Spen modulates lipid droplet content in adult Drosophila glial cells and protects against paraquat toxicity

Glial cells are early sensors of neuronal injury and can store lipids in lipid droplets under oxidative stress conditions. Here, we investigated the functions of the RNA-binding protein, SPEN/SHARP, in the context of Parkinson’s disease (PD). Using a data-mining approach, we found that SPEN/SHARP is one of many astrocyte-expressed genes that are significantly differentially expressed in the substantia nigra of PD patients compared with control subjects. Interestingly, the differentially expressed genes are enriched in lipid metabolism-associated genes. In a Drosophila model of PD, we observed that flies carrying a loss-of-function allele of the ortholog split-ends (spen) or with glial cell-specific, but not neuronal-specific, spen knockdown were more sensitive to paraquat intoxication, indicating a protective role for Spen in glial cells. We also found that Spen is a positive regulator of Notch signaling in adult Drosophila glial cells. Moreover, Spen was required to limit abnormal accumulation of lipid droplets in glial cells in a manner independent of its regulation of Notch signaling. Taken together, our results demonstrate that Spen regulates lipid metabolism and storage in glial cells and contributes to glial cell-mediated neuroprotection.


Scientific Reports
| (2020) 10:20023 | https://doi.org/10.1038/s41598-020-76891-9 www.nature.com/scientificreports/ SPEN/SHARP can act as both a negative and positive regulator of the Notch signaling pathway, which plays a major role in cell fate specification in many species [24][25][26][27] . A recent study showed that SPEN acts as co-repressor of the transcription of genes responsive to Notch signaling 25 by binding to the recombinant binding protein J-kappa (RBP-Jκ) during Drosophila eye development 26 . Conversely, SPEN can act as a positive regulator of Notch signaling by recruiting the lysine methyl transferase 2D (KMT2D) co-activator complex to Notch target genes 27 . Spen has also been shown to promote Notch receptor activation by regulating trafficking of the Notch ligand Delta in intestinal stem cells of adult flies 28 .
Given that Spen promotes survival in glial cells of the developing Drosophila, we hypothesized that Spen expression in glial cells could confer neuroprotection in a model of PD. In the present study, we investigated the role of Spen as a potential regulator of lipid metabolism and Notch signaling in adult Drosophila glial cells, and determined the possible involvement of spen in the paraquat intoxication model of PD. SPEN has been reported to be expressed by human astrocytes 29 , but whether its expression is differentially regulated in the brains of patients with PD is unknown. To address this, we exploited the findings of a recent meta-analysis of microarray datasets obtained from the SN of PD patients compared with control subjects 30 . From the list of differentially expressed genes (data Supplement 1), we performed a tissue enrichment analysis using the TargetMine webtool 31 . This analysis identified 197 genes that were both upregulated in the SN of PD patients and expressed in astrocytes; one which was SPEN. SPEN is known to be expressed by both astrocytes and neurons in the normal human brain 29 (data Supplement 2); however, it is not clear whether the upregulation of SPEN in the brains of PD patients occurs in glia or neurons. We also identified 469 genes that were both downregulated in the SN of PD patients compared with control subjects and expressed in astrocytes (data Supplement 2). Interestingly, we detected significant enrichment of lipid metabolism-associated genes, as determined using the BioPlanet pathway enrichment tool 32 , in the gene set downregulated, but not upregulated, in the SN of PD patients (data Supplement S3, S4, Table 1). Taken together, these analyses indicate that SPEN expression is upregulated in astrocytes and/or neurons of PD patients compared with normal subjects, which prompted us to investigate its function in the Drosophila paraquat model of PD.

SPEN, the human ortholog of Drosophila spen, is upregulated in the substantia nigra pars compacta of PD patients.
Glia-specific overexpression of spen protects Drosophila from paraquat-induced neurotoxicity. We first investigated whether spen mRNA levels in the brain of adult Drosophila were altered under conditions of paraquat-induced toxicity. RT-qPCR revealed a significant upregulation of spen expression in the brains of paraquat-treated flies compared with control flies (Fig. 1A). To obtain insights on the potential function of Spen upregulation, we examined the survival of flies heterozygous for spen loss-of-function mutations 17 (spen k07612 /+ and spen 03350 /+) after paraquat intoxication. Both of the spen heterozygous mutant lines exhibited a higher sensitivity to paraquat compared with control flies (Fig. 1B, Supplemental Fig. S1), suggesting that Spen protects against paraquat-induced neurotoxicity. As previously reported 33 , spen expression, as revealed by P-lacW inserts, was detectable in both neurons and glial cells of Drosophila adult brain (Supplemental Fig. S2).
To determine whether the neuroprotective function of spen results from its expression in glial or neuronal cells, we generated flies in which spen was selectively knocked down in either cell type using a pan-glial (repo) or pan-neuronal (elav) driver. We found that downregulation of spen in glial cells, but not neuronal cells, increased the sensitivity of male flies to paraquat (Fig. 1C, Supplemental Fig. S3). Knockdown of spen in adult glia using an alternative genetic method, the temperature-sensitive TARGET system 45 , had the same effect of increasing Drosophila sensitivity to paraquat (Fig. 1D). Conversely, glia-specific overexpression of spen protected against paraquat toxicity (Fig. 1C). Collectively, these results show that Spen expression in glial cells protects Drosophila from paraquat-induced toxicity.

Spen regulates the Notch signaling pathway in adult Drosophila glial cells. Because Spen/SPEN
has been shown to positively or negatively regulate Notch signaling pathway, depending on the molecular context and cell type 25,27,28 , we next determined how Spen regulates Notch pathway in adult glial cells. To this end, we knocked down spen expression using an Eaat1 (excitatory amino acid transporter 1) driver, which is mainly expressed in astrocyte-like glial cells 34,35 , and monitored Notch pathway activation using a reporter transgene, Notch Responsive Element (NRE)-GFP. This transgene carries a minimal promoter containing Su(H)-DNA binding sites upstream of the EGFP coding sequence 36 (Fig. 2A). As previously reported, basal expression of www.nature.com/scientificreports/ NRE-GFP, reflecting basal activity of the Notch pathway, was observed in glial cells of control flies (Eaat1 > +), particularly in the antennal lobe 37 (Fig. 2B,C, Supplemental Fig. S4) and the dorsal part of the central brain.
Interestingly, however, basal activity of Notch signaling was strongly diminished in flies with glia-specific expression of spen RNAi (Fig. 2C,D). These data indicate that Spen is required for basal Notch activity in glial cells in the adult Drosophila brain. To extend these observations, we performed an epistasis experiment between the Notch intracellular domain (Notch intra ), an activator of the Notch pathway, and spen RNAi in Eaat1+ cells. As expected, Notch intra expression resulted in a strong and uniform induction of NRE-GFP reporter expression; however, spen knockdown significantly reduced the induction of NRE-GFP by Notch intra , indicating that Spen is necessary to maintain Su(H)-dependent Notch signaling in adult glia (Fig. 2D,E). The induction of NRE-GFP was not associated with a change in glial cell survival rate, as assessed by quantification of Repo-positive cells in control and spen knockdown flies (Fig. 2F). Collectively, these results show that Spen acts as a positive regulator of Notch signaling in adult Drosophila glial cells (Fig. 2G).  www.nature.com/scientificreports/

Spen regulates the number, size, and localization of lipid droplets in glial cells. Drosophila
glial cells have been shown to accumulate LDs under conditions of oxidative stress 6,7 , and Spen is known to control lipid metabolism in the fat body of Drosophila larvae 11 . We thus investigated whether Spen expression affects LD expansion and/or accumulation in Drosophila glial cells. To this end, we developed an ImageJ 39 macro to quantify fluorescence from BODIPY staining of LDs in serial confocal sections of adult Drosophila brain.
Quantification was based on an automated detection of BODIPY-positive particles that were distinguished from non-specific fluorescence by successive iterations. This new method allows not only discrimination of LDs from doublet foci on successive confocal stacks but also evaluation of the density, size, and circularity of LDs (see "Materials and methods" section for details). As visualized in the fluorescence micrographs (Fig. 3A) and quantified in the antennal lobe neuropil (Fig. 3B,C), spen knockdown in glial cells induced a significant increase in LD number and size. A similar accumulation of LDs was seen when spen RNAi was expressed using the pan-glial driver repo-GAL4, but not the pan-neuronal driver elav-GAL4 (Supplemental Fig. S5). The accumulation of LDs induced by glia-specific knockdown of spen was also detected when LDs were visualized by staining of wholemount brains for PLIN2 40,41 , a peridroplet protein anchored to the phospholipid monolayer surrounding LDs (Fig. 3D). In addition to the immunostaining approach, we expressed the UAS-PLIN1::GFP genetically encoded LD reporter 40,41 under the control of the Eaat1-GAL4 driver and confirmed that glia-specific spen knockdown induced accumulation of LDs in the brain of adult flies (Fig. 3E,F). Interestingly, inhibition of the canonical Notch pathway in glia by knockdown of Su(H) did not affect the number or size of LDs (Supplemental Fig. S6).
Collectively these results suggest that LD accumulation in flies with spen knockdown occurs specifically in glial cells and is not due to the inhibition of canonical Notch signaling.

Discussion
Several studies have reported that dysregulation of lipids, including LDs, is a component of the glial cell response to stress during neurodegeneration 6,[8][9][10] , including that associated with PD 42 . Here, we identified SPEN/SHARP as an astrocyte-expressed gene that is significantly overexpressed in the SN of PD patients compared with control subjects. In Drosophila, spen expression in glial cells mediates the resistance of Drosophila to paraquat treatment. We also show that Spen is a positive regulator of Notch signalling in adult glial cells. Finally, we found that gliaspecific expression of spen regulates LD accumulation in glial cells, in a manner independent of its regulation of Notch signaling. Collectively, our results suggest that the regulation of lipid metabolism by Spen contribute to the glia stress response. SPEN was one of many astrocyte-expressed genes found to be significantly differentially expressed in the brains of PD patients compared with control subjects. Interestingly, the differentially expressed genes were enriched in genes involved in lipid metabolism, which was of particular interest given the previously reported role of Spen as a regulator of lipid storage in adipocyte-like cells in Drosophila. Indeed, Spen was identified in two independent screens as a modulator of fat content in Drosophila larvae and adults 43,44 . A more recent study showed that spen manipulation in adipocyte-like cells correlated with alterations in the expression of key metabolic enzymes, supporting a role for Spen in energy catabolism 11 . In the present study, we found that glia-specific silencing of spen also affected lipid metabolism, as reflected by an increase in the number and size of LDs in Drosophila glial cells. While the mechanism underlying this effect is unclear, it is possible that spen downregulation may lead to a decrease in LD degradation or an increase in expression of triacylglycerol biosynthesis genes. Further investigations will be required to determine if Spen acts as a positive or negative regulator of lipid metabolism in different physiological and pathological contexts.  www.nature.com/scientificreports/ Spen has been shown to function as a negative regulator of Notch signaling at the morphogenetic furrow during eye development in Drosophila 26 . Here, we showed that Spen positively regulates Notch signaling in adult Drosophila glial cells. Spen-dependent activation of the Notch pathway has also been observed in intestinal stem cells of adult flies 28 , substantiating the role of Spen as a positive regulator of Notch in adult tissues. Thus, Spen appears to differentially regulate Notch signaling in cell type-, context-, and developmental stage-specific manners. The ability of Spen to differentially affect Notch signaling may be mediated via recruitment of intermediate positive or negative regulatory factors, resulting in activation or repression of Notch responsive-genes expression, respectively.
Our results show that the accumulation of LDs induced by spen knockdown is independent of Notch/Su(H) signaling, as reflected by the lack of effect of glia-specific Su(H) knockdown on LD size or number. Rather, Spen may directly affect mRNA stability or splicing of lipid metabolism genes to promote gene expression independently of Notch 11 . Finally, it is possible that the LD accumulation may be due to increased oxidative stress in spen-mutant flies, as previously observed in flies with defective mitochondrial respiration or after exposure to hypoxia 6,7,9 . Our results showing that spen-mutant flies exhibited increased sensitivity to paraquat toxicity are consistent with a mechanism involving oxidative stress.
Our findings are also in accord with a role for glia-expressed Spen in lipid metabolism in the context of PD pathophysiology. Using a data-mining approach, we found that genes differentially expressed in the SN of PD patients are not only enriched in astrocyte-expressed genes (e.g., SPEN/SHARP) but also include a significant number of genes annotated with the Gene Ontology terms "phospholipid metabolism", "lipid and lipoprotein metabolism", and "sphingolipid metabolism". These results point to a potential pathological role for lipid metabolism in PD, which is in accordance with a meta-analysis of genome-wide association studies of PD 46 . Among the lipid-related genes identified here to be differentially expressed in the SN of PD patients are several that could contribute to the regulation of LD formation and/or fate. For example, AGPAT4 (1-acylglycerol-3-phosphate O-acyltransferase 4) is involved in the synthesis of precursors of TAGs, the major lipid component of LDs. Similarly, SCD5 (stearoyl-CoA desaturase 5), a recently identified new target for PD treatment 47 , catalyzes free fatty acid desaturation and plays an important role in the early steps of LD formation. Finally, Arf79F and schlank, the Drosophila orthologs of two lipid-related genes found to be downregulated in the SN of PD patients, have been shown to control the homeostasis of LDs in Drosophila 48,49 .
Taken together, our data support a central role for spen expressed in glial cells in the control of lipid metabolism and resistance to paraquat-induced toxicity in Drosophila. Further studies on the function of Spen will contribute to our understanding of the involvement of lipid dyshomeostasis in neurodegeneration.  17 and were used here as heterozygotes. UAS-PLIN1::GFP 40 was obtained from RP Kuhnlein (University of Graz, Austria). The EP line spen-GS2279 (Kyoto DGRC Stock Center) was used to overexpress spen, and is referred to as UAS-spen. The UAS-spen RNAi line was a gift from KM Cadigan 50 and was previously characterized in studies of Drosophila retina development 14 . spen mutants and transgenic flies were outcrossed to a w 1118 control stock. We used the temperature-sensitive TARGET system 45 to restrict spen RNAi expression to adult glial cells. Briefly, flies carrying repo-GAL4, tubulin-GAL80 ts , and UAS-spen RNAi were raised at 18 °C to inhibit GAL4 activity and switched to 29 °C as adults to induce expression of spen RNAi .

RT-qPCR.
Total RNA was isolated from 25 to 35 Drosophila heads using RNeasy mini kits (Qiagen) and reverse transcribed with oligo(dT)15 primers and the ImProm-II Reverse Transcription System (Promega)  www.nature.com/scientificreports/ according to manufacturers' instructions. Quantitative PCR reactions were carried out on a StepOnePlus system (Applied Biosystems) using FastStart Universal SYBR Green Master (Roche Applied Science). Efficiency (E) of the primer sets was assessed using serial dilutions of cDNA preparations. Standard curves were generated to quantify mRNA abundance and PCR cycle numbers (Ct) for calculation of the relative mRNA expression level (Qr = E − Ct) 51 . Values were normalized to Rp49 mRNA levels. Primers for qPCR were: spen forward 5′-TTC GTT GTG GGA TAG CAG CA-3′ and reverse 5′-CGT TCG AAG CTG TTT GTC G-3′ and Rp49 forward 5′-ATC GTG AAG AAG CGC ACC AAG-3′ and reverse 5′-ACC AGG AAC TTC TTG AAT CCG-3′.
Immunostaining. Drosophila heads were removed and placed in a drop of fresh Hemolymph-Like 3 dissection buffer 52 (HL3) supplemented with d-glucose (120 mM). The proboscis was removed and the cuticle was opened to access the brain, and the brains were dissected and fixed overnight at 4 °C in 1% paraformaldehyde (PFA) diluted in HL3 medium. Fixed brains were washed 3 times for 10 min each in phosphate-buffered saline (PBS) containing 0.5% Triton X-100 and 5 mg/ml bovine serum albumin, and then incubated for 1 h in the same buffer containing 4% normal goat serum to prevent non-specific antibody binding (blocking solution). The brains were then incubated overnight at 4 °C with the following primary antibodies diluted in blocking solution: For experiments with the NRE-GFP-expressing Drosophila line, the heads were removed and placed in a drop of fresh PBS. The brains were dissected and fixed in 4% PFA/PBS for 20 min at room temperature. The brains were then washed in PBS and mounted directly in Vectashield medium (AbCys). As controls for the NRE-GFP experiments, brains from Drosophila not expressing GFP were also processed to evaluate background fluorescence.
Image processing. Images of whole-mount brains were acquired at the LYMIC-PLATIM Imaging and Microscopy Core Facility of SFR Biosciences (UMS3444, ENS de Lyon, France). For PLIN1::GFP, Repo, and BODIPY fluorescence, images were acquired on a Zeiss LSM800 confocal microscope and analyzed with the ImageJ 39 software (see section below). For NRE-GFP fluorescence, images were acquired on a Leica epifluorescence microscope. In each experiment, the mean GFP fluorescence of control brains (Eaat1-GAL4/+) was used to normalize the results. The NRE reporter is a synthetic construct with three copies of SPS sites taken from the E(spl) regulatory region 36 . SPS (Su(H) paired sites) are binding sites for the Notch activity-dependent transcription factor Su(H).

Automated image analysis.
We developed an ImageJ macro (https ://gitbi o.ens-lyon.fr/dclue t/Lipid _Dropl ets) to identify fluorescent particles on confocal stacks using Drosophila brains stained with the lipidbinding dye BODIPY 493/503 (Molecular Probes, D-3922) or labeled with the glial cell-specific marker Repo. The program requires ImageJ 39 v1.49 g or higher and is based on an iterative detection of the brightest particles followed by removal of "doublets" of the same particle over the stack. Briefly, the program first delineates the brain region of interest and then identifies particles within that region along the stack. The signal is intensified using the Gaussian blur function and the maximum entropy treatment. The "Max-Entropy" threshold method 53 is then applied to detect the particles of interest. The detected particles are stored in a transient matrix and removed from the image. The next iteration is able to detect less bright particles. Finally, the program removes all doublets of the same particle along the z-dimension of the stack (keeping the largest candidate as the best) to enable optimal counting of labeled particles. The program can calculate multiple parameters, such as particle density, size, and circularity.
Paraquat-induced PD model. Paraquat medium was prepared fresh shortly before each experiment.
Paraquat (Sigma, 36541) was added at 10 or 20 mM to PBS (as indicated in the corresponding figure legends) containing 0.8% low-melting agarose (Sigma, A9414) and 10% sucrose (Sigma, S0389). Three-day-old male flies were fasted for 4 h on 0.8% agarose medium and then transferred to 10 or 20 mM paraquat-containing medium for 5-7 days for survival experiments or the indicated times for RT-qPCR analysis. At least three independent experiments were performed, each with n ≥ 20 flies per condition per experiment.
Lipid droplet staining. Heads were removed from 6-day-old flies and placed in a drop of fresh HL3 dissection buffer supplemented with d-glucose (120 mM). Brains were dissected and fixed overnight at 4 °C in 1% PFA/HL3 medium. Fixed brains were washed 3 times for 10 min each in PBS/01.% Triton X-100 and then incubated overnight at 4 °C with 15 mg/ml BODIPY 493/503 (Molecular Probes, D-3922) diluted in the same buffer. The brains were washed 3 additional times with PBS/0.1% Triton X-100, mounted in Vectashield medium (AbCys) on a bridge slide, and stored at − 20 °C until imaged.
Statistical analysis. Data are presented as the means ± standard deviation from three independent experiments unless noted. Statistical analyses were performed using R (R Core Team) and Prism software (GraphPad, San Diego, CA). The statistical tests applied are given in the figure legends.