Multiplex Reverse-Transcription Loop-Mediated Isothermal Amplification Coupled with Cascade Invasive Reaction and Nanoparticle Hybridization for Subtyping of Influenza A Virus

Considering the fatal human victims and economic loss caused by influenza virus infection every year, methodologies for rapid and on-site detection of influenza viruses are urgently needed. LAMP is the most commonly used nucleic acid isothermal amplification technology suitable for on-site use. However, for multiplex LAMP, differentiation of the amplicons derived from multiple targets is still challengeable currently. Here we developed a multiplex RT-LAMP assay for simultaneous amplification of three prominent subtypes of influenza viruses (A/H5, A/H7 and 2009A/H1). The amplicons were further identified by cascade invasive reaction and nanoparticle hybridization in separate target-specific detection tubes (referred to as mRT-LAMP-IRNH). The analytic sensitivities of the assay are 10 copies of RNA for all the three HA subtypes, and the specificity reached 100%. Clinical specimen analysis showed this assay had a combined sensitivity and specificity of 98.1% and 100%, respectively. Overall, the mRT-LAMP-IRNH assay can be used as a cost-saving method that utilizes a simple instrument to detect A/H5, A/H7, and 2009A/H1 influenza viruses, especially in resource-limited settings.

Scientific RepoRts | 7:44924 | DOI: 10.1038/srep44924 RT-PCR (including real-time RT-PCR) is at present the powerful method for detection of influenza viruses [11][12][13][14][15] . However, it requires bulky and expensive equipments, as well as highly skilled technicians, which make these methods not suitable for use in resource limited regions or for field use. Recently, researches have focused on the development of isothermal amplification methods for pathogen detection. As the reaction is conducted under isothermal conditions, it can be carried out with a simple water bath so that a thermal cycler is not required 16 . As the most commonly used isothermal amplification method at present, loop-mediated isothermal amplification (LAMP) method is rapid and sensitive for amplification of DNA at a constant temperature of 60-65 °C 17,18 . Besides targeting DNA templates, LAMP can also be used to amplify RNA template by the use of reverse transcriptase together with DNA polymerase, so-called reverse transcriptase LAMP (RT-LAMP). RT-LAMP methods have been developed to detect various RNA pathogens including influenza viruses [19][20][21][22][23] .
Simultaneous detection of multiple pathogens in one tube is without doubt cost-and time-saving. However, the differentiation of the ladder-like LAMP amplicons derived from multiple targets is still challengeable to date, although previous studies have described several methods for multiplex LAMP detection. These multiplex LAMP methods either used end point analysis, through gel electrophoresis 24 or pyrosequencing 25 , or used real-time detection, through annealing curve analysis 26 , DARQ 27 or MERT-LAMP technique 28 . However, these techniques all require complicated and specialized instrumentations, which diminish the point-of-care testing capability of multiplex LAMP.
Because the aggregation of gold nanoparticles (AuNPs) can cause the change of their optical property, AuNPs have been used as a sensor for DNA detection [29][30][31] . The main merit of this sensor is the visible detection by naked eyes, and thus especially suitable for field use. In this study, we developed a multiplex RT-LAMP assay for simultaneous amplification of three prominent subtypes of influenza A viruses (H5, H7 subtypes of avian influenza A and 2009A/H1N1 viruses), and the multiplex RT-LAMP amplicons were further identified by cascade invasive reaction 32,33 and gold nanoparticle hybridization in separate target-specific detection tubes. Multiplex RT-LAMP coupled with cascade invasive reaction and nanoparticle hybridization (termed as mRT-LAMP-IRNH in brief) provides us a sensitive, specific and cost-saving diagnostic tool for identification of influenza A viruses, especially in resource-limited situations.

Results
Development and optimization of the mRT-LAMP-IRNH assay. At the commencement of this study, we used a real-time turbidimeter enabling observation of primer kinetics to determine the optimal primer sequences, primer concentrations, incubation temperature, as well as incubation time of the multiplex RT-LAMP reaction. The results showed that the optimal multiplex RT-LAMP reaction was obtained with primer sequences shown in Table 1 and the primer concentrations described in Materials and Methods. At 63 °C, the best amplification efficiency was observed for all the three HA subtypes (A/H5, A/H7 and 2009A/H1), and thus 63 °C was considered to be optimal temperature for the multiplex RT-LAMP reaction (data not shown). According to the amplification curves obtained by real-time turbidimeter, 50 min was selected as the optimal incubation time. In mRT-LAMP-IRNH, the multiplex RT-LAMP amplicons were further identified by cascade invasive reaction and nanoparticle hybridization in separate target-specific detection tubes. A schematic of the principles of cascade invasive reaction and gold nanoparticle hybridization is shown in Fig. 1. In the positive reaction, the cleaved hairpin probe cannot trigger the aggregation of AuNPs, leading to a red color of the reaction. Conversely, in the negative reaction, the intact hairpin probe leads to the aggregation of AuNPs, and then the reaction mixture becomes colorless (Fig. 1). Analytical sensitivity of the mRT-LAMP-IRNH. The analytical sensitivity of the mRT-LAMP-IRNH assay was determined by testing ten-fold serial dilutions of viral RNA transcripts (ranging from 10 4 to10 0 RNA copies) on two separate days. After 50 min of amplification, diluted RT-LAMP products were added into the invasive reaction mixture to perform cascade invasive reactions for each target, then the AuNPs were added for hybridization. The detection limits for all the three HA genes (A/H5, A/H7 and 2009A/H1) were 10 1 copies of synthetic RNA by presenting red color in the reaction tubes as shown in Fig. 2a-c. Identical results were obtained on both days indicating that the mRT-LAMP-IRNH assay was both robust and reproducible.
To further evaluate the sensitivity of cascade invasive reaction and nanoparticle hybridization (IRNH) for LAMP product detection, the multiplex RT-LAMP amplifications were also monitored by a real-time turbidimeter. As shown in Fig. 2d-f, the detection limits for A/H7 and 2009A/H1 were both 10 1 copies of synthetic RNA, which were equivalent to their detection limits obtained by mRT-LAMP-IRNH. However, the detection limit for A/H5 was 10 2 copies of synthetic RNA, which was an order of magnitude lower than that obtained by mRT-LAMP-IRNH, indicating the sensitivity of cascade invasive reaction coupled with nanoparticle hybridization was higher than real-time turbidity detection for RT-LAMP product analysis.
Analytical specificity of the mRT-LAMP-IRNH. To assess the potential for the mRT-LAMP-IRNH assay to cross-react with other genetically or clinically related viruses which could cause similar symptoms, a specificity test was conducted using viral RNA extracts from various control viruses. As shown in Fig

Figure 1. The principles of cascade invasive reaction and gold nanoparticle hybridization for detecting multiplex RT-LAMP products.
Step 1 (Primary invasive reaction): A target DNA is firstly hybridized with an upstream probe (Up) and a downstream probe (Dp), forming a one-base overlapping structure at the 3′ -end of the Up. The blue arrow indicates the site of cleavage.
Step 2 (Secondary invasive reaction): The cleaved flaps from the target-specific primary invasive reaction are used to drive a secondary invasive reaction. Then the hairpin probe (Hp) is cleaved by AfuFEN.
Step 3 (Nanopartical hybridization): When the target is present, the cleaved HP cannot trigger the aggregation of AuNPs, leading to a red color of the reaction. Conversely, when the target is absent, the Hp is intact, leading to the aggregation of AuNPs, and then the reaction becomes colorless.
Scientific RepoRts | 7:44924 | DOI: 10.1038/srep44924 Discussion Viral culture paired with serological HA typing is the standard method for detecting and typing influenza A viruses 34 . The main drawbacks of virus culture are that it can only detect live viruses and requires more time and  higher biosecurity. Immunological methods for testing influenza viruses have low skill requirements, but poor sensitivity and specificity. Recently, many molecular diagnostic approaches such as single or multiplex RT-PCR (including real-time RT-PCR) have been developed for subtyping influenza viruses. However, these methods require expensive thermal cycling equipments. Currently, rapid, reliable and affordable point-of-care tests for influenza virus detection are urgently needed, especially in resource limited regions. In this study, we describe a sensitive mRT-LAMP-IRNH assay for the detection of three influenza A viruses (subtypes A/H5, A/H7 and 2009A/H1) for the first time. The read-out can be observed by naked eyes, and no specialized instrument is required, which make this assay especially useful in resource-limited situations such as primary care facilities.
As LAMP belongs to isothermal amplification methods, it can be carried out with a simple water bath and without the need of bulky and expensive equipments, which points to the potential applicability of the assay for clinical point-of-care diagnostic use. LAMP products can be detected by agarose gel electrophoresis, turbidity or fluorescence detection 35,36 , lateral flow dipstick 23 , or even visual inspection 37,38 . These methods are robust and reliable, but detect total amplification products in a reaction and are thus limited to detection of a single target. Multiplex detection is the development trend of pathogen detection technology due to the properties of cost savings and high efficiency. In order to achieve multiplex LAMP detection, methods based on end point analysis 24,25 as well as real-time detection [26][27][28] have been employed to differentiate multiple target sequences, while these strategies all required complicated and specialized instruments. To enable multiplex pathogen detection without the need of specialized instruments, Dou et al. developed a polymer/paper hybrid microfluidic biochip for simultaneous LAMP detection of three meningitis-causing pathogens. Though this assay was not really a multiplex LAMP reaction, by using of microfluidic biochip, three pathogens were simultaneously detected on a chip, and high sensitivity and specificity were achieved within 1 h 39 . In this study, we present an alternative molecular method for multiplex LAMP amplicon detection by combining invader techniques 32,33 and gold nanoparticle probe techniques [29][30][31] . AuNPs have been used as a sensor for DNA detection for years. To improve the sensitivity, we coupled invader techniques with gold nanoparticle probe techniques, which were further used to detect multiplex RT-LAMP amplicons for the first time. Although the identification of the multiplex RT-LAMP amplicons is monoplex, the mRT-LAMP-IRNH assay developed in this study is a cost-saving method that requires no complicated instrument, and is more suitable for multiple-target detection of limited amount and/or precious nucleic acids. This pilot study also provides a strategy for establishing multiplex LAMP assay much more than three plex for field use. Meanwhile, unknown clinical specimens with co-infection can also be distinguished without using any equipments.
Due to the use of six sequence-specific primers per target, similar to that of the ordinary LAMP technique, which recognizes eight conserved regions for specific identification of positive targets, mRT-LAMP-IRNH offers the same high specificity. Moreover, the invasive reaction was a specific DNA detection method by the use of two target-specific probes. The mRT-LAMP-IRNH assay combines the specificities of both LAMP and the cascade invasive reaction, which should provide higher specificity in theory. In this study, we demonstrated the high specificity of the assay by showing the absence of cross-reactivity by analyzing the RNA extracts from various genetically or clinically related viruses which could cause similar symptoms.
The concordance of high analytical sensitivity between LAMP and other sensitive molecular methods has been reported previously 40,41 . LAMP reaction was able to tolerate the inhibitory effect of large amounts of templates, and was less affected by the presence of various salts and inhibitors 42 . In mRT-LAMP-IRNH method, aggregation of AuNPs is achieved by adding hairpin probes complementary to both AuNPs. The cleavage of hairpin probes is triggered by the cleaved flaps from the target-specific primary invasive reaction. Because one LAMP amplicon can yield large numbers of cleaved hairpin probes, the sensitivity of this method is increased significantly. By testing ten-fold serial dilutions of viral RNA transcripts, the analytical sensitivities of the mRT-LAMP-IRNH assay were found to be 10 1 copies of synthetic RNA for all three targets. Previous studies have shown that although multiplex nucleic acid detection tests have the advantage of high efficiency, due to the mutual interference of multiple primers, the analytic sensitivities of these assays tend to decrease. However, through the signal amplification effects of invader techniques and gold nanoparticle probe techniques, the sensitivity of the mRT-LAMP-IRNH assay established in this study was equivalent or even significantly increased as compared to general monoplex LAMP method 23,[43][44][45] . In this study, the detection limit for A/H5 obtained by mRT-LAMP-IRNH was ten times higher than that obtained by real-time turbidity detection, indicating the cascade invasive reaction coupled with AuNPs hybridization increased the sensitivity of LAMP whose results were usually analyzed through turbidity.
The performance characteristics of the mRT-LAMP-IRNH assay were evaluated by testing 88 clinical specimens with the parallel analysis by real-time RT-PCR. Compared to real-time RT-PCR, the sensitivities of the mRT-LAMP-IRNH assay for detecting A/H5, A/H7 and 2009A/H1 were 100%, 100% and 96.4%, respectively. As 50 min was selected for multiplex LAMP reaction, the total detection time of mRT-LAMP-IRNH assay was about 100~120 min which was similar to that of real-time RT-PCR assay. As 120 min might be a little long for field use, we had tried to shorten the invasive reaction and hybridization reaction time, but we found that when the concentration of virus was low, the result might be somewhat ambiguous to the naked eye if the detection time was shortened. In future, we would test more conditions and settings to try to shorten the reaction time, and to combine invasive reaction with hybridization reaction in one step. These would make the assay more timely and practical for field use. In summary, a highly sensitive and specific mRT-LAMP-IRNH assay was developed for the first time in this study which can detect three prominent subtypes of influenza viruses (A/H5, A/H7 and 2009A/H1). The advantages of cost-saving and no requirement of any complicated instrument make this assay more suitable for low-equipment setting laboratory use and for on-site testing. Consequently, this detection method constructed in this study would facilitate initial clinical treatment, infection control, as well as epidemiologic investigations of influenza infection.

Methods
Ethics statement. Written informed consent for using the clinical specimens was obtained from all patients

Design of primers and probes. Multiplex RT-LAMP primers for avian influenza A/H5, A/H7 and 2009A/H1
were designed using conserved regions of the HA gene for each influenza A subtype, respectively. Primers were designed using the PrimerExplorer version 4 program (Eiken Chemical Co., Tokyo, Japan). The feasibility and specificity of the primers were subsequently checked by BLAST search with sequences in GenBank. The sequence information of the final selected multiplex RT-LAMP primers is shown in Table 1. The cascade invasive reaction upstream probes and downstream probes were designed using the Universal Invader Design Software version 1.2.4 (Third Wave Technologies, Inc., Madison, USA) according to the sequences of multiplex RT-LAMP amplification products. The sequence information of the final selected upstream probes, downstream probes, hairpin probes used in secondary invasive reaction, as well as the two probes used for modification of AuNPs is shown in Table 3. All the primers and probes were synthesized by TaKaRa Biotechnology Co. Ltd. (Dalian, China).
Preparation and modification of AuNPs. AuNPs were synthesized by reducing tetrachloroauric acid with trisodium citrate according to the protocol described previously 46 . Transmission electron micrographs were taken to measure the size of synthetic particles (an average diameter of 13 nm). Two types of oligonucleotide probes (Table 3)  Cascade invasive reaction and gold nanoparticle hybridization. For each multiplex RT-LAMP reaction, three target-specific cascade invasive reactions were performed. The 20 μ l invasive reaction was carried out with 0.2 μ M upstream probe, 0.2 μ M downstream probe, 0.2 μ M hairpin probe, 100 ng AfuFEN enzyme which was prepared in our lab as described before 47 , and 5 μ l diluted multiplex RT-LAMP amplification products (20 μ l products plus 30 μ l deionized water) in a reaction buffer containing 10 mM MOPS (pH7.5), 0.05% Tween-20, 0.05% Nonidet P40, and 7.5 mM MgCl 2 . The reactions were run at 85 °C for 1 min followed by 63 °C for 20 min. After cascade invasive reaction, 3 μ l of each AuNPs (30 nM), NaCl (500 mM) and water were added in a volume of 30 μ l. Hybridization was performed at 55 °C for 30 min. The results were observed by naked eyes directly or after briefly centrifuging the products. The positive reaction mixture kept red while the negative reaction mixture became colorless.
Sensitivity and specificity of mRT-LAMP-IRNH. Ten-fold serial dilutions of synthetic RNA transcripts of the three HA genes (avian influenza A/H5, A/H7 and 2009A/H1, ranging from 10 4 to10 0 RNA copies) were used to assess the analytical sensitivity of the mRT-LAMP-IRNH assay. The specificity of the assay was determined by analyzing the RNA extracts from various control viruses mentioned above. Briefly, RNA extracted from influenza viruses A/Nanjing/2/2013(H7N9), A/Jiangsu/1/2007(H5N1), A/Jiangsu/2/2009(H1N1), or each control virus was used as RT-LAMP template, respectively. The amplicons from each RT-LAMP reaction were further identified by cascade invasive reaction and gold nanoparticle hybridization as described before in three separate target-specific detection tubes.

Clinical specimen analysis.
To investigate the feasibility of our methodology for clinical sample analysis, a total of 88 clinical specimens were collected and analyzed by mRT-LAMP-IRNH assay with the parallel analysis by our in-house real-time RT-PCR assays for influenza A/H5, A/H7 and 2009A/H1. The primers and probes used in our real-time RT-PCR assays for A/H5, A/H7 and 2009A/H1 were all recommended by WHO [48][49][50] , and all the three assays had been validated against viral culture and commercial real-time RT-PCR kits. The Real-time RT-PCR was performed using a SuperScript ® III Platinum One-Step qRT-PCR System (Invitrogen) according to the instructions.  Table 3. Probes used for cascade invasive reaction and modification of AuNPs.