenzhen, China) following the standard pipelines of Illumina, and generating 2 × 300 bp paired-end reads.

Raw reads were filtered to remove adapters and low-quality and ambiguous bases. Then paired-end reads were added to tags by the Fast Length Adjustment of Short reads program (FLASH, v1.2.11) to get the tags. All sequences were aligned and clustered into operational taxonomic units (OTUs) with 97% similarity using UPARSE software (v7.0.0.1090), and chimeric sequences were compared with the Gold database using UCHIME (v4.2.40) to detect. Then, the OTU representative sequences were taxonomically classified using the Ribosomal Database Project (RDP) classifier v.2.2, with a minimum confidence threshold of 0.6, and processed using QIIME (v1.8.0) on the Greengenes database v201305.

The optimized sequences were mapped back to the OTU representative sequences using the USEARCH_global method, and the abundance of OTU sequences in each sample was obtained. Alpha diversity were estimated by MOTHUR (v1.31.2) at the OTU level.

Statistical analysis comparing non-categorical or ordinal variables between groups was performed with Mann-Whitney U test, while Chi-square test was used to compare categorical variables between groups. To explore the difference in microbiota before and after anti-TB treatment, paired. T test was performed. The foldchanges of microbiota during anti-TB treatment were calculated, which were used to evaluate the associations between dynamic changes in microbiota and treatment outcomes. All analyses were conducted with a two-sided p-value of 0.05.