Analytica Chimica Acta 826 (2014) 51–60

Contents lists available at ScienceDirect

Analytica Chimica Acta journal homepage: www.elsevier.com/locate/aca

Segmented continuous-flow multiplex polymerase chain reaction microfluidics for high-throughput and rapid foodborne pathogen detection Bowen Shu, Chunsun Zhang *, Da Xing ** MOE Key Laboratory of Laser Life Science & Institute of Laser Life Science, College of Biophotonics, South China Normal University, Guangzhou 510631, China

H I G H L I G H T S

G R A P H I C A L A B S T R A C T

 A high-throughput and rapid microfluidic method for pathogen detection is proposed.  The method offers a simple and convenient way toward highthroughput DNA analysis.  Important parameters of segmented continuous-flow multiplex PCR were investigated.  The proposed method is suitable for high-throughput biomedical monitoring.

On a spiral-channel microfluidic platform, the high-throughput and rapid amplification for multiple foodborne bacterial pathogens was developed via the segmented continuous-flow multiplex PCR. Fig. 1 showed the design principle of spiral-channel segmented continuous-flow multiplex PCR amplification.

A R T I C L E I N F O

A B S T R A C T

Article history: Received 21 November 2013 Received in revised form 16 March 2014 Accepted 8 April 2014 Available online 15 April 2014

High-throughput and rapid identification of multiple foodborne bacterial pathogens is vital in global public health and food industry. To fulfill this need, we propose a segmented continuous-flow multiplex polymerase chain reaction (SCF-MPCR) on a spiral-channel microfluidic device. The device consists of a disposable polytetrafluoroethylene (PTFE) capillary microchannel coiled on three isothermal blocks. Within the channel, n segmented flow regimes are sequentially generated, and m-plex PCR is individually performed in each regime when each mixture is driven to pass three temperature zones, thus providing a rapid analysis throughput of m  n. To characterize the performance of the microfluidic device, continuous-flow multiplex PCR in a single segmented flow has been evaluated by investigating the effect of key reaction parameters, including annealing temperatures, flow rates, polymerase concentration and amount of input DNA. With the optimized parameters, the genomic DNAs from Salmonella enterica, Listeria monocytogenes, Escherichia coli O157:H7 and Staphylococcus aureus could be amplified simultaneously in 19 min, and the limit of detection was low, down to 102 copies mL 1. As proof of principle, the spiral-channel SCF-MPCR was applied to sequentially amplify four different bacterial pathogens from banana, milk, and sausage, displaying a throughput of 4  3 with no detectable crosscontamination. ã 2014 Elsevier B.V. All rights reserved.

Keywords: Microfluidics Segmented-flow Continuous-flow Multiplex polymerase chain reaction High-throughput Foodborne pathogens

1. Introduction * Corresponding author. Tel.: +86 20 85217070 8501; fax: +86 20 85216052. ** Corresponding author. Tel.: +86 20 85210089; fax: +86 20 85216052. E-mail addresses: [email protected], [email protected] (C. Zhang), [email protected] (D. Xing). http://dx.doi.org/10.1016/j.aca.2014.04.017 0003-2670/ ã 2014 Elsevier B.V. All rights reserved.

Infectious diseases caused by foodborne pathogens have attracted considerable attention, due to the significant public health and global economic impact. Among thousands of the recognized

52

B. Shu et al. / Analytica Chimica Acta 826 (2014) 51–60

foodborne pathogenic bacteria, about 20 of different bacteria including Vibrio parahaemolyticus, Salmonella spp., Staphylococcus aureus, Listeria monocytogenes, and Escherichia coli O157:H7 have been found to be responsible for most food-borne outbreaks [1]. These food-borne pathogens must be rapidly detected at all stages of food production, processing, transportation and marketing to ensure food safety along the food chain, thus requiring the rapid analysis of a large number of samples for possible contamination and then subjecting suspected samples to further confirmation [2]. However, the standard method for detection of bacteria mainly relies on specific microbiological identifications that are timeconsuming, labor intensive and low throughput, which constrains the number of samples or targets that can be tested and limits the ability to detect putatively contaminated foods [3,4]. Therefore, it is of great importance to develop high-throughput multiplexed methods capable of simultaneously detecting a number of samples for multiple foodborne pathogens. Recently, much effort has been devoted to developing highthroughput and multiplex systems that utilize various methods to detect bacterial pathogens [5–9]. Among them, the multiplex polymerase chain reaction (PCR), which enables simultaneous identification of several targets incorporation of multiple sets of primers, is promising in term of speed, reliability, high specificity and sensitivity. Furthermore, capitalizing on the advantages of microfluidic technology, such as shorter analysis time, lower reagent and power consumption, higher processing throughput and integration level, the multiplex PCR microfluidic systems have attracted a great deal of attention [10–14]. Generally, the developed microfluidic multiplex PCR are categorized into two types: static ones [10,11] and continuous-flow ones [12,13]. The continuous-flow PCR circumvent the need for repeated heating and cooling of the reaction chamber of the static ones by moving the sample through alternating temperature zones, thus providing much faster amplification speeds, simpler steady-state temperature control and lower fabrication cost. Up to now, several continuous-flow multiplex PCR systems have been demonstrated with different fluidic architectures: close-loop [12,13], bidirectionalflow [14–16], fixed-loop [17–19]. These systems are attractive in demonstrating the significant advantages of using continuous-flow microfluidic platform to perform multiplex PCR for simultaneous testing of multiple samples, including rapid thermal cycling speed and high potential of integrating other functionalities. However, these systems all only focus on the multiplex PCR amplification in a

single reaction mixture per reaction channel and the analysis throughput are still limited, thus more reaction channels are needed for further enhancement of detection throughput, which may be at the price of the enlargement of footprint, complexity of fluid operation and fabrication cost increase [16,20,21]. As a highthroughput sample processing method, the segmented-flow technique involves the use of an immiscible phase such as oil or air to divide the aqueous flow stream into discrete slugs or droplets [22–25], and this technique has been implemented in microfluidic devices to perform high-throughput single-molecule and single-cell PCR [26–29]. However, the PCR amplifications in these microfluidic devices were intended for detection of one target per slug or droplet, which constrains the number of samples or targets to be analyzed in a single test. To address the aforementioned issues, we introduce a segmented continuous-flow multiplex PCR (SCF-MPCR) on a spiral-channel microfluidic device for high-throughput and rapid DNA detection. In our prototype, four target genes in a single multiplex reaction solution were amplified for identification of four foodborne bacterial pathogens (including Salmonella enterica, L. monocytogenes, E. coli O157:H7 and S. aureus). In addition, three segmented continuous-flows were used to simultaneously amplify multiple bacterial pathogens in three different food samples. To our knowledge, we proposed for the first time a novel spiralchannel SCF-MPCR method for high-throughput and rapid detection of foodborne bacterial pathogens. 2. Experimental 2.1. Design principle of spiral-channel segmented continuous-flow multiplex PCR Fig. 1 shows the schematic of design principle for the presented spiral-channel SCF-MPCR. The n segmented flow regimes are sequentially generated in a microchannel, while the channel is coiled around three heating zones those are maintained at constant temperatures for denaturation, annealing and extension. Within the multiplex PCR mixture in each sample segment, m pairs of primers, together with the m kinds of different DNA templates, are included. In order to ascertain whether the amplicons are products of residual contamination, a control segment, containing no DNA template but primers and all other reagents necessary for multiplex PCR, is injected before each sample segment. In addition, a BSA

Fig. 1. Illustration of design principle of spiral-channel segmented continuous-flow multiplex PCR amplification.

B. Shu et al. / Analytica Chimica Acta 826 (2014) 51–60

solution plug is located before the sample or control segment, to alleviate possible PCR inhibition due to the interaction between the PCR mixture and the reactor substrate. To cleanse the carryover from the preceding sample segment to the following one, a washing buffer plug is interposed after each sample segment. When the numbers of primer pairs used in each sample segment are same, the detection throughput of the present microfluidic device should be m  n. Here, the DNA templates and primer pairs in each reaction mixture, and the number of segmented flow regimes can be flexibly selected to meet different application requirements. When the reaction mixtures in the spiral channel are driven to continuously flow through three isothermal zones of the microfluidic device, a high-throughput and rapid segmented continuous-flow multiplex PCR can be realized. According to the throughput rate of 1 sample min 1 (discussed below), the developed multiplex PCR device can perform the number of assays up to 1440 samples per day. Compared to the previously reported microfluidic designs where more reaction channels were added to realize the high-throughput [16,20,21], the method proposed here can further increase the throughput by simply adding the number of segmented flow regimes. Thus, the presented method offers a low-cost and convenient way to high-throughput nucleic acid analysis. 2.2. Assembly of the spiral-channel segmented continuous-flow multiplex PCR device The spiral-channel SCF-MPCR device mainly consists of three components: (1) a spiral-channel continuous-flow PCR system including a disposable polytetrafluoroethylene (PTFE) capillary

53

tube, three copper heating blocks, three resistance cartridge heaters (H1, H2, H3) and three K-type thermocouple sensors (S1, S2, S3), as is described in the left panel of Fig. 2A; (2) a homemade computer-controlled, LabVIEW-based temperature measurement and control system that was developed in our laboratory [16], which is illustrated in the right panel of Fig. 2A; (3) a precision programmable syringe pump system (KDS210, KD Scientific Inc., Holliston, MA, USA) for fluid actuation (Fig. 2B). The PTFE tube (0.5  0.9  3800 mm [i.d.  o.d.  length], Deqing Tonghe Plastic Research Institute, Zhejiang, China) including the inlet and outlet for fluid operation, was wrapped 30 times around the helical machined grooves (1.1 1.1 mm [width  depth]) on the surface of three heating blocks. The heating blocks were fabricated from a copper column (35  18  70 mm [o.d.  i.d.  height]) by the Automation Engineering R&M Centre (Guangzhou, China), with a segment ratio of 1:1:2 for the denaturation, annealing and extension zones. The circulating water was pumped to cool the annealing zone to obtain the necessary annealing temperature and the compact structure of the thermal cycler. To eliminate the temperature crosstalk, the two ends of heating blocks were clamped by the bakelite brackets with an air gap of 7 mm between the three heating blocks. Heater cartridges (6  70 mm [o. d.  height], 100 W, Guangzhou Haoyi Thermal Electrics Factory, Guangzhou, China) were centrally fixed into the heating blocks, and K-type thermocouple sensors (0.005 inch diameter, Omega Engineer, Stamford, CT, USA) were placed in micro-holes drilled into the columnar blocks. The temperature measurement and control system consisted of a PCI-4351 control module (National Instruments), a TBX-68T terminal block (National Instruments)

Fig. 2. (A) Schematic of the microfluidic device with a spiral-channel continuous-flow PCR system and a temperature measurement and control system. (B) Schematic of the segmented continuous-flow PCR including the syringe pump system and the capillary reactor.

54

B. Shu et al. / Analytica Chimica Acta 826 (2014) 51–60

and solid state relay actuator (SSR) modules. The temperature analog signals from the thermocouple through a TBX-68T terminal block were input to the PCI-4351 card and converted to digital ones as the feedback signals of the homemade fuzzy proportional/ integral/derivative (PID) control algorithm that determined the power input to the heaters through SSRs. 2.3. Materials and reagents Four bacterial strains used in this study included Salmonella enterica CMCC50040 (S. enterica), Escherichia coli O157:H7 GW1.0202 (E. coli O157:H7), Listeria monocytogenes CMCC54007 (L. monocytogenes), and Staphylococcus aureus CMCC26003 (S. aureus), which were obtained from Guangzhou Institute of Microbiology (Guangzhou, China). The genomic DNAs were isolated from these bacterial cultures using the TIAMamp Bacterial Genomic DNA Extraction Kit (Tiangen Biotech Co., Ltd. Beijing, China) according to the manufacturer’s instructions. The isolated DNAs were quantified on the Eppendorf Bio-Photometer and were then adjusted to 2.5  106 copies mL 1. Mock food samples were briefly prepared by spiking a known quantity of bacteria (2.0  108 colony forming units (CFU) of each pathogen) in 1 mL of food liquid or homogenate (milk, banana, or sausage). These food samples underwent DNA extraction according to the boiling method [30], and the final supernatant was used for evaluating the possibility of practical application of the presented spiral-channel SCF-MPCR. More detailed information of mock sample preparation and DNA extraction can be found in the Supplementary Information (SI). The sequences of oligonucleotide primer pairs used for multiplex PCR are listed in Table 1, which were synthesized by Invitrogen Biotechnology Co., Ltd. The four target amplicons for multiplex PCR included 317-bp fragment of invA gene for S. enterica , 275-bp fragment of rbfE gene for E. coli O157:H7, 275-bp fragment of hlyA for L. monocytogenes, and 110-bp fragment of nuc gene for S. aureus. The PCR chemicals, including 10  PCR Buffer (100 mM Tris–HCl (pH 8.3), 15 mM MgCl2, 500 mM KCl), deoxynucleotide triphosphates (dNTPs, 2.5 mM each of dATP, dGTP, dCTP, and dTTP) mixture, and Taq polymerase (5 unit mL 1), were all purchased from TaKaRa Biotechnology (Dalian) Co., Ltd. Bovine serum albumin (BSA, Fraction V, Purity 98%, Biotechnology Grade) was bought from Roche Diagnostics GmbH (Mannheim, Germany). GoldViewTM dye was obtained from SBS Genetech Co., Ltd. Beijing, China. DSTM 2000 DNA markers, which contain 2000, 1000, 750, 500, 250, and 100 bp DNA fragments, were provided by Dongsheng Biotech Co., Ltd. Guangzhou, China.

1  PCR buffer and 0.1% Triton-100 (v/v) would be used for static passivation of the capillary inner surface. And the washing buffer for removing the residual DNA from the microchannel and indicating the location of each sample plug consists of 0.2  bromophenol blue and 1  PCR buffer. Additionally, the syringes, single-use PTFE capillary and connectors which are discarded after each experiment, were treated according to the protocol of our previous work [24,25], to ensure a DNA free environment. The segmented continuous-flow regimes are generated by sequentially aspirating defined volumes of aqueous solution and air into the capillary, as depicted in Fig. 3, and the fluid manipulation is similar to droplet generation procedure reported by Chabert and co-workers [29]. Initially, the inlet of capillary as the aspirating tip is located above the “B” tube, and its outlet is connected to the gas-tight syringe of the syringe pump system. In step I, after a desired flow rate is set, the inlet is moved vertically into the “B” tube, and 5 mL of BSA solution are taken. Subsequently, the inlet is withdrawn to its initial position. In step II, the “N” tube is displaced laterally to take up the position of the “B” tube. Afterward, the inlet is moved vertically into the “N” tube and 8 mL of negative control master mix are taken, as shown in step III. During the hanging period of the inlet, an air plug is taken and the two aqueous plugs are naturally separated. Similarly, another 5 mL of BSA solution, 25 mL of master mix with sample, 5 mL of washing buffer and their air plugs are taken to produce an air-segmented continuous-flow regime. By repeating such steps, more segmented continuous-flow regimes can be sequentially generated. Since the hanging period of the inlet is 2 s, the length of air plug varies from 4 mm to 20 mm if a flow rate from 2 mm s 1 to 10 mm s 1 is utilized. In the case of 4 mm s 1, the volume of each air-segmented continuous-flow regime is approximately 54.4 mL, consisting of two 5-mL BSA plugs, 8-mL negative control plug, 25-mL sample plug, 5-mL washing buffer plug, and four air plugs (1.6 mL each). Once PCR is completed, a disposable syringe is connected to the inlet of the capillary to drive the plugs out, and all plugs are collected individually at the outlet for further analysis. 2.5. Multiplex PCR protocol Unless demonstrated otherwise, the 25-mL multiplex PCR cocktails contains 2  PCR buffer, 200 mM of each dNTP, 0.04 ng mL 1 BSA, 400 nM each of the primers, 0.4 unit mL 1 of Taq polymerase and variable amounts of template DNA. The continuous-flow multiplex PCR was carried out for 30 cycles: denaturation at 95  C, annealing at 54  C, and extension at 72  C. Within the reaction channel, unless otherwise stated, the sample plug flowed at a rate of

2.4. Sample preparation and operation of segmented continuous-flow Prior to sample loading, samples (S), negative controls (N), BSA solutions (B) and washing buffers (W) were separately prepared and stored in individual 200-mL PCR tubes to avoid contamination, as is illustrated in Fig. 2B. Here, the BSA solution including 1 ng m L 1 BSA, Table 1 Oligonucleotides used as primers in multiplex PCR. Pathogen (VMGs)

Oligonucleotide sequence (5’–3’)

S. enterica (invA)

Sal-F: CCAGC CGTCT TATCT TGA 317 Sal-R: CATCG CACCG TCAAA G Esc-F: CAGTT TACCA ACCGT CATT 275 Esc-R: GGTGG CTCCT GTGTA TTTTA Lis-F: GCCGT AAGTG GGAAA TC 198 Lis-R: ATAGG CAATG GGAAC TCC Sta-F: CGGCG TAAAT AGAAG TGG 110 Sta-R: CCGTA TCACC ATCAA TCG

E. coli O157:H7(rbfE) L. monocytogenes (hlyA) S. aureus (nuc)

Note: VMGs, virulence and marker genes.

Product size (bp)

Fig. 3. Schematic of operation of segmented continuous flow. Here, the u means the linear flow rate designated on the syringe pump. Air-segmented continuous flow plugs are generated by sequentially aspirating defined volumes of air and aqueous solution.

B. Shu et al. / Analytica Chimica Acta 826 (2014) 51–60

4 mm s 1, corresponding to the residence times of 5 s in the denaturing zone, 5 s in the annealing zone, and 11 s in the extension zone, respectively. As a positive control, the conventional multiplex PCR was performed on a commercial thermal cycler (A200, Longgene, Hangzhou, China), with the same residence time for each temperature step as that of the continuous-flow multiplex PCR protocol. 2.6. Analysis of amplification products After PCR, 5 mL of the products mixed with 1 mL 6  loading buffer were subjected to electrophoresis on 2.5% agarose gel pre-stained with GoldViewTM dye. The gel was run at 5 V cm 1 for 30 min, the gel image was then captured using Gel Doc XR system (Bio-rad) under ultraviolet illumination. The band sizes and yield of resulting amplicons were established using DSTM 2000 DNA markers as reference, where the 750-bp band was 105 ng per lane. The images were further analyzed with Quantity One software (Bio-rad), where the background was subtracted and the brightness of each lane was plotted against the vertical axis, and then all intensity values were normalized to the 750-bp band of the markers. Data are representative of the three repeated experiments. 3. Results and discussion 3.1. Effect of annealing temperatures, polymerase concentrations and flow rates on spiral-channel continuous-flow multiplex PCR For a multi-component biochemical reaction such as PCR, many parameters can affect the reaction performance. In a multiplex

55

PCR, more stringent annealing temperature is required for all the target fragments to be co-amplified. In addition, more amount of polymerase and reaction times (for example extension time) are usually required due to more than one target fragments being simultaneously amplified [31]. Therefore, the effects of these key factors on the spiral-channel continuous-flow multiplex PCR performance were evaluated here. Prior to SCF-MPCR, we have investigated the effect of annealing temperature (Ta) on continuous-flow multiplex PCR in a single segment. And, these Ta values were set at 48  C, 50  C, 52  C, 54  C, 56  C and 58  C, respectively (Fig. 4). As shown in Fig. 4, the continuous-flow multiplex PCR yielded detectable target products at each Ta value without visible unspecific products, confirming a highly specific amplification. In the presented quadruplex PCR, in addition, the yield of each target fragment apparently first increased and then decreased with the change of the Ta value from 48  C to 58  C, and the maximum yield was obtained at 54  C, where the relative fluorescence intensity (RFU) of the S. enterica amplicon was high up to 1.7 and thus the relatively balanced quadruplex PCR amplification seemed to be achieved (lane “54” in Fig. 4A and column “54” in Fig. 4B. Thus, 54  C was selected as the optimal primer/target Ta for the following continuous-flow multiplex PCR. Polymerase is a critical reagent that is responsible for catalyzing the PCR reaction. To characterize the effect of the Taq DNA polymerase concentrations on the presented continuous-flow multiplex PCR, the enzyme concentrations ranging from 0.05 to 0.5 uni mL 1 were utilized (Fig. 5). It is clear that the multiplex amplification efficiency with high concentration of polymerase (for example 0.3–0.5 unit mL 1, “0.3”, “0.4”, “0.5” lanes in Fig. 5A

Fig. 4. Effect of annealing temperatures on spiral-channel continuous-flow multiplex PCR. (A) Representative electropherogram of continuous-flow multiplex PCR products at 48, 50, 52, 54, 56, and 58  C, respectively (Lane 48–58). Lane M: Markers. For these runs, 105 copies mL 1 of each bacterial DNA template and 0.5 unit mL 1 Taq polymerase were used, and the flow rates were 2 mm s 1. (B) The fluorescence intensities from the product’s gel bands normalized to those of the 750-bp marker.

56

B. Shu et al. / Analytica Chimica Acta 826 (2014) 51–60

Fig. 5. Effect of polymerase concentrations on spiral-channel continuous-flow multiplex PCR. (A) Representative electropherogram of continuous-flow multiplex PCR products with 0.05, 0.1, 0.2, 0.3, 0.4, 0.5 unit mL 1 of Taq polymerase, respectively (Lane 0.05–0.5). Lane M: Markers. Lane P: the positive control PCR product from the conventional thermal cycler, where 0.05 unit mL 1 of Taq polymerase was used. For these runs, 105 copies mL 1 of each bacterial DNA template was used, and the flow rates were 2 mm s 1. (B) The fluorescence intensities from the product’s gel bands normalized to those of the 750-bp marker.

and “0.3”, “0.4”, “0.5” columns in Fig. 5A) appeared higher than that with a lower concentration of polymerase using the same continuous-flow multiplex PCR protocol. Here, it is to be noted that higher polymerase concentrations are required for the continuous-flow multiplex PCR to achieve the amplification comparable to that from the positive control (lane “P” in Fig. 5A and column “P” in Fig. 5B). The optimum polymerase concentration is 0.4 unit mL 1 for continuous-flow multiplex PCR; while, the recommended concentration for benchtop multiplex PCR is 0.05 unit mL 1. This observation indicates that within the presented continuous-flow multiplex PCR, although BSA has been used to statically and dynamically coat the reaction channel, there still is a possibility of polymerase loss from the aqueous phase. This may be attributed to that the presented microsystem has a relatively large surface/volume ratio and tube length, and that Taq polymerase itself has a higher adsorption affinity to most surfaces [32]. It should be acknowledged that the use of higher polymerase concentration will increase the cost of consumables for highthroughput SCF-MPCR relative to bench-top multiplex PCR. However, the sample volume used in the SCF-MPCR can be decreased to reduce the consumption of polymerase. In addition, to further minimize the non-specific adsorption, the present BSA passivation method may be required to couple with other surface modifications [33,34]. For a well-defined continuous-flow PCR device, the residence time of the PCR mixture in each reaction zone only depends on its applied flow rate in the channel. In Fig. 6, the effect of different flow rates of the reaction solution on the continuous-flow multiplex

PCR was evaluated. When the flow rate was not higher than 6 mm s 1 (“2”, “4”, “6” lanes in Fig. 6A and “2”, “4”, “6” columns in Fig. 6B), four target DNA fragments were successfully amplified in about 13–38 min after 30 cycles, this time is about one-fifth or half of the time required on the conventional PCR machine (65 min, including the times of each PCR step and temperature ramping times). Therefore, the presented spiral-channel continuous-flow multiplex PCR system is much faster than the conventional PCR machine when it applies to single-sample or low-throughput amplification. If the flow rates were further increased to 8 or 10 mm s 1, the yields of multiplex PCR products significantly reduced and longer target DNA fragments even failed to be detectable by agarose gel electrophoresis (AGE) (“8”, “10” lanes in Fig. 6A and “8”, “10” columns in Fig. 6B). This is primarily attributed to this factor: the synthesis rate of DNA polymerase is usually 60– 100 nucleotides s 1 at 72  C [13], and an extension time of less than 6 s is considered insufficient for the successful amplification of 317-bp S. enterica or 275-bp of E. coli fragment (the flow rate was 8 mm s 1 and 10 mm s 1, the corresponding extension time was 5 s and 4 s). Here, it is interesting to point out that the presented microfluidic device can perform much faster short-fragment triplex or duplex PCR in case of 8 or 10 mm s 1. As is demonstrated in Fig. 6, 198-bp of L. monocytogenes, 110-bp of S. aureus, and/or 275-bp of E. coli gene fragments were simultaneously amplified in 9.3 min or 7.5 min. This amplification speed in the presented microfluidic device has surpassed those of previously reported continuous-flow multiplex PCR devices [12–19]. Considering that faster flow rates can reduce the overall amplification time at the

B. Shu et al. / Analytica Chimica Acta 826 (2014) 51–60

57

Fig. 6. Effect of flow rates on spiral-channel continuous-flow multiplex PCR. (A) Representative electropherogram of continuous-flow multiplex PCR products at 2, 4, 6, 8, and 10 mm s 1 (Lane 2–10). Lane M: Markers. Lane N: the negative control on the continuous-flow multiplex PCR microfluidics, where the flow rate of 2 mm s 1 was used. For these runs, 105 copies mL 1 of each bacterial DNA template and 0.4 unit mL 1 Taq polymerase were used. (B) The fluorescence intensities from the product’s gel bands normalized to those of the 750-bpmarker.

cost of yield and multiplexed ability, therefore, it is necessary to balance processing speed with the limit of detection required. For this reason, a linear flow rate of 4 mm s 1 is chosen for following sensitivity test and application of segmented continuous-flow multiplex PCR. 3.2. Sensitivity of the spiral-channel continuous-flow multiplex PCR For some applications such as detection of multiple foodborne pathogens, it may be essential to obtain a detectable amount of multiplex PCR product quickly from minute amounts of target analytes. Therefore, the sensitivity of the proposed continuousflow multiplex PCR assay was determined using 10-fold dilution series of the genomic DNA templates of four bacterial strains, and the template concentrations changed from 105 to 101 copies mL 1. The sensitivity test revealed that the continuous-flow multiplex PCR assay was able to correctly identify the presence of four pathogens at the initial template concentrations from 105 to 102 copies mL 1 (lanes “105”, “104”, “103”, “102” in Fig. 7A and columns “105”, “104”, “103”, “102” in Fig. 7B). As a result, the minimum concentration of purified DNA that could be simultaneously amplified at 4 mm s 1 on the presented microfluidic device was 102 copies mL 1 (lane “102” in Fig. 7A and column “102” in Fig. 7B). Here, it is necessary to note that among several reported continuous-flow multiplex PCR assays [12–19], the initial template concentrations used to obtain fast and multiplex amplification were ranged from the order of 102–107 copies mL 1. For example, in our previous study [15], an oscillatory-flow multiplex PCR system

has been developed for the simultaneous detection of S. enterica, E. coli O157:H7, and L. monocytogenes, the minimum target concentration that could be detected was about 400, 310, and 630 copies mL 1, respectively. Recently, a more sensitive continuous-flow multiplex PCR assay has been developed [14]. On their oscillatoryflow PCR chip, as low as 102 cells mL 1 bacteria could be detected by amplifying two genes involved in biogenic amine biosynthesis. Consequently, our proposed microfluidic device has achieved an equivalent sensitivity to that reported by Sciancalepore et al., but is capable of simultaneously detecting more target genes for different foodborne bacteria. In addition, the detection limit of the proposed device is comparable to those of the conventional bench-top multiplexed PCR assays reported previously [30,35–38]. Here, it should be also noted that the experiments in Fig. 7 were repeated three times with excellent reproducibility, indicating the robustness of the continuous-flow multiplex PCR system. 3.3. Spiral-channel segmented continuous-flow multiplex PCR for simultaneous detection of multiple bacterial pathogens in three different foods To further demonstrate the high-throughput multiplexing capability of the presented continuous-flow multiplex PCR, the spiral-channel segmented continuous-flow multiplex PCR method has been used to simultaneously detect the bacterial pathogens in multiple food samples. Here, three foods (milk, banana, and sausage) were artificially contaminated with a bacterial culture mixture (about 2.0  108 CFU of each pathogen), and then the DNAs

58

B. Shu et al. / Analytica Chimica Acta 826 (2014) 51–60

Fig. 7. Sensitivity of the spiral-channel continuous-flow multiplex PCR. (A) Representative electropherogram of continuous-flow multiplex PCR products with DNA concentrations ranging from 105 to 101 copies mL 1 (Lane 105–101). Lane M: Markers. Lane N: the negative control on the continuous-flow multiplex PCR. For these runs, 1 mL DNA template of S. enterica, E. coli O157:H7, L. monocytogenes, and S. aureus, together with 0.4 unit mL 1 Taq polymerase, were used, and the flow rate was set at 4 mm s 1. (B) The fluorescence intensities from the product’s gel bands normalized to those of the 750-bpmarker.

isolated from the thermally lysed mixture were subjected to the spiral-channel SCF-MPCR amplification. As illustrated in Fig. 1, the segmented continuous-flow mode for multiplex PCR amplification of different samples is performed. During each test, three segmented continuous-flow regimes were used to detect four bacterial DNA targets in banana, milk, and sausage, respectively. And, the negative-control plug before each reaction sample was used to perform the continuous-flow multiplex PCR without any DNA templates. Fig. 8A shows a typical gel electrophoresis analysis of multiplex PCR products of four bacterial DNA targets in different food samples (lanes “banana”, “milk”, and “sausage” in Fig. 8A). Fig. 8B displays the analytical results of three independent spiralchannel SCF-MPCR amplification experiments. Here, the amounts of multiplex PCR products of the continuous-flow microfluidic system were quantified with the image analysis software (Quantity One) and compared to those of 750-bp band of the DNA markers. Seen from Fig. 8, the presented SCF-MPCR could simultaneously detect four bacterial pathogens in different food environments (lanes “banana”, “milk”, and “sausage” in Fig. 8A and columns “Banana”, “Milk”, and “Sausage” in Fig. 8B), and moreover no detectable amplification products were found in the negative controls (lanes “N” in Fig. 8A and columns “N” in Fig. 8B). The SCFMPCR assay results from one set of mocked samples in Fig. 8 were further confirmed by performing MPCR assays from another set of mocked samples in the conventional bench-top machine and the continuous-flow microfluidic device (Fig. S1 in the SI). By comparing the relative intensities from SCF-MPCR products with

those from the bench-top MPCR products, the presented SCFMPCR has a comparable amplification efficiency (Table S1 in the SI). These results indicate that the proposed spiral-channel SCFMPCR may represent an efficient tool for the high-throughput and rapid screening of food samples. However, it seems that there is some variance in the detection results in the case of banana, milk, and sausage. For example, the PCR product of E. coli O157:H7 in banana was about 27–31% of that in milk and sausage. This probably results from the different losses of bacterial cells during centrifugation and the different recovery ratios of DNA from the bacterial cells in three food samples. Obviously, the food samples used in the presented work were three distinct types of foods. In addition, the presence of PCR inhibitory substances associated with complex food matrices could hinder the reproducibility of detection. To overcome this problem, a high-quality DNA extraction method for the pathogenic bacteria in food samples, such as the magnetic-based extraction [39,40], is preferable to integrate to the presented microfluidic device in the future. 3.4. Limitations and future work The SCF-MPCR microfluidic device presented here is a prototype, designed to potentially provide a high-throughput and rapid analysis system for the presence of multiple pathogens. In practice, the throughput of the SCF-MPCR is dependent on two factors: (1) the processing numbers of PCR reactions in a given time (“n”); (2) the number of PCR reactions that can be multiplexed (“m”).

B. Shu et al. / Analytica Chimica Acta 826 (2014) 51–60

59

Fig. 8. Simultaneous detection of four foodborne pathogens in banana, milk, and sausage using spiral-channel SCF-MPCR. (A) Representative electropherogram of SCF-MPCR products. Lane banana, milk, sausage: 2 mL DNA from thermally lysed bacterial cells in banana, milk, and sausage which were artificially contaminated with S. enterica, E. coli O157:H7, L. monocytogenes, and S. aureus (2.0  108 CFU each bacterium), respectively. Lane M: Markers. Lane N: the negative control on the SCF-MPCR microfluidics. During each run, 0.4 unit mL 1 Taq polymerase and 4 mm s 1 were utilized. (B) The fluorescence intensities from the product’s gel bands normalized to those of the 750-bp marker.

The throughput rate of the presented technique can be limited to the number of reactions that can be run through a single channel in a given time with high amplification efficiency. As noted by Curcio and Roeraade [22], the value of “n” is determined by the plug size, the distance between plugs, and the flow rate. In the prototype test, each analysis regime has a volume of about 55 mL, and a flow rate of 4 mm s 1 is used to carry out the SCF-MPCR amplification. Therefore, a theoretical maximum throughput rate of approximately 0.9 samples min 1 can be achieved (the throughput rate is about 1.2 samples min 1 at 6 mm s 1). In this case, the throughput of the presented device is lower than that of a typical conventional thermocycler, where 96 samples can be amplified in about 60 min (1.6 samples min 1). There are three factors that could limit the throughput rate of the presented device: (1) a relatively large sample volume, (2) additional BSA solution for alleviating PCR inhibition, (3) negative control and washing buffer plugs required due to carryover cross contamination concerns. To circumvent these issues, the current prototype is being further improved by using droplet-based microfluidic technology. In this technology, the sample volume can be reduced to nano-liter or even pico-liter level. In addition, PCR inhibition and carryover contamination can be mitigated. These features enable the SCF-MPCR microfluidic device to perform high-efficient PCR in reduced reaction volumes, which can greatly expand the throughput rate. The SCF-MPCR throughput can also be enhanced by multiplexing, but the value of “m” is limited by the multiplexing capacity

of amplification as well as the multiplexing capacity of detection. For the work described, the upper limit of MPCR product length appears to be 350 bp due to the extension time, while the lower limit should be 50–60 bp given the length of typical primers. Based on the resolution of the bands produced by the AGE, approximately 6–8 different bands could probably be resolved, if amplification products of suitable lengths could be found for the desired target analytes. In addition, cross reactivity and differences in amplification efficiency (and thus competition for reagents) between the primer sets also limit multiplexing ability. Therefore, the practical upper limit of “m” is not more than 8 in the presented device. To further increase the multiplexing number, another detection method would probably be advantageous. For example, fluorogenic probes (i.e. TaqMan probes) can be used to identify the MPCR products, and they improve the throughput by allowing all amplicons to be shorter (thus permitting faster flow rates). In addition, inline detection or detection with a plate reader could be utilized for fluorogenic probe PCR assay, and thus the detection time could be shortened, compared to the offline gel electrophoresis analysis. 4. Conclusions In conclusion, we have developed a new SCF-MPCR amplification strategy on a spiral-channel microfluidic device for highthroughput and fast nucleic acid amplification. By implementing continuous-flow multiplex PCR in a single air-segmented flow (for

60

B. Shu et al. / Analytica Chimica Acta 826 (2014) 51–60

a single sample including the negative control), the presented microfluidic device enables a fast quadruplex PCR amplification in 19 min when a flow rate of 4 mm s 1 is utilized. And in case of 8 or 10 mm s 1 faster triplex and duplex PCR can be completed in 9.3 or 7.5 min. In addition to the rapid multiplexing capability, a sensitivity of 102 copies mL 1 comparable to those achievable in conventional bench-top multiplex PCR assays and most other continuous-flow multiplex PCR assays, can be obtained by using the mixture of genomic DNAs associated with four representative foodborne pathogens. To further improve the throughput of multiplexed analysis, the multiple air-segmented continuous-flow multiplex PCR was performed. However, it should be noted that for the presented microfluidic device, the throughput of SCF-MPCR is limited by the technique’s throughput rate and the multiplexing capacity of amplification/detection. To overcome these limitations, the current prototype can be further improved by droplet microfluidics or inline fluorogenic probe detection. We believe that the proposed SCF-MPCR method could find widespread application in medical, forensic and environmental diagnostics. Acknowledgments This research is supported by the National Natural Science Foundation of China (61072030), the National Basic Research Program of China (2010CB732602), the Key Program of NSFCGuangdong Joint Funds of China (U0931005), and the Scientific Research Foundation of Graduate School of South China Normal University (2013kyjj005). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.aca.2014.04.017. References [1] V. Velusamy, K. Arshak, O. Korostynska, K. Oliwa, C. Adley, Biotechnology Advances 28 (2010) 232. [2] A. Roda, M. Mirasoli, B. Roda, F. Bonvicini, C. Colliva, P. Reschiglian, Microchimica Acta 178 (2012) 7. [3] A.G. Gehring, S.I. Tu, Annual Reviews in Analytical Chemistry 4 (2011) 151. [4] Lui, N. Cady, C. Batt, Sensors 9 (2009) 3713. [5] Y. Gao, P.M. Sherman, Y. Sun, D. Li, Analytica Chimica Acta 606 (2008) 98. [6] L. Rodriguez-Lorenzo, L. Fabris, R.A. Alvarez-Puebla, Analytica Chimica Acta 745 (2012) 10. [7] R. Zhang, H.Q. Gong, X. Zeng, C. Sze, Analytical and Bioanalytical Chemistry 405 (2013) 4277.

[8] C.Z. Li, K. Vandenberg, S. Prabhulkar, X. Zhu, L. Schneper, K. Methee, C.J. Rosser, E. Almeide, Biosensors and Bioelectronics 26 (2011) 4342. [9] N. Ramalingam, Z. Rui, H.B. Liu, C.C. Dai, R. Kaushik, B. Ratnaharika, H.Q. Gong, Sensors and Actuators B-Chemical 145 (2010) 543. [10] E.T. Lagally, C.A. Emrich, R.A. Mathies, Lab on a Chip 1 (2001) 102. [11] J. Yang, Y. Liu, C.B. Rauch, R.L. Stevens, R.H. Liu, R. Lenigk, P. Grodzinski, Lab on a Chip 2 (2002) 179. [12] N. Agrawal, Y.A. Hassan, V.M. Ugaz, Angewandte Chemie International Edition 46 (2007) 4316. [13] Y. Sun, Y.C. Kwok, N.T. Nguyen, Lab on a Chip 7 (2007) 1012. [14] A.G. Sciancalepore, E. Mele, V. Arcadio, F. Reddavide, F. Grieco, G. Spano, P. Lucas, G. Mita, D. Pisignano, Food Microbiology 35 (2013) 10. [15] H.Y. Wang, C.S. Zhang, D. Xing, Microchimica Acta 173 (2011) 503. [16] C.S. Zhang, H.Y. Wang, D. Xing, Biomedical Microdevices 13 (2011) 885. [17] Y.W. Chen, H. Wang, M. Hupert, S.A. Soper, Analyst 138 (2013) 1075. [18] Y.W. Chen, H. Wang, M. Hupert, M. Witek, U. Dharmasiri, M.R. Pingle, F. Barany, S.A. Soper, Lab on a Chip 12 (2012) 3348. [19] W. Wu, K.T. Loan Trinh, N.Y. Lee, Analyst 137 (2012) 2069. [20] P.C. Chen, D.S. Park, B.-H. You, N. Kim, T. Park, S.A. Soper, D.E. Nikitopoulos, M.C. Murphy, Sensors and Actuators B-Chemical 149 (2010) 291. [21] D.S.W. Park, P.C. Chen, B.H. You, N. Kim, T. Park, T.Y. Lee, P. Datta, Y. Desta, S.A. Soper, D.E. Nikitopoulos, M.C. Murphy, Journal of Micromechanics and Microengineering 20 (2010) 055003. [22] M. Curcio, J. Roeraade, Analytical Chemistry 75 (2003) 1. [23] N. Park, S. Kim, J.H. Hahn, Analytical Chemistry 75 (2003) 6029. [24] Y.Y. Li, D. Xing, C.S. Zhang, Analytical Biochemistry 385 (2009) 42. [25] Y.Y. Li, C.S. Zhang, D. Xing, Microfluidics and Nanofluidics 10 (2011) 367. [26] M.M. Kiss, L. Ortoleva-Donnelly, N.R. Beer, J. Warner, C.G. Bailey, B.W. Colston, J. M. Rothberg, D.R. Link, J.H. Leamon, Analytical Chemistry 80 (2008) 8975. [27] Y. Schaerli, R.C. Wootton, T. Robinson, V. Stein, C. Dunsby, M.A.A. Neil, P.M.W. French, A.J. deMello, C. Abell, F. Hollfelder, Analytical Chemistry 81 (2008) 302. [28] R. Hartung, A. Brösing, G. Sczcepankiewicz, U. Liebert, N. Häfner, M. Dürst, J. Felbel, D. Lassner, J.M. Köhler, Biomedical Microdevices 11 (2009) 685. [29] M. Chabert, K.D. Dorfman, P. de Cremoux, J. Roeraade, J.-L. Viovy, Analytical Chemistry 78 (2006) 7722–7728. [30] S. Kawasaki, N. Horikoshi, Y. Okada, K. Takeshita, T. Sameshima, S. Kawamoto, Journal of Food Protection 68 (2005) 551. [31] O. Henegariu, N. Heerema, S. Dlouhy, G. Vance, P. Vogt, Biotechniques 23 (1997) 504. [32] R. Kodzius, K. Xiao, J. Wu, X. Yi, X. Gong, I.G. Foulds, W. Wen, Sensors and Actuators B-Chemical 161 (2012) 349. [33] L. Chen, J. West, P.A. Auroux, A. Manz, P.J.R. Day, Analytical Chemistry 79 (2007) 9185. [34] J.A. Kim, J.Y. Lee, S. Seong, S.H. Cha, S.H. Lee, J.J. Kim, T.H. Park, Biochemical Engineering Journal 29 (2006) 91. [35] A. Germini, A. Masola, P. Carnevali, R. Marchelli, Food Control 20 (2009) 733– 738. [36] D. Zhang, H. Zhang, L. Yang, J. Guo, X. Li, Y. Feng, Journal of Food Safety 29 (2009) 348–363. [37] B. Alarcón, V. García-Cañas, A. Cifuentes, R. González, R. Aznar, Journal of Agricultural and Food Chemistry 52 (2004) 7180–7186. [38] J. Chen, J. Tang, J. Liu, Z. Cai, X. Bai, Journal of Applied Microbiology 112 (2012) 823–830. [39] G. Amagliani, G. Brandi, E. Omiccioli, A. Casiere, I.J. Bruce, M. Magnani, Food Microbiology 21 (2004) 597. [40] C. Delattre, C.P. Allier, Y. Fouillet, D. Jary, F. Bottausci, D. Bouvier, G. Delapierre, M. Quinaud, A. Rival, L. Davoust, C. Peponnet, Biosensors and Bioelectronics 36 (2012) 230.