Direct detection from sputum for drug-resistant Mycobacterium tuberculosis using a CRISPR-Cas14a-based approach

Abstract

The increasing prevalence of multidrug-resistant tuberculosis (MDR-TB) highlights the urgent need for an efficient approach to identify Mycobacterium tuberculosis complex (MTBC) strains resistant to rifampicin (RIF) and isoniazid (INH). In response, we developed a CRISPR-Cas14a MTB RIF/INH platform that can detect the most common mutations associated with RIF and INH resistance. To evaluate the sensitivity and specificity of our CRISPR-Cas14a MTB RIF/INH platform, we carried out a comprehensive assessment using clinical isolates of M. tuberculosis and sputum samples from TB patients, making direct comparisons with phenotypic drug susceptibility testing (pDST). A total of 60 clinical isolates from TB patients were utilized, consisting of 18 RIF mono-resistant, 15 INH mono-resistant, 24 MDR isolates, and 3 fully susceptible isolates. Among the 42 RIF-resistant isolates, our platform accurately identified 39, achieving a sensitivity of 93.3% (95% CI, 80.0-98.5) and a specificity of 100% (95% CI, 81.6–100). Similarly, out of the 39 INH-resistant isolates, the platform successfully identified 38, demonstrating a sensitivity of 97.5% (95% CI, 86.5–99.9) and a specificity of 100% (95% CI, 83.8–100) when compared with pDST. Moreover, in the analysis of 55 sputum samples, our platform accurately identified RIF resistance in 10 out of 12 samples (85.7%) and INH resistance in all 11 samples (100%). Notably, excluding the nucleic acid extraction step, the entire testing procedure can be completed in approximately 1.5 h. These results suggest that the CRISPR-Cas14a MTB RIF/INH platform is a reliable and promising novel tool for detecting RIF and INH resistance in isolates or directly from sputum samples.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12866-025-03899-4.

Introduction

It is estimated that approximately 400,000 new patients were diagnosed with rifampicin-resistant tuberculosis (RR-TB) or multi-drug resistant tuberculosis (MDR-TB) in 2023. However, the global treatment coverage rate for MDR/RR-TB in 2023 was merely 44% (175,923/400,000) [1]. The emergence of resistance to the two most effective first-line anti-tuberculosis drugs, RIF and INH, is a significant concern. The global treatment success rate for drug-resistant TB remains relatively low, at 68%. Therefore, accurate and timely diagnosis is crucial for initiating appropriate treatment strategies and curbing the spread of the disease.

Conventional phenotypic drug-susceptibility tests (pDST) are the gold standard for determining the drug susceptibility in cultures. Nevertheless, they are time-consuming, often taking up to three months to generate results. Molecular testing has emerged as a highly promising approach to bridge the diagnostic gap, offering significantly faster results compared to pDST. The World Health Organization (WHO) Consolidated Guidelines on Tuberculosis (Module 3: Diagnosis - Rapid Diagnostics for Tuberculosis Detection) recommend three primary molecular methods for initial TB diagnosis with concurrent drug-resistance detection: (1) Xpert MTB/RIF and Xpert MTB/RIF Ultra assays (Cepheid, USA), (2) Truenat MTB, MTB Plus, and MTB-RIF Dx assays (Molbio Diagnostics, India), and (3) moderate complexity automated nucleic acid amplification tests (NAATs) [2]. These WHO-endorsed rapid diagnostic platforms for TB and drug-resistant TB detection are grounded in advanced automated technologies that utilize nucleic acid hybridization principles. The guidelines offer stratified recommendations tailored to different TB types, specific patient populations, and diagnostic test categories, each supported by varying levels of evidence and corresponding recommendation grades.

In recent years, CRISPR-Cas-based biosensors have emerged as promising tools for nucleic acid detection, owing their high sensitivity, specificity, and reliability [3, 4]. The efficacy of these biosensors stems from the ability of CRISPR RNA (crRNA) or guided RNA (gRNA) to specifically bind to target sequences, triggering Cas enzymes to cleave both target-specific and nonspecific sequence [5–8]. CRISPR-Cas12a and Cas13a have been reported for detecting drug-resistant mutations in Mtb, yet their limitations in detecting a wide range of mutation sites hinder their practical utility [9, 10]. The FLASH-TB method, based on CRISPR-Cas9, can detect mutations in Mtb associated with resistance to first- and second-line drugs directly from sputum. However, it necessitates next-generation sequencing, leading to increased costs and longer testing times [11]. Cas14a, an RNA-guided programmable nuclease, presents a promising alternative for detecting drug-resistant mutations in Mtb. It can target single-strand DNA (ssDNA) without the requirement of a Protospacer Adjacent Motif (PAM) and demonstrates highly specific and sensitive recognition of ssDNA. This make it valuable for various applications such as detecting single-nucleotide polymorphism (SNP) genotyping and pathogen discrimination [12, 13]. Moreover, Cas14a has a molecular weight only half that of Cas12a, making it more suitable for expression in E. coli.

The majority of RIF-resistant MTBC strains carry mutations in the 81-bp region of the rpoB gene, known as the RIF Resistance-Determining Region (RRDR), particularly at codons 516, 526, and 531 (using E. coli numbering) [14–17]. Approximately 70–90% of MTBC strains resistant to INH have mutations either at codon 315 of the katG gene or at positions − 8 and − 15 of the inhA promoter region [14–17]. Given the potential advantages of CRISPR-Cas14a in SNP detection, our aim was to develop a CRISPR-Cas14a-based platform for the rapid detection of the most common mutations associated with RIF and INH resistance.

Methods and materials

Sample collection and pDST

A total of 60 clinical isolates were sourced from the Center for Tuberculosis Control of Guangdong Province (Guangzhou). These isolates consisted of 18 isolates resistant to RIF, 15 isolates resistant to INH, 24 isolates resistant to both RIF and INH, and 3 fully susceptible isolates. The isolates were cultured in Middlebrook 7H9 liquid medium and incubated at 37℃. Sputum samples were collected from suspected TB patients at Shenzhen Third People’s Hospital. The samples were decontaminated using the NALC-NaOH method, with a final NaOH concentration of 1%. After decontamination, the concentrated sediment was resuspended in 1.5 ml of sterile phosphate buffer (pH 6.8). Subsequently, smears were stained using the Ziehl-Neelsen method. Both solid and liquid culture media were inoculated to monitor the growth of the samples, and any remaining sediment from the decontaminated sputum samples was preserved at -20 °C. Once the cultures grew successfully, species identification and pDST were performed on MTBC isolates. pDST for RIF and INH was carried out using the BACTEC MGIT 960 (Becton Dickinson, USA) in accordance with the manufacturer’s instructions. The WHO recommended ratio method was employed for the DST of RIF and INH, with a critical concentration of 1 mg/L for RIF and 0.1 mg/L for INH.

Determination of gene mutations by PCR sequencing

DNA extraction from clinical isolates was performed using a Bacterial DNA Extraction Kit (Gene Optimal, Shanghai, China). The quality and concentration of the purified DNA were evaluated using the NanoDrop 2000 (Thermo Fisher, MA, United States). Additionally, DNA was extraction from sputum samples via a simple boiling method [18]. The RpoB, katG and inhA gene fragments from both clinical isolates and sputum samples were amplified by PCR using the primers listed in Supplementary Table S1. The resulting PCR products were analyzed by 1% agarose gel electrophoresis and then subjected to direct Sanger sequencing to identify specific gene mutations.

Protein production and purification

The Cas14a1 protein sequence was obtained from a previous study conducted by Lucas Harrington [12]. The Cas14a1 Open Reading Frame (ORF) was codon-optimized and synthesized by Sangon Biotech Co., Ltd (Shanghai, China). The Cas14a.1 ORF was cloned into the pET-28a overexpression vector and then transformed into chemically competent E. coli BL21(DE3). The transformed cells were incubated on LB agar plates containing 50 μg/mL Kanamycin at 37℃ for 14 h. Single colonies were selected and inoculated into 100 mL of LB medium, which was incubated overnight at 37℃ on a shaker at 250 rpm. The overnight culture was used to inoculate a fresh medium at a ratio of 1:100 and further incubated at 37℃ until reaching an OD600 of 0.6 ∼ 0.8. After being cooled on ice, 0.5 mM IPTG was added to induce gene expression and incubated overnight at 18℃. The cells were harvested by centrifugation and resuspended in a wash buffer (10 mM imidazole, 1 M NaCl, 50 mM NaH2PO4·2H2O, pH 7.9 ∼ 8.1). Subsequently, they were lysed by sonication and the supernatant was collected by centrifugation. Protein purification was performed using the Ni-NTA 6FF SefinoseTM resin kit (Sangon Biotech, Shanghai, China) following the manufacturer’s instructions. The purified proteins were desalted and concentrated using Amicon® Ultra-15 centrifugal filter concentrators (Millipore, MA, United States), and then stored at -80℃.

gRNA design and preparation

The gRNA spacers were designed to match the target mutations located in the middle of the spacers, as described in a previous study [12]. To prepare the gRNAs, transcription templates were synthesized with the T7 promoter and crRNA scaffold and then cloned into the pUC19 vector by Sangon Biotech Co., Ltd (Shanghai, China). The transcription templates were obtained by PCR amplification of the target pUC19 vector and then were purified using FastPure Gel DNA Extraction Mini Kit (Vazyme, Nanjing, China). All gRNAs were transcribed in vitro using the T7 High Yield RNA Transcription Kit (Vazyme, Nanjing, China) according to the manufacturer’s instructions. The resulting RNA was purified using VAHTS RNA Clean Beads (Vazyme, Nanjing, China) and quantified using NanoDrop 2000 (Thermo Fisher, MA, United States). The purified RNA was stored at -80℃. The ssDNA containing the target regions, which were used as substrates, was synthesized by Sangon Biotech (Shanghai, China). Specific sequence information can be found in Supplementary Table S1.

Multiplex PCR amplification

PCR primers for each target gene were designed using the Primer Premier V5.0 software. The primers were further analyzed for potential primer-dimer interactions during multiplex PCR. To protect the target strands from degradation by T7 exonuclease during the subsequent detection step, the first four 5’ nucleotides of the forward primers were modified with phosphorothioate (PT), while the reverse primer remained unmodified. A multiplex PCR Kit (Vazyme, Nanjing, China) was used for multiplex PCR amplification. The PCR reaction mixture consisted of 6 μl of primer mix (with a final concentration of 0.2 μM for each primer) and 2 μl of DNA template, in a final volume of 50 μl. The PCR thermocycling parameters were set as follows: an initial denaturation step at 95 ℃ for 3 min, followed by 35 cycles of 95 ℃ for 30 s, 60 ℃ for 30 s, and 72 ℃ for 1 min. The final extension step was performed at 72℃ for 10 min. After amplification, the PCR products were analyzed using 1% agarose gel electrophoresis. The PCR primers are listed in Supplementary Table S1.

CRISPR-Cas14a trans-cleavage assay

The CRISPR-Cas14a trans-cleavage assay was carried out according to the methodology described in a previous study [12]. Briefly, the assay was performed in NEbuffer4.0, consisting of 100 nM Cas14a, 100 nM gRNA, 100 nM FQ reporter, 1 U T7 exonuclease, and 200 nM synthesized ssDNA or 2 ul PCR products, in a total volume of 20 μl. The reaction was monitored at 37℃ for the specified time using a fluorescence plate reader (Varioskan Flash, Thermo Fisher Scientific, MA, USA). Fluorescence measurements were taken with an excitation wavelength of 492 nm and an emission wavelength of 520 nm.

Determination of limit of detection (LOD) and specificity

To determine the LOD, DNA was extracted from the H37Rv strain using the Bacterial DNA Extraction Kit (Gene Optimal, Shanghai, China) and quantified with Qubit assay. The number of DNA copies was calculated using the formula (6.02 × 1023) × (ng/μl × 10^− 9 )/(DNA length × 660) = copies/μl. The purified DNA was then serially diluted as indicated in the results. 2 μl of DNA from each diluted solution was used as template and six replicates were performed at every dose point. To determine the specificity, genomic DNA samples from seven mycobacterial reference strains, including M. tuberculosis H37Rv, M. abscessus (ATCC 19977), M. intracellulare (ATCC 13950), M. avium (ATCC 25291), M. kansasii (ATCC 12478), M. gordonae (ATCC 14470), and M. fortuitum (ATCC 6841) [19], as well as E. coli DH5α were used. For PCR amplification, 100 pg of DNA from each reference strain were used as a template.

Statistical analysis

The t-test was employed to evaluate significant differences between the mutation and wild type. A P-value less than 0.05 was considered statistically significant. P-values less than 0.0001, 0.001, 0.01, and 0.05 were denoted as ****, ***, **, and *, respectively. The bar plots represented the means ± SD. The 95% confidence intervals for simple proportions were calculated using Wilson’s method. Statistical analyses were conducted using GraphPad Prism 10.0.

Results

Principle of the CRISPR-Cas14a MTB RIF/INH platform for gene mutation detection

Figure 1 illustrates the principle of the CRISPR-Cas14a MTB RIF/INH platform for identifying drug-resistance mutations. The process commences with multiple PCR amplification of clinical isolates or samples. Subsequently, the resulting PCR product is directly added to individual CRISPR reaction chambers, which are pre-loaded with a mixture of Cas14a, specific gRNAs, T7 exonuclease (T7E), and an FQ reporter. The reaction is then monitored using a fluorescence plate reader. When interpreting drug resistance outcomes via this platform, fluorescence signals serve as crucial indicators. Two distinct gRNA types, gRNA-WT and gRNA-M, are designed to precisely align with wild-type and mutant sites, respectively. A strong fluorescence signal in the gRNA-WT chamber, accompanied by a weak or absent signal in the corresponding gRNA-M chamber, suggests no mutation at the targeted site. Conversely, a weak or absent signal in the gRNA-WT chamber and a strong signal in the corresponding gRNA-M chamber indicate a mutation at the detected site. Although highly unlikely, two alternative scenarios exist. Strong fluorescence signals in both the gRNA-WT and the corresponding gRNA-M chambers imply the presence of both wild-type and mutant genotypes in the detected sample. In contrast, absent fluorescence signals in both chambers suggest that a mutation may have occurred at unexpected sites.

Fig. 1.

Schematic illustration of the CRISPR-Cas14a MTB RIF/INH platform for gene mutations detection. This figure provides a visual representation of the mechanism by which the CRISPR - Cas14a MTB RIF/INH platform identifies gene mutations related to drug resistance

Multiplex PCR amplification of target genes

To streamline the PCR amplification process and reduce the number of required reactions, a multiplex PCR strategy was employed. This approach enabled simultaneous amplification of the rpoB, katG, and inhA genes within a single PCR reaction. The expected sizes of the PCR products for the rpoB, katG, and inhA genes were 326 bp, 150 bp, and 268 bp, respectively. Figure 2 demonstrates the successful amplification of the three target genes from nine Mtb isolates using multiplex PCR. The resulting bands are distinct, well-defined, and consistent with the expected fragment sizes. To further validate the amplification, Sanger sequencing was performed on these bands. Notably, no non-specific amplification bands were observed in any of the multiplex PCR products, indicating that the primer mixture is suitable for use in subsequent experiments.

Fig. 2.

Multiplex PCR amplification of target genes. Multiplex PCR was employed to amplify the target genes. The PCR products were visualized using 1% agarose gel electrophoresis. Lanes 1–9 represent different parallel Mtb isolates and lane 10 served as the negative control

Designing gRNAs for CRISPR-Cas14a to detect RIF- and INH-resistant mutations

Most strains of M. tuberculosis that are resistant to RIF harbor mutations in the RRDR of the rpoB gene. Specifically, mutations at codons 516, 526, and 531 are frequently associated with RIF resistance [17, 20, 21]. Similarly, the majority of INH-resistant strains have mutations either at codon 315 of the katG gene or at the inhA promoter (-15 C/T) [17, 20, 21]. Consequently, gRNAs were specifically designed to detect these known hotspot sites. All the mutations targeted in this study are summarized in Table 1. To assess the efficacy of these gRNAs in differentiating between wild-type and mutant sites, we conducted CRISPR cleavage assays were conducted using synthesized ssDNA substrates. As depicted in Fig. 3A and B, when the DNA substrates were WT ssDNA, the gRNAs that fully matched WT ssDNA substrates induced significantly stronger fluorescence signals compared to their counterparts that matched the mutated ssDNA substrates. Conversely, when the DNA substrates were mutated, the gRNAs that fully matched the mutated ssDNA substrates exhibited much stronger fluorescence signals than those that matched the WT ssDNA substrates. These findings indicate that the designed gRNA-WT and gRNA-M pairs can accurately distinguish between wild-type and mutant sites.

Table 1.

Summarizing all the mutations that need to be detected in this study

E. Coli codon	M. tuberculosis codon	M. tuberculosis mutation	Amino acid change	Perfectly matched gRNA
rpoB 531	rpoB 450	C1349	-	gRNA12 (WT)
rpoB 531	rpoB 450	C1349T	S450L	gRNA13 (M)
rpoB 531	rpoB 450	C1349G	S450W	gRNA15 (M)
rpoB 526	rpoB 445	C1333	-	gRNA5 (WT)
rpoB 526	rpoB 445	C1333T	H445Y	gRNA6 (M)
rpoB 526	rpoB 445	C1333G	H445D	gRNA7 (M)
rpoB 526	rpoB 445	A1334	-	gRNA9 (WT)
rpoB 526	rpoB 445	A1334T	H445L	gRNA10 (M)
rpoB 526	rpoB 445	A1334G	H445R	gRNA11 (M)
rpoB 516	rpoB 435	G1303	-	gRNA1 (WT)
rpoB 516	rpoB 435	G1303T	D435Y	gRNA2 (M)
rpoB 516	rpoB 435	A1304T	D435V	gRNA4 (M)
-	katG 315	G944	-	gRNA39 (WT)
-	katG 315	G944A	S315T	gRNA40 (M)
-	katG 315	G944C	S315T	gRNA41 (M)
-	inhA_c-777t	C-777	-	gRNA37 (WT)
-	inhA_c-777t	C-777T	-	gRNA38 (M)

Fig. 3.

Design and evaluation of gRNAs for the CRISPR-Cas14a MTB RIF/INH platform to detect RIF- and INH-resistant mutations. (A) ssDNA substrates including wild-type (WT) and mutated versions were used. Fluorescence intensity was measured at 2-min intervals over a period of one hour. (B) After 60 min, the fluorescence intensities among different gRNAs were compared. Statistical significance of the differences was evaluated using the unpaired t-test (n = 3 technical replicates; bars represent mean ± S.E.M, **p < 0.01, ***p < 0.001, ****p < 0.0001)

Evaluation of the sensitivity and specificity of the CRISPR-Cas14a MTB RIF/INH plaform for detecting MTBC

The CRISPR-Cas14a MTB RIF/INH platform is capable of simultaneously detecting rpoB, inhA promoter, and katG fragments, making it suitable for direct diagnosis of TB. To assess the sensitivity of this platform method in detecting MTBC, a serial dilution of DNA from the H37Rv strain was performed to determine the LOD. A negative control (NC) using an equal volume of TE buffer was included. Only the gRNA-WT were selected for LOD determination. The results showed that gRNA1, gRNA12, and gRNA39, targeting rpoB 1303G, rpoB 1349 C and katG 944G, respectively, had a LOD of 5 copies/reaction. Meanwhile, gRNA9 and gRNA37, targeting rpoB 1334 A and WT inhA − 15 C, had a LOD of 10 copies/reaction (Fig. 4A and B). Additionally, gRNA5, targeting rpoB 1333 C, had a LOD of only 20 copies/reaction.

Fig. 4.

Evaluation of the LOD and specificity of the CRISPR-Cas14aMTB RIF/INH platform for detecting MTBC. (A) The fluorescence levels of the CRISPR-Cas14aMTB RIF/INH platform were measured across a dilution series of Mtb H37Rv genomic DNA. A heatmap was generated to display the signals of the platform for each individual gRNA. (B) Fluorescent signals were transformed into to a binary test result using a cut-off signal that was more than 6 standard deviations above the no template control. The LOD was determined by performing six replicate reactions at each dilution. (C) The specificity of each individual gRNA was determined using DNA from six common NTM strains, H37Rv strain DNA, and E. coli DH5α. Negative controls, where reactions were performed without DNA. For LOD determination, six technical replicates were executed, and for the specificity test, three technical replicates were performed. The bars in the graphs represent the mean values ± standard error of the mean (S.E.M), with statistical significance indicated as follows: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

To evaluate the specificity of the gRNA-WT, genomic DNA samples from seven mycobacterial reference strains, including M. tuberculosis H37Rv, M. abscessus (ATCC 19977), M. intracellulare (ATCC 13950), M. avium (ATCC 25291), M. kansasii (ATCC 12478), M. gordonae (ATCC 14470), M. fortuitum (ATCC 6841) and E.coli DH5α were used. It was observed that gRNA1, gRNA5 and gRNA9 induced significantly higher fluorescence in H37Rv compared to the other reference isolates. However, these gRNA-WT also showed relatively high fluorescence in certain NTM isolates compared to the negative. In contrast, gRNA12, gRNA37 and gRNA39 specifically induced strong fluorescence in Mtb H37Rv (Fig. 4C). Overall, these results demonstrate that the CRISPR-Cas14a based platform offers a highly sensitive and specific tool for detecting MTBC, underscoring its potential as a valuable tool in TB diagnosis.

Evaluation of the performance of the CRISPR-Cas14a MTB RIF/INH platform using clinical isolates

To evaluate the performance of the CRISPR-Cas14a MTB RIF/INH platform in detecting mutations conferring resistance to RIF and INH, a study was conducted with a panel of 60 isolates. Based on pDST, among these isolates, 18 were resistant to RIF only, 15 were resistant to INH only, and 24 were resistant to both RIF and INH. Additionally, three isolates were sensitive to all the drugs tested. The genetic mutants of each isolates were characterized via Sanger sequencing. The distribution of mutations in isolates associated with RIF and INH resistance is summarized in Supplementary Table S2. Among the RIF-resistant isolates, the most prevalent mutation detected was rpoB S531L (C1349T), which was identified in 66.7% (28/42) of the isolates. Among the INH-resistant isolates, the katG S315T mutation (G944C) was dominant, found in 92.68% (36/39) of the isolates, while only two isolates (5.13%) harboring the inhA USP − 15 C to T mutation.

Subsequently, the drug-resistant isolates were employed to assess the CRISPR-Cas14a MTB RIF/INH platform. Among the 42 RIF-resistant isolates, 39 were identified as mutant (Fig. 5). Specifically, among these 39 mutant isolates, 28 harbored a mutation at position rpoB C1349T (S531L), 2 had a mutation at position rpoB C1349G (S450W), 3 had a mutation at position rpoB C1333T (H526Y), 3 had a mutation at position rpoB C1333G (H526D), 2 had a mutation at position rpoB codon A1334T (H526L), and 1 had a mutation at position rpoB codon A1334G (H526R). These identification results were consistent with the Sanger sequencing data. In strain 17, weak fluorescence signals were observed for all 6 gRNAs targeting rpoB 1333 and 1334, and in strain 45, weak fluorescent signals were observed for all the 3 gRNAs targeting the rpoB 1349. These results indicated the presence of mutations in these regions. However, clear mutation information for the two isolates could not be determined. Further sequencing analysis revealed that strain 17 had a mutation at position rpoB C1333A (H5266N), and strain 45 had a mutation at position rpoB CG1349TC (S531F). As no specific gRNAs were designed to match these mutations accurately, misdiagnosis occurred. Moreover, isolates 18 was misdiagnosed as sensitive to RIF. In fact, it had a mutation at the position of rpoB T1355G (L452P), which was outside the targeted regions of the gRNA set used in this study. In conclusion, this platform exhibited a sensitivity of 93.3% and a specificity of 100% in detecting RIF-resistance (Table S3).

Fig. 5.

Performance of the CRISPR-Cas14aMTB RIF/INH platform in detecting mutations in clinical isolates. A heatmap is presented to show the fluorescence signals of the platform for 60 clinical isolates. Each individual gRNA is listed on the left side of the heatmap, and their corresponding target loci are listed on the right. The genotype of each isolate was determined based on the gRNA that exhibited the highest fluorescence signal. Three technical replicates were conducted for each isolates. And the fluorescence signals shown in the heatmap are the average values obtained from these three replicates

When it came to detecting INH resistance, the CRISPR-Cas14a MTB RIF/INH platform successfully identified mutations in the katG or inhA gene in 38 out of the 39 INH-resistant isolates. Among these 38 mutant isolates, 36 had a mutation at position katG G944C (S315T), whereas only 2 had a mutation at position inhA USP-15 C > T. These identified mutations were consistent with the results of Sanger sequencing. In strain 45, extremely weak fluorescence signals were observed for all 3 gRNAs targeting the katG 944 site suggesting the presence of mutations in the targeted regions. Nevertheless, we were unable to determine the specific mutation sites within the katG gene for this strain. Sanger sequencing analysis revealed that isolate 45 had a mutation at position katG CG944CA (S315T), which fell outside the scope of the gRNA package used in this study. In summary, this platform demonstrated a sensitivity of 97.5% and a specificity of 100% in detecting INH resistance (Table S3).

Evaluation of the performance of the CRISPR-Cas14a MTB RIF/INH platform using sputum samples

The isolation and culturing of bacteria from clinical samples is time-consuming process. Consequently, our aim was to evaluate the ability of the CRISPR-Cas14a MTB RIF/INH platform to directly detect drug-resistant mutations in clinical samples. A total of 55 sputum samples that tested positive in smear analysis were collected. Susceptibility tests for RIF and INH were performed on culture-positive patient samples. Among the 55 samples, 8 were identified as MDR, 4 showed RIF mono-resistance, 3 showed INH mono-resistance, and 41 were susceptible to both RIF and INH (Table 2).

Table 2.

Performance of the CRISPR-Cas14a RIF/INH platform for the detection of MDR-TB using pDST as a reference (n = 55)

		pDST
		Rifampicin		Isoniazid
		Resistant	Susceptible	Resistant	Susceptible
CRISPR-Cas14a MTB RIF/INH	Resistant	10	0	11	0
	Susceptible	2	43	0	44
		Sensitivity = 85.7% (95%CI 71.2% − 93.9%)		Sensitivity = 100% (95%CI 90.8%-99.8%)
		Specificity = 100% (95%CI 98% − 99.8%)		Specificity = 100% (95%CI 97.9% − 99.7%)

These sputum samples were utilized to assess the CRISPR-Cas14a MTB RIF/INH platform. Among the 12 RIF-resistant samples, the platform accurately detected. However, discrepancies were noted in 2 samples. According to the CRISPR-Cas14a MTB RIF/INH platform, these 2 samples were susceptible, but pDST indicated they were resistant. Sanger sequencing analysis revealed that the misdiagnoses in these two samples were attributed to mutations D435Y (1303 G > T) and L452P (1355 T > C), which fall outside the target region covered by the gRNAs set. As shown in Table 2, the platform did not detect any mutations in any of the RIF-susceptible samples. When compared with pDST, the CRISPR-Cas14a MTB RIF/INH platform exhibited a sensitivity of 85.7% (95% CI 71.2–93.9) and a specificity of 100% (95% CI 98–99.8) in detecting RIF resistance in sputum samples. For INH resistance, no discrepancies were observed between the CRISPR-Cas14a MTB RIF/INH platform and pDST for INH resistance.

Discussion

Globally, the burden of MDR-TB, INH mono-resistant TB and RIF mono-resistant TB remains significant. The global distribution of INH mutations has not been comprehensively investigated. According to the WHO report, approximately 8% of TB patients worldwide have TB that is susceptible to RIF but resistant to INH. In certain regions, the rates of mono-resistance may be even higher, ranging from 10 to 20%. Moreover, individuals with such resistance patterns often experience poorer treatment outcomes when treated with standard first-line regimens [22–24]. Currently, there are limited options in the market for methods capable of simultaneously detecting RIF and INH resistant mutation sites. Hence, the development of tools that can accurately identify resistance to both INH and RIF has become increasingly crucial, as it would enable rapid and personalized therapy. CRISPR-based nucleic acid detection technology, known as the “next generation nucleic acid detection technology”, has been widely used in detecting various nucleic acid targets and even non-nucleic acid targets [4, 25, 26]. However, research regarding its application in detecting drug-resistance mutation is scarce [9, 10]. This is mainly because Cas12a and Cas13b proteins are motif-dependent nucleases. In contrast, Cas14a’s substrate recognition does not depend on PAM motifs and exhibit a lower error tolerance [12]. This suggestes that this Cas effector is suitable for the detection of single-base mutations.

In this study, we developed a CRISPR-Cas14a MTB RIF/INH platform for the rapid detection of mutation conferring resistance to RIF and INH. By utilizing multiplex PCR technology, we can simultaneously amplify three target genes—rpoB, katG, and inhA—within a single reaction tube. This strategic approach significantly reduces the number of required PCR amplifications and accelerates the entire detection process. This platform not only facilitates the simultaneous detection of RIF and INH resistance but also eliminates the need for designing additional gRNA specifically for the Mtb detection. The entire testing procedure, including nucleic acid extraction, can be completed within 2.5 h. In comparison to most molecular WHO-recommended rapid diagnostic tests (mWRD), our method offers a significantly shorter detection time, therby enhancing its potential for rapid and efficient diagnosis in clinical settings (Table S4). A recent study has demonstrated that CRISPR-based detection of multiple targeted loci enhances DNA detection sensitivity compared to single-locus detection [27]. Similarly, our CRISPR-Cas14a MTB RIF/INH platform enables the concurrent detection of multiple regions, potentially further enhancing the sensitivity of Mtb detection. Although this aspect was not fully explored in our current study, our findings suggest promising avenues for future research.

In our investigation involving clinical isolates, we carried out a comparative analysis of the CRISPR-Cas14a MTB RIF/INH platform with Sanger sequencing and pDST. The CRISPR-Cas14a MTB RIF/INH platform exhibited impressive performance. For RIF - resistant mutations, it achieved a detection sensitivity of 93.3% and a specificity of 100%. Regarding INH - resistant mutations, its detection sensitivity exceeded 97%, with a specificity of 100%. Moreover, our results indicated that this platform can directly detect RIF and INH resistance in sputum samples., This significantly shortens the time required for diagnosing drug resistance. Notably, this method does not necessitate costly detection equipment, resulting in a remarkably low detection cost. This feature makes it highly conducive to large - scale promotion and implementation, particularly in impoverished regions (Table S4). However, it should be noted that, currently, our method exhibits a significantly lower level of automation compared to mWRDs for tuberculosis and drug-resistant tuberculosis detection. This is mainly because it operates independently of expensive instrumentation and requires separate nucleic acid extraction and detection procedures (Table S4). Distinct from mWRD tsets, our methodology represents an innovative detection platform. The utilization of Cas14a/gRNA demonstrates superior single-base resolution compared to traditional probe-based methods and nucleic acid hybridization techniques (Table S4). Although the CRISPR-Cas14a MTB RIF/INH platform represents a pioneering progress in rapid TB and drug-resistant TB detection, further optimization and technical enhancements are crucial to meet the operational standards and diagnostic requirements set by mWRDs for clinical TB management.

Despite its promising potential, our platform does have certain limitations. Firstly, gRNA package utilized in our platform is confined to hotspot regions. This implies that it might fail to identify drug-resistance-related mutations occurring beyond these regions, thereby posing a risk of misdiagnosis. Secondly, the current detection throughput of our platform is relatively low. To address this issue, future research will explore the integration of CRISPR technology with microfluidic technology. This integration is expected to enhance detection throughput, reduce detection costs, and optimize overall detection efficiency. Thirdly, the experimental procedures are currently conducted in an open environment, which raises the potential risk of aerosol contamination. In the future, we aim to collaborate with relevant companies to optimize and upgrade the platform into an “all-in-one tuber” or “all-in-one chip” drug resistance detection system. This would help to mitigate the risk of aerosol contamination. Regarding the clinical samples used in this experiment, particularly sputum samples, their number was relatively limited. In our study, we assessed the drug resistance detection capability of our CRISPR platform by comparing it with pDST. All the sputum samples included in the study were bacterial positive in bacterial culture and had undergone subsequent pDST. Due to the challenges associated with collecting this type of sputum sample, it was challenging to obtain a large number of samples within a short period. As a result, the number of sputum samples available for this experiment was limited. However, we are actively engaged in collecting more diverse clinical samples, such as bronchoalveolar lavage fluid and pleural effusion. Once the technology is refined and developed into a user- friendly commercial kit, we plan to conduct multi-center joint testing to further validate its effectiveness.

Conclusion

we have successfully developed a CRISPR-Cas14a platform specifically designed for simultaneous detection of RIF and INH resistance mutations in Mtb. When evaluated using clinical isolates and samples, our platform demonstrated remarkable sensitivity and specificity. This substantial progress in drug-resistant TB detection will notably improve the speed and accuracy of diagnosis. Moreover, our study highlights the promising potential of CRISPR-Cas14a-based nucleic acid methods in identifying mutations associated with other forms of drug resistances.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

CRISPR

Clustered Regularly Interspaced Short Palindromic Repeats

gRNA

Guided RNA

INH

Isoniazid

LOD

Limit of detection

MDR-TB

Multidrug-resistant tuberculosis

MTBC

Mycobacterium tuberculosis Complex

ORF

Open Reading Frame

PAM

Protospacer adjacent motif

pDST

Phenotypic drug susceptibility testing

RIF

Rifampicin

RR-TB

Rifampicin-resistant tuberculosis

RRDR

Rifampin resistance-determining region

SNP

Single-nucleotide polymorphism

ssDNA

Single strand DNA

WHO

World Health Organization

Author contributions

G. X, H. L and H. X initiated and designed the research. G. X and H. X performed the experiments. G. X and G. Z analyzed the data. H. L, H. S, D. L, M. O and P. L prepared materials. G. X wrote the manuscript. G. Z revised the manuscript. All of the authors read and approved the final manuscript.

Funding

This work was supported by the National Natural Science Foundation of China (no. 82300128, 82170009), the National Key Research and Development Plan (No. 2020YFA0907200), the Guangdong Science Fund for Distinguished Young Scholars (No. 0620220214), the Guangdong Scientific and Technological Foundation (No. 2020B1111170014), Sanming Project of Medicine in Shenzhen(No. SZZYSM202311009) and the Shenzhen Scientific and Technological Foundation (No. KCXFZ20211020163545004, RCJC20221008092726022, JCYJ20220530163216036, JCYJ20210324130009024).

Data availability

All data generated or analyzed during this study have been included in this published article and its supplementary information files. If someone wants to request further information or data, please contact Dr. Guoliang Zhang with e-mial: szdsyy@aliyun.com.

Declarations

Ethics approval and consent to participate

This study was conducted in accordance with the ethical standards of the Declaration of Helsinki and approved by the Shenzhen Third People’s Hospital Ethical Committee (Reference No. 2021–016). Written informed consent was obtained from each participant. All the methods used in this study were in compliance with the applicable guidelines and regulations.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.