The 1998 Banff ÁTAS conference drew a large crowd. No wonder. Recent efforts in microinstrumentation, combinatorial chemistry, and microarray and microfluidics technologies have moved further from fantasy and closer to reality. Spurring the work now is the completion of the 97 million-base genome of the Caenorhabditis elegans. Meanwhile, pharmaceutical companies scramble to offer billions of dollars to join forces with the new industry's emerging giants such as Affymetrix, Nanogen, and Orchid Biotech.
While lab chips don't make analytical chemistry any easier, according to D. Jed Harrison, PhD, analytic chemist, and chipmaker at the University of Alberta in Edmonton, they are a way of developing automation. They consume less reagent volume and make the processes faster because of their smaller size. Although the microdevices have potential uses for environmental monitoring, cell sorting, protein separation, biological warfare detection, and more, much interest is focused on DNA sequencing and drug discovery.
In 1986, Roger Ekins, PhD and researcher at University College in London, began serious work on microarrays. As he explains, "Oligonucleotide microarrays are used for DNA analysis, each oligo microspot recognizing and binding a particular DNA fragment. The latter are made fluorescent using a polymerase chain reaction (PCR-based) technique, and the chip is scanned using, typically, a confocal microscope - the method first introduced by my colleagues and myself, and now adopted by Affymetrix. In principle one can construct microarrays comprising tens of thousands of different oligonucleotide microspots, though people in the field are undecided whether such numerically large arrays are necessary for most diagnostic purposes."
But, Joseph Hacia of the National Human Genome Research Institute (NHGRI) who devised an analysis for rapidly developing multiexon PCR amplification protocols in DNA chip-based hybridization, looks at it differently. While diagnostics may not need the large arrays, genomre research does. He and his colleagues used high-density arrays of more than 90,000 oligonucleotide probes, 25 nucleotides in length, to screen for all possible heterozygous germ-line mutations in the 9.17-kb coding region of the ATM gene.
The microdevices perform functions parallel to traditional benchtop steps of DNA sequencing. The goal is to evaluate simultaneously the expression levels of thousands of genes and compare differences in expression between normal and diseased or genetically manipulated cells enabling identification of the levels of proteins that change in a disease state. Compounds that have the desired effect on expression of the relevant genes, such as restoring the expression pattern to normal or mimicking the effect of a known therapeutic, can then be evaluated as drug leads.
The oligo arrays are made on a silicon surface coated with molecules that bind the DNA building blocks A, C, G, and T. These linker molecules are initially blocked with a compound that detaches when exposed to light. Various masks are used to beam the light through so that only certain areas of the chip have the linker molecules unblocked. The chip is then incubated with one of the four bases which binds to the exposed areas. Repeating the procedure and using different masks allows creation of numerous oligos.
As each oligo binds to a stretch of DNA, RNA molecules signaling gene expression from tissues can be isolated. Then the molecules are converted to DNA and fluorescently tagged.
In high density arrays, several microtitre plates can be represented on a single 80mm x 128mm filter, e.g. with a 4x4 pattern, 16 plates are condensed onto a single filter. "So you can have a 96-well plate measuring 2 ÷" x 3 ÷" with each well taking 100 mL of solution," says Harrison. "Then you can run reactions as if you had 96 beakers. The spot chemicals are in fixed locations. When you flush a sample through the array, you know that where it lights up you have a reaction. There's no need for fluidic control."
Yet, as Ekins points out, with the advent of controlled electrical fields being applied to some chips to allow the DNA fragments to find their complementary oligo in just minutes, "some authorities believe that miniaturized electrophoretic methods will prove more useful."
For instance, at Nanogen, one of the companies Harrison sees as being closest to success with their devices, Jing Cheng and his colleagues isolated particular pathogens, including Streptococcus, Staphylococcus, and cultured carcinoma cells, in the blood by using an electrophoretic method. Their one milliliter square silicon chip has a 5-by-5 array consisting of 25 round, platinum electrodes. A blood sample passes through the chip while an oscillating pattern of 10 kHz is applied to the electrodes, causing the pathogen to collect on the electrodes. Fluid washes away the blood cells, isolating the pathogen. A 400-volt shock to the bacteria "basically short-circuits the cell membrane and breaks out the DNA and RNA," according to Cheng. Another wash flushes away the protein, leaving the DNA and RNA intact. The purified genetic material is pumped out and placed on a second electronic chip where it's analyzed. All in 30 minutes. Thus, once a cell's behavior in electric fields is known, cell-sorting can be done independent of filters. And, since the test molecules are electronically concentrated over the test site, a lower concentration of target DNA molecules is required.
In an industry where only one of 10,000 compounds makes it into and through clinical trials, the pursuit of new drugs is a multi-billion dollar business. Microfluidic chips can potentially synthesize thousands of individual molecules in microchannels in minutes, instead of the hours and days traditionally needed.
Traditional approaches to drug discovery involve techniques such as sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE), size exclusion chromatography (SEC), reversed-phase HPLC, capillary electrophoresis (CE) combined with quadrupole or ion trap mass spectrometry (MS/MS), and matrix-assisted laser-desorption ionization time-of-flight spectrometry (MALDI-TOF). But, as Mike Ramsey, PhD researcher at Oak Ridge National Labs explains, the microdevices are highly parallel. Functional elements of microchips entail input and output via pipette, inkjet, electrospray coupled to MS, and MALDI. In addition, electrophoretic, chromatographic, or homogenous separations are possible and detection includes fluoresence, absorbance, and MS.
Microfabricated on planar substrates using standard photolithographic, etching, and deposition techniques, the microdevices contain microchannels, measuring about 80 microns wide and 10 microns deep. Both samples and reagents can be suspended in them or attached to their walls.
"Extremely small solution aliquots of less than 100 pL can be delivered to channels by controlling the potential at each port of a chip," says Ramsey. "We've demonstrated both constant volume dispensers with very high precision of 0.3% rsd at 100 pL volumes and a variable volume dispenser at <2% rsd at volumes >200 pL."
Although multiplexing requires a complex channel manifold with a greater number of channels and interconnections, microchip devices can provide results to several questions on a single sample. "Information can be provided in a serial fashion from a sample by performing a chemical separation, " says Ramsey. "For example, by performing an electrophoretic or chromatographic separation on a chip, several analytes could be separated allowing their quantification, similar to the information provided by benchtop HPLC and CE instrumentation. Alternatively, analyses of different kinds could be performed in parallel channels. This latter example would be a case where particular channels might be dedicated to certain tests."
Nanogen facilitates electronic multiplexing by controlling individual test sites independently, for addressing of capture probes and concentration of test sample molecules, which allows for the simultaneous use of biochemically unrelated molecules on the same chip. Some of Orchid's chips can do as many as 44 simultaneous chemical reactions.
Ramsey is particularly interested in a device that can perform multiplex PCR and electrophoretic sizing because of its capacity for high-throughput genetic analysis. He believes "future effort will be directed toward expanding the number of different DNA samples that can be analyzed in this format. The goal is to adapt this microchip technology to the analysis of mammalian DNAs for research, forensics, and medical diagnostics."
Affymetrix's work with high throughput screening and high density arrays to determine the specific nucleotide sequence of 705 bp of the rpoB gene of Mycobacterium tuberculosis, has demonstrated that arrays that sequence important genomic regions can simultaneously identify species and provide insight into the organism's population structure.
Caliper is developing high throughput screening assays against protein kinases involved in signal transduction for screening large chemical libraries. Because nearly a millionfold less reagent will be needed, screening of difficult-to-isolate targets is possible and more accurate data is obtained than is possible with traditional high throughput systems. This allows earlier screening in the drug discovery process and provides greater confidence in the results.
But, Russ Granzow, director of business development at Orchid Biotech, says that although they can do about 10 polymerase chain reactions, they have found that "multiplex LCR is a better method for application because it doesn't create too much non-specific binding. Besides, we need to look at chips to do part of the process, not necessarily all of it."
Each microdevice extracts substances from and conducts experiments on samples placed on the chip. Samples are manipulated and fed into a mechanical chip reader for analysis. A computer analyzes the resultant data. It's how all this happens that makes the field intriguing.
"If there's anything holding back the commercialization of lab chip devices, it's that a good sampling interface hasn't really been developed," says Harrison. "The interface between the chip and the outside world is an Achilles heel."
Certainly, it is not for want of trying. Experiments with different techniques abound. Harrison uses traditional pressure driven flow to deliver samples without disrupting the microchannels. Researchers at Caliper use sipper samples injected through a capillary attached to a chip. Mathies introduces samples via an eight-channel micropipetter.
Mark Burns, PhD and researcher in the Department of Biomedical Engineering and Chemical Engineering at the University of Michigan, Ann Arbor, and his colleagues, though, have made great strides in dealing with the injection problem. In their integrated nanoliter DNA analysis device, injection is done with a pipette into a fluid-entry port. The use of hydrophobic patches, constructed by reacting silane on a patterned aluminum mask, positioned just beyond the vent line in the injection channels, solved the problem of introducing and metering precise nL-size samples. The patches, in conjunction with an air pressure source, allow isolation and movement of drops of less than 1 nL.
"While some chips use capillary electrophoresis, most microfluidic chips drive the fluids electrokinetically," says Harrison. The electronic process has the distinct advantage of significantly accelerating the rate of hybridization. And, due to flow characterized by small inertial forces, most flow is laminar, permitting the adjacent movement of layers of different fluids without transverse convective mixing.
The microchannels work as microcircuits. Each terminates at a port or reservoir where reactions occur. A computer applies AC electric fields at rates measured in nL per second, and fluid moves through the channel. The magnitude and frequency of the energizing voltages controls the direction and speed of the sample.
While some chips make use of pumps containing electrodes that apply a voltage to pull cells and fluids through channels, electroosmotic flow (EO) can control the convective flow of fluid without pumps by using 1 to 15 kV. The net result of EO pumping, as Harrison says, "is that fluid follows the electro field lines. Since any surface has some residual charge on it, the solution that's injected creates electricity. There's an electric double layer - a thin volume that pulls the solution with it - that might only be 1 - 5 nanometers thick, about the equivalent of 1- 50 layers of water molecules."
Ramsey sees several advantages to using EO flow, including the planar flow profile. Unlike Poiseuille flow where the velocity varies continuously, EO maintains a uniform velocity at all points within the cross section of the tube. The EO flow profile allows more efficient material transport because there are no stagnation regions and minimal axial dispersion.
But as Burns notes, devices that use EO often use a non-cross-linked separation medium. "This allows for electronic manipulation," he says, "but if high voltages are used, that may limit the ability to integrate sensitive electronic components in the same device."
In chips that rely both of microarray and microfluidic technology, microvalves can be used to pneumatically control the flow. For instance, the valves in Orchid's chips are simply breaks in the capillaries that hold back fluids with surface tension. To restart the flow, electrodes protruding into the capillaries give the fluid an electric stimulus, overcoming the tension.
The chip's smaller dimensions allow lower currents and efficient heat separation, resulting in higher efficiency and less time for analysis. Also, because the lateral distances in the channels are minute, diffusion can rapidly separate molecules and small particles according to their diffusion coefficients.
Open channel electrochromatography is done by coating the channel wall with octadecylsilan, which functions as the stationary phase. EO flow then loads the sample into the chip and pumps it through. Micellar electrokinetic capillary chromatography (MECC), which also uses EO, offers some benefits over open channel electrochromatography. It provides a higher stationary phase density in the separation channel, a replaceable separation medium, and requires no coating of the capillary walls. Its current drawback is that micelles elute from the column.
Electrophoresis is widely used to separate molecules by size. In the large-scale method, an electric voltage pulls DNA samples through a porous gel, moving different fragments at different rates, causing them to separate. With capillary electrophoresis (CE) a mixture of ions can be injected into a capillary that contains a buffer. Ramsey explains, "the buffer provides a medium for producing a constant electric field when an electric potential is applied. With the electric field present, the time for the ions to migrate from the injection zone to a detection zone is measured. The migration time or velocity is a measure of the electrophoretic mobility of the ions which depends on the ratio of the charge of a molecule to its hydrodynamic radius or size. Thus, the electrophoretic mobility provides charge to size information."
In addition, reagents can be mixed with samples before or after separation using one of the techniques above. However, in hybridization array devices with external imaging, detection of components can occur without a separation stage.
The velocity of the components through the reactor chamber or reservoir determines the reaction time. By adjusting the applied potentials, the flow is precise enough to accurately determine the residence time of the reagents in the reactor. This makes it possible for the concurrent determination or reaction kinetics for parallel multiple reactions.
DNA assays, assuming the prior use of a PCR technique, generally necessitate the use of particular incubation temperatures to detect single point mutations. In many chips, resistive metal heaters embedded with plastic coating needed to electrically and chemically isolate the electronic components are placed right below the reaction reservoirs. Although this method only one side of the liquid sample, but the vertical variation in liquid temperature at this size scale is < 1Š C.
One of the advantages of Burns' integrated device is the potential solution to the problem of adapting mL-size reaction to nL-size systems since it works with discrete drops of known concentrations and volumes. Microfabricated heaters and temperature sensors then control the temperature in the reaction area for the specified time.
In glass chips, temperature variations can occur and push cells where they're not meant to go. Silicon dissipates heat better than glass, and eliminates the changes in temperature. But, the silicon creates its own reactivity problem. To mask it, various barrier layers are put in place. Peter Wilding, of the University of Pennsylvania School of Medicine in Philadelphia, uses silicon dioxide to "passivate the surface." Researchers at Nanogen put a gel on the walls that makes the silicon nonstick.
"At some point on the chip, quantification of some chemical specie(s) must performed," explains Ramsey. "In general this has been done using laser-induced fluorescence. A particular position in the fluidic circuit is probed for fluorescence material with a known response."
For instance, the chips Affymetrix developed for NHGRI contain probes with artificially constructed genetic sequences identical to a normal gene segment such as BRCA1 and BRCA2. PCR produces multiple copies of the DNA, which is copied into a single stranded RNA sample, affixed with green fluorescent dye, and fragmented into small pieces. A patient's copied RNA sample is tagged with red dye, and each sample is inserted into the probe-containing chip. Nonbinding samples are washed away and the fluorescent patterns are read by a laser. Mutation is noted as a decrease in fluorescence
Nanogen uses strand displacement amplification, a proprietary process of Becton Dickinson, that enzymatically amplifies low number of diagnostic targets in a sample. The amplification simplifies accurate detection.
While most current detection occurs off-chip, which is fine for lab-based instruments, Harrison sees this as problematic for potentially portable units that could be used for environmental monitoring, biological warfare, and even in doctor's offices. "In 1996, there was experimentation with integrating fiber optics into the devices," he says. "But it wasn't practical. The mating of micro optics to microfluidic devices is exciting."
With the high cost of labor and reagents, even incremental improvements in the efficiency and cost of chemical analysis are helpful. Mini-devices that can integrate sample injection, movement, mixing, reaction, separation, and detection offer a remarkable boon. The only external connections needed are input electronic control lines, data output lines, and an air pressure manifold.
However, as Ekins says, "Although DNA analysis is grabbing all the current headlines for a number of reasons, including the interest of pharmaceutical companies to develop drugs appropriate to individual patients' genetic make-up, antibody microarrays are also likely to major impact in obvious areas such as virology, allergy, etc. It should not be forgotten that many substances of biological importance, such as certain hormones, are of heterogeneous molecular composition, and it may well become necessary to distinguish between the different isoforms to make better clinical sense of assay results."