Genotyping Techniques

Research Report Melting Curve Analysis of SNPs (McSNP): A Gel-Free and Inexpensive Approach for SNP Genotyping

BioTechniques 30:358-367 (February 2001)

J.M. Akey, D. Sosnoski1, INTRODUCTION nostic probe once per PCR cycle (25), E. Parra1, S. Dios1, K. Hiester1, have been applied to SNP genotyping 1 The third-generation genetic map methods (6,10,14). The dsDNA-specif- B. Su, C. Bonilla, L. Jin, and  M.D. Shriver1 comprised of single nucleotide poly- ic dye SYBR Green I (Moleclar morphisms (SNPs) is being expeditious- Probes, Eugene, OR, USA) has been University of Texas Houston ly developed (24). SNPs are the most used to analyze the melting curves of Health Science Center, Houston, abundant type of sequence variation in PCR products, which are characterized 1 TX and Penn State University, the human genome and will be useful by a rapid loss of fluorescence as the University Park, PA, USA tools in many diverse applications that temperature is raised through the sam- include disease gene mapping, evolu- ples’ melting temperature (Tm) (17). tion, pharmacogenetics, and forensics Since the Tm is a function of product ABSTRACT (3,11). An impressive SNP resource al- length, sequence composition, and GC ready exists as nearly 300000 have been content, it should be possible to distin- High-throughput methods for assaying deposited into publicly accessible data- guish DNA fragments that differ with DNA variation require two important steps: bases (see http://www.ncbi.nlm.nih.gov/ respect to these parameters by melting (i)discriminating the variation and (ii)de- SNP/, http://snp.cshl.org/, and http://hg- curve analysis (MCA). The general tecting the signal. In this report, we describe base.cgr.ki.se/). However, without paral- principles of MCA were used to devel- a novel SNP genotyping method that we re- lel progress in SNP genotyping technol- op Tm-shift genotyping, an SNP geno- fer to as melting curve analysis of SNPs ogy, their true power and inherent typing assay based on allele-specific (McSNP). McSNP combines a classic ap- benefits will not come to fruition. PCR (6). While this represents an im- proach for discriminating alleles, restriction Novel genotyping methods amen- portant advance in MCA applied to SNP enzyme digestion, with a more recent able to high-throughput analysis should genotyping, it is subject to the inherent method for detecting DNA fragments, melt- ideally be gel-free, robust, inexpensive, limitations of allele-specific PCR, such ing curve analysis. Melting curve analysis is and simple to perform. To this end, as difficulty in reaction optimization. performed by slowly heating DNA frag- these requirements have inspired the In the burgeoning field of SNP geno- ments in the presence of the dsDNA-specific development of a variety of genotyping typing technology, it is easy to forget the fluorescent dye SYBRGreen I. As the sam- assays, including the oligonucleotide elegantly simple and accurate approach ple is heated, fluorescence rapidly decreas- ligation assay (OLA) (12), genetic bit of assaying SNP variation by digesting es when the melting temperature of a partic- analysis (GBA) (15), mass spectro- PCR products with restriction enzymes ular fragment is reached. We show that it is scopy (8), “chip” technology (24), Taq- [i.e., PCR-restriction fragment length possible to determine the composition of Man(14), and dynamic allele specific polymorphism (RFLP) typing]. In this simple mixtures of DNA fragments, such as hybridization (DASH) (10). Although report, we describe how MCA can be those that result from restriction enzyme di- many SNP genotyping methods have applied to classical PCR-RFLP meth- gestions of short PCR products. McSNP is been developed, no single technology ods. Specifically, a simple, inexpensive, well suited for high-throughput genotyping has emerged as being clearly superior and high-throughput assay for SNP because 96 samples can be analyzed and because of limitations such as cost, genotyping that we refer to as MCA of automatically scored in 20 min. Our results complexity, and accuracy. SNPs (McSNP) is presented in which clearly demonstrate that McSNP is a simple, Recently, the principles underlying the melting profiles of restriction en- inexpensive, and accurate means of geno- kinetic PCR (9), which is predicated on zyme-digested PCR products are ana- typing SNP variation. monitoring the fluorescence of a diag- lyzed. In total, we present data on seven

358BioTechniques Vol. 30, No. 2 (2001) SNPs, clearly demonstrating that this ences 7, 5, and 22, respectively. Marker (Hybaid). Other instruments that may method is a practical, robust, and sensi- M89 was obtained from Reference 21. be suitable for MCA include the Gene- tive approach to assaying SNP variation. The samples used in this study repre- Amp5700 and ABI Prism7700 Se- sent a subset of samples that we are us- quence Detection Systems (Applied ing in other studies and are comprised Biosystems), the LightCycler(Roche MATERIALS AND METHODS of African-American, European-Amer- Molecular Biochemicals, Indianapolis, ican, and Hispanic populations. More IN, USA), the iCycler(Bio-Rad Lab- PCR specific details can be found elsewhere oratories, Hercules, CA, USA), the (16,20). Informed consent from each Smart Cycler (Cephied, Sunnyvale, SNPs were amplified by PCR, and subject was obtained for participation CA, USA), and the Sentinel(Strata- most reactions were conducted under a in this study. gene, La Jolla, CA, USA). However, standard set of conditions. PCR was we have not explored the general accu- performed in either a Hybaid Multi- Restriction Enzyme Digestion racy, robustness, and scoring software Block System (Hybaid, Franklin, MA, associated with these machines, which USA) or a GeneAmp 9700 (Applied Restriction enzyme digestions were may differ substantially. Melting curves Biosystems, Foster City, CA, USA) and performed in a final volume of 25 µL, are acquired by ramping the tempera- consisted of the following thermocy- which consisted of 10 µL PCR product, ture from 35°C to 90°C at a rate of cles: 95°C for 5 min, followed by 30 1 U appropriate restriction enzyme, 1× 0.04°C/s and monitoring the change in cycles of 95°C for 30 s, 55°C for 30 s, reaction buffer supplied with the re- fluorescence of SYBR Green I at 520 72°C for 30 s, and a final extension at striction enzyme, and, when recom- nm. The total run time is approximately 72°C for 5 min. PCRs were performed mended by the supplier, 1×bovine 22 min, which can be generally cut in in 25-µL reactions consisting of 25 ng serum albumin (BSA). In particular, half once the Tmis determined and the genomic DNA, 1.5 mM MgCl2, 2.5 µL markers 4019, M89, and 14867 were thermal range adjusted accordingly. It 10×PCR buffer (Life Technologies, digested by the restriction enzyme is during the slow ramp to 90°C that the Rockville, MD, USA), 0.1 mM dNTPs, NlaIII, while markers CKMM, 14319, fluorescence is captured. The raw data 0.04 µM each primer, and 2.5 U Taq FY-null, and LPL were digested by the are first converted to relative fluores- DNA polymerase (Life Technologies). restriction enzymes TaqI, RsaI, StyI, cence by dividing each point by the ini- The exception to these conditions was and PvuII, respectively. We did not find tial fluorescence (i.e., 35°C) (6). Final marker Fy-null, in which the final it necessary to purify the PCR products melting curves are reported as the MgCl2was 2.0 mM. Primer sequences before restriction enzyme digestions, three-point smoothed negative first de- were as follows: CKMM (5′-GCAG- which were performed as recommend- rivative of fluorescence, with respect to GCGCCTACTTCTGG-3′and 5′-AGC- ed by the suppliers for a period of 4–24 temperature versus temperature with a TCATGGTGGAAATGGAG-3′), 4019 h. All restriction enzymes were pur- baseline subtraction. The baseline cor- (5′-CAGGCCAAGAGCGTCCTA-3′ chased from New England Biolabs rection for each data point is calculated and 5′-TGCCACTCTGTGAACAGC- (Beverly, MA, USA). by subtracting the slope from a linear AA-3’), 14319 (5′-CATCTGAGTGC- regression line of the first nine data AAGATAAAAAGGA-3′and 5′-CCC- MCA points. All genotypes were confirmed ACCCCCAAATCATCTAT-3′), FY- by agarose gel electrophoresis. Specifi- null (5′-GCCCAGAACCTGATGGCC- Melting profile reactions were ana- cally, samples were amplified by a sec- CTCATTAGTGCT-3′and 5′-CTGTC- lyzed in a final volume of 50 µL. All re- ond set of PCR primers to produce larg- AGCGCCTGTGCTT-3′), 14867(5′-G- actions contained 5 µL restriction en- er products more suitable for scoring GCAGGACATTCCAAGGCTCTC-3′ zyme product, 5 µL 10×SYBR Green I by agarose gel electrophoresis. These and 5′-CACCCTGGGTTAACACATT- (which comes as a 10000×stock sup- samples were then independently di- CA-3′), LPL (5′-TGCAAGGGTTTT- plied by Molecular Probes), and were gested and scored by agarose gels. GCTTAATTCT-3′and 5′-CAACAA- brought to a final volume of 50 µL in CAAAACCCCACAGC-3′), and M89 double-distilled water. Various destabi- (5′-ACAGAAGGATGCTGCTCAGC- lizing agents were added to these reac- RESULTS TT-3′and 5′-GCAACTCAGGCAAA- tions including formamide, dimethyl GTGAGACAT-3′), in which the bold sulfoxide (DMSO), and urea (all from Restriction Enzyme MCA Profiles denotes the mismatched base to create Fisher Biotech, Fair Lawn, NJ, USA). the putative NlaIII restriction site (see Here, we provide the details of only The experimental protocol of Mc- Results section). SNP markers WI- those additives that had an enhancing SNP is very simple and is outlined in 4019 (dbSNP ID: 2420), WI-14319, effect on MCA. The optimal final con- Figure 1. Figure 2 illustrates typical Mc- and WI-14867 were identified by centrations of these additives were ei- SNP genotype profiles for six autosomal querying the Whitehead Institute ther 10% formamide or 10% DMSO. SNP loci. The peak of a melting curve Genome Center SNP database (http:// All reactions were performed in poly- denotes the Tm of a particular DNA www-genome.wi.mit.edu/). SNP mark- carbonate 96-well plates (Hybaid). fragment, which is defined as the tem- ers LPL, CKMM (dbSNP ID: 5318), Real-time fluorescent monitoring was perature at which half of the fragments and FY-null were obtained from Refer- performed with the DASH machine in solution have denatured. As expected,

Vol. 30, No. 2 (2001) BioTechniques 359 Genotyping Techniques

the observed Tm values of undigested gested allele results in two fragments between them decreases. Our laborato- SNP alleles are higher compared to the with the same or different Tm. ry has had success using fragments in digested alleles. Note the clear distinc- the range of 50–150 bp. tion between homozygous and heterozy- Engineering a Restriction Site Moreover, it is necessary to add a gous individuals, in which heterozygous destabilizing agent to ensure that the Tm melting curves are composites of the In many instances, an SNP will not of a fragment is reached during the ex- ° two homozygous melting curves. cause a restriction enzyme site differ- periment (i.e., Tm<100C). We have More specifically, Figure 2, panel A, ence. In these cases, a digestion site for tested formamide, DMSO, and urea for consists of homozygous undigested one of the SNP alleles can be introduced their effects on melting curve profiles. genotypes, and, thus, only one Tmpeak by designing one of the PCR primers The overall effects of these additives are is observed. Figure 2, panel B, contains with a mismatch near the 3′end (13). similar in that they destabilize dsDNA to examples of heterozygous genotypes in For instance, the Y-chromosome SNP, varying degrees and thus lower the Tm. which one allele is digested and the M89, was genotyped with McSNP by However, they have different influences other is undigested. In general, we de- using a primer with a mismatched C 3 on the shape of melting curves. For ex- tected two distinct types of heterozy- bp upstream of the 3′end, thus creating ample, the addition of urea resulted in gous melting curves, in which het- a NlaIII restriction site in one of the alle- broadening of peaks and decreasing the erozygotes formed either two (i.e., les. Individuals with the M89*T allele fluorescence of the sample (data not marker CKMM) or three (i.e., marker will be undigested and produce an 87-bp shown). The effects of DMSO and for- 4019, FY-null, and 14319) resolvable fragment, whereas those with the mamide were found to be superior peaks. In three-peak heterozygotes, one M89*C allele will be digested resulting because they resulted in more sharply high-temperature peak corresponds to in fragments of 20 and 67 bp, as seen in defined peaks. Thus, DMSO and forma- the undigested allele, while the other Table 1. Although a mismatched primer mide have relatively equivalent effects two lower-temperature peaks result theoretically reduces the efficiency of on melting curve characteristics. Al- from the two fragments produced by PCR, these primers are not difficult to though we chose to use formamide, one endonuclease cleavage of the digested work with in practice and often do not could also use DMSO to lower the Tm. allele (Figure 1). Conversely, in two- require optimizations over that of a stan- peak heterozygotes, the two resulting dard PCR. For more specific details, see TmEstimation DNA fragments from endonuclease References 1, 4, and 13. cleavage of the digested allele have Theoretical prediction of the Tmof a similar Tm values and melt simultane- Major Parameters Affecting McSNP DNA molecule is complex and has been ously. Thus, the difference between Profiles the subject of numerous studies (re- two- and three-peak heterozygotes is viewed in Reference 23). We have inves- whether the digested allele gives rise to Critical factors in McSNP include tigated the empirical relationship be- DNA fragments with the same or dif- the size and sequence of PCR products tween the observed and predicted Tm ferent Tm. Note that LPL and 14867 are as well as additives, which destabilize because it is important to have some a actually three-peak heterozygotes, but, the DNA duplex and lower the Tm. In prioriinformation regarding the expect- for presentation purposes, the lowest particular, there is a greater difference ed resolution of digested and undigested Tmpeak was not included in the figure in the Tm between smaller DNA frag- DNA fragments. The two methods used (Table 1). Finally, Figure 2, panel C, ments than there is for larger DNA for predicting Tmwere a simple salt-ad- demonstrates homozygous digested fragments. Theoretical and empirical justed formula (http://www.biotechlab. genotypes. Two possible patterns of ho- studies of DNA denaturation have nwu.edu/OligoCalc.html) (18,19) and a mozygous digested melting curves are shown that, as the size of DNA frag- nearest-neighbor algorithm (http://alces. possible, depending on whether the di- ments increases, the difference in Tm med.umn.edu/rawtm.html) (2).

Figure 1. The experimental protocol of McSNP. The SNP is amplified by PCR and then subjected to restriction enzyme digestion. The identity of each DNA fragment is determined by the unique pattern of melting curve peaks acquired by analyzing the change in fluorescence of the dsDNA-specific dye SYBR Green I as the sample is heated. In this example, the individual is scored as a heterozygote having one undigested and one digested allele (alleles 1 and 2, respectively).

360BioTechniques Vol. 30, No. 2 (2001) Genotyping Techniques

362BioTechniques Vol. 30, No. 2 (2001) Genotyping Techniques

Figure 2. Typical McSNP profiles of the six autosomal SNPs investigated (4019, FY-null, CKMM, 14319, 14867, and LPL).Panels A, B, and C denote ho- mozygous undigested, heterozygous, and homozygous digested genotypes, respectively. Note the clear resolution of homozygous and heterozygous genotypes. No template negative controls showed below background levels of fluorescence and are therefore not included in the figure.

364BioTechniques Vol. 30, No. 2 (2001) Genotyping Techniques

Table 1 presents the average ob- Second, unusual peaks in the scoring tools that have evolved in nature to rec- served and predicted Tm for all DNA range (0.4%) disappeared when the ognize SNPs. The two main disadvan- fragments of the seven SNPs investigat- same samples were amplified, digested, tages of “classic” PCR-RFLP typing are ed based on the nearest-neighbor algo- and scored again with the DASH soft- (i)limited throughput and (ii) partial di- rithm. Similar results were obtained for ware. These problems are attributable gestion of PCR products, which de- the alternative formula described above to artifacts in the amplification or mon- creases accuracy. However, both of (data not shown). The observed Tm is itoring of the Tm profile. Overall, the these concerns are addressed by analyz- the average over one or two 96-well automatic scoring feature of the DASH ing digestion products by McSNP sample plates. There is a high concor- software allows a fast and accurate rather than agarose gel electrophoresis. dance between the observed and pre- classification of genotypes. However, a First, throughput of gel-based meth- dicted Tmlevels for the DNA fragments careful inspection of the automatic ods is limited because it requires sub- (r2= 0.94) shown in Table 1 (Figure 3). scores is recommended to detect and stantial time and labor to pour, load, Note the CKMM digested fragments correct possible errors caused by PCR stain, and interpret gels. On the other (labeled 2a and 2b), which, despite a artifacts or partial digestion. hand, McSNP reactions are self-con- 14-bp difference in size, have predicted tained because it is possible to conduct and observed Tm points that differ by PCR, restriction enzyme digestion, and only 0.2°C and 0°C, respectively. As in- DISCUSSION MCA in the same 96-well plate, with the dicated in Figure 2, these two fragments analysis step requiring less than 20 min. melt together in a single peak. These In this report, we have investigated Second, partial digestion is a poten- observations have important implica- how the general approach of MCA, tial problem in any method using re- tions for the design of McSNP experi- which is based on the natural phenom- striction enzymes and originates pri- ments because they suggest that the var- enon of Tmdifferences between DNA marily from either too many PCR ious Tmprediction formula can be used fragments that differ in sequence and cycles or decreased efficiency of the re- as an accurate tool to estimate a priori size, can be used for genotyping. striction enzyme digestion because of whether or not there will be a resolvable Specifically, we have developed Mc- the PCR components (e.g., Triton X- difference in Tm between two DNA SNP, a simple, high-throughput, and in- 100). Simple solutions to these prob- fragments. We suggest that, as a rough expensive SNP genotyping assay. Fur- lems include performing fewer PCR approximation, two DNA fragments thermore, we have established the cycles and diluting the PCR product in can be resolved if their predicted Tm general robustness of McSNP by geno- a larger volume before restriction en- difference is at least 3°C. typing larger numbers of individuals zyme digestion. Both methods are more and using the genotyping software of feasible for McSNP typing because the Automated Scoring the DASH machine to automatically sensitivity of MCA is considerably score genotypes. higher than gel-based detection (data To investigate the utility and robust- While restriction enzymes may seem not shown). So far, we have seen little ness of McSNP on a larger scale, we archaic compared to some of the recent evidence for partial digestion (<1%; see genotyped 853 individuals for marker technologies developed for SNP geno- Automated Scoring section of Results) CKMM. We used the automatic scoring typing (8,24), it should be noted that and do not anticipate this to be an ob- feature of the DASH software (version they are very accurate, as they rely on stacle for most McSNP experiments. 2.14), which allows genotypes to be “called” automatically after an experi- ment is complete. Of the 853 samples, 37 failed to amplify (i.e., PCR failures). Of the remaining 816 samples, 807 were correctly genotyped and scored (98.9%). Thus, only 1.1% (nine indi- viduals) of the reactions that were am- plified and automatically scored with the DASH software were repeated be- cause of inconsistencies with the ob- served melting curve pattern. These problems can be classified into two cat- egories. First, unusual patterns in the relative size of the melting curves (0.7%) are likely due to partial diges- tion of the PCR products, and after sub- sequent amplification and digestion with increased amounts of enzyme Figure 3. Relationship between observed and expected T points.We found a high correlation be- and/or fewer PCR cycles, the samples m tween observed and predicted Tmpoints. The predicted Tmis based on a salt adjusted nearest-neighbor show the expected melting profiles. algorithm described in the text (2), and the plotted data are from Table 1.

366BioTechniques Vol. 30, No. 2 (2001) An important requirement of the Mc- grant National Institutes of Health grant 15.Nikiforov, T.T., R.B. Rendle, P. Goelet, Y.H. SNP assay is the presence of a restric- nos. DK53958 and HG02154 to M.D.S. Rogers, M.L. Kotewicz, S. Anderson, G.L. tion site generated by the SNP. However, There are several patents pending relat- Trainor, and M.R. Knapp.1994. Genetic Bit Analysis: a solid phase method for typing sin- it has been estimated that SNPs only oc- ed to the contents of this manuscript. gle nucleotide polymorphisms. Nucleic Acids cur in natural restriction sites 50% of the Res. 11:4167-4175. time (12). Thus, the ability to create arti- 16.Parra, E.J., A. Marcini, J. Akey, J. Martin- ficial restriction sites during PCR is crit- REFERENCES son, M.A. Batzer, R. Cooper, T. Forrester, D.B. Allison, R. Deka, R.E. Ferrell, and ical to the general applicability of Mc- 1.Athma, P., N. Fidahusein, and M. Swift. M.D. Shriver.1998. Estimating African Am- SNP. To this end, computer simulations 1995. Single base polymorphism linked to the erican admixture proportions by use of popu- have been performed, and it is estimated ataxia-telangiectasia locus is detected by mis- lation-specific alleles. Am. J. Hum. Genet. that 98% of all polymorphisms occur in match PCR. Biochem. Biophys. Res. Com- 63:1839-1851. either natural or artificial restriction sites mun. 210:982-986. 17.Ririe, K.M., R.P. Rasmussen, and C.T. Wit- 2.Breslauer, K.J., R. Frank, H. Blocke, and twer.1997. Product differentiation by analysis for the commercially available four-base L.A. Marky.1986. Predicting DNA duplex of DNA melting curves during the polymerase cutters (13). Hence, McSNP will be able stability from the base sequence. Proc Natl chain reaction. Anal. Biochem. 245:145-160. to genotype the preponderance of identi- Acad Sci. USA 83:3746-3750. 18.Rychlik, W. and R.E. Rhoads.1989. A com- fied SNPs. 3.Brookes, A.J.1999. The essence of SNPs. puter program for choosing optimal oligonu- Gene 234:177-186. cleotides for filter hybridization, sequencing It is also important to consider the 4.Cohen, J.B. and A.D. Levinson.1988. A and in vitro amplification of DNA. Nucleic cost of a genotyping assay. The reagent point mutation in the last intron responsible Acids Res. 17:8543-8551. and consumable costs of a McSNP ex- for increased expression and transforming ac- 19.Sambrook, J., E.F. Fritsch, and T. Maniatis. periment are very low. The post-PCR tivity of the c-Ha-ras oncogene. Nature 1989. Molecular Cloning: A Laboratory Man- cost per genotype can be broken down 334:119-124. ual. CSH Laboratory Press, Cold Spring Har- 5.Gennarelli, M., G. Novelli, A. Cobo, B. Bai- bor, NY. as follows: $0.06 for the restriction en- get, and B. Dallapiccola.1991. 3′creatine ki- 20.Shriver, M.D., M.W. Smith, L. Jin, A. zyme (an average of 15 commonly used nase (M-type) polymorphisms linked to my- Marcini, J.M. Akey, R. Deka, and R.E. Fer- enzymes), $0.001 for SYBR Green I, otonic dystrophy in Italian and Spanish rell. 1997. Ethnic-affiliation estimation by use $0.0003 for formamide, and $0.06 for populations. Hum. Genet. 87:654-656. of population-specific DNA markers. Am. J. 6.Germer, S. and R. Higuchi.1999. Single- Hum. Genet. 60:957-964. the polycarbonate plates (which can be tube genotyping without oligonucleotide 21.Su, B., J. Xiao, P. Underhill, R. Deka, W. used in the PCR step, further reducing probes. Genome Res. 9:72-78. Zhang, J. Akey, W. Huang, D. Shen et al. costs). Thus, the cost of a McSNP ex- 7.Gotoda, T., N. Yamada, T. Murase, H. Shi- 1999. Y-chromosome evidence for a north- periment is approximately $0.12 be- mano, M. Shimada, K. Harada, M. Kawa- ward migration of modern humans into East- yond that of a PCR, making it one of mura, K. Kozaki, and Y. Yazaki.1992. De- ern Asia during the last Ice Age. Am. J. Hum. tection of three separate DNA polymorphisms Genet. 65:1718-1724. the least expensive SNP genotyping in the human lipoprotein lipase gene by gene 22.Tournamille, C., Y. Colin, J.P. Cartron, and methods available. amplification and restriction endonuclease di- C. Le Van Kim. 1995. Disruption of a GATA Finally, a realistic estimate of sample gestion. J. Lipid Res. 33:1067-1072. motif in the Duffy gene promoter abolishes throughput is 2400 genotypes/day/in- 8.Griffin, T.J., J.G. Hall, J.R. Prudent, and erythroid gene expression in Duffy-negative × L.M. Smith.1999. Direct genetic analysis by individuals. Nat. Genet. 10:224-228. strument (96 samples 25 runs/day, as- matrix-assisted laser desorption/ionization 23.Turner, D.H.1996. Thermodynamics of base suming each run requires 15 min), mass spectrometry. Proc. Natl. Acad. Sci. pairing. Curr. Opin. Struct. Biol. 6:299-304. which translates into 576000 genotypes/ USA 25:6301-6306. 24.Wang, D.G., J.B. Fan, C.J. Siao, A. Berno, year (assuming 48 working weeks). The 9.Higuchi, R., C. Fockler, G. Dollinger, and R. P. Young, R. Sapolsky, G. Ghandour, N. primary bottleneck of the McSNP proto- Watson.1993. Kinetic PCR analysis: real- Perkins et al.1998. Large-scale identifica- time monitoring of DNA amplification reac- tion, mapping, and genotyping of single-nu- col is the PCR step. Therefore, multi- tions. Biotechnology (NY) 11:1026-1130. cleotide polymorphisms in the human plexing McSNP experiments would be 10.Howell, W.M., M. Jobs, U. Gyllensten, and genome. Science 280:1077-1082. advantageous. In principle, it should be A.J. Brookes.1999 Dynamic allele-specific 25.Wittwer, C.T., M.G. Herrmann, A.A. Moss, straightforward to multiplex two loci. hybridization. A new method for scoring sin- and R.P. Rasmussen.1997. Continuous fluo- gle nucleotide polymorphisms. Nat. Biotech- rescence monitoring of rapid cycle DNA am- However, solutions containing three or nol. 17:87-88. plification. BioTechniques 22:130-138. more loci would likely result in complex 11.Kwok, P.Y. and Z. Gu.1999. Single nucleo- melting curve patterns that may substan- tide polymorphism libraries: why and how are Received 10 August 2000; accepted 16 tially overlap. We are exploring the po- we building them? Mol. Med. Today 5:538- October 2000. tential of multiplex McSNP experiments 543. 12.Landegren, U., R. Kaiser, J. Sanders, and and are adapting McSNP to a microar- L. Hood.1988. A ligase-mediated gene detec- Address correspondence to: ray-based format. In conclusion, Mc- tion technique. Science 241:1077-1080. 13.Li, H. and L. Hood. 1995. Multiplex geno- Joshua M. Akey SNP allows SNPs to be genotyped in a Human Genetics Center gel-free high-throughput manner and is type determination at a DNA sequence poly- morphism cluster in the human immunoglobu- Graduate School of Public Health a useful addition to the geneticist’s lin heavy-chain region. Genomics 26: University of Texas Houston Health Science genotyping toolbox. 199-206. Center 14.Livak, K.J., S.J. Flood, J. Marmaro, W. 6901 Bertner Avenue Giusti, and K. Deetz.1995. Oligonucleotides Houston, TX 77030, USA e-mail: [email protected] ACKNOWLEDGMENTS with fluorescent dyes at opposite ends provide a quenched probe system useful for detecting PCR product and nucleic acid hybridization. This work was supported in part by PCR Methods Appl. 4:357-362.

Vol. 30, No. 2 (2001) BioTechniques 367