Deep sequencing of mitochondrial DNA and characterization of a novel POLG mutation in a patient with arPEO

Morphologic analyses demonstrate mitochondrial myopathy

A muscle biopsy from the deltoid muscle at age 66 years showed slight variability in muscle fiber size, scattered ragged red fibers, and 10%–20% fibers with cytochrome c oxidase deficiency (figure 1, A and B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Morphologic analysis, pedigree, and molecular genetics

(A) The muscle biopsy from the patient showed slight variability in muscle fiber size and scattered ragged red fibers (arrows) (Gomori trichrome staining). (B) Several fibers show cytochrome c oxidase (COX) deficiency with normal succinate dehydrogenase (SDH) activity (blue fibers, arrows), whereas a few fibers with mitochondrial proliferation show no COX deficiency (arrowhead) (combined enzyme histochemical staining of COX and SDH). (C) Pedigree of the family: filled symbol indicates affected individual with biallelic variants in POLG, and half-filled symbols indicate healthy heterozygous carriers. M1 indicates the novel variant c.590T>C; p.(F197S) and M2 c.2740A>C; p.(T914P) in POLG (NM_002693.2). (D) Illustration of the novel c.590T>C. p.(F197S) variant in POLG, with chromatogram. (E) Illustration showing the evolutionary conservation of amino acids. Blue bar indicates the mutated residue (p.Phe197Ser).

Molecular genetic analysis identifies POLG variants

ES revealed that the patient had 2 potentially pathogenic heterozygous missense mutations in the gene coding for the catalytic subunit of DNA polymerase gamma (POLG), the novel c.590T>C p.(F197S), and the previously described c.2740A>C; p.(T914P) (NM_002693.2) (figure 1, C and D). The variants were not common in the human population and affected conserved amino acids among species (figure 1E). Genetic analysis of the sons revealed that both were heterozygous carriers for the c.2740A>C; p.(T914P) variant (figure 1C), demonstrating that the mutations are located on different alleles.

The F197S mutant has reduced 3′-5′ exonuclease activity

To further understand the molecular consequences of the mutations, we expressed and purified WT, F197S, and T914P POLγA for biochemical analysis. We previously showed that T914P had no DNA binding affinity.⁷ Given the patient's relatively mild condition, we predicted the F197S mutation to have no impact on DNA binding affinity, which we confirmed using electrophoretic mobility shift assay (figure 2A).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2 The F197S point mutation reduces 3′-5′ exonuclease activity

(A) The F197S mutant has WT-like DNA binding affinity. DNA affinity of POLγ to a primer template (lower panel) was examined by electrophoretic mobility shift assay. Each reaction contained 10 fmol of DNA template and increasing (indicated) amounts of POLγ holoenzyme in a final volume of 15 µL. The reactions were run on 6% native polyacrylamide gels in 0.5 × TBE for 30 minutes at 180 V and visualized by autoradiography. No major differences in binding affinity between WT and F197S were observed. (B) The F197S point mutation disrupts the architecture of the exonuclease pocket. F197 is located in a large hydrophobic pocket directly adjacent to the catalytically essential D198 (upper panel; PDB ID: 3IKM). This residue is strongly conserved among members of the DNA polymerase A family (lower panel; Sc = Saccharomyces cerevisiae; Dm = Drosophila melanogaster; Xl = Xenopus laevis; Hs = Homo sapiens; Mm = Mus musculus). (C) The T7 exonuclease pocket is substantially smaller to accommodate a serine residue at the position equivalent to F197. Mutation of this residue to phenylalanine results in steric clashes, which would likely destabilize the pocket (PDB ID: 1T7P; Hs = Homo sapiens; T7 = bacteriophage T7). (D) A 5′-labeled (³²P) primer with a 3′-mismatch was annealed to a circular single-stranded plasmid (lower panel to the left). POLγA with 3′-5′ exonuclease activity will, in the absence of dNTP, hydrolyze DNA into shorter fragments. As expected, WT degraded the primer (lanes 1–5), whereas EXO- could not due to lack of exonuclease activity (lanes 6–10). T914P showed no activity. The F197S mutant had weak exonuclease activity, only able to hydrolyze a few nucleotides under these conditions. The arrow indicates the partially hydrolyzed primer.

We next considered whether the Phe-197-Ser substitution affected 3′-5′ exonuclease activity. F197 is strongly conserved in the DNA polymerase A family across vertebrates and invertebrates (figure 2B, lower panel). In the 3′-5′ exonuclease domain, the F197 residue is located within a large hydrophobic pocket where it forms numerous van der Waals interactions with the surrounding residues (figure 2B, upper panel). Substitution of the bulky and hydrophobic phenylalanine with the smaller and polar serine would likely cause partial collapse of this pocket and local structural distortions. This may induce a misalignment of the neighboring D198 residue and subsequent reduction in its ability to bind a catalytically essential Mg²⁺ ion. Of interest, the closely related T7 DNA polymerase carries a serine at this position (figure 2C, lower panel) arguing that this amino acid change might not provoke structural or functional changes. Structural analysis, however, revealed that the exonuclease pocket is substantially smaller and that the presence of a phenylalanine would induce severe steric clashes and likely destabilize the exonuclease domain in T7 DNA polymerase (figure 2C, upper panel).

Despite being tolerated in T7 DNA polymerase, we decided to test whether the serine affects the 3′-5′ exonuclease activity of POLγA in vitro. A double-stranded template with a one-nt 3′-flap (figure 2D, lower panel) was incubated with the WT or mutant POLγ variants including a previously characterized exonuclease-deficient mutant D274A (herein EXO-) for comparison.¹¹ As expected, WT POLγ was able to hydrolyze the primer (figure 2D, lanes 1–5), whereas EXO- was unable to degrade the primer (figure 2D, lanes 6–10). F197S was able to degrade the primer, however, to a much lower extent than WT. This demonstrates that in POLγA, unlike the related T7 DNA polymerases, a serine at this position is not well tolerated (figure 2D, lanes 16-20). No 3′-5′ exonuclease activity from T914P was observed (figure 2D, lanes 11–15), likely due to the lack of DNA binding rather than an enzymatic effect of the mutation.⁷

The exonuclease-deficient POLγ (EXO-) in mice causes a 5-fold increase in mtDNA point mutations,¹² prompting us to investigate the mutation load of the patient's mtDNA. We isolated total muscle DNA from the patient and performed deep DNA sequencing. We identified a slight, but significant, increase in mtDNA point mutations in the patient compared with age-matched controls (table), consistent with our in vitro findings. However, the level of heteroplasmy is likely too low to by itself explain the disease phenotype in the patient.

The F197S mutation affects DNA polymerase activity

We next tested the polymerase activities of the mutants. First, we performed a DNA synthesis assay using a circular single-stranded DNA (ssDNA) template (figure 3A, lower panel). Because the F197S mutation is located in the 3′-5′ exonuclease domain, we also included EXO- for comparison. We found that F197S was active as a polymerase, albeit with strongly diminished activity compared with WT and EXO- (figure 3A, lanes 16–20 vs 1–5 and 6–10, respectively). The speed of replication was slower, generating less full-length product due to increased stalling. T914P was unable to support DNA synthesis (figure 3A, lanes 11–15).⁷

• Download figure
• Open in new tab
• Download powerpoint

Figure 3 The F197S mutation affects DNA polymerase activity

(A) DNA synthesis reaction on single-stranded DNA, where POLγA has the possibility to synthesize ≈3,000 nt. WT (lanes 1–5) and EXO- (lanes 6–10) POLγA displayed efficient DNA synthesis. F197S (lanes 16–20) was considerably slower, and T914P (lanes 11–15) was enzymatically dead. The squared bracket shows the positions of disrupted polymerization products. Below: A cartoon showing the template to the left and the fully synthesized product to the right. (B) DNA synthesis reaction on double-stranded DNA, where the replisome has the possibility to produce very long DNA products (>20 kb). The replisome, in the presence of F197S, was considerably slower than in the presence of WT. The T914P mutant was not able to support DNA synthesis. Mixing T914P and F197S (1:1 ratio) together to mimic the patient situation caused an apparent decrease in replication efficiency because only half of the total amount of polymerase (F197S) is active (lanes 16–20).

We also examined DNA synthesis in the context of the replisome (i.e., in the presence of the TWINKLE helicase and mtSSB) using a circular dsDNA template containing a preformed replication fork (figure 3B, lower panel). Here, F197S was able to synthesize long stretches of DNA, although it was not as efficient as WT (figure 3B, compare lanes 11–15 and 1–5). T914P was not able to support DNA synthesis (figure 3A, lanes 11–15 and 3B lanes 6–10).⁷ To mimic the in vivo situation (of the patient), a reaction was included where both disease-associated mutants were mixed at a 1:1 ratio, but keeping the same total amount of protein. This decreased the efficiency of replication further due to only half the amount of POLγA (F197S) being functional (figure 3B, lanes 16–20 vs 11–15).

The F197S mutant causes multiple deletions and duplications in vivo

Disease-causing mutations in POLγ are known to cause mtDNA depletion and/or multiple mtDNA rearrangements. Based on the deep sequencing of isolated total muscle DNA, we found that mtDNA copy number in the patient was comparable to two controls, demonstrating that the patient does not exhibit mtDNA depletion (figure 4, A and C). However, the patient did have multiple deletions in both the major and minor arc (figure 4B). The sum heteroplasmy for the deletions and duplications amounted to 50.7%, which likely explains the patient's phenotype (figure 4D). Of interest, we also found a strong hotspot region for duplications spanning over the promoter region (position 300 nt–1,000 nt) where the top 3 candidate duplications were closely spaced (positions: 310–962, 310–960, and 315–965) (figure 4E) accounting for 21% heteroplasmy (14.7, 2.5, and 3.4% individually) of all deletions/duplications detected.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4 General statistics and distribution of breakpoints in the patient and control samples

(A) Total coverage depth in the nuclear and mitochondrial genome of the samples along with estimated mitochondrial copy number. (B) Mitochondrial DNA (mtDNA) deletions (blue) and duplications (red) predicted from the sequencing data in the patient and control samples where the color intensity reflects the heteroplasmy level. Here, deleted regions are those showing gapped alignment of reads on the circular genome. Duplications are a gapped alignment indicating deletion of a genome fragment with an origin of replication, which is not plausible for a functioning mtDNA molecule. Thus, for duplications, the remaining fragment on the circular genome is considered as duplicated. (C) Coverage depth of sequencing reads along the mitochondrial genome in the patient and control samples. (D) Distribution of heteroplasmy for the detected deletions and duplications and the sum of the heteroplasmy levels for all deletions and duplications in the given samples. (E) Position and heteroplasmy levels of the deletions/duplications detected above 1% threshold heteroplasmy in the patient sample (none detected in the controls).