2.3. Molecular Studies

Total genomic DNA was purified from available tissues (skeletal muscle, blood, skin fibroblasts, and urine) using an automated extractor (Magpurix, Zinexts Life Science Corporation, Taipei, Taiwan). Sequencing of the whole mitochondrial genome was performed in 82/111 subjects using Nextera XT technology (Illumina, San Diego, CA, USA) following the protocol “Human mtDNA Genome”; libraries were sequenced with paired-end 2 × 150 reads on a MiSeq desktop sequencer (Illumina, San Diego, CA, USA). This approach allowed us to sequence the mitochondrial genome at a high depth of coverage (>5000× on average). The clinical significance of the identified mtDNA variants was assessed based on mitochondrial variation databases (Mitomap, www.mitomap.org, accessed on 8 July 2021) and the bioinformatic tools commonly used in our NHS-certified mitochondrial laboratory, such as MitImpact (https://mitimpact.css-mendel.it, accessed on 8 July 2021) and HmtVar (www.hmtvar.uniba.it, accessed on 8 July 2021). In particular, the aligned reads were analyzed for variant calling and heteroplasmy level detection with the Ingenuity Variant Analysis tool (Qiagen, Hilden, Germany). Parameters were the following: minimum basecall quality score >30, analysis threshold at 1%, interpretation threshold at 1%, and minimum read count at 1000× (needed for assessing heteroplasmy). Capillary sequencing confirmed the variants identified in patients and demonstrated segregation in available maternal relatives. Relative read count indicated by Integrative Genomics Viewer software was used to assess levels of heteroplasmy in specific mtDNA variants in tissues and individuals by calculating the percentages of the four nucleotides detected in the same position.

The search for mtDNA deletions (single or multiple) in muscle DNA was performed using a long-template PCR amplification method, amplifying different mtDNA regions using various sets of primers adopting Ranger DNA Polymerase (Meridian Lifescience, Memphis, TN, USA), specifically designed to amplify long genomic DNA templates. A reported qPCR analysis was used to test for the presence of the so-called “common” deletion [21] or to assess gene copy number and define the extent of mtDNA depletion [22]. One limitation is that the method cannot indicate deletions below about 100 bp.

Using methods described elsewhere [23], we analyzed in 27 children the “MitoExome”, a customized multigene panel targeting the coding regions of 1172 genes associated with mitochondrial pathways [24]. Briefly, the panel was designed with the NimbleGen Design software (Roche NimbleGen Inc., Pleasanton, CA, USA), and target enrichment and amplification were performed following the SeqCap EZ HyperCap Library protocol (Roche, Madison, WI, USA), adapted for a MiSeq desktop scanner (Illumina, San Diego, CA, USA). An in-house bioinformatic pipeline, based on the Ingenuity Variant Analysis bioinformatic suite (IVA, Qiagen, Germany), and the web tool Annovar (http://wannovar.usc.edu/, accessed on 8 July 2021) were used to filter variants and prioritize genes/alleles of interest. Data were also prioritized based on minor allele frequencies (defined in gnomAD 2.1, https://gnomad.broadinstitute.org/, accessed on 8 July 2021) and CADD scores (CADD, Combined Annotation Dependent Depletion; https://cadd.gs.washington.edu/snv, accessed on 8 July 2021). Missense variants were also systematically evaluated for the functional consequences in silico using Polyphen2 (http://genetics.bwh.harvard.edu/pph2/, accessed on 8 July 2021) and SIFT (Sorting Intolerant From Tolerant) (http://sift.jcvi.org/, accessed on 8 July 2021) predictions. To evaluate possible deleterious effects of synonymous and missense variants on splicing, we used the Human Splicing Finder web tool (Genomsis, Marseille, France). All sequenced samples had at least 98% of target covered 20× or more.