Our previous data showed that NTE/SWS function is important for both glia and neuronal cells in the brain (Melentev et al., 2021; Ryabova et al., 2021). After downregulation of NTE/SWS in neurons, adult flies show a decrease in longevity, locomotor and memory deficits, and severe progression of neurodegeneration in the brain (Melentev et al., 2021). We have shown that NTE/SWS plays a role in the development of the learning center of the brain involved in short-term and long-term memory storage, olfactory control, and startle-induced locomotion (Melentev et al., 2021). In addition, we found that flies with NTE/SWS deficiency in neurons or glia show mitochondrial abnormalities as well as accumulation of ROS and lipid droplets (Melentev et al., 2021; Ryabova et al., 2021). Now we have decided to determine the cell type in which NTE/SWS plays a determining role in the maintenance of brain health.

Similar to its human counterpart, NTE, which is found in virtually all tissues, including the nervous system (https://www.proteinatlas.org/ENSG00000032444-PNPLA6/tissue), NTE/SWS is ubiquitously expressed in Drosophila brain, detected by immunohistochemical analysis using SWS-specific antibodies (Figure 2—figure supplement 1A). NTE/SWS is a transmembrane phospholipase anchored to the cytoplasmic side of the endoplasmic reticulum to regulate lipid membrane homeostasis. Its cytoplasmic localization makes it difficult to determine precisely in which brain cell type it has more pronounced expression, as neurons and glia have very complex shapes and forms. Therefore, additional markers must be used to discriminate NTE/SWS expression in the brain. To address this, we expressed membrane-bound GFP under control of the sws promoter (sws-Gal4; UAS-CD8::GFP), which allows labeling of membranes of cells in which the sws promoter is active. Importantly, sws was strongly expressed in the glia that surround the brain and form the blood-brain selective permeability barrier (Figure 1E and F). In Drosophila, the BBB is entirely made by two glial cell layers: PG and SPG (Figure 1C and D). With the help of sophisticated SJs, SPG cells form a tight barrier that prevents paracellular diffusion and separates the CNS from hemolymph. Since the BBB is protecting the brain from toxic substances, and NTE/SWS deregulation is associated with toxicity-induced neurodegeneration, we investigated whether NTE/SWS has a functional role in BBB maintenance and its selective permeability.

To test if loss of NTE/SWS affects the barrier structure, we analyzed the expression pattern of Coracle (CoraC), which is a major component of SJs (Yi et al., 2008). In controls, CoraC is strongly expressed by SPG cells, shown as a smooth line at the brain surface (Figure 2A, green arrow). Upon sws loss, the CoraC pattern at the brain surface was broken and contained lesions and membrane aggregations (Figure 2B, magenta arrow).

Previous characterization of the sws loss-of-function mutant showed that NTE/SWS deficiency resulted in the formation of membranous glial structures, especially in the lamina cortex (Kretzschmar et al., 1997). Since NTE/SWS is ubiquitously expressed, we utilized the double driver line (repo, nSyb-Gal4, Figure 2—figure supplement 1F and F′) to achieve its downregulation in both neuronal and glial cells (Figure 2—figure supplement 1C). Since these animals had the same disorganized structure of brain surface as the loss-of-function mutant, we concluded that NTE/SWS functions specifically in the nervous system to preserve brain surface structure. Moreover, downregulation of sws in all glial cells (repo>sws^RNAi) resulted in the same phenotype (Figure 2—figure supplement 2C). At the same time, upon sws downregulation in neurons, we did not observe formation of lesions and membrane clusters in the brain surface (Figure 2—figure supplement 2E), indicating a cell-autonomous function of NTE/SWS in glia to maintain BBB organization.

To test if NTE/SWS has a cell-autonomous role in the brain barrier cells, we used already existing SPG driver lines (moody-Gal4, UAS-CD8::GFP and Gli-Gal4, UAS-CD8::GFP, Figure 2—figure supplement 1D and E) and UAS-sws^RNAi. We found that, similar to pan-glial sws knockdown, its downregulation specifically in SPG cells caused the formation of lesions and membrane clusters within the brain surface (Figure 2C, Figure 2—figure supplement 2D, blue and magenta arrows). Importantly, expression of Drosophila or human NTE in these glia cells rescued this phenotype (Figure 2D and H, Figure 2—figure supplement 2F), demonstrating the conserved function of this protein in SPG cells for brain surface formation and possibly maintenance of the brain barrier.

There have been remarkable recent advancements in the field of protein structure prediction, offering valuable tools for exploring three-dimensional structures with unprecedented effectiveness. We used the AlphaFold2 prediction and the PyMol tools (Jumper et al., 2021) to display predicted structure models of the human NTE and Drosophila NTE/SWS proteins. Both proteins contain a highly conserved patatin-like phospholipase (EST) domain (Figure 2—figure supplement 3, EST domain in magenta). EST domains in NTE/SWS (952–1118) and human NTE (981–1147) demonstrated a remarkably high level of confidence, exhibiting helical structures with predicted local distance difference test scores (pLDDT) exceeding 90 (Figure 2—figure supplement 3).The EST domain exhibits a distinctive architectural pattern comprising three layers of α/β/α structure. Its central region is formed by a six-stranded β-sheet, flanked by α-helices in the front and back. Upon comparing the predicted structures of EST-SWS and EST-NTE, we observed a significant overlap between them (Figure 2—figure supplement 3). These findings offer additional evidence of the high conservation of functional domains in NTE/SWS and the close relationship between these proteins across different species.

Together, the remarkable similarities observed between human and Drosophila SWE/NTE protein structure along with their shared involvement in the formation and maintenance of the brain barrier in Drosophila emphasize their close relationship and suggest a conserved function in BBB maintenance.

Next, we aimed to understand the nature of the SPG phenotype caused by sws deficiency. SPG cells have a very specific shape; they are thin and very large. Fewer than 50 SPG cells surround one adult brain hemisphere and a single SPG cell can cover the size of one half of the imaginal disc of the eye, covering an area equivalent to approximately 10,000 epithelial cells (Limmer et al., 2014; Hartenstein, 2011; Silies et al., 2007). Therefore, to better visualize the defects in surface glia organization upon sws loss, we introduced moody-Gal4, UAS-CD8::GFP (moody>CD8::GFP) constructs into the sws¹ mutant background, which allowed analysis of SPG cell membranes. To our surprise, we observed that almost all lesions that were formed near the brain surface contained membrane material marked by CD8::GFP (Figure 2F). This was in sharp contrast to the control, where SPG membranes formed a distinct GFP-positive line (Figure 2E). Importantly, the same excessive SPG cell membranes were observed inside the lesions formed upon sws downregulation explicitly in SPG cells (Figure 2G), confirming that NTE/SWS is required cell-autonomously in SPG cells for the proper architecture of the surface glia.

Next, we wanted to understand the origin of these excessive membranes observed in sws-deficient glial cells. This task appeared to be quite challenging, as SPG cells form a very thin polarized endothelium, not even reaching 1 μm thickness in most areas (Limmer et al., 2014). In addition, SPG cells localize in very close proximity to each other and to neurons, making the analysis of subcellular protein localization challenging. Therefore, to dissect in more detail the sws-related phenotype of accumulated SPG membranes inside the lesions on the brain surface, we used an electron microscopy approach.

We found that sws mutants showed the formation of various multilamellar bodies in the brain, which were not observed in the control (Figure 3A–B′). These atypical structures ranged in size from 5 to 15 μm and contained concentrically laminated and multilayered membranes (yellow arrows), lipid droplets (red arrows), and other partially degraded organelles or cytoplasmic constituents. We hypothesized that these inclusions most likely correspond to secondary lysosomes in the phase of digesting endosomal cargo, which are a hallmark of lysosomal storage diseases (LSDs).

To authenticate the nature of membranous accumulation in sws mutants, we used endosomal and lysosomal markers. Rab7 is a small GTPase that belongs to the Rab family and controls transport to late endocytic compartments such as late endosomes and lysosomes (Guerra and Bucci, 2016). Immunohistochemical analysis demonstrated that in contrast to controls, where Rab7 was present in relatively small and evenly dispersed throughout the brain late endosomes and lysosomes (Figure 3D, red), in sws-deficient brains, accumulation of Rab7-positive compartments was observed. Moreover, Rab7-positive structures colocalized with atypical membrane aggregates of SPG cells (Figure 3D′, yellow). The same assemblies were observed upon sws downregulation in SPG cells (Figure 3D′, yellow).

Rab7 controls biogenesis of lysosomes and clustering and fusion of late endosomes and lysosomes (Feng et al., 2014). Therefore, to support the idea that these abnormal cellular accumulations are of lysosomal origin, we used an additional marker – CathepsinL – which is a key lysosomal proteolytic enzyme expressed in most eukaryotic cells (Xu et al., 2021). We found that sws loss or its downregulation in barrier-forming glia cells resulted in the appearance of CathepsinL-positive inclusions that colocalized with GFP-labeled membrane aggregates formed in the mutant SPG cells (Figure 3E–E′′, yellow). We conclude that the structures observed upon NTE/SWS deregulation are abnormally enlarged lysosomes.

Next, we quantified the number of brain hemispheres with atypical Rab7- or CathepsinL-positive accumulations. In the control groups, very few (<10%) of the analyzed brains showed accumulation of Rab7 or CathepsinL. However, in mutants with sws loss-of-function and with sws SPG-specific downregulation, a significant increase in the frequency of brains containing Rab7- or CathepsinL-positive aggregates was observed (Figure 3F and G). Since sws-associated neurodegeneration is age-dependent (Melentev et al., 2021; Kretzschmar et al., 1997; Dutta et al., 2016; Sujkowski et al., 2015; Sunderhaus et al., 2019), we tested if abnormal lysosomes positive for Rab7 and CathepsinL increase with age. Analysis of the brains of 15-day-old sws downregulation in SPG cells demonstrated ~2-fold increase in the percentage of brains with lysosomal accumulations within the brain surface in comparison to 1-day-old animals (Figure 3F and G). These data demonstrate for the first time that NTE/SWS-associated phenotypes might be additionally characterized by the excessive storage of cellular material in lysosomes that is accelerated by age.

Importantly, similar abnormal buildup of cellular material in lysosomes have been found in hippocampal neuropil (Akassoglou et al., 2004) and spinal axons of NTE-deficient mice (Read et al., 2009). While these structures have not been specifically described as lysosomal defects, the presence of similar dense bodies containing concentrically laminated and multilayered membranes in NTE-deficient mice suggests that, similar to Drosophila, NTE/SWS-related phenotypes in mammals may also be associated with excessive storage of cellular material in lysosomes. Lysosomal changes and dysfunction have been involved in the initiation and development of numerous diseases, such as cancer, autoimmune, cardiovascular, neurodegenerative, and LSDs (Cao et al., 2021; Hebbar et al., 2017). In particular, LSDs are a group of rare metabolic disorders caused by inherited defects in genes that encode proteins vital for lysosomal homeostasis, such as lysosomal hydrolases or membrane proteins. LSDs often manifest as neurodegenerative disorders. Therefore, next, we wanted to investigate how lysosomal accumulation in SPG cells affects their functions, resulting in progressive brain degeneration.

The main function of SPG cells is to protect the CNS from being exposed to molecules that are harmless to peripheral organs but toxic to brain neurons. SPG cells form a thick polarized endothelium, selective permeability of which is achieved by forming very tight SJs that provide structural strength and a barrier that controls the flow of various solutes from outside the brain (Limmer et al., 2014; Hartenstein, 2011; Silies et al., 2007). Since our data show that the expression pattern of a key SJ protein, CoraC, is dramatically perturbed in sws mutant brains (Figure 2A and B), we decided to test if deregulation of NTE/SWS can affect the ability of SPG cells to form a selective permeability barrier.

As a result of abnormal BBB function, the CNS becomes permeable to small molecules such as dextran-coupled dyes. To test BBB permeability, the 10 kDa dextran dye was injected into the abdomen of flies (Figure 4A). After injection, animals were allowed to recover for at least 12 hr, followed by the dissection and analysis of adult brains. In controls, dextran dye predominantly remained at the outer surface of the brain (Figure 4B and B′). In contrast, the dye was detected inside almost all of the sws¹ mutant brains (Figure 4C and C′). Moreover, the downregulation of sws in different types of glial cells also caused increased permeability of brain barrier in more than 80% of the analyzed brains (Figure 4D). Expression of Drosophila NTE/SWS and human NTE in glia in sws¹ mutant rescued the organization of the surface glia (Figure 2H) and partially rescued the barrier phenotype, suggesting that human NTE and Drosophila NTE/SWS are important for the BBB integrity in Drosophila (Figure 4D). Taken together, our results demonstrate that SPG cells with NTE/SWS deficiency are characterized by defective brain barrier function and lysosomal accumulation of excess cellular material, which includes membranes.

Next, we wanted to understand whether the compromised brain barrier in sws mutants triggers the activation of any cellular stress pathways, including apoptosis, ferroptosis, oxidative stress, ER stress, and inflammation. We treated mutant flies for 14 days with different anti-inflammatory substances and stress suppressors and analyzed whether observed glial phenotypes could be suppressed by any medication. We analyzed CoraC expression and compared the frequencies of abnormal brain surface appearance in the drug-treated versus untreated animals (Figure 5—figure supplement 1A, B). We revealed that sodium salicylate, a non-steroidal anti-inflammatory drug (NSAID) and rapamycin, which activates autophagy by inhibiting Tor (Xu et al., 2017), showed the best ability to suppress surface glia phenotypes in sws mutants (Figure 5—figure supplement 1B, C′). This indicates that an activated inflammatory response is associated with sws deficit.

A leaky BBB allows different toxic substances and bacteria to enter the CNS and affect neurons and glial cells, which can lead to cell death and increased inflammation in mammals (Kim et al., 2012). To test if a permeable brain barrier in general is causing inflammation in Drosophila, we decided to test if an additional mutant with defective BBB has an increased inflammatory response in the brain. We focused on a moody^ΔC17 mutant that has been previously shown to have a defective brain barrier (Bainton et al., 2005).

First, we tested whether the moody mutant shows a phenotype similar to that observed in sws mutants by analysis of the CoraC expression pattern. We observed that the surface brain layer in moody mutants or upon moody downregulation in SPG by moody-Gal4 (moody>moody^RNAi) contained lesions and had an abnormal membrane assembly, resembling CoraC expression pattern in sws mutants (Figure 5—figure supplement 2A–C and Figure 2B, magenta arrows).

Second, we analyzed if anti-inflammatory factors can reduce glial phenotypes in moody mutants, similar to sws mutants. We found that in moody mutants, the surface glia phenotype analyzed using CoraC as a marker could also be suppressed by NSAID and rapamycin (Figure 5A, Figure 5—figure supplement 1D and D′). The fact that anti-inflammatory factors can reduce glial phenotypes in both sws and moody mutants indicates that inflammation, triggered as a result of a compromised brain barrier, plays a role in a feedback loop that exacerbates the abnormal surface glia organization (Figure 5G). At the same time, inflammation inhibitors only partially rescued the BBB phenotype in moody and sws mutants, suggesting the involvement of additional pathways in maintaining the BBB.

Next, we tested whether inflammatory pathways are activated in both mutants with permeable barriers. The molecular mechanisms of innate immunity between flies and mammals are highly evolutionarily conserved. For example, Drosophila Toll and IMD pathways are nuclear factor kappa B (NF-κB)-based signaling pathways that share similarities with the Toll-like receptor and tumor necrosis factor receptor 1 signaling pathways in mammals (Pavlidaki et al., 2022; Kounatidis and Chtarbanova, 2018; Cao et al., 2013; Kounatidis et al., 2017). It has been previously shown that in glial cells, activation of the IMD pathway results in phosphorylation of the NF-κB transcription factor Relish, which is translocated to the nucleus to induce expression of the AMPs Attacin A, Cecropin A, and Diptericin (Winkler et al., 2021; Kounatidis and Chtarbanova, 2018). We performed quantitative PCR (qPCR) analysis and measured the mRNA levels of these AMPs in heads of sws and moody loss-of-function mutants. We found that the mRNA levels of all three AMPs were significantly upregulated in mutants in comparison to relevant controls (Figure 5B). These data demonstrate that both mutants with defective BBB exhibit an increased inflammatory response.

In addition, polyunsaturated fatty acids (PUFAs) have been shown to play a key role in inflammatory processes. Their oxygenated products, called eicosanoids, induce and regulate inflammation via G-protein-coupled receptor (GPCR) signaling pathways (Stanley and Kim, 2018). To find out whether levels of polyunsaturated and saturated fatty acids are changed, we measured levels of FFAs from accurately weighed heads of control flies and mutants with defective BBB (sws¹ and moody^ΔC17). FFAs were measured by gas chromatography-mass spectrometry (GC-MS) as described recently (von Hanstein et al., 2023). We found that both mutants with defective BBB show upregulated levels of linoleic acid, α- and γ-linolenic acid, eicosanoic acid, arachidonic acid, and eicosapentaenoic acid when compared to controls (Figure 5C). Additionally, levels of other FFAs involved in inflammatory response, 9-cis-tetradecenoic acid, palmitic acid, palmitoleic acid, stearic acid, and oleic acid (Miao et al., 2015; Korbecki and Bajdak-Rusinek, 2019) were elevated upon sws or moody loss (Figure 5C). These data show that in both mutants with a compromised BBB, the inflammatory response is accompanied by the accumulation of FFAs.

Given that the loss of sws results in age-dependent neurodegeneration (Kretzschmar et al., 1997), we investigated whether the increased inflammatory response is progressing with age. We performed qPCR analysis and quantified mRNA levels of AMPs (Attacin A, Cecropin A, and Diptericin) in the heads of sws mutants and flies that had sws downregulation only in SPG cells (moody>sws^RNAi) of 15- and 30-day-old flies. We confirmed that the mRNA levels of all three AMPs were significantly upregulated in mutants of both ages in comparison to the relevant controls (Figure 5D, black asterisks). Furthermore, we observed a significant age-related increase in the expression of inflammatory genes in both sws mutants and flies with sws downregulation in SPG cells (moody>sws^RNAi, Figure 5D, red asterisks), thereby illustrating the correlation between age-related NTE/SWS neurodegeneration and inflammatory processes. Importantly, expression of Drosophila NTE/SWS in SPG cells in sws¹ mutant background normalized levels of inflammatory genes expression in flies of both ages (15- and 30-day-old flies), confirming that the increased inflammatory response is a consequence of the defective BBB (Figure 5D, green asterisks). Moreover, downregulating sws in glial cells during adulthood, after BBB formation, resulted in an increased inflammatory response (Figure 5—figure supplement 1). Since previous studies have demonstrated the induction of neurodegeneration by the overactivation of innate immune-response pathways, especially elevated expression of AMPs (Cao et al., 2013), our data showing increased levels of AMPs in aging flies with a defective BBB further strengthen the connection between the BBB, AMPs, and neuroinflammation and reinforce the causative link between BBB breakdown and inflammaging.

Upon infections and autoimmune conditions, macrophages have the capability to infiltrate the brain, aiding in pathogen removal but also posing the potential risk of causing tissue damage. It has been recently shown that the IMD pathway attracts and facilitates the invasion of hemolymph-borne macrophages across the BBB into the inflamed brain during pupal stages (Winkler et al., 2021). To investigate whether the neuroinflammatory response in sws mutants is associated with the entry of macrophages into the brain, we introduced srp(Hemo)3xmCherry, which enables the labeling of macrophages (Cattenoz et al., 2021), into the sws¹ mutant background. In contrast to control brains, we observed the presence of macrophages within the brain in both developing and adult brains of sws¹; srp(Hemo)3xmCherry mutants (yellow arrowheads, Figure 5—figure supplement 1F and G). Moreover, using the anti-NimC1 antibody (Kurucz et al., 2007), macrophage infiltration into the adult brain was detected in flies with sws downregulation specifically in SPG cells (moody>GFP, sws^RNAi, yellow arrowheads, Figure 5E and F). This suggests that the presence of an inflammatory response in mutants with a compromised BBB is associated with macrophage entry into the brain.

However, while both sws and moody mutants have defective BBB, the nature of these mutations and their involvement in cellular processes are very different. Moody is a GPCR that is expressed in SPGs and localizes to the sites of SJ formation (Babatz et al., 2018; Bainton et al., 2005; Schwabe et al., 2005; Li et al., 2021a). Its cellular function is to control continued cell growth of SPG by differentially regulating actomyosin contractility and SJ organization (Li et al., 2021a). NTE/SWS is a transmembrane ER protein that hydrolyzes phosphatidylcholine and binds to and inhibits the C3 catalytic subunit of protein kinase A (Bettencourt da Cruz et al., 2008). To understand how such different mutations could result in similar outcomes, we first analyzed if moody loss would result in lysosomal material accumulation. Electron microscopy analyses demonstrated that unlike in sws mutant brains, no intracellular accumulations with extracellular material were observed upon moody loss (compare Figure 3B and Bʹ and Figure 5—figure supplement 2F). Furthermore, no accumulation of endosomal-lysosomal pathway components such as Rab7 were detected within SPG cells of moody mutants (Figure 5—figure supplement 2G and H). At the same time, as previously described (Babatz et al., 2018; Bainton et al., 2005; Schwabe et al., 2005; Li et al., 2021a), we observed that in the absence of moody, SJs were formed, but they were disorganized (Figure 5—figure supplement 2F, arrow).

We compared in greater detail the SJ organization in both mutants using a molecular component of SJs, Neurexin IV (NrxIV). In comparison to the wild type, upon sws loss, SJs were not properly assembled and exhibited irregular membrane clusters and disruptions (Figure 6A and C). In contrast, the moody mutant exhibited a frayed SJ phenotype (Figure 6B). Since Moody coordinates the continuous organization of junctional strands in an F-actin-dependent manner, as a result of its loss, SJ strands fail to extend properly during cell growth (Figure 6C). While the role of Moody in SJ formation is understood (Babatz et al., 2018; Bainton et al., 2005; Schwabe et al., 2005; Li et al., 2021a), the mechanism by which NTE/SWS may be involved in this process is unclear. The in-depth examination of cell junctional structures in sws mutants using electron microscopy revealed their improper assembly, characterized by the accumulation of irregular membrane clusters and disruptions in septa organization (Figure 6E and F, yellow arrowheads). Cell junctions are a special type of plasma membrane domain whose transmembrane proteins form a complex, mechanically stable multiprotein structure (Giepmans and van Ijzendoorn, 2009). The lipid component of cell junctions exhibits a typical membrane raft structure (Lee et al., 2008; Shigetomi et al., 2023; Nusrat et al., 2000; Simons and Vaz, 2004; Mühlig-Versen et al., 2005).

The main feature of membrane rafts is that they contain an enriched fraction of cholesterol and sphingolipids and are able to dynamically orchestrate specific membrane proteins involved in cell adhesion, signal transduction, protein transport, pathogen entry into the cell, etc. Since NTE/SWS regulates lipid membrane homeostasis, we hypothesized that it influences the composition of membrane rafts. Analysis of SPG membranes in sws-deficient brains shows abnormal clustering of SJs proteins and disorganized membrane domains, implying that NTE/SWS phospholipase plays a role in organizing SPG membrane architecture (Figure 6C). As lysosomes play a crucial role in lipid catabolism and transport, any disruptions in their function can have repercussions on cellular lipid homeostasis, thereby influencing the composition of membrane rafts. To investigate whether the observed SJ phenotype in sws mutants can be replicated by inducing lysosomal dysfunctions, we downregulated in SPG cells several key lysosomal genes: moody>Dysb^RNAi, moody>Npc1a^RNAi, moody>Pldn^RNAi, and moody>spin^RNAi. Significantly, the downregulation of any of these genes led to abnormal formation of SJs and membrane organization in SPG cells (Figure 6—figure supplement 1A–E). This suggests that the lysosomal control of membrane homeostasis has a significant impact on the appearance of SJs.

In summary, our data show that the phospholipase NTE/SWS plays a crucial role in lysosome biogenesis and organization of the architectural framework of BBB membranes. We propose that since NTE/SWS regulates lipid membrane homeostasis, its loss results in the disruption of membrane rafts, which includes SJs, leading to brain barrier permeability. As a result, the inflammatory response accompanied by the accumulation of FFAs is activated in mutant brains, leading to progressive neurodegeneration that can be alleviated by the use of anti-inflammatory drugs.

Fly stocks were maintained at 25°C on a standard cornmeal-agar diet in a controlled environment (constant humidity and light-dark cycle). As controls OregonR and w¹¹¹⁸ lines were used. The sws¹ mutant and the UAS-sws lines were gifts from Kretzschmar et al., 1997. To obtain sws transheterozygotes, sws¹ and sws⁴, obtained from Bloomington Drosophila Stock Center (BDSC 28121), mutant alleles were used. To express transgenes in an sws-dependent manner, an sws driver line (sws-Gal4), obtained from the Kyoto Stock Center (104592), was used. y* w* P{GawB}sws^NP4072/FM7c line was created using the strategy of the Gal4 enhancer trap element P{GawB} insertion (Brand and Perrimon, 1993; Hayashi et al., 2002). To define an expression pattern of the driver lines, a UAS-nlsLacZ, UAS-CD8::GFP transgenic line, kindly donated by Frank Hirth, was used. To induce human NTE gene expression, a UAS-hNTE transgenic line (kindly donated by Robert Wessells) was used. To downregulate sws expression, UAS-sws^RNAi (BDSC 61338) was used. Glia-specific Gal4 driver lines – repo-Gal4, UAS-CD8::GFP/TM6B, Gliotactin-Gal4, UAS-CD8::GFP, and moody-Gal4, UAS-CD8::GFP – were gifts from Mikael Simons. A neuronal Gal4 driver, nSyb-Gal4 was obtained from BDSC (BDSC 51945). In addition, to phenocopy sws loss-of-function in the nervous system, a double driver line was generated (repo-Gal4, nSyb-Gal4, UAS-CD8::GFP/TM6B, Sb), which allowed the expression of the transgenes in both neuronal and glial cells. To induce sws downregulation in glia after the BBB was formed, we used tub-Gal80^ts;repo-Gal4/TM6B driver line. The moody^ΔC17 mutant was a gift from Christian Klämbt. To downregulate moody expression, UAS-moody^RNAi (BDSC 66326) was used. UAS-Dysb^RNAi (BDSC67316), UAS-Npc1a^RNAi (BDSC37504), UAS-Pldn^RNAi (BDSC67884), UAS-spin^RNAi (BDSC27702) lines were used to analyze SJs of lysosomal storage mutants. srp(Hemo)3xmCherry line (kindly donated by Angela Giangrande) was used to analyze the macrophage entry through the BBB.

For analysis of adult brain morphology, 7 μm paraffin-embedded sections were cut from fly heads. To prepare Drosophila brain sections, the fly heads were immobilized in collars in the required orientation and fixed in Carnoy fixative solution (6:3:1=ethanol:chloroform:acetic acid) at 4°C overnight. Tissue dehydration and embedding in paraffin was performed as described previously (Kucherenko et al., 2010). Histological sections were prepared using a Hyrax M25 (Zeiss) microtome and stained with hematoxylin and eosin as described previously (Shcherbata et al., 2007). All chemicals for these procedures were obtained from Sigma-Aldrich.

Fly brains of 1- and 15-day-old animals were dissected in 1× phosphate buffered saline (1× PBS) and then fixed in 4% formaldehyde diluted in 1× PBS for 20 min at room temperature. Next, brains were washed with PBT (0.2% Triton X-100 in 1× PBS) four times, followed by block with PBTB (2 g/L bovine serum albumin, 5% normal goat serum, 0.5 g/L sodium azide) for 1 hr at room temperature and then incubated at 4°C with primary antibodies diluted in PBTB on nutator overnight. The following day, samples were washed with 1× PBT four times followed by block for 1 and 2 hr incubation with secondary antibodies at room temperature. Next, samples were washed four times with PBT (one of the washes contained DAPI to mark nuclei). Lastly, medium (70% glycerol, 3% n-propyl gallate in 1× PBS) was added to samples for later mounting on the slides. The following primary antibodies were used: mouse anti-Repo (1:50), mouse anti-CoraC (1:50), and mouse anti-Rab7 (1:50), rat anti-DE-Cadherin (1:50) from the Developmental Studies Hybridoma Bank (DSHB); chicken anti-GFP (#ab13970, 1:1000) and rabbit anti-mCherry (#ab167453, 1:1000) from Abcam; mouse Anti-β-Galactosidase (#Z3781, 1:200) from Promega; rabbit anti-SWS (1:1000 from Doris Kretzschmar); mouse anti-CathepsinL (#1515-CY-010, 1:400) from R&D Systems; rabbit anti-NrxIV (1:1000 from Christian Klämbt); mouse anti-NimC1 (1:300 from István Andó). The following secondary antibodies were used: goat anti-chicken Alexa 488 (1:500), goat anti-rat Alexa 488 (1:500), goat anti-rat Alexa 647 (1:500), goat anti-rabbit Alexa 488 (1:500), and goat anti-rabbit Alex 568 (1:500) from Thermo Fisher Scientific; goat anti-mouse IgG2a Cy3 (1:400), goat anti-mouse IgG1 647, and goat anti-mouse IgG1 Cy3 (1:500) from Jackson ImmunoResearch Laboratory. For visualization of cell nuclei, DAPI dye was used (1:1000, Sigma). Samples were analyzed using a confocal microscope (Zeiss LSM 700). For making figures, Adobe Photoshop software was used.

Total RNA was extracted from 15- and 30-day-old fly brains using Trizol reagent (Invitrogen) following the manufacturer’s protocol. To detect sws mRNA, the forward and reverse primers AGACATACGCCGTGAATACCG and GCGACGACTGTGTGGACTTG, respectively, were used. To detect expression of innate immunity factors, the following forward and reverse primers were used: Attacin A (forward and reverse primers CACAACTGGCGGAACTTTGG and AAACATCCTTCACTCCGGGC, respectively), Cecropin A (forward and reverse primers AAGCTGGGTGGCTGAAGAAA and TGTTGAGCGATTCCCAGTCC, respectively), and Diptericin (forward and reverse primers TACCCACTCAATCTTCAGGGAG and TGGTCCACACCTTCTGGTGA, respectively). As an endogenous control for qPCRs, Ribosomal Protein L32 (RpL32) with the following forward and reverse primers AAGATGACCATCCGCCCAGC and GTCGATACCCTTGGGCTTGC, respectively, was used. The threshold cycle (CT) was defined as the fractional cycle number at which the fluorescence passes a fixed threshold. The ΔCT value was determined by subtracting the average RpL32 mRNA CT value from the average tested CT value of target mRNA, correspondingly. The ΔΔCT value was calculated by subtracting the ΔCT of the control sample from the ΔCT of the experimental sample. The relative amounts of miRNAs or target mRNA is then determined using the expression 2^−ΔΔCT.

Flies were injected into the abdomen with a solution containing 10 kDa dextran dye labeled with Texas Red (#D1864) from Molecular Probes. Flies were then allowed to recover for more than 12 hr before the dissection, followed by the analysis for dextran dye presence in the brain. Fly heads of 15-day-old animals were dissected in 1× PBS and then fixed in 4% formaldehyde diluted in 1× PBS for 1 hr at room temperature. Then fly brains were dissected in 1× PBS and fixed in 4% formaldehyde diluted in 1× PBS for 20 min at room temperature. Next, brains were washed with PBT (0.2% Triton X-100 in 1× PBS) four times, followed by block with PBTB (2 g/L bovine serum albumin, 5% normal goat serum, 0.5 g/L sodium azide) for 1 hr at room temperature and then washed two times with PBT (one of the washes contained DAPI to mark nuclei). Lastly, medium (70% glycerol, 3% n-propyl gallate in 1× PBS) was added to samples for later mounting on the slides.

TUDCA (#580549), 4-PBA (#567616), valsartan (#PHR1315), fenofibrate (#F6020), sodium salicylate (#S3007), rapamycin (#R0395), deferoxamine mesylate salt (#D9533), liproxstatin-1 (#SML1414), and sphingosine (#860025P) from Sigma-Aldrich were added to 5% glucose solution at a final concentration shown by Supplementary file 5. Then, these glucose-dissolved components were fed by micropipettes to the flies that were kept for 14 days on a diet food without any sugar. To visualize the uptake of chemicals, solutions were also colored by 2.5% wt/vol of Brilliant Blue (#80717) from Sigma-Aldrich.

Accurately weighed heads of 15-day-old flies were treated with 1000 µL aliquots of acetonitrile in autosampler glass vials (1.8 mL), the samples were sealed and vortexed several times and then stored in a refrigerator overnight (4°C). Next day, the samples were warmed up to room temperature and centrifuged (10 min, 3345×g, 4°C). Aliquots (950 µL) of the clear supernatants were decanted carefully transferred to autosampler glass vials (1.8 mL). The samples were spiked with 10 µL aliquots of a 1000 µM stock solution of sterculic acid (C19H34O2; 10 nmol; 8-cyclopropen-octadecenoic acid corresponds to C19:1) which served as the internal standard (IS) for all FFAs. The solvent was evaporated entirely under a stream of nitrogen gas. The solid residues were reconstituted in anhydrous acetonitrile (100 µL). Then, 10 µL Hünig base (N,N-diisopropylethylamine) and 10 µL 33 vol% pentafluorobenzyl (PFB) bromide in anhydrous acetonitrile were added. Subsequently, the FFAs were derivatized by heating for 60 min at 30°C to generate the PFB esters of the FFAs. Solvents and reagents were evaporated to dryness under a stream of nitrogen gas. The residues were treated with 1000 µL aliquots of toluene and the derivatives were extracted by vortex-mixing for 120 s. After centrifugation (10 min, 3345×g, 4°C), 300 µL aliquots of the clear and colorless supernatants were transferred into microvials placed in autosampler glass vials (1.8 mL) for GC-MS analysis. A standard control sample containing 1 mL acetonitrile, 1 µL 10 mM arachidonic acid (C20:4, 10 nmol), and 10 µL 1 mM IS (10 nmol) was derivatized as described above for the fly samples after were evaporated to dryness under a stream of nitrogen gas. After centrifugation (10 min, 3345×g, 4°C), 100 µL of the clear and colorless supernatant were transferred into an autosampler glass vial (1.8 mL), diluted with toluene (1:10, vol/vol), and subjected to GC-MS analysis as described below.

GC-MS analyses were performed on a GC-MS apparatus consisting of a single quadrupole mass spectrometer model ISQ, a Trace 1210 series gas chromatograph, and an AS1310 autosampler from Thermo Fisher (Dreieich, Germany). A fused-silica capillary column Optima 17 (15 m length, 0.25 mm ID, 0.25 µm film thickness) from Macherey-Nagel (Düren, Germany) was used. Aliquots of 1 µL were injected in the splitless mode. Injector temperature was kept at 280°C. Helium was used as the carrier gas at a constant flow rate of 1.0 mL/min. The oven temperature was held at 40°C for 0.5 min and ramped to 210°C at a rate of 15°C/min, and then to 320°C at a rate of 35°C/min. Interface and ion-source temperatures were set to 300°C and 250°C, respectively. Electron energy was 70 eV and electron current 50 µA. Methane (constant flow rate of 2.4 mL/min) was used as the reactant gas for negative-ion chemical ionization. The electron multiplier voltage was set to 1300 V. Authentic commercially available reference compounds were used to determine the retention times of the derivatives and to generate their mass spectra. The selected ions [M−PFB] − with mass-to-charge (m/z) ratios and retention times of the derivatives are summarized in Supplementary file 6b. Quantitative measurements were performed by selected-ion monitoring (SIM) of the ions listed in Supplementary file 6bwith a dwell time of 50 ms and SIM width of 0.5 amu for each ion in three window ranges. The results of the GC-MS analyses of the control standard sample that contained 10 nmol arachidonic acid and 10 nmol IS are summarized in Supplementary file 6c. The highest peak area ratio of FFA to the internal standard (FFA/IS) was obtained for arachidonic acid (0.098). This in accordance with the ratio observed in the standard curve (Figure 5—figure supplement 3A). A lower FFA/IS was obtained for palmitic acid (0.026). As palmitic acid was not externally added to the control standard sample, it is assumed that palmitic fatty acid is ubiquitously present as a contamination in the laboratory materials. An FFA/IS value of 0.027 was obtained for a fatty acid, which co-elutes with nonadecanoic acid (C19:0). As this fatty acid was not externally added to the control standard sample nor it is expected to be a laboratory contamination, it can be hypothesized that it is a contamination in the commercially available preparation of the IS which is a quasi C19:0 fatty acid. The FFA/IS values of the other FFAs are remarkably lower (<0.0065), which suggest that they cannot be considered as appreciable contaminations (Figure 5—figure supplement 3B).

After dissection, brains of 15-day-old flies were fixed overnight immediately by immersion in 150 mM HEPES containing 1.5% glutaraldehyde and 1.5% formaldehyde at pH 7.35. Preparation for transmission electron microscopy was done as described (Mariani et al., 2022). Imaging was done in a Zeiss EM 900 at 80 kV, equipped with a side-mount CCD camera (TRS).

To analyze the activation of the inflammatory response, real-time qPCR analysis of AMPs mRNA levels from heads of each genotype was performed. AVE ± SEM was calculated. The experiments were performed in at least two biological replicates for each genotype. Two-tailed Student’s tests were used to test for statistical significance (*p<0.05, **p<0.005, ***p<0.001, see Supplementary file 1).

To analyze the frequency of brain hemispheres with defective brain surfaces, Z-stack confocal images of the entire adult brain were captured. The brain surface was identified by CoraC expression. The numbers of brain hemispheres exhibiting a normal brain surface, those containing lesions, or those with both lesions and membrane clusters on the brain surface were quantified. For the comparison of observed phenotypes, two-way tables and chi-squared tests were used (*p<0.05, **p<0.005, ***p<0.001, see Supplementary file 2).

To assess the frequency of brain hemispheres with the accumulation of Rab7-positive or CathepsinL-positive structures in surface glia, Z-stack confocal images of the entire adult brain were captured. The surface glia were identified by moody-Gal4, UAS-CD8::GFP expression. The numbers of brain hemispheres with Rab7 or CathepsinL accumulation in the surface glia were quantified. For the comparison of observed phenotypes, two-way tables and chi-squared tests were used (*p<0.05, **p<0.005, ***p<0.001, see Supplementary file 3).

To analyze the frequency of brain hemispheres with a permeable BBB, Z-stack confocal images of the entire adult brain were captured. The permeable BBB was identified by 10 kDa dextran dye labeled with Texas Red localization inside the fly brain. The numbers of brain hemispheres with a permeable BBB were quantified. All experiments were performed in at least three biological replicates for each genotype. For the comparison of observed phenotypes, two-way tables and chi-squared tests were used (*p<0.05, **p<0.005, ***p<0.001, see Supplementary file 4).

To analyze the frequency of brain hemispheres with defective brain surfaces in in vivo Drosophila treatment assays, the brain surface was identified by CoraC expression. The numbers of brain hemispheres with formed lesions and membrane clusters on the brain surface were quantified. The reduction in the percentage of the glial phenotype, assayed by CoraC expression pattern in sws and moody mutants treated with different chemicals compared to untreated mutants, was quantified. All experiments were performed in at least three biological replicates for each genotype. For the comparison of observed phenotypes, two-way tables and chi-squared tests were used (*p<0.05, **p<0.005, ***p<0.001, see Supplementary file 5).

To analyze the changes in FFA levels in fly mutant heads, GS-MS measurements of FFAs were performed. One-way ANOVA tests were used for statistical analysis (*p<0.05, **p<0.005, ***p<0.001, see Supplementary file 6).