Study of an FBXO7 patient mutation reveals Fbxo7 and PI31 co‐regulate proteasomes and mitochondria

Mutations in FBXO7 have been discovered to be associated with an atypical parkinsonism. We report here a new homozygous missense mutation in a paediatric patient that causes an L250P substitution in the dimerisation domain of Fbxo7. This alteration selectively ablates the Fbxo7‐PI31 interaction and causes a significant reduction in Fbxo7 and PI31 levels in patient cells. Consistent with their association with proteasomes, patient fibroblasts have reduced proteasome activity and proteasome subunits. We also show PI31 interacts with the MiD49/51 fission adaptor proteins, and unexpectedly, PI31 acts to facilitate SCFFbxo7‐mediated ubiquitination of MiD49. The L250P mutation reduces the SCFFbxo7 ligase‐mediated ubiquitination of a subset of its known substrates. Although MiD49/51 expression was reduced in patient cells, there was no effect on the mitochondrial network. However, patient cells show reduced levels of mitochondrial function and mitophagy, higher levels of ROS and are less viable under stress. Our study demonstrates that Fbxo7 and PI31 regulate proteasomes and mitochondria and reveals a new function for PI31 in enhancing the SCFFbxo7 E3 ubiquitin ligase activity.

Mutations in FBXO7 have been discovered to be associated with an atypical parkinsonism.We report here a new homozygous missense mutation in a paediatric patient that causes an L250P substitution in the dimerisation domain of Fbxo7.This alteration selectively ablates the Fbxo7-PI31 interaction and causes a significant reduction in Fbxo7 and PI31 levels in patient cells.Consistent with their association with proteasomes, patient fibroblasts have reduced proteasome activity and proteasome subunits.We also show PI31 interacts with the MiD49/51 fission adaptor proteins, and unexpectedly, PI31 acts to facilitate SCF Fbxo7 -mediated ubiquitination of MiD49.The L250P mutation reduces the SCF Fbxo7 ligase-mediated ubiquitination of a subset of its known substrates.Although MiD49/51 expression was reduced in patient cells, there was no effect on the mitochondrial network.However, patient cells show reduced levels of mitochondrial function and mitophagy, higher levels of ROS and are less viable under stress.Our study demonstrates that Fbxo7 and PI31 regulate proteasomes and mitochondria and reveals a new function for PI31 in enhancing the SCF Fbxo7 E3 ubiquitin ligase activity.

Introduction
Parkinson's disease (PD) is the second most common neurodegenerative disorder in the world, affecting more than 6 million people worldwide, where 90% of cases are idiopathic.Over the past two decades, an increasing number of mutations have been identified and associated with familial forms of PD.Study of the functions of these so-called PARK genes provides new opportunities to understand the mechanisms underlying familial and sporadic forms of Parkinson's disease [1,2].Since 2008, when the first case of Parkinsonian pyramidal syndrome in an Iranian family was linked to a mutation in FBXO7, many other pathological recessive mutations of FBXO7 have been identified [3][4][5][6].These mutations are clustered at the N-and C-terminal domains of Fbxo7, and a wide range of symptoms have been reported [2,[6][7][8][9][10][11][12][13][14][15][16][17][18].The age of the onset is also variable with cases arising in childhood and adolescence to young adults and older patients.Some patients, for example, with L34R mutations have a disease-onset and presentation similar to idiopathic cases of Parkinson's disease [6].
Fbxo7 was originally identified as a substrate recognition subunit of Skp1-Cullin-1-F-box (SCF)-type E3 ubiquitin ligases.It has a C-terminal proline-rich (PRR) domain and an N-terminal ubiquitin-like (Ubl) domain, which are involved in recruiting substrates for ubiquitination by the SCF Fbxo7 ligase [19].Although hundreds of substrates of SCF Fbxo7 ligase have been identified, many are not subject to proteasomal degradation [20].Moreover, some Fbxo7 interacting proteins are not ubiquitinated, but rather scaffolded and stabilised by association with Fbxo7.Examples include cell cycle regulators, Cdk6 and p27, and consequently, Fbxo7 augments G1 phase cyclin D/Cdk6 activity [21][22][23].Fbxo7 also regulates the levels of PINK1/PARK6, another autosomal recessive gene associated with Parkinson's disease [24].Fbxo7 acts in a common pathway with PINK1 and Parkin/PARK2 to enable stress-activated mitophagy, and it also affects basal mitochondrial function [25,26].
In addition to two substrate recruiting domains, Fbxo7 also contains an FP (Fbxo7-PI31) dimerisation domain that enables its homo-dimerisation and hetero-dimerisation with the proteasome regulator, PI31/PSMF1 [27,28].Fbxo7 and PI31 share a similar organisational structure; both contain an FP domain and a PRR domain [28].In many cell types, a reduction in Fbxo7 levels correlates with lower PI31 levels [19,29,30].
Multiple global proteomic profiling and interactome studies have demonstrated the association of Fbxo7 and PI31 with each other and the proteasome [31][32][33][34].The PRR domain of PI31 mediates its interaction with a subunits of the 20S proteasome, and recent structural studies detail insertion of the C terminus of the PI31 in multiple species in the barrel of the proteasome, indicating an inhibitory effect [35][36][37].However, the expression of PI31 on proteasome activity in cells remains controversial [38].Fbxo7, has been reported to ubiquitinate a proteasome subunit a2/PSMA2, regulating proteasome assembly [39].A mouse with conditional loss of Fbxo7 in specific neuronal populations showed early-onset motor deficits and premature death which was attributed to reduced proteasome activity [39].More recent work has detailed a role for PI31 in the axonal transport of proteasomes [40].These studies suggest a multi-faceted role for Fbxo7 in the assembly, activity, and localisation of proteasomes in neurons.
Here we report a new homozygous point mutation in FBXO7 in a paediatric patient from an Israeli family presenting with a combination of upper and lower motor neuronal problems and dystonia.At their initial clinical presentation at 5 months of age, whole exome sequencing was performed, and a novel point mutation c.749T>C in the FBXO7 gene (p.Leu250Pro, NM_ 012179.4) was identified.We investigated the molecular effects of the L250P point mutation, located in the FP domain of Fbxo7, which selectively ablates its interaction with PI31.We found mutant Fbxo7-L250P and PI31 levels were significantly reduced in patient cells, which have decreased levels of proteasome subunits and proteasome activity.PI31 enhances SCF Fbxo7 ligase activity towards its substrates, and MiD49 and MiD51, fission adaptor proteins, are interacting partners for PI31.Although MiD49 and MiD51 both interacted with Fbxo7, PI31 enabled only MiD49 ubiquitination by SCF Fbxo7 ligase.MiD49 and MiD51 levels were significantly reduced in L250P patient cells, although no effects on the dynamics of the mitochondrial network were evident.Instead, patient cells had significantly reduced mitochondrial mass, respiration, and mitophagy and higher levels of cellular and mitochondrial ROS, and reduced viability under stress.Our data reveal the dual functions of Fbxo7 and PI31 in regulating both proteasomal and mitochondrial function (Box 1).

Demographic data and detection of a new pathogenic FBXO7 point mutation (L250P)
A female patient, the first offspring of healthy consanguineous parents, who are first degree cousins, initially presented at the age of 5 months with decreased limb movements and evolving truncal hypotonia.Her neurological examination revealed head drop and axial hypotonia with reduced facial and limb movements, generalised hyper-reflexia associated with bilateral ankle clonus and positive Babinski sign consistent with a pyramidal syndrome.Brain MR imaging at the age of 1.5 years showed thinning of corpus callosum, but otherwise was unremarkable.
Disease course was marked by the development of limb rigidity and fluctuating dystonic postures, but cognitive and communication skills remained relatively preserved.Ophthalmological examination at the age of 2.5 years displayed bilateral optic pallor with decreased visual evoked potential responses consistent with an evolving optic atrophy.
Given the neurological phenotype of the child dominated by axial hypotonia, limb rigidity, decreased spontaneous movements and generalised pyramidal signs along with a homozygous variant in FBXO7, a gene known to be associated with parkinsonian pyramidal syndrome, a dopamine transporter (DAT) SPECT imaging was performed which showed lack of uptake in the striatum.She started treatment with L-DOPA with gradual dose increase.She showed an initial good response with an increased range of limb movement, but the long-term clinical effect waned.In addition, she developed progressive kyphoscoliosis that is related to her rigidity and urinary retention that necessitated regular catheterisation.A repeated brain MRI at 3.5 years was similar to her initial imaging.On her last examination at approximately 5 years of age, the child expresses very slow but steady progress in her developmental milestones.She is able to sit with support, stand in a walker, grasp objects and bring them to her mouth.She utters two words and communicates with her family.However, in addition to her initial motor impairment, the patient developed moderate global developmental delay, and her recent developmental quotient (DQ) is 40, where a score < 75 indicates a developmental delay.

L250P point mutation localises to the FP domain of Fbxo7 and ablates Fbxo7 hetero-dimerisation with PI31
The L250P mutation is located within the FP domain (residues 180-324) of Fbxo7, which enables its homo-dimerisation and hetero-dimerisation with and FP domain in PI31 [28].The FP domains of both are structurally similar adopting an a/b-fold with a central five-stranded anti-parallel b-sheet flanked by two a-helices in the N-terminal and three a-helices at the C terminus [27,28,41].The FP domain uses an ab-interface for FP-FP-domain interaction and dimerisation, although different modalities for interacting have been proposed (ab, aa or bb-see ref. [27]).For example, the Fbxo7 interface has been suggested to use the b-interface (Fig. 1A, left-hand panel, green surface).Leucine 250 is a buried hydrophobic core residue within the central beta strand of the Fbxo7 FP

Box 1. The paper explained
Problem: The discovery of monogenetic mutations associated with early onset PD allows researchers to focus in on specific pathways to understand the causes of genetic and sporadic cases of PD.It is apparent that several of the genetic forms of PD damage multiple downstream physiological processes, including mitochondria, protein homeostasis and autophagy, stress responses, calcium signalling, metabolism, and intracellular trafficking.This makes it difficult to understand how dysfunction in any particular process affects others and how they link together to cause disease.Results: We identified a novel missense mutation in the dimerising domain of the ubiquitin ligase, Fbxo7, which prevents is interaction with a protein, PI31.PI31 is a regulator of the proteasome, and we find patient cells have reduced proteasome activity.Unexpectedly, we find this missense mutation also changes the ability of this ubiquitin ligase to modify a class of proteins involved in controlling the mitochondrial network.The patient cells also have reduced levels of mitochondria, respiration, and mitophagy and higher levels of intracellular ROS, and are more vulnerable to cellular stresses.Impact: By understanding how downstream physiological processes are co-ordinately affected by pathological mutations occurring in genetic cases of parkinsonism, we can decide which are the most proximal and actionable pathways for designing therapeutics.
domain b-sheet.Substitution by proline (L250P) would be particularly disruptive to the b strand resulting in a bulge or kink to the strand.The kink arises as the proline sidechain prevents a hydrogen on the amide nitrogen, so it cannot act as a hydrogen-bond donor, thereby disrupting the main chain hydrogen-bonding network of a sheet.Also, the proline sidechain places a steric constraint on the neighbouring residues because of its pyrrolidine ring, perturbing its secondary structure (Fig. 1A).Thus, the L250P mutation would be predicted to both partially destabilise the Fbxo7 FP domain as well as disrupt dimerisation through its b-interface with the FP domain of PI31 [27,28,41].
To explore the effect of the L250P mutation on Fbxo7 dimerisation with PI31, co-IP experiments were performed.HEK293T cells were transfected with plasmids expressing Flag-tagged WT or mutant Fbxo7.Anti-Flag immunoprecipitates of total cell lysates were assayed by immunoblotting.Endogenous PI31 associated with WT Fbxo7 but not L250P Fbxo7, indicating that the L250P mutation ablated their interaction (Fig. 1B).We also tested whether the L250P mutation affected Fbxo7 interaction with Skp1, which binds the F-box domain adjacent to the FP domain.Both alleles of Fbxo7 interacted with Skp1, indicating that only the interaction with PI31 is affected.Moreover, the ability of Fbxo7-L250P to bind Skp1 suggests it can be incorporated into an SCF-E3 ubiquitin ligase, although lacking PI31.These data show PI31 hetero-dimerises with Fbxo7 exclusively via its b-interface and raises the possibility that PI31 acts as a co-factor to enhance SCF Fbxo7 ligase activity.
Co-immunoprecipitation experiments were also conducted to test the effect of L250P mutation on Fbxo7 homo-dimerisation through either an ab or bb interface.
For these experiments, HEK293T cells were transfected with FLAG-WT or L250P alleles of Fbxo7 along with HA-tagged WT or Fbxo7-L250P alleles.Anti-FLAG immunoprecipitates were assayed for the presence of WT or mutant Fbxo7-HA protein.We noted the Fbxo7 L250P-HA constructs were less well expressed compared to WT Fbxo7-HA constructs.Nonetheless, both WT and mutant Fbxo7 were detected in WT and mutant FLAG-immunoprecipitates (Fig. 1C), indicating the L250P mutation does not prevent Fbxo7 homodimerisation.These data indicate Fbxo7 is not misfolded by the L250P mutation and that homodimerisation of Fbxo7 is not dependent exclusively on the b-interface of its FP domain and suggest alternate interfaces mediate this.

Patient fibroblasts with L250P mutation show reduced expression of Fbxo7 and PI31
We next measured the effect of the Fbxo7 L250P mutation on its endogenous expression levels by immunoblotting and observed a marked reduction of mutant Fbxo7 protein compared to WT (Fig. 1D).Previous studies showed that in many different cell and tissue types, when Fbxo7 expression is reduced, there is a coincident reduction in endogenous PI31 levels [21,29,30].Moreover, as Fbxo7 interaction with PI31 was specifically lost as a consequence of the L250P mutation, we hypothesised PI31 expression levels would be reduced in patient fibroblasts.Total cell lysates from patient and control fibroblasts were immunoblotted for PI31 levels, and steady state PI31 expression was significantly lower in patient cells (Fig. 1D).To measure the protein stability, we performed a time course of treatment with the cycloheximide.The half-life of PI31, was unaffected in patient Fig. 1.The Fbxo7-PI31 (FP) domain missense mutation L250P compromises Fbxo7-PI31 interaction and proteasome activity.(A) Close-up of the structure of Fbxo7 FP domain highlighting the position of the L250 residue within the central beta strand of the Fbxo7 FP domain bsheet.The Fbxo7 FP domain is on the left (green) and PI31 on the right (cyan a-interface).The b-interface is the proposed Fbxo7-PI31 hetero-dimerisation surface.(B) Co-immunoprecipitation of endogenous PI31 with Flag-Fbxo7 wild-type (WT) or L250P in total cell lysates from HEK293T cells transfected with Flag-Fbxo7 WT or L250P.Immunoblots of total lysate prior to anti-FLAG immunoprecipitation are also shown (n = 2).(C) Co-immunoprecipitation of Flag-Fbxo7 WT or L250 and HA-Fbxo7 WT or L250P in total cell lysates from HEK293T cells transfected with Flag-Fbxo7 WT or L250P and HA-Fbxo7 WT.Immunoblots of total lysate prior to anti-FLAG immunoprecipitation are also shown (n = 3).(D) Immunoblotting for the expression of Fbxo7 and PI31 proteins from cell lysates of control or patient L250P fibroblasts.GAPDH serves as a loading control.Graphs of quantification of relative Fbxo7 (n = 4) and relative PI31 levels (n = 5).Mean AE SD, **P < 0.01, ***P < 0.001.(E, F) Cycloheximide (CHX) chase on control and patient L250P fibroblasts.Total cell lysates were prepared after the indicated treatment time and the levels of PI31 (E) and Fbxo7 (F) were analysed by immunoblot (n = 3).Quantification of Fbxo7 relative to tubulin loading control is shown below.(G) Immunoblot analysis for the proteasome subunits PSMA7, PSMC4 and PSMB1 in total cell lysates from control and patient L250P fibroblasts.Tubulin is used as a loading control (n = 6).(H) Proteasome activity assay performed on total cell lysates from control or patient L250P fibroblasts.MG132 treatment is used to confirm the efficiency of the assay by blocking the proteasome and diminishing the signal (n = 2).Mean AE SEM, **P < 0.01.(I, J) Image of proteasome activity labelling in gel using the fluorescent probe Me4BodipyFL-Ahx3Leu3VS and quantification of catalytic subunits relative to tubulin loading control (n = 3).Mean AE SEM, *P < 0.05, **P < 0.01.cells compared to controls (Fig. 1E).By contrast, Fbxo7 stability was reduced in the patient cells.Whereas in control cells, Fbxo7 levels decreased by 50% within 2 h, but remained detectable throughout the 8 h time course.When equivalent amounts of control and patient cell lysates were tested, Fbxo7-L250P was present at much lower levels, and undetectable by 2 h (data not shown).When twice the amount of lysate was assayed, mutant Fbxo7 was undetectable after 2 h (Fig. 1F), indicating a more rapid decay than WT Fbxo7.These data show that the L250P mutation prevents Fbxo7 interaction with PI31 and destabilises the expression of Fbxo7, while PI31 expression is concomitantly reduced in patient cells.

Reduced proteasome levels and activity in Fbxo7-L250P patient fibroblasts
Fbxo7 and PI31 both associate with proteasomes, and because their levels were significantly reduced in patient cells, we tested the effect on proteasomes.We first tested for the steady state levels of proteasome subunits by immunoblotting lysates for PSMA7, PSMB1, and PSMC4 (Fig. 1G).PSMB1 and PSMC4 showed a reduction in levels, while PSMA7 levels were similar between the two lines.These data suggest that proteasome activity might be altered in patient cells, so we tested its activity using a chymotrypsin-like activity assay using a peptide substrate, Suc-LLVY-AMC, which in a two-step assay, generates a luminescent signal when proteolytically cleaved.Lysates from patient fibroblasts showed a significant 15% reduction compared to control fibroblasts (Fig. 1H).A different assay for proteasome activity entails the use of a fluorescent Me4BodipyFL-Ahx3Leu3VS probe which covalently binds the N-terminal threonine residue of the proteolytically active subunits (b1/2/5 and b1i/2i/5i) in the proteasome, enabling their visualisation (Fig. 1I).Lysates made from patient fibroblasts incubated with the activity probe had a 30-50% reduction in levels of constitutive and stress-induced catalytic b subunits (Fig. 1J).Immunoblotting for steady state levels of LMP2 (b1i) and LMP7 (b5i) showed these were also reduced (data not shown).PI31 has been reported to function as a regulator of the stress-induced immunoproteasome, which contains the alternate b subunits [42][43][44].These data together indicate that patient cells have reduced proteasome activity due to lower expression of both constitutive and alternate proteasome subunits.

Fbxo7 and PI31 interacts with the mitochondrial fission adaptors MiD49 and MiD51
The Fbxo7 L250P mutation prevented PI31, but not Skp1 binding, which raised the possibility that PI31 acts as a co-factor for SCF Fbxo7 ligase.To identify SCF Fbxo7 substrates reliant on PI31, we cross-referenced the Bio-GRID database for PI31/PSMF1 interactors with our own experiments to identify SCF Fbxo7 substrates using protein arrays.A 2005 study reported PI31 and MiD49/SMCR7/MIEF2 as high confidence interacting proteins [34,45], and we previously identified MiD49 as a lower confidence in vitro substrate of SCF Fbxo7 ligase [20].MiD49 and the closely related MiD51 protein are important fission adaptor proteins that recruit the Drp1 GTPase to multiple organelles, including mitochondria, ER, lysosomes, undergoing fission (Fig. 2A) [46][47][48].To validate an interaction between the MiD49 and PI31, we first performed co-IP assays.MiD49-Myc was cotransfected with either FLAG-tagged WT PI31 or a PI31 mutant lacking the C-terminal proteasomeinteracting domain (DC, aa1-189) in HEK293T cells expressing either a control shRNA or a shRNA targeting Fbxo7 expression (Fig. 2B).A reduction in a cleaved form of PI31 was apparent upon Fbxo7 KD [49].Importantly, MiD49 was detected in anti-FLAG immunoprecipitates for both WT PI31 and PI31DC and in cells with undetectable Fbxo7 expression.The interaction between PI31 and MiD49 was also verified using in vitro pull-down assays where GST-PI31 fusion proteins were used to pull down MiD49 made using programmed reticulocyte lysates (Fig. 2C).Together these data suggest PI31 and MiD49 interact with each other independently of Fbxo7 and the C terminus of PI31.

SCF Fbxo7 ubiquitinates MiD49, but not MiD51
The in vivo and in vitro interaction data suggest PI31 might recruit MiD49 and/or MiD51 to the SCF Fbxo7 ligase.To test whether Fbxo7 interacted with MiD49/51, HEK293T cells were co-transfected with FLAG-tagged Fbxo7 and Myc-tagged MiD49/51.Both proteins were detected in anti-FLAG immunoprecipitates of Fbxo7 (Fig. 2D,E).Of note, a higher molecular weight smear, indicative of polyubiquitination, was detected in immunoblots for MiD49-Myc (Fig. 2D), but not MiD51-Myc (Fig. 2E), suggesting MiD49, but not MiD51, is a substrate of SCF Fbxo7 ligase.We mapped the Fbxo7-MiD49 interaction using Fbxo7 truncations of functional domains within Fbxo7 in co-IP assays (Fig. 2F).The smallest construct tested, containing only the FP dimerisation and F-box domains (aa169-398), interacted with both MiD49 (Fig. 2G) and MiD51 (Fig. 2H).As Fbxo7 lacking its F-box domain interacted with both MiD proteins, these data suggest the FP dimerisation domain of Fbxo7 mediates interactions with MiD49 and MiD51.
We directly tested the ability of SCF Fbxo7 ligase to ubiquitinate MiD49 and MiD51 using in vitro ubiquitination assays.Increased higher molecular weight smearing indicative of ubiquitination was detected for MiD49 upon addition of the WT Fbxo7 ligase but not the ligase-dead DF-box variant (Fig. 3A).No such higher molecular weight smearing was detected for MiD51 (Fig. 3B).We also conducted in vivo ubiquitination assays where HEK293T cells with reduced Fbxo7 levels were co-transfected with Myc-tagged MiD49 or MiD51, FLAG-tagged Fbxo7 constructs, and HA-tagged ubiquitin.Myc-tagged MiD49 or MiD51 was isolated from cells via anti-Myc immunoprecipitation and probed by immunoblotting.MiD49 (Fig. 3C), but not MiD51 (Fig. 3D), showed increased ubiquitination in the presence of the WT Fbxo7 ligase but not the ligase dead DF-box allele.Thus, despite being structurally similar and both proteins interacting with Fbxo7, only MiD49 was ubiquitinated by SCF Fbxo7 ligase.

The Fbxo7-L250P mutation disrupts the ubiquitination of MiD49
We tested whether the Fbxo7-L250P mutation, which removes the PI31 binding, affected Fbxo7 interaction with the MiD proteins.HEK293T cells with constitutive knockdown of Fbxo7 were co-transfected with MiD49 or MiD51 and constructs encoding Fbxo7 with pathological mutations (T22M, L250P, R378G and R498X).None of the mutations prevented Fbxo7 interaction with either MiD49 (Fig. 3E) or MiD51 (Fig. 3F).However, while the WT ligase and other pathological Fbxo7 alleles showed higher molecular weight species of MiD49, this was much reduced specifically in the presence of Fbxo7-L250P (Fig. 3E).In vivo ubiquitination assays confirmed this reduced ability of SCF Fbxo7-L250P to ubiquitinate MiD49 in cells (Fig. 3G).We further tested the effect of the point mutation against a range of SCF Fbxo7 substrates.The ability of SCF Fbxo7-L250P , to ubiquitinate GSK3b was comparable to WT SCF Fbxo7 ligase demonstrating its capacity to function (Fig. 4A); however, its ability to ubiquitinate other substrates, Tomm20, both in vitro and in vivo (Fig. 4B,C) and Rpl23 in vivo, was impaired (Fig. 4D).These data suggest PI31 is not required for the Fbxo7-MiD49 or MiD51 interaction but is required for SCF Fbxo7 ubiquitination of MiD49.In addition, SCF Fbxo7-L250P ubiquitination of other substrates, like Tomm20 and Rpl23 was reduced.Thus, the mutation results in an E3 ligase with an altered substrate range compared to WT.

Fbxo7 stabilises MiD49/51 protein levels
We next tested the effect of the L250P mutation on MiD proteins in the patient fibroblasts.Surprisingly, we observed 60% and 65% decreases in MiD49 and MiD51 levels, respectively (Fig. 5A), indicating Fbxo7 and/or PI31 stabilises MiD protein levels.We infer this stabilising effect of Fbxo7 on the MiD proteins is not dependent on its ubiquitination since SCF Fbxo7-L250P does not ubiquitinate MiD49, and MiD51 is not an SCF Fbxo7 substrate.
To test whether knock-down of Fbxo7 levels has the same effect on MiD49/51 expression as the L250P mutation, we immunoblotted lysates from HEK293T cells with constitutively expression of a shRNA targeting Fbxo7.We found an approximately 60% decrease in endogenous MiD49 protein levels, indicating Fbxo7 stabilises MiD49 levels (Fig. 5B).A similar effect of reduced Fbxo7 expression was seen on MiD51 levels which were 55% lower.Consistent with a stabilising effect of Fbxo7 on MiD49, overexpression of WT or ligase-dead Fbxo7 in Fbxo7 KD cells increased MiD49 levels 2-fold (Fig. 5C).As Fbxo7 has been reported to bind and stabilise another interacting protein, EYA2, from ubiquitination by another SCF-type E3 ligase, SCF Fbxw7 [50], we tested whether Fbxo7 also protected MiD49 from ubiquitination.Transfection of MiD49 in HEK233T cells with Fbxo7 KD revealed higher levels of MiD49 ubiquitination compared to cells with endogenous Fbxo7 levels, and ubiquitination intensity was increased further upon proteasomal inhibition (Fig. 5D).A stabilising effect of Fbxo7 was also seen in cycloheximide chase experiments on cells treated with hydrogen peroxide to induce ROS and oxidative stress.Where control cells maintain high levels of MiD49 over 6 h, cells lacking Fbxo7 show markedly reduced MiD49 stability (Fig. 5E,F).Taken together, these data indicate that although Fbxo7 can ubiquitinate MiD49, its levels are stabilised by Fbxo7, under basal conditions and upon exposure to oxidative stress.

Mitochondrial content and function are reduced in L250P fibroblasts
Our data suggest the Fbxo7 L250P mutation might affect the mitochondrial network because of decreased MiD49/51 levels.To test this, patient and control fibroblasts were stained with MitoTracker Green FM, which stains total mitochondrial mass, and mean fluorescence was analysed by flow cytometry.Surprisingly, quantification of MitoTracker Green staining showed an approximately 40% decrease in total mitochondrial content in patient fibroblasts compared to controls (Fig. 6A).This decrease in staining was not due to smaller cell size between the control and patient fibroblasts.To confirm this finding, mitochondrial mass was also measured by qPCR for mtDNA levels.The average copy numbers of three mitochondrial genes (COXI, CYTB and ATP6) were measured and normalised to two regions of genomic DNA (ATP5B and CHR10).Consistent with measurements of mitochondrial staining by FACS analysis, there was an approximately 50-60% decrease in the levels of mtDNA in patient fibroblasts compared to control cells, indicating patient fibroblasts have a reduced mitochondrial mass (Fig. 6B).
We next visualised the mitochondrial network by staining fibroblasts with MitoTracker Green FM and performing confocal microscopy (Fig. 6C).Images were processed using the IMAGEJ PlugIn MINA, and in agreement with our findings on mass, the overall mitochondrial footprint, which is the area of mitochondrial signal, was significantly lower in the Fbxo7-L250P fibroblasts compared to controls (Fig. 6D).However, no other parameters, including the mean branch length, network size, or the number of individual mitochondria, were significantly altered, suggesting that the L250P mutation does not significantly affect mitochondrial network structure/dynamics.
To test if there were any changes in membrane polarisation between the patient and control fibroblasts, we stained cells with TMRE, a fluorescent dye whose accumulation is dependent on mitochondrial membrane potential.When normalised to overall mitochondrial mass, we did not detect any differences in membrane potential (Fig. 6E).
Given the significant decrease in mitochondrial mass, we evaluated oxidative phosphorylation using a Seahorse XF Cell Mito Stress Test to measure cellular oxygen consumption.Basal respiration, maximal respiration, ATP production and respiratory spare capacity were all found to be significantly lower in the Fbxo7-L250P fibroblasts (Fig. 6F).These data indicate a smaller mitochondrial footprint and reduced oxygen consumption in patient fibroblasts.Because of these findings, we tested whether there were alterations in the levels of mitophagy in patient cells by transducing cells with the mito-QC reporter.This reporter is a fusion of mCherry-GFP to aa101-152 of FIS1 which localises to the outer mitochondrial membrane [51].Under basal conditions, mitochondria are red and green fluorescent, and upon mitophagy, the GFP signal on mitochondria is quenched in the low pH environment of the lysosome.Thus, mitochondria in lysosomes appear red (Fig. 7A).Patient cells showed significantly reduced levels of basal mitophagy, both the raw percentages and normalised to the reduced levels of mitochondrial content per cell (Fig. 7B).Moreover, patient cells have significantly higher levels of Parkin (Fig. 7D), potentially a compensatory increase of this mitophagy regulator when Fbxo7 levels are reduced.When treated with the PINK1/Parkin-independent mitophagy inducer, an iron chelating agent, deferiprone (DFP), patient fibroblasts underwent a significant increase in the fold change in mitophagy, similar to control cells, although they did not achieve the same overall levels seen in control cells (Fig. 7B,C).Thus, patient cells show significant decreases in basal mitophagy and DFP-induced mitophagy, although they are able to induce mitophagy.
Finally, to gain insight into a potential mechanism to account for the large decrease in mitochondrial mass despite a concurrent reduction in basal mitophagy in patient cells, cell lysates were immunoblotted for the transcription factors involved in mitochondrial biogenesis.Both PGC1a and PPARc were significantly decreased in the patient cells (Fig. 7E).These results indicate that mitochondrial homeostasis is compromised in these patient cells.

Fbxo7-L250P patient fibroblasts show increased levels of ROS
Deficiencies in Fbxo7 have been shown to be associated with increases in the level of cytosolic ROS [25,52,53], and our data indicate defective mitophagy in patient cells.To investigate if ROS levels were altered, we treated cells with the ROS indicator H 2 DCFDA.Non-fluorescent H 2 DCFDA is converted to fluorescent DCF following oxidation by ROS; thus, increased fluorescence is indicative of higher levels of oxidative stress.Consistent with previous findings on Fbxo7 reducing ROS levels, patient fibroblasts had two-fold higher levels of cellular ROS compared to control fibroblasts (Fig. 7F).To further assay specifically for mitochondrial ROS, cells were incubated with MitoSOX, a derivative of dihydroethidium, which accumulates in mitochondria, and upon oxidation by superoxide, produces a fluorescent signal.No significant differences in total mitochondrial ROS levels were detected, suggesting mitochondrial ROS is not likely to account for the two-fold increase in cellular ROS detected with H 2 DCFDA.However, when normalised to mitochondrial mass, the MitoSOX signal in patient fibroblasts was approximately double that of control cells (Fig. 7G).These data indicate that patient fibroblasts have elevated levels of mitochondrial and cellular ROS, although we did not detect any differences in cell viability as a result of this increase in ROS between the control and patient fibroblasts under basal conditions.However, the patient fibroblasts were found to be less viable following treatment with hydrogen peroxide and proteasome inhibitors, but not to mitochondrial toxins, rotenone and antimycin A (Fig. 7H) or to other cell stressors, tunicamycin or brefeldin A (data not shown).Consistent with the reduced proteasome activity (Fig. 1H,I,J) and higher intracellular ROS (Fig. 7F,G) in the patient cells, these data suggest that the patient cells are more sensitive to proteasome inhibition and exposure to hydrogen peroxide than control cells, but both cell lines respond equivalently to mitochondrial inhibition.

Discussion
In this study, we report on a patient who was diagnosed at 5 months of age with clinical features suggestive of a complicated pyramidal syndrome.Whole exome sequencing revealed a new homozygous point mutation, c.749T>C in FBXO7 (p.Leu250Pro, NM_ 012179.4),which has not been reported previously.Bi-allelic loss of function mutations in FBXO7 are associated with an early onset parkinsonian-pyramidal syndrome termed PARK15 (OMIM# 260300) typically presenting during the juvenile period [4,6].Although the majority of patients reported so far initially presented within the second decade of life or later with mainly slowly progressive spastic paraparesis and variable parkinsonian symptoms including bradykinesia, limb rigidity, hypomimia and action or resting tremor, the disease spectrum has enlarged with the increased use of wide genomic sequencing in patients with undiagnosed neurological impairments.Therefore, PARK15 disease spectrum has increased accordingly to include variable cognitive decline, dysphagia, urinary incontinence, ophthalmological abnormalities, abnormal MRI findings including cerebral and cerebellar atrophy and variable response to levodopa ranging MiD49 and MiD51 protein levels quantified relative to GAPDH and tubulin respectively (n = 3).Mean AE SEM, **P < 0.01, ***P < 0.001.(C) HEK293T cells expressing either control or Fbxo7-targeting shRNA (KD) were transfected with Fbxo7 constructs (wild-type; WT and DF-box).Endogenous MiD49 protein levels in cell lysates were analysed by immunoblot and levels were quantified using IMAGEJ (n = 2).Mean AE SEM.(D) Control and Fbxo7 KD HEK293T cells were transfected with MiD49-Myc and treated with either DMSO or MG132 for 4 h prior to harvesting.MiD49 ubiquitination levels in total cell lysates (no MG132 treatment) were analysed by immunoblotting and quantified using IMAGEJ (n = 5).Mean AE SEM, **P < 0.01.(E) Immunoblots of HEK293T cells stably expressing control (À) and Fbxo7 shRNA (+) were treated with 250 lM H 2 O 2 for up to 6 h.Cell lysates were resolved by SDS/PAGE and immunoblotted as indicated.(F) Quantification of relative levels of MiD49 over time in response to H 2 O 2 (n = 4).For each experiment, MiD49 density was normalised to the density of the loading control.The 0 h time point was set to a value of 1 for each cell line, and the ensuing timepoints were normalised relative to the 0 h time point.Error bars are SD, *P < 0.05, **P < 0.01, ***P < 0.001.
from no improvement to some beneficial effect [6,9,15,16].A single report described two siblings with infantile onset PARK15 associated with poor cognitive outcome, severe motor impairment and early lethality [11].Our patient presented an unusual and to our knowledge, the most severe phenotype with infantile onset and significant neurological impairment with minimal L-DOPA responsiveness and cognitive impairments.Degenerating axons in the optic nerves have been described in conditional mouse lacking Fbxo7 specifically in the myelinating cells of the central and peripheral nervous system [54], highlighting an important role for Fbxo7 in maintaining the optic nerve fibres.At the age of 2.5 years, an evolving optic atrophy has also been detected in this patient, suggesting that this new point mutation could have an effect on Fbxo7 functions in the optic nerve axons.Moreover, in other mouse models lacking Fbxo7 specifically in the dopaminergic neurons, decreased numbers of fibres innervating the striatum are observed [30,39], suggesting that Fbxo7 could have a more global role in axonal growth in different parts of the nervous system, and this would explain the abnormalities on the DAT SPECT scan.The causes for the variable phenotype of our patient are currently unidentified but may be associated with this specific FBXO7 mutation and perhaps additional genetic factors that play a role in Fbxo7 function.
The L250P mutation lies within the FP domain of Fbxo7, and it was previously reported that mutation of V253E prevents Fbxo7 interaction with PI31 [28].Consistent with these previous studies on the mode of interaction of the FP domains of Fbxo7 and PI31, the L250P mutation specifically abolishes their heterodimerisation.In characterising the fibroblasts from this patient, we have shown the levels and stability of Fbxo7-L250P and PI31 are decreased, highlighting the importance of the b-interface for the interaction between the two proteins.Of note, the L250P mutation did not prevent homo-dimerisation, suggesting other regions of Fbxo7 mediate this and that homo-dimerisation is not sufficient to stabilise Fbxo7-L250P protein levels.
A positive correlation between Fbxo7 and PI31 expression has been observed here (Fig. 1D) and in multiple studies [21,29,30,49].Additionally, multiple groups report Fbxo7 and PI31 interact with each other and affect the proteasome in a number of ways, ranging from assembly and activity to trafficking [29,39,40].Additionally, a preprint reporting that PI31 transgene expression rescues motor neuron degeneration caused by Fbxo7 loss in mice hints at a capacity for PI31 to bypass the need for Fbxo7 activity or a redundancy in their functional relationship, although no mechanism is proposed (biorxiv: 10.1101/ 2020.05.05.078832v1).Our study in patient fibroblasts bearing a pathogenic point mutation adds to this picture showing reduced Fbxo7 and PI31 protein levels and the loss of their hetero-dimerisation correlated with lower levels of specific proteasome subunits, including alternate immunoproteasome subunits, and significantly reduced proteasome activity.These observations underscore the multiple ways in which Fbxo7 and PI31 can affect the proteasome.
Given that PI31 is usually thought of as a proteasome regulator, a surprising finding from our studies is that PI31 affects the degree of the ubiquitination of many substrates of the SCF Fbxo7 E3 ligase.Since we showed mutant Fbxo7-L250P could bind to Skp1, we reasoned it could be incorporated into an active E3 ubiquitin ligase.We discovered the MiD proteins as interactors of PI31, via its FP domain, and showed SCF Fbxo7-L250P ubiquitination of MiD49 is much reduced compared to the WT ligase.In principle, other interacting proteins of PI31 could similarly be recruited for ubiquitination.PI31 appears to selectively modulate the efficiency of ubiquitination of substrates as evidenced by SCF Fbxo7-L250P ligase reduced ubiquitination of Tomm20, Rpl23 and MiD49 while is ubiquitination of GSK3b was unaffected.Interestingly, MiD49 still interacted with Fbxo7-L250P but was not ubiquitinated, suggesting PI31 binding promotes a conformational change that permits MiD49 ubiquitination.These data argue PI31 enhances SCF Fbxo7 ubiquitination of some of its substrates, and the L250P mutation reduces the ubiquitination of Fbxo7 substrate repertoire.The functional consequences of MiD49 ubiquitination remain to be determined; however, we did not find evidence for its ubiquitin-mediated degradation.Rather, the levels of MiD49 and MiD51 protein were significantly reduced in the patient fibroblasts and Fbxo7 KD cells, suggesting Fbxo7 and/or PI31 stabilise their levels or prevent their degradation.Fbxo7 may protect MiD49 from other E3 ligases, e.g., MARCH5, whose ubiquitination has been shown to promote proteasomal degradation, either by preventing ligase binding, or alternatively by blocking access to ubiquitination sites [55,56].Such a mechanism is similar to Fbxo7 function in the EYA2-Fbxw7 axis [50].
MiD49 and MiD51 are adaptor proteins which recruit Drp1 to catalyse the fission of organelles, including mitochondria, ER, and peroxisomes.Although some studies report loss of MiD49 and MiD51 levels can cause an elongated, hyper-fused mitochondrial network [48,57,58], this was not the case in the patient fibroblasts.Since the MiD proteins function in a redundant fashion with other adaptors Fis1 and Mff, it is possible that these proteins compensated for the decreased expression of the MiD proteins in these cells.The presence of multiple Drp1 adaptors suggests cell type-specific or contextual roles for these proteins [46,59,60].More recent studies support this idea, e.g., MiD49 has been shown to be involved in localised mitochondrial fission caused by damage to the plasma membrane, while MiD51 is important for coordinating the Bcl2 family of proteins on mitochondria in response to apoptotic stimuli [61,62].Thus, the phenotypes of MiD49 and MiD51 loss may be responsive to the physiological state of the cell.In this context, Fbxo7 appeared important for maintaining MiD49 levels upon exposure to hydrogen peroxide.
Despite the lack of an effect on connectivity of the mitochondrial network, the L250P patient fibroblasts showed a significant reduction in mitochondrial mass, in respiration, and in basal mitophagy levels.The latter is consistent with results reported in Drosophila by Sanchez-Martinez et al. [63] who show a ~50% reduction in basal mitophagy in nutcracker, the Fbxo7 ortholog, mutant larvae.However, they contrast results by Kraus et al. [64], who reported no change in basal mitophagy using Keima reporters, in Fbxo7 KO iNeurons or HeLa cells.While these differences may be due to the varying nature of the mito-QC vs the Keima reporters, we also have observed that using CRISPR-Cas9 to make clonal KO Fbxo7 results in cellular adaptations during clonal selection.In the patient fibroblasts, we found high levels of Parkin, suggesting a compensatory up-regulation of this stress-mitophagy regulator, and we have reported the difficulty in reducing or over-expressing Fbxo7, for example, in T cell lines [65], and speculate that this is due to its inverse correlation with ROS levels [25,52,53], reducing cell viability.We also find patient cells induce mitophagy in response to DFP although not to the higher levels seen in control cells.Our results also indicate no significant difference in the ratio of polarised/total mitochondria between patient and control cells, yet patient cells had impaired mitochondrial function and higher levels of cytosolic and mitochondrial ROS.Impaired mitochondrial respiration and higher levels of cytosolic ROS have been reported in another patient fibroblast line bearing homozygous R378G mutations in Fbxo7 [25].We speculate that the reduced biogenesis of mitochondria might arise from regulatory feedback to maintain lower levels of mitochondria producing ROS.Our study of this patient's fibroblasts demonstrates the involvement of Fbxo7 in regulating proteasomes and mitochondria, two of the commonly dysregulated pathways associated with neurological diseases.Our data implicate PI31 as a co-regulating factor with Fbxo7 which impacts on both pathways and argue that understanding the cross-talk between them is important for understanding the basis of neurodegeneration which warrants further studies.

Ethical approval
Informed consent for participation in the study was obtained from the parents of the patient.The study was approved by the Emek Medical Center ethics committee (study no.EMC-0067-09) and conforms to the standards set by the Declaration of Helsinki.

Whole exome sequencing
Exonic sequences were enriched with the SureSelect Human All Exon 50 Mb V5 Kit (Agilent Technologies, Santa Clara, CA, USA).Sequences were generated on a HiSeq2500 (Illumina, San Diego, CA, USA) as 125-bp paired-end runs.Read alignment and variant calling, were performed with DNAnexus (Palo Alto, CA, USA) using default parameters with the human genome assembly hg19 (GRCh37) as reference.Exome analysis of the proband yielded 55 million reads, with a mean coverage of 979.Following alignment to the reference genome [hg19] and variant calling, variants were filtered out if the total read depth was < 89 and if they were off-target (> 8 bp from splice junction), synonymous, or had minor allele frequency (MAF) > 0.005 in the gno-mAD database.Homozygous variants of interest that survived filtering are provided in Table 1.

Cell lines
HEK923T and HFF-1 control fibroblasts (Human Foreskin Fibroblasts) were purchased from ATCC (Gaithersburg, MD, USA), which uses short tandem repeat profiling to authenticate cell lines.Cells were maintained in DMEM supplemented with 10% heat-inactivated foetal bovine serum (Gibco, Thermo Fisher Scientific, Waltham, MA, USA), 100 UÁmL À1 penicillin and streptomycin (Thermo Fisher Scientific) at 37 °C in a humidified 5% CO 2 atmosphere, and regularly tested for mycoplasma contamination.Cell lines stably expressing shRNA to human FBXO7 were generated as described previously [20] and selected with 2 lgÁmL À1 puromycin (Sigma-Aldrich, Haverhill, Cambridgeshire, UK).

DNA constructs
pcDNA3 vectors expressing full length, truncated or DF-box Fbxo7 with an N-terminal Flag tag or C-terminal HA tag have been previously described [21,24].Flag-Fbxo7 L250P construct in pcDNA3 has been engineered by site-directed mutagenesis and cloned into Flag-pcDNA3 vector.Fbxo7 L250P-HA construct was generated by subcloning Fbxo7-L250P from the Flag-pcDNA3 vector into pcDNA3-HA vector.

Cell transfections
HEK293T cells were split 24 h prior transfection when cells were at 50-80% confluency.For each 10 cm dish, 2-6 lg of plasmid DNA was diluted in 300 lL of Optimem media (Invitrogen, Thermo Fisher Scientific) and mixed with the transfection reagent (Fugene or PEI) at a ratio of 3 lgÁlL À1 and incubated at RT for 15 min before adding to cells.Cells were transfected for 36-48 h prior to harvesting.
Fbxo7 was immunoprecipitated from cell lysates using 2 lL of anti-Fbxo7 antibody (Aviva), or isotype matched  control, for 1 h at 4 °C with rotation, then 20 lL Protein A/G PLUS-Agarose (Santa Cruz Biotechnology) was added and samples were incubated for a further 3 h.Beads were washed four times in lysis buffer and resuspended in 29 Laemmli loading buffer.For co-immunoprecipitation assays, cells were lysed as above.An anti-FLAG Ò M2 Affinity Gel (Sigma-Aldrich) was washed twice with lysis buffer then added to lysates and incubated for 4 h at 4 °C with rotation.Beads were washed four times in lysis buffer and beads were resuspended in 30 lL of 29 Laemmli loading buffer.

In vitro ubiquitination assays
To purify HA-tagged substrates for in vitro ubiquitination, HEK293T cells were transfected with HA-Tomm20 or HA-GSK3b which was immunoprecipitated with anti-HA agarose (Sigma-Aldrich) 48 h after transfection.Substrate was eluted with 300 lgÁmL À1 HA peptide (Sigma-Aldrich) and stored at 20 °C in 15% glycerol.Myc-tagged MiD49 and MiD51 were produced via in vitro transcription/translation (IVT) (Promega, Chilworth, Southampton, Hampshire, UK).For in vitro ubiquitination assays, a ubiquitin mix was prepared with 100 nM E1 (UBE1; Bio-Techne, Minneapolis, MN, USA), 500 nM E2 (UbcH5a; Bio-Techne), 20 lM human recombinant ubiquitin (Santa Cruz Biotechnology) and 2 mM ATP (Bio-Techne) in 19 ubiquitin conjugation reaction buffer (Bio-Techne).This was incubated for 5 min at RT then added to 100 nM SCF and 1 lL substrate and incubated for 1 h at 30 °C.The entire 10 lL reaction was mixed with an equal volume of 29 Laemmli loading buffer, resolved by SDS/PAGE and analysed by immunoblotting.

In vivo ubiquitination assays
HEK293T cells were transfected with tagged ubiquitin, FLAG-Fbxo7 and substrate constructs.Cells were treated with 25 lM MG132 for 4 h before being harvested and lysed in RIPA lysis buffer.For MiD49/51 ubiquitination assays, lysates were incubated with 2 lL anti-Myc-tag antibody for 1 h at 4 °C with rotation.Twenty microliter Protein A/G PLUS-Agarose (Santa Cruz Biotechnology) was then added to samples and incubated for a further 2 h.For Tomm20 ubiquitination assays, 20 lL anti-HA agarose (Sigma-Aldrich) was added to lysates for 3 h at 4 °C with rotation.Beads were washed 4 times in lysis buffer and resuspended in 29 Laemmli loading buffer.

GST protein purification
GST-tagged protein constructs were transformed into FB810 Escherichia coli.Bacterial cultures were grown in 2xTY medium containing 100 lgÁmL À1 ampicillin at 37 °C until an optical density of between 0.4 and 0.6 was reached.Protein expression was induced with 1 mM IPTG and cells were incubated at 30 °C for 3 h.Cells were pelleted by centrifugation (3000 g, 15 min, 4 °C) then resuspended in 4 mL TBS buffer with cOmplete, EDTA-free protease inhibitor cocktail (Roche, Basel, Switzerland) and lysozyme (Sigma-Aldrich).Cell suspensions were disrupted by sonication (3 9 20 s) on a SoniPrep 150.Triton X-100 in TBS was added to a final concentration of 1% and samples were centrifuged at 18 000 g, 25 min, 4 °C.Glutathione Sepharose 4B beads (GE Healthcare) were added to supernatants and rotated at 4 °C for 2 h to capture GST-tagged proteins.Beads were washed in 0.5% Triton X-100.Glycerol was added to a final concentration of 15% and stored at À20 °C.

Proteasome activity assay
Proteasome-Glo using Suc-LLVY-AMC for chymotrypsinlike activity assay (G8621; Promega) was used for proteasome assays.One confluent 10 cm dish of cells was washed in PBS before trypsinisation.The collected cells were washed twice in culture media to remove trypsin.Cells were lysed in a lysis buffer (10 mM Tris-HCl, 10 mM NaCl, 2 mM EDTA, 0.5% Triton).Fifty micrograms of protein lysates in 50 lL total volume was used to mix with 50 lL assay reagent.Control samples were treated with the proteasome inhibitor MG132.Samples were loaded in triplicates into a white-walled multi-well plate.After 30 min of incubation, plates were read in a luminometer.
For assays using the activity probe Me4BodipyFL-Ahx3Leu3VS, human fibroblasts were either left untreated or treated with 25 lM MG132 for 2 h before the addition of 500 nM proteasome probe (Me4BodipyFL-Ahx3Leu3VS, UbiQ) to the cell media and incubated for an additional hour.Cells were detached using trypsin (400 lL for a 6 cm dish) and harvested.Cells were lysed in NP40 lysis buffer (50 mM Tris-HCl (pH 7.4), 150 mM NaCl,1% NP40).Protein concentration was determined using a Pierce BCA protein assay kit (Thermo Fisher Scientific).Thirty micrograms of proteins were loaded on a 12% SDS/PAGE gel.The gel was imaged on an Azure 600 imager (Azure Biosystems, Dublin, CA, USA) before the transfer of the proteins on a PVDF membrane to blot for loading control proteins used for normalisation.

Cycloheximide (CHX) chase assay
Cells were seeded equally onto a 6-well plate.At around 80% confluence, the first well was harvested, and the rest of the cells were treated with 100 lgÁmL À1 of CHX.Each subsequent well was harvested at timepoints of 2, 4, 6, 8 and 12 h post-treatment, and frozen on dry ice.Cells were stored at À80 °C until time for lysis in 50 lL RIPA lysis buffer.Samples were resolved by SDS/PAGE and immunoblotted.

mtDNA quantification
Genomic DNA was harvested from cells (Wizard Genomic DNA Isolation Kit; Promega).qPCR reactions were performed in triplicate using SYBR Ò Green JumpStart TM Taq ReadyMix TM (Sigma-Aldrich).The copy number of three mitochondrial genes COX1 (Cytochrome C oxidase 1), CYTB (Cytochrome B) and ATP6 (ATP synthase 6) was normalised against the copy number of the nuclear gene ATP5B (ATP synthase subunit beta) and a nuclear gene desert region of chromosome 10 to calculate the relative mtDNA copy number (oligonucleotide sequences from [51]).

Seahorse Mito stress test
The Seahorse Cell Mito Stress Test was performed according to the manufacturer's instructions.In brief, the day prior to the assay, cells were harvested and plated into a Seahorse XF96 Cell Culture Microplate (Agilent Technologies) at 1000 cells/well and incubated at 37 °C overnight.On the day of the assay, standard cell culture media was replaced with Seahorse XF DMEM medium (Agilent Technologies), supplemented with 2 mM glutamine (Thermo Fisher Scientific), 10 mM glucose (Thermo Fisher Scientific) and 1 mM pyruvate (Thermo Fisher Scientific), at pH 7.4 and cells were incubated at 37 °C for 1 h in a non-CO 2 incubator.Mitochondrial oxygen consumption rate was analysed using the Seahorse XF Cell Mito Stress Test Kit (Agilent Technologies) with 1 lM oligomycin, 1 lM FCCP and 1 lM rotenone/antimycin A. The assay was run at 37 °C on a Seahorse XF96 analyser (Agilent Technologies).OCR measurements were normalised to cell number which was quantified using the CyQUANT TM Cell Proliferation Assay Kit (Thermo Fisher Scientific).Data were analysed using WAVE software (Agilent Technologies).

DCFDA staining for total cellular ROS
Equal numbers of cells were harvested, washed and resuspended in OptiMEM and incubated with 2.5 lM H 2 DCFDA for 30 min at 37 °C.After 30 min, cells were washed and resuspended in PBS and propidium iodide was added to a final concentration of 0.4 lgÁmL À1 .Mean DCF fluorescence in live cells was measured via flow cytometry.

MitoSOX assay
Equal numbers of cells were harvested, washed, and resuspended in HBSS and incubated with either 100 nM Mito-Tracker Green FM (Thermo Fisher Scientific) for 30 min or 2.5 lM MitoSOX (Thermo Fisher Scientific) for 15 min at 37 °C.Cells were washed and resuspended in PBS and mean fluorescence was measured via flow cytometry.Mito-SOX fluorescence was normalised to mitochondrial content.

Mitochondrial membrane potential assay
Equal numbers of cells were harvested and incubated with 100 nM MitoTracker Green FM in DMEM for 30 min at 37 °C.Cells were washed with PBS and incubated with 25 nM TMRE (Abcam) in DMEM for 15 min at 37 °C.Mean fluorescence was measured via flow cytometry.TMRE fluorescence was normalised to mitochondrial content.

Generation of the MitoQC cell lines
The MitoQC construct (mCherry-GFP-FIS1101-152 in pBABE [51]) was transduced into the human fibroblast.In summary, the construct was co-transfected with the VSV-G expressing plasmid into ECO-Phoenix cells (ATCC) using FuGENE (Promega) following manufacturer's instructions.Virus was harvested 48 and 72 h after transfection and filtered through a 0.42 lm filter to remove any ECO-phoenix cells.10 lgÁmL À1 polybrene is added to the viruses and the virus was applied into the cells.Cells were selected with 250 lgÁmL À1 hygromycin.

Fig. 5 .
Fig.5.MiD protein levels are stabilised by Fbxo7 expression.(A) Immunoblots of endogenous MiD49 and MiD51 protein levels in control and patient L250P fibroblast lysates.Protein levels quantified relative to tubulin loading control (n = 3).Mean AE SEM, **P < 0.01, ***P < 0.001.(B) Immunoblots of endogenous MiD49 and MiD51 protein levels in HEK293T cells expressing control and Fbxo7 shRNA.MiD49 and MiD51 protein levels quantified relative to GAPDH and tubulin respectively (n = 3).Mean AE SEM, **P < 0.01, ***P < 0.001.(C) HEK293T cells expressing either control or Fbxo7-targeting shRNA (KD) were transfected with Fbxo7 constructs (wild-type; WT and DF-box).Endogenous MiD49 protein levels in cell lysates were analysed by immunoblot and levels were quantified using IMAGEJ (n = 2).Mean AE SEM.(D) Control and Fbxo7 KD HEK293T cells were transfected with MiD49-Myc and treated with either DMSO or MG132 for 4 h prior to harvesting.MiD49 ubiquitination levels in total cell lysates (no MG132 treatment) were analysed by immunoblotting and quantified using IMAGEJ (n = 5).Mean AE SEM, **P < 0.01.(E) Immunoblots of HEK293T cells stably expressing control (À) and Fbxo7 shRNA (+) were treated with 250 lM H 2 O 2 for up to 6 h.Cell lysates were resolved by SDS/PAGE and immunoblotted as indicated.(F) Quantification of relative levels of MiD49 over time in response to H 2 O 2 (n = 4).For each experiment, MiD49 density was normalised to the density of the loading control.The 0 h time point was set to a value of 1 for each cell line, and the ensuing timepoints were normalised relative to the 0 h time point.Error bars are SD, *P < 0.05, **P < 0.01, ***P < 0.001.

Fig. 6 .
Fig. 6.Mitochondrial content and function are reduced in patient fibroblasts.(A) Control and patient L250P fibroblasts stained with MitoTracker Green FM (MTG) and average mitochondrial content was analysed by flow cytometry (n = 10).Graph of the mean FSC as an indicator of cell size AE SEM, ***P < 0.001; ns, non-significant.(B) Genomic DNA was isolated from control and patient L250P fibroblasts and mtDNA levels were quantified via qPCR.Three mtDNA loci were analysed and normalised to two regions of genomic DNA (n = 5).Mean AE SEM, **P < 0.01.(C) Representative images of control and patient L250P fibroblasts stained with MitoTracker Green FM and imaged via confocal microscopy.Size bar 20 lm; inset size bars 5 lm (n = 3).(D) Cells were stained with MitoTracker Green and imaged via confocal microscopy.Parameters of mitochondrial network morphology were quantified using the MINA IMAGEJ PlugIn.Approximately 20-30 images were analysed per repeat for each cell line (n = 3).Mean AE SEM, *P < 0.05.(E) Control and L250P patient fibroblasts were incubated with 25 nM TMRE and 100 nM MitoTracker Green.Mitochondrial membrane potential (as indicated by TMRE fluorescence) was measured via flow cytometry and normalised to mitochondrial content (MitoTracker Green fluorescence) (n = 5).Mean AE SEM; ns, nonsignificant.(F) Representative Seahorse assay profile and quantification of control and L250P patient fibroblasts analysed by Agilent Seahorse Mito Stress Test (n = 3).Mean AE SD, ***P < 0.001.

Fig. 7 .
Fig. 7. Decreased mitophagy and mitochondrial biogenesis in patient cells.(A) Representative images of mCherry and GFP fluorescence in cells expressing MitoQC reporter.Mitochondria appear yellow in merged images.Treatment with depolarising agent CCCP induces mitophagy which results in red mito-lysosomes.Treatment with Bafilomycin A, which prevents lysosome acidification, reduces the appearance of red mito-lysosomes.Scale bar represents 20 and 3 lm in the inset magnifications.(B) Quantification of flow cytometry analysis of the mCherry/GFP ratio in control and patient L250P fibroblasts, treated with 1 mM DFP for 22 h.Mean AE SD, ****P < 0.0001.(C) Quantification of the fold change in mitophagic cells in control fibroblasts and patient cells.Data are normalised to mitochondrial content.(D) Immunoblot analysis for Parkin in cell lysates from control and patient L250P fibroblasts.GAPDH is used as a loading control (n = 3).Quantification of protein levels normalised to tubulin.Mean AE SD, ****P < 0.0001.(E) Immunoblot analysis for the transcription factors PGC1a and PPARc in cell lysates from control and patient L250P fibroblasts.Tubulin is used as a loading control (n = 3).Quantification of protein levels normalised to tubulin.Mean AE SD, **P < 0.01, ****P < 0.0001.(F) Levels of ROS in control and patient L250P fibroblasts was measured via flow cytometry.Total cellular ROS was assayed by treating cells with H 2 DCFDA and measuring the resulting DCF fluorescence (n = 5).Error bars are SEM, *P < 0.05.(G) Mitochondria-specific ROS was assayed by staining cells with MitoSOX.Resulting fluorescent signal was normalised to mitochondrial content (MitoTracker Green) (n = 3).Error bars are SEM, *P < 0.05.(H) Control and patient L250P fibroblasts were treated with a panel of cellular stressors and stained with propidium iodide.Cell viability was measured via flow cytometry (n = 5).Error bars are SEM, *P < 0.05, **P < 0.01, ***P < 0.001.

Table 1 .
Homozygous variants of interest detected in patient fibroblasts by analysis of whole exome sequencing.