created: 10/09/97, last updated: 18/09/97,© 1997 ABRF
Linda Wood Ballard, Genomics Core Facility, Huntsman Cancer Institute
Eleven years ago when Kary Mullis envisioned the polymerase chain reaction, he could not have envisioned the multitude of uses geneticists would find for it. Coupled with gel technology borrowed from DNA sequence analysis, the PCR has initiated a quantum leap in exploring the genome.
Introduction
Genetic analysis involves the detection of allelic variations or polymorphisms in the genome by a genetic marker. Polymorphisms are found both in genes and in anonymous (non-coding) regions of the genome and reflect such occurrences as base pair changes, insertions, deletions, local rearrangements, and repeats. Three classes of polymorphic markers are in common use:
RFLP's or restriction fragment length polymorphisms. Restriction enzymes recognize and cleave specific DNA sequences. A single base-pair change at such a location will result in DNA fragments of different lengths upon digestion with the enzyme. These polymorphisms are visualized using the effective but cumbersome Southern blot technology (33, 35).
VNTR's or variable number tandem repeats. These markers also utilize restriction enzymes and Southern blots. However, the restriction enzymes cleave identical sites, and the polymorphism results from the fact that the cleavage sites encompass a hypervariable region in which a short segment of DNA (11-60 bp) is repeated a variable number of times (20).
STR's or short tandem repeats. In 1989 it was recognized that the two-to-four base pair repeat regions (CA, CAG, AAAG, AGAT, etc) prolific in the genome were polymorphic and could be easily amplified using the new PCR technology (9, 10, 34). Product is visualized directly on a polyacrylamide sequencing gel.
Genotyping with polymorphic markers is done for several purposes. One major purpose is the mapping of markers or a gene. Several sophisticated mapping strategies have been developed for dealing with family studies, sib pair studies, and exclusion mapping in the general population. The basis of these is linkage analysis.
Linkage Analysis Linkage analysis takes advantage of the phenomenon of meiotic recombination. Loci which are physically close together on an allele are not as likely to be separated by meiotic recombination as those that are more distant. Loci on different chromosomes or which are on the same chromosome but separated by great distance segregate independently and the frequency of recombination is 50 percent. A recombination fraction less than 50 percent indicates loci which are linked. The recombination fraction is used as a measure of genetic distance. A statistical likelihood test is performed and is usually expressed as a LOD (logarithm of the odds) score. Obviously, the more informative meioses seen, the better the evidence for or against linkage. A LOD score of 3 is generally accepted as evidence of linkage (23).
Genetic distance and physical distance are not necessarily the same. This is due to several factors. For example, a double recombinant will show no recombination and underestimate distance, there are sexual differences in recombination, and there are recombination "hot spots" in some regions of the genome. PCR-based markers offer an advantage in that they enable a direct link between the physical and linkage maps (22).
Linkage mapping pre-supposes the ability to unambiguously determine which allele is associated with a locus. The more polymorphic a locus is, the more likely it is to be heterozygous for an individual and, thus, informative for linkage.
Obviously, with the advent of more highly polymorphic markers, it is possible to collect fewer individuals or type fewer markers for the same number of informative alleles. A high-throughput genotyping facility is of great economic advantage to a researcher, since it is much less expensive to genotype more markers than to collect and process additional DNA samples.
Mutation Detection Another purpose for genotyping is detection of gene mutations. Cardiff's Human Gene Mutation Database (Institute of Medical Genetics in Cardiff) had compiled nearly 10,000 published gene mutations by the end of 1996. The expense of sequencing has pressed the development of several alternate detection strategies. In addition to the small insertion/deletion mutations (6), researchers have used SSCP (18), heteroduplexes (12), oligonucleotide ligation products (30), and others with varying success to search for new mutations or screen for known mutations.
Finally, genotyping can reveal other types of differences between tumor and normal samples. For example, a polymorphic marker can reveal allelic imbalance (the loss or gain of an existing allele) (16); and microsatellite instability, possibly indicating a DNA replication or repair defect, is evidenced by new alleles at a di-, tri- or tetra-nucleotide repeat sites (5).
Overview of fluorescent genotyping
In the PCR-based genotyping commonly performed by a core facility, short DNA primers (which are homologous to regions flanking an area of interest in the genome) are synthesized. The study area may be of interest because it contains part of a gene or has a polymorphic region which could be used for linkage studies.
The primers are added to the target substrate in the polymerase chain reaction (PCR). Briefly, the substrate DNA is heat denatured, the reaction is cooled to allow the primers to anneal to their complementary regions, and Taq polymerase is used to extend the DNA strand from the annealed primers. This cycle is then repeated 25 to 30 times, resulting in an exponentially increasing product defined by the 5' ends of the primers.
The amplified product is size-sorted by polyacrylamide gel electrophoresis (PAGE). If one of the primers is tagged with radioactive 32P, the gel can be fixed and exposed to X-ray film producing an autoradiogram. More typically today, one primer is fluorescently tagged and the product is visualized as it passes by a scanning laser during electrophoresis. By judicious choice of PCR primers, multiple alleles can be represented by different sized PCR products and can be analyzed on a single gel lane. Labeling PCR products with different color dyes also allows one to analyze mutiple alleles in a single gel lane.. This ability to multiplex genotypes both by size and color is the key to high-throughput methods in use today.
The automated systems developed for DNA sequencing that use fluorescent tags to visualize sequencing products (e.g. the PE/ABD 373 and 377, LI-COR, Pharmacia ALF, and Molecular Dynamics FluorImager DNA sequencers) have been modified to process genotyping data. New software, specific for the needs of genotyping, was created for these platforms. A size standard, run beside the samples or in each lane, allows the software to accurately calculate the size of the PCR product. The digitized image is then displayed as a peak on an electropherogram (Figure 1).
Figure 1. Fluorescently tagged PCR products of different size ranges and colors run simultaneously in the same lane. The first electropherogram shows three genotypes tagged fluorescently with PE/ABD's phosphoramidite dye 6FAM, the second shows genotypes with the dye TET, and the third with the dye HEX. The last electropherogram is the internal size standard GS500 which is essentially a 50 base pair ladder with four additional closely spaced fragments.
These automated systems, now in service in many core laboratories, permit complex and previously time-consuming procedures such as scanning the genome for disease or gene linkage to be accomplished rapidly.
Specific methods
The following protocols, and other information pertinent to genotyping, are derived from our experience at the Genomics Core Facility at the University of Utah. This core laboratory provides high-throughput fluorescent genotyping on a cost-recovery basis to the university research community. Our genotyping services include designing, purifying, quantifying and optimizing conditions for primers, performing PCR, designing multiplexes to optimize through-put, running multiplexed product on the PE/ABD 373 DNA Sequencer, and gel analysis with Genescan and Genotyper software.
Many laboratories contemplating adding genotyping to their services have already been performing DNA synthesis or DNA sequencing. The Genomics Core Facility comes from a different background, having its roots in the marker development and genotyping groups producing markers for the human genome mapping project. Since we use PE/ABD DNA Sequencers, much of the information presented here is specific to that format.
Primers
Mapped primer sets that cover the genome are available commercially (PE/ABD, Research Genetics, Genpak). The individual researcher can use these or compose a panel tailored to a specific project. Two Web sites with good primer information are the Genome Database Browser at http://gdbwww.gdb.org, and PE/ABD's at http://www.markermap.com.
Heterozygosity for STR markers of all motifs correlates positively with the number of perfect repeats at the locus. STR markers with di-nucleotide repeats commonly show a "ladder" of lower bands, which can make allele identification difficult. It is generally observed that the more polymorphic the marker is, the more prominent the ladder. Several mechanisms have been proposed for this, including slippage and out-of-register template switching (21).
Tetra-nucleotide repeat markers are not only highly polymorphic, but also provide a more unambiguous reading, lacking the characteristically prominent "ladder" of di-nucleotide repeat markers. They do, however, have a larger size spread than the di-nucleotide markers, which means that fewer may be size multiplexed in one lane. Be aware that many markers based on tetra-nucleotide repeats also include other di- or mono-nucleotide repeats which contribute to the polymorphism. This increases heterozygosity, but may present the researcher with the challenge of identifying alleles differing by a single base pair. A particularly useful group of tetra-nucleotide repeat markers are those with the AGAT motif. They tend to amplify robustly with few if any spurious bands, generating discrete 4 bp "steps" of alleles.
Fluorescent Markers Many of the earlier fluorescent systems had the goal of replacing 32P with less hazardous detection methods. They were quite versatile in that they could capture information from Southern blots stained with SYBR Green or from actual dried polyacrylamide gels containing fluorescent or end-labeled PCR products (19). Polymorphic systems could be multiplexed, but only by size, and gel resolution was best at a window determined by the acrylamide percentage and gel running time.
The PE/ABD 373 DNA sequencing system used four different dyes, which could be "seen" by a scanning laser as the fragments moved by the laser during electrophoresis and were distinguished by the software. These NHS ester dyes (5FAM, JOE, TAMRA, and ROX) required first, the attachment of an aminohexyl linker at the 5' end of a primer during synthesis, then a second independent reaction step to covalently attach the dye to the linker and finally, purification steps to remove unreacted dyes and unlabeled primer. The labeled yield was approximately 30 percent.
This dye set was adapted for genotyping by reserving the ROX tag for the internal size standard and using the other fluorescent tags for primers. The size standard GS2500 was made from lambda phage cut with the restriction enzyme Pst. This standard was less than optimal because both strands were present and labeled producing doublet peaks and because there were large gaps between fragments at 286, 361 and 470 base pairs.
The next generation dyes, the phosphoramidites 6FAM, HEX and TET, could be added to the 5' end of the primer during synthesis, greatly decreasing the labor involved and increasing coupling yield to approximately 90 percent. The crude oligonucleotides are useful for genotyping without a purification step, although the PE/ABD 3948 Nucleic Acid Synthesis and Purification System conveniently purifies and quantifies the primers. A fourth dye, the NHS ester TAMRA, is used as the size standard and is purchased from PE/ABD. This dye set requires an upgrade to the "B" filter on the 373A instrument. This system has been working robustly for us for five years.
We quantify our fluorescently tagged primers on a spectrophotometer and then aliquot 10 tubes of primers, each sufficient to amplify one 96-well tray, plus several additional tubes of concentrated stocks. These are lyophilized and stored at -20°C until resuspended for use. A 200-nmol synthesis produces enough primer to PCR eighty to ninety 96-well trays.
DNA Quality
The DNA substrate for genotyping should be clean and of high quality. High quality means it is free from contaminants such as phenol, paraffin, dyes and salts (including excessive EDTA) and not degraded, although the PCR is more forgiving of degraded DNA than is Southern blot technology. It also needs to be of consistent concentration to avoid underloaded or overloaded samples. For fresh tissue, blood and lymphocytes, the DNA Core Facility at the University of Utah has had excellent results with the classic Maniatis protocol (28), but currently uses the Puregene DNA Isolation Kit from Gentra Systems, Inc., because it is faster and does not use hazardous phenol or chloroform.
Low Substrate Concentration Although the PCR can theoretically be performed on one molecule of substrate, low concentration can be responsible for several problems. Increasing cycle number to compensate for low concentration can increase non-specific product formation (1). In working with very low concentration substrate samples, we have observed an artifact in which an allele known from a previous PCR "drops out." In a production setting such as a core facility, low substrate amount increases the risk of seeing low-level contamination from products of previous reactions.
Product Contamination Product contamination can be avoided by physically separating DNA and PCR setup areas from the areas where reaction products are multiplexed and gels are loaded. It is important to use dedicated pipetting tools and guard in other ways against transporting product back to the "clean" area. Plasticware can be decontaminated by soaking one hour in a 10% sodium hypochlorite (Clorox™) solution (26). Robbins Scientific reports that a 10 percent bleach solution rinse will decontaminate its pipetting robot. This method is attractive because of its low cost and simplicity. Some contamination problems may need to be addressed by UV irradiation (29) or by the replacement of dTTP with dUTP in the PCR and the subsequent digestion of leftover product by uracil DNA glycosylase (15).
Archived DNA Samples Slightly degraded DNA is not generally a problem for product less than 1000 bp. However, if DNA is from archived tissue, the quality may be very poor both in terms of degradation and contamination from stains or fixatives which can poison the reaction (2). In this case you may have the difficult situation in which the less sample you use, the better the amplification will be, but the more likely you will be to see artifacts. If the DNA is severely degraded, the concentration of a particular size of substrate is effectively unknown.
Three strategies may be useful: (a) find the most appropriate DNA extraction technique to remove contaminants and make DNA available (31); (b) select primers with the smallest possible product size; and (c) perform identical reactions at different substrate concentrations, for example 20 ng and 200 ng. In any case, with degraded DNA's it is vitally important to be scrupulous in avoiding product contamination and to use a "hot start" PCR to avoid spurious bands (7). "Hot start" is discussed in the following PCR section.
The PCR
Reaction Optimization In a high-throughput lab which is attempting to maximize efficiency by using complex multiplexes, it is vitally important to have a robust PCR with no spurious product. Since reaction conditions vary from lab to lab (different thermocyclers, reagent sources, etc.) it is wise to optimize new primers even if conditions are published. We have found this to be particularly true of primers in the polymorphic alu regions where primer homology to many areas of the genome results in a narrow window of conditions between no product and product with spurious bands.
Many reagents and conditions of the reaction can be modified to produce more or less stringency, but annealing temperature and MgCl2 concentration are most common (27). Stratagene's Robocycler gradient model thermocycler is designed to provide a temperature gradient across the block and may be used to optimize reaction conditions.
Genomics Core Facility Protocol While many markers amplify successfully at a single permissive annealing temperature, the Genomics Core Facility routinely uses a modified "touchdown" PCR (8). After the initial 94°C denaturation step, we begin cycling with an annealing temperature 2°C above the calculated Tm of the primers. Then for 8 cycles we decrease the annealing temperature by 1°C increments, after which we cycle at that annealing temperature for 25 cycles.
To set up our PCR trays, we aliquot 100 ng of substrate into a PCR tube or tray, making sure the droplet is at the bottom of the tube, and allow it to air dry. The trays are then covered and stored for use within a few days. The drying does not decrease the efficiency of the PCR and gives us better control over our total reaction volume in our low humidity environment. We use a 1X reaction mix of 20 ml containing: 10 mM Tris-HCl, pH8.3; 40 mM NaCl; 200 mM of each dNTP; 1.5-4.0 mM MgCl2; 0.5 mM of each primer; and 0.5 U Taq polymerase.
This is put together on ice, vortexed briefly and added immediately to the reaction tray (also on ice) with a multipipettor. The tray is then capped and placed in the thermocycler which has been preheated and paused at the 94°C denaturing step.
Non-templated Nucleotide Addition A problem of increasing significance for automated allele calling (32) is the doublet peak formed when Taq polymerase adds a non-templated nucleotide (generally an adenosine) to the 3' end of the product. The doublet peak can confuse allele calling software necessitating time-consuming manual editing. At worst, the doublet peak prevents the calling of closely spaced alleles. A rough survey of several hundred of our markers showed this doublet to some degree in about 1/3. The ratios of "true" and "+A" products tends to be primer specific (3).
The product ratios can be influenced somewhat by reaction conditions as shown in Figure 2 (32). Higher stringency and short or no extension times (including time spent at room temperature after cycling) favor the "true" allele. Lower stringency and longer extension times, including an extended soak at 72°C after cycling, promote the "+A" product. It is generally easier to successfully drive the reaction toward the "+A" than toward the "true" allele. However, manipulating reaction conditions does not eliminate the problem for all primers and could be a source of confusion and increased error if identical conditions are not used for subsequent reactions.
Figure 2. Non-templated A addition. Examples of doublet peaks produced when Taq polymerase adds a non-templated nucleotide (generally an adenosine) to the 3'-end of the PCR product. All three electropherograms show the same sample amplified with the same set of primers, but with differing PCR conditions. The first was amplified using 1.0 mM MgCl2 and a 2-step cycle with no extension step. The second used 2.0 mM MgCl2 with a 2-step cycle, and the third used 1.25 mM MgCl2 with a 3-step cycle and an additional extension step of 30 minutes at 72°C.
Enzymatic methods for eliminating the non-templated adenosine may also be useful. Post-PCR addition of T4 DNA polymerase removes the "+A" overhang, but necessitates a second step of addition and incubation which would decrease laboratory efficiency. Use of enzymes with 3'-5' exonuclease activity holds some promise, although it increases expense somewhat.
Recently, a strategy of primer modification was developed which completely eliminates the non-templated adenosine problem. Called "PIG-tailing", the strategy involves the addition of a consensus sequence favoring the "+A" to the 5' end of the reverse primer. A thorough discussion of the "+A" problem, some predictive help for choosing primers and the "PIG-tailing" protocol are found in the excellent papers by Smith (32), Brownstein (3) and Magnuson (17).
Multiplex PCR Multiplexing the PCR reaction is attractive for increasing throughput and decreasing reaction costs. It may be essential for labs working with DNA that is in short supply. It does, however, require additional optimization to find markers that will amplify evenly together and not produce additional spurious bands. Two fairly new products have shown promise in the lab for increasing success: Perkin Elmer's AmpliTaq Gold™ and Clonetech's TaqStart Antibody™ both permit a rigorous "hot start" without the need for additional complicated steps.
"Hot start" A PCR technique called "hot start" was developed to prevent a first round of non-specific primer annealing and extension as the reaction is heated from 4°C up to the denaturing temperature. It is accomplished by adding an essential reaction component after the reaction has reached the denaturing temperature or by the use of a Taq enzyme modified to be heat activated. A common protocol involves aliquoting all reagents except for one into the PCR tube or well, melting a wax pellet over the reaction mix, cooling to form the wax barrier, adding the other reaction component on top of the wax barrier, and then beginning the thermocycling. The reaction components will only mix and begin the PCR when the wax has melted during the initial denaturing step.
Perkin Elmer's AmpliTaq Gold is activated by heating for 9 to 12 minutes at 92°C to 96°C. Its buffer conditions are markedly different from regular AmpliTaq (particularly pH) and it generally requires more MgCl2. The specific mechanism is proprietary (AmpliTaq Gold product insert). TaqStart Antibody is a monoclonal antibody which deactivates Taq polymerase at ambient temperature. The initial denaturing step in the PCR reverses the deactivation of the enzyme, allowing the reaction to proceed (14).
Robotics
When handling huge numbers of samples there is an obvious need to rely on robotics both to increase productivity and because the risk of human error is substantial.
PCR Format To this point, the standard PCR format for the high-throughput lab has been the 96-well tray. Now, however, several manufacturers are offering 384-well capability, either with a modified thermocycling block (MJR) or by using a 96-well tray with divided wells (Robbins Scientific). This format virtually mandates the use of robotics for DNA handling and PCR setup. Beckman's Biomek 2000 and Robbins Scientific's Hydra-96 Microdispenser both have this capability. The AmpliTaq Gold and TaqStart Antibody previously mentioned allow PCR reactions to be set up at room temperature without compromising target product purity.
Gel Loading Gel loading has typically been a manual procedure, utilizing a single pipette with flat tips or Hamilton's Multi-Channel Gel Loading Syringe. The move to include more lanes per gel (PE/ABD's format has gone from 24 lanes to 48/64 with 96 lanes in development) makes this manual step more onerous. The use of the multi-channel syringe requires some creativity in setting up the tray to accommodate the difference in spacing from the tray to the gel. Both PE/ABD and LI-COR are developing tools for multiple loading of wells. BIORAD's Geneloader II is capable of loading the PE/ABD 373's 24-cm well-to-read gels. This robot has been discontinued, but some labs have modified it to load shorter or longer gels and even retro-fitted it for use with the PE/ABD 377 DNA Sequencer.
Visualization
Multiplexing product The Genescan software available with the PE/ABD instruments visualizes peaks in a relatively narrow quantity range. The peak height scale is from 0-6000 arbitrary relative fluorescence units, but the actual peak height should be between 100 and 4000 to be readable. Underloaded samples can be overwhelmed by low background from other product of the same color. Overloaded sample will appear truncated and can fool the allele calling software into "seeing" extra alleles. Additionally, overloaded sample can interfere with genotypes of the same size range in other colors. Either requires remultiplexing and rerunning the entire gel, since attempting to identify problem lanes or samples to rerun is even more time consuming than reprocessing and is prone to human error.
An appropriate protocol for balancing multiplexed product can be very different from lab to lab depending on the type of experiments being done. If the same primers are being used for many multiples of trays of similar samples, balancing the reaction products quickly becomes a matter of routine. If primers and substrate are changing frequently, some percentage of gels will have to be sacrificed as test gels. We test 1 ml of two to four diluted samples from each PCR tray. This allows us to identify failed PCR's and to look for any non-specific product. We then calculate the average amount of product which will produce a homozygous peak of 1200 to 1500, typically 0.1 to 2.0 ml. Some labs find it effective to test product on agarose gels.
Sometimes, reduced cycle number or poor amplification results in the need to load more than 4 ml of total PCR product. This amount or more loaded on a gel will result in salt effects, such as anomalous size standard peaks and gel distortion. The product can be desalted with a variety of small spin columns (Amicon's Microcon, Qiagen's Quickspin) or by ethanol precipitation or with Millipore's new 96-well float dialysis trays.
Data management
Many labs are struggling at the interface between the huge amount of digitized images which can be produced by running efficient multiplexes and the methods used to analyze it. With the advent of 384-well PCR trays and 96-lane gels, a single technician with adequate robotic equipment could realistically produce 2880 genotypes per day, given a multiplex of 10 markers. This is, indeed, high-throughput. The challenge, then, is finding the management tools to keep up!
Current PE/ABD Genescan and Genotyper software must still be termed "semi-automated" for most applications. (Genescan still requires manual lane calling.) Some researchers have replaced PE/ABD software with their own (11, 25) and others have written software to interface with the PE/ABD software which addresses a specific need (4). A barcode systems is used at Sequana Therapeutics to track both sample and the PCR plate (13).
Cybergenetics has recently introduced their TrueAllele™ software and genotype service (www.cybergenetics-inc.com). This is offered both as licensed software and as a genotyping service which may be of particular interest to sequencing core laboratories needing to do a limited amount of genotyping. The software handles files from a variety of genotyping instruments, tracks lanes, sizes and scores alleles (25).
The latest generation of software from both PE/ABD (Genotyper 2.0, GenoPedigree, GenBase) and LI-COR (Gene ImagIR™ Genetic Analysis Software), offer a relational database with the ability to handle pedigree and marker information. Some combination of these may finally offer the researcher an analysis capability with speed comparable to the laboratory techniques.
Conclusion
The dual technologies of PCR and gel electrophoresis have provided researchers with a robust tool for genetic analysis. Improvements such as maximization of reactions per tray, lanes per gel and genotypes per lane have brought these techniques to an edge of efficiency unimagined in the days of DNA digests and Southern blots. Supporting robotics are maximizing technician effort. There will be increasing refinements&emdash;software improvements, capillary electrophoresis, more complete automation&emdash;as the technology hits its stride.
Recent developments in the form of DNA arrayed on microchips involve very different methods and techniques. The ability to read single base-pair changes expands the resource of markers to every polymorphic sequence in the genome and promises complete mutation detection. Automation is integral to the system rather than superimposed upon it. Microchip array technology could signal the next quantum leap in high-throughput genetic analysis.
References
Linda Wood Ballard may be contacted at the Genomics Core Facility, Huntsman Cancer Institute, University of Utah, 15 N. 2030 E. Room 2100, Salt Lake City, UT 84112-5330, Tel: (801) 581-3875, E-mail: linda.ballard@genetics.utah.edu
Return to the The ABRF Home Page