Plasma exosomal miR-122-5p_R-1, miR-23b-3p_R + 1, and miR-15a-5p_R-1 are associated with multidrug-resistant tuberculosis

Abstract

Background

Patients with multidrug-resistant tuberculosis (MDR-TB) who are resistant to at least both rifampicin and isoniazid, lack effective treatment options in clinic. The gold standard for the diagnosis of MDR-TB is drug sensitivity test, which is time-consuming and has a relatively low positive detection rate. Screening early diagnostic biomarker for MDR-TB is urgent need in clinical practice.

Methods

A total of 33 patients with MDR-TB, healthy controls and drug-sensitive tuberculosis (DS-TB) were included in this study. Total plasma exosomal RNA was extracted from the subjects, and the MDR-TB plasma-specific exosomal miRNAs were obtained by Illumina sequencing.

Results

There were 644 and 647 differentially expressed miRNAs in the plasma exosomes of MDR-TB patients obtained by sequencing and biogenic analysis compared with DS-TB patients and healthy controls, respectively. Differential miRNAs are mainly involved in the biological function of regulation of transcription and protein binding, and enriched in the pathways in cancer and MAPK signaling pathway. Moreover, seven plasma exosomal miRNAs in MDR-TB patients were significantly different from those in DS-TB patients and healthy controls. Among them, three of the miRNAs (hsa-miR-122-5p_R-1, hsa-miR-23b-3p_R + 1, and hsa-miR-15a-5p_R-1) were found to be in target relationship with MDR-TB related genes (NTRK2, KIDINS220, NCKAP1, MAPK9, NFAT5, ATF6 and SLC11A2) by target gene prediction analysis. Further the bioinformatic analysis showed that hsa-miR-122-5p_R-1 targets the protein PGLYRP2, a diagnostic biomarker identified in our previous study.

Conclusions

We suggest that hsa-miR-122-5p_R-1, hsa-miR-23b-3p_R + 1, and hsa-miR-15a-5p_R-1 are closely related to MDR-TB.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12879-025-11487-0.

Introduction

Pulmonary tuberculosis (TB), caused by mycobacterium tuberculosis (MTB), is a contagious disease. The development of drug resistance, especially multidrug-resistant tuberculosis (MDR-TB), significantly increases the challenges associated with its diagnosis and treatment. In 2023, there were approximately 400,000 new cases of multidrug-resistant/Rifampicin-resistant tuberculosis (MDR/RR-TB), accounting for 3.7% of all newly diagnosed TB patients [1]. Currently, the gold standard for diagnosis of MDR-TB is drug susceptibility testing (DST) for Mycobacterium tuberculosis complex (MTBC). However, DST takes 1–6 weeks, delaying timely treatment [2–5]. Alternative methods each have limitations: the redox-indicator methods are prone to bacterial contamination, with high false positives and low specificity [6]; phage amplified biologically assay (PhaB) suffer from poor specificity [7]; PCR-restriction fragment length (PCR-RFLP) technology can only detect two mutation loci [8, 9]; and the WHO-recommended Xpert MTB/RIF can only identify rifampicin resistance and shows reduced sensitivity in smear-negative and pediatric cases [10]. Line probe assay (LPA) can detect resistance to rifampicin and isoniazid but are primarily limited to smear-positive patients and may still yield false negatives in some cases [11–13]. Overall, current diagnostic tools for MDR-TB remain constrained by long turnaround times, poor sensitivity and specificity, and limited applicability, particularly in smear-negative patients—who comprise up to 85.65% of MDR-TB cases [14]. Therefore, there is an urgent need for rapid, sensitive, and specific molecular biomarkers to improve MDR-TB diagnosis in clinical settings.

Exosomes are 40–100 nm vesicles secreted by cells and present in various body fluids, including plasma and serum [7]. In vitro studies suggest that exosomes from MTB-infected cells participate in host immune regulation [15]. Singh [16] showed that exosomes contribute to macrophage migration and cytokine release following MTB infection. Bhatnagar [17] found that exosomes from infected macrophages can induce pro-inflammatory responses and playing a role in immune surveillance. Exosomes carry non-coding RNAs, mRNAs and proteins with biological relevance, and exosomal miRNAs have shown diagnostic potential in several diseases due to their stability in circulation. Madhavan [18] identified serum exosomal miR-1246, miR-4644, and others as diagnostic biomarkers for pancreatic cancer. Tan [19] found that plasma exosomal miR-223, miR-339 and miR-21 can serve as novel biomarkers for the early atherosclerotic thrombosis. Kim [20] reported miR-619-5p as a diagnostic indicator for non-small cell lung cancer (NSCLC). In TB, exosomal miRNAs like miR-766-3p, miR-376c-3p, miR-1283, and miR-125a-5p have shown good diagnostic value (AUC: 0.80 to 0.89) [21]. In HIV/TB co-infected, elevated levels of miR-20a, miR-26a, and others have been noted [22]. These miRNAs may affect TB pathogenesis by modulating immune-related pathways, including MAPK and PI3K-Akt [23, 24]. However, studies on exosomal miRNAs in MDR-TB are scarce; only one report to date has explored their role, focusing on treatment monitoring rather than diagnosis [25].

In this study, we isolated plasma exosomal miRNAs from MDR-TB patients and compared them with drug-sensitive tuberculosis (DS-TB) patients and healthy controls. Seven miRNAs specifically expressed in MDR-TB were obtained, among which three (hsa-miR-23b-3p_R + 1, hsa-miR-15a-5p_R-1, and hsa-miR-122-5p_R-1) were found to target MDR-TB-related genes. Therefore, we suggest that these three plasma exosomal miRNAs may be associated with MDR-TB.

Materials and methods

Patients and health controls

TB diagnostic criteria: According to the diagnostic criteria for tuberculosis issued by the Ministry of Health of China, patients who meet one of the following criteria can be diagnosed as TB and were included in the study [26]: (1) Positive acid-fast bacillus test (including sputum smear or culture); (2) Negative tuberculosis sputum test, but chest X-ray and CT scan examination with typical manifestations of active TB findings; (3) Pulmonary pathological lesions confirm tuberculosis through histological examination of lung tissue; (4) In patients clinically suspected of having TB, other pulmonary diseases have been excluded after follow-up imaging (e.g., serial chest X-rays), and the diagnosis of TB is supported by the evolution of radiographic findings.

MDR-TB patients were confirmed with DST for MTBC. Inclusion criteria for MDR-TB patients: At least twice positive test results for acid-fast bacillus in smear, and the bacterial type identification present as mycobacterium tuberculosis; MTBC test result showed resistance to both H (isoniazid) and R (rifampicin) through liquid DST. Based on these criteria along with comprehensive clinical evaluation (including medical history, physical examination, and imaging findings), the patients were confirmed as MDR-TB cases and enrolled in the study.

All MDR-TB and DS-TB patients included in this study were newly diagnosed and treatment-naïve. Plasma samples were collected from MDR-TB patients, DS-TB patients, and healthy controls (healthy blood donors). All the subjects were aged 18–65 years old, HIV negative, and with normal liver and kidney function. Patients with hepatitis B, diabetes, asthma, history of tuberculosis, other congenital diseases, chronic inflammation, autoimmune disease, surgical history, or other diseases were excluded.

Fasting blood were collected in the morning, and the plasma samples were acquired in EDTA- anticoagulant tubes and centrifuged at 3000 rpm, 4 °C, for 10 min within 4 h. Plasma samples were dispensed into sterile centrifuge tubes and stored in refrigerator at −80 °C.

A total of 33 samples (Table 1) were collected in this study. The research was approved by the Ethics Committee of South China University of Technology School of Medicine (2018-No.5) and the Ethics Committee of Zhejiang University School of Medicine (2017-No.023), and written informed consent was obtained from all subjects before blood sampling. All the procedures were performed in compliance with the Helsinki Declaration.

Table 1.

Characteristics of MDT-TB patients compared with TB patients and healthy controls

	Healthy Control (N = 11)	TB (N = 11)	MDR-TB (N = 11)	P Value
Age, age range (Mean ± SD)	45.36 ± 17.85	47.55 ± 17.44	51.27 ± 13.72	0.7856 a
Gender: female, no. (%)	5(45.45)	5(45.45)	3(27.27)	0.8801 b
Positive sputum smears, no. (%)	ND	7(63.64)	6(54.55)	/
Positive sputum culture, no. (%)	ND	4(36.36)	6(54.55)	/
Resistant to INH by DST, no. (%)	ND	0(0)	11(100)	/
Resistant to RFP by DST, no. (%)	ND	0(0)	11(100)	/
TC (mmol/L)	4.45 ± 1.02	4.18 ± 1.07	4.04 ± 0.93	0.5982 a
TG (mmol/L)	1.02 ± 0.28	1.10 ± 0.63	0.99 ± 0.23	0.5703 a
HDL (mmol/L)	1.36 ± 0.15	1.26 ± 0.14	1.36 ± 0.16	0.1989 a
LDL (mmol/L)	2.68 ± 0.49	2.81 ± 0.43	2.72 ± 0.34	0.8186

DST Drug susceptibility testing, INH Isoniazide, RFP Rifampicin, TC Total cholesterol, TG Triglyceride, HDL High-density lipoprotein, LDL Low-density lipoprotein, N number of subjects, ND Not determined

^a P-value for Kruskal-Wallis test, ^b P-value for chi-square test

*P < 0.05. **P < 0.01. *** P < 0.001

Plasma exosome extraction and exosomal RNA isolation and purification

Plasma exosomes and exosome-derived total RNA were extracted and purified from MDR-TB patients, DS-TB patients, and healthy controls using the SeraMir™ Exosome RNA Amplification Kit (SBI). For each subject, 500 µL of plasma sample was mixed with 120 µL ExoQuick reagent by inverting the mixture three times. The mixture was then incubated at 4 °C for 30 min, followed by centrifugation at 13,000 rpm for 2 min. After discarding the supernatant, the pellet containing the exosomes was retained. Subsequently, 350 µL of solution was added to the pellet, vortexed for 15 s, and allowed to stand at room temperature for 5 min until completely dissolved. Next, 200 µL of anhydrous ethanol was added, vortexed for 10 s, and the mixture was transferred to the spin-collection tube provided with the kit. This was centrifuged at 13,000 rpm for 1 min, and the flow-through was discarded. The column was then washed by adding 400 µL of washing solution, centrifuging at 13,000 rpm for 1 min, and discarding the effluent. This washing step was repeated once more (centrifuging at 13000 rpm for 1 min and discarding the effluent), followed by an additional centrifugation at 13,000 rpm for 2 min, and dry the column. Finally, the collection tube was replaced with a 1.5 mL RNase-free centrifuge tube, and the exosomal RNA was eluted using 30 µL of elution solution. Elution was achieved by centrifuging at 2000 rpm for 2 min, and the exosomal RNA was collected by a final centrifugation of the eluate at 13,000 rpm for 1 min.

Small RNA sequencing library preparation and Illumina sequencing

Small RNA sequencing library of each subject was prepared by using the TruSeq Small RNA Sample Prep Kits (Illumina, San Diego, USA). According to the characteristics of miRNA that the 5’ end is a phosphate group and the 3’ end is a hydroxyl group, a 3’ adapter and 5’ adapter of adenylated single-stranded DNA are attached to the small RNA successively by using T4 RNA ligase 2. Among them, the 5’ adapter of the 3’ end is rAPP, and the 5’ end can capture small RNAs with 5’ phosphate groups. T4 RNA ligase 2 requires pre-adenylated adapters during ligation reactions without ATP, to reduce self-association between miRNAs and the adapter sequences. The small RNA sequences of 5’ and 3’ adapters were reverse transcribed with RT primers complementary to the 3’ end; then the cDNA sequences generated by reverse transcription were amplified by PCR; then the PCR products were recovered in 140–160 bp length to complete the construction of the small RNA library. Finally, the constructed small RNA library was sequenced using the Illumina Hiseq4000 with a single-read of 1*50 bp length.

Bioinformatics analysis

We performed a rigorous quality control process on the raw small RNA sequencing data. Adapter sequences were trimmed, and reads shorter than 18 nucleotides were removed. Sequences with atypical miRNA characteristics generated in sample preparation and ligation adapters, as well as impure sequences generated by optical digital processing during sequencing, were removed by miRNA data analysis software ACGT101-miR (LC Sciences, Houston, Texas, USA). Next, we aligned the reads against multiple reference databases—including mRNA, Rfam (for filtering out non-miRNA small RNAs such as rRNA, tRNA, snoRNA, and snRNA), and Repbase (to remove repeat-associated sRNAs)—to eliminate non-miRNA sequences (Supporting information Figure S1-S2). Most of the small RNAs with 20–24nt length were screened, consistent with the typical size range of Dicer enzyme products (Supporting information Figure S3). Then, by comparing analysis in the miRBase21.0 database with Bowtie software, the known human miRNAs were identified, and new miRNA sequences were discovered. The results with relatively high confidence were selectively retained according to the genetic background of the species of tested samples. Comparative quantitative analysis of exosomal miRNA expression levels between MDR-TB patients, DS-TB patients, and healthy controls was performed, to obtain the relative quantitative ratios of exosomal miRNAs between the experimental groups. miRNAs were filtered based on fold change (fold change ≥ 1.20 for upregulation or ≤ 0.80 for downregulation) and statistical significance (P ≤ 0.05).

The target genes of the obtained differential exosomal miRNAs were predicted by searching in the TargetScan database and miRanda database. Then, Gene ontology (GO) was analyzed online (http://www.geneontology.org) to determine the functions annotation of differential miRNAs, and miRNAs were classified into three categorizations based on their biological process, cellular component, and molecular function. The pathway enrichment information of differential exosomal miRNA was obtained by KEGG analysis, and the recognition sequences of miRNAs and their target genes were predicted through sequence comparison and bioinformatics analysis. In addition, genes that are associated with MDR-TB were obtained in The Human Gene Database (GeneCards).

Statistical analysis

The clinical data was analyzed with GraphPad Prism software 10.0, and the two-tailed P values < 0.05 were considered statistically significant. Parametric data were analyzed by T-test (between two groups), One-way ANOVA (between three or more groups) for means, and chi-square test for the composition ratios. Nonparametric data were analyzed using the Mann-Whitney U test (between two groups) and the Kruskal-Wallis test (between three or more groups). The study samples provided at least 84.32% power to identify significant differences between MDR-TB patients, DS-TB patients and healthy controls at a statistical support level of α = 0.05 with an effect size f of 0.6 applying a post hoc power analysis calculated by Gpower3.1.9.

Results

Clinical characteristics of MDR-TB patients

A total of 33 patients were enrolled in this research, including 11 healthy controls (mean age 45.36 ± 17.85 years), 11 DS-TB patients (mean age 47.55 ± 17.44 years), and 11 MDR-TB patients (mean age 51.27 ± 13.72 years) from the Shaoxing Municipal Hospital and Zhejiang Hospital in 2015–2018. Females account for 45.45% of healthy controls, 45.45% of DS-TB patients, and 27.27% of MDR-TB patients. The clinic data, including positive sputum smears, positive sputum culture, resistance to INH by DST, resistance to RFP by DST, total cholesterol (TC), Triglyceride (TG), high-density lipoprotein (HDL), low-density lipoprotein (LDL) was analyzed and shown in Table 1.

Transcriptomic results

The sequencing results of 11 samples from each group were combined for analysis. It is shown that the expression levels of plasma exosomal miRNAs from DS-TB patients were significantly different with that of healthy controls. However, there were fewer exosomal miRNAs differentially expressed in the MDR-TB patients compared with healthy controls. While, the expression levels of exosomal miRNAs between MDR-TB patients and DS-TB patients showed significant differences (see Fig. 1 for the volcano map of differential genes).

Fig. 1.

Volcano plots and histogram of differentially expressed exosomal miRNAs in MDR-TB, DS-TB, and healthy control groups

The statistical analysis of sequencing data was conducted based on differential expression analysis using fold change and P value thresholds. Specifically: In the DS-TB vs. healthy control comparison, a total of 266 exosomal miRNAs were detected. Among these, 122 miRNAs were differentially expressed at P ≤ 0.05, including 93 upregulated and 29 downregulated miRNAs. At a stricter significance level (P ≤ 0.01), 70 miRNAs were upregulated and 19 miRNAs were downregulated. In the MDR-TB vs. healthy control comparison, a total of 647 exosomal miRNAs were detected. Among them, 14 miRNAs showed differential expression at P ≤ 0.05 (6 upregulated, 8 downregulated), and 3 miRNAs were significantly downregulated at P ≤ 0.01. In the MDR-TB vs. DS-TB comparison, a total of 644 exosomal miRNAs were identified. Among them, 98 miRNAs were differentially expressed at P ≤ 0.05, including 26 upregulated and 72 downregulated. At P ≤ 0.01, 16 miRNAs were upregulated and 51 miRNAs were downregulated.

Functional classification and pathway enrichment maps of MDR-TB differential exosome miRNAs were obtained

Bioinformatics analysis of the differentially expressed exosomal miRNAs, including gene cluster analysis and regulatory gene prediction, were performed. Functional classification of differentially expressed exosomal miRNAs was obtained by GO annotations, and the pathway enrichment information was obtained by KEGG enrichment analysis (Fig. 2). Most of the differential exosomal miRNAs are enriched in the function of zinc ion binding, transcription (DNA-templated), and signal transduction (Fig. 2A). Moreover, KEGG enrichment analysis showed that differential exosomal miRNAs were mainly enriched in pathways in cancer, MAPK signaling pathway, and regulation of actin cytoskeleton, endocytosis pathways (Fig. 3A).

Fig. 2.

GO enrichment and annotation analysis of differentially expressed exosomal miRNAs

Fig. 3.

Pathway enrichment analysis of differentially expressed exosomal miRNAs using KEGG enrichment analysis. A, Pathway enrichment analysis of differentially expressed exosomal miRNAs; B, Predicted target region between hsa-miR-122-5p_R-1 and PGLYRP2 3’-UTR were analyzed and showed, and the seed region of the predicted target region was also showed. C-D, PPI networks between the candidate exosomal miRNAs of MDR-TB and their target genes, according to the functional classification of GO annotation (C) and KEGG pathway enrichment analysis (D). E, PPI network between the candidate exosomal miRNAs and the biomarker proteins of MDR-TB obtained in our previous researches. PPI: protein protein interaction

Candidate exosomal miRNAs for MDR-TB

A total of seven significantly differential exosomal miRNAs (hsa-miR-92a-3p, hsa-miR-15a-5p_R-1, hsa-miR-122-5p_R-1, hsa-miR-15b-5p, hsa-let-7 g-5p, hsa-miR-23b-3p_R + 1 and hsa-miR-4433b-5p_R + 1) for MDR-TB were obtained compared with both DS-TB and healthy controls in this study (Table 2). Among them, the expression levels of hsa-miR-23b-3p_R + 1 and hsa-miR-4433b-5p_R + 1 in the plasma exosomes of MDR-TB patients were significantly lower than those in DS-TB patients and healthy controls; and the expression levels of hsa-miR-122-5p_R-1 in MDR-TB patients were significantly higher compared with DS-TB patients and healthy controls. Morever, the expression levels of hsa-miR-92a-3p, hsa-miR-15a-5p_R-1 and hsa-miR-15b-5p in MDR-TB patients were significantly lower than healthy controls, but significantly higher than DS-TB patients. In addition, the expression levels of hsa-let-7 g-5p in MDR-TB patients were significantly higher than healthy controls, but significantly lower than DS-TB patients.

Table 2.

MDR-TB differential exosome MiRNAs

MiRNA name	MiRNA sequence (5’−3’)	P value (t test)	MDR-TB Vs Healthy control	MDR-TB Vs DS-TB
hsa-miR-23b-3p_R + 1	ATCACATTGCCAGGGATTACCACT	3.46E-02	Down	Down
hsa-miR-4433b-5p_R + 1	ATGTCCCACCCCCACTCCTGTT	4.68E-02	Down	Down
hsa-miR-122-5p_R-1	TGGAGTGTGACAATGGTGTTT	2.01E-02	Up	Up
hsa-miR-92a-3p	TATTGCACTTGTCCCGGCCTGT	4.99E-03	Down	Up
hsa-miR-15a-5p_R-1	TAGCAGCACATAATGGTTTGT	1.67E-02	Down	Up
hsa-miR-15b-5p	TAGCAGCACATCATGGTTTACA	2.23E-02	Down	Up
hsa-let-7 g-5p	TGAGGTAGTAGTTTGTACAGTT	2.86E-02	Up	Down

MDR-TB Multidrug-resistant tuberculosis, DS-TB Drug-sensitive tuberculosis

Three miRNAs may be associated with MDR-TB

Functional association of the seven candidate exosomal miRNAs and their predicted target genes of MDR-TB were analyzed according to the bioinformatics results of GO annotation and KEGG pathway enrichment analysis. GO analysis showed that the target genes of seven differentially expressed exosomal miRNAs were significantly enriched in biological processes such as transcriptional regulation, transcription and signal transduction. These target genes were mainly localized in the nucleus, cytoplasm and cytoplasmic matrix. At the molecular function level, target genes with protein-binding, metal ion-binding and DNA-binding activities had the highest percentage (Fig. 2B). We retrieved genes associated with MDR-TB from GeneCards and performed a Venn analysis with the predicted target genes of differentially expressed exosomal miRNAs identified in MDR-TB. This analysis yielded eight overlapping target genes (NTRK2, KIDINS220, NCKAP1, MAPK9, NFAT5, ATF6 and SLC11A2) potentially involved in MDR-TB. The corresponding miRNAs were hsa-miR-23b-3p_R + 1, hsa-miR-15a-5p_R-1, and hsa-miR-122-5p_R-1 (Fig. 3C and D). Meanwhile, we analyzed the functional association of the MDR-TB biomarker proteins, sCD14, PGLYRP2, FGA, TGFBI, PCSK9, and CCL14 obtained in our previous researches, with the seven candidate exosomal miRNAs of MDR-TB, and found that the potential biomarker protein PGLYRP2 of MDR-TB was the target gene of hsa-miR-122-5p (Fig. 3E). The predictive recognition sequences between candidate exosomal miRNAs (hsa-miR-23b-3p_R + 1, hsa-miR-15a-5p_R-1 and hsa-miR-122-5p_R-1) with their target genes (NTRK2, KIDINS220, NCKAP1, MAPK9, NFAT5, ATF6 and SLC11A2) related to MDR-TB and biomarker protein PGLYRP2 of MDR-TB were obtained by bioinformatics analysis (Fig. 3B, Figure S4-S6).

Searching in the GeneCards (THE HUMAN GENE DATABASE) database to screen the target genes related to MDR-TB. The result of sequence comparison showed that hsa-miR-23b-3p_R + 1 has one more nucleobase (T) compared with the hsa-miR-23b-3p in the 3’ end sequence; and the hsa-miR-15a-5p_R-1 has one nucleobase (G) less than the hsa-miR-15a-5p in the 3’ end sequence, and the hsa-miR-122-5p_R-1 has one nucleobase (G) less than the hsa-miR-122-5p in the 3’ end sequence. While, hsa-miR-4433b-5p_R + 1 has one more nucleobase (T) than the hsa-miR-4433b-5p in the 3’ end sequence (Table 3).

Table 3.

Sequence comparison of the MiRNAs

MiRNA name	MiRNA sequence (5’−3’)
hsa-miR-15a-5p	TAGCAGCACATAATGGTTTGTG
hsa-miR-15a-5p_R-1	TAGCAGCACATAATGGTTTGT
hsa-miR-23b-3p	ATCACATTGCCAGGGATTACCAC
hsa-miR-23b-3p_R + 1	ATCACATTGCCAGGGATTACCACT
hsa-miR-122-5p	TGGAGTGTGACAATGGTGTTTG
hsa-miR-122-5p_R-1	TGGAGTGTGACAATGGTGTTT
hsa-miR-4433b-5p	ATGTCCCACCCCCACTCCTGT
hsa-miR-4433b-5p_R + 1	ATGTCCCACCCCCACTCCTGTT

Correlation analysis between the candidate exosomal miRNAs with clinical data

Furthermore, spearman correlation between clinical characteristics with differentially expressed exosomal miRNAs, and between different exosomal miRNAs, were analyzed. The results showed significant negative correlations between TG with hsa-miR-15a-5p_R-1 (rs = −0.7297, P = 0.0108), and between TG with hsa-miR-15b-5p (rs = −0. 6414, P = 0.0334). In addition, hsa-miR-92a-3p showed significant positive correlations with hsa-miR-15a-5p_R-1 (rs = 0.8277, P = 0.0017) and hsa-miR-15b-5p (rs = 0.8523, P = 0.0009), significant negative correlations with hsa-let-7 g-5p (rs = −0.7155, P = 0.0133), hsa-miR-23b-3p_R + 1 (rs = −0.8000, P = 0.0031) and hsa-miR-4433b-5p_R + 1 (rs = −0.6570, P = 0.0281). Hsa-miR-15a-5p_R-1 showed significant positive correlations with hsa-miR-15b-5p (rs = 0.9533, P = 5.6060E-06), and negative correlations with hsa-let-7 g-5p (rs = −0.7368, P = 0.0097) and hsa-miR-4433b-5p_R + 1 (rs = −0.7175, P = 0.0129). Hsa-miR-15b-5p showed significant negative correlations with hsa-let-7 g-5p (rs = −0.7039, P = 0.0156), hsa-miR-23b-3p_R + 1 (rs = −0.7295, P = 0.0108) and hsa-miR-4433b-5p_R + 1 (rs = −0.6951, P = 0.0176). Hsa-let-7 g-5p showed significant positive correlations with hsa-miR-23b-3p_R + 1 (rs = 0.6702, P = 0.0240) and hsa-miR-4433b-5p_R + 1 (rs = 0.7780, P = 0.0048) (Table S1).

Discussion

In this study, illumina high-throughput sequencing was applied to screening plasma exosomal miRNA biomarkers for MDR-TB patients compared with DS-TB patients and Healthy controls. A total of 647 and 644 plasma exosomal miRNAs were obtained from MDR-TB patients compared with DS-TB patients and healthy controls, respectively. Among them, seven miRNAs (hsa-miR-23b-3p_R + 1, hsa-miR-4433b-5p_R + 1, hsa-miR-122-5p_R-1, hsa-miR-92a-3p, hsa-miR-15a-5p_R-1, hsa-miR-15b-5p and hsa-let-7 g-5p) in MDR-TB patients were significant differentially expressed with both DS-TB patients and healthy controls. It was also showed that the target genes of the three miRNAs (hsa-miR-23b-3p_R + 1, hsa-miR-15a-5p_R-1 and hsa-miR-122-5p_R-1) were closely related to MDR-TB by bioinformatics analysis.

Sequence analysis showed the three miRNAs (hsa-miR-23b-3p_R + 1, hsa-miR-15a-5p_R-1 and hsa-miR-122-5p_R-1) isomiRs differ slightly at the 3′ end from their canonical miRNAs, consistent with previously reported isomiRs [27–29]. IsomiRs are natural variants generated through processes like imprecise cleavage, 3′ additions, RNA editing, or SNPs [30, 31]. Studies indicate that most of isomiRs regulate the same target mRNAs with the same seed sequences, they may differ in abundance, stability, and potentially affecting their regulatory activity [32, 33]. Emerging evidence suggests isomiRs can act as disease-specific biomarkers with distinct biological roles [30].

Hsa-miR-122-5p is involved in lipid metabolism and drug resistance. It is upregulated in fatty liver patients and correlated with lipoprotein levels [34]. It has also been implicated in drug resistance in cancers via regulation of Bcl-2, CDKs, and CLIC1 [35, 36], and associated with chemosensitivity by targeting ASCT2 and DUSP4 [37, 38]. In MDR-TB, we found the exsomal hsa-miR-122-5p_R-1 significantly upregulated. Its predicted target gene, PGLYRP2—a known MDR-TB biomarker—plays a role in immune activation through TLR2 and TLR4 [39, 40]. Despite lacking one nucleotide compared to the canonical form, hsa-miR-122-5p_R-1 may contribute to abnormal lipid metabolism and drug resistance in MDR-TB patients.

MiR-23b-3p has been reported to be involved in lung diseases and drug resistance. It has been shown that hsa-miR-23b-3p is elevated in cancer stem cells in lung cancer [41]. It is also significantly increased in extracellular vesicle of granuloma patients, but the free hsa-miR-23b-3p in plasma were significantly decreased than that of lung adenocarcinoma patients [42]. MiR-23b-3p is also reported to be involved in drug resistance in a variety of diseases through pathways involving ATG12, GLS1, and MALAT1 [43–46]. SNHG16 similarly promotes resistance through miR-23b-3p modulation [47]. Currently, studies on the mechanism of miR-23b-3p resistance have focused on drug resistance in various cancers, and there is no study related to TB. Our data showed that exsomal hsa-miR-23b-3p_R + 1 is significantly downregulated in MDR-TB and targets several MDR-TB-related genes (NTRK2, KIDINS220, NCKAP1, MAPK9, NFAT5, ATF6, and SLC11A2), suggesting a potential role in pathogenesis with MDR-TB.

MiR-15a-5p has been reported to be involved in lung diseases and drug resistance [48, 49]. It is downregulated in NSCLC [50], chronic lymphocytic leukemia [51, 52], pituitary adenoma [53], prostate cancer [54] and ovarian cancer [55], and increased in acute promyelocytic leukemia [56, 57]. It influences chemoresistance through regulation of PDCD4, BTG2, and autophagy [58, 59], and through regulated by cirRNA and HIF-1α [60, 61]. In this study, exosomal hsa-miR-15a-5p_R-1 in MDR-TB patients was significantly lower than healthy controls, but significantly higher than DS-TB patients, and negatively correlated with TG levels, linking it to lipid metabolism. The target genes (NTRK2, KIDINS220, NCKAP1, MAPK9, NFAT5, ATF6 and SLC11A2) of hsa-miR-15a-5p_R-1 were related to MDR-TB, reinforcing its association with MDR-TB.

Among the miRNAs identified, hsa-miR-92a-3p, hsa-let-7 g-5p, hsa-miR-15b-5p, and hsa-miR-4433b-5p_R + 1 were significantly altered in MDR-TB patients. Notably, hsa-miR-15b-5p showed a negative correlation with triglyceride levels. However, current bioinformatic analysis revealed limited evidence linking these miRNAs directly to MDR-TB, and their biological roles require further investigation.

Conclusions

In summary, seven MDR-TB-associated exosomal miRNAs (hsa-miR-92a-3p, hsa-miR-15a-5p_R-1, hsa-miR-122-5p_R-1, hsa-miR-15b-5p, hsa-let-7 g-5p, hsa-miR-23b-3p_R + 1, and hsa-miR-4433b-5p_R + 1) were identified by sequencing. Among them, hsa-miR-122-5p_R-1, hsa-miR-23b-3p_R + 1, hsa-miR-15a-5p_R-1 were predicted to target genes associated with MDR-TB (NTRK2, KIDINS220, NCKAP1, MAPK9, NFAT5, ATF6 and SLC11A2). Importantly, hsa-miR-122-5p_R-1 also targets PGLYRP2, a potential early diagnostic biomarker identified in our previous study, suggesting these miRNAs may play important roles in MDR-TB pathogenesis. A key limitation is the insufficient exosomal RNA quantity for qRT-PCR validation of sequencing results. Additionally, the biological relevance of the identified miRNAs to MDR-TB requires further investigation in larger cohorts.

Abbreviations

MDR-TB

Multidrug-resistant tuberculosis

DS-TB

Drug-sensitive tuberculosis

MTB

Mycobacterium tuberculosis

MDR/RR-TB

Multidrug-resistant/Rifampicin-resistant tuberculosis

DST

Drug susceptibility testing

MTBC

Mycobacterium tuberculosis complex

PhaB

The phageamplified biologically assay

PCR-RFLP

PCR-restriction fragment length

LPA

Line probe assay

MTB

Mycobacterium tuberculosis

NSCLC

Non-small cell lung cancer

MGIT

Mycobacterial Growth Indicator Tube

TC

Total cholesterol

TG

Triglyceride

HDL

High-density lipoprotein

LDL

Low-density lipoprotein

Authors’ contributions

TT J, JC L and KP X designed the study, writing the manuscript; TT J, YZ, LY Z, LG, YT H, HH contributed to data collection and data management, and data analysis; LY S and LL W contributed to the diagnosis and inclusion of all the subjects; TT J, LY S, JC L, and KP X contributed to the funding of the study.

Funding

This work was supported by Guangzhou Ruiqian Biological Technology Co., LTD (grant number 20230330); and The Zhejiang Province Traditional Chinese Medicine Science and Technology Plan Project (grant number 2025072985).

Data availability

RNA-Seq raw data that support the findings of this study have been deposited in the Sequence Read Archive (SRA) repository, the Persistent WEB LINK TO DATASETS is: https://www.ncbi.nlm.nih.gov/sra/?term=, and the BioSample accession TO DATASETS is PRJNA1213895.

Declarations

Ethics approval and consent to participate

The research was approved by the Ethics Committee of South China University of Technology School of Medicine (2018-No.5), and the Ethics Committee of Zhejiang University School of Medicine (2017-No.023), and written informed consent was obtained from all subjects before blood sampling. All procedures were in accordance with the ethical standards of the Committee responsible for human experimentation (institutional and national) and followed the Declaration of Helsinki.

Consent for publication

Written informed consent for publication was acquired from the patients for the publication of the identity revealing information.

Competing interests

The authors declare no competing interests.