Heterozygous SSBP1 start loss mutation co-segregates with hearing loss and the m.1555A>G mtDNA variant in a large multigenerational family

The m.1555A>G mitochondrial DNA variant causes maternally inherited deafness, but shows highly variable clinical penetrance. Using exome sequencing, Kullar et al. identify a hypomorphic mutation in SSBP1 that segregates with hearing loss in a family transmitting m.1555A>G, and serves as a trans-acting genetic modifier of clinical penetrance.


Introduction
Mitochondrial dysfunction causes hearing loss in isolation and as a feature of multi-systemic mitochondrial disease. The mitochondrial variant m.1555A4G in the 12S ribosomal RNA gene MTRNR1, is present at an estimated 1 in 385 (0.26%) of the European population, and is necessary but not sufficient to cause maternally inherited deafness (Prezant et al., 1993;Rahman et al., 2012). Aminoglycosides are a recognized modifier factor but cannot account for all hearing-impaired carriers in multigenerational pedigrees, implicating additional co-segregating genetic factors (Bykhovskaya et al., 1998;Guan et al., 2006).
Here, we report a multi-generational family where a heterozygous start loss mutation in the core mitochondrial DNA (mtDNA) replisome protein gene, single stranded binding protein 1 (SSBP1), co-segregated with the m.1555A4G variant and the phenotype. This provides an explanation for the variable clinical penetrance of the disorder.

Materials and methods Patients
Forty-six individuals (21 female: 25 male) carrying the m.1555A4G mtDNA variant from Northern Finland (Fig.  1A) were previously described (Hakli et al., 2013). Individuals in Generation IV (nine females: 10 males) either have normal hearing (n = 9) or sensorineural hearing loss [n = 10, moderate high frequency (2-8 kHz) hearing loss n = 3, moderate pan-frequency (0.25-8 kHz) hearing loss n = 6, profound pan-frequency hearing loss n = 1]. The mean age of hearing loss diagnosis was 3.7 years (range 1.6-5.4 years). There was no history of aminoglycoside usage. Neurological examination of all individuals was otherwise normal. There was no clinical evidence of either proximal or distal myopathy. DNA was available from 25 individuals from Generation III and IV. We subsequently also identified the children of two fathers in Generation III [Subject III-6 (P1), father in Family D, n = 9 children; and Subject III-10 (P4), father in Family E, n = 6 children; mean age of children = 10.3 years (range 1-19 years); Fig. 1A(i)]. Fibroblast cell lines were established from Subjects III-6 (P1) and III-5 (P2). We also studied fibroblasts and DNA from an unrelated individual carrying m.1555A4G (Subject P3). Skeletal muscle biopsy was obtained from Subjects III-10 (P4, age 38 years) and III-8 (P5, age 41 years).

Molecular genetics
Exome sequencing, variant calling and filtering to isolate heterozygous candidate variants with a minor allele frequency (MAF) 51% were performed as previously described (Keogh et al., 2015). Sanger sequencing was performed to confirm segregation of the SSBP1 variant. Pyrosequencing (Qiagen) was undertaken for allelic quantification of m.1555A4G. Long-range PCR was used to detect mtDNA deletions using primers (D1R m.19-1 and D2F m.1650-1671) amplifying a product encompassing almost the complete mtDNA. mtDNA copy number was determined using real time PCR TaqMan Õ assays targeting MT-ND1 or MT-CO3 and the nuclear genes, B2M or RNaseP as described (Grady et al., 2014). Total DNA was prepared by phenol-chloroform extraction and precipitation for 7S DNA analysis, then linearized by PvuII digestion before electrophoresis on 0.7% agarose gels and Southern hybridization (Kornblum et al., 2013). RNA was extracted from fibroblast cell lines using RNeasy Õ Mini Kit (Qiagen) and cDNA was synthesized using High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems). Quantification of gene expression was performed using the TaqMan Õ Gene Expression Assay using transcript specific primers for SSBP1 and MT-CYB with normalization to GAPDH.

Intra-mitochondrial translation assay
Metabolic labelling of mitochondrial proteins was performed as previously described (Van Haute et al., 2016). Loading was assessed by western blotting for TOMM20.

SSBP1 protein levels
Western blotting revealed decreased steady-state SSBP1 levels in fibroblasts from Subjects P1 and P2 compared to the m.1555A4G cell line (Subject P3) and controls (Subjects C1-3) (Fig. 1E). No significant difference in the mRNA expression of SSBP1 was found between patient and control fibroblasts (Fig. 1F), consistent with the variant abolishing SSBP1 translation without an effect on transcription.

Muscle and fibroblast mtDNA analysis
Haematoxylin and eosin staining of the muscle biopsy from Subject P5 was unremarkable whereas sequential COX-SDH staining revealed evidence of respiratory chain deficiency with a low number of COX-deficient and COX-intermediate fibres (5/174, 3.4%, Fig. 2A) (Murphy et al., 2012). Long-range PCR of muscle DNA from Subjects P4 and P5 showed the presence of multiple mtDNA deletions (Fig. 2B), and muscle mtDNA copy number was reduced by $60% in Subjects P4 and P5 compared to controls (Fig. 2C), both suggestive of disordered mtDNA maintenance. We did not detect mtDNA copy number changes in patient blood or fibroblast DNA [ Fig. 3A(i)], thus demonstrating tissue specificity.

Respiration and growth analysis
Measurement of cellular OCR in patient and control fibroblasts revealed no significant difference in basal OCR but a trend of lower maximal respiration in Subjects P1, P2 and P3 compared to controls (Fig. 2D). Subject P1, P2 and P3 cells had significantly greater doubling times than controls when galactose was used as a carbon source (Fig. 2E and F).

7S DNA analysis
Mammalian mtDNA molecules contain a triple-stranded region (D-loop) found in the major non-coding of many mitochondrial genomes, formed by stable incorporation of a third, short DNA strand known as 7S DNA. The exact function of 7S DNA is unknown; however, it has been proposed to play a role in replication as an intermediate of prematurely-terminated heavy (H-) strand synthesis and moreover, perturbations in the steady-state levels of 7S DNA have been observed in a mtDNA maintenance disorder (Kornblum et al., 2013;Nicholls and Minczuk, 2014). Previous work has shown that SSBP1 is required for mtDNA replication and regulates the mtDNA D-loop by modulating the synthesis of 7S DNA (Ruhanen et al., 2010). We analysed the abundance of 7S DNA relative to genome length mtDNA molecules in patient and control fibroblasts. Subjects P1 and P2 had significantly reduced 7S DNA abundance compared to Subject P3 and controls. There was no difference in the level of full-length mtDNA relative to nuclear 28S DNA in Southern blots, confirming the lack of difference in mtDNA copy number in Subject P1, P2 and P3 fibroblasts compared to controls [ Fig. 3A and A(i)].

Mitochondrial mtDNA transcription and protein synthesis in fibroblasts
Measurement of de novo intra-mitochondrial protein translation by incorporation of 35 S radiolabelled methionine revealed markedly reduced global mitochondrial protein synthesis in Subject P1, P2 and P3 cells compared to controls [ Fig. 3B and B(i)]. Reduction in protein synthesis in patient cells was accompanied by a reduction in the steady state levels of mtDNA-encoded complex IV subunits, P2 and P3 [Fig. 3D and D(i)]. There was no difference in steady state levels of nuclear encoded mitochondrial proteins ATP5A and SDHA (Fig. 3D) or in MT-CYB mRNA expression (Fig. 3C). Together these findings reveal a defect in mitochondrial translation mediated by m.1555A4G that was no more severe in the presence of the SSBP1 variant.

Discussion
Here we describe a heterozygous start loss mutation in SSBP1 co-segregating with hearing loss in a maternal pedigree transmitting the m.1555A4G variant. The background frequency of m.1555A4G in northern Finland has been estimated to be significantly less than in other European populations (0.0047% versus 0.26%, Fisher's exact P 4 0.001) reducing the likelihood that our observations are the result of a chance finding (Lehtonen et al., 2000;Rahman et al., 2012).
The SSBP1 mutation reduced SSBP1 levels, decreased 7S DNA in fibroblasts, and was associated with multiple deletions of mtDNA and mtDNA depletion in skeletal muscle. Fibroblasts from these patients also showed reduced intra-mitochondrial protein synthesis in keeping with the co-existing m.1555A4G variant, leading to reduced proliferation rates under conditions of forced mitochondrial respiration (Guan et al., 1996).
Patients with defects in POLG, another mitochondrial maintenance gene encoding the mitochondrial DNA polymerase, pol, are thought to accumulate mtDNA deletions by replication stalling at homopolymeric tracts. It has been proposed that SSBP1 reduces arrests within these tracts, and thus suppresses mtDNA deletion formation (Mikhailov and Bogenhagen, 1996). Given that SSBP1 also coats the H-strand during replication, low SSBP1 levels in our patients may increase mtDNA replication stalling and non-specific replication initiation, compromising replication fidelity leading to tissue-specific mtDNA deletion and depletion (Miralles Fuste et al., 2014). The observed reduction in 7S DNA in Subjects P1 and P2 fibroblasts is in keeping with dysfunctional mtDNA replication. The absence of detectable mtDNA deletions in fibroblasts is well recognized in mtDNA maintenance disorders (Stewart et al., 2011), probably because the deletions are rapidly lost in rapidly dividing cells.
Of note, we specifically searched for a common variant in TRMU (c.28G4T p.A10S, gnomAD frequency = 0.097), that has previously been suggested to modify the phenotype in a subset of m.1555A4G carriers (Guan et al., 2006;Meng et al., 2017). Reflecting this allele frequency, the variant was found in both hearing impaired (3/8, 37.5%) and normal hearing (2/5, 40%) individuals in the family, so cannot account for the phenotype in our patients. Similarly, the SSBP1 variant is unlikely to be pathogenic in isolation given that 10/15 (67%), of the children in Families D and E carry the SSBP1 variant in conjunction with wild-type m.1555A and all have normal hearing. In addition, the heterozygous knockout mouse Ssbp1 tm1a(KOMP)Wtsi displays only a mild phenotype, with hearing no different to wildtype littermates (Brown and Moore, 2012).
Our findings are therefore consistent with an additive effect of the SSBP1 mutation and m.1555A4G, with a combined effect on mtDNA translation and mtDNA maintenance causing a tissue-specific phenotype. Although there is both microscopic and molecular evidence of muscle disease, the patients do not display overt clinical signs of myopathy. The clinical features in these individuals were limited solely to the auditory system.
Taken together, these data suggest rare trans-acting alleles are important modifiers of the m.1555A4G phenotype; this should be taken into consideration for appropriate genetic counselling of carriers and their families.