Single Isothermal Assay for Multi-Site Mutation Detection of Rifampicin Resistance in Mycobacterium tuberculosis

Abstract

Antimicrobial drug resistance is an escalating global health burden, often driven by multiple genetic changes within key resistance-associated genes. Achieving multiplex capability of mutation detection while maintaining simplicity and affordability is critical, particularly in point-of-care and resource-limited settings. Here, we introduce a strategy for multi-site mutation detection using single isothermal amplification of a nucleic acid fragment spanning multiple mutations in the rifampicin resistance-determining region (RRDR) of the rpoB gene, encompassing codons 516 and 526 in Mycobacterium tuberculosis. This unified design eliminates competition among targets amplified by multiple primer sets. Site-specific hybridization probes enable accurate discrimination between wild-type and mutant sequences, while an integrated self-calibration probe provides normalization to mitigate variability from sample concentration and sample matrix interference. To validate this approach, we applied it to detect rifampicin (RIF) resistance mutations at codons 516 and 526 of the rpoB gene in Mycobacterium tuberculosis, which are two key targets for molecular diagnostics and surveillance. Using 42 artificial DNA fragments, which included both wild-types and mutants with single- or two-site mutations, the assay achieved 100% accuracy in discriminating between wild-type and mutant sequences for codon 516. On the other hand, sequences harboring mutations at codon 526 were identified with 100% accuracy, compared to 94% accuracy for wild-type sequences. Overall, the system achieved a 100% positive percent agreement (PPA) for drug-resistance sequences and 97% negative percent agreement (NPA) for drug-sensitive sequences based on these 42 samples. These findings suggest that this method has the potential to provide a reliable framework for multi-site mutation detection.

1. Introduction

Antimicrobial resistance (AMR) has emerged as one of the most pressing global health challenges, threatening the effectiveness of current treatments and increasing morbidity, mortality, and healthcare costs worldwide. The World Health Organization has identified AMR as a critical priority, with drug-resistant infections projected to cause millions of deaths annually if left unchecked [1]. Rapid and accurate detection of resistance-conferring mutations is essential for guiding effective therapy in many cases and curbing the spread of resistant pathogens. However, many resistance mechanisms involve multiple genetic loci, making multiplex detection capability a key requirement for diagnostic platforms. While molecular methods involving variations in PCR have advanced beyond the traditional culture-based Drug Susceptibility test (DST), their reliance on thermal cycling equipment and sophisticated laboratory infrastructure limits their impact in resource-limited settings [2]. High costs, lengthy protocols, and the need for specialized training create a diagnostic gap in resource-constrained regions where drug resistance is most prevalent [3,4].

Research in isothermal nucleic acid amplification technologies (INAATs) has been gaining momentum as an easy-to use alternative to traditional technologies like PCR. Operating at constant temperature, these technologies eliminate the need for thermocyclers and enable the use of simpler and portable diagnostic platforms. Techniques such as loop-mediated isothermal amplification (LAMP), recombinase polymerase amplification (RPA), and nucleic acid sequence-based amplification (NASBA) have demonstrated robust performance in pathogen detection and have been successfully deployed in field settings [3,4,5,6]. However, INAATs present an inherent challenge for mutation detection. Unlike, for instance, allele-specific PCR, which achieves high-stringency primer discrimination and polymerase primer extension through thermal cycling and denaturation-renaturation steps, INAATs operate at constant temperature. This isothermal condition eliminates the stringency conferred by thermal cycling, allowing amplification to proceed even when mismatches occur in mutation-specific or allele-specific primer binding regions, thereby compromising accurate mutation discrimination [7,8]. Overcoming this limitation requires strategies involving the use of additional specialized enzymes [9,10], such as exonuclease or endonuclease, for mutation-specific cleavage or complex control reactions [11,12]. Such requirements substantially increase assay complexity and costs. Furthermore, in terms of multiplex capability, INAATs face another challenge: INAAT approaches typically require multiple primer sets to target different genetic variants or necessitate separate reactions for each mutation site of interest [13,14], when multiplex testing relies on amplifying multiple targets by different primer sets, competition for reaction components among these targets will reduce individual target sensitivity, will have high risk of cross-reactivity among lots of primers, and ultimately compromise overall assay performance [13,14,15].

To overcome the challenge of mutation detection using isothermal amplification, we have developed a dual-probe strategy via molecular beacon hybridization, which provided a straightforward solution for distinguishing mutations without prior knowledge of their specific mutation direction [15]. However, our previous approach still required two separate assays to interrogate two mutations distributed across spatially distant sites, not addressing the challenge of multiplex capability of INAATs in mutation detection. In this work, we present a new method for detecting multi-site mutations in one assay. Briefly, instead of generating separate amplicons for each mutation site, our new method amplifies a single nucleic acid fragment that spans multiple mutation loci of interest (Figure 1). This feature tackles the challenge in isothermal amplification reactions targeting multiple sequences simultaneously by eliminating competing intra-target amplification. It also simplifies primer design and reduces the risk of primer cross-reactivity. Within one amplicon, we deploy site-specific hybridization probes (i.e., a molecular beacon probe) designed to discriminate between wild-type and mutations based on differential binding kinetics and subsequent signal generation. Signal generation is triggered upon successful probe-target binding, allowing for real-time or end-point detection of multiple mutations in a single reaction. By focusing the amplification on a single fragment while deploying multiple detection probes, the method maintains high analytical specificity while significantly simplifying the reaction setup. To enhance the reliability and robustness of the assay, we have incorporated a self-calibration probe based on our previously developed normalization methodology (Figure 1). This feature compensates for variations in signal intensity that may arise from differences in sample concentration, matrix interference effects, reagent lot-to-lot variability, or instrument-specific biases.

We demonstrate the applicability of this new multi-site mutation detection platform by targeting the rifampicin (RIF) resistance-determining region (RRDR) of the rpoB gene in Mycobacterium tuberculosis, a key marker for multi-drug resistant tuberculosis (MDR-TB) [16]. Codons 516 (Asp516) and 526 (His526) in the rpoB gene, based on the E. coli numbering system, are critical determinants of rifampicin resistance in Mycobacterium tuberculosis. Mutations at codon 526 typically confer high-level resistance by causing significant structural changes that block rifampicin binding, while alterations at codon 516 often lead to low-to-moderate resistance through subtle modifications of the drug-binding pocket. These sites are among the most frequently observed in resistant clinical isolates and are key targets for molecular diagnostics and surveillance [17,18,19,20]. Our approach enables a simultaneous detection of mutations at codons 516 and 526 for RIF-drug resistance. The principles applied in this work are broadly applicable and can be easily applied to the detection of other pathogens and drug-resistance targets.

2. Materials and Methods

2.1. Plasmids and Oligos

DNA fragments containing wild-type sequences and mutations at desired locations in the rpoB gene were synthesized and cloned into the pUC19 cloning vector by GenScript (Piscataway, NJ, USA) (Table S1). The plasmids were quantified using the Bio-Rad ddPCR platform (Bio-Rad, Hercules, CA, USA) as per Bio-Rad ddPCR user instructions. All primer candidates were screened using the NEB^® LAMP Primer Design Tool (v1.3.0, New England Biolabs, Ipswich, MA, USA) to minimize self-dimer, cross-dimer, and hairpin formation. Molecular beacons were evaluated using IDT OligoAnalyzer™ to assess beacon–primer and beacon–beacon interactions, as previously validated [15]. HPLC-purified unmodified primers and fluorescent-quencher modified probes used in the study were purchased from IDT (Coralville, IA, USA) or Sigma Aldrich (Woodlands, TX, USA) and resuspended using TE buffer (Sigma Aldrich, Milwaukee, WI cat# 93283, USA). All primers and probe sequences targeting the MDRR of rpoB gene in Mycobacterium tuberculosis used are listed in Table S2.

2.2. Isothermal Amplification Assay

Lyo-ready LAMP mix from Meridian BioSciences (Luckenwalde, BB, Germany cat#MDX097) was used for all studies here. Final LAMP reaction contains 1X-Lyo-ready LAMP mix, and 1.6 μM each of FIP and BIP primers, 0.2 μM each of F3 and B3 primers, 0.8 μM each of LF and LB primers, 0.4 μM each of Probe C, Probe I₁ and Probe I₂ (as per the test conditions), 8.0 mM MgSO₄ and synthetic plasmids or ATCC Mycobacterium tuberculosis (MTB) genomic DNA (ATCC, Vanassas, VA, USA cat# ATCC25177D). Nuclease-free water (Sigma Aldrich, Milwaukee, WI, USA cat# W4502-500 mL) was added if necessary to adjust the volume of reaction to 15 µL. Real-time fluorescence measurements for HEX (Probe C), Cy5 (Probe I₁) and FAM (Probe I₂) signals, were collected at 66 °C over 60 min with a 1 min interval using Bio-Rad’s CFX96 Touch Real-Time PCR Detection System (Bio-Rad, Hercules, CA, USA). RFU (Relative Fluorescence Unit) values collected at 10 min and 40 min with both the baseline and threshold set to auto for all channels were used for Z-score calculations (see Section 2.3).

2.3. Statistical Analysis

Samples were classified as wild-type or mutant based on Z-score calculations described in our previous work [15]. Briefly, the Z-score for each sample is calculated by using the baseline RFU values at 10 min mark and end-point RFU values at the 40 min mark of the calibrator probe and Probe I₁ or Probe I₂ in the following equation:

$$Z=\left|{\displaystyle \frac{X-\mu }{\sigma }}\right|,$$

(1)

where

X: S_sample;

µ: mean of S_ref;

σ: standard deviation of S_ref.

The values of S for the reference and test samples are calculated using the following equation:

S = (f_ien/f_ibn)/(f_ce/f_cb)

(2)

n: indicator probe number; f_cb: baseline signal of “calibrator” probe (here is Probe C), f_ce; end-point signal of “calibrator” probe (here is Probe C); f_ibn: baseline signal of “indicator” probe n (here is Probe I₁ or Probe I₂); f_ien: end-point signal of “indicator” probe n (here is Probe I₁ or Probe I₂).

For sample classification, a threshold Z-score value (Z_T) was determined for each indicator probe based on signals obtained from reference wild-type MTB genomic DNA. Samples yielding Z-scores exceeding this threshold (Z > Z_T) were classified as mutation-positive for that site, while those with Z-scores at or below the threshold (Z ≤ Z_T) were classified as mutation-negative for that site.

3. Results

3.1. A Model of Multi-Site Mutation Detection via Single-Fragment Amplification

In our previous dual-probe method for single-plex mutation detection by isothermal amplification, molecular beacon probes were designed to target single-stranded DNA regions, i.e., the circular regions, also referred to as loop regions in the dumbbell-shaped LAMP amplicon as shown in Figure 1. The “indicator” probe (Probe I) targeted the mutation site of interest in one loop region, while the “calibrator” probe (Probe C) targeted the conserved sequence in the other loop region. The key reason for this is the single stranded nature of the loop regions provides ease of access for probe hybridization-to-detection without requiring strand displacement or denaturation. This approach is justified because hybridization-based detection of double-stranded DNA generally necessitates additional steps, such as thermal denaturation and subsequent renaturation, to enable probe-target binding [21]. These requirements introduce workflow complexity and hinder the development of a true one-pot assay, thereby limiting applicability in field-deployable settings. However, positioning the probe-targeting site in a single-stranded region limits primer design; in our prior work, we had to design two separate primer sets to detect two spatially distant mutation sites, to ensure that each mutation site was positioned within a single-stranded region. Advantageously, LAMP’s optimal temperature range (60–70 °C) is relatively high, which theoretically might destabilize short duplex regions within amplicons, facilitating probe hybridization to internal target sites. Moreover, the strong strand displacement activity of Bst polymerase used in LAMP can further facilitate probe access into dsDNA region during amplification. Together, these features indicate that probes may remain effective even when targeting regions that are not single-stranded in LAMP, while maintaining the simplicity of a one-pot detection workflow. In this study, we aimed to explore the flexibility of probe design—beyond restricting it to the loop region—and to evaluate the feasibility of detecting multi-site mutations by simple amplification using a single primer set. A three-probe system was proposed consisting of two mutation-specific indicator probes and one internal calibrator probe (Probe C). The indicator probes (Probe I₁ and Probe I₂) were designed to detect two mutations at two sites, while Probe C served as an internal signal normalization. Based on our previous validation studies with individual probe components, we established expected fluorescence signal patterns for different mutation scenarios. Figure 1A shows the expected signals from the three probes in the absence of any mutations at either site. Figure 1B shows expected signals when a single mutation is present either at site one or site two. When mutations are present at both the codons, a reduction in signals for both Probe I₁ and Probe I₂ is expected as shown in Figure 1C.

3.2. Expanded Probe Design Space: Functional Detection Beyond Single-Stranded Targets

First, we validated that the probe remained effective when targeting a double-stranded region. Using the same probe (Probe I₁: targeting codon 516 mutations), we designed two primer sets (Table S1) to amplify rifampicin (RIF) resistance-determining region (RRDR) of the rpoB gene in Mycobacterium tuberculosis (MTB). With the primer set 1 (Figure 2A), the probe will bind to the loop region in the amplified product; with the primer set 2 (Figure 2B), the probe will bind to the non-loop region overlapping with the B1 region in the amplified product. As shown in Figure 2, without extra denaturing steps, Probe I₁ can detect amplified products either through binding to an intact single-stranded region (Figure 2A) or invading a double-stranded region (Figure 2B). As expected, when the sample contained a mutation at codon 516, a reduced signal of Probe I₁ was observed (green curves with circle marker: mutant; solid green curves: wild-type). This data confirmed that probes designed to recognize regions outside the loops can detect the amplicon and distinguish between the wild-type and mutant sequences, thereby supporting the feasibility of multi-site mutation detection with a single LAMP amplification reaction.

3.3. Multiplex Probe Integration Does Not Compromise Assay Sensitivity

Next, we examined whether incorporating additional probes for multi-site mutation detection could lead to assay sensitivity loss. In prior work, we have shown that the presence of two probes (Probe C and a single indicator probe) does not affect LAMP assay sensitivity [15]. For this study, three probes were included: Probe C and Probe I₂, both targeting single-stranded region, and Probe I₁, which was designed to bind a dsDNA region (Figure 3A,B). We then compared assay sensitivity in reactions containing Probe C and Probe I₂ only versus reactions containing Probe C, Probe I₁, and Probe I₂. As shown in Figure 3C,D, both configurations (i.e., two probes and three probes) achieved a 3/3 detection rate at 4000 and 2000 copies/reaction, and a 2/3 detection rate at 200 copies/reaction. These findings indicate that adding a third probe for increasing the mutation detection multiplexing capability does not compromise nucleic acid detection sensitivity.

3.4. Performance of Multi-Site Mutation Detection Model for Rifampicin Resistance in Mycobacterium tuberculosis

Finally, we evaluated our multi-site mutation detection model in MTB RIF drug resistance. We first established a reference wild-type training data set (n = 6) to determine Z-score threshold (Z_T) using MTB wild-type genomic DNA (see Section 2.3). For the test cohort sample, due to the unavailability of Biosafety Level -1/2 MTB genomic material carrying mutations at codon 516 and 526, we used seven synthetic nucleic acid fragments representing wild-type and different mutations at codon 516 and 526 of the rpoB gene in Mycobacterium tuberculosis. Among these, one fragment corresponded to the wild-type (drug-sensitive, WT: Asp-516, His-526) sequence, two fragments carried mutations exclusively at codon 516 resulting in distinct amino acid substitutions (Mut 1: Tyr-516, His-526; Mut 2: Val-516, His-526), two fragments carried mutations exclusively at codon 526 with different amino acid substitutions (Mut 3: Asp-516, Tyr-526; Mut 4: Asp-516, Leu-526), and the remaining two fragments carried combined mutations at codons 516 and 526 (Mut 5: Val-516, Leu-526; Try-516, Leu-526). All mutations were selected based on their clinical relevance, representing dominant rifampicin (RIF) resistance cases [19,20], see nucleic acid sequence details in Table S2. To ensure robust statistical analysis, each synthetic fragment was treated as an independent replicate (n = 6). Primer set 2 and the three probes mentioned above were used in this evaluation. Probe C (calibrator probe) hybridizes to a highly conserved sequence, while Probe I₁ (indicator probe 1) targets codon 516, and Probe I₂ (indicator probe 2) is designed for an additional mutation site: codon 526.

As shown in Figure 4, in wild-type DNA samples (Asp-516, His-526) containing no mutations at either target codon, all three probes exhibited robust fluorescence signals throughout the amplification process. Both indicator probes (I₁ and I₂) generated strong signals comparable to the calibrator probe, indicating successful hybridization to their respective wild-type target sequences. The calibrator probe consistently produced stable signals across all tested conditions, confirming its suitability as an internal control. The multiplex probe system demonstrated exceptional specificity in discriminating between different mutation patterns across six mutant plasmid constructs. Fluorescence signals from Probe I₁, designed to detect codon 516 mutations, exhibited wild-type-level fluorescence signals when analyzing Mut 3 (Asp-516, Tyr-526) and Mut 4 (Asp-516, Leu-526) plasmids, confirming the preservation of the wild-type aspartic acid residue at position 526 in these constructs. Conversely, significant signal reduction was observed with Mut 1 (Tyr-516, His-526) and Mut 2 (Val-516, His-526) templates, indicating the detection of the tyrosine and valine substitution at codon 516. The performance of Probe I₂ mirrored this specificity for codon 526 detection. Decreased fluorescence signals were exclusively observed with Mut 3 and Mut 4 plasmids, both harboring mutations at position 526 (tyrosine and leucine substitutions, respectively), while maintaining robust signals with wild-type sequences at this site in Mut 1 and Mut 2 plasmids. The selective response pattern highlights the probe’s target-specific binding capability and its utility for mutation identification. Both probes successfully differentiated between various amino acid substitutions at the target codon, demonstrating their independence from substitution type in mutation detection. The system’s capacity for simultaneous multi-site mutation detection was confirmed through analysis of Mut 5 (Val-516, Leu-526) and Mut 6 (Tyr-516, Leu-526) constructs. Both plasmids exhibited concurrent signal reduction in Probe I₁ and Probe I₂, indicating successful identification of mutations at both target codons within a single reaction. This multi-detection capability represents a significant advancement for comprehensive resistance profiling applications.

Following visual inspection in Figure 4, fluorescence data were further normalized using the calibrator probe, and Z-scores were calculated as described in the Section 2 and our previous work [15]. A Z-score threshold (Z_T) was determined for probes I₁ (Z_T1 = 3) and I₂ (Z_T1 = 3) based on a reference wild-type training set (Table S3). Among the 42 samples in the test cohort, at codon 516, all 18 wild-type sequences were correctly identified as wild-type, and all 24 mutant sequences were correctly identified as mutant, yielding 100% accuracy for both categories. At codon 526, 17 of 18 wild-type sequences were correctly classified (94% accuracy), while all 24 mutant sequences were correctly identified (100% accuracy). Overall, the assay achieved 100% sensitivity for detecting drug-resistant mutations (true positive rate) and 97% specificity for identifying drug-sensitive sequences (true negative rate) (Figure 5 and Table S4).

4. Discussion

A global rise in antimicrobial resistance has increased the need for rapid, accurate and point-of-care diagnostic tools. Traditional diagnostic methods used involve the use of sophisticated instruments which makes them less suited for resource-limited settings with a heavy burden of drug-resistance. Recently, isothermal technologies like LAMP, RPA, etc., have gained traction in pathogen detection. One major factor for this is their cost-effectiveness and demonstrated operational feasibility in peripheral laboratories [2,22,23]. However, multi-site mutation detection by isothermal amplification assays faces challenges from primer competition and preferential amplification of certain targets, while existing strategies rely on primer optimization, balancing primer concentration, and incorporating probe-based detection systems (e.g., DARQ-LAMP or QUASR-LAMP) to ensure target-specific signal generation has been implemented [24,25,26,27]. Our results demonstrate that a single LAMP primer set combined with multi-probe detection can accurately identify rifampicin resistance mutations at multiple codons within the rpoB RRDR. This approach addresses a fundamental challenge in multiplex LAMP technology: the technical difficulties associated with simultaneously amplifying multiple targets. The literature has extensively documented that multiplex assays suffer from primer competition, differential amplification efficiencies, and increased optimization complexity [28,29,30]. Each target in LAMP requires 4–6 primers (two outer primers, two inner primers, and optionally two loop primers), meaning that multiplex detection of even two targets requires 8–12 primers in a single reaction [28]. This primer complexity creates several documented hurdles including primer-primer interactions, competition for polymerase and reaction components, and preferential amplification of one target over another [3,4]. Gadkar et al. (2018) specifically demonstrated that in multiplex LAMP reactions, preferential amplification of one target over another is common, leading to unreliable detection of lower-abundance targets [30]—a critical concern for mutation detection where mutant alleles may be present at lower frequencies than wild-type sequences in heteroresistant populations.

We have demonstrated a fundamentally different strategy for multiple mutation detection on a single amplified fragment to overcome this competition. Our assay introduces several key innovations: (a) simplified reagents: requires only a polymerase, eliminating additional enzymes or nucleases; (b) single primer set: minimizes amplification competition, improving efficiency and reliability. (c) dual-probe ratio metric readout: combines an internal calibrator probe with mutation-specific indicator probes to enhance accuracy and reduce false results by compensating for signal variability.

While these findings highlight the promise of our approach, several important limitations should be acknowledged. First, the assay cannot differentiate resistance-conferring mutations from silent mutations at the same loci. In other words, it can only indicate whether a sequence differs from the wild-type, without specifying the exact nucleotide or amino acid change. For example, as shown in Figure 4, substitutions at codon 526 (His → Leu vs. His → Tyr) or codon 516 (Asp → Tyr vs. Asp → Val) cannot be distinguished. The single false-positive result observed in our study (1/18 wild-type samples at codon 526) occurred at a Z-score near the classification threshold, highlighting the importance of threshold optimization and the potential for borderline classifications. Although clinical sample variability and expanded target diversity across M. tuberculosis lineages (single base substitutions, transversions, deletions, or insertions) may in theory produce varying signal intensities in indicator probes and calibrator probes, comprehensive evaluation with large sample cohorts is required to fully characterize these effects.

Second, the method cannot determine whether a sample represents a heterozygous mixture or a homozygous population. Consequently, drug-resistant clones present at low abundance may escape detection. Finally, this approach may not be suitable when mutation hotspots are widely spaced and cannot be amplified by a single primer set, as isothermal amplification technologies have inherent limitations on amplicon size.

In conclusion, the single fragment amplification spanning multiple mutation loci along with the internal calibrator and mutation indicator probe system is a promising tool for isothermal-based mutation detection technology. By enabling the identification of drug-resistance in resource-limiting settings, this method has the potential to contribute to global efforts to tackle rising drug resistance. Further refinement and validation of the assay across a broader spectrum of clinically relevant mutation variants, as well as the design of additional targets to encompass other rifampicin resistance-associated mutation sites, will be required to fully realize the potential of this method for integration into existing MTB rifampicin resistance diagnostic and surveillance frameworks.

5. Patents

The work described in this manuscript is the subject of a Revvity, Inc., pending patent application.