OUP user menu

Robustness of a loop-mediated isothermal amplification reaction for diagnostic applications

Patrice Francois, Manuela Tangomo, Jonathan Hibbs, Eve-Julie Bonetti, Catharina C. Boehme, Tsugunori Notomi, Mark D. Perkins, Jacques Schrenzel
DOI: http://dx.doi.org/10.1111/j.1574-695X.2011.00785.x 41-48 First published online: 1 June 2011

Abstract

We evaluated the robustness of loop-mediated isothermal amplification (LAMP) of DNA for bacterial diagnostic applications. Salmonella enterica serovar Typhi was used as the target organism and compared with a real-time quantitative PCR (qPCR) for testing assay performance and reproducibly, as well as the impact of pH and temperature stability. This isothermal amplification method appeared to be particularly robust across 2 pH units (7.3–9.3) and temperature values (57–67 °C). The detection limit was comparable to that observed using optimized home-brew qPCR assays. The specificity of the amplification reaction remained high even at temperatures markedly different from the optimal one. Exposing reagents to the ambient temperature during the preparation of the reaction mixture as well as prolonging times for preparing the amplification reaction did not yield false-positive results. LAMP remained sensitive and specific despite the addition of untreated biological fluids such as stool or urine that commonly inhibit PCR amplification. Whereas the detection of microorganisms from whole blood or a blood-culture medium typically requires extensive sample purification and removal of inhibitors, LAMP amplification remained more sensitive than conventional qPCR when omitting such preparatory steps. Our results demonstrate that LAMP is not only easy to use, but is also a very robust, innovative and powerful molecular diagnostic method for both industrialized and developing countries.

Keywords
  • loop-mediated isothermal amplification (LAMP)
  • strand displacement
  • PCR
  • performance
  • diagnostic
  • Salmonella

Introduction

The detection and identification of bacterial pathogens from clinical samples is crucial to determine the etiology of infection and to direct antimicrobial therapy. Much work has been carried out to develop rapid, economic, reliable and standardized bacterial detection methods for routine clinical use. To date, reference methods for bacterial detection or identification rely mainly on culture-based approaches that are time-consuming and labor intensive (Tang et al., 1997). These methods are now challenged by sequence-specific DNA amplification strategies, which can significantly decrease the time required to obtain results, especially for non- or slow-growing organisms (Tang et al., 1997). The speed of molecular testing allows the early implementation of specific and effective therapy, with the potential to improve outcomes and decrease drug-related selection pressure and costs.

Recent efforts in high-throughput sequencing have contributed to the elucidation of numerous genomes or specific characteristics of bacterial pathogens, enabling molecular-based detection and identification. Most molecular assays are based on the utilization of the PCR (Saiki et al., 1985; Mullis et al., 1986). PCR is one of the most sensitive techniques, but is labor-intensive and often requires extensive sample preparation to eliminate amplification inhibitors (Fredricks & Relman, 1998). Most PCR-based identification assays rely on the amplification of species-specific genes (Johnson et al., 2003) such as elongation factors (Schneider et al., 1997), RNA polymerase (rpoB) (Drancourt & Raoult, 2002) or ribosomal DNA genes (Goldenberger et al., 1997; Xu et al., 2003). These targets offer enough sequence variation for species identification, but require either laborious and time-consuming post-PCR target identification strategies, such as gel electrophoresis or amplicon sequencing (Barry et al., 1991; Roth et al., 2000; Ruiz et al., 2000), or specific labeled identification reagents built into a real-time PCR system, which are expensive. PCR methods are also sensitive to contamination (Corless et al., 2000) and inhibition by the compounds present in the template material. Other detection schemes dedicated to bacterial detection have been proposed that rely on fluorescence-based strategies or massively parallel hybridization technologies that offer the possibility to perform multiplex determinations (Palacios et al., 2006). However, these techniques require skilled operators and the usage of costly equipment (Anthony et al., 2000; Corless et al., 2001; Francois et al., 2006; Ortu et al., 2006).

LAMP, a closed-tube, real-time, Loop-mediated isothermal AMPlification system that allows qualitative visual detection of amplification results, has been described recently (Notomi et al., 2000; Mori et al., 2001) and is particularly simple to perform using only a heating block. Multiple LAMP assays have been designed for microbial pathogen detection (Hara-Kudo et al., 2007; Kuhara et al., 2007; Qiao et al., 2007; Yoneyama et al., 2007). With the goal of confirming the core features of LAMP that might make this a more generalized method for molecular diagnosis in developing countries, we evaluated the performance and robustness of LAMP assays in comparison with conventional PCR. Limit of detection, linearity of the amplification and stability over a large range of pH or temperature values, as well as potential effects due to the presence of several known inhibitors of conventional PCR amplification were investigated using kits to detect Salmonella enterica serovar Typhi. The same kit was also evaluated for its potential to detect bacterial growth in commercial blood-culture bottles with minimal handling.

Materials and methods

LAMP

LAMP reagents were obtained from Eiken Chemical Company (Japan). The mixture consists of Bst polymerase, concentrated buffer and the four LAMP primers recognizing six regions of the target gene (for additional details, see http://loopamp.eiken.co.jp). The final reaction volume was 25 µL, including a volume of 5 µL for the sample. This volume was used to test a broad range of pH values. The sample was added in a volume of 4 µL and supplemented by concentrated Tris-buffer (1 M) at adjusted pH (6.3–9.5) in a volume of 1 µL. The standard isothermal amplification protocol was defined by the manufacturer and consisted in a single step of 60 min at 65 °C for Salmonella detection.

Detection of LAMP amplification

Two different devices were used to evaluate the results of LAMP amplification (Mori et al., 2001, 2006). Turbidity caused by the accumulation of magnesium pyrophosphate, and amplification byproduct, was measured using an LA-380 real-time turbidimeter (Eiken Chemical Company). Fluorescence caused by the unquenching of calcein during the reaction was detected using an endpoint fluorescence detector (Eiken Chemical Company). Turbidimetry provides a real-time assessment of the reaction and was used for detection sensitivity, elongation time and preincubation experiments. End-point fluorescence tests were performed after isothermal amplifications of S. Typhi DNA in 0.2-mL tubes using a thermal cycler (MJResearch, Bio-Rad, Hercules, CA) set at 65 °C for 60 min. Because of protein precipitation, turbidity assays in the presence of whole blood generated false-positive signals and studies with blood were therefore assessed by the end-point fluorescence detection.

Real-time qPCR analysis

All experiments were performed using an SDS7500 (ABI, Applied Biosystems, Santa Clara, CA). Primers and probes specific for the invA gene of S. Typhi were designed using primer express 1.5 software (Applied Biosystems) with default parameters. The sequences were Styinva_JH02_F: TCGTCATTCCATTACCTACC and Styinva_JH02_R: AAACGTTGAAAAACTGAGGA for primers and TCTGGTTGATTTCCTGATCGCA for the probes (coupled to FAM and TAMRA on the 5′- and 3′-ends, respectively). Preliminary experiments were performed to determine oligonucleotide concentrations showing the best sensitivity, specificity and linear range of detection. All qPCR reactions using purified DNA were performed using enzymatic kits purchased from ABgene Inc. (Rochester, NY). This optimized assay was used in all experiments after proving at least as sensitive as a previously published assay (Rodriguez-Lazaro et al., 2003) (comparative data not shown).

Clinical tests from stool and blood cultures

Stool samples were obtained from the central bacteriology laboratory and tested directly after fast decantation using low-speed centrifugation (Sethabutr et al., 2000). In these assays, the enzymatic mixture was adjusted to 20 µL. The reaction was completed by 5 µL of a mixture composed of Salmonella DNA (1 µL positive control from the Eiken kit) and a stool sample either neat or diluted twice in water.

Four different blood-culture bottles were evaluated: aerobic and anaerobic commercial bottles; Bactec (bioMérieux, Durham, NC); and BacT/Alert (Becton-Dickinson, Basel, Switzerland). Serial dilutions of blood-culture bottles were performed in a final volume of 40 µL, and then pure and diluted samples were spiked with 1 µL of Salmonella DNA (control positive from the kit). A volume of 5 µL was then added to 20 µL of the enzymatic mixture to perform the LAMP assay. Simultaneously, qPCR experiments were performed on the same bacterial DNA dilutions.

To simulate positive blood culture, a 0.5 McFarland suspension of S. enterica was prepared and blood culture bottles were spiked at an approximate titer of approximately 40 bacteria mL−1. Bottles were then incubated at 37 °C and a 1-mL volume was sampled using a syringe every hour for an 8-h period. All samplings, including the initial inocula, were plated on blood agar for enumeration. After centrifugation at 6000 g for 5 min, the supernatant was removed and the pellet was rinsed with 1 mL saline solution (0.9%). After the addition of 100 µL of Tris-EDTA (10 mM Tris–1 mM EDTA) containing 100 mg of acid-washed 100-µm-diameter glass beads (Sigma-Aldrich Chemie GmbH, Buchs, Switzerland), the mixture was vortexed at maximal power for 30 s as described previously (Francois et al., 2005). Glass beads were removed by rapid centrifugation at 2000 g for 30 s and diluted supernatants were subjected to LAMP or qPCR amplification.

Results

Variation of amplification temperatures

Assay performance (detection sensitivity and specificity) remained consistent at reaction temperatures between 57 and 67 °C (Table 1). Below and above those temperatures, detection sensitivity declined when using a standard isothermal amplification protocol (60 min amplification). At a temperature of 48 °C, amplification was never obtained despite a 60-min incubation.

View this table:
Table 1

Performance of LAMP under different physico-chemical conditions

Amplification temperature (turbidity)67 °C65 °C63 °C60 °C57 °C48 °C
Purified Salmonella DNA (5 pg)+++++
Purified Salmonella DNA (0.5 pg)++++or −
Positive control DNA Salmonella+++++
Control DNA E. coli
Negative control
Elongation time (turbidity)60 min120min180 min
Positive control DNA Salmonella+++
Control DNA E. coli+ or −
Negative control
Preincubation (turbidity)On iceRT (10min)RT(20min)37 °C (10 min)37 °C (20 min)37 °C (30 min)
Purified Salmonella DNA (0.5 pg)++++++
Positive control DNA Salmonella++++++
Negative control
pH (turbidity)pH 6.8pH 7.3pH 7.8pH 8.3pH 8.8pH 9.3pH 9.6No Tris
Positive control DNA Salmonella++++++
Purified Salmonella DNA 10 pg++++++
Negative control
Hemin (fluorescence)10 µM5 µM1µMNo hemin
Positive control DNA Salmonella++++
Purified Salmonella DNA 10 pg++++
Negative control
NaCl (turbidity)100 mM200 mM400 mM
Positive control DNA Salmonella++or −
Negative control
Anticoagulants (turbidity)HeparinHeparin (1 : 20)EDTAEDTA (1 :20)CitrateCitrate (1 : 20)
Positive control DNA Salmonella++++
Negative control
Blood (fluorescence)1 µL0.1 µL0.01 µLBlood (turbidity)1 µL0.1 µL0.01 µL
Positive control DNA Salmonella+++++++
Negative control++
Stools or urine (turbidity)1 µL0.1 µL0.01 µL
Positive control DNA Salmonella+++
Negative control
  • Two experiments generated positive signals, whereas two others were negative.

  • Signal recorded positive by the device, but visual inspection revealed very uncharacteristic curves.

  • All experiments were performed at least three times. The results indicated in the table were obtained in all replicates, except where indicated.

Variation of amplification time

Robust results were observed with elongation periods varying from 60 min (standard protocol) to 120 min. For 180 min, negative controls containing only water or nonspecific DNA remained negative, but amplification was sporadically observed for invA in the presence of Escherichia coli ATCC25922 DNA, another Enterobacteriaceae used as a negative control strain.

Variation of sample preparation time

In the standard procedure, the addition of compounds the sample was performed in a cold rack (4 °C) and required <5 min when performed by an experienced technician. Experimentally, samples were held for up to 30 min at room temperature (22 °C) as well as at 37 °C for various periods of time. The latter incubation generated false-positive results in control negative samples (Table 1). Incubation for short periods of time at 37 °C or for a longer period at 20 °C did not yield false-positive results as observed for qPCR assays. Even the limited enzymatic activity of Taq polymerase during the assembly of qPCR assays often requires the addition of uracyl-N-glycosidase to prevent such false-positive results.

Variation of pH

Tests were performed following the manufacturer's instructions over a range of pH conditions varying from 5.7 to >9.5. Negative results were observed in all the tests performed at pH≤7.3. Alkalinity altered the assay results only for a pH of 9.5 or higher.

Sensitivity (limits of detection) and linearity

Serial dilutions of purified S. Typhi DNA showed that LAMP reproducibly detected 500 fg, but failed to detect 50 fg (approximately eight genome equivalent copies). This is a detection threshold similar to that reported by another group targeting the same invA gene by qPCR (Rodriguez-Lazaro et al., 2003). However, using our home-brew qPCR assay, the limit of detection consistently reached 50 fg (Fig. 1a and b). The qPCR assay displayed linearity over the large range of genome copies tested. On the contrary, the LAMP assay was nonlinear when plotting input DNA against the time to positive signal (Fig. 1c).

Figure 1

Detection of Salmonella DNA Detection of the invA gene from Salmonella enterica serovar Typhi was achieved using real-time PCR (a) or LAMP amplification (b), the latter assessed by turbidity measurement. Both techniques evaluated the linearity of the detection procedure (c) as a function of genome copy numbers introduced into the sample. Real-time PCR yields linear signals, whereas LAMP amplification was nonlinear over the range of DNA amounts tested.

Potential utilization for the direct detection of infectious agents

In order to assess its capacity to avoid time-consuming sample preparation steps before amplification (Fredricks & Relman, 1998; Fukushima et al., 2003), the standard LAMP reaction was run after the direct addition of a number of potential inhibitory biological fluids. The addition of 1 µL of pure urine or decanted stool has no impact on the amplification efficiency of 1000 gene copies. Whole blood in limited amounts (1 µL per reaction) did not inhibit the detection of Salmonella DNA when using the turbidimetry as the readout, despite the fact that turbidity curves were altered in shape and amplitude at low DNA input concentrations (Fig. 2ac). Serial dilutions of whole blood, as well as hemin, NaCl and blood-culture media, were tested to evaluate compatibility with LAMP amplification (Table 1). Table 1 provides the ranges of concentrations for NaCl, hemin and N-actyl cysteine compatible with the LAMP amplification allowing the detection of 1000 gene copies.

Figure 2

Effect of the sample matrix on LAMP detection Illustrative examples of LAMP amplifications of Mycobacterium DNA are depicted, using either turbidity (a–b) or the fluorescence detector device (c). Rapid detection of Mycobacterium DNA was obtained using standard conditions (a) while the presence of traces of whole blood (b) showed significant alteration of the shape of the curve. Fluorescence detection of LAMP amplification allowed detecting 1000 copy genomes of Mycobacterium DNA in samples 1–6. However, positive samples showed different visual aspects: samples 1 and 2 were spiked with 1.0 and 0.1 µL of EDTA-containing whole blood, whereas samples 3 and 4 contained 1.0 and 0.1 µL of pure urine, respectively. Simultaneously, a negative control (5) and a positive control (6) are performed for all experiments.

Anticoagulants are also potential inhibitors of enzymatic reactions: EDTA and heparin affected the LAMP amplification at the concentrations used to collect blood samples. A minimum dilution of 20-fold is sufficient to obtain robust results with either substance. The presence of citrate did not affect LAMP amplification efficiency (Table 1). However, we noticed that positive samples could show unusual fluorescence, as depicted in Fig. 2c. Despite this observation, the difference in the signals between positive and negative samples remained obvious.

Blood-culture medium was tested extensively to determine the maximum amount that could be processed for the optimized detection of bacterial genomic DNA using LAMP. By applying serial dilutions, we determined that under our defined conditions, a 20-fold dilution was required before the amplification of a 4 µL input volume. By comparison, a dilution of at least 2000-fold was necessary for detecting positive signals using conventional qPCR amplification. Media from four different types of aerobic or anaerobic bottles were tested and showed similar results (not shown). Detections of Salmonella organisms in spiked blood-culture bottles were then performed on a 20-fold dilution of a 4 µL volume of blood culture. Following this procedure, a titer of 380 bacteria mL−1 was reached after 7 h of incubation. Considering the dilution required for LAMP amplification, the experiment was performed on a volume equivalent to 0.2 µL of the culture media. This translated into a detection threshold of 1.5–2 bacteria mL−1 of medium, without any time-consuming or elaborated purification strategies. Using the same conditions, qPCR failed to detect any positive signal.

Discussion

Several molecular amplification methods and procedures have been developed in order to detect or identify microbial pathogens from various origins (Corless et al., 2001; Klaschik et al., 2002; Mackay et al., 2002; Fang & Hedin, 2003). PCR or real-time PCR methods using the Taq polymerase are utilizable either for the detection of or the quantitative evaluation of pathogens (Mackay et al., 2002; He & Jiang, 2005). However, these methodologies require high marginal costs as well as investment in expensive equipment, and are highly susceptible to inhibition and false positives (Fredricks & Relman, 1998; Monis & Giglio, 2006). Elaborate strategies have been described to overcome these problems, but all such strategies translate into increased time, steps and cost as well as decreased sensitivity (Klein et al., 1997; Fredricks & Relman, 1998). In this study, we showed the stability of isothermal amplification using LAMP, which yielded reproducible results through a broad range of temperatures, elongation times and pH values. This robustness is especially desirable when harsh processing reagents are used (e.g. NaOH extraction) or when biological samples are used without extensive DNA purification. The presence of moderate concentrations of known Taq polymerase inhibitors such as whole-blood, hemin, blood culture media, N-acetyl cystein, NaCl and anticoagulant or anti-complement compounds was tolerated by the LAMP assay. In our study, the LAMP fluorescent detection method remained robust in the presence of >5% whole blood, whereas Taq polymerase was totally inhibited with <1% blood (Panaccio & Lew, 1991). In terms of linearity, real-time PCR performed particularly well with linear reactions over more of four orders of magnitude input DNA. For the evaluation of the linearity of LAMP amplification, time to obtain a positive result (instead of cycle threshold) showed poor linearity. Moreover, multiplexing ability has yet to be demonstrated for LAMP. Thus, while this technology appears to be useful for rapid detection, it is currently of limited use for quantitative assessments or for multiplexing.

Another important aspect of LAMP was its tolerance of various elongation periods, as well as changes in conditions during reagent preparation. Resistance to prolonged warming of the master mix is a critical difference from conventional PCR assays, which require cold blocks for the preparation of reaction mixtures or the use of UNG to avoid false-positive results. The LAMP assay tested in this study also appears to be particularly robust to physicochemical parameter modifications, generating reliable results in a range of 2 pH units or 10 °C changes in incubation temperature. Note, however, that under such extreme conditions, a dose-dependent effect was recorded and the detection of the lowest amount of target (0.5 pg) was not observed at 57 °C. This observation is probably due to an altered efficiency of the amplification reaction performed under different conditions than the optimal ones recommended by the manufacturer.

Biological samples also contain well-described inhibitors of Taq (Panaccio & Lew, 1991; Fredricks & Relman, 1998), but surprisingly, the LAMP performed well despite the presence of untreated urine or stools, materials that typically demand processing via filtration or with chaotrophic salts (Klein et al., 1997). Robust amplification and detection (via turbidity or fluorescence) was obtained after the direct use of 1/20 dilutions of blood culture media (aerobe and anaerobe bottles). Under similar conditions, qPCR failed to yield any amplification product.

LAMP is not only sensitive, but fast, and highly robust under circumstances of impure preparations and variable incubation times. These are ideal characteristics for a test deployed in the clinical setting or in the developing world. In comparison, PCR applications offer quantitative capability and some limited multiplexing — but given its cost, specialized equipment needs and stringent technical requirements, PCR appears unlikely to prove adaptable to widespread deployment in the clinic or in the developing world, where the need for rapid diagnosis of emerging infections is most urgent. We note that nucleic acid sequence-based amplification (NASBA) is another isothermal technique that shares some advantages with LAMP and has been used to detect pathogens in clinical samples (e.g. Mugasa et al., 2009), but, like PCR, NASBA requires a pre-test nucleic acid extraction step, which adds time, cost and complexity to analysis compared with LAMP. Preliminary data indicate that LAMP is able to fill this gap. In summary, these validated LAMP assays proved highly robust to multiple physical and chemical conditions, clearly outperformed an optimized qPCR assay in a proof-of-concept assay and promised fulfillment of performance and stability requirements for every diagnostic laboratory.

Acknowledgements

This work was supported by FIND and by grants from the Swiss National Science Foundation 3100A0-116075/1 (P.F.) and 3100A0-112370/1 (J.S.).

Footnotes

  • Editor: Patrik Bavoil

References

View Abstract