The CD36 scavenger receptor Bez regulates lipid redistribution from fat body to ovaries in Drosophila

Scavenger receptors (SR) are a family of transmembrane proteins that are structurally diverse and have a wide range of ligands. They were first discovered in macrophages as receptors for oxidized Low Density Lipoprotein (oxLDL) (Goldstein et al., 1979; Brown et al., 1980). Further studies have shown that SR act as multifunctional receptors implicated in a range of cellular processes. In addition to modified lipoproteins such as oxLDL, many other endogenous, mainly polyanionic, ligands have been identified, such as long chain fatty acids, oxidized phospholipids and extracellular matrix proteins (Rigotti et al., 1995; Silverstein and Febbraio, 2009). SR can be sorted into eight different classes depending on their predicted structure, which is determined by the presence of different conserved motifs (Krieger, 1997; Plüddemann et al., 2007). Class B scavenger receptors (SR-B) are characterized by a hairpin structure, formed by two transmembrane domains and one extracellular loop with multiple glycosylation sites. The extracellular CD36-domain includes numerous ligand-binding sites (Canton et al., 2013) and its sequence is highly conserved between the different members of this SR class. Its tertiary structure is defined by disulfide bonds between conserved cysteine residues, which are essential for the recognition of ligands (Gruarin et al., 1997). By contrast, the sequences of the relatively short intracellular N- and C-terminal tails are not conserved. These tails interact with cytosolic proteins and regulate cellular signaling. SR-B can bind to a wide range of ligands, such as lipoproteins, long chain fatty acids and oxidized phospholipids. In contrast to the other SR classes, they are also able to bind to native lipoproteins such as LDL and high density lipoprotein (HDL) and thereby contribute to the regulation of lipid metabolism. Vertebrate SR-Bs include the proteins Cluster of differentiation 36 (CD36), Scavenger receptor class B member I (SRBI) and Lysosomal integral membrane protein II (LIMP-II). CD36 and SRBI are located at the plasma membrane, whereas LIMP-II is a protein of the lysosomal membrane with a function in the import of glucocerebrosidase (Heybrock et al., 2019).

CD36 is a multifunctional protein that regulates the uptake and export of fatty acids in gut cells, adipocytes and muscle cells (Daquinag et al., 2021). It is also expressed in taste bud cells, where it acts as a lipid sensor for the detection of dietary lipids (Laugerette, 2005). Fatty acid transfer from adipocytes to ovarian tumors has been shown to depend on the CD36-dependent lipid uptake by adipocytes (Ladanyi et al., 2018). CD36 is also required for ovarian angiogenesis and folliculogenesis: CD36-deficient mice display hypervascularized ovaries and have fewer offspring than controls (Osz et al., 2014). However, this function has been linked to binding of thrombospondin 1, an angiogenesis regulator, and not to the lipid binding property of CD36.

In the invertebrate Drosophila melanogaster, 14 SR-B homologs have been identified (Nichols and Vogt, 2008). The high number of SR genes suggests a diversification of functions in insects. Many of these genes remain uncharacterized until now, but some have been functionally analyzed. Croquemort (Crq) is expressed in macrophages and takes part in the phagocytosis of apoptotic cells during embryogenesis and the morphogenesis of the central nervous system (Guillou et al., 2016). Debris buster (Dsb) is a protein of the lysosomal membrane that is required for trafficking of extracellular matrix components important for airway physiology and for clearing of degenerating dendrites (Han et al., 2014; Wingen et al., 2017). Neither inactivation nor afterpotential-D (NinaD) and Scavenger receptor acting in neural tissue and majority of rhodopsin absent (Santa-maria) were identified as transporters for carotenoids, precursors for the synthesis of vitamin A, which itself is modified and then embedded in the optical pigment Rhodopsin. NinaD is expressed in the gut, where it mediates the uptake of carotenoids, which are required for visual chromophore synthesis. Santa-Maria is expressed in specific neurons and glia cells and locates to the plasma membrane, where it mediates the uptake of β-carotene that afterwards is converted to vitamin A (Dewett et al., 2021). Sensory neuron membrane protein 1 (SNMP1) is expressed in olfactory neurons and in steroidogenic organs such as the ovaries, testes and the prothoracic gland (PG). Silencing of SNMP1 via RNAi expression in the PG and in the follicle cells of the ovary results in a reduced content of lipid droplets in these tissues (Talamillo et al., 2013).

Lipid shuttling in both human blood and Drosophila hemolymph is mediated by lipoproteins. In Drosophila, apoLpp and apoLTP belong to the apoB family of apolipoproteins. The product of the apoLpp gene (CG11064) gives rise to apoLI and apoLII that assemble the Lipophorin lipoprotein (Lpp). Human APOB is the major lipoprotein in LDL particles. Lpp binds to Lipid transfer protein (LTP) in the gut and is loaded with dietary lipids. These are distributed to the fat body for storage and to other organs (Palm et al., 2012). The Drosophila fat body stores lipids in lipid droplets from where they are mobilized, e.g. upon starvation, to nurse other tissues. Fasting induces a morphological change of fat body lipid droplets as they become larger and fewer when lipids are mobilized (Ugrankar et al., 2019).

Nutrition is closely linked to reproduction. Malnutrition in women heavily impacts reproduction, with both obesity and undernutrition affecting fertility (Silvestris et al., 2019). This link also exists in Drosophila, and Drosophila females store high amounts of the neutral lipid triglyceride (TG) to support oocyte maturation. Germline ablation leads to increased lipid storage (Rodrigues et al., 2023). Uptake of neutral lipids takes place between stages 8 and 10 of oocyte maturation: at stage 10, egg chambers massively accumulate neutral lipids in both nurse cells and follicle cells. Lipoproteins mediate the transfer of lipids from the hemolymph to ovaries (Sieber and Spradling, 2015). In ovaries, Lpp receptors bind LTP, which is required for neutral lipid accumulation (Rodríguez-Vázquez et al., 2015). The metabolic signaling that allows the transfer of storage lipids from adipocytes to ovaries is still incompletely understood. Here, we identify CG3829 as one of the few of the 14 Drosophila SR-B that contain a conserved lipid binding residue homologous to CD36. We show its role for the export of dietary lipids from the fat body to the circulation. We named the protein Bez (Between Emp and Zip, pronounced ‘Beth’) and found that in Bez mutants, ovaries degenerate owing to missing lipid transfer from the fat body. Bez mutant adipocytes accumulate lipids and are unable to mobilize TG, and Bez-mediated lipid export from the fat body is required to allow lipid accumulation during oocyte maturation. We demonstrate that Bez transfers labeled lipids to the lipoprotein Lpp, thereby distributing storage lipids to ovaries. We thus establish a role for this SR-B in the metabolic regulation of reproduction.

Bez is one of 14 CD36-like protein family members in Drosophila. We performed an AlphaFold analysis of its predicted protein structure (Fig. 1A). Similar to the human CD36-encoded protein platelet glycoprotein 4, CG3829 contains two transmembrane domains and a large extracellular barrel-like putative lipid binding domain with high sequence similarity to CD36 (Fig. S1). Of note, Bez is the only fly SR-B that contains two conserved lysine residues of the hydrophobic pocket that, in mammalian CD36, has been linked to lipid uptake (Kuda et al., 2013). The SR-B Peste (CG7228) and Santa-Maria contain one of these lysine residues, which are absent in the other Drosophila SR-B. In addition, Bez contains two disorganized amino acid stretches at both termini. We analyzed the expression pattern of the Bez protein in embryos and detected expression at the apical membrane of the embryonic gut (Fig. S2A) and at the yolk sac (Fig. S2B). Strong expression was also seen in the fat body (Fig. 1B-D). In the embryonic fat body (Fig. 1B), marked by the lipid droplet surface protein Perilipin-1 (Plin1, also known as Lsd-1; Beller et al., 2010), Bez was found primarily at the plasma membrane of adipocytes. To confirm this localization, we generated bez-RNAi clones in the fat body. Wild-type cells (GFP-negative) showed high Bez expression at the plasma membrane (illustrated by colocalization with the plasma membrane protein α-Spectrin), whereas bez-RNAi clones (marked by the GFP signal) lacked Bez staining (Fig. 1C, asterisks). Bez also localized to the plasma membrane of the adult fat body adipocytes, marked by Plin1 (Fig. 1D). Taken together, the Bez protein structure and expression pattern suggest a conserved function in lipid binding similar to that of CD36.

To investigate the molecular function of Bez, we used a mutant in which the P-element P(EP)CG3829^G8378 was inserted into the open reading frame (mutant allele bez^EP). By mobilizing this EP element we generated a second allele (bez^jo2) featuring a stop codon in the sequence encoding for the first transmembrane domain at the position of amino acid 89 (Fig. 2A). The resulting Bez mutants (bez^EP/jo2, bez^−/−) did not express the Bez protein in embryos (Fig. S3). Bez homozygous mutants are semi-viable, but females are sterile, and larvae show a strong developmental delay: 50% of heterozygous controls pupariated after ∼4.5 days at 25°C, whereas 50% of homozygous Bez mutants pupariated after almost 6 days after egg laying (AEL). Similarly, adult flies hatched ∼1.5 days later than the heterozygous and wild-type controls (Fig. 2B). Expression of the UAS-Bez construct in the fat body (cg-Gal4) reversed the developmental delay of Bez mutants (Fig. 2B, ‘rescue’). Next, we expressed RNAi against Bez in different tissues. Expression of bez-RNAi in the midgut (mex-Gal4) or in hemocytes (hml-Gal4) did not phenocopy the developmental delay of Bez homozygous mutants. By contrast, expression in the fat body (cg-Gal4) fully phenocopied the developmental delay compared with its control (cg-Gal4>w, Fig. 2C). This shows that Bez function in the fat body is essential for normal development.

To analyze the cause for female sterility of Bez mutants, we investigated Bez mutant fecundity, egg morphology, and ovary morphology. Drosophila ovaries consist of parallel tubes, the ovarioles, that contain developing egg chambers of advancing developmental stages. Egg chambers consist of nurse cells, oocyte and a surrounding layer of follicle cells. Up to stage 9, the nurse cells occupy ∼75% of the space in the egg chamber, whereas in stage 10 egg chambers, the oocyte grows to ∼50% of the egg chamber (Hudson and Cooley, 2014). We found that Bez mutants laid fewer and smaller eggs and that these eggs collapsed, with no larvae hatching from them (Fig. 3A). We dissected ovaries of Bez mutant females and found that the ovaries were small and degenerated compared with those of control females (Fig. 3B). We then asked whether Bez was expressed in ovaries and performed quantitative real-time PCR of RNA isolated from whole flies and from isolated ovaries (Fig. 3C). We found that Bez was not expressed in ovaries, suggesting that Bez functions remotely from the fat body to sustain ovary maturation. We analyzed oogenesis by immunofluorescent staining of dissected ovaries and found that oocyte maturation during oogenesis was blocked in Bez mutants: although control ovarioles contained egg chambers of all stages including mature eggs, Bez mutant ovarioles contained no mature eggs and no egg chambers older than stage 10. Localization of Orb, an RNA-binding protein that establishes polarity during oogenesis (Lantz et al., 1994), was broader and less defined in the degenerated egg chambers of Bez mutants (Fig. 3D). Degenerated egg chambers were apoptotic and showed fragmented DNA, as shown by the fragmented DAPI staining (indicated by x). We asked whether Bez knockdown in specific tissues would be sufficient to phenocopy the low fecundity of Bez mutants. We found that expression of bez-RNAi in the fat body (FB-Gal4 or cg-Gal4) as well as ubiquitous expression (c355-Gal4) was sufficient to block fecundity (Fig. 3E). Expression of bez-RNAi in the fat body (FB-Gal4 or cg-Gal4) was also sufficient to block oocyte maturation and to induce egg chamber apoptosis, similar to Bez mutants (Fig. 3F; Fig. S4A,B). Our results show that Bez function in adipocytes is essential for oocyte maturation and thereby for female fecundity.

Egg chambers display a shift in their lipid content between stages 8 and 10. Stage 10 egg chambers store high amounts of neutral lipids in oocyte and nurse cells (Sieber and Spradling, 2015). To investigate whether defects in lipid storage cause ovary degeneration in Bez mutants, we stained neutral lipids with BODIPY. We confirmed that stage 10 egg chambers massively take up storage lipids in heterozygous controls (Fig. 3F, bez^+/−; stage 8-9 egg chambers are marked with asterisks, stage 10 egg chamber is marked with an arrow). Degenerated oocytes of homozygous Bez mutants did not contain increased amounts of neutral lipids (Fig. 3F, bez^−/−). We quantified the fluorescence intensity of BODIPY in egg chambers between stages 8 and 10, which showed that Bez homozygous mutant egg chambers did not increase their lipid content (Fig. 3H). This missing lipid uptake was phenocopied by expression of bez-RNAi in the fat body (cg-Gal4, Fig. 3G,H). By contrast, the stage 8 to stage 10 lipid transition was independent from midgut expression of Bez: midgut depletion of Bez allowed for lipid accumulation and did not trigger egg chamber degeneration (mex-Gal4, Fig. 3G,H). Next, we used a colorimetric assay to measure TG of whole male flies. Fat body-specific depletion of Bez by RNAi led to significantly reduced total TG content compared with controls (Fig. 4A).

We hypothesized that Bez is required for lipid mobilization from adipocytes and that, under Bez loss of function, impaired lipid mobilization leads to insufficient lipid supply to egg chambers. We thus asked whether Bez mutants are unable to mobilize lipids upon starvation. Control female flies can survive for up to 4 days without food, with 50% of flies surviving after 76 h of starvation. Male flies are less starvation-resistant, with 50% of flies surviving after 40 h of starvation (Fig. 4B). Both female and male Bez mutants are significantly less starvation-resistant than controls. To investigate whether acute depletion of Bez from the fat body is sufficient to trigger starvation sensitivity, we expressed bez-RNAi under the control of the temperature-sensitive Gal80 system in females. At 18°C, Gal80 expression suppresses Gal4 and thus the expression of bez-RNAi. At 29°C, Gal80 no longer represses Gal4, allowing for the expression of bez-RNAi in the fat body. The resulting acute depletion of Bez in adipocytes increased starvation sensitivity compared with controls (Fig. 4C). Next, we repeated the experiment and measured TG levels in whole male and female flies. At 18°C, the TG content was reduced in response to nutrient restriction after 6 h in both male and female flies, although not quite significantly (Fig. 4D′). At 29°C, Bez depletion in adipocytes reduced the organismal TG content in both male and female flies (Fig. 4D″), similar to the result in Fig. 4A. Control flies reduced their TG content in response to nutrient restriction. By contrast, TG levels in male (Fig. S5A) and female (Fig. 4D″) flies with Bez depletion in adipocytes remained at similar levels under starvation. This shows that acute and adipocyte-restricted depletion of Bez impairs lipid mobilization upon starvation.

We hypothesized that impaired lipid mobilization in Bez mutants would affect lipid droplet morphology. We used larval fat body tissue of control animals and Bez mutants and stained lipid droplets with BODIPY. The lipid droplet size was semi-quantified as the BODIPY-stained area (Fig. 5A,B). Bez mutant fat body tissue contained significantly enlarged lipid droplets; however, their number per cell remained unchanged. Of note, this matched the lipid droplets phenotype in adipocytes of loss-of-function mutants of Lpp, a Drosophila lipoprotein homologous to APOB (Palm et al., 2012). We repeated the experiment in Bez mutant fat body tissue with Bez reconstituted in distinct clones allowing for direct comparison of lipid droplet morphology within the same tissue (Fig. S6A,B). Lipid droplet staining confirmed that these organelles are large in Bez mutant adipocytes, whereas in adjacent rescue clones the droplets are smaller. Similarly, bez-RNAi leads to larger lipid droplets (Fig. S6C). To rule out overexpression effects, we used the UAS-Bez rescue construct in a wild-type background, where it did not alter lipid droplet morphology (Fig. S6D).

Next, we asked whether the difference in lipid droplet morphology in Bez mutant fat body tissue corresponds to differences in lipid content. We determined the total lipid content of isolated fat body tissue by mass spectrometry and found that Bez mutant tissue showed lipid accumulation compared with controls (Fig. 5C). Bez mutants accumulated lipids of both types, the neutral lipids [diglyceride (DG), TG] and phospholipids [phosphatidylethanolamine (PE), phosphatidylcholine (PC); Fig. S6E]. To investigate how Bez affects uptake and incorporation of fatty acids into the lipidome of the fat body, we incubated isolated tissue of control flies and Bez mutants with alkyne-labeled oleic acid (alkyne oleate, FA 19:1;Y), a traceable analog of the abundant natural oleic acid (Kuerschner and Thiele, 2022). Upon click-reaction a mass spectrometry lipidome analysis was performed (Wunderling et al., 2021). Bez mutants showed a trend of increased total uptake and incorporation of alkyne oleate into the TG pool generating labeled TG;Y (Fig. 5D), suggesting that Bez might play a role in determining lipid biosynthetic routes. Levels of labeled phospholipids (PE;Y and PC;Y) were also elevated in the mutant tissue, but our measurements did not reach significance. Taken together, these data show that Bez mutant adipocytes enhanced lipid uptake and storage, but were unable to mobilize these lipids. This suggests that lipid export from adipocytes to target tissues (such as ovaries) is impaired in Bez mutants.

How does Bez affect the lipid content of egg chambers without being present in ovaries? Lipids are transported between organs by lipoproteins in both Drosophila and humans. A homolog of APOB is apoLpp, an apolipoprotein contained in Lipophorin (Lpp). We used a fly line that expresses Lpp fused to GFP under the control of its endogenous promoter. We first analyzed whether Bez colocalized with Lpp at the adipocyte plasma membrane. Using Airyscan super-resolution microscopy, we found that both Bez and Lpp were organized in distinct punctate domains at the adipocyte plasma membrane, several of which colocalized (Fig. 6A; Fig. S7A). Next, we asked whether Bez has an impact on the lipid composition of the lipoprotein particles. We used freshly isolated hemolymph lipoproteins from Lpp-GFP reporter animals of wild-type or Bez mutant background and analyzed their lipid content by mass spectrometry. In hemolymph lipoproteins, DG is the abundant lipid class (Palm et al., 2012). To test lipoprotein enrichment in our preparation, we compared the DG to TG ratio of hemolymph lipoproteins and fat body tissue. We found that our preparation had a DG to TG ratio of ∼20, compared with ∼0.02 in isolated fat bodies (Fig. S7B), suggesting low contamination from e.g. lipid droplets. Although similar in protein content (Fig. 6B), particles from Bez mutants showed a significantly reduced lipid content (Fig. 6C). We then assessed the distribution of Lpp in the gut and fat body. The apical membrane of the gastric caeca, four tubular structures of the anterior midgut, was marked by Lpp-GFP in control animals, whereas in Bez mutants this membrane showed an even more intense signal (Fig. 6D). In the fat body, Lpp-GFP was found in a regular pattern at the plasma membrane of adipocytes, similar to the Bez protein (Fig. 6E). In the absence of Bez, this regular pattern was lost and Lpp-GFP presented in aggregates at the plasma membrane.

As Bez localizes to the plasma membrane of adipocytes it becomes exposed to hemolymph and lipoprotein particles. To corroborate the interaction of Lpp from the hemolymph with the membrane receptor Bez, we isolated hemolymph from Lpp-GFP-expressing flies for incubation with Bez mutant fat body tissue that contained Bez rescue clones (Fig. 7A). In fat bodies incubated with hemolymph from wild-type control flies, only background staining was visible in the GFP channel (Fig. 7A, upper panels). Similarly, Bez mutant cells incubated with hemolymph from Lpp-GFP flies showed only minor signal for GFP (Fig. 7A, lower panels). By contrast, the rescue clones showed a strong staining for Lpp-GFP, demonstrating a Bez-dependent recruitment of Lpp to the plasma membrane of adipocytes.

Does lipid transfer from adipocytes to Lpp depend on Bez? To address this question, we used the oleate analog alkyne oleate (FA19:1;Y) and traced its transfer from fat body tissue to Lpp. To this end, we incubated fat body tissue from control and Bez mutant animals with the tracer, yielding labeled lipid metabolites. After washing, the tissue was incubated for 60 min with naïve Lpp before separation of tissue and Lpp. The particles were washed and analyzed by mass spectrometry to identify their content of labeled lipids (Fig. 7B). Lipid transfer is represented by a red line between the median measurements from fat body and Lpp samples (Fig. 7C). From all labeled lipids, ∼27% were transferred onto Lpp from control adipocytes. As shown before (Fig. 5D), Bez mutant adipocytes readily incorporated alkyne oleate into their lipidome (Fig. 7C). In the samples obtained from Bez mutant tissue, we detected one outlier that showed higher amounts of alkyne oleate in the Lpp sample than in the fat body samples. By contrast, all other samples showed no transfer of the tracer onto Lpp. As shown in Fig. 5D, incorporation of the tracer into DG is not significantly altered between control and Bez mutant adipocytes. Naïve Lpp incubated with control tissue acquire ∼33% of labeled DG;Y, whereas no labeled DG;Y was transferred from Bez mutant adipocytes (Fig. 7D). Note that the outlier from the Bez mutant sample, as well as the highest value from the wild-type control sample, were omitted from the analysis but are shown in Fig. S7C. Taken together, these experiments show that Bez is required for productive lipid transfer from fat body to Lpp.

In sum, we suggest that Bez is required for lipid export from adipocytes, and that by interaction with lipoproteins it regulates the distribution of lipids to ovaries, where these lipids are required for oocyte maturation (Fig. 7E). We thus demonstrate a role for a previously uncharacterized scavenger receptor in the remote metabolic regulation of ovary development.

The CD36 scavenger receptor plays a crucial role in fatty acid transport. CD36 is required to incorporate dietary lipids into chylomicrons, which allows the transfer of fatty acids from the gut into the circulation. Fatty acids transported by LDL are taken up into adipose tissue by LDL receptors, and mobilization of lipids from adipocytes again requires CD36 (Daquinag et al., 2021). However, whether CD36 interacts with lipoproteins in this process is yet unresolved. CD36 is also required for adipogenesis, and patients with a CD36 deficiency show impaired chylomicron formation, reduced lipid utilization and storage, and increased lipolysis (Zhao et al., 2018). Here, we describe a novel role for Bez, a largely uncharacterized CD36-like lipid scavenger receptor in lipid export from adipocytes of Drosophila melanogaster. We show that, in the absence of Bez, lipids accumulate in the fat body, the major lipid-storing organ of Drosophila with functional homology to liver and adipose tissue. Lipids are trapped in Bez mutant adipocytes, as becomes apparent by enlarged lipid droplets and by an inability to mobilize lipids in response to starvation. Lipid export from storing tissue is not only important to catabolize lipids for energy gain in times of nutrient scarcity, but also serves to nurse target tissues such as imaginal discs, ovaries and the brain. The requirement of lipids for oocyte development becomes apparent during maturation from stage 8 to stage 10, when egg chambers accumulate neutral lipids (Sieber and Spradling, 2015). Bez mutants are viable, but females are sterile and lay very few, collapsed eggs. Bez mutant ovaries degenerate, which can be seen by the absence of mature eggs and the presence of fragmented DNA from approximately stage 10 onwards, as well as missing neutral lipid accumulation. Of note, it is sufficient to deplete Bez from the fat body to reduce fecundity and induce starvation sensitivity. This highlights the importance of storing lipids in for further distribution, as dietary lipids directly from the gut cannot substitute. Bez is not expressed in ovaries, which precludes a direct role for Bez in ovary lipidation, but further underlines its role in remote lipid dispatch from the fat body.

Lipid transport through hemolymph is mainly mediated by Lpp, a lipoprotein with APOB homology. APOB is the characteristic apolipoprotein in human chylomicrons and LDL (Olofsson and Borèn, 2005). CD36 can interact with lipoproteins: an important example is the role of CD36 as a receptor for oxLDL on macrophages. Upon binding of oxLDL to macrophage CD36 in a blood vessel, the macrophages convert to foam cells and migrate into the intima of the blood vessel, which marks the first step of a cascade leading to the formation of atherosclerotic plaques (Libby, 2021). It is unclear to date whether CD36–lipoprotein interaction also serves nutrient allocation in healthy individuals. Here, we characterize one of the 14 Drosophila SR-B as a mediator of storage lipid distribution. We show that Lpp colocalizes with Bez at the adipocyte plasma membrane and that, in the absence of Bez, adipocytes can no longer trap Lpp at the plasma membrane. By employing a traceable oleic acid analog (clickable alkyne oleate), we demonstrate that lipid transfer from fat body tissue to isolated Lpp is impaired in Bez mutants: although Bez mutant fat body accumulates lipids that incorporate alkyne oleate, they transfer them at a lower rate to Lpp. Of note, we measured this in a time frame of 60 min, which leaves the possibility that lipid transfer is slower, but not completely abolished, in Bez mutants.

We propose the following model for the function of Bez: located to the plasma membrane of adipocytes, but also gut cells, Bez mediates the export of lipids. The lipids are transferred to other organs as an energy source during nutrient scarcity or energy-demanding developmental processes. For lipid mobilization, Bez interacts with Lpp, the major lipid carrier in hemolymph that can also bind to specific receptors in target organs such as ovaries. In the absence of Bez, adipocytes fail to sufficiently supply lipids. Of note, impaired lipid uptake in mutants for Lpp receptors leads to a similar phenotype of ovary degeneration (Parra-Peralbo and Culi, 2011). Surprisingly, none of the other eight SR-B (out of 14 family members) expressed in the fat body (Herboso et al., 2011) can substitute; neither can intestinal Bez. To our knowledge, this study describes a role in lipid export for one of these SR-B for the first time. Other SR-B are involved in the uptake of their substrates, e.g. NinaD and Santa-Maria in carotenoid uptake or Emp in endocytosis-dependent protein uptake (Pinheiro et al., 2023). Our study suggests that Bez is also required for Lpp interaction and lipid transfer from the gut, as Lpp accumulates in gastric caeca of Bez mutants. Although lipid export from gut cells might also be mediated by Bez, it does not affect oocyte maturation. Midgut-specific depletion of Bez neither induces a developmental delay nor ovary degeneration. Dietary lipids are thus insufficient to promote oocyte maturation. This puzzling finding may relate to specific characteristics of the mobilized lipids. We speculate that lipids traffic sequentially and in a coordinated way from the gut to the fat body and, upon remobilization, further to different target organs.

Our study opens the paths to several exciting research questions of inter-organ communication: how are lipids sorted and allocated between gut, fat body and target tissues? Tissues like brain and ovaries have completely different lipid compositions; how much does uptake of dietary lipids and remodeling in the fat body contribute to their lipid signature? Is there a buffering system for the incorporation of lipids of a specific chain length? For example, will dietary medium- or long-chain lipids affect the composition of ovary lipids, or be adjusted in the fat body? The combination of Drosophila as a tractable genetic model that allows straightforward in vivo and ex vivo manipulation of specific tissues with alkyne lipid tracing by high resolution and high content analysis is a promising approach to answer these questions.

The following fly stocks were used in this study: from the Bloomington Drosophila Stock Center (BDSC) at Indiana University – bez^EP (#28496), w (#6329), c355-Gal4 (#3750), cg-Gal4 (#7011), lsp2-Gal4 (#6357), V32-Gal (#4937), prd-Gal4 (#1947), UAS-Dcr-2 (#24651), UAS-Flp (#4540), Ga4.Act5C (FRT.CD2) (#4780), Gal4-Act5C (FRT.CD2) and UAS-RFP (#30558); from the Vienna Drosophila Resource Center (VDRC) – the bez-RNAi lines P(GD2224)v42872 (#42872) and P(KK111090)VIE-260B (#103492) and the apoLpp-GFP line PBac(fTRG00900.sfGFP-TVPTBF)VK00033 (#318255) that expresses apoLpp tagged with sfGFP at the C-terminus; from the National Institute of Genetics (NIG-FLY) at Japan – the bez-RNAi line # 3829R-2. The group of Irene Miguel-Aliaga (Francis Crick Institute, UK) kindly provided the mex-Gal4 line. The group of Ronald Kühnlein (University of Graz, Austria) kindly provided the stocks: y*w*; P{w[+mW.hs]=GawB} FB P{w[+m*]UAS-GFP1010T2}#2; P{w[+mC]=tubPGal80[ts]2; Sb1 e1 and w*; P{w[*mW.hs]=GawB} FB; +; and UAS-Plin1-EGFP (w*;P{w+mC UAS-plin1::EGFP}/TM3).

The bez^EP allele has the EP-element P{w[+mC]=EP}CG3829[G8378] of 7.987 kb inserted in the third exon of the bez gene locus, 317 bp downstream of the translation initiation site ATG (Fig. 2A). We generated the null bez^jo2 allele by mobilization of the EP-element that carries a mini-white gene (w⁺). Candidates were selected by loss of the mini-white transgene and by the sterility phenotype found in homozygous bez^EP mutant females. PCR of genomic DNA of the bez^jo2 allele showed that 35 nucleotides of the EP-element remained inserted in the Bez coding region and introduced a premature termination code after amino acid 89. The apoLpp-GFP line was combined with the bez^jo2 allele to yield bez^−/−; Lpp-GFP.

The genotype of the bez-RNAi stock used in this study is w; UAS-bez^3829R-2; UAS-Dcr2, UAS-bez^GD2224. This line was generated by combining the UAS-Dcr and two bez-RNAi lines: #3829R-2 from NIG and #42872 from VDRC. Similar, but weaker phenotypes were obtained with both independent RNAi lines, as well as with the VDRC line #103492.

The generation of the conditional bez-RNAi knockdown was done using the TARGET system (McGuire et al., 2004). The crosses between the Gal4 driver and the bez-RNAi flies were carried out at the restrictive temperature of 18°C, at which the Gal80 protein is active, represses Gal4 and the UAS-bez-RNAi constructs are not expressed. Adult flies were collected in the first 24 h after eclosion, separated by sex and shifted to 29°C, temperature, at which the Gal80 protein is inactive and the UAS-bez-RNAi constructs start to be expressed. The flies were kept at 29°C for 4 days before experiments were performed.

Clones in the fat body were generated by combining prd-Gal4 and UAS-Flp with an Actin-Gal4 STOP cassette [Gal4-Act5C (FRT.CD2)] and UAS constructs. The prd enhancer is active during embryonic development and drives the expression of the flipase that, through the excision of a STOP cassette, allows the expression of the UAS constructs (the fluorescent reporters UAS-GFP or UAS-RFP and UAS-bez-RNAi for knockdown or UAS-Bez for rescue experiments) independent from the original prd enhancer. The prd-Gal4 has a transient, but very broad, expression pattern in cells that will differentiate into fat body cells, which allow the recovery of many clones in adipocytes that have been induced very early in development.

Anti-Bez antibody was generated after guinea pig immunization with the peptide SHTKDAEMSMPARQESDR-Cys (amino acids: 2-19) at Pineda Antibody Services (Berlin). The affinity purified anti-Bez antibody was used at 1:100 dilution. Specificity is shown in Figs 1C, 7A and Fig. S3. The following additional primary antibodies were used: rabbit anti-GFP (1:100; Santa Cruz Biotechnology, sc-8834), mouse anti-RFP (1:200; Santa Cruz Biotechnology, sc-393257), goat anti-E-cadherin (1:100; Santa Cruz Biotechnology, sc-6458), rabbit anti-plin1 (1:100, gift from R. Kühnlein, University of Graz, Austria), mouse anti-α-spectrin (1:100; 3A9, Developmental Studies Hybridoma Bank) and mouse anti-orb [1:100; 4H8 and 6H4 (1:1), Developmental Studies Hybridoma Bank]. Conjugated secondary antibodies Alexa488 (Molecular Probes, A37570) and Cy3 (Jackson ImmunoResearch, 115-165-003) were used at 1:200. Conjugated Alexa647 (Invitrogen, 17216523) was used at 1:100. Stainings with DAPI, Nile Red and BODIPY 558/588_C12 (Sigma Aldrich, D-3835) were performed for 2 h at room temperature during the incubation with the secondary antibodies. Embryos were mounted in Fluoromount medium (Southern Biotech). Fluorescent images were obtained on a Zeiss (LSM710) confocal microscope and Airyscan super-resolution images were obtained on a Zeiss LSM 880. Quantification of lipid droplet size and fluorescence intensities were carried out with ImageJ. For colocalization analysis, we used the Plugin JACoP. For quantification of lipid droplet size, we followed the protocol by Ugrankar et al. (2019). Cells from five images from at least three different experiments were analyzed.

For pUAST-Bez generation, Bez coding sequence was amplified using the primers CCGCGGCCGCTGATGTCACATACCAAAGATGCAGAG and GCTCTAGATTACGTTCCCTGATGGATGCCCCCA from the GH19047 cDNA (Drosophila Genomics Resource Center) and inserted in pUAST using NotI and XbaI. Transgenic flies were generated using standard P-element transformation (O'Connor and Chia, 1993).

Quantitative real-time RT-PCR was performed with the iQ5 Real-Time PCR Detection System from Bio-Rad, using the IQ SYBR Green Supermix (Bio-Rad). The RNA was extracted using the NucleoSpin RNA II kit (Macherey-Nagel) and the cDNA were synthesized using the QuantiTect Reverse Transcription Kit (Qiagen). cDNA samples were run in triplicates and experiments were repeated with independently isolated RNA samples. The mRNA amounts were normalized to rpL23 (rp49) mRNA values.

For the quantification of egg depositions, crosses of a UAS-bez-RNAi line and different Gal4-driver were performed with ten females and transferred to 29°C after 2 days. Eggs were collected in apple juice agar plates coated with a thin layer of yeast in intervals of 24 h. To determine production rates, the total number of eggs was counted over the following 2 days.

For the starvation assay, 20 4-day-old flies that were previously separated by sex were transferred to starvation vials containing starvation medium (1% agar in PBS) that only provide a water supply. Mortality rates were scored at 12-h intervals by counting the number of dead flies as diagnosed by the lack of sit-up response and finally determined as a percentage of the total population. The starvation assays with bez^EP mutants were performed at 25°C. For assays with conditional bez-RNAi, flies were maintained at 29°C.

We collected 8-10 adult flies in 2 ml screw caps and froze them at −80°C. They were homogenized in 600 μl PBT using a Precellys 24 Homogenizer (Bertin). After homogenization, the tubes were centrifuged at maximal speed at 4°C and the supernatant was transferred to a cold Eppendorf tube (if necessary, they were centrifuged a second time to remove debris). Then 10 μl were removed from the supernatant and stored at −80°C in Eppendorf tubes for protein determination using the Pierce BCA Protein Assay Kit. Protein content was calculated based on the albumin standard curve.

The remaining supernatant was heated for 10 min at 70°C and 20 μl of the heated supernatant, as well as 20 μl glycerol standards (0.125, 0.25, 0.5 and 1 mg/ml) and a PBST (0.05% Tween-20 in PBS) blank, were transferred to Eppendorf tubes in duplicate. To measure free glycerol, one of the samples was treated with 20 μl of PBST only and the other sample was treated with 20 μl of a triglyceride reagent (TGR). The TGR contains a lipase that digests the TAGs and relieves the glycerol backbone. In the next step, all the probes were incubated for 45 min at 37°C. After incubation, the tubes were centrifuged for 3 min at full speed at room temperature and 30 μl of the probes and standards were transferred to a clear-bottomed 96-well plate and 100 μl of free glycerol reagent were added to each well using a multichannel pipette. The plate was incubated at 37°C for 5 min sealed with parafilm. Air bubbles and condensate were removed by centrifuging the plate (1000 g for 5 min in a Heraeus Megafuge 1.0R). Absorbance at 540 nm was measured using a TECAN plate reader. Samples and standards were performed in triplicate. The average of the triplicate was subtracted by the average of the blank. Triolein-equivalent standard curves were constructed for the TGR-treated and untreated standards. Free glycerol content of the samples was calculated based on these standard curves. Glycerol concentration of the TGR-treated samples was subtracted by the concentration of the untreated samples. The difference in glycerol concentration hereby represents the amount of TAGs that were digested by the enzymatic reaction. Finally, the values were normalized to protein levels.

The hemolymph purification protocol was adapted from Palm et al. (2012): freshly prepared hemolymph was used for all experiments. Thirty third instar larvae were washed extensively in PBS to eliminate the food attached to the cuticle. In 250 µl PBS with protease inhibitors (cOmplete Protease Inhibitor Cocktail, Roche), the cuticles of the larvae were carefully ripped open to allow hemolymph bleeding. The liquid was transferred to an Eppendorf cup and centrifuged for 30 min at 4°C and 1500 g. The supernatant was transferred to a fresh Eppendorf cup and centrifuged for 30 min at 4°C and 16,000 g. Of note, we omitted further purification by an isopycnic gradient to maintain the functionality of the lipoproteins. The supernatant was transferred to a new tube and incubated immediately with dissected fat body. For lipidomics analysis, lipids of 50 µl freshly isolated Lpp were extracted by addition of 100 µl LC-MS grade water. Then 500 µl extraction mix [5/1 methanol/CHCl₃ (LC-MS grade)]+20 µl OMICS internal standard were added as previously described (Thiele et al., 2019). Samples were sonicated for 30 min in a bath sonicator and centrifuged for 2 min at 20,000 g. The supernatant was decanted into a fresh tube and 300 µl CHCl₃ and 600 µl 1% acetic acid in LC-MS grade water were added. Samples were shaken manually for 10 s and centrifuged at 20,000 g for 5 min. The lower phase was transferred into a fresh tube and dried in a speed-vac (Eppendorf Concentrator Plus) for 15 min at 45°C and redissolved in 500 µl spray buffer [8/5/1 isopropanol/methanol/H₂O (all LC-MS grade)+10 mM ammonium acetate+0.1% acetic acid (LC-MS grade)] until mass spectrometry analysis.

A previously published protocol (Thiele et al., 2012) to preload Drosophila fat body tissue with alkyne oleate was optimized. Click reaction and mass spectrometry analysis were carried out as previously described (Thiele et al., 2019; Wunderling et al., 2023). Briefly, fat body tissue from control and Bez mutant larvae was dissected in triplicate in hemolymph-like buffer (HL3A; 115 mM sucrose, 70 mM NaCl, 20 mM MgCl₂, 10 mM NaHCO₃, 5 mM KCl, 5 mM HEPES, 5 mM trehalose, pH 7.2; Paradis et al., 2022). Tissue was preloaded with 50 µM alkyne oleate (FA 19:1;Y) in HL3A when incubated on a nutator for 60 min. An aliquot was taken to analyze the fat body lipidome and quantify alkyne oleate uptake and metabolism. The remaining tissue was washed extensively with HL3A to remove adherent alkyne lipid tracer. Isolated Lpp (see above) was added 1/10 to the preloaded fat body tissue and incubated together on a nutator for 60 min. The fat body tissue was carefully removed, first manually and then by passaging through a cell strainer. Flow-through was centrifuged at 4°C and 3000 g for 30 min and the supernatant was removed. Lipids were extracted as described above but additional internal standards for alkyne labeled metabolites were included (Thiele et al., 2019). After drying the sample, lipids were resolved in 10 µl CHCl₃ and sonicated for 5 min. Then 40 µl of C171 click mix [prepared by mixing 10 µl of 100 mM C171 in 50% methanol (stored as aliquots at −80°C) with 200 µl 5 mM Cu(I)AcCN₄BF₄ in AcCN and 800 µl ethanol] were added and the samples were incubated at 40°C overnight (16 h). After this, 200 µl CHCl₃ and 200 µl water were added and the samples were centrifuged at 20,000 g for 5 min. The upper phase was discarded and samples were dried in a speed-vac for 15 min at 45°C. Samples were resolved in 500 µl spray buffer and analyzed by mass spectrometry.

Bar charts represent mean and standard deviation (s.d.). Error bars in curves represent standard error of the mean (s.e.m.). Boxes in box plots represent the interquartile range and median, whiskers represent minimum and maximum. Green squares in box plots represent single data points. We used Microsoft Excel and GraphPad Prism for bar charts and curves and Origin Pro 8G for box plots. We used the software GraphPad Instat for our statistical analyses. Two-sided, unpaired Student's t-test was applied for normally distributed data in single comparisons, assuming heteroscedasticity. One-way ANOVA with Tukey-Kramer post-test was used for multiple comparisons. The Kolmogorov-Smirnov test was applied to test normality, and Bartlett's method was used to test for equal standard deviations within groups. *P<0.05, **P<0.01, ***P<0.001. A minimum of three biological replicates was used for each analysis.

We thank Marko Brankatschk for valuable discussion of the data and procedures. We thank Mohammed Yaghmour, Darla Dancourt Ramos and Valeria Gulayeva for help with experiments, and Almut Wingen and Dominic Gosejacob for discussion. We thank Christoph Thiele for providing the alkyne oleate and acknowledge the LIMES MS facility. We thank Ronald Kühnlein for sharing the Plin1-GFP and fat body-Gal4 fly line and Plin1 antibody.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.202551.reviewer-comments.pdf