MATERIALS AND METHODS

Over 1000 fresh frozen colorectal adenocarcinoma specimens were collected in the Department of Medical Genetics, University of Helsinki between May 1994 and June 1998. The microsatellite instability (MSI) status of the samples was previously determined using BAT26 mononucleotide marker.^23,24 We chose all 23 samples which had displayed LOH in chromosome 22q13 in our previous genome wide analysis of 104 MSI negative CRCs with 372 microsatellite markers,²² and analysed them by genomic sequencing of ACO2 in tumour and normal DNA. Because LOH at the locus of interest was an inclusion criterion, normal DNA was sequenced in addition to tumour DNA to exclude loss of ACO2 variants during tumour progression. This approach allowed detection of all putative somatic and germline sequence changes. Tumour DNA was extracted from fresh frozen specimens. A pathologist evaluated the proportion of tumour tissue histologically before DNA extraction. All samples displayed over 60% carcinoma tissue. Normal DNA was obtained either from blood or normal colon mucosa. The specimens representing normal mucosa were always derived from a separate site rather than from tumour margins. Fifteen of the samples were sporadic and eight had one first degree relative affected with CRC. Because of the MSI negativity and mild family history of cancer, these eight familial samples have not been screened for known genes predisposing to CRC. Familial adenomatous polyposis, Peutz-Jeghers syndrome, and juvenile polyposis were, however, clinically excluded. The age of the patients varied from 30 to 89.

Somatic and germline mutations were searched for by direct sequencing of the 18 coding exons of ACO2. Primers were designed for each exon using GenBank sequence (www.ncbi.nlm.nih.gov). Primer sequences and GenBank accession numbers are listed in table 1. The PCR reactions were carried out in a 50 μl reaction volume containing 100 ng genomic DNA, 1 × AmpliTaq Gold PCR buffer (Applied Biosystems (AB) Division, Foster City, CA), 300 μmol/l of each dNTP (Finnzymes, Espoo, Finland), 0.8 μmol/l of each primer, and 2 units of AmpliTaq Gold DNA polymerase (AB). MgCl₂ concentration was 2.25 mmol/l for all exons except for exons 6 and 15. For exon 6 it was 3 mmol/l and for exon 15 1.5 mmol/l. The PCR cycles for exons 1–4, 6–11, and 16 were 35 cycles at 95°C for 45 seconds, 57°C for 45 seconds, and 72°C for one minute. For exons 5, 12–15, 17, and 18 the annealing temperature was 62°C. Predenaturation was performed at 95°C for 10 minutes and final extension was 72°C for 10 minutes for all exons. The PCR products were run on a 2% agarose gel (BioWhittaker Molecular Applications, Rockland, ME) to verify the specificity of the reactions and purified using NucleoSpin Multi-96 Extract Kit (Macherey-Nagel, Düren, Germany).

Direct sequencing of the PCR products was performed using cycle sequencing with Big Dye 3 Terminator chemistry, and reactions were run on an ABI 3100 capillary sequencer (AB) according to the manufacturer’s instructions.

RESULTS

Twenty-three colon cancer patients were analysed for somatic and germline mutations of ACO2 by genomic sequencing. Sequencing showed two silent changes. Eight out of 23 (35%) patients had a 212A>C change (T64T) in exon 3 (fig 1). In one patient the variant was homozygous. Sixteen out of 23 (70%) patients had a 690C>T change (L224L) in exon 5. Two of the patients had a homozygous change. Both polymorphisms have been reported previously.¹⁸ Occurrence of the variants was not associated with family history of CRC. The two variants were not predicted to affect splicing when analysed with Splice Site Prediction by Neural Network and NetGene2 programs (http://www.fruitfly.org/seq_tools/splice.html, http://genome.cbs.dtu.dk/services/NetGene2/).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

ACO2 variant in exon 3. The upper panel shows the heterozygous C/A variant in normal DNA. The lower panel shows the deletion of the C allele in the respective tumour DNA.

All selected cases had showed LOH close to the ACO2 locus when analysed with an adjacent microsatellite.²² This allelic loss typically included ACO2 as evidenced by loss of heterozygosity in most of the informative tumour sequences (fig 1). In 16 tumours at least one of the polymorphic sites was informative, and LOH was observed in 14 of these 16 tumours. All seven samples that had a heterozygous variant in exon 3 displayed LOH in the respective tumour. Five of these had lost the A allele and two had lost the C allele. Ten out of 16 samples, which had a variant in exon 5, showed LOH. Of these 10 patients, five had lost the C allele and five had lost the T allele. In four cases the locus was heterozygous but the tumour did not show unambiguous LOH. Thus, a minimum of 14 of the 104 (13.5%) tumours in the original LOH tumour study of unselected microsatellite stable samples had lost one allele of ACO2.

DISCUSSION

Several genes relevant to colorectal cancer development have been identified during the last two decades. The first such genes were TP53 and APC, which were identified in 1989 and 1991, respectively.^2,4 Germline mutations in APC predispose to familial adenomatous polyposis (FAP)² and somatic mutations in APC can already be seen in the earliest neoplastic lesions.²⁵ Like APC, many genes which are necessary for cancer development function as tumour suppressor or caretaker genes. The first step in identification of such genes has often been the detection of deletions in the respective chromosomal region. In addition to the well known 5q, 18q, and 17p deletions, losses of chromosomes 1p, 8p, 15, and 22q are frequently observed in colorectal cancer.^9–11 These regions most likely contain key tumour suppressor genes, which have not yet been identified.

Genes associated with mitochondrial energy metabolism have not been considered very relevant to cancer development, since many oncogenes and tumour suppressor genes are part of important signal transduction pathways rather than associated with energy metabolism. More recently, however, the role of mitochondrial DNA and Krebs’ cycle genes in tumorigenesis has received increasing attention. In 1997, Polyak et al²¹ reported a high frequency of somatic mutations in mitochondrial DNA in colorectal cancer cells.²¹ Furthermore, germline mutations in nuclear encoded mitochondrial proteins predisposing to hereditary paragangliomas and HLRCC were observed in SDH and FH, respectively.^12–15 Most mutations affecting FH and SDH have been found in the germline. Somatic mutations appear to be rare; however, in both enzymes somatic inactivation has been reported as well.^14,17 Both SDH and FH are components of the tricarboxylic acid cycle and lose function in tumorigenesis like classical tumour suppressor genes, raising the question of how important mutations of housekeeping and mitochondrial proteins are in the pathogenesis of common tumour types. The mechanism by which Krebs’ cycle gene defects contribute to tumorigenesis is as yet unclear. It has been shown that a moderately increased level of oxygen radicals is highly mitogenic.²⁶ Mutations in mitochondrial proteins may lead to increased levels of reactive oxygen, thus giving cells a growth advantage.²¹ Another possible explanation is hypermutability resulting from oxidative damage.¹⁵

Related proteins can play role in very different tumour types. SDH and FH are both components of the tricarboxylic acid cycle. However, mutations in SDH predispose to paragangliomas, while mutations in FH predispose to leiomyomas and renal cell cancer.^12–15 In addition to hereditary predisposition, these proteins appear to play a role in sporadic tumorigenesis as well.^14,17 It is therefore reasonable to hypothesise that mutations in other Krebs’ cycle proteins may also have a role in tumorigenesis. We studied the possible role of ACO2 in colorectal cancer. Twenty-three patients, who showed deletions near the ACO2 locus in chromosome 22q13, were screened for somatic and germline variants, but pathogenic mutations were not found. We observed two silent polymorphisms in exons 3 and 5, both of which have been reported previously.¹⁸ A 212A>C (T64T) change in exon 3 was found in 35% (8/23) of the patients. A 690C>T (L224L) change in exon 5 was detected in 70% (16/23) of the patients. Both changes appear to be fairly common in the Finnish population. The respective allele frequencies among the American white population were 31% and 46%,¹⁸ indicating that the 690C>T change is more common among Finns. The occurrence of the variants did not correlate with family history of the disease.

Fourteen out of 16 informative tumour sequences showed LOH at the ACO2 locus (fig 1). This confirms that ACO2 resides in the commonly deleted region. The losses were not targeted to a particular allele. In one case, two flanking microsatellite markers showed LOH, but LOH was not detected in the respective tumour sequence. It is possible that this patient has a homozygous deletion at the ACO2 locus.

Although no pathogenic mutations were found, it is possible that methylation has a role in the inactivation of ACO2. To our knowledge, the ACO2 promoter has not been reported previously. The sequence approximately 5000 bases upstream from the initiation codon ATG, however, is predicted to contain two CpG islands (22-−207 and −295-−519 bp from ATG) when analysed with the CpGPlot program (http://bioweb.pasteur.fr/seqanal/interfaces/cpgplot.html). It is, therefore, possible that deletion of one allele and promoter CpG hypermethylation of the other might have occurred in some lesions, providing biallelic inactivation of ACO2.

Our results show that ACO2 is often deleted in colorectal cancer but is unlikely to be the true target of the deletions unless haploinsufficiency of ACO2 promotes tumorigenesis. Further work is required to identify the as yet unknown 22q gene(s) inactivated during colorectal cancer development.