Validation of Next-Generation Sequencing of Entire Mitochondrial Genomes and the Diversity of Mitochondrial DNA Mutations in Oral Squamous Cell Carcinoma

Abstract

Background: Oral squamous cell carcinoma (OSCC) is mainly caused by smoking and alcohol abuse and shows a five-year survival rate of ~50%. We aimed to explore the variation of somatic mitochondrial DNA (mtDNA) mutations in primary oral tumors, recurrences and metastases.

Methods: We performed an in-depth validation of mtDNA next-generation sequencing (NGS) on an Illumina HiSeq 2500 platform for its application to cancer tissues, with the goal to detect low-level heteroplasmies and to avoid artifacts. Therefore we genotyped the mitochondrial genome (16.6 kb) from 85 tissue samples (tumors, recurrences, resection edges, metastases and blood) collected from 28 prospectively recruited OSCC patients applying both Sanger sequencing and high-coverage NGS (~35,000 reads per base).

Results: We observed a strong correlation between Sanger sequencing and NGS in estimating the mixture ratio of heteroplasmies (r = 0.99; p<0.001). Non-synonymous heteroplasmic variants were enriched among cancerous tissues. The proportions of somatic and inherited variants in a given gene region were strongly correlated (r = 0.85; p<0.001). Half of the patients shared mutations between benign and cancerous tissue samples. Low level heteroplasmies (<10%) were more frequent in benign samples compared to tumor samples, where heteroplasmies >10% were predominant. Four out of six patients who developed a local tumor recurrence showed mutations in the recurrence that had also been observed in the primary tumor. Three out of five patients, who had tumor metastases in the lymph nodes of their necks, shared mtDNA mutations between primary tumors and lymph node metastases. The percentage of mutation heteroplasmy increased from the primary tumor to lymph node metastases.

Conclusions: We conclude that Sanger sequencing is valid for heteroplasmy quantification for heteroplasmies ≥10% and that NGS is capable of reliably detecting and quantifying heteroplasmies down to the 1%-level. The finding of shared mutations between primary tumors, recurrences and metastasis indicates a clonal origin of malignant cells in oral cancer.

Fig 1. Percentage of the minor component on heteroplasmic positions as obtained with Sanger sequencing and next-generation sequencing on an Illumina HiSeq platform.

The samples contained pre-defined mixtures of the DNA of two lab members. In Sanger electropherograms, heteroplasmies were neither detected in the 1+49 mixture nor in the 1+99 mixture.

Fig 2. Bland-Altman plot for depicting the agreement between NGS and Sanger sequencing heteroplasmy measurements.

On the x-axis, the average of next-generation and Sanger sequencing heteroplasmy estimates is plotted. On the y-axis, the difference between next-generation and Sanger sequencing heteroplasmy estimates is plotted. The mean difference is indicated as green line, the 95% limits of agreement (average difference ± 1.96 standard deviation of the difference) are indicated as red dotted lines. For heteroplasmies 10%, which were detected with NGS, but could not be found in Sanger electropherograms, the value for the Sanger estimation was set to $\frac{10}{\sqrt{2}}$.

Fig 3. Phylogenetic tree representing the benign profiles of 28 oral cancer patients and seven team members.

The sequences are stored in GenBank (accession numbers KC286583 –KC286589 corresponding to Lab005 –Lab011 and KC286590 –KC286617 corresponding to MKG01 –MKG28). The position of the rCRS is indicated for reading off sequence motifs. Mutations are transitions unless a base is explicitly indicated. The prefix ‘‘@” designates reversions, whereas suffixes indicate transversions (to A, G, C, or T), indels (.1, d), and heteroplasmies (R, Y). Point heteroplasmic mutations that were observed in benign tissue samples are highlighted in grey.

Fig 4. Distribution of NGS-typed heteroplasmies across the mitochondrial genome.

Frequencies were calculated as number of heteroplasmies per gene region divided by the total number of observed heteroplasmies. The data from our study were compared with data from head and neck cancers and with all cancer samples from Ju et al. [27].

Fig 5. Heteroplasmic mutations as seen with Sanger sequencing in MKG05 and MKG15.

The benign tissue was taken from the resection border of the tumor.

Fig 6. Heteroplasmic mutations as seen with Sanger sequencing in MKG20, MKG21 and MKG27.

The benign tissue was taken from the resection border of the tumor.

Fig 7. Distribution of point heteroplasmy levels in tumor and benign tissue samples.