3856 Electrophoresis 2006, 27, 3856–3863

Marek Minarik1 Research Article

Lucie Benesova1
Lucie Fantova1
Jiri Horacek2 Parallel optimization and genotyping of
Jiri Heracek3
Anu Loukola4 multiple single-nucleotide polymorphism
1
Genomac International, markers by sample pooling approach using
Laboratory for Molecular Genetics,
Prague, Czech Republic cycling-gradient CE with multiple injections
2 rd
3 Faculty of Medicine,
Charles University, Increasing importance of single-nucleotide polymorphisms (SNPs) in determination of
Center for Neuropsychiatric Studies,
disease susceptibility or in prediction of therapy response brings attention of many
Prague, Czech Republic
3
Faculty Hospital, molecular diagnostic laboratories to simple and low-cost SNP genotyping methodol-
Department of Urology, ogies. We have recently introduced a mutation detection technique based on analysis
Prague, Czech Republic of homo- and heteroduplex PCR fragments resolved in cycling temperature gradient
4
National Public Health Institute,
conditions on a conventional multicapillary-array DNA sequencer. The main advan-
Department of Molecular Medicine,
Helsinki, Finland tage of this technique is in its simplicity with no requirement for sample cleanup prior
to the analysis. In this report we present a practical application of the technology for
genotyping of SNP markers in two separate clinical projects resulting in a combined
Received May 11, 2006 set of 44 markers screened in over 500 patients. Initially, a design of PCR primers and
Revised June 9, 2006
conditions was performed for each SNP marker. Then, optimization of CE running
Accepted June 23, 2006
conditions (limited just to the proper selection of temperature cycling) was performed
on pools of 20 DNA samples to increase the probability of having each of the two
allele types represented in the sample. After selecting the optimum conditions,
screening of markers in patients was performed using a multiple-injection approach
for further acceleration of the sample throughput. The rate of successful optimization
of experimental conditions without any pre-selection based on the SNP sequence or
melting characteristics was 80% from the initial SNP marker candidates. By studying
the failed markers, we attempt to identify critical factors enabling successful typing.
The presented technique is very useful for low to medium sized SNP genotyping
projects mostly applied in pharmacogenomic research as well as in clinical diag-
nostics. The main advantages include low cost, simple setup and validation of SNP
markers.

Keywords: Cycling-gradient CE / Multiple injection / Prostate cancer / Schizophrenia /

SNP DOI 10.1002/elps.200600289

1 Introduction recent decade has seen a tremendous speed of devel-

opment of various methodologies and instrumentation for
Genotyping of specific single-nucleotide polymorphisms SNP genotyping [1]. Among other parameters, increase in
(SNPs) is becoming increasingly important in the diag- sample throughput has been the driving factor behind the
nosis of genetic risk factors for inherited disease suscep- new technology development. This is certainly the case in
tibility as well as in prediction of treatment response. The whole-genome SNP typing (and haplotyping) studies,
where a vast number of potential marker candidates is
Correspondence: Dr. Marek Minarik, Genomac International Ltd, screened in each sample. For such demanding applica-
Laboratory for molecular genetics, Bavorska 856, CZ-15541 Prague, tions, dedicated systems using massive parallel hybridi-
Czech Republic zation and direct readout are employed either in the clas-
E-mail: [email protected]
sic microarray format [2], utilizing functionalized
Fax: 1420-224-458-021
microbeads [3] or picotiter plates [4]. With resulting hun-
Abbreviations: CGCE, cycling-gradient CE; Tm, melting tempera- dreds of thousands of genotypes obtained in each
ture experiment, the ultrahigh-throughput platforms inherently

© 2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com

Electrophoresis 2006, 27, 3856–3863 CE and CEC 3857

require the ability to process enormous amounts of gen- In 2003 we have originally introduced the cycling gra-
erated data. Therefore, such systems currently find their dient CE (CGCE), sometimes also referred to as the
main use in pharmaceutical industry and large research cycling temperature CE, as a simplified CE version of
centers capable to absorb the associated equipment classic temperature gradient gel electrophoresis with
cost. For common clinical diagnostics, disease risk increased sample throughput [15]. The method is
assessment or therapy response prediction, the required based on electrophoretic separation of homoduplex
throughput is usually lower as the number of individual and heteroduplex DNA fragment forms under partial
polymorphisms with confirmed functional relevance and melting conditions in denaturing gel matrix. The two
clinical outcome is still relatively low [5]. This opens a strands in each fragment are held together on one end
possibility for utilization of low to medium throughput by a 40 bp artificial high-temperature melting domain
technologies exhibiting sufficient capacity to deliver up to (GC clamp) attached to a 100–150 bp target sequence.
hundreds, rather than thousands, of assays per experi- The melting of the target sequence is achieved by
ment. applying periodical cycles of temporal temperature
gradients. Similar to the other methods based on par-
Technologies for low- to medium-throughput SNP geno- tial denaturation, the CGCE/cycling temperature CE
typing often employ tools and protocols commonly technique exhibits superior resolution allowing to
accepted in other molecular biology applications such as separate wild-type and mutant homoduplex and het-
DNA sequence analysis or detection of somatic muta- eroduplex forms for any mutant appearing within the
tions. Experimental setup is usually based on existing target sequence. The method has already been routi-
instrumentation such as standard PAGE, UV/Vis or fluo- nely applied to detect unknown somatic mutations
rescence spectrometers, PCR thermocyclers with or from cancer tissue [16, 17].
without real-time quantification capability, etc. Some of
the techniques are based on DNA electromigration and In this work, we present further validation of the technique
performed with the help of CE as the separation platform. by optimization of experimental conditions with sub-
Since the attainment of the human genome, CE instru- sequent screening of indiscriminate sets of SNP markers
mentation has been routinely used in molecular biology. in larger patient cohorts. Following optimization and
Capillary electrophoretic systems with 1, 8, 16 or 96 cap- elimination of unsuccessful markers from an initial group
illary arrays became routine inventory in many labora- of marker candidates, we present screening of 13 SNPs in
tories. Over the past years, their original mainstream use a homogeneous population of 85 patients diagnosed with
in DNA sequencing and microsatellite genotyping (mainly schizophrenia and another 31 SNPs in a population of 487
for human identification and paternity testing) has gradu- prostate cancer patients and controls. In these two clin-
ally been extended to other applications, most impor- ical projects we demonstrate the utilization of sample
tantly in the area of discovery and screening of somatic pooling in parallel assay optimization of multiple SNP
mutations and typing of SNPs. Various adaptations of markers [18] and utilization of multiple injection to
common gel electrophoresis methods and applications increase sample throughput and to shorten the overall
previously developed on home-built CE instruments have analysis times.
thus become routine on commercial systems. The most
popular CE applications, other than sequencing or
microsatellite genotyping, utilize fragment sizing by RFLP
[6], amplified fragment length polymorphism (AFLP) [7], or 2 Materials and methods
following multiplexed ligation [8, 9] and allele separation
by SSCP [10] or heteroduplex analysis [11]. A special 2.1 Patients and SNP markers
group of methods employs separation due to partial DNA
melting in CE-adapted versions of denaturing gradient gel The present work includes data from two separate pro-
electrophoresis techniques [12–14]. The universal appli- jects aiming at inherited factors of disease predisposition.
cability of CE instrumentation and simple transition Blood samples from patients diagnosed with schizo-
among sequencing, genotyping and mutation detection phrenia were obtained from the Center for Neu-
applications is appealing to medium throughput environ- ropsychiatric Studies at the Prague Psychiatric Center,
ments such as core laboratories or diagnostic centers. and blood samples from a group of prostate cancer
The CE-based SNP typing methods usually exhibit a patients and controls with clinically confirmed benign
higher cost per single genotype in comparison to the prostatic hyperplasia were obtained from the Urological
previously mentioned ultrahigh capacity platforms; how- clinic, Faculty Hospital in Prague. Both projects were
ever, the main advantage is in their immediate availability approved by the ethical committee of the 3rd Faculty of
and the low setup cost. Medicine, Charles University, Prague.

© 2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com

3858 M. Minarik et al. Electrophoresis 2006, 27, 3856–3863

SNP markers studied in genetic predisposition of schi- 3 Results and discussion

zophrenia covered the following genes: COMT, BDNF,
RGS4, DTNBP, NRG1, DRD2, DRD3, HT2A, SERT and The purpose of this work was to evaluate practical appli-
MAG. For the prostate cancer study we have focused cation of the CGCE technique in a typical clinical project.
on selected polymorphisms in the Androgene Receptor Unlike in other reports in which data is often presented
gene as well as three other genes involved in testos- only in a limited subset of optimized markers, the results
terone synthetic pathway: CYP17A1, SRD5A2 and presented here document real test of the approach on
CYP3A4. considerable sets of SNPs examined in large groups of
patients. We choose to proceed methodically with no bias
based on intermediate results. The processes of running
2.2 DNA isolation and PCR amplification the experiments and data evaluation were separated; the
person executing the CE runs was blinded to the resulting
Genomic DNA was isolated from patient blood samples data.
using the GMC-ISOBL extraction kit (Genomac Interna-
tional, Prague, Czech Republic). All PCR reactions were Both clinical projects reported here were directed at dis-
performed using standard Taq polymerase (Qiagen, ease predisposition. Sequence information on all SNP
GmbH, Hilden, Germany), PCR primers were obtained markers was first obtained by searching literature and
from VBC-Genomics, GmbH (Vienna, Austria). Primers public databases. Using the sequence information sur-
with GC-clamp extension were within 60–65 bp range, rounding the SNP position, the theoretical Tm of the frag-
regular primers were within the 20–22 bp range. Analyzed ment was calculated. To reflect the additional denaturing
PCR fragments were labeled with fluorescein. Standard due to urea in the CE gel matrix, an experimentally deter-
PCR conditions with typical 35 cycles were applied. The mined constant of 217C was deducted from the theoreti-
PCR program was always concluded with a heteroduplex cal value. For each marker a target PCR segment was
formation step consisting of denaturation at 957C for designed with an ideal length of 100–120 bp (not including
5 min, annealing at 657C for 30 min and slow cooling to the primer sites and GC-extension). The typical workflow
47C. is presented in Fig. 1. The optimization phase consisted
of two steps: allele detection and subsequent validation.
Theoretical melting temperatures (Tm) were calculated In the first step, 20 randomly selected patient DNA sam-
using WinMelt software (Medprobe, Oslo, Norway). In ples were mixed at equivalent concentrations. A set of
order to account for the effect of chemical denaturant PCR reactions, specific to each SNP marker, was then
present in the CE gel matrix, 217C was deducted from the performed on the pooled sample [20]. The resulting PCR
value obtained from WinMelt [19]. products were analyzed by CGCE using three tempera-
ture gradient programs consisting of 2 min cycles at 627C
around Tm21, Tm and Tm11, with Tm specific for each in-
2.3 Cycling gradient CE dividual marker. Naturally, different SNP markers exhibit-
ing the same Tm could be examined in parallel within the
All CGCE experiments were performed on a standard same 96-well plate, reducing the overall number of test
MegaBACE™ 1000 96-capillary DNA analysis system runs required. After the first round of test runs, most
(GE Healthcare Bio-Sciences, Piscataway, USA) using markers already revealed characteristic temperature-de-
GMC-MB running buffer and GMC-MTRX denaturing pendent shifts when comparing positions of homo- and
separation polymer (Genomac International, Prague). hetero-duplex peaks at different temperature gradient
Alternatively a standard long-read sequencing linear settings. In cases where the homo- and heteroduplex
polyacrylamide (LPA) matrix (GE Healthcare Bio-Sci- forms could not be identified, a set of the next 20 DNA
ences) was used in some CGCE experiments resulting samples was pooled, amplified and subjected to the
in identical resolution. During the optimization phase, same CGCE temperature gradient programs. Markers
temperature within the capillary array was monitored that did not show homo- and heteroduplex profiles
by a digital thermometer equipped with an external Pt- after the first or second round (by which a total of 40
sensor. Cycling temperature profiles were pro- random samples has already been investigated) were
grammed by MegaBACE temperature gradient creator assigned as failed. Since the migration order of individual
(MBTG) version 2.0. (Genomac). For the screening duplex fragments given by the melting energies of
phase, the MegaBACE was equipped with a robotic homoduplexes can be directly translated into the base
plate autoloader Caddy™ 1000 (Watrex Praha, Prague, types (more stable G/C eluting initially and less stable
Czech Republic), to allow for unattended automated A/T eluting later), it was possible at this point to directly
overnight operation. assign genotypes. However, in order to exclude the

© 2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com

Electrophoresis 2006, 27, 3856–3863 CE and CEC 3859

Figure 1. Scheme of SNP opti-

mization and validation work-
flow.

possibility of a false positive due to the presence of candidates analyzed in the present work is summarized in
another SNP at a different position within the same target Table 1 with detailed information on marker type, allele
sequence, identity of each marker was confirmed by se- frequency and resulting CGCE quality.
quencing. Since sequencing only reveals a minor allele if
present in approximately 30–50% ratio, it was performed After the above described optimization phase was com-
on PCR pools of at most five samples [21]. If no hetero- pleted, SNP markers from A, B and C quality groups could
zygous position was detected a new pool of another five finally be evaluated in large numbers of patient samples.
samples was taken. During screening, each sample was first analyzed to dis-
tinguish homozygous status (only a single homoduplex
From the initial group of 55 SNP candidates, the optimi- peak in the electropherogram), heterozygous status
zation phase produced positive results (unambiguous (characteristic pattern of homo- and heteroduplex peaks)
identification of all duplex forms) in 44 markers. This or to uncover failure of the PCR reaction (no peaks). If a
represents a success rate of 80%. All SNPs could be homozygous status was detected, the sample was sub-
divided into four groups according to the quality of sequently injected following heteroduplex formation with
resulting peak profiles (see Fig. 2). The first group includ- a PCR product of a known homozygous control. If the
ed markers that showed clearly separated peaks for all second analysis again resulted in a single homoduplex
duplex forms and due to the distinct patterns their scoring peak, the genotype of the tested sample was the same as
could be done automatically by the software (Fig. 2A). A the homozygous control. On the other hand, if the second
second group included markers whose heterozygous analysis produced a characteristic heteroduplex pattern,
profiles did not always reveal all four duplex forms, but then the inspected sample was assigned genotype
where the separated alleles could still be assigned with opposite to the control.
minimal intervention (Fig. 2B). The last group contained
markers showing irregular profiles with asymmetrical In order to speed up the screening phase of the project,
peaks or additional system peaks (Fig. 2C); however, we decided to employ a multiple-injection technique. This
based on clear differences in peak patterns between technique, originally used to increase sample throughput
homozygous and heterozygous control, genotypes could of MTHFR mutation screening [22], takes advantage of
still be obtained after manual inspection and careful the fact that during CE analysis, separating analytes
comparison of individual sample profiles. The remaining usually occupy only a small portion of the entire capillary
markers did not produce any conclusive peak patterns length. Consequently, the remaining open distance allows
and therefore did not pass the CGCE test (Fig. 2D). Out of injecting a fresh sample mixture even before the separa-
the 44 SNP markers passing the CGCE test, 24 (55%) tion of previously injected analytes is completed. Frag-
were assigned as top quality, 12 (27%) as medium quality, ments from different injections are thus separated within
and 8 (18%) as low quality. An overview of all SNP marker individual non-overlapping regions in the capillary. We

© 2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com

3860 M. Minarik et al. Electrophoresis 2006, 27, 3856–3863

Figure 2. Examples of quality

level types observed for CGCE
separation of various SNP
markers. (A) Top-quality mark-
ers suitable to automated scor-
ing by software, (B) medium-
quality markers capable of
identification with minimum
intervention, (C) low-quality
markers requiring manual com-
parison of patterns, (D) failed
markers which did not lead any
information useable to allele
scoring.

have previously illustrated application of this approach in injections could be performed even while the previously
CGCE for the detection of somatic mutations, where the introduced samples passed though the detector with no
repeated injections were carried out in synchronization distortion of the resulting electropherograms. The number
with the temperature gradient cycles. The early as well as of possible injections performed within one analysis run
the late injected samples were thus subjected to the same was limited by the longevity of the separation gel matrix
temperature conditions [15]. In this work we have further and the buffering capacity of the running buffer. In addi-
refined this approach. Prior to each subsequent injection, tion, periodical voltage interruptions, during which a
the CE voltage was interrupted and the data collection longitudinal diffusion in the gel occurs, also affected the
was paused. Following the injection, data collection and overall resolution by peak-broadening, especially if large
the CE voltage were renewed. As a result, subsequent numbers of injections were performed within a single run.

© 2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com

Electrophoresis 2006, 27, 3856–3863 CE and CEC 3861

Table 1. Overview of all SNP marker candidates evaluated in this study

Gene SNP ID Type Major allele frequency Tm (7C) CGCE quality

Prostate cancer project
CYP17A1 rs2486758 C/T T 0.818 71 HIGH
rs4919683 A/C C 0.589 74 MEDIUM
rs3740397 C/G C 0.614 75 MEDIUM
rs4919687 A/G G 0.796 76 HIGH
rs743575 A/C A 0.789 75.4 HIGH
rs3824755 C/G G 0.732 76 HIGH
rs284847 C/T Not available 76 HIGH
rs743572 A/G A 0.562 77.4 MEDIUM
rs6162 A/G G 0.608 76.8 MEDIUM
rs6163 A/C C 0.562 73 LOW
ARa) rs6152 A/G G 0.762 78.5 2
rs1337076 G/T G 1.000 74 2
SNP30 C/T Not available 76.5 2
SRD5A2 rs2300697 C/T T 0.622 65 MEDIUM
rs4952219 A/G G 0.598 62 HIGH
rs12470143 C/T C 0.533 69.2 HIGH
rs2208532 A/G A 0.594 66.4 MEDIUM
rs2268796 A/G G 0.542 69.8 LOW
rs2300703 C/T C 0.538 64.4 LOW
rs4952220 A/C A 0.541 64.6 MEDIUM
rs612224 G/T T 0.714 69 MEDIUM
rs1042578 A/G G 0.896 65 LOW
SNP4 C/T Not available 66.5 HIGH
rs413836 A/G G 0.750 2 2
rs1475054 C/G C 0.740 70 HIGH
rs12712317 A/C Not available 70.5 HIGH
SNP17 A/G Not available 73.4 HIGH
rs676033 A/G Not available 74.8 LOW
rs11403363 T/2 Not available 2 2
CYP3A4 rs12333983 A/T T 0.599 2 2
rs2242480 A/G G 0.639 72 2
rs4646437 C/T C 0.679 75.5 HIGH
SNP15 C/G Not available 68 2
rs28988605 T/2 T 0.663 2 2
rs1851426 C/T C 0.869 74 HIGH
Schizophrenia project
COMT rs737865 C/T T 0.792 73.2 MEDIUM
rs4680 A/G A 0.517 78 MEDIUM
rs165599 A/G Not available 78.5 LOW
BDNF rs6265 A/G G 0.825 73.5 HIGH
RGS4 rs10917670 C/T Not available 68.2 HIGH
rs951436 A/C A 0.575 69.2 HIGH
rs951439 C/T C 0.517 71.4 HIGH
rs2661319 C/T Not available 75 HIGH
rs10917672 C/T Not available 75 HIGH
DTNBP rs2619538 A/T T 0.610 72.5 HIGH
rs12204704 A/G Not available 72.5 HIGH
rs2743852 C/G Not available 70.2 MEDIUM
rs12525702 C/T Not available 69 LOW
NRG1 SNP8NRG221533 C/T Not available 62.5 HIGH
DRD 2 rs1801028 C/G C 0.980 2 2
DRD 3 rs6280 C/T T 0.650 77.2 HIGH
5-HT2A rs6313 C/T C 0.562 70.8 HIGH
SERT rs4795541 N* Not available 2 2
MAG rs720308 A/G A 0.844 74 MEDIUM
rs720309 A/T T 0.845 73.2 LOW

a) Androgene Receptor gene

© 2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com

3862 M. Minarik et al. Electrophoresis 2006, 27, 3856–3863

Therefore, under given experimental conditions, the analysis time can be calculated as ttotal = t0 1 ti6n
maximum number of injections was set at four. Using this where n is the number of injections. Therefore, each
condition no notable distortion of resolution was consecutive injection increases the total analysis time
observed. A typical result from such experiment is pre- only by the ti., while in a conventional series of CE runs
sented in Fig. 3 showing a four injection run with various the time is increased by t0 1 ti. In the present work, the
SNP markers. In the presented case the alleles were first peak (usually a primer) typically appeared 25 min
assigned with the use of an internal heterozygous stand- after the start and the last peak (usually a residual sin-
ard. An increase in one of the wild-type or mutant homo- gle-stranded fragment) eluted in approximately 45 min.
duplex peaks indicates a homozygous sample, symme- Hence, for every subsequent injection the net gain in
trical proportion of the two peaks indicates a hetero- analysis time was approximately 25 min. In addition,
zygous sample call. time needed to replace gel matrix and run pre-
conditioning [23] was also saved. Figure 4 demonstrates
An estimated gain in sample throughput using multiple- the difference in overall sample throughput between a
injection method can be determined using a simple normal series of CE runs and a series of multiple runs,
calculation. For a given analysis during which the first each including four sample injections. The plot reflects
fragments elute after a “dead” time of t0, and where the the t0 and ti values, additional time required to perform
time between the first and the last eluting peak is ti, the each sampling within one multiple-injection experiment
total analysis time can be expressed as t0 1 ti. In a as well as setup time needed to replace gel matrix and
standard series of multiple CE runs the total time can be to perform pre-conditioning prior to each CE run. It is
expressed as ttotal = (t0 1 ti) 6 n, where n is the number clear that with the multiple injections the overall
of runs. If multiple injections are performed, the total throughput is close to double.

Figure 3. Genotyping of multi-

ple SNP markers by CGCE with
multiple injections using hetero-
zygous internal control. The
labels indicate allele calls.

Figure 4. Effect of CE with mul-

tiple-injection approach to SNP
screening sample throughput.
(A) Processing time required for
regular screening, and (B) time
required for screening using four
injections per analysis.

© 2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com

Electrophoresis 2006, 27, 3856–3863 CE and CEC 3863

4 Concluding remarks with clinical diagnosis of schizophrenia. In addition, we

have succeeded in finding optimum conditions for analy-
Most current clinical applications of SNP genotyping are sis of another 31 SNPs and genotyped a group of 274
focused on investigating relatively small groups of 10–100 prostate cancer patients and 213 controls with con-
markers. Those include for example pharmacogenomic formed benign prostatic hyperplasia. Clinical outcome of
testing for targeted cancer therapy or genetic risk the two projects will be discussed in separate reports.
assessment in disease predisposition. In these cases,
recently developed high-throughput SNP typing plat- The authors would like to thank Patrik Sekerka for assis-
forms producing thousands of genotypes per analysis are tance with the CE experiments and Assoc. Prof. Michael
not optimal. Automated multicapillary DNA sequencers, Urban for helpful discussions regarding clinical aspects of
widely introduced during the human genome project, are the prostate cancer. This project was supported by the
now standard in most molecular biology laboratories. Internal Grant Agency of the Czech Ministry of Health
These instruments are rarely running at full capacity, project no. NR/8039–3.
therefore there is a demand for new applications.

Sample pooling is an effective approach to create sam- 5 References

ples with both alleles represented on the resulting PCR
product mixture. In the current study the overall rate for [1] Syvanen, A. C., Nat. Genet., 2005, 37, 5–10.
first-take successful optimization of experimental param- [2] Hardenbol, P., Baner, J., Jain, M., Nilsson, M. et al., Nat.
Biotechnol. 2003, 21, 673–678.
eters in a set of 55 different SNP marker candidates was
[3] Shen, R., Fan, J. B., Campbell, D., Chang, W. et al., Mut.
80%. Identification of the factors behind the low quality is Res. 2005, 573, 70–82.
not straightforward. The poor performance of some [4] Leamon, J. H., Lee, W. L., Tartaro, K. R., Lanza, J. R. et al.,
markers is likely related to the secondary structure Electrophoresis 2003, 24, 3769–3777.
effects, hence, to the actual target sequence surrounding [5] Lenz, H. J., J. Clin. Oncol. 2004, 22, 2519–2521.
the SNP position. We expect that with an additional effort, [6] Ulfelder, K. J., Schwartz, H. E., Hall, J. M., Sunzeri, F. J.,
Anal. Biochem. 1992, 200, 260–267.
primer design and experimental conditions could even-
[7] Vos, P., Hogers, R., Bleeker, M., Reijans, M. et al., Nucleic
tually be further evaluated to enable analysis of additional Acids Res. 1995, 11, 23, 4407–4414.
markers. Especially in cases where a functional relevance [8] van Eijk, M. J., Broekhof, J. L., van der Poel, H. J., Hogers, R.
specific to a given polymorphism is key for subsequent C. et al.,Nucleic Acids Res. 2004, 32, e47.
clinical decisions, the effort in further optimization would [9] Tobler, A. R., Short, S., Andersen, M. R., Paner, T. M. et al.,J.
be justified. The additional steps may include reposition- Biomol. Tech. 2005, 16, 398–406.
[10] Andersen, P. S., Jespersgaard, C., Vuust, J., Christiansen,
ing of the GC clamp from one primer to the other as sug-
M., Larsen, L. A., Hum. Mutat. 2003, 21, 116–122.
gested recently [17] or complete redesign of the PCR by [11] Tian, H., Brody, L. C., Landers, J. P., Genome Res. 2000, 10,
repositioning the SNP position within the target 1403–1413.
sequence. Results of a separate subset of experiments [12] Khrapko, K., Hanekamp, J. S., Thilly, W. G., Belenkii, A. et
(data not shown) suggest that other techniques utilizing al., Nucleic Acids Res. 1994, 22, 364–369.
partial melting, such as dHPLC or denaturing gradient gel [13] Schell, J., Wulfert, M., Riesner, D., Electrophoresis 1999, 20,
2864–2869.
electrophoresis, are also likely to fail in case of the mark-
[14] Larsen, L. A., Christiansen, M., Vuust, J., Andersen, P. S.,
ers that did not lead positive results in CGCE testing. In Comb. Chem. High Throughput Screen. 2000, 3, 393–409.
those instances a method based on a principle other than [15] Minarik, M., Minarikova, L., Bjorheim, J., Ekstrom, P. O.,
denaturation should be used, such as selective restriction Electrophoresis 2003, 24, 1716–1722.
(RFLP) or allele-specific PCR. By using multiple-injection [16] Minarik, M., Minarikova, L., Hrabikova, M., Minarikova, P. et
al., Electrophoresis 2004, 25, 1016–1021.
in each run, we have dramatically shortened the overall
[17] Hinselwood, D. C., Abrahamsen, T. W., Ekstrom, P. O.,
time required to screen multiple SNPs in hundreds of Electrophoresis 2005, 26, 2553–2561.
samples. Further improvement in sample throughput and [18] Sham, P., Bader, J. S., Craig, I., O’Donovan, M., Owen, M.,
level of automation in allele calling can be achieved by Nat. Rev. Gen. 2002, 3, 862–871.
using alternative fluorescent tags for labeling of the tested [19] Blake, R. D., Delcourt, S. G., Nucleic Acids Res. 1996, 24,
sample and the internal standard or even to better distin- 2095–2103.
guish individual strands within the same PCR fragment. [20] Amos, C. I., Frazier, M. L., Wang, W., Am. J. Hum. Genet.
2000, 66, 1689–1692.
The overall feasibility of the technique has been pre- [21] Wilkening, S., Hemminki, K., Thirumaran, R. K., Bermejo, J.
L. et al., BioTechniques 2005, 39, 853–858.
sented on two separate clinical projects scrutinizing the
[22] Ulvik, A., Refsum, H., Kluijtmans, L. A., Ueland, P. M., Clin.
role of genetic polymorphisms in disease predisposition. Chem. 1997, 43, 267–272.
We have optimized 13 SNP markers and performed [23] Griess, G. A., Hardies, S. C., Serwer, P., Electrophoresis
screening of a homogeneous population of 85 individuals 2005, 26, 102–111.

© 2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com