The synaptic vesicle protein Mover/TPRG1L is associated with lipid droplets in astrocytes

Crucial brain functions such as neurotransmission, myelination, and signaling pose a high demand for lipids. Lipid dysregulation is associated with neuroinflammation and neurodegeneration. Astrocytes protect neurons from lipid induced damage by accumulating and metabolizing toxic lipids in organelles called lipid droplets (LDs). LDs have long been considered as lipid storage compartments in adipocytes, but less is known about their biogenesis and composition in the brain. In particular, proteins covering the LD surface are not yet fully identified. Here, we report that the presynaptic protein Mover/TPRG1L, which regulates the probability of neurotransmitter release in neurons, is a component of the LD coat in astrocytes. Using conventional and super‐resolution microscopy, we demonstrate that Mover surrounds naive and oleic acid induced astrocytic LDs. We confirm the identity of astrocytic LDs using the neutral lipid stains Bodipy and LipidTox, as well as immunofluorescence for perilipin‐2, a known component of the LD coat. In astrocytes, recombinant Mover was sufficient to induce an accumulation of LDs. Furthermore, we identified point mutations that abolish targeting to LDs and show similarities in the required binding sequences for association to the presynapse and LDs. Our results show that Mover is not only a presynaptic protein but also a candidate for LD regulation. This highlights the dual role of Mover in synaptic transmission and regulation of astrocytic LDs, which may be particularly important in the context of lipid‐related neurological disorders.


| INTRODUCTION
Mover (Mossy fiber associated vertebrate-specific protein; also called TPRG1L and SVAP30) is a non-transmembrane protein consisting of 266 amino acids expressed in various organs such as the heart, liver, and testis, with the highest expression being detected in the brain (Burré et al., 2006;Kremer et al., 2007).Mover is heterogeneously distributed in the mouse brain, with high abundance in the hippocampus, the amygdala, and the ventral pallidum (Wallrafen & Dresbach, 2018).A proteomic screen of the anterior cingulate cortex, showed a 2.4-fold increase in expression of Mover (listed as novel protein RP11-46F15.3) in schizophrenic patients compared to healthy individuals (Clark et al., 2006).In neurons, Mover is located to the presynaptic terminal and associated with synaptic vesicles (Ahmed et al., 2013).Mover knockout mice show altered synaptic transmission at the calyx of Held and hippocampal mossy fiber terminals (Pofantis et al., 2021;Viotti & Dresbach, 2019), implying that Mover modulates synaptic activity.
Synaptic transmission does not depend on neurons alone.Astrocytes contribute to synapse formation and synaptic plasticity (Sancho et al., 2021), in part by forming a tripartite synapse, where astrocyte processes associate with a presynaptic terminal and its postsynaptic counterpart (Araque et al., 1999).Moreover, astrocytes support neurons with metabolites, and recent studies demonstrate that they play a crucial role in various processes such as inflammation, elimination of waste products and neutralizing reactive oxygen species (ROS) (Giovannoni & Quintana, 2020;Islam et al., 2019).
Astrocyte-neuron coupling is not limited to synaptic terminals.Cytotoxic lipid species that could cause damage to neuronal membranes and mitochondria are released from neurons via the apolipoprotein E (APOE) carrier and accumulate in astrocytes, where they can be stored in specialized lipid storage compartments called lipid droplets (LD) (Ioannou et al., 2019).
LDs are lipid storage organelles found in nearly all species (Murphy, 2012).They are highly abundant in adipose tissue, but are present in various other cell types as well.Research of the past decade has shed light on novel protective functions and implications for LDs in diseases of the nervous system (Welte, 2015).For example, in the Drosophila larval brain they protect a stem cell niche from hypoxia derived ROS damage (Bailey et al., 2015).Glia specific mitochondrial dysfunction in the adult Drosophila resulted in an accumulation of LDs and a progressive neurodegeneration (Cabirol-Pol et al., 2018), implicating LDs also in pathologic conditions.Ioannou et al. (2019) demonstrated that the accumulation of astrocytic LDs serves a neuroprotective function by sequestering toxic fatty acids derived from neurons.LDs metabolize lipids and form dynamic contacts with other cellular organelles for exchange of substrates (Valm et al., 2017).
LDs consist of a core, composed of triacylglycerols and sterol esters, that is surrounded by a phospholipid monolayer (Olzmann & Pedro Carvalho, 2019).The monolayer surface is covered with proteins and enzymes that are necessary for trafficking, organelle interaction, lipid conversion, and regulation of LDs (Zehmer et al., 2009).
Perilipins are a key group of LD proteins consisting of five isoforms, which regulate distinct aspects of LD generation and stabilization (Itabe et al., 2017).The protein composition of the LD surface is thought to contribute to the diversity of LD functions.
Here, we report that Mover is a novel component of the LD surface in astrocytes.Co-localization with LD markers demonstrated that endogenous Mover is indeed found surrounding oleic acid (OA) induced astrocytic LDs, and recombinant Mover retains the ability to target LDs.STED microscopy revealed a heterogeneous distribution around the LD surface with the formation of local clusters.

Overexpression of Mover in astrocytes induced an accumulation of
LDs independent of OA induction.Furthermore, using recombinant Mover constructs harboring point mutations, we identified regions of Mover necessary for LD binding.These results highlight that Mover not only modulates synaptic transmission at presynaptic terminals, but is also involved in LD regulation in astrocytes.
Cortical astrocytes from rat were plated at a density of 2.5 Â 10 4 to 5 Â 10 4 cells per well of 24-well plates in Minimum Essential Media (Gibco, MEM) supplemented with 10% horse serum (Gibco), 4.65 g/L Glucose (Sigma), and 2 mM L-Glutamine on glass coverslips coated with 0.04% PEI.The astrocyte culture medium was changed twice a week.For astrocyte cultures, the microglial ratio was assessed using IBA1 and GFAP stainings and was in three independent sets below 4% (3.39%, 1.28%, and 1.88%; see also Supplement data 1).Cortical astrocytes from WT or KO mice were plated at a density of 2 million cells per 75 cm 2 flasks in MEM supplemented with 10% horse serum, 4.65 g/L Glucose and 2 mM L-Glutamine.The astrocyte culture medium was changed twice a week.After 2 weeks, cells were harvested, frozen in astrocyte culture medium supplemented with 10% DMSO and stored into liquid nitrogen.Astrocytes from WT and KO mice were thawed together and plated at a density of 5 Â 10 4 cells per well in astrocyte culture medium on glass coverslips coated with 0.04% PEI.The astrocyte culture medium was changed twice a week.
After 6 days, astrocytes from WT and KO Mover mice were treated for 24 h with 400 μM of OA.
Cells were passaged every 2 or 3 days.For immunofluorescence, cells were counted and plated at a density of 1-2 Â 10 4 cells per well in a 24-well plate on 0.04% PEI coated glass coverslips.

| Transfection
HEK cells were transfected with several Mover constructs 1 day after plating using PEI transfection in Opti-MEM medium (Gibco).Per coverslip, 1 μg DNA was mixed with 1% PEI in a 1:3 (v:v) ratio and incubated for 30 min at room temperature and carefully applied to each coverslip.Cortical astrocyte cultures were transfected at day in vitro 13 using lipofectamine 2000 (Invitrogen) transfection following manufacturer's instructions.The medium of the cells was changed with fresh pre-warmed medium before the transfection to ensure an exact volume per well.0.8 μg DNA diluted in transfection buffer were incubated per coverslip.The medium was changed after 4 h of incubation.

| Oleic acid treatment
OA treatment was used to increase the number and size of LDs.OA complexes with bovine serum albumin (Sigma, BSA) were prepared with an adapted protocol from Listenberger et al. (2016).Briefly, a $4.6 mM solution of OA-BSA complexes was obtained by adding dropwise 20 mM sodium oleate in a prewarmed 5% solution of BSA in DPBS.The solution was sterile filtered and stored at À20 C. The solution of OA-BSA complexes will be referred to as "oleic acid" or "OA" in the following.Cells were treated with 400 μM OA in culture medium for 11-24 h.

| Acute brain slices
Acute brain slices were prepared from C57BL6/J mice at 11-15 weeks of age following a protocol modified from Mathis et al. (2011)

| Statistic
The number and size of ring-shaped Mover structures were quantified using the multi point tool from FIJI (Schindelin et al., 2012).Rings 3 | RESULTS

| Mover forms ring-shaped structures in astrocytes
In immunostained primary hippocampal co-cultures, containing neurons and astrocytes, Mover co-localizes with the presynaptic marker protein synaptophysin (Ahmed et al., 2013;Kremer et al., 2007)   Mover is not only a presynaptic protein but is also associated with organelles in astrocytes.
To verify that this observation reflects an association of Mover with astrocytes in brain tissue, we examined mouse hippocampal brain slices by confocal microscopy.Immunostaining for Mover and GFAP revealed a clear association of Mover puncta with astrocytes (Figure 2).
Mover immunosignals were found in astrocytic processes or in the soma (Figure 2a).Tracking the signals across sequential z-planes confirmed that Mover was located inside astrocytes (Figure 2b).This result demonstrates that Mover is found in astrocytes also in situ.
Because of the characteristic shape of the Mover signal in cultures, we suspected that these ring-shaped structures might represent LDs.The number and size of LDs can be increased by treating cultured astrocytes with excess fatty acids (Nakajima et al., 2019).We therefore went back to in-vitro cultures and applied OA to hippocampal co-cultures to see if this affects the ring-shaped Mover structures we observed in astrocytes in a similar manner.
In untreated cultures, astrocytic ring-shaped Mover structures appeared small and low in numbers (Figure 3a-d), whereas application of OA rapidly increased their number and size (Figure 3e-h).We quantified these differences and observed a significantly higher number (Figure 3i) and an increase in the diameter (Figure 3j) (number: STED microscopy further supports the observation that ring-shaped Mover is associated with LDs (Figures 5 and 6).In particular, STED analysis revealed that endogenous Mover does not appear as closed rings around OA induced LipidTox-positive LDs, but as a broadly distributed punctate staining around the LD surface (Figure 5a-d), similar to previously reported results for the LD proteins perilipin-2 and perilipin-5 (Gemmink et al., 2018).Furthermore, we occasionally observed that Mover formed local clusters on the LD surface (Figure 5e-g) or appeared at the contact site between LDs (Figure 5h).These results raise the possibility that Mover is arranged in nanodomains on the LD surface.We then stained for Mover immunofluorescence in cortical astrocyte cultures that expressed monomeric GFP (mGFP) as a control (Figure 6a) or the recombinant mGFP-Mover (tagged with mGFP at the N-terminus) (Figure 6b).As expected, the antibody revealed punctate staining for endogenous Mover in mGFP expressing astrocytes (Figure 6a).In contrast, the immunostaining covered the entire LD surface in mGFP-Mover expressing astrocytes (Figure 6b).This suggests that there is additional capacity for Mover recruitment on the LD surface in control astrocytes, which recombinant Mover can occupy.Moreover, in mGFP-Mover overexpressing astrocytes, the LDs appeared to be easier to detect than in control astrocytes, even without OA (Figure 6).This led us to hypothesize that overexpressing Mover affects LDs.

| Effect of Mover deletion on LD formation in astrocytes
To

| Effect of Mover overexpression on LD formation in astrocytes
In order to perform a quantitative gain of function assay, we gen- IRES driven membrane targeted GFP as reporter.We used the GFP signal to delineate the borders of transfected cells.A plasmid encoding only a membrane-targeted GFP (GFP-F) was used as a control.
We then expressed these constructs in rat cortical astrocyte cultures and quantified the number of LDs within transfected cells (Figure 8a).To apply stringent criteria for quantification we only regarded structures as LDs when they were both LipidTox-positive and visible in DIC (Figure 8b).We observed significantly more LDs in the cells overexpressing Mover compared to control cells expressing GFP-F (Figure 8) (Mover overexpressing cells: 114.5,67,198 [median,25th and 75th percentile]; control cells: 12, 6, 44 [median, 25th and 75th percentile]; p-value <.0001, Wilcoxon-Mann-Whitney test, two-tailed) (Figure 8c).This result explains why the LDs were easier to detect in cells overexpressing Mover than in control cells (Figure 6b) and demonstrates that although Mover is not essential for the formation of LDs it is sufficient to increase the number of LDs in astrocytes.

| Identification of Mover sequences necessary for LD association
To determine Mover sequences necessary for LD association, we transfected HEK293 cells with recombinant Mover constructs.
We investigated the ability of each recombinant Mover construct to associate with LDs after OA treatment.In addition to the N-terminal  et al., 2018).However, while endogenous Mover was organized in nano-clusters on the LDs, perilipin-2 signal appeared to be more homogenously distributed.
Interestingly, we observed that Mover sometimes labels LDs (identified by LipidToxRed staining) that are perilipin-2 negative in hippocampal co-culture.This was, however, not the case in cortical astrocyte cultures, where almost all LDs were immunopositive for both Mover and perilipin-2.We conclude from this observation that not all astrocytic LDs exhibit the same surface protein coating.The differences we observed are consistent with results from CHO K2 cells, Huh7 cells, in mouse liver, and in mouse brown adipose tissue, where LDs were separated by size and distinct surface protein compositions (Zhang et al., 2016).Coat identity could depend on the size of the LD, the time elapsed after their formation, or their interaction with a distinct organelle in the cell.Furthermore, the brain-specific LD protein GRAF1a shows similarities with Mover in this respect: recombinant GRAF1a surrounds most astrocytic LDs after OA treatment, and a subset of GRAF1a-positive LDs are devoid of perilipin-2 or perilipin-3 (Lucken-Ardjomande Häsler et al., 2014).In other aspects, Mover differs from GRAF1a: STED microscopy revealed GRAF1a is mainly found at LD contact sites but also forms long vesicular structures.In contrast, we found that Mover forms nanodomains across the entire LD surface.In addition, we demonstrated that endogenous Mover is found around astrocytic LDs, without the need of overexpression.From these results we conclude that, while Mover shows characteristics similar to GRAF1a, it localizes to different substructures on the LD, suggesting that Mover is involved in different mechanisms.
Synaptic vesicles and LDs are highly specialized organelles that are distinct in their structure, function, and biogenesis.One major difference is their membrane structure: while synaptic vesicles have a phospholipid bilayer membrane, LDs exhibit a phospholipid monolayer membrane.Thus, the presence of Mover around both synaptic vesicles and LDs is intriguing.The dual association of Mover on both structures is reminiscent of Annexin VI, which was first identified as interaction partner of the synaptic vesicle protein synapsin 1 (Inui et al., 1994), and more recently has been identified as a LD binding protein involved in hepatocytic LD formation (Cairns et al., 2017).
Likewise, α-synuclein, initially found to regulate synapse maintenance (Murphy et al., 2000), has also been shown to modulate the triacylglyceride turnover in brain LDs (Cole et al., 2002).Parkinson's disease related α-synuclein variants partially lost their affinity for LDs or their ability to regulate triacylglyceride metabolism (Cole et al., 2002), likely contributing to the disease.repeats that generate amphipathic helices (Rowe et al., 2016).Mover was not detected at the ER and does not contain a BAR domain.
However, Mover contains a domain with structural similarity to pleckstrin-homology domains, called hSac2 domain, spanning amino acids 53-163 and covering the entire exon 2 (from amino acids 93-151).In the protein sac2, the hSac2 domain lacks phospholipid binding capacity, but supports subcellular targeting to early endosomes through dimerization (Hsu et al., 2015).In line with the observed lack of phospholipid binding, our results indicate that the hSac2 domain is not sufficient for the binding of Mover on the LD surface.On the other hand, deletion of exon 2, and thus removal of the hSac2 domain, prevented targeting of Mover to LDs in our study.
This indicates that the hSac2 domain, while not sufficient, may contribute to the binding of Mover to LDs.
Strikingly, a single mutation exchanging the phenylalanine F206 for an arginine abolished the binding of Mover to LDs.This phenylalanine is localized in the calmodulin-binding sequence of Mover.This sequence is predicted to form an amphipathic helix.Thus, Mover may use a binding mode reminiscent of the mode used by perilipins, where amphipathic helices are the major binding sites.Intriguingly, phenylalanine F206 is also essential for both dimerization and targeting of Mover to the presynapse (Akula et al., 2019).Therefore, this short region of Mover is apparently involved in multiple functions, including calmodulin (CaM) binding, dimerization, targeting to presynaptic terminals in neurons, and targeting to LDs in astrocytes.A simple explanation for the importance of F206 would be that its hydrophobicity contributes to lipid binding or insertion within the phospholipid layer.
Alternatively, F206 may be important for the overall folding of Mover.

| Mover overexpression induces LDs accumulation in astrocytes
Knockout of Mover did not prevent LD formation in astrocytes.This is not surprising given that some species, such as Caenorhabditis elegans and Drosophila, do not have Mover-related genes in their genomes but do form LDs. Thus, Mover may rather be a regulator of LDs.
We found that Mover overexpression induces accumulation of LDs in cortical astrocytes.Under normal conditions, LD number in cells is low, except for specialized fat storage cells such as adipose tissue.However, under stress, cells might accumulate LDs as a response and potential protective mechanism (Smolič et al., 2021).How Mover leads to the accumulation of LDs is not clear.The biogenesis of LDs comprises at least three steps: Triacylglycerides are synthesized in the ER and accumulate between the two membrane leaflets, leading to the budding of the ER as nascent LDs.Immature LDs then grow by additional triacylglyceride synthesis at the LD.Accordingly, the surface phospholipid layer needs to expand too (Olzmann & Pedro Carvalho, 2019;Wilfling et al., 2014).
Finally, LDs can grow by fusion with other LDs or by substrate exchange with other organelles, with which LDs form highly dynamic contacts (Valm et al., 2017).Although we did not observe any evidence of Mover localization to any organelle other than LDs, we cannot exclude that small fractions of Mover act at the ER.If Mover acts at a stage downstream of LD budding it may increase the size of existing LDs in a way that previously existing LDs become detectable after Mover overexpression.Mechanistically, this effect could involve increased triacylglyceride synthesis, reduced lipolysis, or increased LD fusion.
It will be interesting to see in future studies whether the increase in Mover expression, which occurs in the anterior cingulate cortex of schizophrenic patients (Clark et al., 2006), leads to an increase in astrocytic LDs and whether this reflects disorder-induced dysregulation or rather a protective role of Mover.

| CONCLUSION
In summary, our results demonstrate a novel localization for the pre-
Abberior Expert Line STED Instrument microscope using avalanche photodiodes (APD) detectors was used for transfected Mover in cortical astrocyte cultures.STED images were acquired with UPLSAPO-HR 100Â APO, 1.4 NA immersion objective in oil.The samples were excited with a 561 nm, 40 MHz pulsed excitation laser (Abberior Instruments) for LipidTox and a 640 nm, 40 MHz pulsed excitation laser (Abberior Instruments) for STAR 635P.A STED laser, 775 nm (1.2 W) pulsed laser from MPB Communications Inc., was used for STED acquisitions.Images were acquired with the Imspector software.The resolution for the STAR635P fluorophore and the 775 nm depletion laser was below 70 nm, for STAR580 the resolution was below 140 nm.
2.7.3 | Confocal microscopyBrain slices were imaged at a Leica SP5 confocal microscope (Leica DM 6000) equipped with a single photon Ar-laser (488 nm excitation) and a HeNe-laser (633 nm excitation).A 20Â immersion objective was used for image acquisition.
were manually selected and characterized as rings with a hole; closed circles were visually excluded.The selection of the astrocyte cells was based on GFAP-positive staining.As the immunofluorescence signal of Mover in epifluorescence images produced closed ring-shaped structures, those structures were referred to as "Mover rings."For further analysis, a subset of rings was blindly and randomly chosen by assigning random numbers to each indexed ring, and selecting the top 5-10 rings per region.The diameters were determined in microns using FIJIs straight tool.Mover rings were counted in two set of stainings: Mover + GFAP (four independent preparations [N = 4 animals] with 84 [ÀOA] and 70 [+OA] regions analyzed) or Mover + Bodipy + Perilipin2 (three independent preparations [N = 3 animals] with 52 [ÀOA] and 61 [+OA] regions analyzed).Each region had a size of 154.76 Â 154.76 μm.For quantification, results were grouped in untreated (ÀOA) and OA treated (+OA) cells.To test for normality of the data, we applied a Shapiro-Wilk test.In GraphPad Prism 9, results were arranged in scatter plots showing the median and quartiles.For LD quantification in cortical astrocytes, structures were counted as LDs if they were visible by differential interference contrast (DIC) and contained a positive LipidTox staining.ROIs were selected based on the GFP staining for transfected cells or on the GFAP staining for astrocytes from WT and KO Mover mice.LDs in Mover overexpressing group, control group KO and WT astrocytes were counted using the multi point tool from ImageJ while blinded for the condition.For data analysis, the two-tailed Wilcoxon-Mann-Whitney test for independent samples (for Mover ring and diameter, and control vs. overexpression analysis) and Kruskal-Wallis test with Dunn's comparison (for multiple comparison in WT vs. KO data), p-value = .05,were performed in GraphPad Prism 9.

F
I G U R E 1 Mover immunofluorescence appears as ring-shaped structures in astrocytes.(a) Epifluorescence microscopy of a hippocampal neuron-astrocyte co-culture stained with MAP2 for neurons (blue), synaptophysin (magenta), and Mover (green).(b-e) Zoom of the box in (a), highlighting the ring-shaped Mover structures; merge (b) and single channel images (c-e).The arrow indicates the extraneuronal ring-shaped Mover structure; the arrowheads indicate co-localization of synaptic Mover with synaptophysin around MAP2 stained neurons.(f) Epifluorescence microscopy of hippocampal neuron-astrocyte culture; stained with GFAP for astrocytes (blue), synaptophysin (magenta), and Mover (green).(g-j) Zoom of the box in (f); merge (g) and single channel images (h-j).The ring-shaped Mover structure is located within the GFAP stained astrocytes.Scale bars = 20 μm.F I G U R E 2 Mover is present in astrocytes of acute brain slices.(a) Maximum intensity projection of horizontal mouse brain section in the hippocampal formation.The z-volume is 46.8 μm with a step size of 0.2 μm.Panels show an overview of the field with DAPI (gray), astrocytes (marked with GFAP, magenta), and Mover (green), as well as a zoom of the box shown in the overview.(b) Single plane images of the astrocyte shown in (a).The white arrows/arrowheads indicate Mover signal that can be seen appearing on the astrocyte process.Slice thickness 350 μm.Scale bars = 20 μm.These ring-shaped Mover structures were then found to be present in GFAP-positive astrocytes (Figure 1f-j).Extra-neuronal Mover structures were remarkably larger than synaptic Mover puncta (Figure 1b-e, arrowheads) and were not associated with the synaptic protein synaptophysin (Figure 1b-e, arrow).These observations suggested that

p
< .0001,Wilcoxon-Mann-Whitney test, two-tailed; diameter: p < .0001,Wilcoxon-Mann-Whitney test, two-tailed) of ring-shaped F I G U R E 3 OA treatment increases the number and size of Mover-positive ring-shaped organelles.(a) Representative overview of untreated hippocampal co-cultures stained with GFAP (magenta) and Mover (green).(b-d) Zoom of the box in (a) show few small ring-shaped structures within the astrocyte; merge (b) and single channel images (c, d).(e) Representative overview of OA treated hippocampal co-cultures stained with GFAP (magenta) and Mover (green).(f-h) Zoom of the box in (e) show larger ring-shaped Mover structures within the astrocyte; merge (f) and single channel images (g, h).(i) Representation of the number of ring-shaped Mover structures per region in absence (ÀOA) or presence (+OA) of OA treatment.(j) Representation of the diameter of ring-shaped Mover structures per region in absence (ÀOA) or presence (+OA) of OA treatment.Scatterplot showing median and quartiles.Statistical evaluation was done using a two-tailed Wilcoxon-Mann-Whitney test for 52-84 regions in 3-4 different preparations (n = 4 for Mover + GFAP staining, n = 3 for Mover + Bodipy + Perilipin2 staining).Asterisks represent significant differences with ***p ≤ .001.Scale bars = 20 μm.Mover structures in astrocytes after OA treatment: Untreated cells showed 10 (6, 15) (median, 25th and 75th percentile) ring-shaped Mover structures per region with a diameter of 1.03 (0.89, 1.23) μm (median, 25th and 75th percentile), compared to 66 (47, 87) ringshaped Mover structures with a diameter of 1.29 (1.10, 1.57) μm (median, 25th and 75th percentile) in OA treated cultures (Figure 3i,j).These experiments demonstrated striking similarities in the behavior of LDs and ring-shaped Mover structures that led us to further investigate their identity.3.2 | Mover is associated with LDs To confirm the presence of Mover around LDs we stained OA treated hippocampal co-cultures with LD markers.The LD associated protein perilipin-2 (Heid et al., 1998) was selected as specific staining of the LD surface.Mover was associated with both perilipin positive (Figure 4, arrows) and perilipin negative structures (Figure 4, arrowheads).Thus, one pool of Mover is associated with LDs while another pool of Mover may be associated with different organelles or with F I G U R E 4 Co-localization of Mover with LD markers Bodipy, LipidTox, and perilipin-2.(a) Hippocampal co-cultures stained for Mover (green), GFAP (blue), and perilipin-2 (magenta).(b-e) Zoom of the box in (a) showing partial overlap of Mover with perilipin-2.Merge (b) and single channel images (c-e).(f) Hippocampal co-cultures stained for Mover (green), Bodipy (blue), and perilipin-2 (magenta).(g-j) Zoom of the box in (f) showing overlap of Mover with LD marker Bodipy.Merge (g) and single channel images (h-j).(k) Hippocampal co-cultures stained for Mover (green), LipidTox (blue), and perilipin-2 (magenta).(l-o) Zoom of the box in (k) showing overlap of Mover with LD marker LipidTox.Merge (l) and single channel images (m-o).Arrows indicate overlap of Mover and perilipin-2, while arrowheads indicate Mover positive, but perilipin-2 negative structures.Scale bars = 20 μm.F I G U R E 5 STED microscopy reveals association of endogenous Mover with LDs.(a) OA induced LDs from hippocampal co-cultures stained with LipidTox (magenta) and for Mover (green) imaged by STED microscopy.(b-d) Zoom of the box shown in (a).Merge (b) and single channel images (c, d) show the distribution of endogenous Mover around the LDs and the neutral lipid staining inside.(e) OA induced LDs stained with LipidTox (magenta) and Mover (green) imaged by STED microscopy.(f-h) Zoom of three regions of interest shown in (e) indicating the distribution of Mover in nanodomains around LDs. Scale bars = 1 μm.F I G U R E 6 STED microscopy shows targeting of recombinant Mover to the LD surface in cortical astrocyte cultures.(a) Distribution of Mover immunofluorescence (green) around LipidTox (LTR) stained LDs (magenta) in mGFP transfected control astrocytes in presence (+OA) of OA treatment compared to untreated (ÀOA) cells.(b) Distribution of Mover immunofluorescence (green) around LipidTox (LTR) stained LDs (magenta) in recombinant-Mover overexpressing astrocytes in presence (+OA) or in absence (ÀOA) of OA treatment.Mover overexpression was sufficient to cause a marked increase in the visibility and contrast of LDs compared to untreated mGFP transfected control, where LDs appear shapeless.Scale bars = 1 μm.perilipin negative LDs.To test this, we labeled the LDs' neutral lipid core using the fluorescent dyes Bodipy and LipidTox.Virtually all ring-shaped Mover structures surrounded accumulations of neutral lipids, as verified by both Bodipy (Figure 4f-j) and LipidTox stainings (Figure 4k-o).These results indicate that Mover localizes to the surface of astrocytic LDs and that some of them are not associated with perilipin-2.
erated a new plasmid encoding untagged Mover together with an F I G U R E 7 Cortical astrocyte cultures from WT and KO Mover mice do not show difference in the number of LDs without or after OA treatment.(a) Representative image of cortical astrocytes from WT or KO mice untreated or treated with 400 μM OA for 24 h.(b) Quantification of the LD number per astrocyte in WT and KO without or with OA treatment indicates no significant difference between WT and KO Mover astrocytes in LD number.Scatterplot showing median and quartiles.N = 2 independent astrocyte cultures for WT and for KO.n = 22-27 astrocyte per cultures.ns = non-significant difference with p > .05,Kruskal-Wallis test, with Dunn's correction.Scale bars = 10 μm.
tagged mGFP-Mover (Figure 9a-d), the two C-terminal tagged Mover constructs Mover-mGFP (tagged with monomeric GFP) (Figure 9e-h) and Mover-eGFP (tagged with conventional eGFP) (Figure 9i-l) co-localized around LDs.To exclude a contribution of the GFP tag to the LD association, we also tested a C-terminal myc-tagged Mover construct that showed association to LDs after staining for Mover (Figure 9m-p).These results suggest that Mover does not need a free N-terminal or C-terminal domain to interact with LDs.Likewise, a truncated Mover construct missing the first 51 amino acids (mGFP-52-266) still associated with LDs (Figure 9q-t), which indicates that the N-terminal 51 amino acids are not required for LD binding.F I G U R E 8 Mover overexpression induces accumulation of LDs in astrocytes.(a) Cortical astrocyte cultures were transfected either with a recombinant Mover-GFP construct (overexpression) or with a GFP control (control).LDs were stained with LipidTox (magenta).Astrocytes were delineated (white dashed line) using the membrane targeted GFP signal (memGFP) (green); bottom panels show the respective DIC channel.(b) Zoom of the boxes in DIC channel in (a) reveals LD accumulation after Mover overexpression.(c) Quantification of the LD number per astrocyte in overexpression and control condition reveals a marked increase of LDs after Mover overexpression.Scatterplot showing median and quartiles.Statistical evaluation was done using a two-tailed Wilcoxon-Mann-Whitney test.Asterisks represent significant differences with ***p ≤ .001.N = 3 independent experiments.n = 119 GFP-F control astrocytes and 158 Mover overexpressing astrocytes.Scale bars = 10 μm.F I G U R E 9 Different recombinant Mover constructs associate to LDs in HEK293-cells.OA treated HEK cells were transfected with mGFP-Mover (a-d), Mover-mGFP (e-h), Mover-eGFP (i-l), Mover-myc (m-p), mGFP-52-266 (q-t), and Mover-5p-eGFP (u-x).Recombinant Mover signal (green) from the respective GFP tag, except for Mover-myc, where Mover antibodies were used, and LipidTox (LTR) stained LDs (magenta) show strong association of different constructs to LDs.First panel per row on the left gives an overview, the other panels zoom on the indicated boxes with a merge and the respective single channel images.Scale bars = 10 μm.F I G U R E 1 0 Extended regions of Mover are required for targeting to LDs in HEK293-cells.OA treated HEK cells were transfected with Mover-4cam-eGFP (a-d), F206R-eGFP (e-h), mGFP-52-266-F206R (i-l), mGFP-hSac2 (m-p), Mover-DE2-mGFP (q-t), and mGFP (u-x).Recombinant Mover signal (green) from the respective GFP tag and LipidTox (LTR) stained LDs (magenta) reveal no association of the probed constructs with LDs, similar to GFP control.First panel per row on the left gives an overview, the other panels zoom on the indicated boxes with a merge and the respective single channel images.Scale bars = 10 μm.We then explored if phosphorylation is necessary for LD association and expressed a phosphorylation-deficient Mover construct with mutations at five predicted phosphorylation sites (Mover-5p-eGFP, including the following mutations: T13A, S14A, T64A, S221A, and Y257F).This phosphorylation-deficient mutant did not lose its ability to associated with LDs either (Figure9u-x), suggesting that Mover does not require phosphorylation at these five distinct sites to bind to LDs.To further explore regions required for LD association, we tested three possibilities arising from our previous work.First, introducing four point mutations (F206R, K207E, K215E, and K219E) abolishes calmodulin binding of Mover(Akula et al., 2019).A construct harboring these mutations (Mover-4cam-eGFP) failed to associate with LDs (Figure10a-d).Interestingly, a single point mutation (F206R) was sufficient to abolish LD association, both in full length (Figure10e-h) and in the truncated construct mGFP-51-266 (Figure10i-l).These results indicate that F206 is essential for LD association.Second, in silico analysis predicts a conserved domain, called hSac2 domain, spanning amino acids 53-163 of Mover.A construct encoding only the hSac2 domain did not bind to LDs (Figure10m-p).Third, database entries suggest that at the cDNA level a Mover isoform exists that lacks exon 2, which encodes amino acids 94-150 and contains a large part of the hSac2 domain.A construct lacking exon 2, and thereby disrupting the hSac2 domain, failed to bind to LDs (Figure10q-t).Together these data indicate that in addition to F206 the amino acids 94-150 are required for LD association.Moreover, they show that the hSac2 domain is required but not sufficient for LD binding.All deletion constructs that failed to bind to LDs were distributed in a way similar to soluble mGFP (Figure10u-x).4 | DISCUSSIONAstrocytic LDs protect neurons against lipid stress.The biogenesis and function of LDs relies on the composition of their protein coat.In this study, we identified the presynaptic protein Mover as a novel component of the astrocytic LD coat using several LD markers.We showed that the association with LDs is disturbed by large deletions in the exon 2 region and by point mutations in the calmodulin binding region of Mover.Furthermore, while knockout of Mover does not affect astrocytic LDs, overexpression of Mover induces LD accumulation in astrocytes, highlighting Mover as a candidate regulator of astrocytic LDs.4.1 | Mover is a novel component of astrocytic LDs Several lines of evidence indicate that Mover is a protein of the LD surface in astrocytes.First, using epifluorescence microscopy, we found endogenous Mover surrounding LDs identified by two distinct neutral lipid stains LipidTox and Bodipy.Similarly, recombinant Mover retained the ability to bind to LipidTox-positive LDs, both in astrocytes and HEK cells.To strengthen our observation that Mover is part of the LD coat, we labeled perilipin-2, a known part of the LD surface, and saw that it co-localized well with Mover.Second, after OA treatment, Mover behaves as expected for a LD associated protein, in that the number and size of the ring-shaped structures formed by Mover increased.Third, super-resolution STED microcopy confirmed the presence of endogenous and recombinant Mover on the LD surface.The punctate STED signals found across the LD surface is reminiscent of the STED signal of perilipin-2 (Gemmink

4. 2 |
Binding of Mover around LDs involves its predicted calmodulin domainSeveral modes of binding to LDs have been described: some proteins containing hairpin structures translocate from the ER to LDs during LD budding(Kory et al., 2016); GRAF1a binds to LDs via its membrane binding BAR domain (Lucken-ArdjomandeHäsler et al., 2014); perilipins associate with existing LDs through n-terminal 11mer synaptic protein Mover on the surface of astrocytic LDs.Overexpression of Mover increased the number of LDs, suggesting that Mover affects LD biogenesis.This introduces Mover as a novel component of astrocytic LDs and a candidate regulator of the mechanisms driving accumulation of LDs in astrocytes.AUTHOR CONTRIBUTIONS TD, JK and FD designed the experiments.JK performed Mover staining in co-cultures and in brain slices and identification of Mover sequences necessary for LD association.JK established the OA treatment.TD performed the cloning of Mover recombinant constructs.FD established the rat and mice astrocyte cultures and performed the comparison of WT and KO astrocytes.JK and FD performed the STED experiments.TTD performed gain of function experiment.JK wrote the first draft of the manuscript.JK, FD, TTD and TD wrote the manuscript.
test if LD formation is affected in the absence of Mover, we com- percentile]; adjusted p-value = >.99,Kruskal-Wallistest, two-tailed).Thus, WT and KO astrocytes had similar capability to respond to OA treatment.These results indicate that Mover is not required for formation of LDs in astrocytes.