Molecular pathophysiology of human MICU1 deficiency

MICU1 encodes the gatekeeper of the mitochondrial Ca2+ uniporter, MICU1 and biallelic loss‐of‐function mutations cause a complex, neuromuscular disorder in children. Although the role of the protein is well understood, the precise molecular pathophysiology leading to this neuropaediatric phenotype has not been fully elucidated. Here we aimed to obtain novel insights into MICU1 pathophysiology.


INTRODUC TI ON
Mitochondrial disorders are genetically heterogeneous and normally manifest as multisystem diseases with very heterogeneous clinical presentations based on the underlying molecular defect. Many syndromes caused by mutations affecting mitochondrial proteins are associated with a vulnerability of skeletal muscle, in addition to the nervous and ocular system. Mitochondrial diseases associated with myopathies are progressive disorders that produce significant disabilities and may cause premature death. However, within the clinical spectrum of multisystem mitochondrial diseases, the muscular phenotype may be outweighed by other clinical aspects such as neurodegeneration [1].
The mitochondrial Ca 2+ uptake protein 1 (MICU1), encoded by the MICU1 gene is the primary Ca 2+ -sensing regulator of the pore of the mitochondrial Ca 2+ uniporter formed by the mitochondrial calcium uniporter protein (MCU) and Essential MCU regulator (EMRE) [2][3][4]. Loss-of-function mutations in the MICU1 gene cause an early onset multisystem disease characterised by proximal myopathy, learning difficulties and a progressive extrapyramidal movement disorder (OMIM #615673) [5]. MICU1 (−/−) mice either die at birth, likely because of impaired neuronal respiratory control, [6] or are viable but manifest marked ataxia and muscle weakness [7]. At the cellular level, mitochondrial Ca 2+ overload, resulting in altered mitochondrial morphology, reduced ATP-level and increased vulnerability to stress-induced cell death were documented [6,7]. In vitro studies of the functional consequences of MICU1 mutations in patient-derived fibroblasts showed mitochondrial fragmentation related to altered Ca 2+ -induced phosphorylation of DRP1 that initiates fission [8] and influences both cell survival and apoptosis. Thus, it has been assumed that MICU1 serves as a signal-noise discriminator in mitochondrial Ca 2+ signalling that controls oxidative phosphorylation [9], a process of importance in tissues with dynamic energetic demands such as brain and skeletal muscle.
Although altered cellular Ca 2+ homeostasis may impact on proper protein function and abundance, as well as on lipid homeostasis, the overall cellular protein and lipid composition in human MICU1-deficient cells or tissues has not been studied yet. Thus, we aimed to explore the biochemical consequences of MICU1 deficiency in patient-derived cells and muscle tissue. We report two patients with biallelic MICU1 mutations. In a patient-derived lymphoblastoid cell line, the lack of cooperative activation of mitochondrial Ca 2+ uptake were lost in MICU1-deficient cells and that 39 proteins were altered in abundance. Several of those proteins are linked to mitochondrial dysfunction and/or perturbed Ca 2+ homeostasis, also impacting on regular cytoskeleton (affecting Spectrin) and Golgi architecture, as well as cellular survival mechanisms.

Conclusions:
Our findings (i) link dysregulation of mitochondrial Ca 2+ uptake with muscle pathology (including perturbed lipid homeostasis and ER-Golgi morphology), (ii) support the concept of a functional interplay of ER-Golgi and mitochondria in lipid homeostasis and (iii) reveal the vulnerability of the cellular proteome as part of the MICU1-related pathophysiology.

K E Y W O R D S
Mitochondrial degeneration, lymphoblastoid cell proteomics, Spectrin, metabolic diseases, mitochondrial myopathy Uniparental disomy for chromosome 10 (upd [10]) was excluded in index patient 2 by microsatellite analysis of highly polymorphic markers located on chromosome 10. A list of markers used may be provided upon request.

Analysis of copy numbers of exons encoding MICU1
Copy number variants of exons 1, 2 and 3, covering the presumed rearrangement of the MICU1 gene (NM_006077), were analysed in the DNA of patient 2 and his parents using qPCR. The respective qPCR primer sequences and conditions are available upon request. For the detection of larger genomic imbalances, the CytoScan® HD Array (Affymetrix, Santa Clara/CA, USA) was applied. Only CNVs >50 kb with a mean marker distance of <5 kb were considered.

Generation of a lymphoblastoid cell line
Lymphoblastoid cell lines were generated from peripheral blood samples derived from the index patient and the healthy sibling as described previously. [10] Measurements of mitochondrial Ca 2+ uptake in permeabilised lymphoblasts Fluorometric measurements of cytosolic [Ca 2+ ] c were performed as previously described [2]. Briefly, saponin-permeabilised lymphoblasts (2.4 mg) were resuspended in 1.5 mL of intracellular medium containing 120 mM KCl, 10 mM NaCl, 1 mM KH 2 PO 4 , 20 mM Tris-HEPES at pH 7.2 and supplemented with proteases inhibitors (leupeptin, antipain, pepstatin, 1 mg/ml each), 2 mM MgATP, 2 µM thapsigargin (Enzo) and maintained in a stirred thermostated cuvette at 35 °C.

Immunoblotting of MICU1 and EMRE in lymphoblastoid cells
Western blotting of the permeabilised lymphoblast cells was carried out as described before [11].

Proteomic profiling
As proteomic profiling is a powerful tool to obtain unbiased insights into pathophysiological processes, label-free protein quantification utilising liquid chromatography coupled to tandem mass spectrometry has been carried out on lymphoblastoid cells derived from one MICU1 patient and his healthy sibling (as control). For further details see Data S1.

Light, fluorescence and electron microscopy of patient-derived muscle biopsy
Seven-micrometre-thick sections were cut from cryopreserved muscle biopsy tissue. These sections were used for H&E staining and for enzyme histochemistry (NADH, ATPase, COX-SDH) as well as for immunofluorescence (IF). Fluorescence studies were carried out as described previously [12]. Light and electron microscopic investigations of the patient-derived muscle biopsy were carried out as described previously [12,13]. Further immunological studies were performed on 7-µm-thick sections cut from muscle biopsy tissue fixed in formalin and embed-

Coherent anti-Stokes Raman scattering and Second
Harmonic Generation spectroscopy, non-linear unmixing and statistical evaluation of muscle fibre calibres CARS and second harmonic generation (SHG) measurements were performed on a modified Leica TCS SP8 CARS with an APE pico-Emerald laser system. Five-micrometre-thick sections were cut from FFPE tissue blocks and thoroughly deparaffinated. No further sample preparation was applied.
Non-linear unmixing of CARS data was performed as described previously [14] and the statistical evaluation was done on the basis  Immunoblot studies on whole muscle protein extracts SDS-PAGE and subsequent western blot studies on whole protein extracts from patient-derived muscle biopsies have been carried out as described previously [15]. Primary antibodies against β-Spectrin

Clinical presentation of the patients
The index patient was a girl, born as the first child of Turkish consanguineous parents ( Figure 1A) after an uneventful pregnancy (parents had five prior miscarriages; Figure 1A). Whereas development of free walking was observed at age 18 months, speech development was delayed with her first words occurring at age 24 months and first The second patient is a previously health boy who at the age of 22 months, in the course of an acute upper respiratory tract infection, suddenly developed acute ataxia which spontaneously regressed within 2 days. Laboratory studies revealed an elevated CK value of 10,000 U/l which persisted during follow-up. At the age of 32 months, when he first presented to our hospital, he showed bilateral enlargement of the calf muscles, normal tendon reflexes, grossly normal muscle strength with no Gower's sign and normal ability to climb stairs.
He tended to walk on tiptoes, although no fixed pointed foot was present. At age 40 months, the parents reported he had an impaired sensitivity to pain and an increased tendency to fall, possibly related to a marked motor restlessness combined with persistent pointed foot.
This restlessness combined with reduced attention was still noted at age 4 years. At the age of 5 years, at last follow-up, there was no Gower's sign but moderate proximal weakness with reduced exercise capacity attributed to exercise-related proximal muscle pain. Electroand echocardiography proved normal.

Genetic testing
Whole genome sequencing of index patient 1 revealed a homozygous G > A change within the coding region of the MICU1 gene on chromosome 10 (chr10: 74268012; c.553C>T; exon 6 of 12; Figure 1C). This nucleotide change results in a premature stop codon at amino acid position 185 p.(Arg185*), which has been described previously as being a pathogenic founder mutation in 13 patients from the Middle East [16]. This homozygous mutation is expected to lead to deficiency of the MICU1 protein, which was validated in the immortalised lymphoblastoid cells by immunoblotting ( Figure 6B) and highly likely to be responsible for the disorder manifesting in the index patient. Segregation of the mutation in the family could be demonstrated ( Figure 1C).
In patient 2, next generation sequencing (NGS)-based testing of myopathy genes using an Illumina enrichment kit revealed homozy-  Figure 1E). The exact size of the deletion was not determined, and SNP array analysis failed to detect this relatively small deletion. Analysis of parental blood samples revealed that the mother is a heterozygous carrier of the c.52C>T (p.(Arg18*)) sequence variant, whereas the deletion is of paternal origin ( Figure 1D,E). The patient is therefore compound heterozygous for two loss-of-function variants in the MICU1 gene.

Biopsy findings
Muscle histology including H&E, NADH, COX-SDH, ATPase and toluidine blue staining showed fibre calibre variation, occasionally internalised myonuclei, fibre-type grouping, groups of lesioned basophilic fibres, focally increased NADH-labelling consistent with irregular focal accumulation of mitochondria and/or sarcoplasmic reticulum in some muscle fibres, groups of COX-negative fibres and subsarcolemmal increase of enzyme histochemical activity ( Figure 2).  Figure S1).
The evaluation of the unmixing enables us to compare different parts of the fibre. Here, we compare the resulting spectra of the interior and the border of the fibre ( Figure 4B, Figure S1). The spectra of the interior of control (endmember 1) and disease control Furthermore, CARS microscopy revealed the presence of areas with low signal intensity, sized between 0.5 and 1.5 µm (Figure 4).
Those areas, which most likely represent (autophagic) vacuoles, show considerably lower peaks for all three wave numbers mentioned above, thus suggesting that those are not densely packed with aggregates of lipids or proteins. However, the disease controls also present with vacuoles, but with diameters ranging from 0.5 to 20 µm. Furthermore, rimmed vacuoles were found in one disease control and filled vacuoles in the two disease controls (black arrows in Figure S1).
Moreover, degenerating muscle fibres presented with granular appearance in CARS microscopy defined by a dominant peak F I G U R E 4 CARS microscopic findings in the patient-derived muscle biopsies. (A) CARS and SHG contrast images of MICU1 patient and control muscle biopsies. The MICU1 patient-derived biopsies show clusters of muscle fibres with particularly pronounced spot patterns (white frames) not identified as such in the control and disease control muscles. Cells within these cluster present with increased protein (2930 cm −1 (ѵ s (=CH 3 )) and lipid (2889 cm −1 (ѵ as (=CH 2 ), 2840 cm −1 (ѵ s (=CH 2 )), as well as SHG signals. Scale bars 60 µm. (B) Round degenerating muscle fibres (black arrows) and fibres with reduced eosin staining (white arrow) were detected by H&E staining. In the superimposed CARS/ SHG images (scale bar 200 µm), round muscle fibres of similar structure were also detectable, sometimes as groups or clusters (uppermost arrow). These cells have an internal granular structure (scale bar 5 µm) and granules have a higher intensity at 2847 cm −1 than at 2889 cm −1 indicating a lower lipid order (1). Besides, there are also elongated accumulations with a higher lipid order (2). Both structures show comparable protein signals at 2930 cm −1 . Outside of these structures, the intensity of the three peaks is distinctively lower (3). The round muscle fibres are surrounded by SHG signals, maybe hinting to collagen or collagen-like structures. (C) Image of muscle fibre of a MICU1 patient with autophagic vacuoles taken at 2885 cm −1 (lipid spectra). Spectra are shown from different locations: low intensities of protein and lipids are measured within the vacuoles (1). Two typical spectra recorded around the vacuoles are also presented, which mostly differ in intensity (2, 3). Scale bar 2.5 µm. (D) In the superimposed CARS/SHG images (scale bar 200 µm), round muscle fibres of similar structure were also detectable; sometimes as groups or clusters (uppermost arrow). These cells have an internal granular structure (scale bar 5 µm) and granules have a higher intensity at 2847 cm −1 than at 2889 cm −1 indicating a lower lipid order (1). Besides, there are also elongated accumulates with a higher lipid order (2). Both structures show comparable protein signals at 2930 cm −1 . Outside of these structures, the intensity of the three peaks is distinctively lower (3). The round muscle fibres are surrounded by SHG signals, maybe hinting to collagen or collagen-like structures. (E) Comparison of the average muscle fibre calibres in the comparison of the controls with MICU1 patients. Statistical details are given in supplemental document 1. Star: difference to all controls is of high statistical significance MOLECULAR PATHOPHYSIOLOGY OF MICU1 DEFICIENCY | 847 a wave number 2840 cm −1 suggesting that lipid organisation is lower than observed in the above mentioned small-calibre fibres ( Figure 4B). In addition, degenerating fibres are characterised by a SHG-signal suggestive for the presence of structured protein aggregates, a hallmark of cell degeneration [18,19]. To quantify changes in muscle fibre calibres, the diameter of muscle cells were

Lymphoblastoid patient cells lack MICU1 and display dysregulation of mitochondrial Ca 2+ uptake
Lymphoblastoid cells derived from the homozygous patient 1 (red), healthy sibling (black) and heterozygous father (grey) were permeabilised in suspension and were used for fluorometric measurements of mitochondrial Ca 2+ uptake ( Figure 6A)  | phenocopied the pattern previously observed with other MICU1deficient cells [2,6]. Lysates prepared from the cells were used for immunoblotting that showed loss of the MICU1 band in the homozygous patient ( Figure 6B). Notably, the MICU1 bands as well as the mitochondrial Ca 2+ uptake phenotypes were similar in the healthy sibling (control) and the heterozygous father ( Figure 6AB).
Immunoblot for EMRE that is required for the assembly of the MCU pore also showed a decrease in the homozygous patient, replicating a likely compensatory decrease in the uniporter abundance, which has been documented previously in cell lines [7,11,20] but not in patients [9]. This result has indicated that loss of MICU1 can initiate changes in the abundance of other proteins which are not directly targeted by the genetic impairment.

MICU1 mutations alter protein compositions in lymphoblastoid cells
Proteomic profiling utilising lymphoblastoid cells derived from patient 1 and the healthy sibling were carried out and allowed the

MICU1 deficiency-associated phenotypes
Loss-of-function point mutations have been identified in MICU1 as the cause of a rare neuromuscular and neurodevelopmental disease in children [13]. So far, the clinical spectrum of MICU1 patients included progressive proximal muscular dystrophy, elevated CK levels, generalised muscle weakness, positive Gower's manoeuvre, mild intellectual disability or learning difficulties and variable features such as episodic ataxia, skin involvement, ptosis, hypometropia, pendular nystagmus, bilateral optic atrophy, microcephaly and peripheral axonal neuropathy as well as elevated transaminase enzymes [5,9,16].
In addition, two cousins with a homozygous deletion involving exon 1 (chr 10: 74,385,085-74,387,860, size: 2,775 nucleotide pairs) of MICU1 were described: one case presented with fatigue and lethargy associated with mild hypotonia, and global muscle weakness among other symptoms such as episodes of accompanied viral infections but normal psychomotor development. Remarkably, the other case carrying the same homozygous deletion presented with a more pronounced phenotype characterised by additional migraines, pendular nystagmus, cataracts and mild learning difficulties [22]. Muscle biopsy studies commonly revealed rare atrophic fibres and increased internal nuclei. The variable clinical presentation resulting from MICU1 deficiency in this family underlines the broad clinical spectrum of the disease and shows that a clear genotype-phenotype correlation did not emerge. Further 'larger' genetic rearrangements associated with the clinical manifestation of a MICU1 phenotype include a heterozygous intragenic duplication of exons 9 and 10 [16] and a homozygous deletion of exon 1 [22].
Here, we describe two new MICU1 patients, one carrying the recently described homozygous Middle Eastern founder mutation  [23], and these organelles in turn form a functional continuum due to the formation of mitochondria-associated ER membranes (MAMs), also involved in the transport of Ca 2+ [24][25][26] and based on lipid synthesis [17], an influence of MICU1 deficiency on proper cellular lipid homeostasis can be hypothesised. In this context, immunoblot studies of Seipin (BSCL2), a protein mediating the formation and/ or stabilisation of endoplasmic reticulum-lipid droplets, revealed increased abundance in MICU1 patient-derived muscles ( Figure S2).
In addition, the absence of a respective lipid peak in degenerating muscle fibres suggests that the increased appearance of organised

Mitochondrial and cytoskeletal pathology caused by deficiency of functional MICU1
The dysregulation of 16 proteins (40% of all vulnerable proteins) can be directly linked to altered mitochondrial Ca 2+ handling and thus MICU1 dysfunction resulting in ultrastructural mitochondrial pathology (swollen mitochondria, cristae-disorganisation and build-up of aggregates within the mitochondrial matrix; Figure 3). This biochemical finding accords with the proven perturbed Ca 2+ homeostasis in the patient-derived lymphoblastoid cells ( Figure 6) and provides novel insights into the biochemical consequences. Importantly, perturbed Ca 2+ homeostasis has already been described in fibroblasts derived from MICU1 patients. [9] A growing body of evidence suggests that mitochondrial morphology and function are modulated by the cytoskeleton via mostly uncharacterised pathways [27] and our proteomic findings hint towards a vulnerability of cytoskeleton upon loss of MICU1 by altered abundance of SPTBN1, SPTAN1, DBN1, VIM and COTL1. Vulnerability of Spectrin could also be observed by immunological investigations in the muscle biopsies of our two MICU1 patients ( Figure 5). Interestingly, Ca 2+ is known to modulate the stabilisation of Actin-Spectrin complexes [28] and altered mitochondria with impaired capacity to buffer receptor-gated Ca 2+ fluxes have already been linked to promotion of Ca 2+ -activated proteases and subsequent degradation of the Spectrin meshwork [29], thus confirming a tight connection between mitochondrial and Spectrin homeostasis. In addition, DBN1, VIM and COTL1 are also proteins that bind to actin in a Ca 2+ -dependent manner [30][31][32], which functioning in concert with Spectrins, modulate actin dynamics. Of note, immunoblotting studies on patient-derived muscle protein extracts revealed decreased abundance of further costameric proteins including Dystrophin, β-Dystroglycan and Desmin supporting the concept of a perturbed anchoring of the actin cytoskeleton to the sarcolemma as part of MICU1-associated pathophysiology. In addition, perturbed anchoring of the Actin cytoskeleton to organelles and/ or the sarcolemma is suggested by the irregular sarcoplasmic increase of ITGB4 and ACTN3, two proteins involved in anchoring of Actin to a variety of intracellular structures, in MICU1 patients ( Figure S2).
However, additional functional und structural studies on single isolated fibres would be needed to elucidate this pathobiochemical interplay more precisely. Given that impairment of cytoskeletal dynamics and mitochondrial dysfunction are commonly observed in many neurodegenerative and neuromuscular disorders [1,[33][34][35][36][37], it seems to be plausible that vulnerability of regular cytoskeletal composition results from MICU1 deficiency-based mitochondrial dysfunction and significantly contributes to the phenotypical manifestation of loss-of-function MICU1 mutations in different tissues. Remarkably, β-III-Spectrin spinocerebellar ataxia type 5 mutation reveals a dominant cytoskeletal mechanism that underlies dendritic arborisation [38].

ER-Golgi pathology and the presence of autophagic vacuoles caused by loss of functional MICU1
Altered abundance of ER-Golgi proteins might accord with mitochondrial vulnerability and could represent a secondary pathophysiological mechanism: proteomic profiling revealed increased Syntaxin-5 (SNX5), which belongs to the SNARE complex facilitating the transport of cholesterol to mitochondria [39] (lipid accumulation has been identified in MICU1 patients based on our CARS microscopic studies), and SNX5 is also involved in vesicle tethering and fusion at the cis-Golgi membrane to maintain the stacked and inter-connected structure of the Golgi apparatus. Interestingly, our immunological KOHLSCHMIDT eT aL. | studies on muscle biopsy specimen confirmed an increase of SNX5, as well as an increase of POC1A, an ER-Golgi-resident protein, in a proportion of MICU1-mutant muscle cells ( Figure S2). In addition, electron microscopy showed a build-up of proliferated and de-organised ER-Golgi structures (Figure 3). A potential impact of altered Golgi architecture-as observed in the muscle biopsies-on regular vesicular transport can also be deduced from our proteomic findings through the altered abundance of vesicular trafficking regulators such as VPS37B and AP-3 complex subunit mu-1 (AP3 M1): VSP37B is a Ca 2+ -dependent binding protein of the endosomal sorting complexes required for transport (ESCRT) required for vesicular trafficking processes [40]. AP3 M1 is part of the AP-3 adaptor complex, which associates with the Golgi, facilitates the budding from the Golgi and may be directly involved in trafficking to the endosomal or lysosomal system [41]. The expression of the AP-3 complex is essential for the biogenesis of acidic Ca 2+ and polyphosphate organelles (so called acidocalcisomes) including lysosomes [42] with various functions including autophagy [43]. Interestingly, the results of our Niban has recently been linked to autophagy-induction activated by perturbed Ca 2+ homeostasis and ER stress [44]. Potential consequences of MICU1 mutations on the intracellular Ca 2+ dynamics in ER-mitochondria crosstalk [45] are further supported by the identified increased abundance of Endoplasmin, a Ca 2+ -buffering chaperone of the ER [46], in both MICU1 patient-derived lymphoblastoid cells and muscle. Also, the increase of BiP and Calreticulin in patientderived muscle cells accords with this concept.

Activation of pro-apoptotic and pro-survival mechanisms caused by MICU1 mutation
Apart from a role of AP-3 in Golgi maintenance and protein clearance, a correlation between AP-3-function and the interferon signalling pathway has been described [47], and dysregulation of interferon proteins (along with caspases and B-cell lymphoma-protein 2 (BCL2); Tab. 1) as observed in our proteomic data is known to lead to cellular apoptosis by inducing the expression of several apoptotic regulators (including caspases and BCL2) via a mitochondrial pathway [48]. Along this line, increased abundance of SBP10 and TNF receptor-associated factor 1 (TRAF1) accords with potential induction of apoptosis [49,50]. In this context, it is important to note that a recent study demonstrated the control of TRAF1 along with CAMK4 (also increased in our in vitro system) on the molecular genetic level upon muscle loading highlighting their involvement in muscle fibre hypertrophy and thus muscle cell survival [51]. Induction of apoptotic processes in MICU1 patient-derived muscle cells is indicated by the combined increase of CASP1 and CytC ( Figure S2), in turn supporting the relevance of proteomic findings obtained in the lymphoblastoid cells. However, an increased abundance of proteins with neuroprotective properties have also been identified, in agreement with the concept of activation of pro-survival mechanisms: Uridine 5'-monophosphate synthase (UMPS) is involved in synthesising UMP from orotate and deficiency of UMPS has been linked to hereditary orotic aciduria, which is associated with some degree of physical and mental retardation [52]. Succinate-semialdehyde dehydrogenase (ALDH5A1) catalyses one step in the degradation of the inhibitory neurotransmitter gamma-aminobutyric acid [53]. Additionally, MTHFD2 (human MTDC), also elevated in the lymphoblastoid cells of our patient, modulates neuronal development via mitochondrial formate production [54,55] and thus most likely antagonises mitochondrial vulnerability in tissue upon loss of functional MICU1.

Biochemical and clinical synopsis
We hypothesise that impaired Ca 2+ dynamics and concomitant in- However, biochemical studies on cells and/or muscle biopsies of further MICU1 patients confirming the data described here are needed to finally declare our findings as pathophysiological cascades resulting from the loss of functional MICU1.

E THI C S APPROVAL :
The ethics committee of University Medicine Essen  had granted ethical approval. The legal guardians of the patients gave consent for the performed analyses.