Significance of somatic mutations and content alteration of mitochondrial DNA in esophageal cancer

Paired tumors and surrounding tissues were previously banked specimens that were surgically removed from patients with histologically confirmed esophageal cancer (18 esophageal squamous cell carcinomas, one adenosquamous carcinoma and one adenocarcinoma). The age of the patients (1 female and 19 male) were from 38 to 73 years old with the mean age of 59 ± 14. The specimens were obtained from the tumor bank of the Pathology Department of Changhua Christian Hospital, Changhua, Taiwan, according to an institutionally approved protocol. DNA was isolated from frozen tissues using proteinase K digestion and phenol/chloroform extraction.

Comprehensive mutation analysis of the entire mitochondrial genome by temporal temperature gradient gel electrophoresis (TTGE) has been described previously [7, 29, 30]. Thirty-two pairs of overlapping primers were used to amplify the entire 16.6-kb mitochondrial genome [29]. The DNA fragments vary in size from 306-bp to 805-bp with an average of 594-bp [7, 29, 30]. Each fragment has an average of 74-bp on each end overlapping with the neighboring fragments. PCR products were denatured at 95°C for 30 s and slowly cooled to 45°C for a period of 45 min at a rate of 1.1°C/min. TTGE was performed on a Bio-Rad D-Code apparatus. The polyacrylamide (acrylamide:bisacrylamide 37.5:1) gels were prepared in 1.25 × Tris-Acetate-EDTA buffer containing 6 mol/L urea. Electrophoresis was carried out at 135 V for 4–6 h at a constant temperature increment of 1–2°C/h. The beginning and ending temperature were determined by computer simulation from the melting curve (50% denatured) of the DNA fragment [7, 29, 30]. On TTGE analysis, a single band shift represents a homoplasmic DNA alteration, and a multiple-banding pattern represents a heteroplasmic mutation. The DNA fragments from tumor and surrounding tissue of the same patient were analyzed side-by-side.

Any DNA fragments showing differences in banding patterns between tumors and surrounding samples were sequenced to identify the exact mutations. The DNA sequencing was performed by using the big dye terminator cycle sequencing kit (Perkin Elmer) and an ABI 377 (Applied Biosystems) automated sequencer. The results of DNA sequence analysis were compared with the complete human mitochondrial sequence deposited in GenBank (accession number NC_001807) http://​www.​mitomap.​org; http://​www.​ncbi.​nih.​gov[31] using MacVector 7.0 (Oxford Molecular Ltd, Oxford, England) software. Any DNA sequence differences between tumor and matched normal mtDNA were scored as somatic mutations. Sequence variations found in both tumor and matched normal mtDNA but different from that recorded in the GenBank were scored as germline variations. Those not recorded in MitoMap database were categorized as novel mtDNA variations, and those appear in the database were categorized as reported polymorphisms or mutations.

The mtDNA content was determined using real time quantitative PCR analysis [32] on ABI sequence detection system 7700 (Applied Biosystems) and was expressed as ratio of copy number of mtDNA to copy number of β2 microglobulin (β2M) gene (nDNA). The mtDNA/nDNA ratio in tumor was divided by the ratio in the surrounding non-cancerous tissue to obtain the ratio of mtDNA level in tumor to that in normal tissue. The ratio of >1 means that the mtDNA content in the tumor is higher than that in the surrounding tissue. The ratio of <1 means that the mtDNA content is reduced in the tumor when compared to its surrounding non-cancerous tissue.

Fisher's exact test is used to analyze whether there is association between the mtDNA content change in tumors and the presence of mutations. Non-parametric Mann-Whitney test was also used to test if mtDNA content alteration in tumor is different in patients with or without mtDNA mutations. P-values less than 0.05 are considered statistically significant.

To test if somatic mtDNA mutation is a general phenomenon in esophageal cancer and to characterize the mutation spectrum, the TTGE mutation detection method was used to screen the entire mitochondrial genome of tumor and surrounding normal tissues. Figure 1 depicts the four novel somatic mtDNA mutations. Panels A-C each showed single band to multiple band change from surrounding normal tissue to tumor tissue indicating a homoplasmy to heteroplasmy alteration, while panel D showed an alteration from a heteroplasmy to homoplasmy. Direct sequencing of the DNA fragment allowed the identification of the mutations as shown in Figure 1 and listed in Table 1. Eleven out of 20 (55%) tumors harbored somatic mtDNA mutations with a total of 14 mutations. Nine mutations were found in the D-loop region, one in 12S rRNA and four in mRNA. Eleven tumors were found to have somatic mutations (Table 1). All nine mutations in the D-loop involved an insertion or deletion within the poly C stretch of the nucleotide positions (np) 303–309 hot spot region. One of them, the T at nucleotide position 310 was changed to C, resulting a homopolycytidine stretch of 15 C's. The long stretch of poly C might increase the instability of this region. Two of the somatic mutations changed from the homoplasmic state in surrounding tissue to a mutant homoplasmic state in tumor. Three changed from heteroplasmic state in normal to homoplasmic state in tumor, and 6 changed from homoplasmic wild type to heteroplasmic mutant. Three had heteroplasmy in both the surrounding tissue and the tumor, but the proportions of the mutant mtDNA in the paired tissues are different.

When the sequence of surrounding tissue was compared with revised Cambridge reference mtDNA sequence deposited in GenBank (accession No: NC_001807) and MitoMap http://​www.​mitomap.​org, numerous germline sequence variations were found. A total of 185 distinct germline variations have been identified from the sequenced fragments. These do not represent all the sequence variations in the specimens analyzed since only the DNA regions that show somatic mutations, either homoplasmic or heteroplasmic, by TTGE were sequenced. Ten of these variations are novel (Table 2), and 175 (data not shown) of them have been recorded in the Mitomap database. Eight of these novel variations occurred in mRNA region, and five were missense variations (Table 2). Among them, the most remarkable is A11976C in NADH dehydrogenase subunit 4 (ND4) where tyrosine at amino acid position 406 is replaced by serine. Other novel missense variations are M47T in NADH dehydrogenase subunit 4 (ND4L), S531T in NADH dehydrogenase subunit 5 (ND5), S219N in ATP synthase subunit 6 (ATPase 6), and E109D in cytochrome c oxidase subunit II (COII). The reported polymorphisms are mostly unremarkable. Other frequent germ-line polymorphisms are C16233T, T16189C, T16519C in D-loop region, and C12705T in ND5, and T9540C in cytochrome c oxidase subunit II (COII) are silent variations with no amino acid change.

The frameshift mutation, 10941del TAACAACCCCC, causing frame-shift and premature termination at amino acid 110 of ND4, is expected to be a deleterious mutation. It involves a homoplasmy to heteroplasmy change. G10500A somatic mutation changes a moderately conserved alanine residue to threonine in ND4L (Fig. 2A). The A9182G mutation in ATPase6 is a back change from asparagine to serine. The wildtype is serine, but it is asparagine in the patient's normal tissue and changes back to serine in the tumor. The amino acid is located in a moderately conserved region (Fig. 2B). One silent mutation G9377A in COIII (cytochrome c oxidase subunit III) does not change the amino acid. However, it does not exclude the possibility that the nucleotide substitution may cause impairment in RNA processing due to improper precursor RNA folding. Another mutation, A1544T in 12S rRNA occurs at a stem region. This mutation causes the disruption of a critical base pair. It may destabilize the rRNA structure.

In addition to somatic mtDNA mutations, novel missense germline mutations may also cause affect on mitochondrial function. For instance, A11976C changes an evolutionary highly conserved hydrophobic tyrosine to hydrophilic serine at amino acid position 406 of ND4 subunit (Fig. 3A), would potentially cause functional effect. Another example is the E109D mutation in COII (Fig. 3B). Although it involves a conserved amino acid change, the shortening of hydrocarbon chain may have subtle effects on mitochondrial function. Accumulation of the subtle effect over time may increase an individual's risk of developing cancer. All five novel missense germline variations were not detected in 19 breast cancers, 31 neurofibromas, and 15 medulloblastomas, all of Caucasian origin in previous studies [7, 11, 13]. They were also not present in 18 oral cancers and 20 liver cancers of Taiwanese patients [8, 12].

To investigate if the down-regulation of mitochondrial function in cancers is due to regression of mtDNA biogenesis, the relative amount of mtDNA in tumor compared to that in the surrounding non-cancerous tissue was measured by real-time quantitative PCR analysis. Table 3 lists the results and clinicopathological characteristics of each tumor. The mean value of tumor/normal mtDNA ratio in the tumors with mtDNA mutation was 1.32 ± 0.31 and that of the tumors without mtDNA mutation was 1.45 ± 0.22. Thus, tumor with or without somatic mtDNA mutations show no significant difference (p value = 0.59 using non-parametric Mann-Whitney test) in mtDNA content change from normal to tumors. The majority (6 out of 11) of tumors with somatic mtDNA mutations have either elevated (>2) or reduced (<0.5) mtDNA content in tumors compared to surrounding normal tissue. However the p value (0.27) does not reach a statistically significant level as analyzed by Fisher's exact test. Analysis of a larger sample size will be required. The patient (case E18) whose mtDNA content is severely reduced in tumor harbored a frameshift deleterious mutation (10941 del 11bp). He was also the youngest patient (38 years of age). His tumor was at stage III and metastasized to lymph nodes. He expired 12 months after surgery. The patient (case E18) had a previous history of tongue cancer. The patient (case E05) who had the most reduced mtDNA content in tumor, harbored the A1544T novel mutation in 12S rRNA gene. His tumor was at stage III and had metastasized. He expired 4 months after the surgery. These two cases are consistent with the possibility that mtDNA mutations are associated with impaired biogenesis of mitochondrial genome. Four tumors had more than 2 fold increase in mtDNA content. Two of them (case E10 and E17) had 303–309 hot spot poly C mutation. Both patients had metastatic tumors, and died post-surgery. Two other patients had small non-metastatic tumors.

The relationship of the mtDNA content changes in tumor with tumor size, and metastatic status is depicted in figure 4. The majority of tumors display an inverse relationship between tumor size and mtDNA content change (Fig. 4). Eight out of 11 tumors with mtDNA mutations became metastatic, whereas only 4 out of 9 tumors without somatic mtDNA mutations became metastatic. The analysis of odds ratio showed that the patients with somatic mtDNA mutation has 3.33 times higher risk of metastasis compared to patients without somatic mutations. However, due to a small sample size, it does not reach the statistic significant level (odds ratio: 3.33, 95% CI: 0.515–21.6). In order to reach statistical significance, the study of a larger sample size will be necessary.