Rapid Molecular Detection of Multidrug-Resistant Tuberculosis by PCR-Nucleic Acid Lateral Flow Immunoassay

Several existing molecular tests for multidrug-resistant tuberculosis (MDR-TB) are limited by complexity and cost, hindering their widespread application. The objective of this proof of concept study was to develop a simple Nucleic Acid Lateral Flow (NALF) immunoassay as a potential diagnostic alternative, to complement conventional PCR, for the rapid molecular detection of MDR-TB. The NALF device was designed using antibodies for the indirect detection of labeled PCR amplification products. Multiplex PCR was optimized to permit the simultaneous detection of the drug resistant determining mutations in the 81-bp hot spot region of the rpoB gene (rifampicin resistance), while semi-nested PCR was optimized for the S315T mutation detection in the katG gene (isoniazid resistance). The amplification process additionally targeted a conserved region of the genes as Mycobacterium tuberculosis (Mtb) DNA control. The optimized conditions were validated with the H37Rv wild-type (WT) Mtb isolate and Mtb isolates with known mutations (MT) within the rpoB and katG genes. Results indicate the correct identification of WT (drug susceptible) and MT (drug resistant) Mtb isolates, with the least limit of detection (LOD) being 104 genomic copies per PCR reaction. NALF is a simple, rapid and low-cost device suitable for low resource settings where conventional PCR is already employed on a regular basis. Moreover, the use of antibody-based NALF to target primer-labels, without the requirement for DNA hybridization, renders the device generic, which could easily be adapted for the molecular diagnosis of other infectious and non-infectious diseases requiring nucleic acid detection.


Introduction
Multidrug-resistant tuberculosis (MDR-TB) is defined by the resistance of Mycobacterium tuberculosis (Mtb) to at least the two most potent antimicrobials against TB infection, Hospital, Mahidol University, Thailand. Outer primers were designed, using NCBI Primer-BLAST [12], to obtain specific regions of the target genes; RpoB-OF' (5'-CGCTGTTGG AAAACTTGTTC-3'), RpoB-OR' (5'-CTCCAGGAAGGGAATCATCG-3'), and KatG-OF' (5'-GGCGGACCTGATTGTTTTCG-3'), KatG-OR' (5'-GAGACAGTCAATCCCGATGC-3'). The amplified products of the target genes were ligated with pGEM T-Easy Vector (Promega, USA) ( Table 1) and incubated overnight at 4°C, as per the protocol provided by the manufacturer. Competent E. coli DH5-α cells were transformed with the ligated plasmids and then grown on LB agar culture plate containing ampicillin, IPTG, and X-gal at 37°C overnight (16-18 hrs) for blue/white selection. The white colonies were picked and inoculated in LB broth with ampicillin and incubated overnight (12-16 hrs) at 37°C, with vigorous shaking (250 rpm). The overnight grown bacterial culture was then used for extracting the plasmid DNA (pDNA), using a plasmid extraction mini-kit (Favorgen, Taiwan). The presence of inserted genes was confirmed using restriction enzyme EcoR1-HF and by DNA sequencing (First Base, Singapore). Stocks of plasmids in DH5-α E. coli cells were made in 50% glycerol and stored at -80°C. The stored E. coli cells were regrown in LB broth (ampicillin) and pDNA extracted for use (mini-kit, Favorgen). The concentrations of the extracted pDNA were measured using a spectrophotometer (NanoDrop 8000, Thermo Scientific, USA).

PCR Amplification
Amplification of the target genes, rpoB (locus Rv0667) and katG (locus Rv1908c) [13,14], were performed in two separate assays using labeled primers ( Table 2). For the rpoB assay, RpoB-IF' and RpoB-IR' ( Table 2) were designed to amplify a conserved region within the rpoB gene as Mtb DNA control. A set of rpoB mutant primers (forward) ( Table 2) [15], together with Rpo-B-IR', were also designed for specific mutation detection to determine Mtb RIF resistance in the rpoB assay. For the katG assay, KatG-IF' and KatG-IR' ( Table 2) were designed for the amplification of a conserved region within the katG gene as Mtb DNA control, together with 315T-F' (katG mutant forward) and KatG-IR' ( Table 2), designed for the S315T mutation detection to determine Mtb INH resistance in the katG assay (Table 2). RpoB-IF' and Rpo-B-IR', as well as KatG-IF' and KatG-IR' have also been designed to recognize a conserved region within the rpoB and katG genes, within the Mycobacterium tuberculosis complex (MTBC). All primers were designed using NCBI Primer-BLAST [12]. For the rpoB assay, individual forward primers targeting rpoB codons 531, 526 and 516 were first experimented separately to evaluate their efficacy and stringency. The individual rpoB primers were then combined into one single assay for multiplex PCR, for the simultaneous detection of all three codons. For the katG assay, a single forward primer (315T-F') targeting the S315T mutation was combined with KatG-IF' and KatG-IR' for semi-nested PCR. All tests were initially performed using pDNA, with a concentration of 10 8 copies (1 ng) added per PCR reaction. The experimental conditions were optimized for primer concentrations and amplification parameters. The optimized conditions were validated with Mtb isolates with known mutations; for the rpoB gene, 4 isolates for codon 516, 6 for codon 526, and 4 for codon 531 were used for the validation of the rpoB assay, and for the katG gene, 4 isolates with S315T mutation were used for the validation of the katG assay (   control (-C), which consisted of PCR reagents without the addition of templates (NO TEM-PLATE), were included to test for possible background appearance or false positives. For the rpoB assay, the FITC-labeled rpoB forward primers, together with the biotin-labeled RpoB-IR' ( Table 2), generate PCR products of 189 bp (codon 531), 203 bp (codon 526) and 235 bp (codon 516) in size (Fig 1). The DIG-labeled RpoB-IF' ( Table 2) and the biotin-labeled RpoB-IR', generate PCR product of 314 bp in size (Fig 1). For the katG assay, the FITC-labeled 315T-F' and the biotin-labeled KatG-IR' ( Table 2), generate PCR product of 335 bp (codon 315) in size (Fig 1). The DIG-labeled KatG-IF' ( Table 2), together with the biotin-labeled Kat-G-IR', generate PCR product of 630 bp in size (Fig 1). With the use of the labeled primers, the PCR amplification products are rendered as dual labeled amplicons (either biotin-FITC, or biotin-DIG, on either end of the amplicon). After PCR amplification, 10 μl of PCR product were used for NALF detection. For analytical comparison of the results, 5 μl of PCR product were also used to run gel electrophoresis, and were visualized using 3% agarose gel (SeaKem, Lonza Inc., USA) infused with Gel-Red dye (Biotium, USA).

Nucleic Acid Lateral Flow (NALF) Development
The NALF device has been designed for the indirect detection of amplified PCR products through the use of antibodies against primer-tags; FITC, DIG, and biotin (Fig 2). Monoclonal antibodies (mAbs) against each tag molecule were produced in-house using the mouse hybridoma technology, with specific recognition to FITC, DIG, and biotin (data not shown).
The NALF device consists of sample application pad, conjugate pad, nitrocellulose membrane and adsorption pad, which are assembled together in a plastic housing. NALF detection was performed individually for the rpoB and katG assays, using separate NALF devices for each assay. Following PCR, 10 μl of PCR product, either from the rpoB or the katG amplification assay, were mixed with 90 μl of NALF buffer solution and added to the NALF device at the sample pad. The mixture was then dispensed onto the conjugate pad, which hosts gold-nanoparticles (AuNPs) conjugated with anti-biotin mAbs (AuNP-anti-biotin) to detect dual labeled amplicons by binding with biotin (Fig 2). The complex flows along the nitrocellulose membrane towards the test-lines, T1 and T2. Detection of MDR-TB by PCR-NALF The first test-line (T1) is composed of anti-FITC mAb for capturing biotin-FITC-labeled amplicons to indicate RIF resistance in the rpoB assay or INH resistance in the katG assay ( Fig  3); for rpoB, the appearance of T1 corresponds to PCR product sizes 189 bp, 203 bp or 235 bp on agarose gel, and for katG, to 335 bp on agarose gel (Fig 1). The second test-line (T2) hosts anti-DIG mAb to capture biotin-DIG-labeled amplicons for Mtb DNA control, which must appear in all cases for the results to be valid (Fig 3); for rpoB, the appearance of T2 correlates with 314 bp on agarose gel, and for katG, to 630 bp on agarose gel (Fig 1). The excess AuNPsanti-biotin, unbound by amplicons, flow pass the two test-lines and are captured at the control line (C) by anti-mouse IgG to ensure the correct operation of the device. The absorbent pad functions as a wick to maintain the flow rate and direction, while preventing any back flow of fluid [16]. NALF results were read after 10 minutes of PCR product addition.

Least Limit of Detection (LOD) Determination
The least limit of detection (LOD) is defined by the minimum number of copies of target genes required in each PCR reaction for a successful detection of the PCR products. The least limit of detection was determined using purified genomic DNA extracts from the H37Rv Mtb isolate, as well as Mtb isolates with known mutations (Table 1) as PCR amplification templates. To determine the LOD of the rpoB and katG assays, serial dilution of the templates was performed as follows; 10 ng (10 6 copies of DNA), 1 ng (10 5 copies of DNA), 0.1 ng (10 4 copies of DNA), 0.01 ng (10 3 copies of DNA), 0.001 ng (10 2 copies of DNA). All NALF results were compared with the results of agarose gel electrophoresis for confirmation.

KatG PCR Amplification and Detection by NALF
The determination of INH resistance in this study targets a single mutation (S315T) in the katG gene. Because the target mutation is singular, the use of just one mutation determining primer, 315T-F' (Table 2), in the katG assay should be adequate. The functionality of katG primers and the ability to identify WT and MT templates were evaluated with the H37Rv Mtb isolate, along with four Mtb isolates containing the S315T mutation, via semi-nested PCR. The amplicons were detected using the NALF device and confirmed with agarose gel electrophoresis. The NALF results and the corresponding agarose gel electrophoresis results for all four Mtb isolates indicate the correct identification of MT and WT templates. Primer Design for the rpoB Assay RpoB, unlike katG, carries multiple mutation sites. To simultaneously detect all the target sites in one PCR reaction, the incorporation of several site-and mutation-specific primers into a single assay is required. The standard primer design of a single base mismatch between primer and non-target template, especially for codon 526 with four different target mutation types (Table 1), may not be enough to ensure correct template distinction. Therefore, the Yaku-Bonczyk primer design method, entailing an intentionally incorporated additional mismatch, has been explored in the study [15,53]. The added mismatch is expected to enhance specificity, leading to a better discrimination against non-complementary DNA. For the eventual selection of rpoB primers, for mutation detection within the rpoB assay, two primer sets were designed and assessed against each other; SM (single mutation) primers with the standard design of a single base mismatch at the 3'-terminus to complement MT templates, and DM (double mutation) primers with an additional strategically located mismatch at the third position from the 3'-terminus in accordance with the Yaku-Bonczyk method (Fig 6, Table 2).

Mutant Primer Selection for the rpoB Assay
The SM and DM primers were tested with MT and WT templates in semi-nested PCR. Each primer was tested in a separate assay composed of RpoB-IF', RpoB-IR', and the target primer of interest for mutation detection ( Table 2). The test results (Fig 7) show that SM primers were able to anneal to their specific targets, and were able to correctly identify the MT and WT templates with high efficacy. The appearance of an unaccountable non-specific band of approximately 270 bp on agarose gel in Fig 7 did not lead to a false positive/negative or background on NALF, and therefore, did not interfere with the test. DM primers, on the other hand, did not perform as expected. They failed to anneal to their targets, providing negative results (data not shown). DM primers were, therefore, exempted from further experimentation.

Multiplex PCR for the Simultaneous Detection of RIF Resistance
Determining Codons in the rpoB Gene SM primers, which have been selected for the rpoB assay, were combined into a single assay for multiplex PCR, which would allow all RIF resistance determining mutations at target codons 531, 526 and 516 to be detected simultaneously. Initially, the combined assay was tested using pDNA for condition optimization (data not shown). The optimized conditions were then validated and tested for reproducibility using the H37Rv (WT) Mtb isolate and Mtb isolates with known mutations (4 isolates for codon 516, 6 for codon 526, and 4 for codon 531) ( Table 1). Fig 8 is a representation of the results, which reaffirms the efficacy, stringency and the specificity of the SM primers in the rpoB assay. Least Limit of Detection (LOD) Determination The least limit of detection (LOD), which has been described in the study as the minimum number of copies of target genes required in each PCR reaction for a successful detection of the PCR products, was determined for both the rpoB (using SM primers) and the katG assays. The results indicate the LOD to be 10 4 copies of DNA (0.1 ng) per PCR reaction via NALF detection, and 10 5 copies of DNA (1 ng) per PCR reaction via detection by gel electrophoresis (Fig  9). The NALF test-lines illustrate a clear loss of intensity as the concentration of template DNA in the reactions diminished. Below the concentration of 10 4 copies of DNA, the NALF testlines were no longer visible. A lower LOD determined for NALF, compared to agarose gel electrophoresis, suggests a higher sensitivity of the device.

Discussion
NALF technology is an interesting tool for enabling simplification of molecular diagnosis. Recent studies have successfully field-tested NALF for point-of-care molecular detection of malaria with promising results [54][55][56]. Preliminary tests have also been conducted for a lateral The use of SM primers demonstrate an effective template distinction with the appearance of NALF T1 and T2 for all MT templates to indicate RIF resistant Mtb isolates, which relates to the inner and outer DNA bands on agarose gel for the MT templates. As for WT templates, the assays correctly resulted in the appearance of NALF T2 only, to indicate RIF susceptible Mtb isolates, which corresponds to PCR product size 314 bp on agarose gel. flow based quantitative detection of amplified HIV-RNA with clinical relevancy [57]. This proof of concept study, using the above principle, attempted to develop a molecular diagnostic alternative for MDR-TB to complement the use of conventional PCR, targeting low-resource and peripheral healthcare settings that already routinely employ PCR machines. The development of NALF, to meet the objective, was fixated upon simplicity and rapidity. These qualities were achieved through the strategic design to minimize operator involvement, and to present the results in the most non-complex manner. The approach uses gold nanoparticle (AuNP) conjugated with anti-biotin antibodies to detect dual labeled amplicons, and anti-FITC/DIG antibodies to capture and immobilize the amplicons on the test-lines for visual indication (Fig  2). This technique eliminates the requirement for operator intervention. Moreover, the simplicity is also projected through the concise display of NALF results, using only two test lines; T1 for RIF resistance indication in the rpoB assay or INH resistance indication in the katG assay, and T2 for Mtb DNA control (Fig 2).
This NALF design contrasts other commercially existing DNA strip tests for MDR-TB such as GenotypeMTBDRplus which has 27 reaction zones on a single test strip, with 21 DNA probes for mutation detection and 6 probes as control [8]. All 27 lines representing rpoB, katG and inhA are displayed on the same test strip and have to be individually interpreted by the operator. The detailed and complex nature of the results renders it more suitable for epidemiological rather than diagnostic work. In comparison, the NALF device has no technical specification or interference requirements, making it more appropriate for diagnostics.
Another important advantage of using anti-tag antibodies on NALF rather than DNA probes is that it permits the indirect capture of target amplicons without necessitating DNA hybridization. This advantage makes the device generic, allowing broader application. The device can, thus, be adapted for the diagnosis of any disease requiring nucleic acid detection, based on the strict use of primers conjugated with specific tags; biotin for detection and FITC/ DIG for capture (Fig 2). This generic nature allows the device to be mass produced at low cost, serving the objective of targeting low-resource settings. This design differs from the lateral flow designs adopted in previous studies in the determination of HIV and malaria infection where sequence specific oligonucleotide probes were used to capture HIV-RNA and P. falciparum-DNA [54][55][56][57].  Detection of MDR-TB by PCR-NALF As for primer design, we initially reasoned that the standard design of incorporating a single 3'-terminal mismatch to avoid primer hybridization with non-complementary DNA may be inadequate for the determination of single-nucleotide polymorphisms (SNPs). For this study, incorrect hybridization could lead to the misidentification of Mtb drug resistance, and therefore, MDR-TB. To minimize the possibility, the study explored the Yaku-Bonczyk primer design strategy entailing an additional intended mismatch to enhance specificity [15,53]. The study compared the functionality of the SM (single mutation) and DM (double mutation) primers ( Table 2, Fig 7) to evaluate their performance in the selection for the rpoB assay. The criterion was based on specificity, efficacy, stringency and also their technical influence on the design of the NALF device. DM primers, with an additional strategic mismatch (Fig 6), were predicted to perform with a higher discrimination effect based on their success in other studies [15,53,58]. The results, however, defeated the expectations wherein the DM primers failed to anneal to their targets, providing false negative results. In this study, the DM primers were 15-17 bases in length. Comparatively, the primers employed in a previous study, used as guidance in the design of DM primers, is about 38 bases in length [15]. This distinction could have been the root cause of the second mismatch considerably weakening the annealing bond between Detection of MDR-TB by PCR-NALF the DM primers and their complementary MT template, resulting in complete non-binding [59]. SM primers, on the other hand, with the standard design of a 3'-terminal base complementing the MT templates (Fig 6), selectively annealed to their targets, resulting in a successful identification of the drug resistant (MT) and the drug susceptible (WT) Mtb isolates (Fig 7). The rpoB SM primers were, therefore, selected for use in multiplex PCR.
From the broad compilation of the global MDR-TB epidemiological statistics (Fig 4), 85-95% of RIF resistance was found to be accompanied by INH resistance; therefore, RIF resistance is popularly used by several existing commercial tests, such as XpertMTB/RIF and INNO-LiPA, as the surrogate marker for MDR-TB [43,46,60]. However, many studies point towards the benefit of testing for both RIF and INH resistance to improve the comprehensiveness of the results [38,[61][62][63]. Retrospective studies have claimed that RIF monoresistance is on the rise in South-Africa, and testing for RIF resistance alone has led to the misidentification of MDR-TB [61,64]. Additionally, the incorporation of an INH assay would also support the determination of INH monoresistance. A recent study in China showed that early diagnosis of INH monoresistance allowed for the tailoring and implementation of specific therapies to prevent the development of MDR-TB [52]. The results of the study showed positive treatment outcomes for both patients diagnosed early with INH monoresistance as well as the drug-susceptible TB patients. A similar study conducted in the United States in 2009 also presented with comparable results [49]. INH resistance, however, is characterized by mutations in several genes (Fig 4) and aiming to detect all these genes would undesirably increase the complexity of the PCR-NALF test, based on the current design. This study, therefore, selected to focus only on the S315T mutation detection of the katG gene, an INH resistance conferring mutation with the highest global prevalence [29][30][31][32][33][34].
The current PCR-NALF test also presents an additional benefit of choice by dividing rpoB and katG into two separate assays. The option offers convenience. This design contrasts the test feature of GenotypeMTBDRplus where multiple probes for rpoB, katG and inhA are all pre-lined on one single test strip [8]. RpoB and katG assays in this study also share the same PCR thermocyling conditions, permitting the PCR amplification of both assays at the same time, if the user chooses to. The same test design could also be adapted for the determination of resistance to second-line drugs for the indication of extensively drug-resistant tuberculosis (XDR-TB), such as for the detection of fluoroquinolone(s) resistance, which is caused by single base mutations in the gyrA and gyrB genes [65,66]. The addition would add to the choice, as well as the coverage of the test.
Another important factor defining the validity of a molecular test is the least limit of detection (LOD), which signifies the sensitivity of the test. The LOD of the current test, for both the rpoB and katG assays, was found to be 10 4 genomic copies per PCR reaction (0.1 ng of DNA) (Fig 9) using NALF. Even though the least LOD is relatively high, suggesting the need for further optimization, the current result of 10 4 genomic copies could potentially be considered within the clinically relevant range. Sputum samples from TB patients with positive AFB staining of 1+ and 2+, based on the standard manual for laboratory technicians, is estimated to contain approximately 1-3x10 4 and 3-5x10 4 Mtb bacilli per ml of sputum, respectively [67]. Moreover, further studies using sputum samples are underway to help define the sensitivity and specificity of the test.
In spite of the promising results of the study, several aspects could still be improved. The LOD, which reflects the analytical sensitivity of the PCR-NALF test, could further be reduced through additional optimization, to enable a feasible detection of lower DNA concentrations. The extent of rpoB mutation coverage in the study could also be widened by incorporating the detection of more codons such as codons 533, 522, 513 and 511 within the rifampicin-resistance determining region [27,28,44,47]. In addition, further incorporation of inhA and ahpC genes, which account for 20-40% of INH resistance (Fig 4), could also improve the overall comprehensiveness of the test results.
In conclusion, this proof of concept study demonstrates the potential use of PCR-NALF as a molecular diagnostic alternative for the detection of MDR-TB. The simplicity, rapidity and ease-of-use could prove beneficial for low-resource settings that already employ PCR machines. Furthermore, the generic nature of the NALF device provides great diagnostic potential in adapting the test for the detection of other infectious and non-infectious diseases that require nucleic acid identification.