Somatic mitochondrial DNA mutations in Chinese patients with osteosarcoma

Abstract

Somatic mutations in mitochondrial DNA (mtDNA) have been long proposed to drive the pathogenesis and progression of human malignancies. Previous investigations have revealed a high frequency of somatic mutations in the D-loop control region of mtDNA in osteosarcoma. However, little is known with regard to whether or not somatic mutations also occur in the coding regions of mtDNA in osteosarcoma. To test this possibility, in the present study we screened somatic mutations over the full-length mitochondrial genome of 31 osteosarcoma tumour tissue samples, and corresponding peripheral blood samples from the same cohort of patients. We detected a sum of 11 somatic mutations in the mtDNA coding regions in our series. Nine of them were missense or frameshift mutations that have the potential to hamper mitochondrial respiratory function. In combination with our earlier observations on the D-loop fragment, 71.0% (22/31) of patients with osteosarcoma carried at least one somatic mtDNA mutation, and a total of 40 somatic mutations were identified. Amongst them, 29 (72.5%) were located in the D-loop region, two (5%) were in the sequences of the tRNA genes, two (5%) were in the mitochondrial ATP synthase subunit 6 gene and seven (17.5%) occurred in genes encoding components of the mitochondrial respiratory complexes. In addition, somatic mtDNA mutation was not closely associated with the clinicopathological characteristics of osteosarcoma. Together, these findings suggest that somatic mutations are highly prevalent events in both coding and non-coding regions of mtDNA in osteosarcoma. Some missense and frameshift mutations are putatively harmful to proper mitochondrial activity and might play vital roles in osteosarcoma carcinogenesis.

Osteosarcoma is the most common primary bone neoplasm in children and adolescents, with a peak incidence at the age of 10–19 years (Hameed & Dorfman 2011). This aggressive tumour usually stems from the metaphysis adjacent to the growth plates of long bones and displays a high propensity for rapid local invasion and early formation of distant metastases, preferentially to the lungs (Unni 1998). Distinct from many other sarcomas genetically hallmarked by specific chromosomal translocations, osteosarcoma is characterized by complex genomic instability, such as a high degree of aneuploidy, an accumulation of unbalanced chromosomal rearrangements and multiple regions of aberrant amplifications and deletions (Martin et al. 2012). In the past two decades, introduction of intensified multiagent chemotherapy regimens and wide surgical excision of tumours have dramatically improved 5-year survival of patients with localized disease to 65–70% (Chou et al. 2008). However, patients with distant metastases at initial diagnosis or those who encounter recurrence after first-line treatment still have extremely poor prognosis largely due to chemoresistance, with a long-term survival rate of less than 20% (Chou et al. 2008). This disappointing outcome highlights the urgent need of developing alternative strategies for early diagnosis and more effective treatment of this fatal malignancy.

Mitochondria are ubiquitous multifunctional organelles in eukaryotic cells that exert pivotal roles in energy metabolism, free radical generation, calcium buffering and apoptosis (Wallace et al. 2010). Mitochondria possess their own unique genome, namely mitochondrial DNA (mtDNA), which, unlike nuclear DNA (nDNA), replicates independently of the cell cycle and is present in hundreds to thousands of copies per mammalian cell (Yu 2011). Human mtDNA is a 16,569-bp, closed-circular, double-helical molecule that encodes 13 core polypeptide components of respiratory enzyme complexes, two ribosomal RNAs (rRNAs) and a set of 22 transfer RNAs (tRNAs) required for intramitochondrial protein synthesis (Li et al. 2012). The displacement (D)-loop located at nucleotide positions 16,024-576 is a non-coding area with essential regulatory elements governing mtDNA duplication and transcription (Li et al. 2012). Due to inefficient DNA repair mechanisms, the lack of protective histones and its physical proximity to high concentration of endogenous reactive oxygen species (ROS) in the mitochondrial inner membrane, mtDNA is remarkably vulnerable to ROS and other types of genotoxic insults and hence acquires mutations at a much greater rate (10-fold to 200-fold) than nDNA (Cline 2012). Mutant mtDNAs can coexist with wild-type copies within a single cell in a state known as ‘heteroplasmy’ or completely overtake all normal molecules (‘homoplasmy’) (Yu 2012). Alongside numerous sequence alterations present in nDNA, somatic mtDNA mutations have been extensively detected in a broad spectrum of primary human tumours, with the highest prevalence in the D-loop control region and have been long proposed as attractive cancer biomarkers (Yu 2012). Somatic mutations and damage to mtDNA may potentially lead to abnormal oxidative phosphorylation (OXPHOS) and trigger overproduction of carcinogenic ROS, which in turn serve to accelerate the rate of mtDNA mutations and amplify oxidant injury (Chandra & Singh 2011). It has been increasingly accepted that this scenario may be involved in driving the pathogenesis of at least some cancer entities (Chandra & Singh 2011).

Earlier studies from our group and others have demonstrated that osteosarcoma tissues carried a high frequency of somatic mutations in the D-loop region of mtDNA, predominantly in two hypervariable sites (HVS1 and HVS2) as well as the homopolymeric C stretch (D310) between nucleotide positions 303 and 309 and contained decreased mtDNA copy number in comparison with normal bone tissues (Guo & Guo 2006; Yu et al. 2012a). In addition, somatic D-loop mutation is plausibly a critical determinant amongst others leading to lowered mtDNA quantities in osteosarcoma (Yu et al. 2012a). In an attempt to further investigate the prevalence and distribution of mtDNA mutations in osteosarcoma, we extended our previous work by screening for somatic mutations across the entire mitochondrial genome in 31 cases of patients with osteosarcoma and assessed the potential functional significance of these mutations. We also reviewed the clinical information about the patients in our series and explored the potential association of somatic mtDNA mutations with diverse clinicopathological features of osteosarcoma.

Materials and methods

Patients and tumour specimens

Thirty-one patients (18 boys and 13 girls) who were diagnosed with osteosarcoma and underwent surgical resections at the Department of Bone and Soft Tissue Tumor, Tianjin Medical University Cancer Hospital and at the Department of Surgical Oncology, Central Hospital of China National Petroleum Corporation from June 2005 to October 2010 were recruited in this study. These are the same cohort of patients reported previously, and somatic mutations in the D-loop region have been identified in these individuals (Yu et al. 2012a). All tumour tissues were evaluated histologically by two experienced pathologists, including 22 osteoblastic osteosarcomas, 6 chondroblastic osteosarcomas and 3 fibroblastic osteosarcomas. Primary neoplasms were located in the distal femur (n = 19), the proximal humerus (n = 6), the proximal tibia (n = 4), the proximal fibula (n = 1) and the distal ulna (n = 1). Eight patients had metastasis at the time of diagnosis. The size of tumour was measured as the maximum tumour diameter on radiographic images including computerized tomography (CT) scans and magnetic resonance imaging. Tumour tissues were snap-frozen in liquid nitrogen immediately after surgical resection and stored at −80 °C prior to DNA extraction. As references, 5–10 ml of peripheral blood samples was drawn from all subjects and kept at −80 °C.

Ethical approval

This study was approved by the Hospital Ethics Committee and performed according to ethical procedures. All patients or their guardians provided written informed consent.

DNA extraction

Total DNA from the tumour tissue samples was prepared using the phenol/chloroform extraction and ethanol precipitation method, whereas total DNA from 2 ml of blood specimens was extracted using a QIAamp DNA blood Midi kit (Qiagen, Shanghai, China). The final DNA was dissolved in sterilized Milli-Q water and quantified using a NanoDrop spectrophotometer (Fisher, Wilmingto, DE, USA).

PCR amplification of the entire mitochondrial genome

Total DNA was subjected to polymerase chain reaction (PCR) amplification using 11 primer sets originally established by Levin et al. (1999). These primer sets amplify overlapping DNA fragments ranging from 982 to 2134 bp in length that span the complete sequence of the human mitochondrial genome. PCR was performed on a GeneAmp® PCR 9700 DNA Thermal Cycler (Applied Biosystems, Foster City, CA, USA) in a volume of 50-μl reaction mixture containing 100 ng DNA template, 200 μM of each dNTP, 20 pmol of each primer, 10 μl 5 × Phusion HF buffer (Finnzymes) and 2 U Phusion high-fidelity DNA polymerase (Finnzymes). The PCR condition consisted of an initial incubation at 95 °C for 5 min, 30 cycles of 30 s at 95 °C, 30 s at 58 °C and 2 min at 72 °C and a final extension step at 72 °C for 10 min. Amplified PCR products were checked by 0.8% agarose gel electrophoresis and purified by a QIAquick PCR purification kit (Qiagen). In each run, total DNA isolated from mtDNA-deficient T47D breast cancer ρ⁰ cells (Yu et al. 2007) was employed as a negative control to ensure the specific amplification of mtDNA, instead of mitochondrial pseudogenes present in nDNA.

Nucleotide sequencing analysis

Bidirectional sequencing of the purified PCR products was performed using the same procedure as we described previously (Yu et al. 2009). The results of DNA sequencing analysis were compared with the published revised Cambridge sequence in the MITOMAP database (http://www.mitomap.org) using the MegAlin program of the DNAstar software package. Any sequence differences between tumour and matched peripheral blood samples were scored as somatic mutations. The level of heteroplasmy of individual somatic mtDNA mutation was determined according to the method described by Tseng et al. (2011). All somatic mutations found were validated at least twice by additional independent PCR experiments and sequencing.

Statistical analysis

The statistical analysis was carried out with the Statistical Package for the Social Sciences (spss) software package 11.5 (SPSS Inc., Chicago, IL, USA) for Microsoft Windows. The association between somatic mtDNA mutations and clinicopathological parameters of osteosarcoma was examined using the Fisher's exact test. Statistical significance was set at P < 0.05.

Results

Somatic mtDNA mutations in osteosarcoma

Pairwise comparison between tumour and corresponding blood mtDNA showed that 10 of 31 patients with osteosarcoma (32.3%) displayed somatic mutations in the coding regions of the mitochondrial genome, with a total number of 11 mutations. The eleven screened somatic mutations included nine point mutations (C587T, A3411T, T5344C, G7565A, A9000C, G9182A, G12773A, C13977T and T14909A), one single-base deletion (9485delC) and one mononucleotide deletion (8238insA) (Table 1). All but two of the somatic changes were present in heteroplasmic form. In combination with the 29 somatic mtDNA mutations previously identified within the D-loop region in the same patient cohort (Yu et al. 2012a), we found that 71.0% (22/31) of cancerous tissues harboured at least one somatic mtDNA mutation. Of the 40 detected somatic mutations, 29 (72.5%) were located in the D-loop regulatory segment, two (5.0%) were in the tRNA coding regions, two (5.0%) were in the ATP synthase 6 (ATPase 6) gene and 7 (17.5%) were situated in the mRNA genes encoding components of mitochondrial respiratory complexes (Table 1). Eight patients had somatic mutations in both D-loop and coding regions of the mitochondrial genome.

Table 1.

Somatic mutations identified in the D-loop and coding regions of mtDNA from 31 osteosarcoma patients

	D-loop mutations*		Mutations in the coding regions
Patient code	Position	Mutation†	Position	Mutation†	Gene	Amino acid change
1	16,293	A→G
	150	C→T
3	544	C→T	7,565	G→A	tRNAAsp	No
			9,000	A→C	ATPase 6	No
4	16,257	C→T
	303–309	8C→9C
5	16,319	G→A
7	16,223	C→T	12,773	G→A	ND5	Glycine→Aspartic acid
8	16,132	delA	3,411	A→T	ND1	Lysine→Asparagine
	303–309	7C→9C
	548	C→T
9	189	A→G
11			9,485	delC	COIII	Frameshift
12	303–309	8C→7C
13	303–309	9C→8C
14	207	G→A	5,344	T→C	ND2	Phenylalanine→Serine
16	73	A→G
	146	T→C
18	16,362	T→C	8,238	InsA	COII	Frameshift
	303–309	7C→8C
19	303–309	9C→8C
21	16,217	C→T	9,182	G→A	ATPase 6	Serine→Asparagine
23	16,293	A→G	587	C→T	tRNAPhe	No
	263	A→G
24	303–309	9C→8C
26	303–309	8C→7C
27	152	T→C
28			14,909	T→A	CytB	Tyrosine→ Asparagine
30	16,146	A→G
	16,183	A→C
	303–309	9C→8C
31	303–309	7C→8C	13,977	C→T	ND5	No

^*: ref Yu et al. (2012a);

^†: Peripheral blood samples → tumour tissues.

COII, cytochrome c oxidase subunit II; COIII, cytochrome c oxidase subunit III; CytB, cytochrome b; ND1, NADH dehydrogenase subunit 1; ND2, NADH dehydrogenase subunit 2; ND5, NADH dehydrogenase subunit 5.

Potential functional consequences of somatic mtDNA mutations screened in osteosarcoma

Nine somatic mutations (81.8%) amongst the eleven mutations discovered within the protein coding sequences were missense or frameshift mutations that have potentials of inducing persistent mitochondrial defects and changing the capacity of mitochondrial respiratory function. The A3411T point mutation (81% heteroplasmy, patient 8) present in the NADH dehydrogenase subunit 1 (ND1) gene causes a substitution of amino acid residue from lysine to asparagine. The T5344C mutation (60% heteroplasmy, patient 14) results in a substitution from phenylalanine to serine in the ND2 polypeptide. The homoplasmic G9182A transition (patient 21) occurring in the ATPase 6 gene leads to a serine-to-asparagine amino acid replacement. Moreover, the G12773A (homoplasmic, patient 7) and T14909A (54% heteroplasmy, patient 28) mutations induce amino acid replacement from glycine to aspartic acid and tyrosine to asparagine in the ND5 and cytochrome b (CytB) subunits respectively. These five missense point mutations may presumably have negative effects on mitochondrial respiratory function and OXPHOS activity because evolutionary comparisons indicated that they were all placed at highly conserved sites of the affected mitochondrial genes coding for subunits of the electron transport chain (Figure 1a).

Figure 1.

Somatic mutations in the coding regions of mtDNA in osteosarcoma. The A3411T (NADH dehydrogenase subunit 1, ND1), T5344C (ND2), G9182A (ATPase 6), G12733A (ND5) and T14909A (cytochrome b, CytB) missense mutations result in amino acid substitutions from lysine to asparagine, phenylalanine to serine, serine to asparagine, glycine to aspartic acid and tyrosine to asparagine at highly evolutionarily conserved amino acid residues of their respective affected polypeptide constituents of respiratory enzyme complexes (up to six vertebrate sequences are aligned, a). The 8238delA mutation found in the COII (cytochrome c oxidase subunit II) gene leads to a frameshift variation and creates a premature termination of protein synthesis, whereas the frameshift 12418insA mutation in the COIII gene results in the truncated polypeptide of 106 amino acid residues (b). The C587T and G7566A mutations, which were respectively identified in the DHU stem of tRNA^Phe and the TψC stem of tRNA^Asp, might have a possible harmful effect on the tRNA stability via interfering with their secondary or tertiary structure (c). GenBank accession numbers for the mitochondrial genome sequences of human, chimpanzee, dog, mouse, rat and zebrafish are J01415, GU112741, EU789787, FJ803909, GU024175 and AC024175.

The 8238insA mutation (42% heteroplasmy, patient 18) is a frameshift alteration in the cytochrome c oxidase subunit II (COII) gene. This chain-terminating mutation results in an AGG stop codon and thus gives rise to premature termination of COII protein synthesis (Figure 1b). The 9485delC mutation (35% heteroplasmy) detected in patient 11 is a single nucleotide deletion situated in a stretch of nine T's beginning from nucleotide position 9478 in the COIII gene. This mutational change elicits a shift in the coding frame, thereby producing a truncated polypeptide of 106 amino acid residues (Figure 1b). As these two mutations disrupt proper production of subunits that are key components of respiratory enzyme complexes, they may potentially impair the electron transport chain activity and mitochondrial respiratory function.

The C587T mutation (73% heteroplasmy) in the tRNA^Phe gene was found in patient 3. This non-coding mutation interrupts a conserved Watson–Crick base pairing in the DHU stem of tRNA^Phe and could be expected to modify the stability or the secondary/tertiary structure of the clover leaf of the DHU loop (Figure 1c). Likewise, the G7565A mutation in the tRNA^Asp gene (29% heteroplasmy, patient 23) disturbs a conserved Watson–Crick base pair in the TψC stem of tRNA^Asp with a harmful possibility of altering the secondary or tertiary structure of the clover leaf of the TψC loop (Figure 1c).

Somatic mtDNA mutations and clinicopathological features of osteosarcoma

We also sought to probe the possible correlation between somatic mtDNA mutations and various clinicopathological parameters of osteosarcoma in our series. However, no significant relationship was identified between the presence of somatic mtDNA mutations and various clinicopathological variables of osteosarcoma (P > 0.05, data not shown), including age, gender, primary tumour site, tumour size, histological subtype and tumour metastasis.

Discussion

Tumour development and progression are at least partially ascribed to the accumulation of various genetic and epigenetic alterations in the nuclear genome, such as abnormalities in oncogenes or tumour suppressors and loss of heterozygosity. For example, somatic variations occurring in multiple canonical tumour-promoting genes, such as c-MYC, c-MET, p53 and RB, have been postulated to be critically implicated in stimulating the oncogenesis of osteosarcoma as well as the maintenance of the malignant phenotype during tumour progression (Oda et al. 2000; Letson & Muro-Cacho 2001; Cui et al. 2011; Martin et al. 2012). Nonetheless, whether certain sequence variations in the mitochondrial genome are also in association with the pathophysiology of osteosarcoma has not been clarified. A considerable body of literature has pointed out that somatic mtDNA point mutations occur at a fairly high frequency in human cancers and approximately 60% of tumours carry such mutations (Yu 2012). The experimental data obtained in this work, combined with our previous investigation in the D-loop region (Yu et al. 2012a), showed that 72.5% (29/40) of somatic mutations in osteosarcoma are in the D-loop, 5% (2/40) are in the tRNA genes and the remaining 22.5% (9/40) are in the mRNA genes contributing products to the electron transport process. Furthermore, the occurrence of somatic mutations in the mtDNA coding regions of osteosarcoma (11/40, 27.5%) was markedly lower than that found in the mutational ‘hotspot’ D-loop region (29/40, 72.5%). The general distribution pattern of these mutations across the whole mitochondrial genome is in agreement with earlier observations in several other forms of cancer (Yu 2012). Given that the appearance of somatic mtDNA mutation is not tightly correlated with distinct clinicopathological factors of osteosarcoma, it is conceivable that these mutations may evolve independently of distinct clinicopathological features and are likely to represent a general phenomenon occurring in the tumorigenic process of osteosarcoma.

In our study population, we found that 55.2% (16/29) of the mutations in the D-loop region and 36.4% (4/11) of the mutations in the coding regions emerged at short base-repetitive sequences (e.g. poly-A, poly-C and poly-T). One possible reason for this intriguing phenomenon is that these mutations are seemingly originated from slippage and/or mis-incorporation due to a deficient activity of mtDNA polymerase γ during mtDNA replication and repair. It has been documented that mtDNA polymerase γ is highly prone to oxidative damage, and a repression of its repair capacity may invoke an increased occurrence of mtDNA mutations (Graziewicz et al. 2002; Longley et al. 2005). This parallels a recent study from Singh's group showing that mtDNA polymerase γ frequently carried mutations in primary breast tumour samples and breast cancer cell lines (MCF7 and MDA-MB-231), and these mutations were accompanied by reduced mitochondrial activity and an augmented ROS leakage, which could further induce mtDNA mutation and boost tumorigenesis (Singh et al. 2009). Moreover, Mambo and colleagues provided evidence that the damage profile of the mtDNA was region dependent, and the D-loop region, in particular the polycytidine stretch at the nucleotide positions 303-309, is considerably more prone to DNA-damaging agents than other areas of mtDNA probably owing to its inefficient repair mechanisms (Mambo et al. 2003). Collectively, taking into account the fact that sequence alteration(s) in the mono- or dinucleotide repeat tracts is one of the striking features of genomic instability, we assume that there might exist a close link between somatic mtDNA mutations and mtDNA instability in osteosarcoma, similar to the findings demonstrated in other tumour types including Ewing's sarcoma (Lee et al. 2005; Tseng et al. 2011; Yu et al. 2012b). Enhanced oxidative stress in the mitochondria may be a likely contributing factor amongst others accounting for the generation of these mutations and the induction of mtDNA instability in osteosarcoma.

According to the guideline recommended by Bandelt et al. (2009), we checked the MITOMAP, mtDB (http://www.mtdb.igp.uu.se) and GenBank (http://www.ncbi.nlm.nih.gov/genbank) as well as conducted more straightforward Google and Yahoo searches for all the screened mutations within the coding regions and found that two of them, G9182A in ATPase 6 and 9485delC in COIII, were respectively detected as somatic mtDNA mutations in oesophageal cancer and oral cancer (Tan et al. 2003, 2006). However, functional evaluation of these point mutational changes and their consequential effects on malignant phenotype of osteosarcoma cells have not been executed so far. To our best knowledge, the majority of somatic mutations (C587T, A3411T, T5344C, G7565A, A9000C, G12773A, C13977T and 8238insA) that we found in our samples have not hitherto been observed in human neoplastic tissues. Some of these newly identified point mutations, if not all, perhaps could impose a profound effect on the OXPHOS system in osteosarcoma cells because they have potentials to either lead to the transition of highly evolutionarily conserved amino acid residues or cause tRNA structural instability. Additionally, the 8238insA mutation, which results in a deleted COIII subunit consisting of 106 amino acid residues, is also likely to compromise the activity of the respiratory chain complex IV and trigger mitochondrial dysfunction, eventually contributing to tumour development. Indeed, introduction of pathogenic T8993G mutation in the ATPase 6 gene into PC3 cancer cell line through trans-mitochondrial cybrid transfer was able to provide a substantial proliferative advantage of the resulting cybrids both in vitro and in vivo (Petros et al. 2005). Similarly, the frameshift 12418 insA mutation in the ND5 gene with was shown to decrease respiratory function of trans-mitochondrial osteosarcoma 143B cells. The cybrids containing the heteroplasmic 12418insA mtDNA exhibited significantly increased tumour growth (Park et al. 2009). The question of whether these novel mutations have a primary and causative link to the onset and development of osteosarcoma or simply represent a secondary bystander effect mirroring nuclear genomic instability during tumorigenesis awaits to be elucidated.

In summary, our current data, plus previous findings in the non-coding part of mtDNA, suggest that a high frequency of somatic alterations occur throughout the entire mtDNA in osteosarcoma. Some of these acquired mutations were functional relevant mutations in tRNA and polypeptide-coding genes with potential detrimental impacts on the mitochondrial OXPHOS system. These experimental results add further support to the concept that qualitative and quantitative mtDNA variations may be actively involved in oncocytic transformation and/or progression of osteosarcoma. Large population-based studies ideally from different ethnic background are definitely needed to confirm the findings in this study. Additionally, we are at present in the process of identifying potential pathogenic mutations using phylogenetic tools (Bandelt et al. 2001; Salas et al. 2005). In an aim to dissect possible functional significance of these somatic mtDNA mutations in a common nuclear background, we will establish several trans-mitochondrial cybrids (Kaipparettu et al. 2010) by fusing the nuclear donor human osteosarcoma 143B ρ⁰ cells with enucleated cytoplast (the mitochondrial donor) derived from patients' neoplastic cells carrying specific potentially pathogenic mutations. These functional analyses may eventually open a window for a better understanding of these mutations in the multistep development and progression of osteosarcoma.

Funding source

This study was supported in part by an internal research grant (to Z.Q.) and a Chinese government award (to M.Y.).

Conflict of interest

The authors have no conflict of interest of any kind related to the work presented in this article.