Electron microscopy

Open skeletal muscle and skin biopsies were performed after informed parental consent. For ultrastructural examination of skeletal muscle and cultured skin fibroblasts, samples were fixed in 2.5% glutaraldehyde buffered with 0.1 M sodium cacodylate (pH 7.2), postfixed in 1% osmium tetroxide, dehydrated in ascending grades of alcohol, processed through propylene oxide and embedded in Epon^™ resin. Ultrathin sections were cut with a diamond knife on a Leica Ultracut UCT Ultramicrotome, placed on copper grids and stained with uranyl acetate and lead citrate. Examination was carried out with a JEOL 1400 transmission electron microscope.

Live cell imaging

Live cell imaging using confocal microscopy was used to visualize tetramethyl rhodamine methyl ester (TMRM)-labelled mitochondria in fibroblasts. Primary skin fibroblasts cultured under standard conditions as described previously (Fassone et al., 2010) were labelled with 25 nM TMRM (Invitrogen) prior to visualizing mitochondria with the use of a Zeiss LSM 700 ×63 oil immersion objective. Identical detector/gain settings were used for all samples. To determine membrane potential, mean fluorescence intensity (background corrected) of TMRM-labelled mitochondria was analysed using ImageJ (n = 3; >27 cells analysed for each n).

Mitochondrial DNA analysis

The whole mitochondrial genome was sequenced in muscle from Patient 1 using Sanger sequencing, and mtDNA copy number per diploid nuclear genome was determined in patient and control fibroblasts by Droplet Digital^™ PCR (Bio-Rad). The primers and probes used in this study have been published previously (Bai and Wong, 2005).

Candidate nuclear gene sequencing

Candidate nuclear genes (DRP1, MFF and FIS1) previously implicated in mitochondrial fission were analysed by Sanger sequencing of DNA extracted from blood from Patients 1 and 2.

Homozygosity mapping

Genome-wide SNP (single nucleotide polymorphism) array was performed using Illumina HumanCytoSNP 12 in Patient 1, Patient 2 and both parents to search for regions of homozygosity by descent that might contain the mutated gene.

Whole exome sequencing and filtering criteria

The whole exome was sequenced in Patients 1 and 2 using the Illumina HiSeq 2000 platform. Sample preparation and enrichment was performed according to Agilent's SureSelect Protocol Version 1.2. Concentration of each library was determined using Agilent's QPCR NGS Library Quantification Kit (G4880A). Samples were pooled prior to sequencing, with each sample at a final concentration of 10 nM. Read files (Fastq) were generated from the sequencing platform via runs to an average 50× coverage on an in-house HiSeq 2000 system (Illumina). Raw fastq files were aligned to the GRCh37 reference genome using novoalign version 2.08.03. Duplicate reads were marked using Picard tools MarkDuplicates. Calling was performed using the haplotype caller module of GATK (https://www.broadinstitute.org/gatk, version 3.1-1), creating gVCF formatted files for each sample. The individual gVCF files for the exomes discussed in this study, in combination with ∼3000 clinical exomes (UCL-exomes consortium), were combined into merged VCF files for each chromosome containing on average 100 samples each. The final variant calling was performed using the GATK GenotypeGVCFs module jointly for all samples (cases and controls). Variant quality scores were then recalibrated according to GATK best practices separately for indels and SNPs. Resulting variants were annotated using ANNOVAR. Candidate variants were filtered based on function (non-synonymous, presumed loss-of-function or splicing) and minor allele frequency (<0.5% minor allele frequency in our internal control group, as well as the NHLBI exome sequencing data set).

STAT2 transduction

Lentiviral transduction of patient fibroblasts with wild-type STAT2 was performed to determine whether the mitochondrial fission defect could be rescued by exogenous wild-type STAT2. The human STAT2 gene was cloned into the self-inactivating HIV1-derived lentiviral vector plasmid as previously described (Demaison et al., 2002). The plasmid uses the spleen focus-forming virus promoter to drive expression of the STAT2. Expression was linked to enhanced green fluorescent protein through an internal ribosomal entry site.

HEK293T cells were cultured in Dulbecco’s modified Eagle medium (Invitrogen) supplemented with 10% foetal calf serum (Sigma). These cells were seeded in T175 flasks and grown overnight to reach 80% confluency. Each flask was transfected with 50 µg vector plasmid, 17.5 µg vesicular stomatitis virus (VSV) envelope plasmid (pMDG), and 32.5 µg gag/pol packaging plasmid (pCMVΔ8.74) using 2.5 nM polyethylenimine. Viral supernatant was harvested 48 and 72 h after transfection, filtered at 0.22 μm and concentrated by ultracentrifugation at 23 000g for 2 h at 4°C using a Beckmann ultracentrifuge. Virus particles were resuspended in Opti-Mem^® and stored at −80°C. The number of viral infectious particles was calculated by flow cytometry to measure GFP in HEK293T cells, 72 h after transduction with serial dilutions of virus.

STAT2 silencing

Silencing the STAT2 gene in SHSY5Y cells was achieved by transducing the cells with the pGIPZ shRNAmir vector, as previously described (Fassone et al., 2010). Stably transduced clones were selected in 1 µg/ml puromycin (Gibco-Invitrogen). STAT2 knockdown efficiency was tested by quantitative PCR and immunoblot analyses.

Transcript expression

Total RNA was extracted from patient and control cells using RNAqueous^®-4PCR (ABI). For each sample, 1 μg of total RNA was reverse transcribed to cDNA using the high capacity RNA to cDNA kit (ABI). Oligonucleotide primers for quantitative real time PCR were designed using the universal probe library (Roche Molecular Biochemicals), and synthesized at Sigma-Genosys. Quantitative real time PCR was performed using the Power SYBR^® green mix (ABI) on a StepOne^™ quantitative PCR machine (ABI). The target genes analysed were DRP1, MFN1, MFN2 and OPA1, with ACTB (encoding β-actin) as a reference gene. Relative quantification of gene expression in patient samples was carried out against control 1 (C1).

Protein expression

For immunodetection of STAT2, DRP1 (and phospho-DRP1) and the fusion proteins MFN1, MFN2 and OPA1, cell pellets of fibroblasts cultured in Dulbecco’s modified Eagle medium containing GlutaMAX^™-I and 50 mM glucose (Life Technologies) were extracted on ice in PBS, 1.5% n-dodecyl-β-d-maltoside. Equal concentrations of protein were denatured in Laemmli sample Buffer and 10 mM dithiotreitol, resolved on 4–15% Mini-Protean^® TGX Stain-Free^™ gels with Precision Plus Protein standards and blotted onto 0.2 µm PVDF membranes with a Trans-blot^® Turbo^™ Transfer System (all from Bio-Rad). Blots were blocked in PBS, 10% skimmed milk powder (Fluka) and incubated overnight with primary antibody. For the phospho-immunoblots, the primary antibodies used were: rabbit anti-Phospho-DRP1 (Ser616), rabbit anti-Phospho-DRP1 (Ser637), rabbit anti-DRP1 (D6C7), rabbit anti-STAT2 (Sigma) and anti-phosphorylated STAT1 antibodies (all from Cell Signaling). For western blots the primary antibodies used were: anti-DRP1 (BD Transduction Laboratories), anti-MFN1 (Abcam), anti-MFN2 (Abcam), anti-OPA1 (BD Transduction Laboratories), anti-MTCO2 (Abcam), anti-TOM20 (Santa Cruz Biotechnology) and anti-β-actin (Abcam). Blots were incubated with the appropriate secondary antibody (polyclonal goat anti-mouse or anti-rabbit IgG/HRP from Dako) and were developed for 5 min with Clarity^™ Western ECL substrates (Bio-Rad) and visualized on a ChemiDoc^™ MP imager (Bio-Rad).

DRP1 localization

Cells were grown on glass chamber slides (Ibidi), fixed with paraformaldehyde and incubated with anti-DRP1 primary antibody (Cell Signaling) overnight and anti-rabbit FITC conjugated secondary antibody (Sigma) for 1 h. Nuclei were stained with Hoechst and mitochondria were stained with anti-TOM20 (Santa Cruz Biotechnology). Cells were viewed under a Zeiss LSM 700 ×63 oil immersion objective using appropriate filters.

Fluorescence-activated cell sorting for detection of STAT1 phosphorylation

To detect STAT1 phosphorylation, native and transduced patient fibroblasts were either left unstimulated or stimulated with 10⁵ units/ml of IFNα (Stratech Scientific Limited). Cells were then lysed, fixed, washed and permeabilized before adding 5 μl anti-STAT1 phosphorylated tyrosine antibody (BD Bioscience) as previously described (Walshe et al., 2009). Ten thousand fibroblasts were acquired and analysed (FACSCalibur and CELLQest software 2.1.1, BD Biosciences). The percentage change in phosphorylated cells was calculated by subtracting the percentage of unstimulated cells from that of the stimulated cells. Experiments were performed in triplicate and expression levels were expressed as mean ± standard error (SE).

Apoptosis assay

Apoptosis was analysed in fibroblasts derived from the STAT2 deficient index patient (Patient 1) and compared to fibroblasts derived from healthy individuals. Cultured fibroblasts were left unstimulated or stimulated with 100 U/ml IFNα (PBL Interferon Source), 10 µg/ml anti-fasL IgM (activating antibody from Millipore) to crosslink the fasL receptor or phytohaemagglutinin (10 µg/ml, BioStat). Five hours post stimulation, cells were stained with Annexin V (Life Technologies) and 7AAD (BD Biosciences). Ten thousand cells were acquired on a FACSCalibur and analysed using CellQuest Pro (BD Biosciences) and the percentage of apoptotic cells defined as annexin V+7AAD+.

Histological analyses

Light microscopy of skeletal muscle was unremarkable but electron microscopy of muscle and fibroblasts from both children revealed abnormally long mitochondria, between 8–10 µm in length (Fig. 1A), suggesting defective mitochondrial fission.

Homozygosity mapping and candidate gene analysis

Genome-wide SNP array revealed several 1 Mb regions of homozygosity shared by the siblings, and an extended region of homozygosity (10.6 Mb) on chromosome 12 (position: 52 648 837–63 282 047). Sanger sequencing excluded mtDNA mutations and mutations in candidate genes known or predicted to cause defective mitochondrial fission (DRP1, MFF and FIS1) (Waterham et al., 2007; Shamseldin et al., 2012; Shen et al., 2014). Absolute quantification of mtDNA copy number per cell showed a 2-fold increase in Patient 1 compared to five paediatric controls [increased from a mean ± standard deviation (SD) of 658 ± 136 in controls to 1310 in Patient 1].

Whole exome analysis

Whole exome sequencing did not demonstrate any pathogenic mutations in known or putative mitochondrial disease genes in these siblings, but did reveal a novel homozygous stop-gain mutation c.1836 C > A (p.Cys612Ter) in STAT2, located within the region of extended homozygosity on chromosome 12.

Molecular and functional characterization of mutated STAT2

The homozygous c.1836 C > A (p.Cys612Ter) mutation, located within the SH2 domain of STAT2 which is critical for transcriptional function, was confirmed by Sanger sequencing and shown to be heterozygous in both parents (Fig. 1B).

STAT2 transduction

Confocal microscopy demonstrated dense, elongated mitochondria in patient fibroblasts compared to controls in 80% confluent cell cultures (Fig. 2) We found that mitochondria from a clinically similar but unrelated patient (Patient 3), previously reported with a homozygous splice mutation in STAT2 (c.381 + 5 G > C), were also elongated and tubular (Fig. 2), confirming the association between STAT2 deficiency and mitochondrial elongation. Furthermore patient cells were transduced with wild-type STAT2 and the delivery of the correct version of STAT2 was confirmed by green fluorescent microscopy (Fig. 3A). Analysing the mitochondrial length using TMRM (Fig. 2) and quantitative methods such as ImageJ (Fig. 3B) and IMARIS X64 software (Fig. 3C) revealed an average 3-fold decrease in mitochondrial length in patient fibroblasts compared to controls following transduction with wild-type STAT2.

STAT2 silencing

To confirm the specific role of STAT2 as a determinant of mitochondrial structure, SHSY5Y cells were transduced with siRNA targeting STAT2. Quantitative analysis showed a 4-fold increase in mitochondrial length in STAT2-knockout SHSY5Y cells compared to wild-type (Fig. 2B and C).

Transcript expression

We observed no difference in expression of genes encoding fission (DNM1L, also known as DRP1) and fusion (MFN1, MFN2 and OPA1) proteins between patient and control fibroblasts (Fig. 4A).

Protein expression

Western blot analysis revealed that STAT2 protein was undetectable in fibroblasts from all three patients Patients 1–3 (Fig. 5A), confirming nonsense mediated decay. We examined the effects of STAT2 deficiency on proteins involved in mitochondrial fission and fusion. Steady-state levels of DRP1 were similar in cytoplasmic protein extracts of cultured fibroblasts from Patient 1 and Patient 3, compared to six control subjects. However, the outer and inner mitochondrial membrane fusion proteins MFN1, MFN2 and OPA1 were increased in Patient 1 and Patient 3 compared to the six control fibroblasts (Fig. 4B). Further analyses demonstrated that the steady-state levels of mitochondrial cytochrome c oxidase subunit II (MT-CO2) and the translocase of outer mitochondrial membrane 20 (TOM20, encoded by TOMM20) were similarly increased in the patients (Fig. 4B).

DRP1 localization

For DRP1 to exert its effects on initiation of mitochondrial fission, it first needs to be localized to the mitochondria. Using confocal microscopy, we examined DRP1 localization in Patient 1 and control fibroblasts, and found DRP1 to be co-localized with the outer mitochondrial membrane protein TOM20 in both patient and control fibroblasts. Thus a deficiency of STAT2 protein does not appear to affect DRP1 localization to the mitochondria (Fig. 4C).

DRP1 phosphorylation

To further investigate the underlying cause of the long mitochondria, we measured activation of DRP1, the main fission factor in human mitochondria. As DRP1 levels and localization were normal in patient cells, we hypothesized that DRP1 might be inactive in patient cells and therefore unable to complete the fission process. DRP1 activity is dually regulated by phosphorylation at two key serine residues. Post-translational modification by phosphorylation at serine residue 616 (P-DRP1^S616) is known to activate DRP1, whereas phosphorylation at serine 637 (P-DRP1^S637) is associated with the inactive state of the GTPase. Reduced P-DRP1^S616 and increased P-DRP1^S637 was observed in all three patients compared to controls (Fig. 5), indicating that DRP1 is inactive in STAT2 deficiency. Introduction of wild-type STAT2 into patient fibroblasts using lentiviral transduction reversed the phenotypes; we observed increased P-DRP1^S616, decreased P-DRP1^S637 and shorter mitochondria, reflecting a state of increased fission. Conversely, knock down of STAT2 in SHSY5Y cells led to increased P-DRP1^S637and reduced P-DRP1^S616 (Fig. 5) and extremely elongated mitochondria (Fig. 2B), recapitulating the phenotype in STAT2-deficient fibroblasts.

STAT1 phosphorylation detected by fluorescence-activated cell sorting

As well as STAT2, phosphorylation of another STAT protein called STAT1, was also disrupted in Patient 1 and Patient 2 fibroblasts compared to controls and recovered significantly after transducing the cells with wild-type STAT2 (Fig. 6A and B), confirming disturbance of the IFNα pathway and explaining the susceptibility to viral infection in these patients.

Apoptosis assay

ImageJ analysis of TMRM stained fibroblasts revealed reduced mitochondrial membrane potential in all three patient fibroblasts Patients 1–3 (Fig. 6C). We therefore investigated apoptosis in Patient 1 cells compared to healthy controls. Induction with phytohaemagglutinin and anti-fasL both induced comparable apoptosis in patient and control cells. However, the STAT2 deficient cells failed to undergo apoptosis in response to IFNα (Fig. 6D). Both phytohaemagglutinin and anti-fasL induction of apoptosis resulted in the activation of caspase 3. These results indicate that there is not an inherent apoptotic defect in these cells but that IFNα induced apoptosis is severely impaired, consistent with previous observations in STAT2 knockdown cell models (Romero-Weaver et al., 2010).