Mitochondrial morphology and cellular distribution are altered in SPG31 patients and are linked to DRP1 hyperphosphorylation

Abstract

Hereditary spastic paraplegia, SPG31, is a rare neurological disorder caused by mutations in REEP1 gene encoding the microtubule-interacting protein, REEP1. The mechanism by which REEP1-dependent processes are linked with the disease is unclear. REEP1 regulates the morphology and trafficking of various organelles via interaction with the microtubules. In this study, we collected primary fibroblasts from SPG31 patients to investigate their mitochondrial morphology. We observed that the mitochondrial morphology in patient cells was highly tubular compared with control cells. We provide evidence that these morphological alterations are caused by the inhibition of mitochondrial fission protein, DRP1, due to the hyperphosphorylation of its serine 637 residue. This hyperphosphorylation is caused by impaired interactions between REEP1 and mitochondrial phosphatase PGAM5. Genetically or pharmacologically induced decrease of DRP1-S637 phosphorylation restores mitochondrial morphology in patient cells. Furthermore, ectopic expression of REEP1 carrying pathological mutations in primary neuronal culture targets REEP1 to the mitochondria. Mutated REEP1 proteins sequester mitochondria to the perinuclear region of the neurons and therefore, hamper mitochondrial transport along the axon. Considering the established role of mitochondrial distribution and morphology in neuronal health, our results support the involvement of a mitochondrial dysfunction in SPG31 pathology.

Introduction

Hereditary spastic paraplegia (HSP) is a large and heterogeneous family of inherited neurodegenerative disorders, characterized by a length-dependent distal axonopathy of the corticospinal tracts, resulting in progressive lower limb spasticity and weakness (1,2). These features are related to a dying-back axonal degeneration that precedes cell-body degeneration in the longest fibers of the corticospinal tracts and dorsal columns, sometimes associated with a loss of cortical neurons and anterior horn cells (3,4). The severity of HSP is highly variable in terms of age at onset, its course, and phenotypic presentation (5). To date, 76 genes or loci (named SPG, a term that is also used to describe the associated HSP types) have been identified as being involved in HSP determinism. SPG causative genes encode various proteins involved in numerous intracellular functions, e.g. organelle/vesicle trafficking and dynamics, organelle shaping, cytoskeletal organization, or lipid metabolism (6,7). However, and despite an intense research focus, the functions of many genes and their relationship with the disease onset and progression are still unclear (6).

REEP1 encodes Receptor Expression-Enhancing Protein 1 (REEP1), which was initially described as involved in odorant receptor trafficking (8,9). Different mutations in REEP1 have been found in autosomal dominant HSP (SPG31) (10–12). Following the identification of REEP1 mutations, Zuchner et al. proposed that REEP1 localized at the mitochondrial membrane using immunofluorescence (10). However, this localization of REEP1 is still debated. Indeed, Park and colleagues showed that REEP1 localized in the ER (13). They demonstrated that REEP1 modulates the shape of ER through physical interaction with SPASTIN and ATLASTIN, mutated in SPG4 and SPG3A, respectively, and at the organelle/microtubule interface (13). Using REEP1-KO mice, Beetz and colleagues verified that REEP1 participates in the ER membrane shaping in the neurons (14). A recent study also suggested a role of this protein in the regulation of lipid droplets (15–17). Furthermore, ectopic REEP1 expression leads to its various localization profiles, arguing in favor of a ubiquitous intracellular localization of this protein (18). Taken together, these results emphasize the notion that REEP1 likely controls the shape/dynamics of different endomembranes, including ER, trafficking vesicles, and lipid droplets, through a microtubule-associated role. Recently, another study reconciled the discrepancies concerning the subcellular localization of REEP1 by showing that the protein regulates the contact between ER and mitochondria at mitochondria-associated membrane sites (MAMs) (19).

MAMs are very important for mitochondrial morphology. Many mitochondrial shaping proteins, such as the mitochondrial fusion protein MFN2 and mitochondrial fission protein DRP1, are located at or recruited to these sites (20,21). Our group has previously reported that mitochondria of a patient who bore truncated REEP1, with amino acids 36–201 deleted (p.V36Sfs*4), had an abnormal tubular shape (12). In addition and of note, this patient’s fibroblasts and muscle biopsy exhibited a reduced oxidative phosphorylation activity compared with healthy controls, in agreement with previous work showing that hyperfused mitochondrial networks are associated with oxidative phosphorylation deficiency (22). Interestingly, alterations of mitochondrial function and shape have already been characterized in HSP with mutations in PGN (SPG7) and HSP60 (SPG13), which encode mitochondrial proteins, Paraplegin and HSP60, respectively (23–25). Moreover, a protein interacting with REEP1, SPASTIN, has been shown to strongly affect mitochondrial dynamics and morphology through the regulation of microtubule dynamics (26,27).

Although REEP1 involvement in the morphology of various organelles has been established (13), REEP1-dependent regulation of mitochondrial dynamics in SPG31 remains unclear. Here, we demonstrate that mitochondrial dynamics are directly affected by REEP1 dysfunction in SPG31 and that this alteration impacts neuron health.

Results

REEP1 mutations result in altered mitochondrial morphology

In a previous study, we have reported that mitochondrial morphology and mitochondrial respiration were altered in fibroblasts and muscle biopsy from one SPG31 patient (p.V36Sfs*4) (12). To further investigate the role of REEP1 in mitochondrial morphology, fibroblasts were obtained from different healthy individuals (controls) and from five SPG31 patients who carried different mutations in REEP1 gene, including truncating and missense mutations (Table 1). REEP1 mRNA expression is mainly in the brain (Fig. 1A). However, we also confirmed the endogenous expression of both REEP1 mRNA and protein in patient and control primary fibroblasts (Fig. 1B and C). No significant differences were observed between patients and control fibroblasts regarding mRNA and protein expression levels. Examination of mitochondrial morphology revealed that mitochondria were highly reticulated and elongated in fibroblasts from the different patients (Fig. 1D and E). In average, 55.9 ± 3.0% of patient cells displayed tubular network compared to 26.5 ± 2.1 in control cells (Values are Means ± SEM, n = 4–5, P = 0.0159 control vs patient, Mann-Whitney non-parametric test). Furthermore, in all SPG31 patient cells, we found that the levels of various mitochondrial proteins (SDHA, ATP5A, and mtTFA) were increased 3- to 5-fold (Supplementary Material, Fig. S1A). A similar increase was observed for the mitochondrial DNA/nuclear DNA ratio but citrate synthase activity were not different between patients and control cells (Supplementary Material, Fig. S1B and C). Finally, electron microscopy revealed increased mitochondrial length and area in P2 (SPG31 Patient 2) cells compared with control (Supplementary Material, Fig. S1D and E) as well as the presence of mitochondrial budding and swollen ER in patient’s fibroblasts (Supplementary Material, Fig. S1F). Taken together, these observations support the notion that mitochondrial morphology is altered in SPG31 patient cells.

Figure 1

Fibroblasts of SPG31 patients display normal REEP1 expression but aberrant, tubular mitochondrial morphology. (A) Quantitative expression of REEP1 mRNA in different human tissues assayed by by RT-qPCR. Values are normalized to reference genes, as specified in Materials and Methods. They are presented as mean ± SEM (n = 3). (B) Quantitative expression of REEP1 mRNA in patients (P1–5) and controls (C1–3) assayed by RT-qPCR. Values are normalized to reference genes, as specified in Materials and Methods. They are presented as mean ± SEM (n = 3). (C) Western blot analysis of REEP1 expression in fibroblasts of SPG31 patients (P1–5) and controls (C1–2). (D) Representative immunofluorescence micrographs of mitochondrial staining in SPG31 patient and control fibroblasts. Mitochondrial fluorescent labelling was achieved using specific antibodies against mitochondrial protein TOM20. Bars, 20 µm. (E) Quantification of mitochondria in control (C1–4) or patient (P1–5) cells according to three morphotypes: fragmented (white), tubular (black), and intermediary (grey). Bars represent the percentage of cells for each phenotype (> 100 cells were counted in 5 independent experiments).

Table 1

Clinical description of patients (P1-5) and controls (C1-4) involved in this study.

	Patients					Controls
N°	P1	P2	P3	P4	P5	C1	C2	C3	C4
REEP1 Mutation	c.106delG, p.V36Sfs*4	c.166G>A, p. D56N	c.124T>C , p.W42R			none	none	none	none
Sex	M	M	F			F	F	F	M
Family	N08 1658	FSP 321		FSP-418		–	–	–	–
Phenotype	Pure HSP	Pure HSP	Complex HSP (neuropathy)			–	–	–	–
Age at exam (yrs)	48	56	49	45	34	26	46	18	54
Age at onset (yrs)	25	5	18	–	15	–	–	–	–
Disease duration (yrs)	23	61	31	–	19	–	–	–	–
LL Spasticity	Moderate	Moderate	Moderate	Mild	Moderate	–	–	–	–
LL reflexes	( ++)	No	( +++)	( ++)	( +++)	–	–	–	–

REEP1 mutation inhibits mitochondrial fragmentation in a DRP1 phosphorylation-dependent manner

To investigate the molecular mechanism linking REEP1 and the regulation of mitochondrial morphology, we performed a mitochondrial fusion assay using a mitochondrial membrane potential uncoupler, carbonyl cyanide-4-(trifluoromethoxy)phenylhydrazone (FCCP). After 30 min of FCCP treatment, the mitochondrial network was fully fragmented in the entire cell population in both patient and control samples (Fig. 2A). After FCCP washed out, mitochondrial morphology progressively recovered its initial phenotype through promotion of intramitochondrial fusion (Fig. 2A and B). Within 90 min, the tubular mitochondrial network was restored in patient cells, which signified that mitochondrial fusion was not impaired in these cells (Fig. 2B). Therefore, we asked whether the abnormally elongated mitochondrial morphology could have resulted from inhibition of mitochondrial fission. Although the level of DRP1, a key protein involved in mitochondrial fission, was low in cells obtained from the patient carrying the truncating mutation (P1, p.V36Sfs*4), no significant differences were observed between the controls and patients (Fig. 2C). Expressions of mitochondrial fusion proteins, MFN1 and 2 and OPA1, were not modified in patients cells (Fig. 2C). We next analysed the phosphorylation status of serine 637 (S637) of DRP1, known to be involved in the mitochondrial fission process (28). Interestingly, all patients showed high levels of DRP1-S637 phosphorylation (Fig. 2C and D). The DRP1-S637/total DRP1 ratios were 2–5-fold higher in patients compared with the control samples, with a mean value of 3.23 ± 0.08 (P < 0.01, control mean vs. patient mean, n = 4) (Fig. 2D). No changes were observed in the other DRP1 phosphorylation site (S616 residue) (not shown). Accordingly to previous data showing that S637 phosphorylation hampers DRP1 GTPase activity as well as its hetero- and homo-oligomerization (29–31), we found that high molecular weight DRP1 complexes were less prominent in patient cells than in controls (Fig. 2E).

Figure 2

Mitochondrial fission is altered in SPG31 fibroblasts. (A) Mitochondrial fusion assay with control and patient fibroblasts. Cells were stained with MitoTracker reagent and mitochondrial shapes were analysed by live-imaging before FCCP treatment (10 µM), 10 min after FCCP treatment, and then 0, 30, 60, and 90 min after FCCP washout, as indicated. (B) Percentage of cells with fragmented mitochondrial networks in patient P1 and P2 (grey lines) and control C1 and C2 (black lines), quantified 0, 30, 60, and 90 min after FCCP washout (n = 3, 40 cells). (C) Representative western blot showing the total expression of DRP1 and its serine 637 (S637) phosphorylation levels. Levels of OPA1, MFN1 and MFN2 were also analysed and actin was used as loading control. Western blots were performed using total protein extracts from patient and control fibroblasts. (D) Quantification of DRP-S637 phosphorylation/total DRP1 ratios obtained by western blots. Bars represent the ratios for different patient (black) and control (white). Each bar is the mean ± SEM of 4 independent experiments. (E) Representative DRP1 oligomerization profile, as investigated by western blotting in control and patient cell extracts in the presence of protein crosslinker (BMH). Relative of protein loading was verified by immunoblotting of TOM20 and DRP1 in the same but BMH-untreated extracts. These results are representative of 3 independent experiments.

The possible link between DRP1 phosphorylation and the alteration of mitochondrial morphology in SPG31 patients was further investigated using both pharmacological and molecular biology approaches. Since DRP1-S637 phosphorylation depends on cAMP-dependent protein kinase (PKA) (29), patient and control cells were treated with a selective PKA inhibitor (H89) or vehicle (DMSO). Consistently with previous studies (32), we found that H89 significantly decreased DRP1 phosphorylation at S637 in patients, by 50 ± 14%, compared with the vehicle and didn’t affect the level of total DRP1 (Fig. 3A). The H89 treatment rescued the mitochondrial morphology observed in patients (Fig. 3B and C). Then, patient and control cells were transfected with yellow fluorescent protein (YFP)-tagged DRP1-S637A or (YFP)-tagged WT-DRP1 constructs (Fig. 3D). DRP1-S637A is a mutated form of DRP1 that cannot be phosphorylated (28). YFP-tagged DRP1-S637A accumulated in the mitochondria and this accumulation was associated with significantly decreased mitochondrial tubular morphology in patient cells (Fig. 3D and E). These decreases were significantly smaller in YFP-WT-DRP1 transfected cells showing the importance of the phosphorylation residue. Altogether, these results supported our hypothesis that DRP1 hyperphosphorylation at S637 is associated with the alteration of the mitochondrial morphology in SPG31 patients and, importantly, these results indicated that these alterations are reversible.

Figure 3

Tubular mitochondrial morphology in SPG31 patient fibroblasts depends on DRP1-S637 phosphorylation. Tubular mitochondrial morphology in SPG31 patient fibroblasts depends on DRP1-S637 phosphorylation. (A) Representative western blotting verifying the pharmacological inhibition of DRP1-S637 phosphorylation by PKA inhibitor (H89). Cells were treated for 1 h with 20 µM H89. This image is representative of 5 independent experiments. (B) Rescue of mitochondrial morphology in patient H89-treated fibroblasts analysed by immunofluorescence with TOM20-labelled mitochondria. The images are representative of 4 independent experiments. (C) Mitochondrial morphology after DMSO (dark grey) and H89 treatment (white) (n = 4, >100 counted cells). (D) Analyses of mitochondrial morphology (TOM20, red) by immunofluorescence in SPG31 and control fibroblasts transfected with DRP1-S637A-YFP or WT-DRP1-YFP (yellow). (E) Quantification of tubular morphology in patient cells transfected with DRP1-S637A-YFP (white), WT-DRP1-YFP (light grey) and untransfected patient cells (dark grey) (n = 4–6, 40 to 80 counted cells by experiment). Results are given as mean ± SEM (*P < 0.05, **P < 0.01, ***P < 0.001, one-way ANOVA, Sidak’s multiple comparaison test). Bars, 0.2 µm.

REEP1 directly regulates DRP1 phosphorylation

To further clarify the link between REEP1 and DRP1 phosphorylation, we first analysed the localization of REEP1. Indeed, REEP1 localization is highly discussed in the literature (8,10,13,16,19). Using commercially available antibodies, we found that REEP1 partially colocalized with the mitochondria (Supplementary Material, Fig. S2A and B). However, these antibodies displayed unspecific nucleus staining in immunofluorescence assays, precluding reliable quantification. Therefore, we used exogenously expressed REEP1-Myc to investigate REEP1 localization. In HeLa cells expressing REEP1-Myc, intracellular distribution of REEP1 was highly heterogeneous (Fig. 4A). Non-mitochondrial localization of REEP1 was observed in 60% of transfected cells, while in the remaining 40% of cells REEP1 colocalized with the mitochondria (Fig. 4A). The overall Pearson’s colocalization factor (REEP1 vs mitochondria) was 0.45 ± 0.04 in cell expressing WT REEP1-myc. In contrast, in cells expressing mutated REEP1 (W42R, D56N or Δ), REEP1 displayed a higher degree of mitochondrial localization. Pearson’s colocalization factors were equal to 0.78 ± 0.03, 0.86 ± 0.04, and 0.80 ± 0.04, respectively, in these cells (Fig. 4B). While the mitochondrial localization of Δ36-201-REEP1-Myc could be unspecific, caused by a residual transmembrane domain, the results obtained with W42R- and D56N-REEP1 were not unexpected. Indeed, recently, Lim et al. proposed that REEP1 contains subdomains for both mitochondrial and ER localization (19). The N-terminal domain (1 − 115 amino acids) of the protein was shown to induce ER localization, whereas the middle domain (116 − 157 amino acids) promoted mitochondrial localization (19). These results suggested that REEP1 mutations altering the ER localization domain enhance mitochondrial localization of this protein. Accordingly, the presence of REEP1 in cytosolic and mitochondrial-enriched fractions was assessed by cell fractionation and mitochondrial localization of REEP1 was verified (Fig. 4C). Transfection efficiency and protein expression were verified in total cell lysate (Supplementary Material, Fig. S2C). As the mitochondrial compartment is associated with ER through the MAMs, we used a mild detergent (digitonin) to remove these ER membranes and we found that both mutant and WT REEP1 strongly bound to the purified mitochondria (Fig. 4D). On the other hand, Δ-REEP1-Myc was easily removed from the mitochondrial membrane, showing again that this truncated protein behaved differently than proteins with missense mutations. Then, we analysed how REEP1 mutation impact DRP1 phosphorylation. While DRP1-S637 phosphorylation levels were strongly increased in these cells expressing shRNA against REEP1 (Supplementary Material, Fig. S3A and B), the downregulation induced also a significant decrease of DRP1 levels (Supplementary Material, Fig. S3A). These results suggested that REEP1 expression level participated to the control DRP1 stability, possibly by affecting its phosphorylation status. Given this effect of shRNA on DRP1 stability, we used constructs to ectopically express WT-REEP1 or REEP1 carrying pathological mutations in the cell lines. We observed that the stability of DRP1 was not affected in cells expressing mutant proteins (Fig. 4E). We also found that rates of DRP1-S637 phosphorylation were increased by 78 ± 23% and 48 ± 12% in HeLa cells expressing mutated REEP1, D56N-REEP1-Myc and W42R-REEP1-Myc, respectively, compared with cell expressing REEP1-Myc (Fig. 4F). In comparison, ectopic expression of Δ36-201-REEP1-Myc (‘Δ’) induced only a moderate and non-significant increase in DRP1-S637 phosphorylation (37 ± 10%) (Fig. 4E and F). This moderate increase could have been attributed to a compensatory effect of endogenous expression of HeLa REEP1. Importantly, the PKA overall activity was unchanged in cells transfected with REEP1 mutant constructs (Supplementary Material, Fig. S3C). Finally, by immunoprecipitation, we found that myc-REEP1 interact with PGAM5, a mitochondrial serine/threonine-protein phosphatase known to dephosphorylate DRP1 at S637 residue (Fig. 4G). Although this interaction need to be further investigated, it provides an molecular link between REEP1 and the phosphorylation status of DRP1.

Figure 4

REEP1 mutations promote it mitochondrial localization and increase DRP1 phosphorylation. (A) Analysis of mitochondrial REEP1 localization in HeLa cells transfected with Myc-tagged wild-type REEP1 (WT), Myc-tagged Δ36-201-REEP1 (Δ), Myc-tagged D56N-REEP1 (D56N), and Myc-tagged W42R-REEP1 (W42R). The mitochondria and REEP1 were visualized by immunostaining of TOM20 (Green) and Myc-tag (Red), respectively. Bars, 0.25 µm. (B) The degree of colocalization of mitochondria (TOM20 staining) and REEP1 (Myc-tag staining). Each dot represents the Pearson’s coefficient for a single cell (n = 4, ***P < 0.001, mutant vs. WT, Krustal-Wallis multiple comparison test). (C) Cytosolic and mitochondrial distribution of Myc-tagged-REEP1. Cytosolic and mitochondrial fractions were obtained using cells transfected with WT, Δ, D56N-, or W42D-REEP1. (D) Mitochondrial subfractionation of HeLa cells transfected with the different REEP1 mutants, REEP1 WT, or untransfected cells (NT). UBCQ2 and CPT1A were used as inner mitochondrial and outer mitochondrial membrane markers, respectively, and calregulin was used to stain mitochondria-associated ER. REEP1 was detected using anti-Myc antibody. (E) The effect of pathological REEP1 mutations on DRP1 phosphorylation levels assessed by western blotting. HeLa cells were transfected with either Myc-tagged wild-type REEP1 (WT), Myc-tagged Δ36-201-REEP1 (Δ), Myc-tagged D56N-REEP1 (D56N), Myc-tagged W42R-REEP1 (W42R), or GFP (CRTL). Ectopic REEP1-Myc expression was confirmed using anti-Myc antibodies, and endogenous REEP1 expression (asterisk) was verified using anti-REEP1 antibodies. The upper band corresponds to exogenous Myc-tagged REEP1 and the lower band corresponds to endogenous REEP1. (F) Quantification of DRP1-S637/total DRP1 ratios for the different conditions. Each bar is the means± SEM of 6 independent quantifications of the ratio obtained from western blot, (*P < 0.05 mutant vs. WT, N.S; non-significant, Krustal-Wallis multiple comparison test). (G) Interaction between REEP1 and endogenous PGAM5 was revealed by immunoprecipitation with extracts of HEK cells transfected with Myc-tagged REEP1-WT or Myc-tagged REEP1-mutants, followed by western blotting for PGAM5 (n = 3).* corresponds to light chain of IgG.

Pathological REEP1 mutations alter mitochondrial morphology and distribution in neurons

REEP1 is highly expressed in the central nervous system and SPG31 is a neurodegenerative disease. Therefore, we decided to investigate the link between REEP1 and mitochondrial morphology in primary neurons. Mouse primary cortical neurons were co-transfected with the different REEP1 mutants and mitochondrial DSRed (Discosoma sp. red fluorescent protein; Fig. 5A). In neurons transfected with WT REEP1-Myc, we found that REEP1 was partially colocalized with the mitochondria, at precise sites (Fig. 5B), and the overall degree of colocalization (Pearson’s coefficient) was 0.37 ± 0.07 (Fig. 5C). The colocalization between REEP1 and the organelle increased when neurons expressed mutated REEP1 (Pearson’s coefficient between 0.59 ± 0.03 and 0.64 ± 0.03) (Fig. 5A−C). In these cells, the mitochondria were hyperfused and aggregated in the cell body. We found that 46 ± 3% of all mitochondria were in the cell body of neurons expressing W42R-REEP1, 47 ± 2% in neurons expressing D56N-REEP1, and 43 ± 3% in neurons expressing Δ-REEP1-Myc (Fig. 5D). In contrast, in neurons expressing WT-REEP1, 27 ± 2% of mitochondria were located in the soma. In parallel, mitochondrial distribution along the axon was strongly impaired and only 11 ± 2%, 9 ± 1%, and 11 ± 2% of the mitochondria were found beyond 200 µm from the soma of neurons expressing Δ-REEP1-Myc, W42R, and D56N, respectively. In cells expressing WT-REEP1, this fraction represented 26 ± 3% of all mitochondria (Fig. 5D). Importantly, these results indicate that REEP1 mutations lead to the sequestration of mitochondria in the soma of neurons and, therefore, hamper mitochondrial trafficking toward areas with high energy demands, such as neurites or synaptic junctions.

Figure 5

REEP1 regulates mitochondrial distribution in neurons. (A) Analysis of primary cortical mouse neurons expressing WT or mutant REEP1 (green) and mito-DSRed (red). The nucleus is stained using DAPI (blue). (B) Line scale showing REEP1 (green) colocalization with mitochondria (red). (C) Degree of colocalization between mitochondria, stained for TOM20, and Myc-REEP1, stained using anti-Myc antibodies. Each dot represents the Pearson's coefficient for a single cell (n = 3, ***P < 0.001, mutant vs. WT, Krustal-Wallis multiple comparison test). (D) Quantification of mitochondrial distribution along the neurons according to the expression of WT or mutant REEP1 (*P < 0.001, #P < 0.05; mutant vs. WT, n = 3).

Discussion

In this study, we have investigated the role of REEP1 in the maintenance of mitochondrial morphology. We provided evidence that mitochondrial morphology is altered in SPG31 patient cells due to a hyperphosphorylation of the fission protein DRP1 at serine 637. The alteration of mitochondrial shape is associated with the impairment of mitochondrial distribution in neurons.

Alteration of mitochondrial morphology is implicated in the progression of SPG31

HSPs arise gradually to progressive degenerations of the upper motor neurons. Genes causing these diseases are numerous and they are involved in various cellular functions (2,6). In several cases, the molecular mechanisms leading to neuronal degeneration remain to be dissected. This is also the case for SPG31 because it is not yet clear how REEP1 mutations fit in with degeneration of the motor neurons. In the present study, we have postulated that alterations of mitochondrial morphology previously observed in cells obtained from SPG31 patient (10) might be involved in the progression of this disease. Indeed, alterations of various mitochondrial functions, including energy metabolism, calcium buffering, or reactive oxygen species production, all of which depend on mitochondrial shape, are known to participate in or directly induce neurodegeneration (6,33,34). Furthermore, mitochondrial morphology is associated with its bidirectional transport within the cell and this transport is particularly important for neurons because of the elongated architecture of these cells (35). Altered mitochondrial morphology is responsible for axonal neuropathies, e.g. Charcot-Marie-Tooth type 2A or Dominant Optic Atrophy, which are caused by mutations in mitochondrial fusion proteins, MFN2 and OPA1, respectively. In contrast, only one mutation in genes related to the fission machinery has been reported to cause a disease (36). This mutation, in DNML1 encoding the mitochondrial fission protein DRP1, was linked to a severe metabolic and brain dysfunction (36). Fibroblasts from a patient bearing this mutation exhibited highly tubular mitochondrial phenotype, very similar to what we observed in SPG31 patient fibroblasts (36). As it has been discussed by Timmerman et al., HSP and Charcot-Marie-Tooth neuropathies have overlapping molecular basis (37) and our finding of REEP1-dependent regulation of mitochondrial morphology supports this notion.

As a key protein involved in mitochondrial morphology, DRP1 plays an important role in mitochondrial trafficking in highly polarized cells, such as neurons, and therefore, it contributes to neuronal health (38,39). Mitochondria were enlarged and aggregated in a primary neuronal culture derived from DRP1-KO mice embryos, suggesting that DRP1 deficiency is associated with the abnormal distribution of mitochondria within these cells (40). Accordingly, loss of DRP1 activity in post-mitotic neurons impaired mitochondrial energy metabolism and induced neuronal degeneration (41). In our study, we demonstrate that REEP1 mutations hamper DRP1-dependent fission. This leads to the clustering of mitochondria in the neuronal cell body combined with a decline of mitochondrial migration along the axon. These results are similar to previously reported observations by Li and colleagues that DRP1 dysfunction induces mitochondrial depletion in dendrites (38). Lack of mitochondrial transport promotes local mitochondrial energy crisis at different, specific points along the axons and leads to progressive motor neuron degeneration. In Drosophila melanogaster, the impairment of DRP1 function depleted mitochondria from the synapses at neuromuscular junctions, leading to energy crisis at these sites and alteration of neurotransmission induction during intense stimulation (42). Therefore, by inference, defects of mitochondrial migration observed in SPG31 cells, and, most probably, in other SPGs, might affect energy levels along the axons and at neuromuscular junctions, and impair neurotransmission in SPG patients. To conclude, we propose that REEP1-dependent mitochondrial morphology contributes to motor neuron health and, consequently, alteration of this function participates in motor neuron degeneration and the progression of SPG31.

REEP1 is involved in endomembrane trafficking through interaction with dynamin-like proteins

The biological role of REEP1 is yet to be delineated. Accumulating data suggest that REEP1 is an upstream actor of organelle morphology and trafficking. For instance, REEP1 was shown to regulate the expression of G-protein-coupled receptor on the cell surface by promoting vesicle trafficking (8). REEP1 contains a microtubule-binding domain and interacts with microtubules and microtubule-associated proteins, e.g. SPASTIN and ATLASTIN, which are also involved in HSP determinism (SPG4 and SPG3A, respectively) (13). This is in agreement with all of the previously reported REEP1-associated functions, namely vesicle trafficking, and ER and mitochondrial shaping and mobility that require interaction with the microtubules.

Converging lines of evidence support a functional interaction between REEP1 and members of dynamin-like protein family, including DRP1. Here, by using pharmacological and molecular biology approaches in different experimental systems, including primary cells from SPG31 patients and model cell lines, we have found that REEP1 dysfunction promotes DRP1 phosphorylation at serine 637. Furthermore, we discovered a physical interaction between REEP1 and mitochondrial phosphatase PGAM5. This interaction is reduced when REEP1 is mutated. PGAM5 promotes the dephosphorylation of DRP1-S637 (43–45), thus, regulating mitochondrial morphology (45). Interestingly, PGAM5 KO-mouse displayed a Parkinson’s-like movement phenotype and dopaminergic neurodegeneration as shown by Lu et al (46). Our results suggest that REEP1 sequesters PGAM5 to block DRP1 dephosphorylation. Thus, REEP1 mutations induced neuron degeneration through PGAM5-dependent function. Using PGAM5 KO mouse, Lu et al. show that this neurodegeneration involved PINK1 and mitochondrial autophagic degradation. REEP1 participates to the trafficking/shape of different organelles, its involvement to autophagic processes is strongly possible. This protein could play the role of cargo because it is associated µ-tubules or could deliver lipids for the formation of autophagosome because of its MAM localization (47). Such autophagic functions are currently discussed regarding HSP associated proteins, spastizin and spatacsin (48).

Dynamin-like proteins regulate membrane shapes and trafficking by interaction with the microtubules via conserved domains (49). ATLASTIN is a dynamin-like protein located in the ER and the ALSASTIN/REEP1 interaction regulates ER shape by establishing a physical link between the organelle and microtubules (13). On the other hand, the recruitment of DRP1 to the mitochondria occurs at the interface of mitochondria, ER, and microtubules (21), and mitochondrial fission involves the DRP1/microtubule interaction (50). Therefore, one hypothesis that might be worth exploring in the future would be that REEP1 regulates mitochondrial morphology by physically linking the microtubules and DRP1.

In support of the multi-organelle–associated function of REEP1, different organelle-interacting domains have been recently identified in this protein. In the present study, we have analysed REEP1 missense mutations (W42R, D56R) and a truncating mutation (p.V36Sfs*4) that alter its ER targeting domain (amino acids 1–115) (18), according to Lim et al. Interestingly, we found that these mutant proteins displayed a high degree of mitochondrial localization. Beetz et al. recently found a deletion in REEP1 that caused autosomal-dominant distal hereditary motor neuropathies, and this deletion was located in the mitochondrial domain. Such mutated REEP1 did not localize to the mitochondria (18). Other HSP-associated proteins were shown to alter both mitochondrial and ER morphologies and to disrupt the microtubules. For instance, mutations of SPASTIN induced clustering of mitochondria around the nucleus (26,51). This was also the case for Reticulon-2, a protein mutated in SPG12, which shares domain homology with REEP1. In the fruit fly, its loss led to alteration of both ER and mitochondrial morphology and trafficking, in a microtubule-dependent manner (52). In consequence, mitochondrial organization at neuro-muscular junction was disrupted (52).

Collectively, these findings suggest that REEP1 interacts with various intracellular organelles via its membrane domains, and orchestrates the shaping and mobility of these organelles through its interaction with the microtubules. However, it remains to be established how REEP1 regulates the organelle shape and whether REEP1–dependent regulation of organelle homeostasis involves protein phosphorylation mechanisms. For instance, there is no evidence that ATLASTIN or SPASTIN could be phosphorylated in a REEP1-dependent manner as we observed here for DRP1.

Our study provides evidence that mitochondrial morphology is altered in SPG31 patients because of hyperphosphorylation of the mitochondrial fission protein DRP1. It would be interesting to verify these findings in a larger group of subjects, using different cell and transgenic animal models, to investigate their universality. Crucially, we demonstrated that the observed mitochondrial morphology aberrations could be pharmacologically rescued in vitro, and it is therefore imperative that these observations are tested in vivo. This study comprises an important step forward for developing a cure for HSP and related neuropathies.

Materials and Methods

Chemicals

All chemicals were analytical grade. H89 was purchased from Cell Signalling (20 µM final concentration), and FCCP from Sigma Aldrich (10 µM final concentration).

General molecular biology protocols

Total DNA was extracted from cultured cells using the DNeasy Tissue Kit according to manufacturer’s recommendation (Qiagen). RT quantitative PCR (RT-qPCR) was performed as follows. Total RNA was extracted from cells and DNase I-treated to remove genomic DNA contamination (Qiagen RNeasy Plus Mini Kit), according to the manufacturer’s instructions. First strand cDNA synthesis was carried out in 20 μl volume using 1 µg total RNA, with 50 U M-MuLV Reverse Transcriptase (RNase H^–) and Oligo d(T)₁₆ (Applied Biosystems). The cDNA was stored at −20 °C until used in RT-qPCR analysis. RT-qPCR was performed using IQ^TMSYBR^® Green supermix (Biorad) and thermocycler CFX96 Touch Real-Time PCR system (Biorad). Primers were designed based on human sequences deposited in Genbank and are listed in the Supplementary Material, Table S1. All primers were screened by gel electrophoresis and melt curve analysis, to confirm amplification of a single cDNA fragment with a correct melting temperature and size. qPCR efficiencies for all the primer pairs were evaluated using serial dilutions for the same fibroblast cDNA sample. RT-qPCR expression of the target genes was calculated using the comparative ΔΔCT method and is presented as arbitrary units, normalized to endogenous references (geometric averaging of the two internal control genes RPLP0 and GUSB) with the CFX Manager Software.

DNA constructs

Full-length clones of REEP1 and mutant Δ36-201-REEP1 (called Δ) were obtained by RT-PCR of human fibroblast cDNA, and cloned in-frame with C-terminal Myc tag into the pRRLsin-PGK-MCS-WPRE vector. For the mutants REEP1^D56N and REEP1^W42R, QuikChange II XL Site-Directed Mutagenesis Kit (Agilent Technologies) was used to introduce each point mutation. REEP1 shRNA vectors (MISSION shRNA Plasmid DNA) were purchased from Sigma Aldrich. Drp1^S637A–YFP vector was a kind gift from Dr. Scorrano (University of Padova). Primer and clone sequences are provided in the Supplemental Material.

Patient cells, cell culture and transfection

Primary fibroblasts were obtained from arm skin biopsies of five heterozygous patients and four healthy, age- and gender-matched subjects. All patients were recruited through the European and Mediterranean network for spinocerebellar degenerations (SPATAX, http://spatax.wordpress.com). Written informed consent was obtained from all subjects to participate in genetic and functional studies, according to the Declaration of Helsinki. The study fulfilled our institutions’ ethical rules for human studies (RBM 01-29 and RBM 03-48). Primary fibroblasts and cell lines (HeLa and HEK) were cultured in DMEM medium containing 25 mM glucose and supplemented with 10% FBS, 1 mM sodium pyruvate, MEM non-essential amino acids, and 100 U/ml penicillin/streptomycin. The cells were transfected using Fugene HD (Roche), according to the manufacturer’s protocols.

Mice were used to establish primary hippocampal cultures. Mice (C57BL/6) were purchased from Janvier (France) and were housed with free access to food and water and a normal light/dark cycle. All experiments were conducted in compliance with all experimental procedures were approved by the Committee on Animal Health and Care of INSERM and French Ministry of Agriculture and Forestry. P0-P1 mouse brains were extracted directly after sacrifice in PBS containing 0.6% glucose and 0.5% BSA, and the hippocampi were dissected. Dissociation of the neurons was performed using a kit for postnatal neuron dissociation (Milteny Biotec) following the manufacturer's instructions. Cells were seeded onto 0.5 mg/ml poly-l-lysine-coated coverslips and maintained in neurobasal A medium (Gibco) containing 2 mM glutamine, 120 μg/ml penicillin, 200 μg/ml streptomycin, and B27 supplement (Invitrogen). Neuronal transfection was conducted over 4–5 days, using a standard calcium phosphate transfection protocol, and cultured for up to 7–9 days. All cells were cultured at 37 °C in the presence of 5% CO₂. Experiments were performed 48 h after transfection.

Immunoprecipitation, cell fractionation and protein crosslinking assays

For co-immunoprecipitation assays, cells were lysed in lysis buffer (1% Triton X-100, 50 mM Tris pH 7.4, 150 mM NaCl, 10 mM EDTA, and protease inhibitors (Sigma)]. Debris was removed by centrifugation for 20 min at 16,000 × g. Supernatants were incubated for 4 h with anti-Myc agarose beads at 4 °C. The beads were washed five times with Tris-buffered saline/Tween (0.05%). Proteins were eluted with 2× Laemmli sample buffer (Sigma).

The different steps of subcellular fractionation were performed at 4 °C. Cells were harvested in mitochondrial isolation buffer (10 mM Tris HCl, pH7.4, 210 mM mannitol, 70 mM sucrose, 1 mM EDTA) supplemented with protease inhibitor (Sigma), and homogenized by passing through a 26-gauge syringe (20 strokes). Aliquots were saved and called total cell lysate fraction (TCL) in order to analyse protein expression. The samples were centrifuged at 500 × g for 5 min at 4 °C. The supernatant was next centrifuged at 10,000 × g for 15 min at 4 °C to obtain the heavy membrane pellet enriched in mitochondria. The resulting supernatant was stored as the cytosolic fraction. The pellets were resuspended 1ml and submitted to another centrifugation cycle. Purified mitochondrial fraction was obtained by treating the mitochondria-enriched fraction with 1% digitonin in isolation buffer. Then, samples were centrifuged for 30 min at 16,000 × g and the obtained pellet retained as MAM-free mitochondria.

For protein crosslinking assays, cells were mechanically disrupted using syringes (26 gauge needles). The nuclei and unbroken cells were cleared from the extract by centrifugation (700 × g, 10 min). Cleared supernatants were incubated with 10 mM bismaleimidohexane (BMH, Pierce) or DMSO, for 40 min at RT. Crosslinking reactions were stopped by the addition of dithiothreitol (50 mM final concentration).

Western blotting

Samples were prepared in 2× Laemmli sample buffer (Sigma) supplemented with protease and phosphatase inhibitors (Sigma). 10–40 µg proteins were analysed by western blotting using conventional methods. Briefly, proteins were separated by electrophoresis (120 V, 1 h) and transferred onto polyvinylidene difluoride membranes using TransBlot Turbo (Biorad). Membranes were blocked with 5% milk in PBS-Tween (0.05%). Proteins were detected using specific antibodies against DRP1, CPT1A, UBQC2, TFAM (abcam); TOM20, CALREGULIN (Santa Cruz Biotechnology); MYC (Roche), DRP1-S637 (Cell Signaling), and REEP1 (Sigma-Aldrich, SAB2101977), diluted in 5% milk/PBS-Tween(0.05%). β-actin (Sigma-Aldrich) staining was used as the loading control. HRP-conjugated anti-rabbit or anti-mouse antibodies (Biorad) were used as secondary antibodies. HRP signal was visualized using chemiluminescent substrates (Thermofisher) and acquired with Chemidoc MP imaging system (Biorad).

Electron microscopy, immunofluorescence and live imaging

For electron microscopy, fibroblasts were fixed at 22 °C for 2 h with 2.5% glutaraldehyde in 0.2% phosphate buffer, pH 7.4, then post-fixed in 1% osmium tetroxide solution for 30 min at 4 °C, and embedded in EPON resin. Semi-thin sections (1 μm) were cut and stained with 1% toluidine blue. Ultrathin sections (50 nm) were cut and stained with uranyl acetate for examination under JEOL 1200 EX electron microscope (Peabody, MA) at 80 kV. Mitochondrial length and area were measured on several fields (magnification: 25000 and 60000) using ImageJ (NIH).

Immunofluorescence analyses were performed according to a classical protocol. Briefly, cells were grown on coverslips (Labtek) up to 70% confluence. They were then fixed in paraformaldehyde (4%, in PBS) and permeabilized using Triton-X100 (0.15%, in PBS). Coverslips were incubated with 10% BSA in PBS for 45 min and then with primary antibodies in blocking buffer for 1–2 h. Alexa-Fluor 488, Alexa-Fluor 568, and Alexa-Fluor 647 coupled antibodies were used to detect primary antibodies. Images were acquired using a microscope (Zeiss, AxioVision) with a 63× objective. Z-sections (at 0.2 μm intervals) covering the entire depth of the cell were acquired. For live imaging, cells were stained with MitoTracker Green or Red (Life Technologies) according to the manufacturer protocol. Images were recorded at 5% CO₂ and at 37 °C. Colocalization analyses were performed by measuring Pearson’s colocalization factors using Colocalization software from Zeiss. Pearson’s factors ranged from –1 to 1, where 1 represents the perfect colocalization. Mitochondrial distribution in neurons was analysed using a homemade plug-in in Fiji (NIH). For each image local maxima corresponding to mitochondria expressing mito-DS-Red were detected and (x,y) coordinates of the markers were recorded. The origin for distance calculation were set manually to the cell body. The distances to the cell body were pooled in Microsoft Excel. Intervals were defined by a range of 50nm, and mitochondrial distributions were classified by their distance to the soma.

Mitochondrial DNA/nuclear DNA ratio determination

These experiments were performed as described previously (53). Briefly, total DNA is extracted using Qiagen DNase tissue kit. Mitochondrial and nuclear DNA content are determined by qPCR (CFX 96 touch thermocycler, Biorad) using SYBRgreen. Used Primers are summarized in the Supplementary Material, Table S1.

Statistical analysis

All data are expressed as the mean ± SEM. Statistical analyses were performed with GraphPad Prism6 (GraphPad Software, Inc.). D'Agostino & Pearson omnibus normality test was used to assess the Gaussian distribution. When normality was verified, an unpaired Student’s t test was used to compare two independent groups and paired t-tests for sequential measurements. Non parametric test was used in other cases. One-way ANOVA followed by multiple comparison test were performed when comparing different groups to a control condition (e.g. mutants to WT).

Supplementary Material

Supplementary Material is available at HMG online.

Acknowledgements

We thank Prof. Lucca Scorrano for DRP1-S637-YFP.

Conflict of Interest statement. None declared.

Funding

Association contre les Maladies Mitochondriales (AMMI); the French association Strümpell-Lorrain (ASL); French association ‘Connaître les Syndromes Cérébelleux’ (CSC); Union Nationale des Aveugles et Déficients Visuels (UNADEV).

Author notes

^†These authors contributed equally to this work.