Introduction

Sphingolipids are key structural and functional components of the cell membrane in all eukaryotic cells and are enriched in glial cells, such as oligodendrocytes, where they comprise up to 30% of all membrane lipids.¹ All complex sphingolipids, like glycosylsphingolipids and sphingomyelin, derive from ceramide.^2–4 In the de novo ceramide biosynthesis pathway, dihydroceramide desaturases, such as DEGS1, produce ceramide from dihydroceramide by catalyzing the formation of a trans double bond between carbons 4 and 5 of the sphingoid backbone, which enhances conformational plasticity.^5,6 In humans, bi-allelic mutations in DEGS1 cause hypomyelinating leukodystrophy-18 (HLD-18), a progressive, often fatal pediatric neurodegenerative disease marked by cerebral atrophy, white matter reduction, and hypomyelination.^7–9 The primary neural cell type impacted by loss of DEGS1 function and the cell biology of how disruption of ceramide synthesis leads to neurodegeneration remain unknown.

The de novo ceramide biosynthetic pathway is well conserved among higher metazoans.^4,10–13 De novo ceramide synthesis occurs in the endoplasmic reticulum (ER) and starts with the rate-limiting activity of the serine palmitoyltransferase (SPT) complex, which condenses serine and palmitoyl-CoA (lauoryl-CoA in flies) to form 3-Ketosphinganine, which is converted to sphinganine by 3-Ketosphinganine reductase. Ceramide synthases condense sphinganine with acyl-CoA to generate dihydroceramide, which is converted to ceramide by dihydroceramide desaturases. Ceramide is then efficiently transported by the specific ceramide transporter CERT from the ER to the Golgi,¹⁴ where it undergoes headgroup modifications to produce complex sphingolipids that eventually translocate to the plasma membrane.¹⁵ Mutations in most members of the SPT complex, ceramide synthases, and DEGS1 lead to neurodegeneration,^4,12,13 identifying the de novo ceramide biosynthesis pathway as a hotspot for neurodegenerative disease mutations.

Consistent with its central role in ceramide biogenesis, reduction or loss of DEGS1 function in human patients or cell lines, mice, zebrafish, and flies drive dihydroceramide accumulation and ceramide depletion.^8,9,16,17 HLD-18 patients display reduced myelin sheath thickness in peripheral nerves, and knockdown of DEGS1 function in zebrafish reduces the number of myelin basic protein-positive oligodendrocytes,^8,9 suggesting DEGS1 regulates Schwann cell and oligodendrocyte development. In Drosophila, genetic ablation of infertile crescent (ifc), the fly DEGS1 ortholog, drives activity-dependent photoreceptor degeneration,¹⁷ suggesting that ifc function is crucial for neuronal homeostasis. In addition, forced expression of a wildtype ifc transgene in neurons, glia, or muscles was shown to rescue the ifc mutant phenotype.¹⁷ Whether ifc/DEGS1 acts primarily in glia, neurons, or other cells to regulate nervous system development then remains to be determined.

Research on DEGS1 points to dihydroceramide accumulation as the driver of nervous system defects caused by DEGS1 deficiency. Pharmacological or genetic inhibition of ceramide synthase function, which should reduce dihydroceramide levels, suppresses the observed reduction of MBP-positive oligodendrocytes in zebrafish and the activity-dependent photoreceptor degeneration in flies triggered by reduction of DEGS1/ifc function.^9,17 Dihydroceramide accumulation may also contribute to other neurodegenerative diseases, as gain of function mutations in components of the SPT complex cause juvenile amyotrophic lateral sclerosis due to elevated sphingolipid biosynthesis, with dihydroceramides showing the greatest relative increase of all sphingolipids.^18–22 How dihydroceramide accumulation alters the cell biology of neurons, glia, or both to trigger myelination defects and neurodegeneration is unclear.

With its vast genetic toolkit, Drosophila is a powerful system in which to dissect the genes and pathways that regulate glial development and function.^23,24 In flies, six morphologically and functionally distinct glial subtypes regulate nervous system development and homeostasis.^25,26 The surface perineurial and subperineurial glia act as physiochemical barriers to protect the nervous system and control metabolite exchange with the hemolymph, carrying out a similar function as the human blood-brain-barrier.^27,28 Residing between the surface glia and the neuropil, cortex glia ensheathe the cell bodies of neuroblasts and neurons in the CNS in protective, nutritive, honeycomb-like membrane sheaths. Ensheathing glia define the boundary of the neuropil and insulate axons, dendrites, and synapses from neuronal cell bodies in the CNS.²⁹ Astrocyte-like glia extend fine membrane protrusions that infiltrate the neuropil and form a meshwork of cellular processes that ensheathe synapses and regulate synaptic homeostasis in the CNS.²⁹ In the PNS, wrapping glia reside internally to the surface glia and insulate axons in the PNS to enhance neuronal signaling and provide energy support.

Using the Drosophila model, we found that ifc acts primarily in glia to regulate CNS development with its loss disrupting glial morphology. Our work supports a model in which inappropriate accumulation and retention of dihydroceramide in the ER drives ER expansion, glial swelling, and the failure of glia to enwrap neurons, ultimately leading to neuronal degeneration as a secondary consequence of glial dysfunction. Given the conserved nature of de novo ceramide biosynthesis, our findings likely illuminate the exact mechanism through which elevated dihydroceramide levels drive neuronal degeneration and cell death in flies and humans.

Results

ifc contributes to the regulation developmental timing and CNS structure

In an EMS-based genetic screen, we uncovered three non-complementing mutations that when homozygous or trans-heterozygous to each other resulted in identical phenotypes, including a 3-day or greater delay in reaching the late-third larval instar stage, reduced brain size, progressive ventral nerve cord elongation, axonal swelling, and lethality at the late larval or early pupal stage (Figure 1A; data not shown). Whole genome sequencing revealed that ifc was independently mutated in each line: ifc^js1 and ifc^js2 encode V276D and G257S missense mutations, respectively, and ifc^js3 encodes a W162* nonsense mutation (Figures 1A and 1B). Sanger sequencing also uncovered a missense mutation, E241K, in the molecularly uncharacterized ifc¹ allele (Figure 1B). All four mutations reside in the fatty acid desaturase domain, the hotspot for mutations in human DEGS1 that cause HLD-18 (Figures 1B-B’).

Figure 1.

As a prior study reported that a CRISPR-generated, gene-specific deletion of ifc, ifc-KO, resulted in early larval lethality¹⁷, we first confirmed the genetic nature of our ifc alleles because they caused late larval-early pupal lethality. Complementation crosses of each ifc allele against a deficiency of the region (Df(2L)BSC184) and ifc-KO revealed that all combinations, including larvae trans-heterozygous for ifc-KO over Df(2L)BSC184, survived to the late larval-early pupal stage and yielded phenotypes identical to those detailed above for the newly uncovered ifc alleles. Flies homozygous for ifc-KO, however, died as early larvae. Further analysis uncovered second site mutation(s) in the 21E2 chromosomal region responsible for the early lethal phenotype of the ifc-KO chromosome (see Methods). When uncoupled from these mutation(s), larvae homozygous for the “clean” ifc-KO chromosome developed to the late larval-early pupal stage and manifested phenotypes identical to the other ifc alleles. This analysis defined the correct lethal phase for ifc and identified our ifc alleles as strong loss of function mutations.

Loss of ifc function drives ceramide depletion and dihydroceramide accumulation

As ifc/DEGS1 converts dihydroceramide to ceramide, we used untargeted lipidomics on whole larvae and the isolated CNS from wild-type and ifc-KO/ifc^JS3 larvae (hereafter termed ifc^-/- larvae) to assess the effect of loss of ifc function on metabolites in the ceramide pathway (Figure 1C). Loss of ifc function resulted in a near complete loss of ceramides and a commensurate increase in dihydroceramides in the CNS and whole larvae (Figures 1E, S1). Sphinganine, the metabolite directly upstream of dihydroceramide, also exhibited a significant increase in its levels in the absence of ifc function, while metabolites further upstream were unchanged in abundance or undetectable (Figures 1E, S1). Ceramide derivatives like sphingosine, CPE, and Glucosyl-Ceramide (Glc-Cer), were reduced in levels and replaced by their cognate dihydroceramide forms (e.g., Glc-DiCer) (Figures 1E, S1). Loss of the enzymatic function of Ifc then drives dihydroceramide accumulation and ceramide loss.

ifc governs glial morphology and survival

To connect this metabolic profile to a cellular phenotype, we assayed ifc function in the CNS. We leveraged the expression of fatty acid binding protein (FABP) as a marker of cortex glia³⁰ and that of the M{3xP3-RFP.attP} phi-C31 “landing pad” transgene, which resided in the isogenic target chromosome of our screen, as a marker of most glia (Figures S2 and S3). In addition, we labeled neuroblasts with Deadpan (Dpn), neurons with ELAV, and axons with N-Cadherin (NCad). In ifc^-/- larvae, we observed a clear reduction in Dpn-positive neuroblasts in the optic lobe, swelling of glia in peripheral nerves, enhanced RFP expression in the CNS, and the presence of large swollen, cortex glia identified by RFP labeling and fatty acid binding protein (FABP) expression (Figures 1F, S4, and S5). In wild-type larvae, cortex glia display compact cell bodies and fully enwrap individual neuronal cell bodies with their membrane sheaths (Figures 1F and S4).³⁰ In ifc^-/- larvae, cortex glia display swollen cell bodies, fail to fully enwrap neuronal cell bodies, displace neurons from their regular arrangement, and appear to contain brightly fluorescent RFP-positive aggregates (Figures 1F, S6). ifc is then necessary for glial development and function in the larval nervous system. We observed identical CNS phenotypes in larvae homozygous mutant for the ifc^js1and ifc^js2 alleles (Figure S7).

To track the impact of ifc on glial morphology, we combined GAL4 lines specific for each glial subtype with a UAS-linked membrane-tagged GFP transgene (Myr-GFP) and the MultiColor FlpOut system.^31–33 Using this approach, we determined that loss of ifc function affects all CNS glial subtypes except perineurial glia (Figures 2E-E’ and 2J-J’). Cortex glia appeared swollen, failed to enwrap neurons, and accumulated large amounts of Myr-GFP⁺ internal membranes (Figures 2A-A’ and 2F-F’). Ensheathing glia (Figures 2B-B’ and 2G-G’) and subperineurial glia (Figures 2D-D’ and 2I-I’) also displayed swollen, disorganized cell bodies and accumulated Myr-GFP⁺ internal membranes. Astrocyte-like glia displayed smaller cell bodies, reduced membrane extensions, and disrupted organization along the dorsal ventral nerve cord (Figures 2C-C’). We conclude that ifc regulates the morphology of most glial subtypes in the larval CNS.

Figure 2.

Next, we asked if loss of ifc function alters the number of each glial subtype. Using the same glial subtype-specific GAL4 lines to drive a nuclear-localized GFP transgene, we counted the total number of all CNS glial subtypes, except perineurial glia, in wild type and ifc^-/- larvae. To remove the small size of the brain in ifc^-/- larvae as a confounding factor, we focused our analysis on the ventral nerve cord. The number of subperineurial glia was unchanged between the two genotypes, but we observed a 12%, 40%, and 72% reduction in the number of astrocyte-like, ensheathing, and cortex glia, respectively, in ifc^-/- larvae relative to wild-type (Figures 2K-N). Our data reveal a broad role for ifc in regulating glial cell morphology and number in the Drosophila larval CNS. Subsequent experiments revealed that a reduction in cell proliferation and an increase in apoptosis both contribute to the observed reduction in the number of cortex glia (Figure S8).

ifc acts in glia to regulate glial and CNS development

Prior work in zebrafish showed that DEGS1 knockdown reduced the number of myelin basic protein-positive oligodendrocytes;⁹ in flies, loss of ifc function in the eye drove photoreceptor degeneration.¹⁷ Neither study uncovered the cell type in which ifc/DEGS1 acts to regulate neural development. To address this question, we used the GAL4/UAS system, RNAi-mediated gene depletion, and gene rescue approaches to see if ifc acts in neurons or glia to control glial development and CNS morphology. First, we used a UAS-linked ifc-RNAi transgene to deplete ifc function in all neurons (elav-GAL4; repo-GAL80) or all glia (repo-GAL4). Focusing on FABP-positive cortex glia due to their easily scorable phenotype, we found that pan-glial, but not pan-neuronal, knockdown of ifc recapitulated the swollen cortex glia phenotype observed in ifc mutant larvae (Figures 3C-D and 3J).

Figure 3.

To complement our RNAi approach, we asked if GAL4-driven expression of a wild-type Drosophila ifc or human DEGS1 transgene rescued the ifc^-/- CNS phenotype. In the absence of a GAL4 driver, the ifc transgene drove weak rescue of the cortex glia phenotype (Figures 3E and 3K), consistent with modest GAL4-independent transgene expression reported for UAS-linked transgenes.³⁴ Pan-neuronal expression of ifc drove modest rescue of the ifc CNS phenotype beyond that observed for the ifc transgene alone (Figures 3F and 3K), but pan-glial expression of ifc fully rescued the ifc mutant cortex glia phenotype and other CNS phenotypes (Figures 3H, 3K, and S9). Identical experiments using the human DEGS1 transgene revealed that only pan-glial DEGS1 expression provided rescuing activity, albeit at much weaker levels than the Drosophila ifc transgene (Figures 3G, 3I, and 3K). Pan-glial expression of the Drosophila ifc transgene was, in fact, ∼15-fold more potent than the human DEGS1 transgene at rescuing the ifc lethal phenotype to adulthood: When ifc was expressed in all glia, 57.9% of otherwise ifc mutant flies survived to adulthood (n=2452), but when ifc was replaced by DEGS1 only 3.9% of otherwise ifc mutant flies reached adulthood (n=1303). No ifc mutant larvae reached adulthood in the absence of either transgene (n=1030). We infer that ifc acts primarily in glia to govern CNS development and that human DEGS1 can partially substitute for ifc function in flies despite a difference in the preferred length of the sphingoid backbone in flies versus mammals.¹⁰

Results of our gene rescue experiments conflict with a prior study on ifc in which expression of ifc in neurons was found to rescue the ifc phenotype.¹⁷ In this context, we note that elav-GAL4 drives UAS-linked transgene expression not just in neurons, but also in glia at appreciable levels,^{35, 36} and thus needs to be paired with repo-GAL80 to restrict GAL4-mediated gene expression to neurons. Thus, “off-target” expression in glial cells may account for the discrepant results. It is, however, more difficult to reconcile how neuronal or glial expression of ifc would rescue the observed lethality of the ifc-KO chromosome given the presence additional lethal mutations in the 21E2 region of the second chromosome.

ifc is predominately expressed in glia and localizes to the ER

Next, we tracked ifc expression in the CNS via RNA in-situ hybridization and an ifc-T2A-GAL4 transcriptional reporter. RNA in-situ hybridization revealed that ifc is widely expressed in the CNS (Figure 4A), most obviously in the distinctive star-shaped astrocyte-like glia (Figure 4B), which are marked by Ebony expression.³⁷ RNA in-situ hybridization was not ideal for tracing ifc expression in other cells, likely due to signal diffusion. Thus, we paired the ifc-T2A-GAL4 transcriptional reporter with a nuclear RFP (nRFP) transgene³⁸ and confirmed authenticity of the ifc-T2A-GAL4 line by its strong expression in astrocyte-like glia (Figure 4C). Using this approach, we observed strong nRFP expression in all glial cells (Figures 4D and S10A) and modest nRFP expression in all neurons (Figures 4E and S10B), suggesting ifc is transcribed at higher levels in glial cells than neurons in the larval CNS.

Figure 4.

Using a fosmid transgene that harbors a GFP-tagged version of ifc flanked by ∼36-kb of its endogenous genomic region and molecular markers for ER, cis-Golgi, and trans-Golgi,^39–42 we found that the Ifc-GFP colocalized strongly with the ER markers Calnexin 99A (CNX99A) and ESYT (Figures 4F-F’, 4G-G’, and S7) and weakly with the cis-Golgi marker GOLGIN84 (Figure S10C) and the trans-Golgi marker GOLGIN245 (Figure S10D). Our results indicate that Ifc localizes primarily to the ER, aligning with the presumed site of de novo ceramide biosynthesis and prior work on DEGS1 localization in cell lines.^8,43

Loss of ifc drives ER expansion and lipid droplet loss in cortex glia

We next asked if loss of ifc function altered ER, Golgi, or lysosome morphology. Focusing on cortex glia, we observed a clear expansion of the ER marker CNX99A (Figures 4H-H’ and K-K’), a mild enrichment of the Golgi markers, GOLGIN84 and GOLGIN245, in diffuse “clouds” (Figures 4I-I’, 4L-L’, S10E-F), and a reduction in expression of the lysosome marker LAMP (Figures 4J-J’ and M-M’) in ifc^-/- larvae. The expansion of ER markers in ifc mutant larvae compelled us to obtain high-resolution views of organelle structure in cortex glia via transmission electron microscopy (TEM). In wildtype, cortex glia display a compact cytoplasm that surrounds a large nucleus (Figures 5A-B) and extend glial sheaths that fully enwrap adjacent neuronal cell bodies (black arrows; Figures 5C-D). In ifc^-/- larvae, cortex glia display enlarged cell bodies with a maze-like pattern of internal membranes (solid white arrows; Figures 5A’-B’ and 5E-E’) and fail to enwrap neurons (hollow white arrows; Figures 5C’-D’). The internal membrane structures appear to assume an ER-like identity, as we observed significant overlap between the ER marker CNX99A and the membrane marker Myr-GFP when Myr-GFP was driven by a cortex glia-specific GAL4 line (Figure 5F). We observed similar yet milder effects on cell swelling and internal membrane accumulation in subperineurial and wrapping glia in abdominal nerves (purple and pink shading, respectively; Figures 5G-G’ and 5H-H’), indicating that loss of ifc drives internal membrane accumulation in and swelling of multiple glial subtypes.

Figure 5.

TEM analysis also revealed a near complete depletion of lipid droplets in the CNS of ifc^-/- larvae (compare Figures 5A, 5C, and 5D to 5A’, 5C’ and 5D’; lipid droplets marked by asterisk; Figure 5M), which we confirmed using BODIPY to mark neutral lipids in FABP-positive cortex glia (Figures 5I-L). In the CNS, lipid droplets form primarily in cortex glia³⁰ and are thought to contribute to membrane lipid synthesis through their catabolism into free fatty acids versus acting as an energy source in the brain.⁴⁴ Consistent with the possibility that increased membrane lipid synthesis drives lipid droplet reduction, RNA-seq assays of dissected nerve cords revealed that loss of ifc drove transcriptional upregulation of genes that promote membrane lipid biogenesis, such as SREBP, the conserved master regulator of lipid biosynthesis, SCAP, an activator of SREBP, and Pcyt1/Pcyt2, which promote phosphatidylcholine and phosphatidylethanolamine synthesis.^45–47 The spliced form of Xbp-1 mRNA (Xbp-1s) (Figure 5J), which promotes membrane lipid synthesis required for ER biogenesis and is activated by the unfolded protein response (UPR),⁴⁸ is also upregulated in the CNS of ifc mutant larvae. Most ER chaperones, which are typically transcriptionally upregulated upon UPR activation,⁴⁹ were however downregulated (Figure S11A), suggesting that in this case misfolded protein is not the factor that triggers UPR activation and increased ER membrane biogenesis upon loss of ifc.

Loss of ifc increases the saturation levels of triacylglycerols and membrane phospholipids

Lipid droplets are composed largely of triacylglycerols (TGs),⁴⁷ and we observed a 5-fold and 3-fold drop in TG levels in the CNS and whole larvae of ifc mutant larvae relative to wild-type (Figure 6A). Our lipidomics analysis also revealed a shift of TGs toward higher saturation levels in the absence of ifc function (Figures 6A-C). Consistent with this, we observed transcriptional upregulation of most genes in the Lands cycle, which remodels phospholipids by replacing existing fatty acyl groups with new fatty acyl groups (Figures S11B-C).^50–53 As TG breakdown results in free fatty acids that can be used for membrane phospholipid synthesis, we asked if changes in TG levels and saturation were reflected in the levels or saturation of the membrane phospholipids phosphatidylcholine (PC), phosphatidylethanolamine (PE), and phosphatidylserine (PS). In the absence of ifc function, PC and PE exhibited little change in quantity (Figures 6D, 6G), but the levels of the less abundant PS were increased three-fold increase in the CNS of ifc mutant larvae relative to wild-type (compare Figure 6J to Figures 6D and 6G). All three phospholipids, however, displayed increased saturation levels: The relative levels of all PC, PE, and PS species were reduced, except for the most saturated form of each phospholipid – 18:1/18:1 – which is increased (Figures 6E-F and 6H-L). This increase was more pronounced in the CNS than whole larvae (Figures 6F, 6I, and 6L), implying that loss of ifc function creates greater demand for lipid remodeling in the CNS.

Figure 6.

Reduction of dihydroceramide synthesis suppresses the ifc CNS phenotype

Following its synthesis, ceramide is transported by CERT from the ER to the Golgi, but CERT is less efficient at transporting dihydroceramide than ceramide.¹³ The expanded nature of the ER in ifc mutant larvae supports a model in which loss of ifc triggers excessive accumulation and retention of dihydroceramide in the ER due to CERT’s inefficiency of transporting dihydroceramide to the Golgi, driving ER expansion and glial swelling. If this model is correct, reduction of dihydroceramide levels should suppress the ifc CNS phenotype. In agreement with this model, we found that the schlank^G0365 loss of function allele dominantly suppressed the enhanced RFP expression (Figure 7M) and CNS elongation phenotypes of ifc (Figure 7N). We also found that glial-specific depletion of schlank suppressed the internal membrane accumulation (Figures 7A-B), reduced lipid droplet size (Figures 7C-D, and 7O), glial swelling (Figures 7E-F, and S12), enhanced RFP expression (Figure 7M), CNS elongation (Figure 7N), and reduced optic lobe (Figure S5) phenotypes observed in otherwise ifc mutant larvae. Our data support the model that inappropriate retention of dihydroceramide in the ER drives ER expansion and glial swelling and dysfunction.

Figure 7.

Glial-specific depletion of ifc function triggers neuronal cell death

Loss of DEGS1/ifc in human and flies has been shown to promote neurodegeneration and neuronal cell death,^8,9,17 but whether neuronal death results from an intrinsic defect in neurons or is induced by glial dysfunction remains unclear. By using Cleaved caspase-3 as a marker of dying cells^54,55 and ELAV to track neurons, we tracked neuronal cell death in the brain and ventral nerve cord of wild-type and ifc^-/-mutant late-third instar larvae. In wildtype, little to no cell death was evident in the brain or nerve cord and neurons appear smoothly packed next to each other; in contrast, in ifc^-/- larvae, significant cell death was apparent in the brain and to a lesser degree in the nerve cord, with Caspase-3 staining often highlighting small perforations in the neuronal cell layer (Figures 7G-H, 7P, and S7). This perforated pattern was associated with and more expansive than Caspase-3 staining, indicating the perforations mark Caspase-positive dying neurons and Caspase-negative dead neurons. Glial-specific depletion of schlank function in otherwise ifc^-/- larvae suppressed the neuronal cell death phenotype (Figures 7I and 7P), supporting the model that loss of ifc function specifically in glia, rather than a specific requirement for ifc function in neurons, drives neuronal cell death. To directly test whether ifc function in glia is required to guard against neuronal cell death, we used the GAL4-UAS system and RNAi mediated gene interference to deplete ifc function specifically in glia (repo-GAL4) or in neurons (elav-GAL4; repo-GAL80) and found that glial-specific, but not neuronal-specific, depletion of ifc function drove significant neuronal cell death in the brain and to a greater extent the nerve cord, a phenotype that was enhanced upon removal of one copy of ifc (Figures 7J-L, 7Q, and S13). Loss of ifc function then triggers glial dysfunction, which in turn drives neuronal cell death.

Discussion

Our work on ifc supports a model in which loss of ifc/DEGS1 function drives glial dysfunction through the accumulation and inappropriate retention of dihydroceramide in the ER of glia with this proximal defect driving ER expansion, glial swelling, and subsequent neuronal cell death and neurodegeneration. A large increase in dihydroceramide would also impact ER membrane structure: ER membranes are typically loosely packed, thin, semi-fluid structures, with sphingolipids comprising just 3% of ER phospholipids.⁴⁷ Sphingolipids in general and dihydroceramide in specific are highly saturated lipids, promote lipid order, tighter lipid packing, membrane rigidity, and thicker membranes.^56,57 Our observation of thick ER membranes in cortex glia in ifc^-/- larvae (Figure 5) aligns with increased dihydroceramide levels in the ER. In this context, we note that of many clinical observations made on an HLD-18 patient, one was widening of the ER in Schwann cells,⁸ a finding that when combined with our work suggests that dihydroceramide accumulation in the ER is the proximal cause of HLD-18.

Although ER expansion represents the most proximal effect of loss of ifc/DEGS1 function and dihydroceramide accumulation on cortex glia, other organelles and their functions are likely also disrupted, minimally as a consequence of ER disruption. For example, the apparent reduction of the lysosome marker, LAMP, in cortex glia of ifc mutant larvae correlates with increased RFP levels and the presence of bright RFP puncta or aggregates in this cell type, suggesting impaired lysosome function contributes to increased RFP perdurance and aggregation. Defects in the activity of multiple organelles may collaborate to elicit the full phenotype manifested by loss of ifc/DEGS1 function, resulting in glial dysfunction and subsequent neuronal cell death.

Increased dihydroceramide levels may contribute to a broader spectrum of neurodegenerative diseases than simply HLD-18. Recent work reveals that gain of function mutations in SPTLC1 and SPTLC2, which encode components of the SPT complex that catalyzes the initial, rate limiting step of de novo ceramide and sphingolipid biosynthesis,⁵⁵ cause juvenile ALS via increased sphingolipid biosynthesis.^18–22 Of all sphingolipids, the relative levels of dihydroceramide were increased the most in patient plasma samples, suggesting that DEGS1 activity becomes limiting in the presence of enhanced SPT activity and that dihydroceramide accumulation contributes to juvenile ALS. Any mutations that increase metabolite flux through the ceramide pathway upstream of DEGS1 may then increase dihydroceramide levels, drive ER expansion and cell swelling, and lead to neurodegeneration, with disease severity predicated on the extent of excessive dihydroceramide accumulation. Model systems, like flies, can harness the power of genetic modifier screens to identify genes and pathways (potential therapeutic targets), that can be tweaked to ameliorate the effect of elevated dihydroceramide levels on neurodegeneration.

Our work appears to pinpoint glia as the cell type impacted by loss of ifc/DEGS1 function, specifically glia that exhibit great demand for membrane biogenesis like cortex glia in the fly and oligodendrocytes or Schwann cells in mammals. In larvae, ifc is expressed at higher levels in glial cells than in neurons, and its genetic function is required at a greater level in glia than neurons to govern CNS development. Our unpublished work on other genes in the ceramide metabolic pathway reveals similar glial-centric expression patterns in the larval CNS to that observed for ifc, suggesting they too function primarily in glia rather than neurons at this stage. In support of this model, a study on CPES, which converts ceramide into CPE, the fly analog of sphingomyelin, revealed that CPES is required in cortex glia to promote their morphology and homeostasis and to protect flies from photosensitive epilepsy.⁵⁹ Similarly, ORMDL, a dedicated negative regulator of the SPT complex, is required in oligodendrocytes to maintain proper myelination in mice.⁶⁰ Given that many glial cell types are enriched in sphingolipids and exhibit a great demand for new membrane biogenesis during phases of rapid neurogenesis and axonogenesis, we believe that glial cells, such as cortex glia, oligodendrocytes, and Schwann cells, rather than neurons manifest a greater need for ifc/DEGS1 function and ceramide synthesis during developmental stages marked by significant nervous system growth.

Will this glial-centric model for ifc/DEGS1 function, and more generally ceramide synthesis, hold true in the adult when neurogenesis is largely complete and the demand for new membrane synthesis in glia dissipates? Recent work in the adult fly eye suggests it may not. Wang et al.⁶¹ argued that the GlcT enzyme, which converts ceramide to glucosylceramide, is expressed at much higher levels in neurons than glia, and that glucosylceramide is then transported from neurons to glial cells for its degradation, suggesting cell-type specific compartmentalization of sphingolipid synthesis in neurons and its degradation in glia in the adult. In the future, it will be exciting to uncover whether genes of the sphingolipid metabolic pathway alter their cell-type specific requirements as a function of developmental stage.

We note that cortex glia are the major phagocytic cell of the CNS and phagocytose neurons targeted for apoptosis as part of the normal developmental process.^23–26 Thus, while we favor the model that ifc triggers neuronal cell death due to glial dysfunction, it is also possible that increased detection of dying neurons arises due at least in part to a decreased ability of cortex glia to clear dying neurons from the CNS. At present, the large number of neurons that undergo developmentally programmed cell death combined with the significant disruption to brain and ventral nerve cord morphology caused by loss of ifc function render this question difficult to address. Additional evidence does, however, support the idea that loss of ifc function drives excess neuronal cell death: Clonal analysis in the fly eye reveals that loss of ifc drives photoreceptor neuron degeneration¹⁷, indicating that loss of ifc function drives neuronal cell death; cortex-glia specific depletion of CPES, which acts downstream of ifc, disrupts neuronal function and induces photosensitive epilepsy in flies⁵⁹, indicating that genes in the ceramide pathway can act non-autonomously in glia to regulate neuronal function; recent genetic studies reveal that other glial cells can compensate for impaired cortex glial cell function by phagocytosing dying neurons⁶², and we observe that the cell membranes of subperineurial glia enwrap dying neurons in ifc mutant larvae (Fig. S14), consistent with similar compensation occurring in this background, and in humans, loss of function mutations in DEGS1 cause neurodegeneration.^7–9 Clearly, future work is required to address this question for ifc/DEGS1 and perhaps other members of the ceramide biogenesis pathway.

Altered glial function may not only derive from dihydroceramide buildup in the ER, but also from altered cell membrane structure due to the replacement of ceramide and its derivates, such as GlcCer and CPE, with the cognate forms of dihydroceramide. Relative to dihydroceramide species, the 4-5 trans carbon-carbon double bond in the sphingoid backbone of ceramide-containing sphingolipids enables them to form more stable hydrogen bonds with water molecules and facilitates their ability to associate with lipids of different saturation levels.⁵ A high dihydroceramide to ceramide ratio has been shown to form rigid gel-like domains within model membranes and to destabilize biological membranes by promoting their permeabilization.^63,64 As even minor alterations to membrane properties can disrupt glial morphology,⁶⁵ such alterations in membrane rigidity and stability may underlie the failure of cortex glia to enwrap adjacent neurons. The observed increase in saturation of PE, PC, and PS in the CNS of ifc mutant larvae may reflect a compensatory response employed by cells to stabilize cell membranes and guard cell integrity when challenged with elevated levels of dihydroceramide.

The expansion of ER membranes coupled with loss of lipid droplets in ifc mutant larvae suggests that the apparent demand for increased membrane phospholipid synthesis may drive lipid droplet depletion, as lipid droplet catabolism can release free fatty acids to serve as substrates for lipid synthesis. At some point, the depletion of lipid droplets, and perhaps free fatty acids as well, would be expected to exhaust the ability of cortex glia to produce additional membrane phospholipids required for fully enwrapping neuronal cell bodies. Under wild-type conditions, many lipid droplets are present in cortex glia during the rapid phase of neurogenesis that occurs in larvae. During this phase, lipid droplets likely support the ability of cortex glia to generate large quantities of membrane lipids to drive membrane growth needed to ensheath newly born neurons. Supporting this idea, lipid droplets disappear in the adult Drosophila CNS when neurogenesis is complete and cortex glia remodeling stops.³⁰ We speculate that lipid droplet loss in ifc mutant larvae contributes to the inability of cortex glia to enwrap neuronal cell bodies. Prior work on lipid droplets in flies has focused on stress-induced lipid droplets generated in glia and their protective or deleterious roles in the nervous system.^66–68 Work in mice and humans has found that more lipid droplets are often associated with the pathogenesis of neurodegenerative diseases,^69–71 but our work correlates lipid droplet loss with CNS defects. In the future, it will be important to determine how lipid droplets impact nervous system development and disease.

Methods

Key resources table

Fly stocks used

Antibodies and lipid marker used

Deposited data

Software used

Method details

Fly husbandry

Flies were raised on standard molasses-based food at 25°C. Unless otherwise noted, wild type is an otherwise wild-type stock harboring the M{3xP3-RFP.attP}ZH-51D insert in an isogenic second chromosome.

Mutagenesis

A standard autosomal recessive forward genetic screen was carried out using 25-30 μm EMS to mutagenize a M{3xP3-RFP.attP}ZH-51D isogenic second chromosome. Homozygous mutant third instar larvae were visually screened under a standard fluorescent microscope for defects in CNS morphology. A detailed description of the screen and all other identified genes will be described elsewhere.

Creation of recombinant lines and identification of second site mutations in ifc-KO chromosome

Standard genetic methods were used to generate fly strains that contained specific combinations of GAL4 and UAS-linked transgenes in the ifc^js3or ifc-KO background. During this process, we uncovered that the ifc-KO deletion could be unlinked from at least one second chromosomal mutation that caused early larval lethality, resulting in homozygous ifc-KO flies that survived to late L3 to early pupa. A subsequent EMS-based F2 lethal non-complementation screen using the M{3xP3-RFP.attP}ZH-51D isogenic second chromosome as target chromosome and screening against the ifc-KO chromosome identified multiple mutations in four complementation groups that led to an early larval lethal phenotype. Whole genome sequencing identified these genes as Med15, lwr, Nle, and Sf3b1; all four genes reside within ∼90 Kb of each other in chromosomal bands 21D1-21E2 near the telomere in chromosome 2L, identifying the site of the associated lethal mutations and suggesting the actual lesion may be a small deletion that removes these genes. The following alleles of these genes are available at Bloomington Stock Center: Med15^js1(Q175*), Med15^js2 (Q398*), Med15^js3 (C655Y), lwr^js1 (I4N), lwr^js2(E12K), Nle^js1(W146R), Nle^js2 (L125P), Nle^js3 (Q242*), Sf3b1^js1 (R1160*), Sf3b1^js2 (Q1264*), Sf3b1^js3 (570 bp deletion at the following coordinates chr2L571720-572290; this deletion removes amino acids 1031 through 1222 and introduces a frameshift into the reading frame).

Gene rescue and in vivo RNAi phenocopy assays

To restrict UAS-linked transgene expression specifically to glia, we used the repoGAL4 driver line. To restrict UAS-linked transgene expression specifically to neurons, we paired the elavGAL4 driver lines, which activates transgene expression strongly in all neurons and moderately in glia, with repoGAL80, which blocks GAL4-dependent activation in glia. We used P{VSH330794} (VDRC 330794) for the RNAi experiments and UAS-ifc⁶⁸ and UAS-DEGS1 (BDSC 79200) for the gene rescue assays. All gene rescue experiments were performed in the ifc^js3/ifc-KO background with the UAS-transgene placed into the ifc^js3 background and the GAL4 drivers into the ifc-KO background.

The GAL4-UAS method was also used to assess the phenotype of each glial subtype. In combination with the UAS-Myr-GFP transgene, which labels cell membranes, we used the following GAL4 lines to trace the morphology of each glial subtype in wild type and ifc^-/- larvae: GMR85G01-GAL4 (perineurial glia; BDSC 40436), GMR54C07-GAL4 (subperineurial glia; BDSC 50472), GMR54H02-GAL4 (cortex glia; BDSC 45784), GMR56F03-GAL4 (ensheathing glia; BDSC 77469 and 39157), and GMR86E01-GAL4 (astrocyte-like glia; BDSC45914). The UAS-Myr-GFP transgene was placed into the ifc^js3background; each glial GAL4 line was placed into the ifc-KO background.

The GAL4-UAS method was used to assess the ability of UAS-ifc and UAS-DEGS1 transgenes to rescue the lethality of otherwise ifc mutant larvae. ifc-KO/CyO Tb; repo-GAL4/TM6B Tb males were crossed to each of the following lines: ifc^js3/CyO Tb; UAS-ifc; ifc^js3/CyO Tb; UAS-DEGS1; ifc^js3/CyO Tb. All adult flies were sorted into Curly and non-Curly and then counted; all non-Curly flies lacked the TM6B Tb balancer indicating they carried repo-GAL4 and thus were likely rescued to viability by glial expression of ifc or DEGS1. Standard Mendelian ratios were then used to predict the expected number of ifc mutant flies if all survived to adulthood. The total number of observed adult ifc mutant flies, identified by lack of Curly wings, was then divided by this number to obtain the percentage of ifc mutant flies that survived to adulthood. 2452 total flies were assayed for the UAS-ifc cross, 1303 for the UAS-DEGS1 cross, and 1030 for the control cross.

DNA sequencing

Genomic DNA was obtained from wild type larvae or larvae homozygous for each relevant mutant line and provided to GTAC (Washington University) or GENEWIZ for next-generation or Sanger sequencing.

RNA in-situ hybridization

ifc RNA probes for in-situ hybridization chain reaction (HCR) were designed and made by Molecular Instruments (HCR™ RNA-FISH v3.0).⁷³ Wild type CNS was harvested and fixed in 2% paraformaldehyde at late L3. The fixed CNS underwent gradual dehydration and rehydration, followed by standard hybridization and amplification steps of the HCR protocol.⁷³ For double labeling with antibody, the post-HCR labeled CNS was briefly fixed for 30 minutes prior to standard antibody labeling protocol to identify specific CNS cell type(s) with highly localized ifc _RNAs.74,75

MultiColor FlpOut labeling of glial subtypes

For glial labeling in the control background, the MCFO1 line was crossed to GMR-GAL4 driver lines for each glial subtype (Supplemental Table 1).³³ For glial labeling in ifc^-/- background, the MCFO1 line was placed in the ifc^JS3 background, each of the five glial-specific GMR-GAL4 lines were placed individually into the ifc-KO background, and then the MCFO, ifc^JS3 line was crossed to each glial specific GAL4, ifc-KO line. Flies were allowed to lay eggs for 24 hr at 25°C, and progeny were raised at 25°C for 4 days prior to heat-activated labeling. On day 4 after egg-laying, F1 larvae were incubated in a 37°C water bath for 5-minutes. When wild-type or ifc^-/- mutant larvae reached late L3, which was day 5-6 for control and day 9-10 for ifc^-/- mutants, the CNS was dissected, fixed, stained, and then analyzed under a Zeiss LSM 700 Confocal Microscope for the presence of clones, using Zen software.

Antibody generation

YenZym (CA, USA) was used as a commercial source to generate affinity purified antibodies against a synthetic peptide that corresponded to amino acids 85-100 (TLDGNKLTQEQKGDKP) of FABP isoform B. Briefly, the peptide was conjugated to KLH, used as an immunogen in rabbits to generate a peptide-specific antibody response, and antibodies specific to the peptide were affinity purified. The affinity-purified antibodies were confirmed to label cortex glia specifically based on comparison of antibody staining relative to Myr-GFP when a UAS-Myr-GFP was driven under control of the cortex specific glial GAL4 driver GMR-54H02 (Figure S3). The FABP antibodies are used at 1:500-1:1000.

Immunofluorescence and lipid droplet staining

Gene expression analysis was performed essentially as described in Patel.⁷⁶ Briefly, the larval CNS was dissected in PBS, fixed in 2.5% paraformaldehyde for 55 minutes, and washed in PTx (1xPBS, 0.1% TritonX-100. The fixed CNS was incubated in primary antibody with gentle rocking overnight at 4°C. Secondary antibody staining was conducted for at least two hours to overnight at room temperature. All samples were washed in PTx at least five times and rocked for an hour before and after secondary antibody staining. A detailed list of the primary and secondary antibodies is available in Supplemental Table 2. Dissected CNS were mounted either in PTx or dehydrated through an ethanol series and cleared in xylenes prior to mounting in DPX mountant.⁷⁷ All imaging was performed on a Zeiss LSM-700 Confocal Microscope, using Zen software.

For lipid droplet staining, the fixed CNS was incubated for 30 minutes at room temperature at 1:200 dilution of 1mg/mL BODIPY™ 493/503 (Invitrogen: D3922). It was then rinsed thoroughly in PBS and immediately mounted for imaging on a Zeiss LSM-700 Confocal Microscope, using Zen software.

Transmission electron microscopy (TEM) imaging

For transmission electron microscopy (TEM), samples were immersion fixed overnight at 4°C in a solution containing 2% paraformaldehyde and 2.5% glutaraldehyde in 0.15 M cacodylate buffer with 2 mM CaCl2, pH 7.4. Samples were then rinsed in cacodylate buffer 3 times for 10 minutes each and subjected to a secondary fixation for one hour in 2% osmium tetroxide/1.5% potassium ferrocyanide in cacodylate buffer. Following this, samples were rinsed in ultrapure water 3 times for 10 minutes each and stained overnight in an aqueous solution of 1% uranyl acetate at 4°C. After staining was complete, samples were washed in ultrapure water 3 times for 10 minutes each, dehydrated in a graded acetone series (50%, 70%, 90%, 100% x4) for 15 minutes in each step, and infiltrated with microwave assistance (Pelco BioWave Pro, Redding, CA) into Spurr’s resin. Samples were then cured in an oven at 60°C for 72 hours and post-curing, 70 nm thin sections were cut from the resin block, post-stained with uranyl acetate and Sato’s lead and imaged on a Transmission Electron Microscope (JEOL JEM-1400 Plus, Tokyo, Japan) operating at 120 KeV.

Lipidomics

Untargeted lipidomics analysis was conducted on whole larva and dissected CNS of wild type and ifc^-/- mutants at the late-third instar stage. Five replicates were prepared for each set of experiments. For whole larvae, at least 15 larvae of each genotype were used for each replicate. For the dissected CNS, at least 50 wild type and 60 ifc^-/- CNS were used per replicate. Immediately following collection or dissection, larvae and the dissected CNS were flash frozen in liquid nitrogen and placed at −80°C.

Lipids were extracted from frozen whole larvae and dissected larval CNS samples by using an Omni Bead Ruptor Elite Homogenizer using acetonitrile:methanol:water (2:2:1; 40 μL/mg tissue). Two Ultrahigh performance LC-(UHPLC)/MS systems were used in this work: a Thermo Vanquish Flex UHPLC system with a Thermo Scientific Orbitrap ID-X and an Agilent 1290 Infinity II UPLC system with an Agilent 6545 QTOF as described previously.⁷⁸ Lipids were separated on a Waters Acquity HSS T3 column (150 x 2.1 mm, 1.8 mm). The mobile-phase solvents were composed of A: 0.1% formic acid, 10 mM ammonium formate, 2.5 μM medronic acid in 60:40 acetonitrile:water; and B = 0.1% formic acid, 10 mM ammonium formate in 90:10 2-propanol:acetonitrile. The column compartment was maintained at 60 °C. The following linear gradient was applied at a flow rate of 0.25 mL min-1: 0 – 2 min, 30% B; 17 min, 75% B; 20 min, 85% B; 23 – 26 min, 100% B. The injection volume 4 μL for all lipids analysis. Data was acquired in positive ion.

LC/MS data were processed and analyzed with the open-source Skyline software.⁷⁹ Lipid MS/MS data were annotated with Agilent Lipid Annotator software.

RNA sequencing and analysis

To determine the CNS-specific transcriptional changes upon loss of ifc, RNA-seq was conducted on five replicates of dissected CNS tissue derived from wild type and ifc^-/- mutant late-third instar larvae. For each replicate, roughly 30-35 dissected CNS of wild type or ifc-/- larvae were used. Invitrogen™ RNAqueous™-Micro Total RNA Isolation Kit (AM1931) was used to extract RNA. Agilent 4200 TapeStation system was used for RNA quality control.

Samples were prepared according to library kit manufacturer’s protocol, indexed, pooled, and sequenced on an Illumina NovoSeq. Basecalls and demultiplexing were performed with Illumina’s bcl2fastq software and a custom python demultiplexing program with a maximum of one mismatch in the indexing read. RNA-seq reads were then aligned to the Ensembl release 76 primary assembly with STAR version 2.5.1a.⁸⁰ Gene counts were derived from the number of uniquely aligned unambiguous reads by Subread:featureCount version 1.4.6-p5.⁸¹ Isoform expression of known Ensembl transcripts were estimated with Salmon version 0.8.2.⁸² To find the most critical genes, the raw counts were variance stabilized with the R/Bioconductor package DESeq2⁸³ and was then analyzed via weighted gene correlation network analysis with the R/Bioconductor package WGCNA.⁸⁴

Statistics

All data are presented as mean ± SEM. Statistical significance between groups was determined using Student’s t-test or one-way ANOVA with multiple comparisons, and with varying levels of significance assessed as * P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001, and ns, not significant.

Supplemental figures and information

Figure S1.

Figure S2.

Figure S3.

Figure S4.

Figure S5.

Figure S6.

Figure S7.

Figure S8.

Figure S9.

Figure S10.

Figure S11.

Figure S12.

Figure S13.

Figure S14.

Table 1.

Acknowledgements

We thank the Iowa Developmental Studies Hybridoma Bank for antibodies, and the Bloomington Stock Center, Vienna Drosophila Research Center, FlyORF, and the National Institutes of Genetics stock center in Japan for countless fly lines. We thank Drs. Christian Klambt, Chih-Chiang Chan, Dion Dickman for reagents. We thank Dr. Tristan Qingyuan Li for comments on the manuscript. We thank the Genome Technology Access Center at Washington University for next-generation sequencing and analysis of RNA-seq samples. We thank Dr. Sanja Sviben, Gregory Strout, and John Wulf II for assistance in TEM studies conducted at the Washington University Center for Cellular Imaging, which is supported in part by Washington University School of Medicine, The Children’s Discovery Institute of Washington University and St. Louis Children’s Hospital (CDI-CORE-2015-505 and CDI-CORE-2019-813), the Foundation for Barnes-Jewish Hospital (3770) and the Washington University Diabetes Research Center (NIH P30 DK020579).

Additional information

Author Contributions

BAW, HL, Yi Zhu, and JBS completed the genetic screen that identified the ifc alleles and completed the initial genetic characterization of these alleles. Yuqing Zhu, JD, and JBS completed all other experiments except for generating the lipidomics data via mass spectrometry, which were completed by KC and GJP. Yuqing Zhu and JBS wrote the manuscript and compiled figures with help from JD with figures. All authors read, commented on, and edited the manuscript.

Funding

This work was supported by grants from the National Institutes of Health (NS036570) to J.B.S., (NS122903) to H.L., and (R35 ES2028365) to G.J.P.

Funding

National Institutes of Health (NS036570)

National Institutes of Health (NS122903)

National Institutes of Health (ES2028365)