Optimized PCR labeling in mutational andmicrosatellite analysis

Diana Liberata Esposito, Raffaele Palmirotta, Maria Concetta Verı,Sandra Mammarella, Franca D’Amico, Maria Cristina Curia, Gitana Aceto,

Stefania Crognale, Beatrice Creati, Renato Mariani-Costantini, Pasquale Battista,and Alessandro Cama*

To optimize the labeling and visualization of PCRproducts we tested different variables, including de-oxynucleotide concentration and ratio, dilution of la-beled product, number of PCR cycles, and use of one-step or nested labeling protocols. Labeling was achievedusing a fixed amount of labeled dATP, whose relativespecific activity was varied by adding increasingamounts of cold dATP. Optimal PCR-labeling intensitywas reached at dATP concentrations between 0.9 and 7.0mmol/L, with a peak at 1.8 mmol/L. This concentrationcorresponded to an optimal ratio between the increasein specific activity and the decrease in DNA yield.Nucleotide imbalances >1:2 were not advantageous.Mutational analysis by single-strand conformationalpolymorphism (SSCP) was used to validate PCR-label-ing protocols. The limiting nucleotide concentrationsdid not affect SSCP. Clear SSCP patterns were obtainedusing DNA templates of different sizes derived fromseveral genes. SSCP patterns obtained using one-step ornested PCR-labeling protocols were equivalent andwere visualized after overnight exposure, using[a35S]dATP as the label. Dilutions of labeled productsranging between 1:10 and 1:2.5 influenced SSCP pat-terns, and the lowest dilution tested produced better-defined and more-intense signals. Optimized SSCP con-ditions allowed the detection of novel and previouslycharacterized nucleotide variants. Clear microsatellitetyping was also obtained using optimized protocols and[a35S]dATP as the label.

Radioactive PCR labeling is widely used in a number ofapplications, which include production of nucleic acid

probes, single strand conformational polymorphism(SSCP)1 analysis, and microsatellite typing (1–4). The twomost common methods used for PCR labeling are eitherend-labeling of primers before the reaction or incorpora-tion of a labeled nucleotide during PCR amplification.Methods based on incorporation are more flexible andinvolve less experimental steps than those based onend-labeling. Although incorporation methods have beenused for several years, there is a broad variation in thePCR-labeling conditions used by different laboratories(1, 2, 4–10). For instance, in SSCP analysis, many labora-tories still use conditions identical to or comparable withthe original PCR-labeling protocol described by Orita etal. (5–9). To improve the incorporation of the labelednucleotide during PCR-SSCP, some of these protocols useunbalanced concentrations (up to 10-fold) of labeled vsunlabeled nucleotides (8, 9). Different concentrations oflabeled and unlabeled nucleotides (with an imbalance ofup to 100-fold) are also used for microsatellite analysis(4, 10). The use of disparate PCR-labeling conditions de-notes a lack of standardization for these procedures.Therefore, studies investigating the influence of differentlabeling conditions should contribute to the selection ofappropriate labeling protocols for specific PCR applica-tions. In fact, suboptimal labeling conditions may lead tohigher costs of the analysis, increased autoradiographicexposure time and/or reduced signals, which may in turninterfere with the interpretation of tests. Moreover, opti-mized PCR-labeling conditions could be advantageous forboth radioactive and nonradioactive applications. In theattempt to optimize labeling and detection conditions formutational and microsatellite analysis, we tested several

Department of Oncology and Neurosciences, Faculty of Medicine, Univer-sity “Gabriele D’Annunzio”, Via dei Vestini 1, 66013 Chieti, Italy.

*Author for correspondence. Fax 39-871-3555-4110; e-mail Cama@unich.it.Received January 28, 1998; revision accepted March 17, 1998.

1 Nonstandard abbreviations: SSCP, single strand conformational poly-morphism; HNPCC, hereditary nonpolyposis colorectal cancer; HIR, humaninsulin receptor gene; APC, adenomatous polyposis coli gene; hMLH1, humanMutL homolog 1 gene; IRS-1, insulin receptor substrate 1 gene; and hMSH2,human MutS homolog 2.

Clinical Chemistry 44:71381–1387 (1998) Molecular Diagnostics

and Genetics

1381

PCR conditions. Standardized PCR-labeling conditionswere also tested using 35S-labeled nucleotides in place ofcommonly used 32P-labeled nucleotides.

Materials and Methodsdna samples and sequences analyzedDNAs used for PCR amplification, SSCP, and microsatel-lite analysis were obtained in previous studies from ourlaboratory concerning the following: non-insulin-depen-dent diabetes mellitus, familial adenomatous polyposis,and hereditary non-polyposis colorectal cancer (HNPCC)(11–13). In some cases, to test the effect of differentvariables on mutational analysis by SSCP, we selectedsamples known to harbor single nucleotide changes in thehuman insulin receptor (HIR) gene, in the adenomatouspolyposis coli (APC) gene, and in the mismatch repairgene hMLH1 (11–13). Genomic DNA (gDNA) was isolatedfrom whole peripheral blood, using standard procedures(14). Genomic DNA for microsatellite analysis was ob-tained by microdissection of paired tumor and nondis-eased tissue samples from paraffin-embedded sections, aspreviously described (15).

To test the performance of PCR-labeling protocols, weselected DNA fragments of various lengths (expressed inbasepairs, bp) derived from the following: exons 4 and 17of the HIR gene (fragments of 149 bp and 304 bp,respectively), exon 15 of the APC gene (450 bp), the codingexon of the insulin receptor substrate-1 (IRS-1) gene (402bp), exon 19 of the hMLH1 gene (169 bp), and exon 6 of thehMSH2 gene (233 bp). Dinucleotide repeat loci D3S1611(134 bp) and D5S107 (165 bp) were selected for microsat-ellite analysis (16, 17). Oligonucleotide primers were gen-erated using an Applied Biosystems DNA synthesizermodel 392–05 (Applied Biosystems). The sequence andannealing temperatures are available upon request. Am-plifications were carried out using a Perkin-Elmer 480DNA thermal cycler (Perkin–Elmer).

pcr amplificationsPCRs were performed using gDNA as a template. Forsome applications (as specified below), to verify theamounts and the quality of templates used for labeling,we used a nested PCR protocol. In these cases, before PCRlabeling, genomic DNA was amplified following an es-tablished PCR protocol; the templates were visualized onan ethidium bromide-stained agarose gel; and theamounts, as well as the quality, of the amplified productswere evaluated. Primary unlabeled PCR reactions (10–25mL), contained 100–250 ng of genomic DNA, oligonucle-otide primers (1 mmol/L), 10 mmol/L Tris-HCl (pH 8.3),50 mmol/L KCl, 2.5 mmol/L MgCl2, Taq polymerase(25 000 U/L), and 200 mmol/L dNTP. Cycling conditionsincluded an initial denaturation step at 94 °C for 5 min,followed by 28–32 temperature cycles at 94 °C for 60 s atan appropriate annealing temperature (based on the Tm ofthe primers) for 90 s, and at 72 °C for 90 s. The finalextension step was at 72 °C for 10 min.

effects of dNTP concentration on pcr labelingand pcr yieldAmplified gDNAs (including a 149-bp and a 304-bpfragment of exons 4 and 17 in the HIR gene and a 450-bpfragment of exon 15 in the APC gene) were used astemplates for the evaluation of PCR labeling and PCRyield. PCR-labeling reactions were performed in a finalvolume of 25 mL, containing 2.5 mL of diluted primaryPCR products (final dilution 1:10 000), 25 pmol of eacholigonucleotide primer, 10 mmol/L Tris-HCl (pH 8.3), 50mmol/L KCl, 2.5 mmol/L MgCl2, 25 000 U/L of Taqpolymerase, and a fixed amount (2.5 mCi) of [a32P]dATP(6 000 000 Ci/mol). Cold dATP at six different concentra-tions (i.e., 0.9, 1.8, 3.5, 7.0, 35, and 70 mmol/L) was used toadjust the specific activity of the labeled nucleotide. Theconcentrations of dCTP, dGTP, and dTTP were doublethose of the cold dATP except at the maximum concen-tration (70 mmol/L), in which all dNTPs were equimolar.The concentration of 70 mmol/L was selected because itwas expected to produce an optimal DNA yield, on thebasis of preliminary experiments and data from previouswork (18). The concentration of dATP added as the label(0.016 mmol/L) was negligible, even compared with thelowest concentration of cold dATP added, and was notconsidered for the purpose of this experiment. Sampleswere denatured at 94 °C for 5 min, followed by 28 cyclesof amplification, as described above. Eight microliters ofamplified PCR products were separated by electrophore-sis on a 2% agarose gel containing ethidium bromide. PCRlabeling was evaluated on dried agarose gels, using aBio-Rad Molecular Imager System (model GS-525, Bio-Rad). An estimate of PCR yield was obtained by analyz-ing the fluorescent signal from ethidium bromide-stainedbands, using NIH Image software (Ver. 1.51).

effect of dNTP concentration and pcr dilutionon sscp patternPCR-labeling reactions for SSCP analyses were performedin a final volume of 25 mL, containing 2.5 mL of primaryPCR products diluted 1:1000 (final dilution 1:10 000), 25pmol of each oligonucleotide primer, 10 mmol/L Tris-HCl (pH 8.3), 50 mmol/L KCl, 2.5 mmol/L MgCl2, 25 000U/L of Taq polymerase, 2.5 mCi of [a32P]dATP (6 000 000Ci/mol), and four different cold dATP concentrations(0.9, 1.8, 3.5, and 7.0 mmol/L). Samples were denatured at94 °C for 5 min, followed by 28 PCR cycles. Samples werediluted 1:10 in 950 mL/L formamide buffer and dena-tured at 94 °C for 5 min, followed by rapid cooling on ice.For SSCP analysis, 2 mL aliquots were electrophoresed for8 h through a nondenaturing 6% polyacrylamide gel. Twoconditions were used for SSCP analysis: 4 °C in a buffercontaining 45 mmol/L Tris-borate and 1 mmol/L EDTA,at 10 W constant power; and 24 °C in the same buffer plus50 mL/L glycerol, at 7 W as previously described (19). Toevaluate the effect of different dilutions on SSCP patterns,labeled PCR products obtained at a concentration of 7mmol/L dATP were diluted 1:2.5, 1:5, and 1:10 in 950

1382 Esposito et al.: Optimization of PCR labeling

mL/L formamide buffer, denatured by heating at 94 °Cfor 5 min, cooled on ice, and subjected to SSCP analysis.

effect of pcr cycles on sscp patternThe effect of the number of PCR cycles on the SSCPpattern was evaluated by nested amplification of a 450-bpfragment of the APC gene. PCR labeling was performed ina final volume of 10 mL, containing unlabeled PCRproduct as a template (final dilution 1:10 000), 10 pmol ofeach primer, 7.0 mmol/L of dGTP, dCTP, and dTTP, 3.5mmol/L of cold dATP, 1.5 mCi of [a35S]dATP (1 000 000Ci/mol), 10 mmol/L Tris-HCl (pH 8.3), 50 mmol/L KCl,2.5 mmol/L MgCl2, and 25 000 U/L of Taq polymerase.Samples were subjected to a variable number of PCRcycles (10–28). Labeled PCR products, diluted 1:2.5 informamide buffer, were denatured at 94 °C for 5 min andsubjected to SSCP analysis, as described above. Gels weredried and autoradiographed for 1 day at room tempera-ture. Parallel experiments to evaluate the effect of thenumber of cycles on PCR yield at reduced dNTP concen-trations were conducted using unamplified gDNA as atemplate.

mutational analysis of IRS-1, hMSH2, and hMLH1genes by sscpOptimized labeling conditions described above were usedin mutational screening of IRS-1, hMSH2, and hMLH1genes. In these assays, we used 0.15 Ci/L of [a35S]dATPas the label, 3.5 mmol/L of cold dATP, and 7.0 mmol/L ofthe other nucleotides. PCR labeling was performed eitherby direct amplification of gDNA (IRS-1 gene), or by usinga nested PCR protocol (hMSH2 and hMLH1 genes). Thir-ty-two cycles were used for direct labeling of gDNA (10mg/L), whereas 16 cycles of secondary PCR (1:10 000,final dilution of primary PCR products) were used fornested amplification protocols. Amplified products werediluted 1:2.5 in formamide buffer and used for SSCPanalysis.

microsatellite analysisBecause microdissected specimens constitute a ratherunsteady source of DNA, we conducted microsatelliteanalysis using a nested protocol, consisting of a nonradio-active external PCR (30 cycles) followed by a radioactivesecondary PCR (18–22 cycles), as we have previouslydescribed (15). Reactions were performed in 10 mL, con-taining 10 mmol/L Tris (pH 8.3), 1.5 mmol/L MgCl2, 50mmol/L KCl, 7.0 mmol/L dGTP, dTTP, and dCTP, 3.5mmol/L dATP, 3 mCi of [a35S]dATP, (1 000 000 Ci/mol),10 pmol of each primer, and 0.25 U of Taq polymerase.Radioactive PCR products were mixed with 1 volume offormamide buffer and subjected to denaturing electro-phoresis in 5% polyacrylamide gels containing 8 mol/Lurea. Gels were dried and autoradiographed for 1–4 daysat room temperature.

ResultIn this study, different variables, including the ratio oflabeled vs unlabeled dNTPs, dilution of labeled product,number of PCR cycles, and use of one-step or nestedlabeling protocols were tested to optimize the labelingand visualization of PCR products. Optimized PCR-label-ing conditions were used for SSCP and microsatelliteanalyses.

effect of dNTP concentration on pcr labelingand pcr yieldTo establish the optimal ratio of labeled dATP vs unla-beled dNTPs, we performed preliminary experimentsusing different dATP/dNTP ratios, ranging from 1:2through 1:8. The 1:2 ratio of labeled dATP vs unlabeleddNTPs yielded an optimal labeling, whereas higher ratioscaused a progressive inhibition of PCR amplification(data not shown). A 1:2 ratio was used in all subsequentPCR-labeling experiments, with the exception of testsperformed at a concentration of 70 mmol/L dNTP, wherea 1:1 ratio was used.

To study the kinetics of PCR labeling, we performedamplifications in the presence of a fixed amount of labeleddATP (see Materials and Methods). The specific activitywas adjusted by adding cold dATP (ranging from 0.9mmol/L through 70 mmol/L) in the reaction buffer. Thekinetics of PCR labeling and DNA yield at varying dATPconcentrations were tested by amplifying a 149-bp and a304-bp fragment of the HIR gene, as well as a 450-bpfragment of the APC gene (Fig. 1 and data not shown).When a 149-bp fragment of the HIR gene was used,optimal PCR-labeling intensity was reached at dATPconcentrations between 0.9 and 7.0 mmol/L (Fig. 1). ThePCR products reached the peak labeling intensity at aconcentration of 1.8 mmol/L dATP, with a 9.3-fold in-crease compared with the labeling obtained using 70mmol/L dATP (Fig. 1). On the other hand, the kinetics ofPCR yield showed a gradual decrease in absolute amountof PCR product as a function of the decrease in dATPconcentration from 70 to 0.9 mmol/L. Relative specificactivities of 8.0, 4.2, 2.8, 2.1, 0.3, and 0.1 (average labeling/yield) were achieved by varying the dATP concentrationfrom 0.9 to 70 mmol/L. In fact, at a dATP concentrationcorresponding to the maximum PCR labeling (1.8 mmol/L), the PCR yield decreased by 4.2-fold compared with thePCR yield obtained at 70 mmol/L dATP (Fig. 1). Compa-rable results were obtained using a 304-bp fragment of theHIR gene and a 450-bp fragment of the APC gene astemplates (data not shown).

effect of dNTP concentration and pcr dilutionon sscp patternThe use of limiting dNTP concentrations may influencethe fidelity of DNA replication by the Taq polymerase,thus affecting the reliability of mutational screening meth-ods sensitive to single nucleotide changes. To verifywhether limiting dNTP concentrations may affect the

Clinical Chemistry 44, No. 7, 1998 1383

detection of nucleotide variants by SSCP, we amplified a304-bp PCR-amplified fragment of the HIR gene, usingtwo DNA specimens, one of which contained a singlenucleotide change (Fig. 2A). The SSCP patterns weresimilar, and the unique conformer of the sample bearinga nucleotide substitution was evident at all the differentdATP concentrations within the range corresponding tooptimal labeling. These findings were confirmed in inde-pendent experiments using PCR products of differentsizes (data not shown).

Dilution of labeled DNA in different volumes of dena-turing solution may influence SSCP patterns and couldaffect the detection of nucleotide changes. Therefore, wetested different dilutions (1:10, 1:5, and 1:2.5) of the 304-bpamplified fragment of the HIR gene in formamide buffer(Fig. 2B). As expected, the intensity of the SSCP bands wasmaximal at the lowest dilution (1:2.5). Although the SSCPpattern changed with increasing dilutions, the uniqueSSCP conformer corresponding to the nucleotide substi-tution (indicated by the arrow) was clearly identified at alldilution ratios.

effect of pcr cycles on sscp patternWhen a 1:10 000 dilution of PCR products in a nestedPCR-labeling protocol is used, ;16 cycles of secondary

PCR should be sufficient to reach amplification plateau.Furthermore, the use of limiting dNTP concentrationscould decrease the number of cycles necessary to reachthe plateau by reducing the PCR final yield. We tested theeffect of varying the number (10–28) of secondary PCRcycles on SSCP patterns. In these experiments, PCR label-ing of a 450-bp fragment of the APC gene was obtained,using a nested amplification protocol under limitingdATP concentrations (3.5 mmol/L). A relatively weakemitter, such as [a35S]dATP, was used as the label. Theclearest SSCP patterns were obtained after 12–16 cycles ofsecondary PCR (Fig. 2C). Increasing the number of PCRcycles did not enhance SSCP intensity, whereas the ap-pearance of spurious bands reduced the definition ofSSCP conformers. Therefore, low dNTP concentrationsused in our PCR labeling protocols did not appear todecrease the number of cycles predicted to be necessary toreach a PCR plateau starting from a diluted PCR. We alsocompared the DNA yields obtained at reduced dNTPconcentrations, using as a template either diluted PCRproducts (1:10 000, final dilution) or gDNA. In agreementwith what observed in SSCP experiments, when amplifiedDNA was used as a template, PCR plateau was achievedwith 16 cycles of secondary PCR. When unamplified

Fig. 1. Effect of dNTP concentration on PCR labeling and PCR yield.A 149-bp segment of the HIR gene was amplified and labeled with [a32P]dATP, as described in Materials and Methods. Shown are the kinetics of PCR labeling (F) andPCR yield (M) as a function of cold dATP concentration. Aliquots of amplified DNA (8 mL) were separated by electrophoresis on a 2% agarose gel containing ethidiumbromide, and the amount of radiolabeled DNA in the corresponding bands was quantitated using a Bio-Rad Molecular Imager System. PCR yield was evaluated analyzingthe fluorescent signals from ethidium bromide-stained bands by the NIH Image software. The data shown are the means of duplicate determinations from arepresentative of three independent experiments.

1384 Esposito et al.: Optimization of PCR labeling

gDNA was used, more cycles (30–34) were necessary tocompensate for the lower number of target copies, bothunder limiting (3.5 mmol/L dATP) and under nonlimiting(70 mmol/L dATP) dNTP concentrations (data notshown).

mutational analysis of IRS-1, hMSH2, and hMLH1genes by sscpTo validate the optimized PCR-labeling conditions inmutational analysis applications, we performed PCR-SSCP of the IRS-1, hMSH2, and hMLH1 genes, usingreduced dNTP concentrations and [a35S]dATP as thelabel.

Analysis of the IRS-1 gene was performed using un-amplified gDNA samples as templates. The PCR was runfor 32 PCR cycles, and 35S-labeled products were sub-jected to SSCP analysis. A unique SSCP pattern (Fig. 3A,lane 4) was evident after overnight film exposure andcorresponded to a novel silent polymorphism of the gene.Analyses of hMSH2 and hMLH1 genes in HNPCC patientswere performed using diluted PCR products as templates.Unique SSCP conformers were evident, after overnightfilm exposure, in the analyses of both the hMSH2 (Fig. 3B,lane 3) and the hMLH1 (Fig. 3C, lane 4) genes. The uniqueconformer of the hMSH2 gene corresponds to a novelnucleotide substitution.

microsatellite analysisOptimized PCR-labeling conditions were applied to mic-rosatellite analysis. PCR labeling was performed usinggDNA obtained from microdissected paraffin-embeddedpaired tumor and nondiseased tissue samples of HNPCCpatients, using [a35S]dATP as the label. Microsatellitetyping obtained using [a35S]dATP as the label (3 mCi)yielded results comparable with those obtained in an

overnight exposure using a lower amount (0.5 mCi) of[a32P]dATP (Fig. 4 and data not shown). Examples oftumor-associated microsatellite alterations at loci D5S107and D3S1611, corresponding to expansions/contractionsof repeats (patients 1 and 2) and allelic loss (patient 3) areshown in Fig. 4.

DiscussionThere is remarkable variability in the conditions of PCRlabeling used by different laboratories (1, 2, 4–10). Theuse of disparate PCR-labeling protocols, despite the wideapplications of this procedure, points out the need forstandardized conditions. In the present study, we ana-lyzed conditions that may affect the efficiency of PCRlabeling and the detection of PCR-labeled products. These

Fig. 2. Effect of dNTP concentration, PCR dilution,and PCR cycles on SSCP pattern.(A) To evaluate the effect of dNTP concentrations on SSCPpatterns, we amplified a 304-bp fragment of the HIR gene inthe presence of [a32P]dATP and cold dATP (0.9–7.0 mmol/L). The samples analyzed in lanes 2, 4, 6, and 8 derive froma patient with a CAC3CAT silent polymorphism at codon1058 of the HIR gene. The unique SSCP band correspond-ing to this polymorphism is indicated by an arrowhead. Theexperiment shown is representative of similar experimentsperformed using PCR fragments of different lengths (149and 450 bp). (B) To study the effect of PCR dilution on SSCPpattern, the samples in lanes 1 and 2 of panel A werediluted 1:2.5 (lanes 1 and 2), 1:5 (lanes 3 and 4), and 1:10(lanes 5 and 6) in a solution containing 950 mL/L deionizedformamide, 1 g/L bromphenol blue, and 1 g/L xylenecyanol. The samples were denatured at 95 °C for 5 min andelectrophoresed for 8 h through a nondenaturing 6% poly-acrylamide gel at 10 W constant power. (C) To verify theeffect of PCR cycles on SSCP pattern, we amplified a450-bp fragment of the APC gene. Four amplified sampleswere diluted 1:10 0000 and subjected to PCR labeling forthe indicated number of secondary PCR cycles (10–28),using [a35S]dATP as the label. A unique SSCP conformer ofthe APC gene is evident in the second lane of each group(arrowheads). The SSCP conformer of the APC gene corre-sponded to a 5-bp frameshift deletion at codon 1309 ofexon 15.

Fig. 3. Mutational analysis of IRS-1, hMSH2, and hMLH1 genes bySSCP.(A) SSCP pattern of the IRS-1 gene obtained using genomic DNA as the templatefor PCR labeling. The unique SSCP conformer in lane 4 (arrowhead) correspondsto a novel A 3 C transversion at nucleotide 3489 of the IRS-1 gene (with nochange of Pro 1163). (B and C) SSCP patterns of hMSH2 and hMLH1 genes,respectively, obtained using amplified DNA as template for PCR labeling. SSCPconformers are detected in lane 3 of (B) and in lane 4 of (C) (arrowheads). Theunique SSCP conformers in lane 3 of (B) and in lane 4 of (C) corresponded,respectively, to a novel C3 T transition at nucleotide 1052 of the hMSH2 gene(with no change of Ala 328) and to a TTC deletion in the 39 untranslated regionof the hMLH1 gene. All autoradiographic images shown in this figure wereobtained after overnight exposure.

Clinical Chemistry 44, No. 7, 1998 1385

conditions were also validated in mutational analysis bySSCP and microsatellite typing applications.

Maximal PCR-labeling intensity was reached at adATP concentration of 1.8 mmol/L. This concentrationcorresponded to an optimal ratio between an increase inspecific activity and a decrease in DNA yield. Whendifferent DNA templates within the size range usuallyused for SSCP and microsatellite analysis (149–450 bp)were used, the optimal dNTP concentration for labelingwas not affected by the length of the PCR fragment. In therange of dATP concentrations between 0.9 and 7.0mmol/L, there was a modest reduction (;25%) in DNAlabeling, compared with the peak obtained at 1.8 mmol/L.Considering that limiting dNTP concentrations may in-hibit the amplification of critical PCR templates and thatincreasing the concentration of dATP up to 7 mmol/L hasmodest effects on the intensity of the signal, for mostapplications it may be advisable to use conditions lessextreme than 1.8 mmol/L. We also considered nucleotideimbalance as a variable affecting PCR-labeling. Someprotocols, in the attempt to increase the specific activity ofPCR labeling without affecting PCR yield, limited onlythe concentration of the cold nucleotide corresponding tothe one used for labeling (4, 8–10). In these protocols, theconcentration of the other nucleotides was within therange that is expected to give optimal PCR yield withimbalance between labeled vs unlabeled nucleotides of upto 100-fold (4). However, it has been shown previouslythat excessive nucleotide imbalance may force nucleotidemisincorporation and induce premature stoppage of thePCR reaction (20). In our hands, there was no increase inlabeling intensity when nucleotide imbalances .1:2 wereused. In fact, in agreement with the work by Innis et al.(20), greater unbalances appeared to inhibit PCR. There-

fore, considering also the possibility that polymeraseerrors may interfere with mutational analysis, the use ofexcessive nucleotide imbalances does not appear to beadvantageous and should be discouraged.

Mutational analysis by SSCP was used to validateoptimized PCR-labeling protocols. We first evaluatedwhether limiting nucleotide concentrations could affectSSCP patterns, for example, by increasing the rate of Taqpolymerase misincorporation. SSCP patterns obtained atfour different dATP concentrations produced similar re-sults, and single nucleotide changes were also evident atthe lowest dNTP concentration used in our labelingprotocols. Therefore, the reduced dNTP concentrationsdid not interfere with SSCP. Another variable consideredwas the influence of DNA templates. Clear SSCP patternswere obtained using DNA templates of different sizesderived from different genes, including APC, IRS-1, HIR,hMLH1, and hMSH2. In addition, SSCP patterns obtainedusing gDNA directly or amplified DNA as templates wereequivalent and were visible after an overnight exposure.However, to obtain maximum DNA labeling, the numberof PCR cycles had to be proportionally increased whengDNA rather than amplified products was used as atemplate. Of note is that the number of PCR cyclesnecessary to obtain clear SSCP patterns was influencednot only by the number of copies of template (i.e., gDNAvs diluted PCR) but also by the particular gene fragmentamplified. In fact, some samples proved more difficult toamplify and required a greater number of cycles (up to 35cycles in one-step labeling protocols and up to 28 cycles ofsecondary PCR in nested protocols) to obtain optimalsignals. However, the number of PCR cycles should bekept as low as possible, because exceeding the optimalnumber reduced the definition of SSCP patterns.

With regard to the influence of the dilution of labeledPCR product on SSCP pattern, the original protocol byOrita et al. (5) used a 1:200 final dilution of labeledproducts in denaturing solution, which implied the use oflarge amounts of labeled nucleotide to obtain visiblesignals (;10-fold those used in our study). In our hands,dilutions of labeled products ranging between 1:10 and1:2.5 had a considerable influence on SSCP patterns;however, SSCP bands as well as anomalous conformerswere, if anything, more evident at lower dilutions. A lowdilution ratio was particularly important to obtain readilyvisible signals with 35S-labeled reactions.

It is important to note that, using optimized conditions,we were able to visualize all the nucleotide changes thatwere already known to be present in our samples. Inaddition, the use of optimized PCR-labeling conditionswas further validated by the detection of novel nucleotidevariants in the IRS-1 and hMSH2 genes.

Microsatellite analysis is widely used in linkage studies(4), in forensic medicine investigations (21), and in thegenotyping of tumors (22). Mononucleotide or dinucle-otide repeats are often used in microsatellite analysis, andelectrophoresis of these markers under denaturing condi-

Fig. 4. Microsatellite analysis of paired nondiseased and tumorsamples from HNPCC patients.Examples of patterns obtained by microsatellite analysis of paired nondiseased(N) and tumor (T) samples derived from HNPCC patients. Genomic DNA wasobtained by microdissection of paraffin-embedded tissues. The microsatellitemarkers D5S107 and D3S1611 were analyzed by PCR labeling, using [35S-a]dATP as the label, and the amplified products were separated on 6%polyacrylamide denaturing gels. Autoradiographic films were exposed overnightto 4 days. Tumor samples in lanes 2 and 4 present replication errors consistingin expansion (lane 2) and contraction (lane 4) of microsatellite alleles (arrow-heads). A banding pattern suggesting the loss of one microsatellite allele in thetumor sample (arrowhead) is shown in lane 6. No microsatellite alteration isobserved in the tumor sample analyzed in lane 8.

1386 Esposito et al.: Optimization of PCR labeling

tion results in the appearance of multiple bands for eachof the polymorphic alleles (23). With currently usedPCR-labeling protocols, the distribution of the signal inmultiple bands requires the use of a strong emitter, suchas 32P, to obtain visible signals in reasonable exposuretimes. We tested whether conditions that had been opti-mized for SSCP could be efficiently used in microsatellitetyping. Under such conditions, we obtained clear typingof microsatellite alleles, which also allowed the detectionof tumor-associated size shifts, which are typically ob-served in samples from HNPCC patients. In addition, theimproved signal efficiency allowed the use of [a35S]dATPas the label. However, to compensate for the distributionof the signal in multiple bands, the concentrations ofsulfur-radionucleotide were twice those used in the cor-responding PCR-SSCP labeling protocol.

In conclusion, we analyzed some of the conditions thatinfluence PCR labeling and detection in the attempt tooptimize protocols for PCR labeling. Optimized conditionswere validated in SSCP mutational screening and microsat-ellite analysis. This study should provide a useful referencefor investigators who use PCR labeling. In fact, optimizedconditions could contribute to cost reductions, decreasedautoradiographic exposure times, improved quality of sig-nals, and may reduce the amounts of radioactivity handledand/or lead to the use of relatively weak emitters. In thisrespect, the improved signal efficiency obtained with opti-mized protocols allowed the use of [a35S]dATP as the labelin SSCP and microsatellite analyses in place of the com-monly used 32P-labeled nucleotides. Although we usedradioactive tracers, the optimized labeling conditions de-scribed in the present study should also prove useful fornonradioactive PCR-labeling procedures.

This work was supported by Associazione Italiana RicercaCancro (AIRC) Special Project “Hereditary ColorectalTumors”; by Consiglio Nazionale Ricerche (CNR) No.96.00529.29.PF39; by Telethon Italy (No. E.295); and byMinistero Univesitá Ricerca Scientifica Technologica (40%and 60% grants to A.C. and P.B.).

References1. Schowalter DB, Sommer SS. The generation of radiolabeled DNA

and RNA probes with polymerase chain reaction. Anal Biochem1989;177:90–4.

2. Mertz ML, Rashtchian A. Nucleotide imbalance and polymerasechain reaction: effects on DNA amplification and synthesis of highspecific activity radiolabeled DNA probes. Anal Biochem 1994;221:160–5.

3. Cotton RGH. Current methods of mutation detection. Mutat Res1993;285:125–44.

4. Peltomaki P, Aaltonen LA, Sistonen P, Pylkkanen L, Mecklin JP,Jarvinen H, et al. Genetic mapping of a locus predisposing tohuman colorectal cancer. Science 1993;260:810–2.

5. Orita M, Suzuki Y, Sekiya T, Hayashi K. Rapid and sensitivedetection of point mutations and DNA polymorphisms using thepolymerase chain reaction. Genomics 1989;5:874–9.

6. Jensen HJ, Jensen LG, Hansen PS, Faergeman O, Gregersen N.High sensitivity of the single-strand conformation polymorphismmethod for detecting sequence variations in the low-densitylipoprotein receptor gene validated by DNA sequencing. Clin Chem1996;42:1140–6.

7. Beutler E, Westwood B, van Zwieten R, Roos D. G3T transition atcDNA nt 110 (K37Q) in the PKLR (pyruvate kinase) gene is themolecular basis of a case of hereditary increase of red blood cellATP. Hum Mutat 1997;9:282–5.

8. Au KS, Rodriguez JA. Rodriguez Jr E, Dobyns WB, Delgado MR,Northrup H. Mutations and polymorphisms in the tuberous scle-rosis complex gene on chromosome 16. Hum Mutat 1997;9:23–9.

9. Berx G, Nollet F, Strumane K, van Roy F. An efficient and reliablemultiplex PCR-SSCP mutation analysis test applied to the humanE-cadherin gene. Hum Mutat 1997;9:567–74.

10. Visser M, Bras J, Sijmons C, Devilee P, Wijnaedts LCD, van derLinden JC, et al. Microsatellite instability in childhood rhabdomyo-sarcoma in locus specific and correlates with fractional allelicloss. Proc Natl Acad Sci U S A 1996;93:9172–6.

11. Cama A, Sierra ML, Quon MJ, Ottini L, Gorden P, Taylor SI.Substitution of glutamic acid for alanine 1135 in the putative“catalytic loop” of the tyrosine kinase domain of the humaninsulin receptor. A mutation that impairs proteolytic processinginto subunits and inhibits receptor tyrosine kinase activity. J BiolChem 1993;268:8060–9.

12. Palmirotta R, Curia MC, Esposito DL, Valanzano R, Messerini L,Ficari F, et al. Novel mutation and inactivation of both alleles ofthe APC gene and genomic instability in desmoid tumors. HumMolec Genet 1995;4:1979–81.

13. Palmirotta R, Verı MC, Curia MC, Casale V, Fracasso P, StiglianoV, et al. Novel allele of the hMLH1 gene bearing a TTC deletion inthe 39 untranslated region. Int J Oncol 1996;9:701–3.

14. Sambrook J, Fritsch EF, Maniatis ET. Molecular cloning: a labora-tory manual. Cold Spring Harbor, NY: Cold Spring Harbor Labora-tory Press, 1989:9.16–9.19.

15. Ottini L, Esposito DL, Richetta A, Carlesimo M, Palmirotta R, VerıMC, et al. Alterations of microsatellites in neurofibromas of vonRecklinghausen’s disease. Cancer Res 1995;55:5677–80.

16. Gyapay G, Morissette J, Vignal A, Dib C, Fizames C, Millasseau P,et al. The 1993–1994 Genethon human genetic linkage map. NatGenet 1994;7:246–339.

17. Weber JL, Kwitek AE, May PE. Dinucleotide repeat polymorphismsat the D5S107, D5S108, D5S111, D5S117 and D5S118 loci.Nucleic Acids Res 1990;18:4035.

18. Saiki RK. The design and optimization of the PCR. In: Erlich HA,ed. PCR technology. New York: W.H. Freeman & Co., 1992:7–16.

19. Cama A, Palmirotta R, Esposito DL, Curia MC, Ranieri A, Ficari F,et al. A novel deletion in exon 15 of the Adenomatous PolyposisColi gene in an Italian kindred. Hum Mutat 1994;3:301–4.

20. Innis MA, Myambo KB, Gelfald DH, Brown MAD. DNA sequencingwith Thermus aquaticus DNA polymerase and direct sequencing ofpolymerase chain reaction-amplified DNA. Proc Natl Acad Sci U SA 1988;85:9436–40.

21. Hammond HA, Jin L, Zhong Y, Caskey CT, Chakraborty R. Evalua-tion of 13 short tandem repeat loci for use in personal identifica-tion applications. Am J Hum Genet 1994;55:175–89.

22. Ottini L, Palli D, Falchetti M, D’Amico C, Amorosi A, Saieva C, et al.Microsatellite instability in gastric cancer is associated with tumorlocation and family history in a high risk population from Tuscany.Cancer Res 1997;57:4523–9.

23. Hauge XY, Litt M. A study of the origin of ’shadow bands’ seenwhen typing dinucleotide repeat polymorphisms by the PCR. HumMol Genet 1993;2:411–5.

Clinical Chemistry 44, No. 7, 1998 1387