The function of Scox in glial cells is essential for locomotive ability in Drosophila

Synthesis of cytochrome c oxidase (Scox) is a Drosophila homolog of human SCO2 encoding a metallochaperone that transports copper to cytochrome c, and is an essential protein for the assembly of cytochrome c oxidase in the mitochondrial respiratory chain complex. SCO2 is highly conserved in a wide variety of species across prokaryotes and eukaryotes, and mutations in SCO2 are known to cause mitochondrial diseases such as fatal infantile cardioencephalomyopathy, Leigh syndrome, and Charcot-Marie-Tooth disease, a neurodegenerative disorder. These diseases have a common symptom of locomotive dysfunction. However, the mechanisms of their pathogenesis remain unknown, and no fundamental medications or therapies have been established for these diseases. In this study, we demonstrated that the glial cell-specific knockdown of Scox perturbs the mitochondrial morphology and function, and locomotive behavior in Drosophila. In addition, the morphology and function of synapses were impaired in the glial cell-specific Scox knockdown. Furthermore, Scox knockdown in ensheathing glia, one type of glial cell in Drosophila, resulted in larval and adult locomotive dysfunction. This study suggests that the impairment of Scox in glial cells in the Drosophila CNS mimics the pathological phenotypes observed by mutations in the SCO2 gene in humans.


Results
Glial cell-specific Scox knockdown reduces locomotive ability. To investigate the effects of a reduction of Scox expression in the nervous system of Drosophila, we knocked down Scox in neuronal cells using two different GAL4 drivers, neuronal Synaptobrevin (nSyb)-GAL4 and embryonic lethal abnormal vision (elav)-GAL4. The neuron-specific knockdown of Scox induced no detectable defects in the locomotive ability of larvae or adult flies ( Supplementary Fig. S1). Therefore, we next knocked down Scox in glial cells, which is the other cell type in the nervous system, under the control of the pan-glial GAL4 driver reverse polarity (repo)-GAL4. Although the number of glial cells in humans is estimated to be more than ten-times that of neurons, in Drosophila, the number is less than one-tenth 10 . However, glial cell-specific Scox knockdown reduced the locomotive ability of both larvae and adult flies ( Fig. 1A-C). In particular, locomotion in adult flies progressively deteriorated each day after eclosion (Fig. 1D). The deficient locomotive behaviors were observed in two different Scox knockdown fly strains whose target sequences do not overlap, excluding possible off-target effects. In order to evaluate the knockdown efficiency of Scox-IR lines, Western blot analysis was carried out (Fig. 1E,F, Supplementary Fig. S2). As glial cells account for only about 10% of cells in the Drosophila CNS, it was difficult to evaluate the knockdown efficiency using extracts from the CNS. We therefore extracted proteins from whole larvae in which Scox was knocked down by the Act5C-GAL4 driver, ubiquitously expressing GAL4 in somatic cells. The expression levels of SCOX significantly decreased in both knockdown fly lines, as previously reported 11 (Fig. 1E,F).

Scox is essential for mitochondrial morphology and function in glial cells in the CNS. The Scox
gene is the Drosophila homolog of human SCO2 that encodes a protein required for the assembly of cytochrome c oxidase on the mitochondrial inner membrane. In mice and humans, a reduction in SCO2 leads to mitochondrial dysfunction 9,12 . Therefore, we examined whether knockdown of Scox affects the morphology and func- ability of adult flies was tested in a climbing assay. Pan-glial cell-specific Scox knockdown reduced climbing scores compared with the control. Climbing assays were performed on days 3, 7, and 14 after eclosion, and climbing scores were calculated on each day. n = 100. *p < 0.05, ***p < 0.0001. (E) The expression level of SCOX was analyzed by Western blotting. Protein extracts from whole larvae were used. The expression levels of SCOX significantly decreased in both knockdown fly lines, as previously reported 11 . (F) Quantification of the expression levels of SCOX compared with α-tubulin. α-tubulin was used as an internal control and the relative intensities of SCOX bands normalized using α-tubulin are shown by each bar. In the SCOX knockdown larvae, expression levels of SCOX decreased compared with the control. n = 4. repo > GFP-IR (UAS-GFP-IR/+; repo-GAL4/+), repo > SCOXIR 20 (UAS-SCOXIR 20 /+; repo-GAL4/+), repo > ScoxIR 7861 (repo-GAL4/UAS-ScoxIR 7861 ), Act5C > w-IR (Act5C-GAL4/+; UAS-w-IR/+), Act5C > SCOXIR 20 (Act5C-GAL4/UAS-SCOXIR 20 ), and Act5C > ScoxIR 7861 (Act5C-GAL4/+; UAS-SCOXIR 7861 /+). www.nature.com/scientificreports/ tion of mitochondria. Dysfunction of mitochondria-associated proteins causes fragmentation or hyperfusion of mitochondria, and negatively affects their function 13,14 . Thus, we examined the mitochondrial morphology in glial cells of the CNS by marking them with a mitochondrial marker, mito-GFP reporter. As a result, fibrous mitochondria were observed in the control, whereas the mitochondria in Scox knockdown flies were rounded and fragmented ( Fig. 2A). We further inspected the mitochondrial morphology in sections at 10 μm from the surface in brain lobes of the Scox knockdown flies using a transmission electron microscope. The size of mitochondria in this region in Scox knockdown flies was 1.75-times larger than that in the control (Fig. 2B,C). As morphological abnormalities of mitochondria suggest impaired mitochondrial function, a decreased ATP level and increased apoptosis may have been induced in the Scox knockdown flies. The mitochondrial respiratory chain complex plays an essential role in ATP production in most eukaryotic cells. Therefore, we measured ATP levels in the Scox knockdown flies with Act5C-GAL4. Quantification of the ATP level extracted from the whole larvae revealed a reduction to 4-11% in Scox knockdown flies (Fig. 3A). Moreover, when Scox was knocked down by Act5C-GAL4, it was pupal lethal (data not shown).
As mitochondria become dysfunctional, they release cytochrome c, which binds to apoptosis protease activating factor 1 (APAF-1) and other proteins to form apoptosomes, thereby activating caspase-9 and caspase-3, and inducing apoptosis 15 . The observed mitochondrial morphological changes and marked decrease in ATP levels in the Scox knockdown flies suggested the induction of apoptosis. We therefore examined the expression level of active caspase-3 in the CNS of Scox knockdown flies by immunostaining. The number of active caspase-3

Scox knockdown in glial cells affects synaptic morphogenesis and function in the NMJ.
The defective locomotive abilities exhibited by Scox knockdown flies may be due to functional abnormalities of motor neurons. Therefore, we examined the morphology and function of neural synaptic terminals at the neuromuscular junctions (NMJs) in larvae with glial cell-specific knockdown of Scox (Fig. 4). We measured the total length of synaptic branches and the number of boutons in the NMJs on the fourth muscle in larvae with glial cell-specific knockdown of Scox. We quantified the length of the synaptic branches of NMJs. The length of each synaptic branch was shortened in the Scox knockdown flies (Fig. 4A,B). Moreover, the number of boutons was significantly reduced compared with the control (p < 0.05) (Fig. 4A,C). On the other hand, neuron-specific knockdown of Scox induced a defect in bouton number, but not in synaptic branch length ( Supplementary Fig. S3).
The presynaptic terminal has a special region called the active zone, where synaptic vesicles fuse to release neurotransmitters into the synaptic cleft and transmit signals to the next synapse or muscle. Therefore, we performed immunostaining with anti-Bruchpilot (Brp) IgG, a marker of the active zone, to investigate the density and size of the active zone in the NMJs of glial cell-specific Scox knockdown flies (Fig. 4D). As a result, the density of the active zone relative to the area of the synapse was reduced in the Scox knockdown flies (Fig. 4D,E) and the size of each active zone was small, suggesting reduced synaptic function in the Scox knockdown flies (Fig. 4F).
Glial cells have many functions, such as protecting and supplying nutrients to neurons, and their abnormalities may lead to the demyelination of neurons and defects in signal transmission [16][17][18] . Therefore, knockdown of Scox may result in deficits in mitochondrial function, which consequently induce defects in glial cell functions, such as protection and homeostasis of neurons, resulting in abnormal morphology and neuronal dysfunction.

Knockdown of Scox in ensheathing glia is responsible for the locomotion deficit. Mammals
have four types of glial cells, whereas Drosophila has five types of glial cells in the CNS that function similarly to mammalian glial cells, and each of the five types of glial cells has one of the functions corresponding to the four types of mammalian glial cells or the blood-brain barrier [19][20][21] . repo-GAL4 is expressed in all glial cells in Drosophila and we noted decreased locomotive ability in flies in which Scox was knocked down by repo-GAL4  www.nature.com/scientificreports/ ( Fig. 1). We investigated which of the five types of glial cells was responsible for the decreased locomotive ability in larvae and adult flies. NP2222-GAL4, NP3233-GAL4, NP6293-GAL4, NP6520-GAL4, and Moody-GAL4 specifically express GAL in cortex glia, astrocyte-like glia, perineurial glia, ensheathing glia, and subperineurial glia, respectively [22][23][24][25] . Among the five types of glial cells, larval crawling speed was reduced only when Scox was specifically knocked down in the ensheathing glia, whereas knockdown in the other types of glial cells caused no difference (Fig. 5A,B, Supplementary Fig. S4). Moreover, adult flies with ensheathing glial cell-specific Scox knockdown exhibited a decrease in locomotion (Fig. 5C). As in the case of knockdown of Scox by repo-GAL4, the locomotion of adult flies progressively deteriorated with each passing day.
In some neurodegenerative diseases, demyelination of the myelin sheaths of Schwann cells and oligodendrocytes is the cause of disease onset 26,27 . As ensheathing glia most closely match oligodendrocytes in mammals, even though myelin is not formed in the Drosophila CNS 10 , dysfunction of Scox may lead to defects in locomotion because of impairment of the protection of nerve axons by glial cells.

Discussion
The Scox gene is ubiquitously expressed throughout development and functions in the mitochondrial respiratory chain reaction to produce ATPs 11,28 . The Scox knockdown flies by Act5C-GAL4, which is a ubiquitous GAL4 driver, exhibited larval locomotive defects (data not shown) and died before hatching from pupae, with a significant decrease in ATP levels. This is consistent with the previous report that the homozygote carrying null mutations in the Scox gene exhibiting a reduction of cytochrome c oxidase (COX) activity to 21% of wild-type, had reduced locomotive ability and died at the beginning of the second larval instar 28 . This suggests that Scox plays essential roles in development. However, the knockdown of Scox in neurons by elav-GAL4 or nSyb-GAL4 did not affect the locomotive ability. In Drosophila, another gene, Surf 1, a homolog of human Surf1, is involved in the COX assembly 29 . Similar to the knockdown of Scox, the knockdown of Surf 1 by Act5C-GAL4 caused locomotive defects and death at the larval stage, whereas knockdown by elav-GAL4 caused no defects in locomotive ability or viability 30 . In addition, homozygous mutants of the several cytochrome c oxidase subunits, such as tenured, levy, and cyclope, which are homologs of COX V, VIa, and VIc, respectively, resulted in a developmental lethal phenotype [30][31][32] . Thus, the lethality during development based on the perturbation of the COX-related genes, at On the other hand, knockdown of Scox in glial cells caused mitochondrial dysfunction, which led to morphological and functional abnormalities at the NMJ, and resulted in larval and adult locomotive disability. We observed swollen mitochondria in glial cells of the Scox knockdown flies (Fig. 2). Defective oxidative phosphorylation led to a decrease in ATP level, causing mitochondrial dysfunction 33 , and toxic substances that inhibit oxidative phosphorylation and reduce the ATP level were reported to cause mitochondrial swelling 34 . Thus it is consistent with our study because the knockdown of Scox, which is an essential protein for the assembly of cytochrome c oxidase in the mitochondrial respiratory chain complex, reduced the ATP level and mitochondrial swelling led to mitochondrial dysfunction. Spinocerebellar ataxia type 1 (SCA1) is an autosomal dominant neurodegenerative disorder, a poly-glutamine disease, caused by the expression of an aberrant form of SCA1. Similar to our study, flies overexpressing the mutant form of SCA3, the Drosophila homolog of SCA1, in glial cells, but not in neurons, exhibited locomotive dysfunction 21,35 . Therefore, some of the genes identified as neurological disorder-causing genes function in glial cells, but not in neurons. In Drosophila, there are six types of glial cells: perineurial glia, subperineurial glia, cortex glia, astrocyte-like glia, ensheathing glia, and wrapping glia 10,19 . To clarify which type of glial cell Scox has functions, we knocked down of Scox using each glial cell-specific GAL4 driver. As a result, only ensheathing glia-specific knockdown of Scox affected both larval and adult locomotive ability. Ensheathing glia engulf neuropils, including axons, dendrites, and synapses, and regulate larval locomotion [36][37][38] . In mammals, axons are wrapped by oligodendrocytes and Schwan cells. Although myelination does not occur in Drosophila, ensheathing glia are suggested to closely match mammalian oligodendrocytes based on the function of wrapping axons 19 . As mutations in SCO2, a mammalian homolog of Scox, cause CMT type 4, which is a group of demyelinating peripheral neuropathy, perturbation of SCO2 in glial cells may play an essential role in the pathogenesis of CMT. Glycolysis affects the maintenance of post-myelinated oligodendrocytes more than mitochondrial respiration, and demyelination disorders resulting from mitochondrial perturbation are linked to increased ROS and aberrant lipid metabolism, but not ATP production [39][40][41] ; therefore, further analysis as to whether our Drosophila model is useful for CMT type 4 is required. Most Drosophila models established thus far for CMT are designed for the axonal type CMT because Drosophila lacks myelin sheaths and Schwan cells 42 . However, the present study suggests that the development of models mimicking some demyelinating type CMT is also possible in Drosophila.
In this study, we suggested that knockdown of Scox, the Drosophila homolog of human SCO2, which is a causal gene for CMT, in glial cells leads to locomotive disability with neuronal morphological and functional defects resulting from mitochondrial impairments. Until now, due to the lack of myelination, the importance of assessing the mechanism of demyelinating disorders in Drosophila has been overlooked. However, as a commonality between mammalian and Drosophila glia linked to ensheathing axons, such as the regulation of axon ensheathment by peripheral glia regulated by EGFR signaling [43][44][45] , was recently found, analyses of the glial roles in neurological dysfunction in Drosophila may increase. Crawling assay. Crawling assays were adapted from the previously reported method with slight modification 46 . Male third instar larvae were washed in phosphate-buffered saline (PBS) to remove food and then placed on an agar plate containing 1.7% agar. After larvae started moving, videos were taken with a digital camera for 1 min. Then, recorded videos were converted to AVI files using a MOV to AVI converter (MPEG-Streamclip), and analyzed by ImageJ (NIH, USA) with the wrMTrck plugin (developed by Dr. Jesper Søndergaard Pedersen) to track the movement of larvae and trace the path of movement.

Methods
Climbing assay. Climbing assays were carried out in the same manner as described previously 47 . Flies were bred at 28 °C to promote the expression of GAL4 and 20 newly eclosed adult male flies were collected for experiments. Flies were put into a glass tube and tubes were tapped to collect flies at the bottom. Flies were allowed to climb the wall for 30 s and videos were filmed of them moving. This process was repeated a total of 6 times. To calculate climbing scores, the height each fly climbed was scored from 0 to 5 points at 2-cm intervals from the bottom. For each fly's climbing score, means and standard errors were calculated, and climbing assays were performed 3, 7, and 14 days after eclosion.

Protein isolation and Western blot analysis.
To detect SCOX, proteins were extracted from 5 larvae.

ATP measurement.
A whole larva was homogenized in 150 µl of ATP assay buffer (ab83355, Abcam), centrifuged, and the supernatant was collected. Then, 10 µl of chilled trichloroacetic acid (ab204708, Abcam) was mixed with supernatants and samples were reacted at 4 °C for 15 min. After centrifugation, 7.5 µl of neutralization solution (ab204708, Abcam) was applied to samples and samples were reacted at 4 °C for 5 min. Fifty microliters of each sample was mixed with 50 µl of Celltiter-Glo (G7570, Promega) and reacted at 25 °C for 10 min. Luminescence was detected using Lumat LB 9507 (Berthold Technologies).
Electron microscopy. Male  Polymers were sliced into 70-nm ultra-thin sections with a diamond knife using an ultramicrotome (Ultracut UCT, Leica) and samples were mounted on copper grids. They were then stained with 2% uranyl acetate at 25 °C for 15 min, washed with distilled water, and stained with lead staining solution (Sigma-Aldrich) at 25 °C for 3 min. A transmission electron microscope (JEM-1400Plus, JEOL) was used to examine grids at an acceleration voltage of 100 kV and digital images (3296 × 2472 pixels) were taken with a CCD camera (EM-14830RUBY2, JEOL).

Data analysis.
In Western blot analysis, ATP measurement, mitochondrial size measurement, crawling assays, and immunostaining of NMJs (synapse and active zone) and active caspase-3, p-values were calculated using the unpaired two-tailed Welch's t-test. In climbing assays, p-values were also calculated by two-way ANOVA using GraphPad Prism version 7. All graphs present the mean ± SEM.