Microfluidics for IVD analysis: Triumphs and hurdles of centrifugal platforms, Part 1: Molecular fundamentals
Despite extensive research on microfluidic nucleic acid IVDs, there are few, if any, such systems on the market.
During the past ten years, research on developing nucleic acid- (NA) and protein-based IVD tests using microfluidics has skyrocketed and has been primarily driven by the rapid progress in molecular biology and molecular diagnostics.1 The complex challenges involved in producing microfluidic systems for molecular diagnostics are usually addressed by developing separately the various pieces of equipment that handle specific individual steps within the total sample-to-answer process. However, this development approach places little or no emphasis on sample preparation, and does not consider how and from where a sample is obtained.2
The neglect of sample preparation is one of the most significant pitfalls that have prevented the widespread use of portable, integrated, microfluidic systems for IVDs and nucleic acid IVDs in particular. This first article in a three-part series defines the fundamental steps required in a sample-to-answer NA system, with an emphasis on sample preparation and a description of the technologies involved. The second article discusses centrifugal microfluidics as a technology that offers many benefits toward developing a microfluidic nucleic acid IVD device. The third article presents a vision of the future for microfluidics using centrifugal platforms as an example, and examines the challenges and solutions that reach beyond centrifugal platforms.
Figure 1. (click to enlarge
) Process flow for a typical sample-to-answer nucleic acid analysis, listing common examples of each step. Depending on targeted analyte, sample characteristics, and assay performance, all steps listed may not be required.
Fundamental Nucleic Acid IVD Process
A common sample-to-answer process flow for nucleic acid IVDs includes sample collection, sample preparation, amplification, and detection (see Figure 1). These steps are discussed in detail below.
Analyte. In performing NA diagnostics, either DNA or RNA can serve as the analyte (see Table I). In the case of DNA, the presence of foreign DNA (e.g., microbial infection) can be detected, or changes in genomic DNA (e.g., cancerous mutations) can be examined. When targeting RNA, many different types may be considered, including messenger RNA (mRNA) and ribosomal RNA (rRNA). rRNA demonstrates a high degree of evolutionary stability and is present in high concentrations inside cells.3 In the case of an extremely low pathogen concentration (e.g., sepsis), rRNA can also be a good choice for microbial detection and identification.
Table I. (click to enlarge
) Analysis-relevant contrasts between DNA and RNA, listing common characteristics of each.
A specific host gene-marker expression response can be detected, which is elicited by the presence of a microbe, by profiling mRNA production. Such detection allows for testing with smaller sample volumes, which is more compatible with microfluidic disposables.4,5 mRNA detection can also be an excellent option when genetically profiling tumor tissue.
Sample Collection and Characteristics. Although this article does not discuss sample collection methods in detail, having an overall understanding of and appreciation for how samples are collected and how they can affect NA analysis is important. Without going into details about specific conditions and diseases, the most common and important sample types include the following: blood, urine, nasopharyngeal (including direct swabs and aspirates), tissue, fecal, saliva, and interstitial fluid. The literature offers a more comprehensive overview of sample collection methods.6
The minimum concentration of an analyte that is needed to run nucleic acid IVDs must be known and considered. The scaling relation between sample sizes and concentrations is shown in Figure 2. With NA diagnostics, the analyte of interest is often present in very low concentrations and, in some cases (e.g., sepsis), can be as diluted as less than 10 DNA molecules per mL of blood when directly targeting microbes.7 This concentration is at the zeptomolar (10-21 M) level, and requires very specific and powerful amplification coupled with sensitive detection. Sample concentration also dictates how much sample needs to be collected. For example, if an assay requires at least 100 copies of DNA for detection, then in the case of sepsis, a minimum of 10 mL of blood is required. Somewhat contradictorily, a microfluidic disposable device must be able to handle and process this macro amount of fluid.
Other important sample parameters to consider include fluidic properties such as viscosity, surface tension, and density. Such properties and, more importantly, their variation among different samples can affect the microfluidic function of the disposable device. In addition, biological inhibitors in the sample must be considered, and interfering compounds should be identified and neutralized or removed.
Figure 2. (click to enlarge
) Scaling of sample concentrations and volumes, showing typical analyte loads for respiratory and blood samples (right), ranges for common detection methods, and concentrations where amplification is typically required.
While average sample characteristics can be deduced or obtained directly from the literature, extreme deviations from the average can cause a microfluidic system to fail. In particular, sick patients for whom the system is designed represent those extreme cases. For example, a microfluidic disposable device may be able to process blood from healthy patients with normal hematocrit levels. However, the disposable device may fail when hematocrit levels are unusually high or low, as is the case with many ill patients. Only by having a keen understanding of a sample and its characteristics can the downstream analysis steps be robustly designed.
Sample Preparation. Once a sample has been collected and introduced into the microfluidic system, the next task in nucleic acid IVDs is preparing the sample for analysis. The most crucial aspect of sample preparation is lysis: the NA must be released from the cells and/or viruses, thereby making it available for processing.
Various forms of lysis are commonly used, including enzymatic lysis, chemical lysis, thermal lysis, electrical-based lysis, and mechanical lysis. Chemical lysis may leave behind residual substances that must be removed or filtered out. However, such methods require little or no power, and can be activated by mixing or reconstituting reagents. While mechanical disruption requires more energy, it does not leave behind any residual chemicals. An example of this method is ultrasonic lysis, which is often powered by piezoelectric devices.
Other mechanical methods require a sample to come into contact with some energy transfer medium (e.g., beads). Nucleic acids may adsorb onto this medium, decreasing the effective concentration and ultimately degrading the amplification and detection steps. However, mechanical lysis is the most effective method for breaking down cells that have thick cell membranes (e.g., Gram-positive microbes) and for successfully extracting intact DNA.8,9 In this respect, a mechanical lysis method called bead beating is the most efficient method and functions by combining cells with an agitated mixture of milling beads.10,11
Many lysis methods are not selective since all cells and/or viruses are disrupted. In some cases, especially when preparing blood samples, performing a separation step before lysis may be necessary. Separation steps are performed at the cellular or viral level, and are usually used when inhibitors in the sample hinder downstream steps. For example, when processing blood, the red blood cells, white blood cells, and plasma can be separated using a centrifugal density gradient. Alternatively, immobilized antibodies can capture particular cells or viruses of interest.
While separation steps greatly reduce the amount of interfering compounds, a purification step at the NA level may also need to be included, in which the NA of interest is isolated. This step occurs after lysis and usually accompanies a concentration step in which beads functionalized with NA capture probes gather the NA of interest when exposed to the lysed sample. The captured NA is eluted (e.g., by heating or pH change) into a defined buffer volume. This solid-phase capture technique serves to purify and concentrate.
In theory, most or all of the sample preparation steps should ensure that a very concentrated and clean sample is delivered to the amplification and/or detection steps. Today, such sample preparation often requires a complex, lengthy, and costly system. Ideally, an integrated microfluidic system designed for blood-based microbe detection should accomplish all of the following tasks: accept a blood sample with a mL volume, perform a centrifuge-based separation step to isolate the blood plasma, perform a mechanical lysis step on the plasma to release microbial DNA, flow the resulting lysate over a column of micro beads functionalized with capture probes that hybridize only with the microbial DNA of interest, and wash an elution solution with a µL volume over the same capture beads to release only the microbial DNA. The result is a much smaller volume of solution containing a clean, pure, and concentrated microbial DNA sample. Such a complex system can be envisioned, and it does not even include the amplification and detection steps to come.
Amplification. Once sample preparation is completed, amplification of the NA of interest (i.e., the target) is required. Amplification most often involves polymerase chain reaction (PCR) or reverse transcriptase-PCR (RT-PCR), in which RNA is reverse-transcribed into DNA and amplified using PCR. While this article will not discuss in detail these amplification methods, it will examine other existing methods and terminology issues that arise when working with NA amplification.
While PCR requires a sample to be thermocycled between at least two different temperatures, other NA amplification methods exist that allow for amplification at a single temperature (i.e., isothermal amplification). Most isothermal amplification methods are enzymatic-based and amplify, much like PCR, the NA target. Examples of such alternative methods include strand-displacement amplification, helicase-dependent amplification, recombinase polymerase amplification, nucleic acid sequence–based amplification, and transcription mediated amplification.12-16
Other isothermal enzymatic amplification technologies do not amplify the targets themselves, but rather the primers. Such is the case for rolling-circle amplification and Expar, the latter of which can produce a 106-fold signal amplification in five minutes as detected by nanobead agglutination.17,18 Other isothermal amplifications are not enzymatic-based: branched DNA amplification relies only on hybridization, Nanosphere's technology uses gold nanoparticles to catalyze silver deposition when a specific NA is present, and fluorescence chain reaction uses polythiophene polymers to detect the NA of interest in very small quantities.19-21
When trying to decide which amplification method to use, or if one is required at all, fully characterizing the sample and the extent of amplification required is necessary. This task is often difficult, and is complicated further by the confusion over the nomenclature that engineers and molecular biologists use when they discuss NA amplification methods. In the analysis that follows, standard PCR is the method used to clarify some of these issues.
Molecular biologists refer to a PCR experiment/assay as “running a reaction” and state the number of starting copies of analyte DNA in a reaction (e.g., running a reaction with 100 copies). But this statement does not include important qualifiers that describe other conditions of the assay, such as the liquid volume of the reaction (e.g., whether the initial 100 copies are dissolved in 50 pL or 50 µL). Such information makes a significant difference in terms of the initial concentration of DNA required, and thus the sample preparation steps needed.
Both biologists and engineers confuse the terms limit of detection (LOD) and sensitivity when conducting and discussing PCR. In the case of amplification reactions, LOD refers to the smallest amount of material that can be replicated to a detectable concentration. LOD can have broader definitions depending on the situation, such as referring to the detection limit of an entire sample-to-answer system.
Sensitivity also has several meanings and can refer to, for example, the slope of the calibration curve for a given detection system. When discussing disease detection, sensitivity is used as a measure of a test's performance, which is defined as the ratio of identified infected patients (i.e., true positives) to all infected patients, identified or not (i.e., true positives and false negatives). In this case, sensitivity is often called clinical sensitivity. In general, LOD and sensitivity can have many different meanings, with each situation-dependent definition requiring the appropriate qualifiers for clarification.
When conducting PCR, sensitivity is often incorrectly used instead of LOD. The statement, “this PCR assay has a sensitivity of 10 copies,” does not contain the appropriate qualifiers. The intent of the statement is to convey that the LOD of this particular PCR assay under specific conditions, which remain unstated, is 10 copies. The specifications should include parameters such as thermocycling conditions, reaction volume, and the capabilities of the detection method (e.g., whether the detection is based on a fluorescence or electrochemical signal). Each parameter comes with inherent limitations. In addition, it must be stated how often the specified LOD can be achieved. For example, is the 10-copy LOD achieved for all reactions or for 95% of all reactions?
Figure 3. (click to enlarge
) Schematic showing the most common forms of optical nucleic acid detection: a) Real-time PCR using intercalating dyes; b) Real-time PCR using hairpin FRET probes that emit fluorescence upon analyte binding; c) Real-time PCR using TaqMan FRET probes that release the FRET pairing upon digestion by a polymerase; d) DNA microarray using a labeled sample; and e) DNA microarray using a capture probe-analyte-reporter-probe sandwich structure.
IVD companies and government funding agencies have allocated many resources to engineer systems that run faster PCR reactions in smaller volumes. But a source of confusion in this area stems from such statements as, “the system can run PCR in five minutes.” Again, the appropriate qualifiers are not provided. This statement often refers only to the hardware's pure thermocycling capabilities, with disregard for the actual PCR assay capabilities. A system capable of performing a single cycle in one second will likely give poor PCR results (i.e., LOD will be poor). In fact, it has been shown that slower, uniformly controlled heating and cooling rates increase the LOD of a PCR reaction.22
Finally, engineers' attempts to run PCR in very small volumes tend to ignore realistic sample concentrations. For example, a system with a single molecule LOD in a pL-sized droplet has a starting analyte concentration of about 1 × 10-12 M. By comparison, a standard yet robust 25 µL PCR assay may exhibit an LOD of 10 starting copies. The latter assay translates to a starting concentration of about 5 × 10-19 M and an LOD that is seven times better than the droplet PCR system, which has significant implications in terms of the type of sample preparation steps needed.
Detection. Once amplification is finished, if it was required at all, the final step in nucleic acid IVDs is detection. The detection step can be incorporated into amplification, and the most common method for doing so is real-time PCR (not to be confused with reverse-transcriptase PCR as described above). Real-time PCR comes in various forms and involves monitoring an increased fluorescence signal as amplification progresses (see Figure 3a-c). Once the accumulated fluorescence signal goes beyond a certain background fluorescence, or threshold, amplification is verified. The cycle number at which this verification occurs is referred to as threshold cycle (Ct). When references are included in a multiplexed fashion (e.g., 10 copies of reference A and 1000 copies of reference B), a starting concentration calibration curve can be created based on Ct values. Such a curve can back calculate the starting concentrations of the analytes of interest. In this case, real-time PCR becomes quantitative and is called qPCR.
The various types of fluorescence labeling for real-time PCR include the use of dyes that bind to double-stranded DNA and become fluorescent (see Figure 3a). An example is the commonly used SYBR Green dye (SYBR Green I). But these dyes have the disadvantage of nonspecificity, and they bind to any double-stranded DNA, so nonspecific products (amplicons) can provide false positives.
An alternative is a fluorescently labeled DNA probe that can provide specificity. These probes have the additional advantage of allowing multiplexing, as different fluorophores can be used for each analyte. They use Förster Resonance Energy Transfer (FRET) probes (i.e., a reporter and quencher) to minimize fluorescence until either a binding or polymerase-extension event occurs. In the former, a hairpin DNA probe is used with FRET quenching near the probe's ends until it is bound with the analyte (see Figure 3b). In the latter, the probe is digested by the polymerase, thus separating the FRET probes. This real-time PCR is currently the most common method used and goes by the trade name TaqMan (i.e., TaqMan probes; see Figure 3c).
Limitations on traditional PCR chemistry and fluorescent probes prevent robust multiplexing beyond 4–6 analytes at a time. When more analytes are of interest, optical DNA microarrays are often used for detection (see Figure 3d-e). Capture NA probes designed to hybridize with the NA of interest are immobilized on a substrate. The sample is fluorescently labeled and applied to the substrate, followed by a wash and/or rinse to remove nonspecifically bound and adsorbed analytes (see Figure 3d). Alternatively, the nonlabeled sample can bind to capture probes, and in a second step, a fluorescently labeled reporter probe binds to the captured sample, forming a sandwich structure. This process has the advantage of added specificity by using dual probes, but the disadvantage of increased complexity (see Figure 3e). Capture probe spots that become fluorescent mark successful hybridization and detection.
Electrochemical microarrays can achieve results similar to optical DNA microarrays, but often with less expensive hardware. However, the hardware does require complex electrical multiplexing, as each detection spot requires at least 2–3 electrodes. These electrochemical detection schemes often utilize enzymatic labels, and the enzymatic products are detected electrochemically. For example, DNA can be labeled with horseradish peroxidase via a biotin-streptavidin chemistry. These methods have the benefit of built-in signal amplification and may negate the need for NA amplification in some cases.
In order to design successfully a sample-to-answer system for nucleic acid IVDs, the desired system's specifications and characteristics must be established. Doing so requires a fundamental understanding of both the engineering and biological sides of the system, particularly when discussing the sample preparation and amplification steps. The second article of this three-part series will lay a similar groundwork for microfluidic devices and centrifugal microfluidic devices in particular. With a full understanding of the significant molecular biology and microfluidic parameters, the third article will piece together a forward-looking description of what an integrated centrifugal microfluidic IVD device for nucleic acids will look like, how it will function, and how it will be manufactured.
1. L Bissonnette, “Next Revolution in the Molecular Theranostics of Infectious Diseases: Microfabricated Systems for Personalized Medicine,” 6, no. 3 (2006): 433-450.
2. P Yager, “Point-of-Care Diagnostics for Global Health,” 10, no. 1 (2008): 107-144.
3. A Fey, “Establishment of a Real-Time PCR-Based Approach for Accurate Quantification of Bacterial RNA Targets in Water Using Salmonella as a Model Organism,” 70, no. 6 (2004): 3618-3623.
4. AM Chinnaiyan, “Molecular Signatures of Sepsis: Multiorgan Gene Expression Profiles of Systemic Inflammation,” A 159, no. 4 (2001): 1199-1209.
5. M Prucha, “Expression Profiling: Toward an Application in Sepsis Diagnostics,” 22, no. 1 (2004): 29-33.
6. J Hosek, “Methods of Mycobacterial DNA Isolation from Different Biological Material: a Review,” Veterinary Medicine 51, no. 5 (2006): 180-192.
7. ME Brecher, “Evaluation of an Automated Culture System for Detecting Bacterial Contamination of Platelets: an Analysis with 15 Contaminating Organisms,” 41, no. 4 (2001): 477-482.
8. RM Taskova, et al., “A Comparison of Cell Wall Disruption Techniques for the Isolation of Intracellular Metabolites from Pleurotus and Lepista sp., Zeitschrift Fur Naturforschung C-a,” 61 (2006): 347-350.
9. KF Jarrell, et al., “A General-Method of Isolating High-Molecular-Weight DNA from Methanogenic Archaea (Archaebacteria),” 38 (1992): 65-68.
10. EG Zoetendal, et al., “DNA Isolation Protocols Affect the Detection Limit of PCR Approaches of Bacteria in Samples from the Human Gastrointestinal Tract,” 24 (2001): 405-410.
11. JAH. Van Burik, et al., “Comparison of Six Extraction Techniques for Isolation of DNA from Filamentous Fungi,” 36 (1998): 299-303.
12. GT Walker, “Strand Displacement Amplification: an Isothermal, In Vitro DNA Amplification Technique,” 20, no. 7 (1992): 1691-1696.
13. M Vincent, “Helicase-Dependent Isothermal DNA Amplification,” 5, no. 8 (2004): 795-800.
14. O Piepenburg, et al., (2006). “DNA Detection Using recombination proteins,” 4, no. 7 (2006): e204.
15. J Compton, “Nucleic Acid Sequence-Based Amplification,” 350 (1991): 91-92.
16. A Stary, “Performance of Transcription-Mediated Amplification and Ligase Chain Reaction Assays for Detection of Chlamydial iInfection in Urogenital Samples Obtained by Invasive and Noninvasive Methods,” 36, no. 9 (1998): 2666-2670.
17. A Fire, “Rolling Replication of Short DNA Circles,” Proceedings of the National Academy of Sciences of the United States of America, 92, no. 10 (1995): 4641-4645.
18. E Tan, et al., “Isothermal DNA Amplification Coupled with DNA Nanosphere-Based Colorimetric Detection,” Analytical Chemistry 77, no. 24 (2005): 7984-7992.
19. ML Collins, “A Branched DNA Signal Amplification Assay for Quantification of Nucleic Acid Targets Below 100 Molecules/ml,” 25, no. 15 (1997): 2979-2984.
20. SJ Park, et al., (2002). “Array-Based Electrical Detection of DNA with Nanoparticle Probes,” 295: 1503-1506.
21. D Boudreau, et al., “Protein Detecting Arrays Based on Cationic Polythiophene–DNA-Aptamer Complexes,” 18, no. 20 (2006): 2703-2707.
22. YH Kim, et al., “Performance Evaluation of Thermal Cyclers for PCR in a Rapid Cycling Condition,” 44 (2008): 495-505.
is a PhD candidate near completion in the biomedical engineering department at the University of California, Irvine, and a co-founder of iGlyko Inc., a medical device company focused on in-hospital glucose monitoring. firstname.lastname@example.org
Regis Péytavi, PhD,
is a project leader at the Centre de Recherche en Infectiologie of Université Laval in Québec City, and the technological coordinator of a multidisciplinary team project focused on integration of nucleic acid assays with centrifugal microfluidics devices. email@example.com
Michel Bergeron, MD,
is professor and chairman of the Centre de Recherche en Infectiologie of Université Laval in Québec City, and the founder of Infectio Diagnostic Inc., which is now BD Diagnostic-GeneOhm. firstname.lastname@example.org
Marc Madou, PhD,
is a chancellor's professor in the mechanical and aerospace engineering department at the University of California, Irvine, and a pioneer in the fields of centrifugal microfluidics and microfabrication both in industry and academia. email@example.com
Copyright ©2009 IVD Technology