HUMAN MUTATION 17:305�316 (2001)

© 2001 WILEY-LISS, INC.

METHODS

Fluorescent Microsphere-Based Readout Technologyfor Multiplexed Human Single NucleotidePolymorphism Analysis and Bacterial IdentificationFei Ye,1* May-Sung Li,1 J. David Taylor,1 Quan Nguyen,2 Heidi M. Colton,3 Warren M. Casey,3

Michael Wagner,2 Michael P. Weiner,1 and Jingwen Chen1

1Department of Genomic Sciences, Glaxo Wellcome Research and Development, Research Triangle Park, North Carolina2Department of Human Genetics, Glaxo Wellcome Research and Development, Research Triangle Park, North Carolina3Department of Analytical Sciences, Glaxo Wellcome Research and Development, Research Triangle Park, North Carolina

For the SNP 2000 Special Issue

Large-scale human genotyping requires technologies with a minimal number of steps, high accu-racy, and the ability to automate at a reasonable cost. In this regard, we have developed a rapid,cost-effective readout method for single nucleotide polymorphism (SNP) genotyping that com-bines an easily automatable single-tube allele-specific primer extension (ASPE) with an efficienthigh throughput flow cytometric analysis performed on a Luminex 100™ flow cytometer. Thisrobust technique employs an ASPE reaction using PCR-derived target DNA containing the SNPand a pair of synthetic complementary capture probes that differ at their 3¢ end-nucleotide defin-ing the alleles. Each capture probe has been synthesized to contain a unique 25-nucleotide identi-fying sequence (ZipCode) at its 5¢ end. An array of fluorescent microspheres, covalently coupledwith complementary ZipCode sequences (cZipCodes), was hybridized to biotin-labeled ASPE re-action products, sequestering them for flow cytometric analysis. ASPE offers both an advantage ofstreamlining the SNP analysis protocol and an ability to perform multiplex SNP analysis on anymixture of allelic variants. All steps of the assay are simple additions of the solutions, incubations,and washes. This technique was used to assay 15 multiplexed SNPs on human chromosome 12from 96 patients. Comparison of the microsphere-based ASPE assay results to gel-based oligo-nucleotide ligation assay (OLA) results showed 99.2% agreement in genotype assignments. Inaddition, the microsphere-based multiplex SNPs assay system was adapted for the identification ofbacterial samples by both ASPE and single base chain extension (SBCE) assays. A series of probesdesigned for different variable sites of bacterial 16S rDNA permitted multiplex analysis and gener-ated species- or genus-specific patterns. Seventeen bacterial species representing a broad range ofgram-negative and gram-positive bacteria were analyzed within 16 variable sites of 16S rDNAsequence. The results were consistent with the published sequences and confirmed by direct DNAsequencing. Hum Mutat 17:305–316, 2001. © 2001 Wiley-Liss, Inc.

KEY WORDS: SNP; allele-specific primer extension; ASPE; single base chain extension; SBCE; micro-spheres; multiplex; flow cytometry; bacterial identification; mutation detection; ZipCode

Received 18 October 2000; accepted revised manuscript 8January 2001.

*Correspondence to: Fei Ye, Department of Genomic Sciences,GlaxoWellcome Research and Development, 5 Moore Drive, Re-search Triangle, NC 27709-3398.

INTRODUCTION

As the DNA sequence of the human genomeis completely elucidated, much attention is be-ing focused on single nucleotide polymorphisms(SNPs), the most abundant form of genetic varia-tion. According to some estimates, the human

306 YE ET AL.

genome may contain >3 million SNPs [Cooperet al., 1985]. Due to their frequency and distri-bution, SNPs are becoming superior geneticmarkers for assembly of a high-resolution map,aiding identification of disease-related loci [Laiet al., 1998]. In addition, SNPs can potentiallybe used for medical diagnostics. The powerful,target-specific pharmaceuticals being developedtoday can bring profound improvements to thelives of many patients, but may have serious sideeffects in certain sub-populations. The geneticvariation behind these differing biological re-sponses may correlate with a small set of SNPsthat could serve as a diagnostic tool to insureprescription of the right medicine to the rightpatient. These facts increase the need for highthroughput genotyping technologies with simplesteps and the ability to automate. Recently, anumber of methods for SNP detection have beendeveloped, including restriction fragment lengthpolymorphism (RFLP) analysis, single-strandconformation polymorphism analysis (SSCP)[Orita et al., 1989], allele-specific oligonucle-otide hybridization (ASO) [Saiki et al., 1989],oligonucleotide ligation assay (OLA) [Landegrenet al., 1988], primer extension assay [Syvanen,1999; Pastinen et al., 1997], Taqman [Livak etal., 1995], molecular beacons [Tyagi et al., 1998],and structure-specific flp nuclease technology[Mein et al., 2000]. A variety of platforms havebeen used to analyze reaction products includ-ing gel electrophoresis, fluorescence polarization[Chen et al., 1999], semiconductor chips [Gilleset al., 1999], high-density oligonucleotide arrays[Fan et al., 2000; Pastinen et al., 2000], and massspectrometry [Fu et al., 1998; Ross at al., 1998].We have adapted the use of fluorescent micro-spheres in flow cytometric analysis [McHugh,1994; Fulton et al., 1997; McDade and Fulton,1997; Kettman et al., 1998] for SNP determina-tion. Previously we demonstrated the proof-of-concept for this SNP-detection platform usingFACS Calibur instrumentation for analysis ofOLA [Iannone et al., 2000] and single base chainextension [Chen et al., 2000] assays. A similarapproach has also been reported recently [Caiet al., 2000]. Here we describe a new SNPs as-say that combines a DNA polymerase reactionnamed allele-specific primer extension (ASPE)

with a microsphere-based detection using a lessexpensive flow cytometer with a 96-well platereader. The assay relies on the sequence-specificprimer extension of two allele-specific captureoligonucleotide probes that differ at their 3′-endnucleotide defining the alleles. A DNA sequence(termed ZipCode) at the 5′-end portion of thecapture probe allows the resulting enzymaticreaction product to be captured by its comple-mentary sequence (cZipCode), which has beencoupled to a specific fluorescent microsphere.The ASPE assay permits multiplexed queryingof different nucleotides, thereby allowing multi-plexing of both alleles of a particular SNP. In thisstudy, we demonstrate that this new readoutsystem is a simple and reliable method that canbe used for high throughput SNP genotyping.

To explore the applications of our micro-sphere-based multiplexed SNP assay, we havealso adapted this assay system for rapid bacterialidentification using 16S rDNA sequence. Theassay uses primers that are placed 5′ upstream ofvariable bases in the 16S rDNA sequence. Theseprimers are coupled with uniquely identifyingsequences termed ZipCodes that are comple-mentary to sequences (cZipCodes) covalentlyattached to fluorescent microspheres. In ourstudy, ASPE or single base chain extension(SBCE) was used to extend the primers with bi-otin-labeled dCTP or ddNTPs, respectively. Thereaction products were hybridized to thecZipCode-microsphere complex and analyzed byflow cytometry. The flow cytometer identifiedthe fluorescent microsphere and measured thepresence of the biotin-labeled dCTP or ddNTPafter tagging with streptavidin-phycoerythrin(SA-PE). The dCTP or ddNTP molecules addedto the specific primers form a pattern which wasanalyzed to determine the bacterial identifica-tion. Using this multiplexed assay, seventeenspecies were divided into seventeen groups basedon their ASPE or SBCE reaction patterns.

MATERIALS AND METHODS

Reagents

Shrimp alkaline phosphatase (SAP) and E.coli Exonuclease I (Exo I) were obtained fromAmersham Pharmacia (Cleveland, OH). Biotin-labeled ddNTPs and biotin-labeled dCTP were

FLUORESCENT MICROSPHERE FOR SNP ANALYSIS 307

obtained from NEN Life Science Products, Inc.(Boston, MA). AmpliTaq, AmpliTaq Gold Taq,and AmpliTaq FS DNA polymerases were pur-chased from Applied Biosystems (Foster City,CA). Taq FS DNA polymerase was obtainedfrom Applied Biosystems as a special order. Plati-num GenoTYPE Tsp DNA polymerase was pur-chased from Gibco/BRL (Rockville, MD).Streptavidin-phycoerythrin (SA-PE) was ob-tained from Molecular Probes (Eugene, OR).Oligonucleotides with 5′ amino groups wereordered from Applied Biosystems. 2-[N-Mor-pholino]ethanesulfonic acid (MES) and 1-Ethyl-3-(3-Dimethylaminopropyl)carbodiimide Hy-drochloride (EDC) were purchased from Sigma(St. Louis, IL) and Pierce (Rockford, IL), respec-tively. Carboxylated fluorescent polystyrenemicrospheres were purchased from LuminexCorporation (Austin, TX).

PCR Amplification

PCR reactions were performed in a 96-wellmicrotiter-plate on a PTC-100 thermal cycler(MJ Research, Waltham, MA). 15 µl of reac-tion mixture contained 20 ng genomic DNA,10 mM Tris-HCl (pH 8.3), 50 mM KCl, 1.5 mMMgCl2, 100 µM dNTPs, 0.2 µM of each primer,and 1.5 units of AmpliTaq Gold DNA poly-merase. The reactions included a 10 min incuba-tion at 95°C, followed by 40 cycles at 94°C for 30sec, 60°C for 30 sec, and 72°C for 30 sec. Aftercycling, the reactions were incubated at 72°C fora final extension of 5 min. To clean up the PCRreaction for SBCE reaction, one unit of SAP andtwo units of Exo I were added to each 10 µl of thepooled PCR products. The mixture was incubatedat 37°C for 30 min, followed by 15 min at 80°Cto inactivate the enzymes.

Coupling of Oligonucleotides to Microspheres

Oligonucleotides used for coupling to micro-spheres were designed according to our previouspublications [Chen et al., 2000; Iannone et al.,2000]. All of these oligonucleotides contain fourelements: i) a 5′ amino group for covalent attach-ment to the carboxylated microsphere surface, ii)an 18-atom spacer (CH3CH2O)6 to minimize po-tential interaction between the oligonucleotidesequence and the microsphere surface, iii) a 10-

base LUCtag sequence (CAGGCCAAGT) tomonitor the coupling efficiency of the oligonucle-otides to the microspheres, and iv) one of a set of25-base complementary ZipCode sequences(cZipCodes). This set of cZipCodes was selectedfrom the Mycobacterium tuberculosis genome andwas checked empirically for absence of cross-hy-bridization between members of the set [Iannoneet al., 2000]. For the coupling of oligonucleotidesto microspheres, 5 x 106 carboxylated micro-spheres in 50 µl of 0.1 M MES buffer were mixedwith 2 nmoles (2 µl of a 1 mM solution) amino-modified oligonucleotide. 10 µl of freshly madeEDC (30 mg/ml in water) was added to themicrosphere/oligo mixture and incubated atroom temperature for 30 min. One additionalfresh 10 µl aliquot of EDC was added and incu-bated for 30 min with occasional sonication. Themicrospheres were then washed with 1 ml of0.1% sodium dodecylsulfate, followed by wash-ing with 1 ml of 0.02% Tween 20, and finallyresuspended in 500 µl TE pH 8.0 [10 mMTris[hydroxymethyl]aminomethane hydrochlo-ride / 1 mM Ethylenediamine-tetraacetic acid]and stored in the dark at 4°C. Coupling efficiencywas assessed by hybridizing coupled microsphereswith a molar excess of biotinylated oligonucle-otide that is complementary to the LUCtag se-quence. The standard procedure was the sameas that detailed below for hybridization of reac-tion products to the microspheres. Effective cou-pling reactions produced microspheres with amean fluorescent intensity (MFI) of 2000 to 4000units. Microspheres with MFI less than 1000were replaced.

Preparation of Bacterial 16S rDNA

Bacterial strains used in this study were ob-tained from the American Type Culture Collec-tion (ATCC). Prior to extraction, each isolate wasstreaked on trypticase soy agar and examined forproper colony morphology. The identities wereverified using the VITEK identification system.Bacterial DNA was isolated using the PrepMan™system (PE Applied Biosystems, Foster City, CA).For 16S rDNA amplification, two separate sets ofhighly conserved primers were utilized for theamplification from different bacterial species asfollows: 27f/1525r, 5′-AGAGTTTGATCMTG-

308 YE ET AL.

GCTCAG-3′/ 5′-AAGGAGGTGWTCCAR-CC-3′ and 66f/1392r, 5′-CAGGCCTAA-CACATGCAAGTC-3′/5′-GGGCGG(t/a)-GTGTACAAGGC-3′ [Marchesi et al., 1998].DNA was amplified in 50 µL reaction mixturescontaining 10 µl of a 1:250 dilution of templateDNA extraction, and PCR buffer containing 100mM Tris-HCl (pH 8.3), 50 mM KCl, and 1.5mM MgCl2. The reaction also contained 200 µMdNTPs, two units of AmpliTaq Gold, and 0.4µM of each primer. The PCR reactions includeda 10-min incubation at 94°C, followed by 30cycles of 94°C for 30 sec, 55°C for 45 sec, and72°C for 90 sec. After cycling, the reactions wereincubated at 72°C for a final extension of 5 min.The DNA fragments amplified using the 16SrDNA universal primers were sequenced usingstandard dideoxynucleotide termination sequenc-ing. The generated 16S rDNA sequences andsome other reference sequences obtained fromGenBank were analyzed with the CLUSTAL Wprogram.

ASPE Reactions

For each SNP (or variable site of 16S rDNAsequence for bacterial identification), a pair ofprobes was designed such that the last of the 3′-end base differed at the polymorphic site (seeTable 1 for bacterial ID). For the ASPE assay,10 µl of pooled, untreated PCR products (10–20 ng of each amplicon) were added to 10 µl of2X ASPE reaction mix containing 40 mM Tris-HCl (pH 8.4), 100 mM KCl, 2.5 mM MgCl2, 25nM positive control capture oligonucleotide, 50nM of each SNP capture oligonucleotide, 1.5units of Tsp DNA polymerase, and 5 µM biotin-dCTP. An initial denaturing step of 2 min at 96°Cwas used, followed by 30 cycles at 94°C for 30sec, 55°C for 1 min and 74°C for 2 min. Reac-tions were held at 4°C prior to the addition ofmicrospheres.

SBCE Reactions

For the SBCE assay, capture probes were de-signed in such a way that their sequence termi-nated one base upstream from the SNP site (forbacterial identification, the probes were chosenfrom conserved regions of the 16S rDNA se-quence and ended one base 5′ from a variable

nucleotide, see Table 1). Two nearly identicalreactions were set up, differing only in the choiceof labeled ddNTP (four reactions for bacterialidentification). SAP/ExoI-treated PCR products(10–20 ng of each amplicon) were assayed in 20µl of reaction volumes containing 80 mM of Tris-HCl (pH 9.0), 2 mM of MgCl2, 12.5 nM of posi-tive control target oligonucleotide, 12.5 nM ofpositive control capture oligonucleotide, 25 nMof each SNP capture oligonucleotide, 2.4 unitsof AmpliTaq FS, 1 µM of the allele-specific la-beled ddNTP, and 1 µM each of the other threeddNTPs. The reactions were incubated at 96°Cfor 2 min and then cycled 30 times at 94°C,55°C, and 72°C for 30 sec at each temperature.Reactions were held at 4°C prior to the addi-tion of microspheres.

Hybridization of Enzymatic Reaction Products

to the Microspheres

To capture each of the labeling reaction prod-ucts via the hybridization between the ZipCodesat the 5′-ends of the probes and the complemen-tary ZipCodes (cZipCodes) coupled to themicrospheres, 1,000 microspheres of each typewere added to each reaction. The concentrationsof NaCl and EDTA were adjusted to 500 mMand 13 mM, respectively. The mixtures were

TABLE 1. Oligonucleotide Sequences of the 3′ EndPortion of 16S rDNA Capture Probes for

Bacterial Identification

Probe Targetname Probe sequencea siteb

p1 CTCCTACGGGAGGCAGCAGT(a/g) 338-357p2 ATGTTGGGTTAAGTCCCG(c/t) 1041-1058p3 GGAATCGCTAGTAATCG(c/t) 1296-1312p4 TGTCGTCAGCTCGTGT(c/t) 1019-1034p5 GATGAGTGCTAAGTGTTAG(a/g) 817-835p6 TCTCAGTTCGGATTGTAG(g/-) 1254-1271p7 GGTCATTGGAAACTGG(g/a) 630-645p8 ACTTTCAGCGGGGAGGAAGG(g/t) 434-453p9 AGGGTTGCCAAGCCGCGAGG(g/t) 1211-1230p10 ACTTATAGATGGATCCGCGC(c/t) 219-238p11 ATTGGTGCCTTCGGGAACTC(a/-) 979-998p12 GAACAAATGTGTAAGTAACT(a/g) 449-468p13 TACCGGATAACATTTTGAAC(c/-) 172-191p14 GAGTGCTCGAAAGAGAACCG(a/-) 981-1000p15 GTAACAGGAAGAAGCTTGCT(g/-) 70-89p16 GAAACTGGCTTGCTTGAGTCT(t/-) 638-658aThe variable bases at the 3′ end of the sequence are shownin lower case.bThe positions of 16S rDNA sequences are based on the E.coli sequence as a reference (GenBank accession #JZ83205).

FLUORESCENT MICROSPHERE FOR SNP ANALYSIS 309

incubated at 96°C for 2 min and 40°C for atleast 1 hr. The microspheres were washed withwashing buffer (150 mM NaCl, 15 mM sodiumcitrate, 0.02% Tween 20) and pelleted at 1,300x g for 5 min. A 96-well pipettor, such as theRobbins Hydra96™, was used for the washingprocess. Biotin labels were developed in 8 µg/mlSA-PE (in washing buffer) at room temperaturefor 30 min in the dark.

Flow Cytometric Analysis

Microsphere fluorescence was measured usinga Luminex 100 cytometer (Luminex Corp, Aus-tin, TX) and associated software. Each micro-sphere type in the 100-microsphere set wasidentified by its characteristic fluorescence of redand infrared wavelengths. The orange fluores-cence associated with the SBCE or ASPE bio-logical reaction on the surface of the microsphereswas collected and converted to the mean fluo-rescence intensity (MFI) value. Since each micro-sphere type also emitted a small amount offluorescence in the orange wavelengths of thereporter channel, the inherent, analyte-indepen-dent orange fluorescence contributed by eachmicrosphere alone was subtracted from the MFIvalue of each sample on the correspondingmicrosphere using a Microsoft Excel spreadsheet.MFI values from two corresponding alleles weremerged, allowing display of the results on a two-coordinate system in which allelic calls weremade using Spotfire software. A minimum of 30microspheres was analyzed per data point.

RESULTS

Microsphere-Based Multiplex SNPs Analysis

Using Allele-Specific Primer Extension

In order to explore the possibility of setting upthe multiplex primer extension in one reactionwell (tube), instead of G, A, T, and C reactionsseparately as in the SBCE assay, we have usedthe ASPE detection strategy in our microsphere-based assay system (Fig. 1). The key differencebetween ASPE and SBCE is the capture probedesign. For ASPE reactions, the last base of thetwo allelic probes is coincident with the polymor-phic site, while the single SBCE probe ends onebase upstream. The advantage hereby gained isthat for each SNP genotype, only one ASPE re-

action is required if the two probes are designedwith different ZipCode sequences, while theSBCE assay requires two reactions.

In our study, the ASPE assay was validatedby genotyping 15 SNPs on human chromosome12 using PCR-amplified target DNA from 96DNA samples. These DNA samples and SNPshad previously been genotyped by gel-basedOLA assays. A set of 30 unique cZipCode se-quences, validated empirically for non-cross-re-activity, was coupled to 30 different microspheres(of 100 possible) for capturing each of the 15SNPs (one pair of capture probes for each SNP).PCR products were pooled together in one wellfor each DNA sample. After the ASPE reactionswere performed with the pooled PCR amplicons,the enzymatic reaction products were capturedonto a set of microspheres for flow cytometricanalysis as described in Material and Methods.Representative ASPE assay results for four SNPsassayed across 96 samples are shown in Figure 2.The ASPE experiment generated signals vary-ing from a low of 400 MFI to a high of 1600MFI. Assay signal intensity was similar to SBCE-generated signal strength. Fourteen of the 15 SNPswere successfully converted to this assay format.The concordance rate of 1344 genotypes obtainedby ASPE and gel-based OLA was 99.2%.

Bacterial Identification by Microsphere-Based

Multiplex SBCE and ASPE Assays Using

16S rDNA

For the SBCE assay, 16 synthetic captureprobes were designed. The 5′ portion of eachprobe contained a unique ZipCode sequence andthe 3′ portion was identical to the conserved re-gions of 16S rDNA sequence, ending one baseupstream of the 16 variable sites (Table 1). Amongthem, probes 10–16 are more specific for a cer-tain genus or species. For each reaction of themultiplex SBCE assay, a pooled mixture of 16capture probes was added to the amplified 16SrDNA product. The probes were extended andlabeled by one base in the presence of the TaqDNA polymerase and biotin-labeled ddNTP, infour separate reactions (ddA, ddG, ddC, andddT). The probes were then captured by hybrid-izing to the complementary sequences (cZip-Codes) attached to the florescent microspheres.

310 YE ET AL.

FIGURE 1.

FLUORESCENT MICROSPHERE FOR SNP ANALYSIS 311

Figure 3 shows the results of the nine variablesites in four bacterial strains. The intensity ofeach of four bases for each 16S rDNA sample islabeled as G, A, T, and C sequentially. In mostcases, the positive signal to background noiseratio was greater than 10. Since each fluores-cent microsphere becomes the address for asingle 16S rDNA variable site, combining these16 probes for a multiplex assay creates a uniquepattern for each given bacteria species or genus.This pattern divides the gram-positive and gram-negative bacteria into smaller groups, and fur-ther into genus and species (Fig. 4).

For the ASPE assay, a pair of probes for eachof the sixteen 16S rDNA variable sites was de-signed such that their last base differs from eachother. In the presence of DNA polymerase,dNTPs, and a small amount of biotin-labeleddCTP, a labeled extension product of the 3′ por-tion of the probe was obtained. This occurs ifthe template included the target sequence. Foreach template sample, only a single two-probereaction was needed. The subsequent hybridiza-

tion to microspheres and fluorescent dye devel-opment procedures were the same as in theSBCE assay. The ASPE results agreed with re-sults from the SBCE assay (Fig. 3), and wereconsistent with direct sequencing data in allcases. Using these multiplexed assays, 17 bacte-rial species were divided into 17 groups basedon their ASPE or SBCE patterns (see Fig. 4).Furthermore, we analyzed three strains of eachof five bacterial species: E. coli, S. aureus, P.aeruginosa, B. cereus, and L. monocytogenes. Allstrains within a species yielded the same pattern,indicating that this microsphere-based SNPsassay system is reliable and accurate.

DISCUSSION

A primary advantage of the fluorescentmicrosphere technology is the ability to multi-plex biological reactions simultaneously in asingle reaction vessel [Kettman et al., 1998].Examining multiple variable sites in the samereaction reduces labor, time, and cost comparedto DNA sequencing or single reaction-basedhybridization methods. Another advantage ofthe microsphere-based readout technology plat-form is that the enzymatic reactions are con-ducted in solution, as opposed to methods basedon solid-surface reactions. This allows us to ob-tain the benefit of true liquid-phase kinetics. Wehave developed several assays to perform accu-rate genotyping, and these assays have beenadapted to the Luminex 100 fluorimeter for highthroughput SNPs analysis. This flow cytometricplatform offers several major advantages overconventional flow cytometers, including several-fold reduction in initial instrument cost, and anincrease in sample throughput by using 96-wellplates instead of single tubes. We have demon-strated that OLA [Iannone et al., 2000] andSBCE [Chen et al., 2000] assays can be multi-plexed using fluorescent microspheres. In an at-tempt to simplify the reaction step in the SBCEassay, in which G, A, T, and C are required to beset up separately for each given DNA template,we have successfully developed an allele-specificprimer extension (ASPE) reaction. In thismethod, a pair of allele-specific primers, whichdiffer from each other at the 3′ end (polymor-phic site) and encode different ZipCode se-

FIGURE 1. Schematic presentation of the microsphere-based SNP assays. DNA fragments containing the poly-morphic site to be typed are amplified by PCR. For theSBCE assay, PCR products containing a SNP were pooledand treated with SAP and exonuclease I. After heat inac-tivation of the enzymes, the PCR products were used inthe SBCE reaction as described in Materials and Meth-ods. For each SNP, one capture probe with a uniqueZipCode sequence was used to assay the two alleles ineach of two separate wells with a different labeled ddNTPper well. The probe was extended and labeled by onenucleotide in the presence of Ampli Taq FS DNA poly-merase. In ASPE reactions, for each SNP, a pair of probeswas designed, differing from each other at their extreme3′ nucleotide (the polymorphic site). In the presence ofDNA polymerase, dNTPs and a small portion of the bi-otin-labeled dCTP, a labeled extension product of the 3′portion of the primer is obtained, only if the template in-cludes the target sequence. For both SBCE and ASPEassays, multiplexed SNP analysis could be achieved bythe employment of different ZipCode sequences for dif-ferent SNPs in the presence of pooled PCR products.Microspheres covalently attached with an oligonucleotideencoding the complement to the ZipCode sequence anda luciferase sequence (SeqLUC) are added to the com-pleted reactions and hybridization reactions are carriedout at 40°C in the presence of NaCl. The microspheresare then subjected to flow cytometric analysis. A mini-mum of 30 microspheres of each type were read and themean fluorescence intensity (MFI) value was used for de-termining the genotypes. The fluorescence signal contrib-uted by the microsphere alone was subtracted from alldata points. [Color figure can be viewed in the onlineissue, which is available at www.interscience.wiley.com.]

312 YE ET AL.

quences at the 5′ end, is used in the same reac-tion. The DNA polymerase will extend only oneprimer if the template DNA sequence is homozy-gous, and both primers will be extended in thecase of heterozygotes. An advantage over theSBCE reaction is the ability to read alleles froma given SNP in one tube. With SBCE, eachnucleotide requires analysis in a separate tubewhen using ddNTP terminators labeled with onefluorochrome. This ASPE advantage is possiblebecause the “query” nucleotide is part of theASPE capture probe while the signal-generat-ing “labeled” nucleotide is the free biotin-dCTP.In the SBCE assay, the biotin-ddNTP serves asboth “query” and “labeled” nucleotide. The ne-cessity of post-PCR cleanup and the addition ofunlabeled nucleotides in the ASPE reaction iseliminated. The residual dNTPs from the tar-

get-generating PCR reaction are further used forthe primer extension. Although for each SNPassay two types of microspheres are needed for apair of capture probes, with the set of 100

FIGURE 2. Multiplexed SNPs analysis using allele-specific primer extension. 1344 genotypes from 96 separate patientsamples were analyzed for multiplexed A/G SNPs. Representative results for one plate of 96 patient samples across fourof 15 SNPs are displayed in cluster plot format. Normalized MFI values for the G-allele are plotted on the Y-axis, while A-allele values appear on the X-axis. [Color figure can be viewed in the online issue, which is available atwww.interscience.wiley.com.]

FIGURE 3. Multiplex ASPE (A) and SBCE (B) assays us-ing nine variable sites of 16S rDNA. PCR products wereamplified individually from S. aureus, P. aeruginosa, B.cepacia, and E. coli using universal primer pair 27f/1525r.10 ng of PCR products were used as template for single-tube ASPE, or for A, C, G, or T (shaded columns) SBCEreactions as described in Materials and Methods. Cap-ture probes used in this experiment are listed in Table 1.For capturing the reaction products, 1000 microspheresfor each SNP were added to the reactions. After hybrid-ization at 40°C for more than 60 min, the samples wereanalyzed by a Luminex 100 flow cytometer. The fluores-cence intensity on the microspheres is displayed on the Y-axis as the mean fluorescence intensity (MFI). [Colorfigure can be viewed in the online issue, which is avail-able at www.interscience.wiley.com.]

FLUORESCENT MICROSPHERE FOR SNP ANALYSIS 313

FIGURE 3.

314 YE ET AL.

microspheres available from Luminex Corp(Austin, TX), 50 SNPs per well and up to 96 as-says could be analyzed on the much less expen-sive LX 100 instrument equipped with an XY-platereader. The reading time of the XY-plate readerfor one microtiter plate is approximately one hour.Over 30,000 genotypes could be generated in aneight-hour day per instrument. An even moredramatic increase in throughput could beachieved through multiplexed PCR amplificationof the targets and use of automation to allow ex-tended operation. We estimate our average costis less than $0.20 per SNP, excluding the cost ofgenerating the PCR target. SBCE and ASPE re-actions are comparable in cost. To minimize sub-

jectivity of genotypic calls, various data-cluster-ing algorithms are currently under developmentthat will allow automatic assignment of genotypesto the different clusters.

Bacterial identification has become an essen-tial tool in areas such as healthcare and foodand water quality testing. Traditional methodssuch as microscopy and culturing techniques aretime consuming and have severe limitations.Molecular methods have recently been devel-oped for bacterial identification. These ap-proaches rely on hybridization to specific DNAfragments or sequence determination of con-served regions from a bacterial genome, usuallyafter amplification of the DNA by PCR. The

FIGURE 4. Design of the capture probes for multiplex ASPE or SBCE assays using 16S rDNA. A: Physical locations of the16S rDNA probes. Based on the multi-alignment of different bacterial 16S rDNA sequences, 16 conserved regions werechosen for SBCE and ASPE assays. For SBCE reactions, probes are designed such that the 3′ end of the primers termi-nates one base 5′ to the variable site. For the ASPE assay, a pair of probes was designed such that the 3′ end differs fromeach other at the variable site. The locations of the probes are listed in Table 1. B: Polymorphic patterns in multiplexedSBCE and ASPE assays. Bacterial species can be divided into 17 groups based on their unique readout patterns.

FLUORESCENT MICROSPHERE FOR SNP ANALYSIS 315

DNA sequence for 16S ribosomal RNA is cur-rently the molecule of choice. There are severalbenefits to analyzing 16S rDNA sequences ofbacteria as opposed to their biochemical prop-erties. First, the 16S rDNA sequence is specificfor bacteria at the species level [Kolbert andPersing, 1999]. Second, bacterial isolates do notneed to be subcultured for DNA analysis. DNAis present and stable in all living cells. Most im-portantly, DNA can be extracted from the origi-nally isolated colony, amplified, and analyzed ina fraction of the time taken for biochemicalanalysis. So far, the reference method for 16SrDNA analysis involves PCR amplification us-ing conserved primers followed by sequencingor hybridization with species-specific probes[Bottger, 1989; Kirschner and Bottger, 1998;Kolbert and Persing, 1999; Marchesi et al., 1998;McCabe et al., 1999; Tang et al., 2000]. Theseapproaches are more sensitive and more specificthan most biochemical identifications, but themethods currently available for 16S rDNAanalysis are still labor-intensive and often notsuitable for routine diagnostics. We have suc-cessfully adapted our microsphere-based SNPassay system for bacteria identification. In ourstudy, by using a combination of capture probesdesigned from 16 variable sites within the 16SrDNA region for multiplex SBCE and ASPEassays, we were able to discriminate the 17 testedbacterial species based on their unique readoutpatterns (see Fig. 4). Little or no compromise offluorescent signal was observed between theuniplexed and multiplexed experiments (see Fig.3). A comparison of our assay results with pub-lished sequence data showed that the identifi-cations were correct. Our results demonstratethat bacterial identification using 16S rDNA canbe performed in a simple, multiplexed fashion.Multiplex sequence determination is demon-strated by simultaneously probing multiple vari-able sites from a single polymerase chain reaction(PCR) product. Species-specific primer exten-sion reactions and high throughput readout tech-nology employing flow cytometric analysis ofmicrospheres can be rapidly completed. Only asingle colony or a small amount of bacterialsample is required, and it costs less than one USdollar for each sample analyzed. Further work

involving the design and validation of moreprobes from additional 16S rDNA variable sitesand other genes will be needed to realize thepotential use of this system for high throughputbacterial identification.

ACKNOWLEDGMENTS

The authors thank Philip Rivers, ArashAfshari, and Eric Lai for reagents and helpfuldiscussion, and the Glaxo Wellcome GenotypingFacility and the Glaxo Wellcome SequencingCore Facility for their services. Thanks also aredue to Dan Burns of the Genetics Directorate,and Ralph McDade, Van Chandler, Mark Chan-dler, Jim Jacobson, and Christy Weiss of LuminexCorporation for many helpful discussions.

REFERENCES

Bottger EC. 1989. Rapid determination of bacterial RNA se-quences by direct sequencing of enzymatically amplifiedDNA. FEMS Microbiol Lett 65:171–176.

Cai H, White PS, Torney D, Deshpande A, Wang A, MarroneB, Nolan J. 2000. Flow cytometry-based minisequencing:a new platform for high-throughput single-nucleotide poly-morphism scoring. Genomics 66:135–143.

Chen J, Iannone MA, Li MS, Taylor JD, Rivers P, Nelsen AJ,Slentz-Kesler KA, Roses A, Weiner MP. 2000. A micro-sphere-based assay for single nucleotide polymorphismanalysis using single base chain extension. Genome Res10:549–557.

Chen XN, Levine L, Kwok PY. 1999. Fluorescence polariza-tion in homogeneous nucleic acid analysis. Genome Res9:492–498.

Cooper DN, Smith BA, Cooke HJ, Niemann S, Schmidtke J.1985. An estimate of unique DNA sequence heterozygos-ity in the human genome. Hum Genet 69:201–205.

Fan JB, Chen X, Halushka MK, Berno A, Huang X, Ryder T,Lipshutz RJ, Lockhart DJ, Chakravarti A. 2000. Parallelgenotyping of human SNPs using generic high-density oli-gonucleotide tag arrays. Genome Res 10:853–860.

Fu DJ, Tang K, Braun A, Reuter D, Darnhofer-Demar B, LittleDP, O’Donnell MJ, Cantor CR, Koster H. 1998. Sequenc-ing exons 5 to 8 of the p53 gene by MALDI-TOF massspectrometry. Nat Biotechnol 16:381–384.

Fulton RJ, McDade RL, Smith PL, Kienker LJ, Kettman JR Jr.1997. Advanced multiplexed analysis with the FlowMetrixsystem. Clin Chem 43:1749–1756.

Gilles PN, Wu DJ, Foster CB, Dillon PJ, Chanock SJ. 1999.Single nucleotide polymorphic discrimination by an elec-tronic dot blot assay on semiconductor microchips. NatBiotechnol 17:365–370.

316 YE ET AL.

Iannone MA, Taylor JD, Chen J, Li MS, Rivers P, Slentz-KeslerKA, Weiner MP. 2000. Multiplexed single nucleotide poly-morphism genotyping by oligonucleotide ligation and flowcytometry. Cytometry 39:131–140.

Kettman JR, Davies T, Chandler D, Oliver KG, Fulton RJ.1998. Classification and properties of 64 multiplexedmicrosphere sets. Cytometry 33:234–243.

Kirschner P, Bottger EC. 1998. Species identification of my-cobacteria using DNA sequencing. Methods Mol Biol101:349–361.

Kolbert CP, Persing DH. 1999. Ribosomal DNA sequencingas a tool for identification of bacterial pathogens. Curr OpinMicrobiol 2:299–305.

Lai E, Riley J, Purvis I, Roses A. 1998. A 4-MB high-densitysingle nucleotide polymorphism-based map around humanAPOE. Genomics 54:31–38.

Landegren U, Kaiser R, Sanders J, Hood L. 1988. A ligase-mediated gene detection technique. Science 241:1077–1080.

Livak KJ, Marmaro J, Todd JA. 1995. Towards full automatedgenome-wide polymorphism screening. Nat Genet 9:341–342.

Marchesi JR, Sato T, Weightman AJ, Martin TA, Fry JC, HiomSJ, Wade WG. 1998. Design and evaluation of useful bac-terium-specific PCR primers that amplify genes coding forbacterial 16S rRNA. Appl Environ Microbiol 64:795–799.

McCabe KM, Zhang YH, Huang BL, Wagnar EA, McCabeERB. 1999. Bacterial species identification after DNAamplification with a universal primer pair. Mol Genet Met66:205–211.

McDade RL, Fulton RJ. 1997. True multiplexed analysis bycomputer-enhanced flow cytometry. Med Dev Diag Indust19:75–82.

McHugh TM. 1994. Flow microsphere immunoassay for the

quantitative and simultaneous detection of multiple solubleanalytes. Meth Cell Biol 42:575–595.

Mein CA, Barratt BJ, Dunn MG, Siegmund T, Smith AN,Esposito L, Nutland S, Stevens HE, Wilson AJ, PhillipsMS, Jarvis N, Law S, de Arruda M, Todd JA. 2000. Evalu-ation of single nucleotide polymorphism typing with In-vader on PCR amplicons and its automation. Genome Res10:330–343.

Orita M, Iwahana H, Kanazawa H, Hayashi K, Sekiya T. 1989.Detection of polymorphisms of human DNA by gel elec-trophoresis as single-strand conformation polymorphisms.Proc Natl Acad Sci USA 86:2766–2770.

Pastinen T, Kurg A, Metspalu A, Peltonen L, Syvanen AC.1997. Minisequencing: a specific tool for DNA analysisand diagnostics on oligonucleotide arrays. Genome Res7:606–614.

Pastinen T, Raitio M, Lindroos K, Tainola P, Peltonen L,Syvanen AC. 2000. A system for specific, high-through-put genotyping by allele-specific primer extension onmicroarrays. Genome Res 10:1031–1042.

Saiki RK, Walsh RS, Levenson CH, Erlich HA. 1989. Ge-netic analysis of amplified DNA with immobilized se-quence-specific oligonucleotide probes. Proc Natl AcadSci USA 86:6230–6234.

Syvanen AC. 1999. From gels to chips: “minisequencing”primer extension for analysis of point mutations and singlenucleotide polymorphisms. Hum Mutat 13:1–10.

Tang YW, Von Graevenitz A, Waddington MG, Hopkins MK,Smith DH, Li H, Kolbert CP, Montgomery SO, Persing DH.2000. Identification of coryneform bacterial isolates by ri-bosomal DNA sequence analysis. J Clin Microbiol 38:1676–1678.

Tyagi S, Bratu DP, Kramer FR. 1998. Multicolor molecular bea-cons for allele discrimination. Nat Biotechnol 16:49–53.