1016 Electrophoresis 2004, 25, 1016–1021

Marek Minarik1 Application of cycling gradient capillary

Lucie Minarikova1
Michala Hrabikova1 electrophoresis to detection of APC, K-ras, and
Petra Minarikova2 DCC point mutations in patients with sporadic
Petr Hrabal3
Miroslav Zavoral2 colorectal tumors
1
Genomac International A previously introduced technique of cycling gradient capillary electrophoresis (CGCE)
2
Department of Internal Medicine, was applied to monitoring of molecular changes during adenoma-carcinoma transition
Central Military Hospital
3 in progression of sporadic colorectal cancer. The purpose of this work was optimiza-
Department of Clinical
Pathology, tion of separation parameters for selected mutation regions in tumor suppressor genes
Central Military Hospital, involved in the early stages of colorectal carcinogenesis, followed by scanning for
Prague, Czech Republic these mutations in clinical tissue samples from patients with adenomatous polyps
and early carcinomas. A total of 47 colorectal tumors in various stages of progression
were examined. Main emphasis was given to evaluation of mutation detection sensi-
tivity and specificity required for effective early disease detection. A total of 7 different
somatic mutations was identified among 32 K-ras mutant samples, 1 inherited muta-
tion and 5 somatic mutations were identified among 15 adenomatous polyposis coli
(APC) mutated samples. None of the two previously reported “deleted in colorectal
carcinomas” (DCC) mutations was found in any of the clinical samples. In addition to
simple optimization of running conditions, CGCE has demonstrated sensitivity and
selectivity allowing detecting small mutant fractions as well as combination of multiple
mutants within a single target sequence.

Keywords: Adenomatous polyposis coli / Colorectal cancer / Cycling gradient capillary electro-
phoresis / K-ras / Mutations DOI 10.1002/elps.200305770

1 Introduction shown in Fig. 1. The initiation of tissue proliferation towards

adenoma is often accompanied by mutations within muta-
Colorectal cancer represents a leading cause of cancer- tion cluster region of exon 15 in the in adenomatous poly-
related death in developed countries (second overall after posis coli (APC) gene. The region extends over 200 codons
cardiovascular diseases). Current studies positively show harboring over 500 mutations currently registered, with the
increasing incidence as well as mortality rates, which can most frequent mutations around codon 1300 and 1450 [6].
be attributed to exposures to various factors including high As illustrated in Fig. 1, somatic mutations in APC are fol-
fat diet, smoking, environmental carcinogens, etc. [1, 2]. A lowed by development of small adenomas often exhibiting
number of reports on colorectal carcinogenesis suggest a substitutions in mutation hotspot of K-ras (codon 12 and
model based on deactivation of various tumor suppressor 13) exon 1). Further proliferation of the adenoma and its
genes during adenoma-carcinoma progression [3, 4]. transformation into malignant carcinoma is indicated by
According to this model, early stages of tumor initiation loss of heterozygosity at locus 18q21.3 containing the
and growth are indicated by accumulation of somatic DCC gene (deleted in colorectal carcinoma) [7]. Scarce
DNA mutations in several tumor suppressor genes, fol- reports of mutations within DCC include one substitution
lowed by loss of one or both alleles [5]. A scheme of molec- in exon 28 and another substitution in intron 14 [8]. Malig-
ular changes during the colorectal carcinogenesis is nant tumors further often exhibit deactivation of p53 pro-
tein through a number of possible TP53 gene mutations,
Correspondence: Marek Minarik, PhD, Genomac International, most of which occur in exons 5–8 [9]. The described dia-
Bavorska 856, CZ-155 41, Prague, Czech Republic gram suggests usability of monitoring somatic alterations
E-mail: [email protected] in APC and K-ras genes for detection of tumor 4 initiation
Fax: 1420-224-458-021 and DCC and TP53 for possible estimation of malignant
Abbreviations: APC, adenomatous polyposis coli, CGCE, potential in large adenomas and developed tumors.
cycling gradient capillary electrophoresis; DCC, deleted in Detecting the above mutant markers in adenomatous
colorectal carcinomas polyps facilitates early detection of the tumor progression

 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Electrophoresis 2004, 25, 1016–1021 CGCE mutation detection in colorectal tumors 1017

most cases (70%), 5% of adenomas were tubulovillous,

while none exhibited villous structures. The levels of dys-
plasia were mild (22.5% of cases), mild to moderate (30%
of cases), moderate (12.5% of cases), and severe (10% of
cases). Each polyp was dissected into smaller parts (3–
5 mm) and each part was then processed separately.
Genomic DNA was isolated from tissue samples using
JetQuick isolation kit (Genomed, Loehne, Germany).
Extracted DNA (,50 ng) was subjected to PCR amplifi-
cation with fluorescently labeled primers using conditions
specific to each mutant (see Table 1). Artificial mutant
standards were prepared by PCR amplification of a wild-
type DNA using a special extension on one of the PCR
Figure 1. Simplified model of molecular genetic changes primers. The extension included the mutated base in
during tumor progression of colorectal cancer. place of the wild-type base. The following mutation
regions were examined; selected mutation hotspots
and might result in a better chance of survival. Especially, within the mutation cluster region of APC gene (codons
detection of low-level point mutations has a significant 1243–1310 and 1413–1465), K-ras hotspot codon 12 and
potential for cancer prevention through early diagnosis. 13 of exon 1, and 2 mutation positions within exon 28 and
intron 14 in DCC [8].
The recently introduced technique of cycling gradient
capillary electrophoresis (CGCE) represented an ideal
automated tool for parallel monitoring of mutants in multi- 2.2 CGCE
ple clinical samples [10]. The present work describes the

CE and CEC
optimization of mutation detection in several tumor sup- After PCR amplification, mutant heteroduplexes were an-
pressor genes involved in colorectal carcinogenesis fol- alyzed in periodically cycling temperature gradient on
lowed by analysis of clinical samples taken from patients MegaBACE capillary-array genetic analyzer (Amersham
during colonoscopy treatment. Biosciences, Sunnyvale, CA, USA) equipped with a
Caddy plate loading robot (Watrex Praha, Czech Repub-
lic) for unattended operation. The separation took place in
2 Materials and methods a standard denaturing gel matrix (MegaBACE long-range
matrix; Amersham Biosciences) containing 7 M urea. The
2.1 Sample preparation
running conditions included injection for 120 s at 3 kV and
DNA samples were collected from large polyps (. 10 mm) running at 6 kV for 90 min. The cycling gradient tempera-
in colon and rectum from patients by polypectomy during ture profiles were created using MBCS software Version
colonoscopy examination. Tubular adenoma was found in 2.0 (Genomac International, Prague, Czech Republic).

Table 1. Optimized conditions for mutation analysis in selected mutation regions

Target Primer sequence Theoretical CGCE

melting temperature
temperature range

APC1 5’-GTTCATTATCATCTTTGTCATCAGC-3’ 717C 51–497C

5’-FL-[GC]-TTTATTTCTGCTATTTGCAGGGTA-3’
APC2 5’-FL-[GC]-CCATGCCACCAAGCAGAAGTA-3’ 707C 52–507C
5’-TCTCTTTTCAGCAGTAGGTGCTT-3’
K-ras 5’-ATGACTGAATATAAACTTGTG-3’ 707C 52–507C
5’-FL-[GC]-CCTCTATTGTTGGATCATATTC-3’
DCC–C9 5’-FL-[GC]-TTTTCAACACACAATCCCTTT-3’ 657C 45–437C
5’-TCATGCAAACTTACCCATTATGA-3’
DCC–D114 5’-TCATCACTGTGTTTTCTATTTCAGG -3’ 737C 53–517C
5’-FL-[GC]-ACAGACACAGGAAGCAAA-3’

[GC] denotes a high-melting clamp: CGCCCGCCGCGCCCCGCGCCCGTCCCGCCGCCCCCGCCCG.

FL denotes labeling by fluorescein.

 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1018 M. Minarik et al. Electrophoresis 2004, 25, 1016–1021

The temperature ranges were optimized for each marker

(target sequence). The optimization is described in Sec-
tion 3. All primers and CGCE running conditions used in
this work are listed in Table 1.

3 Results
The analysis of somatic point mutations shows potential
for early detection of colorectal tumors through mutation
analysis of bioptic tissue samples. We have previously
introduced CGCE for high-throughput screening of sin-
gle-nucleotide polymorphisms (SNPs) using multiple
injection approach [10]. In the present work, we demon-
strate the applicability of this method in clinical research
by detecting several point mutations in tissue samples
taken during regular colonoscopy examination. All muta- Figure 2. Optimization of CGCE running parameters.
DCC intron 14 mutant standard analyzed at different
tion regions analyzed in this study were previously found
ranges of cycling temperatures: (A) 437C–417C, (B) 447C–
to be related to colorectal carcinogenesis. 427C, (C) 467C–447C, (D) 487C–467C. An optimum temper-
Prior to detecting selected mutation markers, it is neces- ature range of 447C–427C was used for mutation analysis
sary to optimize temperature conditions for all selected in clinical samples. Similar optimization was performed on
all mutation regions used in the study (see Table 1).
target sequences. Mutation detection techniques based
on separation of homo- and heteroduplexes often require
careful optimization of separation conditions including
homo-and hetero- duplexes (labeled 1–4) were analyzed
optimum melting temperature [11] in electrophoresis or
by CGCE at four different gradient cycling ranges. It can
temperature in combination with mobile phase gradient
be seen that at lower temperature range the wild-type (1)
composition in chromatography [12]. Similarly, in a tradi-
and mutant (2) homoduplexes are not completely
tional single-sweep temperature gradient approach
resolved. Clearly the best resolution was achieved when
(TGCE), the gradient range as well as gradient slope has
cycling between 42 and 447C. This is in agreement with
to be optimized in order to find the best conditions to
theoretical predictions, indicating natural melting temper-
resolve all forms (wild-type and mutant homoduplexes
ature of the DCC intron 14 target sequence of 647C since
and two heteroduplexes) potentially present in the sample
the 7 M urea in the separation matrix effectively lowers the
[13]. In CGCE, the only parameter to optimize is a range of
DNA melting temperature by approximately 217C [15, 16].
temperatures while the individual gradient cycles are per-
At a gradient higher than the optimum, the two heterodu-
formed rapidly with no delay. It was demonstrated pre-
plexes (3 and 4) co-elute. Using the identical procedure,
viously that for high reproducibility fast cycling is prefer-
conditions were developed for all target sequences
ential to slower temperature sweeps [14]. The maximum
examined in this study. The final optimum conditions for
achievable frequency of gradient cycling is usually limited
all markers are listed in Table 1. These conditions were
by the rate of temperature control inside the capillary
used to examine the clinical samples.
oven and is typically lower with increasing range of the
cycling gradient. Optimization of a new marker usually One of the most important parameters of each mutation
involves performing several runs using different tempera- detection technique is detection sensitivity expressed in
ture ranges of cycling gradient. During optimization, terms of a fraction of mutated DNA copies detectable in
cycling temperatures are usually selected 27C above and an excess of wild-type DNA [17]. While conventional DNA
below the theoretical melting temperature of the target sequencing typically detects a presence of minor allele at
DNA segment. An example of optimization of the temper- a concentration above 20%, techniques based on spatial
ature cycling conditions is shown in Fig. 2. The PCR prod- separation of mutant and wild-type (such as denaturant
uct of DCC gene wild-type target sequence was mixed at gradient gel electrophoresis, DGGE, or single-strand con-
a 1:1 ratio with an artificial point-mutant standard (see formation polymosphism, SSCP) are capable of detecting
Section 2). The intron 14 mutant, labeled as C9, was pre- low-level mutations in fractions down to 10% or lower
viously found in colorectal carcinoma patients [8]. After [18]. An example of detection sensitivity using CGCE is
mixing, 5 min denaturation was performed at 957C fol- shown in Fig. 3. Two clinical samples, positive for muta-
lowed by reannealing of wild-type and mutant strands at tion in K-ras mutation hotspot (codon 12 and 13), showed
657C for 60 min. The resulting four combinations of pattern of heteroduplex peaks at the optimum separation

 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Electrophoresis 2004, 25, 1016–1021 CGCE mutation detection in colorectal tumors 1019

Table 2. Overview of results of mutation analysis in 47

clinical tissue samples

Gene Location
Ascendens Descendens Sigmoideum/
(15 tumors) (7 tumors) rectum
(25 tumors)

APC 1 (7%) 3 (43%) 5 (20%)

K-ras 5 (33%) 3 (43%) 15 (60%)
DCC 0 (0%) 0 (0%) 0 (0%)

Figure 3. Illustration of CGCE sensitivity. K-ras mutant

region (exon 1) was examined in two patients with differ-
ent levels of mutated DNA fraction. From the high mutant
fraction (A), the G?A substitution at codon 13 could
clearly be identified by sequencing (complimentary
strand was sequenced). If the same mutant is present in
a low lever (B), it is still clearly visible in the CGCE trace
but not detectable by sequencing.

temperature range of 527C to 507C (Figs. 3A and B). The

first sample clearly contains a larger fraction of the
mutated DNA compared to the second sample. While in
the first case the mutations could be identified by rese-
quencing of the PCR product (Fig. 3A, bottom), no muta- Figure 4. Overview of various types of APC mutations
found during clinical CGCE of 47 tissue samples. Upper
tions were found when sequencing the second sample
left trace, nonmutated reference. Bottom right trace, dou-
containing lower fraction of the mutant (Fig. 3B, bottom). ble-mutant with one inherited mutation (T?A substitution
The fraction presented in Fig. 3B was estimated at , 10% in codon 1442) and a second somatic mutation (C?T
level. This estimation was based on a separate calibration substitution in codon 1450). The inherited codon 1442
experiment in which different fractions of artificial mutant mutation was also found separately in blood of another
and wild-type were mixed (data not shown) and is in patient (bottom left).
agreement with similar calibration experiments published
previously [19].
resulting in a double-mutant (Fig. 4, bottom right). The two
The optimized conditions were used for analysis of the concurrent mutations produce a pattern of multiple peaks
five selected mutation regions in a total of 47 tissue sam- representing wild-type and mutant forms with two pairs of
ples. An overview of mutants found during the clinical heteroduplexes for each of the mutants plus an additional
study is summarized in Table 2. Mutations in APC gene heteroduplex pair resulting from mutual combination of
were found in a total of 9 samples (9/47, 19%). Out of the two mutant sequences. The most frequently found
these 9 samples, 6 different mutations were identified. mutations were in K-ras oncogene. Out of 47 tissue sam-
An overview of the 6 different APC mutations and one ples, 22 have exhibited a K-ras mutation. Seven different
double-mutant sample is shown in Fig. 4 together with a types of somatic mutations were identified from the 22 K-
nonmutated sample used as a reference. The exact muta- ras positive samples. An overview of K-ras mutations is
tion positions, denoted as the codon number and the presented in Fig. 5. It can be seen that most mutations
mutation type in each electropherogram window, were were localized in codon 12, while only one was from
confirmed by sequencing. Mutation in codon 1442 (T?A codon 13. Similarly to the above APC gene, a double-
substitution) was confirmed as inherited by its presence mutant was found in one of the patients (Fig. 5, bottom
also in the blood sample taken from the same patient. left). While one of the mutations was identified as G?A
The identical inherited mutation was found in tissue as substitution in codon 12, it was not possible to identify
well as blood samples of yet another patient. In the tissue the second mutation, clearly detected in the CGCE elec-
sample, however, it was in combination with another tropherogram (see Section 4). None of tumors was posi-
somatic mutation at codon 1450 (C?T substitution) tive for any of the two DCC mutations. When comparing

 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1020 M. Minarik et al. Electrophoresis 2004, 25, 1016–1021

hibit multiple mutations at once. Distinguishing the differ-

ent point mutations has a prognostic value, since different
mutations often exhibit different impact on malignancy
and survival rates [20]. In the presented example, both
mutations are clearly distinguishable from the resulting
combination of heteroduplex forms of individual mutants.
The presence of two mutants within a single APC target
sequence are typical for a heavily mutated cancer tissue.
The fact that one of the mutations was inherited (found
also in blood of the patient) certainly suggests the initial
phases of tumor suppressor inactivation [21]. The impor-
tance of this inherited mutation, which was also found in
another patient’s blood sample, is now under study. Also
a subject to further investigation is an unidentified second
Figure 5. Overview of various types of K-ras mutations K-ras mutation from a sample shown in Fig. 5, bottom left.
found during CGCE analysis of 47 tissue samples. Upper By mixing with a set of know mutations, it was prelimi-
left trace, nonmutated reference trace. Bottom left trace,
narily concluded that the mutation is probably at the exact
mutant identified as G?A substitution in codon 12 and
same position as the first mutation (G ? A at codon 12)
unidentified mutant, presumably from the same codon
(see text for details). but must likely involving a different base substitution. This
theory will be examined in more details, however, the
inability to identify the second mutation further confirms
mutation frequency based on tumor location, we have the usability of CGCE over direct sequencing.
found more mutated tumors in colon descendens (30/47,
64%) than in colon ascendens (15/47, 32%). In the presented study, 28 mutants were identified from a
total of 47 tumor samples, 8 samples exhibited both APC
and K-ras mutations. The overall higher frequency of K-ras-
positive samples compared to APC positives can be attrib-
4 Discussion uted to the localized hot spot within the codons 12 and 13
The application of cycling gradient significantly simplifies of K-ras allowing virtually 100% of K-ras mutants to be
the optimization of separation conditions in comparison detected. In addition, it may be possible to identify various
to a single-sweep gradient. The optimization only involves K-ras mutants using an internal artificial mutant standard
finding the temperature range and does not require com- as an alternative to direct sequencing [22]. With APC gene,
plicated matching of the gradient duration (slope) to the it is expected that some portion of samples with mutation
duration of migration of fragments in the capillary. An outside the monitored mutation cluster region escapes
important feature for clinical applications is mutant detec- the assay. In contrast with the expectations, no sample
tion sensitivity. Bioptic tissue samples are usually col- was positive for any of the two DCC mutations monitored
lected from areas of macroscopic changes or early ade- in the study [8]. Finally, the overall higher frequency of the
nomas, which often contain only a small portion of cells APC and K-ras mutants in colon descendens combined
carrying mutated DNA. This requires sensitive techniques with sigmoideum and rectum (25/32, 78%) compared to
capable of detecting low levels of mutants. The presented colon ascendens (6/14, 43%) is in an agreement with the
example showed capability to unambiguously detect clinically observed higher tendency of the malignant poly-
mutated fractions of ,10%. The maximum sensitivity poid growth in the left colon [23].
reached in an unrelated experiment was 1% (data not
shown). A possibility to further increase the sensitivity CGCE represents a viable tool for mutation detection. The
(lower the mutant fraction detected) is now under investi- technique exhibits high sensitivity and specificity, which is
gation. One potential approach is enrichment of hetero- often required for mutation analysis in clinical samples
duplex forms using fraction collection followed by PCR such as bioptic or resection tissue etc. The method does
reamplification of the isolated fractions. not require any sample manipulation prior to capillary
electrophoretic separation and is compatible with two
Specificity of mutation detection techniques becomes commercially available capillary array sequencing instru-
crucial when more than one mutation present within the ments [24, 25]. The main advantage of the technique over
target sequence is to be detected. Such situation is com- currently existing TGCE analogues is in simple optimiza-
mon in the case of monitoring alterations in tumor sup- tion of separation conditions and a possibility of multi-
pressor genes in dysplasic colon tissue, which often ex- plexing by multiple-injection approach [10].

 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Electrophoresis 2004, 25, 1016–1021 CGCE mutation detection in colorectal tumors 1021

The authors would like to thank Dr. Cyril Salek for his assis- [12] Jones, A. C., Austin, J., Hansen, N., Hoogendoorn, B., Oef-
tance in designing mutant primers. This work was sup- ner, P. J., Cheadle, J. P., O’Donovan, M. C., Clin. Chem.
1999, 45, 1133–1140.
ported by the Internal Grant Agency of the Czech Ministry
[13] Schell, J., Wulfert, M., Riesner, D., Electrophoresis 1999, 20,
of Health (IGA MZ) No. NK/6469-3.
2864–2869.
[14] Bjorheim, J., Minarik, M., Gaudernack, G., Ekstrom, P. O., J.
Received September 17, 2003 Sep. Sci. 2003, 26, 1163–1168.
[15] Bullock, A. N., Henckel, J.DeDecker, B. S., Johnson, C. M.,
Nikolova, P. V., Proctor, M. R., Lane, D. P.,Fersht, A. R., Proc.
5 References Natl. Acad. Sci. USA 1997, 94, 14338–14342.
[16] Blake, R. D., Delcourt, S. G., Nucleic Acids Res. 1996, 24,
[1] Czech Cancer Registry Report, Prague, Czech Repub-
2095–2103.
lic2001.
[2] Cancer Facts and Figures 2003, American Cancer Society. [17] Prosser, J., Trends Biotechnol. 1993, 11, 238–246.
[3] Gryfe, R., Swallow, C., Bapat, B., Redston, M., Gallinger, S., [18] Schwaab, R., Oldenburg, J., Lalloz, M. R., Schwaab, U.,
Couture, J., Curr. Probl. Cancer 1997, 21, 233–300. Pemberton, S., Hanfland, P., Brackmann, H. H., Tuddenham,
[4] Houlston, R. S., J. Clin. Pathol. 2001, 54, 206–214. E. G., Michaelides, K., Hum. Genet. 1997, 101, 323–332.
[5] Leslie, A., Carey, F. A., Pratt, N. R., Steele, R. J., Br. J. Surg. [19] Bjorheim, J., Minarik, M., Gaudernack, G., Ekstrom, P. O.,
2002, 89, 845–860. Anal. Biochem. 2002, 304, 200–205.
[6] Beroud, C., Soussi, T., Nucleic Acids Res. 1996, 24, 121–
[20] Morrin, M., Kelly, M., Barrett, N., Delaney, P., Gut 1994, 35,
124.
1627–1631.
[7] Boland, C. R., Sato, J., Appelman, H. D., Bresalier, R. S.,
Feinberg, A. P., Nat. Med. 1995, 1, 902–909. [21] Hesketh, R., The Oncogene and Tumour Suppressor Gene
Core Factsbook, Academic Press, San Diego, CA 1997.
[8] Cho, K. R., Oliner, J. D., Simons, J. W., Hedrick, L., Fearon,
E. R., Preisinger, A. C., Hedge, P., Silverman, G. A., Vogel- [22] Bjorheim, J., Gaudernack, G., Giercksky, K. E., Ekstrom, P.
stein, B., Genomics 1994, 19, 525–531. O., Electrophoresis 2003, 24, 63–69.
[9] Béroud, C., Verdier, F., Soussi, T., Nucleic Acids Res. 1996, [23] Mensink, P. B., Kolkman, J. J., Van Baarlen, J., Kleibeuker, J.
24, 147–150. H., Dis. Colon Rectum 2002, 45, 1393–1396.
[10] Minarik, M., Minarikova, L., Bjorheim, J., Ekstrom, P. O.,
Electrophoresis 2003, 24, 1716–1722. [24] Application Note No. 63-0048-87, Amersham Biosciences,
Uppsala 2002.
[11] Khrapko, K., Hanekamp, J. S., Thilly, W. G., Belenkii, A.,
Foret, F., Karger, B. L., Nucleic Acids Res. 1994, 22, 364– [25] Li, Q., Liu, Z., Monroe, H., Culiat, C. T., Electrophoresis
369. 2002, 23, 1499–1511.

 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim