Gene transcript amplification from cell lysates in continuous-flow microfluidic devices
- First Online:
- Cite this article as:
- Gonzalez, A., Ciobanu, D., Sayers, M. et al. Biomed Microdevices (2007) 9: 729. doi:10.1007/s10544-007-9083-1
- 119 Views
Continuous-flow analysis, where samples circulate encapsulated in a carrier fluid is an attractive alternative to batch processing for high-throughput devices that use the polymerase chain reaction (PCR). Challenges of continuous-flow prototypes include the hydrodynamic and biological incompatibility of the carrier fluid, microchannel fouling, sample carryover and the integration of a nucleic acid extraction and reverse transcription step. We tested two homemade, continuous-flow thermocycler microdevices for amplification of reverse-transcribed messages from cell lysates without nucleic acid extraction. Amplification yield and specificity were assessed with state-of-the-art, real-time quantitative equipment. Carryover contamination between consecutive samples was absent. Amplification specificity and interference by genomic DNA were optimized by primer design. Robust detection of the low-copy transcript CLIC5 from 18 cells per microliter is demonstrated in cultured lymphoblasts. The results prove the concept that the development of micro-total analysis systems (μ-TAS) for continuous gene expression directly from cell suspensions is viable with current technology.
KeywordsPCRGene expressionMicrofluidic devicesContinuous-flowμ-TAS
The wealth of genomic information generated recently is setting the stage for unprecedented advances in biomedical sciences. Gene expression profiling with microarrays is the tool of choice for the characterisation of otherwise indistinguishable phenotypes, and is capable of identifying patterns of expression of clinical relevance (Segal et al. 2005). Whereas large-scale gene profiling analysis systems and algorithms continue to improve, platforms for analysis of selected subsets of informative genes can offer a more appropriate technology, once differential expression has been screened. Most of these systems are based in quantitative RT-PCR, which meets the precision, speed, cost and practicability demands of routine analysis (Macfarlane and Dahle 1993; Gallagher et al. 2005; Goulter et al. 2006). Current robotic thermocyclers based on batch analysis can achieve considerable throughput but as the use of expression profiling or genetic biomarkers increases, they are likely to fall short of demand in the future.
Continuous-flow processing offers notable advantages for the analysis of large numbers of samples, allowing in addition more input flexibility and acceleration of the analytical process (Brivio et al. 2006). Microfluidic devices capable of sequence amplification from pre-extracted nucleic acid samples have been presented, with various levels of performance (Kricka and Wilding 2003; Roper et al. 2005). Microdevices made of glass, silicon or polymeric materials can be useful in certain field and clinical applications of a qualitative nature. Reusable systems, on the other hand, can better suit the demands of centralized or reference laboratories to process large numbers of samples with quantitative competence. PCR thermocycler prototypes where samples flow inside polymeric tubing or microchannels can achieve fast analysis times and high throughput (Obeid et al. 2003; Park et al. 2003; Hashimoto et al. 2004). These prototypes are prone to surface imperfections that can affect reaction efficiency and precision. The serious problems of material biocompatibility are only partially overcome with passivation strategies (Schneegass and Kohler 2001). Carryover contamination between consecutive samples, something crucial in PCR given its amplification power, can conceivably be eliminated by passivation, but a washing step between samples is needed. In addition, microchannel fouling has been observed after continued use (Obeid et al. 2003). In contrast, biphasic, continuous-flow systems, where samples circulate as encapsulated liquid microreactors in a carrier fluid (Urban et al. 2006) seem to be a suitable alternative, potentially capable of overcoming both the biocompatibility and the carryover obstacles (Dorfman et al. 2005).
An additional drawback for highly processive gene expression analysis is the automatization of the nucleic acid extraction step and its integration with reverse transcription and PCR amplification. From an engineering perspective, the extraction step increases design complexity. From the analytical point of view, reproducible analysis requires that the nucleic acid extraction from the sample, potentially of much reduced volume, is complete. Although relatively simple systems for automatized DNA sample preparation have been devised (Legendre et al. 2006), the reproducibility of the process has not been thoroughly documented. When the sample is a cell suspension (e.g. blood), extraction may be substituted by using cell lysis reagents compatible with subsequent reactions. Chomczynski has reported one such solution compatible with DNA amplification (Chomczynski and Rymaszewski 2006). Amplification without extraction has been reported also for RNA (Pastorino et al. 2005). This growing interest in extraction-free amplification is also evidenced from the development of commercially available cell lysis solutions compatible with downstream PCR and RT-PCR (see Section 2). With the use of these solutions it is thus plausible to obtain gene expression data directly from cell samples in a PCR microdevice. The aim of the present study was to test this concept, using self-manufactured microfluidic thermocyclers that use cDNA obtained directly from lysates of cultured lymphoblasts for amplification in biphasic, continuous-flow mode.
2 Material and methods
2.1 Description of the continuous-flow PCR microdevices
Further testing took place on a second thermal cycler endowed with an optical window to accommodate fluorescence reading. The operating thermal zones required for PCR are maintained at constant temperature by aluminum blocks, three zones on an upper plane and three more on a lower plane (Fig. 1(b)). A preheat region at the entrance to the thermal cycler serves to uninhibit thermoactivatable Taq polymerase prior to the first round of amplification. The denaturation and extension blocks are maintained at 95 and 72°C, respectively, with thermofoil heaters. The annealing block is maintained at the necessary temperature by circulating water, both for the upper and the lower thermal planes. In the transition region between both thermal planes, a short segment provides space for an optical area to monitor fluorescence in real time, which is currently under development. The aluminum blocks are held by polycarbonate side plates to provide structural integrity, allowing for an insulating air gap to be maintained between all thermal blocks. Biocompatible FEP Teflon capillary tubing, 0.76 mm I.D. (Upchurch Scientific) is wrapped around the thermal blocks, allowing two rounds of amplification as the reaction mixture passes through each turn of the coiled tube. Thermal control is achieved using LabView 7.1 software, which monitors each heater by means of embedded thermocouples, providing voltage to the control circuitry as required.
2.2 Sample loading and general amplification protocol
The plasmid vector pGEM5zf (+) from Promega was amplified using the primers forward 5′-AGGGTTTTCCCAGTCACGACGTT-3′ and reverse 5′-CAGGAAACAGCTATGACC-3′ (MWG Biotech). Both the LightCycler SYBR Green PCR kit (Roche) and the ABi SYBR Green Core PCR kit (Applied Biosystems) were tested. The results for plasmid DNA amplification presented here correspond to those using the LightCycler kit for yielding a stronger signal. The thermal profile was 95°C for 10 min and 30 cycles of (95°C—30 s, 55°C—30 s, 72°C—30 s), followed by one cycle (60–95°C at 2% ramp rate) to generate a dissociation curve. The results were analyzed by the Sequence Detection System (SDS) software, version 2.2.1, integrated in the AB7900HT (Applied Biosystems).
To assess the reproducibility of the thermocycler, amplification yields obtained with 10 pg of plasmid DNA were measured under identical conditions over 20 consecutive runs. The amount of plasmid DNA was selected so that amplification by cycle 30 was in the upper region of the linear phase. Carryover contamination was measured by analyzing the amplification signal obtained by substituting the DNA template by water and processing it immediately after loading a DNA sample.
2.3 Cell lysis and RT-PCR amplification of CLIC5 transcripts
The leukemia human cell line REH (ATCC® N: CRL-8286™) was used for experiments involving transcript detection. Cells were grown in RPMI-1640 adjusted to contain 10% FBS, 2 mM L-glutamine, 1.5 g/l sodium bicarbonate, 4.5 g/l glucose, 10 mM HEPES and penicillin-streptomycin antibiotic mixture. A cell suspension was washed with PBS, pelleted and resuspended in Cells-To-Signal lysis buffer (Ambion), and incubated at room temperature for 5 min. Five-fold serial dilutions of the cell lysate (from 1,000 cells per microliter downwards) were subsequently prepared with the same buffer (initially) or PBS, once it was found to lead to identical results. Three microliters of each of these solutions were reverse transcribed for 30 min at 42°C in a final volume 10 μl with MMLV reverse transcriptase, using either the specific primers used for PCR (see below) or random decamers. Three microliters of the synthesized cDNA were used subsequently for PCR amplification of the chloride channel gene CLIC5, using forward (5′-CTATGATATCCCGGCTGAGATGAC) and reverse (5′-CACGGGCATAGGCGTTCTT) primers. The samples were cycled for a total of 60 cycles of 95°C (15 s) and 60°C (60 s) in the AB7900HT system or in the microdevice with same flow rate as described in the previous section, but setting the annealing temperature at 60°C, in a reaction volume of 18 μl (corresponding to 90, 18, 3.6, 0.72, and 0.14 lysed cells per microliter). The ABi SYBR Green Core PCR kit (Applied Biosystems) was used in these experiments to follow the recommendations of the Cells-to-signal manufacturer. Dissociation curves were obtained as before and PCR yield, estimated as the −dF/dT peak, was compared between the AB7900HT system and the home-manufactured microdevice. For assay optimization, we also used a pair of intron-spanning primers to avoid the contribution of genomic DNA to the amplification signal. The primers used here were: forward, 5′-GGAGATTGACGCCAACACTTG and reverse, 5′-TTGCGGTATTTCTTGGCCAC. Results are presented in the form of dissociation curves (i.e. −dF/dTvs. temperature during the dissociation cycle) generated by the SDS software.
3.1 Amplification of plasmid DNA in continuous-flow PCR
3.2 RT-PCR of CLIC5 transcripts in crude cell lysates and protocol optimization
We were able to detect fluorescence signal in lysates down to, theoretically, 1 cell per reaction. However, amplification of cell lysates where the RT enzyme was omitted did show significant amplification of the primary peak, meaning that amplicon was being generated from genomic DNA templates. New primers, spanning exons 4 and 5 of the CLIC5 gene, separated by an intron of 10,972 bp, were used next, so that under the present PCR conditions, genomic DNA would not be amplified. With this new primer pair, a peak at 84°C was detected in the dissociation curve. An additional peak at 76°C revealing primer-dimer amplification was also detected with low amounts of cells. To minimize the chances for primer-dimer formation, the primer content in the reaction was lowered to a final concentration of 0.05 μM, which showed to optimally amplify CLIC5 transcripts directly from cell lysates at any cell content. Thereafter, parallel amplification plots at the different cell dilutions and a single peak at 84°C was consistently obtained (Fig. 4(b)).
3.3 Detection of CLIC5 transcripts from crude cell lysatesin the continuous-flow microdevice
PCR is the preferred technique for the detection of microorganisms, mutations and polymorphisms, sequencing, evaluating livestock pedigree or establishing criminal activity. The scope of applications extends further when considering the possibilities of gene expression profiling (Wong and Medrano 2005; Kubista et al. 2006). Therefore, automation and miniaturization of the PCR is nowadays a valuable goal in biotechnology. Current thermocyclers based on batch analysis can deal with limited numbers of samples but continuous-flow can provide an attractive alternative to discrete analysis. In addition, contact-free processing, where samples flow suspended in a carrier fluid, can avoid carryover contamination and deleterious effects derived from surface interactions or evaporation. On the other hand, though this alternative presents considerable advantages, precise fluidic control and liquid handling become more challenging from an engineering perspective.
We show here that continuous-flow biphasic PCR thermocyclers can generate reproducible results from lysates of just a few cells. The first prototype used in this study was similar to previously designed microdevices (Park et al. 2003; Dorfman et al. 2005) but fitted with capillary tubing of smaller diameter, and the second one added the possibility of real-time fluorescence monitoring. Our experiments add though, the following novelties: (1) the use of cell lysates instead of extracted nucleic acids, showing their compatibility with carrier fluid; (2) carry-over and reproducibility is thoroughly assessed in continuous-flow; (3) optimization and artifact minimization of a real, low-abundance transcript detection; (4) quantitative, instead of semiquantitative gel electrophoresis evaluation of the results. In any case, sample size was lower than that reported previously in a similar system (Dorfman et al. 2005). At the speeds typically used here (around 1 mm/s), a typical PCR cycle was completed in every droplet in 32 min, and a theoretical processing potential for about 3,600 such droplets per hour. Droplet stability and PCR efficiency were maintained at flow velocities below 1.5 mm/s but higher velocities induced flow disruption and dramatic decrease in efficiency as reported by others (Hashimoto et al. 2004). Carryover contamination from sample to sample was tested thoroughly because is one of the main strengths of a biphasic flow system. Negative (no-template) samples loaded through tubing previously exposed to positive samples were initially found to be easily contaminated. Therefore, two separate syringe pump systems were set up, one for positive and another one for negative samples, to test the ability of the system to avoid contamination exclusively from the liquid flow. With this configuration, negative samples introduced after positive samples never showed signs of contamination, and this was consistently observed throughout a series of experiments over 20 days. We are thus confident that the encapsulation of the PCR droplet in the immiscible carrier fluid is safe for the rigourous sample handling demands of PCR. This design obviates the need to introduce a wash step between samples (Obeid et al. 2003; Park et al. 2003) and additionally, avoids microchannel fouling. The results stress, on the other hand, the need for a network of parallel injection ports that must be cleaned or conditioned prior to the next sample input.
Quantitative nucleic acid extraction from limited amounts of biological tissue can be challenging. Often neglected, but as important, is the fact that sampling and extraction carry inherent variability, amounting on preanalytical error. This is of singular relevance for quantitative gene expression analysis, because PCR amplification, typically of the order of 108, can easily magnify random preanalytical error to intolerable levels. Indeed, variability of RNA extraction among methods is well known (Burgener et al. 2003). Therefore, we believe that any serious attempt to achieve standardizable technology should include a defined sample preparation step to minimize and delimit this error (Hoorfar et al. 2004). Potentially complete sample preparation techniques such as electrokinetic ejection of RNA from single cells have been reported (Han and Lillard 2000), although quantitative information about them is not available. An alternative to extraction is to perform amplification in the sample collection buffer, bypassing the extraction step. Using such strategy, several investigators have successfully quantified viruses by RT-PCR (Bisset et al. 2001; Pastorino et al. 2005), and detected mammalian genes in single cells (Matsunaga et al. 2003) or in microdissected frozen tissue (Klebe et al. 1996; To et al. 1998). For single-cell PCR, where extraction can be particularly problematic, one-step analysis is perhaps the only realistic approach for quantitative analysis of gene expression. This could be exploited to integrate all the necessary analytical operations, from collection to readouts, in a single device. Interest in this concept of ‘one-solution’ analysis is highlighted from newly released commercial products such as the ‘cells-to-signal®’ kit from Ambion used in this study, which includes a cell lysis buffer compatible with RT-PCR. Crude lysate dilutions did show successful amplification of the CLIC5 transcript in a commercial AB7900HT system. Generation of primer-dimer artifacts during amplification was observed when using specific primers in the reverse transcription step, but could be obviated by using random decamers instead. Integration of reverse transcription and PCR amplification can be done with enzymes such as Thermus thermophilus (Tth) and other DNA polymerases (Myers and Gelfand 1991). However, with the use of specific primers and SYBR Green PCR chemistry, the occurrence of primer-dimer artifacts is, as shown here, more likely, showing the need to optimize each primer pair-based reaction individually. Adjusting the primer concentration was necessary to eliminate primer-dimer artifacts, especially important at low template concentrations. Genomic DNA, also present in the lysate, could interfere with RNA quantitation, but this problem can be eliminated in most cases with judicious primer design. After optimization, the amplification curves in serially diluted crude cell lysates, as performed in the AB7900HT system, were parallel and showed proportional decreases in Ct value as the cell dilution increased, demonstrating a linear dynamic range of over three orders of magnitude. We could similarly detect CLIC5 transcripts in crude lysates in our continuous-flow microdevice. With optimized intron-spanning primers, robust amplification of 18 cells per microliter was achieved in 60 cycles with no amplification in the negative sample. This indicates that a sensitivity of one cell per 55 nl was reached for detection of CLIC5 transcripts in REH cells. This figure would need to be confirmed experimentally in one cell-containing droplets but the results indicate that the Cells-to-signal lysis buffer provides the means to do so. The amplified product showed one single peak at the predicted temperature, confirming its specificity. The reasons why we could not detect lower amounts of cells in the current experiments are unclear, but they may be related with specific PCR inhibitors present in white blood cells (Al-Soud and Radstrom 2001). Protocol improvement would be needed to neutralize these effects and to improve amplification efficiency. Signal to noise ratios in the microdevice were found to be comparable in the microdevice and in the AB7900HT machine, even though some photobleaching of SYBR Green during sample handling may have occurred. It is possible to achieve better amplification efficiency in continuous flow PCR than in static wells, since heat exchange and reaction kinetics can only improve in the microscale (Urban et al. 2006), although formal comparison data are not available yet, in particular for PCR and RT-PCR. We haven’t assayed other PCR chemistries because we estimated the present results as a first step to test whether cDNA solutions (which contain large amounts of detergents and other hydrophobic chemicals) behave similarly to idealized nucleic acid solutions as tested extensively and reported in the literature. Hydrolysis probe-based PCR relies on the same chemical reaction (primer extension), although using the 5′-nuclease activity of a particular class of Taq polymerase as signal generator. If the temperature profile is able to reproducibly amplify these cDNA solutions, we anticipate that other chemistries will present little difference, although this needs to be contrasted experimentally. In fact, other chemistries have been used with overlaid mineral oil, an old PCR technique, without effect on PCR yield (Walsh et al. 2006).
The obvious development for these devices now is to use a multimodular system, where trains of droplets from one sample can be sorted with new microfluidic circuits to feed the thermocycler with one droplet per sample. At the present stage though, the microdevices do not present such capability. Rather, the focus of this report has been put on the generation of reproducible amplifications from cDNA.
With development of the presented approach, quantitative gene expression can be directly analyzed in cells in suspension (e.g. white blood cells) in an integrated continuous flow thermocycler. Differential cell separation using flow cytometry or other microtechnology (Shevkoplyas et al. 2005), real-time monitoring and optimization for each individual gene, along with further integration, will be necessary to achieve such objective. However, limitations due to carryover contamination between samples, microchannel clogging, surface interactions and sample preparation reported in the recent literature can be overcome with available technology and can help in the design of future prototypes of analytical continuous-flow microdevices based on PCR.
We thank Paddy O’Regan for device manufacture, Angela Morris and Dr. Fiona Gilchrist for help with cell culture and Dr.Eric Dalton for help with computerized thermal control of the device. DC and AG were supported by the Marie Curie Research Program ToK FP6 (MTKD-CT-2004-509790).