Research Report
`
`High-Throughput Single Nucleotide
`Polymorphism Genotyping by Fluorescent
`5¢¢ Exonuclease Assay
`
`BioTechniques 27:538-552 (September 1999)
`
`Phillip A. Morin, Robert Saiz
`and Atousa Monjazeb
`Axys Pharmaceuticals,
`La Jolla, CA, USA
`
`ABSTRACT
`
`polymorphisms
`nucleotide
`Single
`(SNPs) are the class of DNA variants that
`are most common in the genome. Technolo-
`gies are needed that allow relatively high-
`throughput, high-quality genotyping of
`SNPs, with rapid assay development for in-
`dividual SNPs of interest. We have imple-
`mented an accurate and high-throughput
`SNP genotyping system using a commer-
`cially available fluorescent 5¢ exonuclease
`assay. Optimization of the assay system for
`low-volume reactions and low-probe con-
`centrations has reduced assay costs by
`75%. Using a simple assay optimization
`process, we successfully developed genotyp-
`ing assays for 92% of assays attempted
`(309 out of 335) and have generated over
`200 000 genotypes.
`
`538 BioTechniques
`
`INTRODUCTION
`
`Single nucleotide polymorphisms
`(SNPs) are the next frontier in genome-
`wide polymorphism screening for gene
`mapping, positional cloning and associ-
`ation studies (1,5,16), and the rush to
`obtain and screen SNPs has spawned
`both private and public research into
`new technology to meet those goals
`(10–12). These whole-genome screen-
`ing methods, however, are orders of
`magnitude beyond the needs of many
`research programs aimed at identifying
`and/or screening dozens or hundreds of
`SNPs in target populations. In addition,
`these methods might not allow the flex-
`ibility to rapidly and inexpensively
`implement a survey of a novel poly-
`morphism or to screen all of the poly-
`morphisms in a gene region.
`For these reasons, technologies are
`needed that allow relatively high-
`throughput, high-quality genotyping of
`SNPs, with rapid assay development,
`for individual SNPs of interest. Current
`popular methods include restriction
`fragment-length polymorphism analysis
`of polymerase chain reaction products
`(RFLP-PCR) (with gel electrophoresis
`for fragment detection), allele-specific
`oligonucleotide (ASO) hybridization
`(3,17,18), oligonucleotide ligation as-
`say (OLA) (2,20), “minisequencing”
`(14) and the 5¢ -exonuclease assay, or
`TaqMan(cid:212)
`(4,6,7,9). Cost, number of
`steps, complexity of reactions, preci-
`sion, accuracy, speed of completion and
`potential for automation are all impor-
`tant considerations for deciding which
`system will be used in particular set-
`tings, and Syvänen and Landegren (19)
`reviewed these factors for ASO, OLA
`
`and minisequencing in the context of
`several laboratory settings.
`We have evaluated and/or used all of
`the above methods, and we adopted the
`5¢ exonuclease assay as our system of
`choice for high-throughput SNP screen-
`ing. This assay takes advantage of the 5¢
`exonuclease activity of Taq DNA poly-
`merase to cause cleavage of allele-spe-
`cific probes, which are hybridized to
`template DNA during PCRs. The
`probes are double-labeled with reporter
`(FAM or TET) and quencher (TAMRA)
`dyes at the 5¢ and 3¢ ends, respectively,
`and cleavage of the probes causes in-
`crease in the fluorescence of the re-
`porter dyes in solution. Differential hy-
`bridization of the two allele-specific
`probes results in cleavage of the probes
`only if they are perfectly annealed to
`their respective allele templates, result-
`ing in increase of fluorescence of one or
`both reporter fluors that can be plotted
`and segregated to determine the tem-
`plate genotype. The advantages of this
`assay over ASO and OLA include (i)
`single-tube amplification and detection,
`(ii) fluorescence signal proportional to
`PCR product amplification, (iii) lower
`reagent and labor costs per genotype,
`(iv)genotype assignment based on
`clearly separated clusters of fluorescent
`values and (v) simple, automatable as-
`say implementation procedures. We
`used the LS-50B fluorescent spectrom-
`eter with a plate reader and the ABI
`PRISM(cid:210) 7700 Sequence Detector (both
`from PE Biosystems, Foster City, CA,
`USA) for data acquisition and analysis
`(see Discussion).
`This manuscript outlines our meth-
`ods and modifications to the protocols
`for the 5¢ exonuclease assay system,
`
`Vol. 27, No. 3 (1999)
`
`1 OF 9
`
`THERMO FISHER EX. 1037
`
`

`
`Research Report
`
`current and potential throughput in our
`laboratory and automation systems for
`increasing efficiency. These modifica-
`tions and the supporting data were gen-
`erated over 3 years of assay develop-
`ment and genotyping, so different
`experiments were conducted with dif-
`ferent assays and sample subsets.
`
`MATERIALS AND METHODS
`
`Primer and Probe Design
`
`Oligonucleotide PCR primers and
`double dye-labeled probes were de-
`signed using Primer Express(cid:212)
`(Version
`1 for Macintosh(cid:210) Power PC; PE Biosys-
`tems), and assay design and conditions
`are primarily based on the allelic dis-
`crimination protocol from PE Biosys-
`tems (8). Default conditions in the pro-
`gram were used for reaction conditions,
`and the following parameters were set
`before selecting primers: primer con-
`centration = 250 nM; primer melting
`temperature (Tm) range = 58(cid:176) –60(cid:176) C;
`primer optimal Tm = 59(cid:176) C; maximum
`primer Tm difference = 2(cid:176) C; probe does
`not have a 5¢ ‘G’; probe Tm must be 7(cid:176)
`greater than primer Tm. Other parame-
`ters were changed if necessary to allow
`the program to select primers near the
`probe. The OLIGO 5.0 program (Na-
`tional Biosciences, Plymouth, MN,
`USA) has also been used effectively.
`Using the default conditions, probe se-
`quences were designed to have Tms of
`76(cid:176) C, and primers were designed for a
`Tm of 69(cid:176) C. Empirically, optimized an-
`nealing temperatures (Ta) for assays de-
`signed with either program were found
`to be approximately 60(cid:176) – 62(cid:176) C in the
`PCR conditions outlined below.
`Probes were selected in Primer Ex-
`press to obtain a theoretical Tm of 67(cid:176) –
`1(cid:176) C, with the polymorphic site as close
`to the center of the oligonucleotide as
`possible to provide the greatest differ-
`ence in Tm between perfectly matched
`probe and template and the mismatch.
`Slight changes in probe length were
`made to keep the Tm constant for the
`two allele-specific probes (e.g., a probe
`with an ‘A’ might be 1–3 nucleotides
`[nt] longer than the alternate allele
`probe containing a ‘G’ at the polymor-
`phic site). This design should allow per-
`fectly matched probes to hybridize
`
`540 BioTechniques
`
`strongly to the template DNA before the
`PCR primers anneal and to be cleaved
`rather than displaced by the Taq DNA
`polymerase during the extension phase.
`Mismatched probe/template pairs will
`be unstable at the PCR cycling condi-
`tions, and thus will not be cleaved.
`Once probe sequences were select-
`ed, all possible PCR primer pairs were
`determined automatically by Primer
`Express, and a pair was selected based
`on the optimal criteria of the program
`(as indicated by the Penalty score),
`product length (£ 150 bp) and by having
`the forward primer close (£ 30 bp if pos-
`sible) to the 5¢ end of the probe. The
`distance between the forward primer
`and the probe (which both anneal to the
`same strand of the template DNA) is
`important if they are further than ap-
`proximately 150 bp apart. We have
`found that signal intensity often falls off
`substantially if the forward primer is
`more than 100 bp from the probe bind-
`ing site (data not shown). The two alter-
`nate allele probes were synthesized
`with the reporter dyes FAM and TET,
`respectively, on the 5¢ nucleotides, and
`with TAMRA and a blocking phosphate
`on the 3¢ nucleotide, and then HPLC-
`purified. Primer oligonucleotides were
`unmodified and unpurified. For some
`AT-rich sequences, probe sequences
`can be over 40 nt. Such long probes
`tend to have higher background fluores-
`cence, but some work well regardless.
`All probes were synthesized by PE
`Biosystems or Integrated DNA Tech-
`nologies (IDT; Coralville, IA, USA).
`PCR primers were synthesized by Oper-
`on Technologies (Alameda, CA, USA).
`
`PCR Annealing Temperature
`Optimization
`
`A test for PCR product amplification
`was conducted for each primer pair to
`determine the optimal amplification
`temperature. Replicate PCRs were per-
`formed on a single DNA sample (20 ng
`of DNA dried into wells of 96-well PCR
`plates) at 58(cid:176) , 60(cid:176) and 62(cid:176) C using 200
`nM final primer concentration, in 1·
`TaqMan PCR Master Mix (PE Biosys-
`tems), without assay probes. The PCR
`products were electrophoresed through
`2% agarose gel in the presence of ethid-
`ium bromide and visualized by fluores-
`cence in UV light, and the Ta that pro-
`
`duced the highest product yield was se-
`lected for all future amplifications.
`
`Fluorescence Optimization
`
`Fluorescence signal intensity was
`optimized by testing an array of primer
`concentrations [forward:reverse (nM):
`300:50, 300:300, 300:900, 900:50,
`900:300 and 900:900] using standard
`assay conditions (below). This subset
`of concentration pairs for the two
`primers was derived empirically to re-
`duce the number of reactions that need
`to be run for each assay optimization.
`The PCR products, amplified in tripli-
`cate (with both probes present) for each
`set of concentrations, were analyzed in
`the ABI PRISM 7700 Sequence Detector
`for the change in the fluorescence of
`one or both of the reporter dyes relative
`to the quencher dye (D RQ), and the
`concentrations that produced the maxi-
`mum D RQ were selected for the assay.
`
`Assay Conditions
`Genomic DNA (5 m L of 4 ng/m L
`DNA in low EDTA TE [10 mM Tris,
`0.1 mM EDTA, pH 8.0]) for each sam-
`ple was aliquoted into 96-well PCR
`plates (Robbins Scientific [Sunnyvale,
`CA, USA] or Corning Costar [Cam-
`bridge, MA, USA]) dried at 80(cid:176) C for
`about 30 min and stored at ambient con-
`ditions until used. PCR cocktails includ-
`ing both probes (100 nM each) were set
`up using the derived primer concentra-
`tions and 1· TaqMan(cid:212)
`PCR Master
`Mix for the 7700, or Buffer II (100 mM
`Tris, pH 8.0, 500 mM KCl, 1% gelatin),
`with 7.5 mM magnesium chloride, the
`dNTPs (dA, G, C, U; 1:1:1:2 ratio), 0.05
`U/m L AmpliTaq Gold(cid:212)
`(PE Biosys-
`tems) and 0.01 U/m L Uraci-DNA-Gly-
`cosylase for the LS-50B. Five, ten or
`twenty microliters of the PCR cocktail
`were added to each well of the 96-well
`plate, and the plates were sealed with ei-
`ther caps or transparent plastic sealers
`(Seal-PLT-100; Elkay Products, Shrews-
`bury, MA, USA). PCR amplification
`was achieved under the following cy-
`cling conditions: 2 min at 50(cid:176) C, 10 min
`at 95(cid:176) C, then 40 cycles of 94(cid:176) C for 30 s,
`60(cid:176) C (or empirically derived optimal
`Tm) for 30 s. After cycling, samples re-
`mained at 4(cid:176) C until removal from the
`PCR thermal cyclers for analysis.
`
`Vol. 27, No. 3 (1999)
`
`2 OF 9
`
`THERMO FISHER EX. 1037
`
`

`
`Research Report
`
`542 BioTechniques
`
`Data Acquisition and Analysis
`
`Before we acquired the ABI PRISM
`7700 Sequence Detector, we used the
`LS-50B Fluorometer with plate reader
`for data acquisition and analysis on 256
`of the assays summarized in this report.
`Excel(cid:210) macros for data analysis were
`written by Jeff Marmaro (PE Biosys-
`tems) (9) and modified by P.A.M. This
`software uses a set of control genotypes
`as references to call alleles. The assays
`were done in 20-m L reaction volumes,
`diluted to 50 m L with deionized water
`and transferred to reflective white
`plates for fluorescence detection.
`For 79 assays, analyzed on the ABI
`PRISM 7700 Sequence Detector, plates
`were placed into the sequencer and ana-
`lyzed for post-PCR allelic discrimina-
`tion without normalization. The dye
`component data for each well are dis-
`played graphically and in a table, and
`alleles can be assigned directly from the
`graphical display using a selection func-
`tion in the software and a drag-down
`menu of genotype options, then export-
`ed for spreadsheet or database storage.
`
`RESULTS
`
`Number of Assays
`To date, we have designed 5¢ exonu-
`clease assays for over 330 SNPs, and
`we typically genotype between 300 and
`3000 individuals per assay. We use pub-
`licly available polymorphisms (Table 1)
`and novel SNPs derived from rese-
`quencing projects to design new assays
`following simple rules outlined by PE
`Biosystems (8).
`
`Assay Conditions
`
`Maximizing the fluorescence signal
`created by cleavage of the dual-labeled
`probes is critical, especially when using
`the ABI PRISM 7700, as it is less sensi-
`tive than the LS-50B Fluorometer and
`other fluorometers. Optimal tempera-
`ture and primer concentrations for the
`PCRs are determined using (i) multiple
`temperatures to find the highest tem-
`perature at which we obtain high PCR
`product yield (detected by agarose gel
`electrophoresis), and (ii) primer con-
`centration matrices for PCRs in the
`
`Vol. 27, No. 3 (1999)
`
`3 OF 9
`
`THERMO FISHER EX. 1037
`
`

`
`Research Report
`
`presence of the fluorescent probes to
`determine the primer concentrations
`that result in the greatest fluorescent
`signal of one or both of the probes from
`one DNA sample. Assays are designed
`to use a PCR primer Ta of approximate-
`ly 60(cid:176) C, and 95% (75 out of 79) of as-
`says optimized for the ABI PRISM 7700
`have optimal Tas between 58(cid:176)
`and
`62(cid:176) C. We have tried several primer
`concentration matrices, and find that
`96% (51 out of 53) of assays have opti-
`mal concentrations falling within a
`300· 900 nM matrix (forward:reverse
`primer concentration: 300:300, 15%;
`300:900, 25%; 900:300, 17%; and 900:
`900, 40%). The differences between as-
`says amplified with the empirically
`derived optimal PCR primer concen-
`trations and alternative concentrations
`are primarily in the signal intensity of
`one or the other fluorescent dye, which
`will tend to skew the heterozygote
`genotype cluster toward one or the oth-
`er homozygote cluster.
`
`PCR Cocktail Reagents
`
`The 7700 ABI PRISM is less sensi-
`tive than the LS-50B Fluorometer and
`uses different software for fluorescent
`dye spectral analysis, so signal intensi-
`ty is more important when using the
`Model 7700. Initial assays were per-
`formed using PE Biosystem’s Buffer II
`(see Materials and Methods) with 7.5
`mM magnesium chloride; however, the
`manufacturer now markets a propri-
`etary 2· buffer mixture that includes a
`passive dye (ROX) as an internal stan-
`dard, produces high signal intensity for
`most assays and includes the dNTPs,
`AmpliTaq Gold and Uracil-DNA-Gly-
`cosylase in the mixture. Thus, only
`probes, primers, DNA and water need
`to be added for each reaction. This is
`convenient and increases quality con-
`trol in assay preparation. We have fur-
`ther increased efficiency and consisten-
`cy among samples by diluting all
`samples to 4 ng/m L, aliquoting 20 ng of
`DNA into replicate PCR plates using a
`96-syringe pipettor (Robbins Scientif-
`ic) and drying the replicate plates. This
`allows us to add the PCR cocktail (also
`using the 96-syringe pipettor) to plates
`with the DNA already in place, pre-
`venting sample mix-up and reducing
`variation in pipettor volume.
`
`544 BioTechniques
`
`Reaction Volume
`
`To date, we performed about 200 of
`our assays in 20-m L volumes, instead of
`50 m L as suggested by PE Biosystems,
`as initial tests indicated no loss of geno-
`type quantities or quality (data not
`shown). To quantify the success rate us-
`ing 20-m L reactions, we surveyed all 46
`assays genotyped during the 7-month
`period just before we switched to 10-m L
`reactions. From those assays, there were
`53 165 genotypes generated from 639
`plates, with an overall scorable geno-
`type rate of 93.7% (standard deviation
`[SD]: 0.07%/plate or £ 6.6 genotypes
`depending on how many samples were
`on the plate). We have recently shown
`that 10- and 5-m L volumes can be used
`in the same 96-well plates without sig-
`nificant loss of fluorescence or allelic
`discrimination (Figure 1A). For 40 as-
`says performed with 5- or 10-m L vol-
`umes, we have achieved a mean of
`94.3% scorable genotypes per 96-well
`plate (339 plates genotyped). This fur-
`ther indicates that reaction volume
`down to 5 m L does not affect the scora-
`bility of the data, but we must caution
`that in some thermal cyclers, the small
`volume reactions fail more frequently.
`The PCRs for these assays have been
`performed in 96-well plates (from Rob-
`bins Scientific or Corning Costar) in a
`PTC-100(cid:212)
`Thermal Cycler (MJ Re-
`search, Watertown, MA, USA) and
`with GeneAmp(cid:210) PCR Systems 9600
`and 9700 with thermal lids (both from
`PE Biosystems). We evaluated the use
`of mineral oil overlays for small volume
`(5 m L) PCR, but found that it was not
`necessary for amplification, and it re-
`duced the fluorescent signal detected by
`the ABI PRISM 7700 (data not shown).
`
`Signal Strength and Allelic
`Discrimination
`
`PCRs were optimized (see Materials
`and Methods) to produce the maximum
`signal-to-noise ratio. Probe concentra-
`tion affects overall signal intensity and,
`to a much lesser extent, allelic discrimi-
`nation. The suggested probe concentra-
`tion is 200 nM, but we have found that
`little resolution is lost by using 100 nM
`of each probe, and twice as many reac-
`tions can be performed (Figure 1B).
`Where optimal conditions cannot be
`
`found using these initial steps, signal
`strength and allelic discrimination can
`sometimes be increased by increasing
`primer concentrations, addition of PCR
`cycles and/or longer annealing times
`(e.g., 60 s vs. 30 s) (data not shown).
`
`Assay Reproducibility and Accuracy
`
`To test the reproducibility of the as-
`says, we selected five assays and geno-
`typed four plates in duplicate using 10-
`m L volumes (on separate days with new
`PCR cocktail prepared each time). Out
`of 3760 genotypes total, there was one
`inconsistent genotype call. Forty-four
`genotypes (1.2%) were called “null” be-
`cause they either failed to amplify in one
`replicate or were not given a genotype
`designation because they were question-
`able (i.e., fell outside of the expected
`clusters of points). The overall success
`rate (percent accurately called geno-
`types) per assay ranged from 98.1% to
`99.6%. Figure 2 shows a single dupli-
`cate set of plates for each assay.
`To estimate the accuracy of the as-
`says, we examined the Mendelian inher-
`itance for 26 assays genotyped on a sin-
`gle 10-generational pedigree of 555
`individuals from the island of Tristan de
`Cunha, from which we have sampled
`265 individuals who were living on the
`island in 1997. This pedigree is extreme-
`ly well studied, so we are confident that
`there are no incorrect relationships. We
`found seven genotypes, (0.1%) out of
`6653, that were not compatible with
`Mendelian inheritance in this pedigree.
`
`Failed Assays
`
`Assay quality is dependent on ampli-
`fication of the product and appropriate
`annealing and cleavage of the probes in
`the PCR. Although primer picking pro-
`grams help to select primers and probes
`that have optimal base composition, Tas
`and lack of secondary structure, they are
`not perfect, especially given the high
`magnesium concentrations and other
`assay reagents that affect the PCR. We
`have found that in the majority of cases,
`our assays work with no additional opti-
`mization beyond the initial temperature
`and primer concentration matrices (see
`Materials and Methods). The most fre-
`quent reason for failure is lack of prod-
`uct amplification, which can be over-
`
`Vol. 27, No. 3 (1999)
`
`4 OF 9
`
`THERMO FISHER EX. 1037
`
`

`
`Research Report
`
`546 BioTechniques
`
`Vol. 27, No. 3 (1999)
`
`5 OF 9
`
`THERMO FISHER EX. 1037
`
`

`
`Research Report
`
`come by additional changes to the PCR
`conditions (e.g., lower annealing tem-
`perature, higher primer concentration,
`etc.) or by redesigning the PCR primers.
`In a few cases in which assays have not
`
`been resurrected by primer redesign, the
`most frequent putative cause (3 assays
`out of 4 analyzed) has been long probe
`sequences (>40 nt) designed in A/T-rich
`DNA regions.
`
`Figure 1. (A) Results from two assays completed on the same 94 DNA samples, using 5-, 10-, and 20-m L
`PCR volumes. Graph scales have been held constant between different volumes for each assay. The three
`clusters of points in each graph represent the allele 1 homozygotes (lower right), heterozygotes (middle),
`allele 2 homozygotes (upper left) and low or no-amplification samples and controls (lower left). Each
`plate contains one pooled DNA sample, which often falls between the cluster of heterozygote data points
`and the most common allele data points. (B) Typical results for probe concentrations of 100 and 200 nM
`(volume = 20 m L). The graph scales have been changed to show that relative distribution of points re-
`mains constant even though the overall fluorescent intensity changes with probe concentration.
`
`548 BioTechniques
`
`DISCUSSION
`
`The data quality of this assay sys-
`tem, along with the 1-tube assay that
`requires no post-PCR processing,
`makes the 5¢ exonuclease assay very at-
`tractive for high-throughput genotyp-
`ing. ASO hybridization systems with
`high-density arrays can produce higher
`throughput; however, signal and back-
`ground variation among samples and
`assays have caused a higher rate of
`genotype ambiguity and miscalls for us
`in the past. Our system of checking for
`Mendelian inheritance problems in ex-
`tended pedigrees has shown that we
`rarely have any incorrect genotypes
`from 5¢ exonuclease assays.
`We have addressed two of the prima-
`ry limitations of this system: (i) the high
`cost of reagents and (ii) the ability to au-
`tomate PCR setup and amplification
`through reduction of reaction volumes.
`We have designed over 340 systems for
`the 5¢ exonuclease assay and completed
`genotyping of 300–3000 DNA samples
`for most of them (>200000 genotypes).
`To date, using the methods described
`here, over 95% of the assays have been
`optimized, and they produced high-
`quality genotype data, most of which
`has been verified by checks for Men-
`delian inheritance in extended pedi-
`grees. Our primary modifications of PE
`Biosystems protocols for these assays
`have included drying DNA into plates,
`PCR product optimization and an ab-
`breviated primer concentration opti-
`mization step, and reduced volume
`PCRs. Our optimization procedure re-
`quires 33 PCRs (when done in tripli-
`cate) and nine agarose gel lanes per as-
`say. This is a reduction of 33 PCRs and
`gel lanes from the optimization protocol
`recommended by PE Biosystems for al-
`lelic discrimination (15), which requires
`64 PCRs (but does not suggest checking
`products on agarose gels).
`The cost per reaction is approxi-
`mately $0.42 for the 10-m L reaction
`volume (including all consumable ma-
`terials), assuming full use of the probes
`(about 10 000 reactions). The 2· PCR
`mixture from PE Biosystems costs
`$0.34/reaction, and the probe costs vary
`by manufacturer, but can generally be
`purchased for about $400 each or $800
`for an assay with enough probe to per-
`form at least 10 000 10-m L reactions.
`
`Vol. 27, No. 3 (1999)
`
`6 OF 9
`
`THERMO FISHER EX. 1037
`
`

`
`Research Report
`
`Figure 2. Results from five assays completed in duplicate, 10-mm L reaction volumes. Five 96-well plates were genotyped in duplicate for each assay for esti-
`mates of reproducibility, and one representative set of duplicate plates is shown for each assay. The Cyp1A1 SNP assay detected no homozygotes for the rare al-
`lele for the plate shown here. The two clusters are for the common homozygotes and the more rare heterozyotes.
`
`550 BioTechniques
`
`Vol. 27, No. 3 (1999)
`
`7 OF 9
`
`THERMO FISHER EX. 1037
`
`

`
`These costs can be decreased by using
`a homemade buffer mixture instead of
`the 2· Master Mix (PE Biosystems) or
`by developing even smaller volume
`PCR amplification and fluorescence
`detection methods.
`The primary bottlenecks in this pro-
`duction line are primer and probe syn-
`thesis (typically 3 days for primers, 5–7
`days for probes), optimization steps and
`availability of thermal cyclers needed to
`amplify 5000–10 000 samples.
`The time needed for primer and
`probe synthesis has been decreased by
`changes in probe synthesis techniques
`that have eliminated post-synthesis
`HPLC purification, addition of TAM-
`RA to each probe and a second HPLC
`purification (13).
`Optimization steps have been imple-
`mented to maximize the chances that an
`assay will work reliably when applied
`across hundreds or thousands of sam-
`ples. The possible permutations in each
`
`optimization, however, have been grad-
`ually reduced after optimization of
`dozens of assays to minimize the
`reagents, labor and time involved. We
`originally tested five annealing temper-
`atures for each assay (56(cid:176) , 58(cid:176) , 60(cid:176) , 62(cid:176)
`and 64(cid:176) C), a matrix of nine probe
`concentrations (100, 500 and 900 nM),
`and a matrix of nine primer concentra-
`tions (50, 300 and 900 nM) for assay
`optimization. We have empirically de-
`rived a subset of optimization steps and
`dropped the probe concentration matrix
`altogether, for a smaller set of steps
`which yields robust results for 95% of
`our assays (see Materials and Methods).
`Volume reduction presents several
`cost-saving and labor-saving benefits.
`First, the PCR components (2· Master
`Mix), including enzymes and passive
`dye standard (ROX), which make up ap-
`proximately 80% of the disposable ma-
`terials costs (excluding probes) with 20-
`m L reactions, are reduced to 66% of
`
`disposable materials costs with 10-m L
`reactions. Second, probe synthesis is a
`fixed cost that is about $0.16/reaction
`when £ 5000 genotypes are processed at
`the 20-m L volume, but reducing volume
`to 10 m L obviously allows double the
`number of genotypes for the same cost
`—a 50% reduction in the per genotype
`cost. Third, mixing and aliquoting of
`smaller volumes is more easily automat-
`ed for robotic workstations, thereby re-
`ducing labor and potential for human er-
`ror. Finally, small volumes can be cycled
`more efficiently in the thermal cyclers,
`so more plates can be cycled through
`each machine in a day, and automatic
`transfer of plates into and out of a PCR
`thermal cycler is ultimately achievable
`for amplifying large numbers of plates
`without the need for a person to feed the
`plates into multiple PCR machines.
`Hardware and software for analysis
`of 5¢ exonuclease assay results are cur-
`rently available with the ABI PRISM
`
`8 OF 9
`
`THERMO FISHER EX. 1037
`
`

`
`Research Report
`
`7700 Sequence Detector. This system
`provides software for rapid analysis of
`the data. An alternative to this special-
`ized and expensive instruments is a
`simple fluorometer with the ability to
`detect multiple fluorescent emission
`wavelengths in the 500–600-nm range.
`Excel macros, developed by PE
`Biosystems (9) and modified by us, can
`be used to organize the fluorescent
`emission data, to separate alleles based
`on controls by multicomponent analy-
`sis of the wavelengths and to automate
`genotype calls. This system (using the
`LS-50B Fluorometer with plate reader)
`has worked well for us; however, it re-
`quires transfer of samples from PCR
`plates to reading plates, assumptions
`and cut-offs for allele calling and man-
`ual checking of genotypes to be sure
`that the allele calling was accurate. As
`most fluorometers are more sensitive
`than the ABI PRISM instruments, assays
`developed for fluorometers do not need
`
`to be optimized as much as the assays
`we have described here. The benefit of
`the fluorometer-based system is that it
`can make use of equipment and soft-
`ware presently available in many labo-
`ratories, but at the expense of precise
`and user-friendly hardware and soft-
`ware that allow high throughput in
`analysis and quality control.
`The 5¢ exonuclease assay system in-
`corporates (i) ease of assay design, (ii)
`commercial availability of oligonu-
`cleotide probes, instruments and analy-
`sis software and (iii) scalability for labo-
`ratories producing <100 genotypes to
`those producing ten-thousand genotypes
`per week. Although this system is un-
`likely to meet the needs of genome-wide
`SNP screening as proposed by several
`authors (1,5,15), for population-level
`genotyping of specific polymorphisms,
`candidate genes and disease genes, this
`system offers ease of use, throughput
`and accuracy not previously available.
`
`ACKNOWLEDGMENTS
`
`We thank Shelley Webster at Axys
`and Gianfranco DeFeo and Chris Grim-
`ley of PE Biosystems for technical
`help. We also thank Chris Hinckel for
`discussion and advice and Andy Wat-
`son, Penny Isabella and several anony-
`mous reviewers for useful suggestions.
`
`REFERENCES
`
`1.Collins, F.S., M.S. Guyer and A.
`Chakravarti. 1997. Variations on a theme:
`cataloging human DNA sequence variation.
`Science 278:1580-1581.
`2.Delahunty, C., W. Ankener, Q. Deng, J. Eng
`and D.A. Nickerson. 1996. Testing the feasi-
`bility of DNA typing for human identification
`by PCR and an oligonucleotide ligation assay.
`Am. J. Hum. Genet. 58:1239-1246.
`3.Handelin, B. and A.P. Shuber. 1996. Simul-
`taneous detection of multiple point mutations
`using allele-specific oligonucleotides, p.
`9.4.1-9.4.5. In N.C. Dracopoli, J.L. Haines,
`B.R. Korf, D.T. Moir, C.C. Morton, C.E. Seid-
`man, J.G. Seidman and D.R. Smith (Eds.),
`Current Protocols in Human Genetics. John
`Wiley & Sons, New York.
`4.Holland, P.M., R.D. Abramson, R. Watson
`and D.H. Gelfand. 1991. Detection of specif-
`ic polymerase chain reaction product by utiliz-
`ing the 5¢ to 3¢ exonuclease activity of Ther-
`mus aquaticus DNA polymerase. Proc. Natl.
`Acad. Sci. USA 88:7276-7280.
`5.Lander, E.S. 1996. The new genomics: global
`views of biology. Science 274:536-539.
`
`6.Lee, L.G., C.R. Connell and W. Bloch. 1993.
`Allelic discrimination by nick-translation
`PCR with fluorogenic probes. Nucleic Acids
`Res. 21:3761-3766.
`7.Livak, K., S.J.A. Flood, J. Marmaro, W.
`Giusti and K. Deetz. 1995. Oligonucleotides
`with fluorescent dyes at opposite ends provide
`a quenched probe system useful for detecting
`PCR product and nucleic acid hybridization.
`PCR Methods Appl. 4:357-362.
`8.Livak, K., J. Marmaro and S. Flood. 1995.
`Guidelines for designing TaqMan(cid:212)
`fluoro-
`genic probes for 5¢ nuclease assay. PE Applied
`Biosystems, Foster City, CA.
`9.Livak, K.J., J. Marmaro and J.A. Todd.
`1995. Towards fully automated genome-wide
`polymorphism screening. Nat. Genet. 9:341-
`342.
`10.Marshall, E. 1997. ‘Playing chicken’ over
`gene markers. Science 278:2046-2048.
`11.Marshall, E. 1997. Snipping away at genome
`patenting. Science 277:1752-1753.
`12.Marshall, E. 1997. The hunting of the SNP.
`Science 278:2047.
`13.Mullah, B., K. Livak, A. Andrus and P.
`Kenney. 1998. Efficient synthesis of double
`dye-labeled oligodeoxyribonucleotide probes
`and their application in a real time PCR assay.
`Nucleic Acids Res. 26:1026-1031.
`14.Pastinen, T., A. Kurg, A. Metspalu, L. Pel-
`tonen and A.-C. Syvänen. 1997. Minise-
`quencing: a specific tool for DNA analysis
`and diagnostics on oligonucleotide arrays.
`Genome Res. 7:606-614.
`15.PE Biosystems. TaqMan(cid:210) Allelic Discrimi-
`nation Demonstration Kit Protocol. Part No.
`4303267. Foster City, CA.
`16.Risch, N. and K. Merikangas. 1996. The fu-
`ture of genetic studies of complex human dis-
`eases. Science 273:1516-1517.
`17.Shuber, A.P., L.A. Michalowsky, G.S. Nass,
`J. Skoletsky, L.M. Hire, S.K. Kotsopoulos,
`M.F. Phipps, D.M. Barberio and K.W.
`Klinger. 1997. High throughput parallel
`analysis of hundreds of patient samples for
`more than 100 mutations in multiple disease
`genes. Hum. Mol. Genet. 6:337-347.
`18.Shuber, A.P., J. Skoletsky, R. Stern and
`B.L. Handelin. 1993. Efficient 12-mutation
`testing in the CFTR gene: a general model for
`complex mutation analysis. Hum. Mol. Genet.
`2:153-158.
`19.Syvänen, A.-C. and U. Landegren. 1994.
`Detection of point mutations by solid-phase
`methods. Hum. Mutat. 3:172-179.
`20.Tobe, V.O., S.L. Taylor and D.A. Nickerson.
`1996. Single-well genotyping of diallelic se-
`quence variations by a two-color ELISA-
`based oligonucleotide ligation assay. Nucleic
`Acids Res. 24:3728-3732.
`
`Received 24 March 1999; accepted 21
`May 1999.
`
`Address correspondence to:
`Dr. Phillip A. Morin
`Axys Pharmaceuticals, Inc.
`11099 North Torrey Pines Road, Suite 160
`La Jolla, CA 92037, USA
`Internet: morin@axyspharm.com
`
`Vol. 27, No. 3 (1999)
`
`9 OF 9
`
`THERMO FISHER EX. 1037