Introduction

The gradual accumulation of microtubule-associated protein tau (MAPT/tau) aggregates is a hallmark of Alzheimer’s disease (AD) and related tauopathies. Misfolded tau species exhibit prion-like behavior by self-replicating and spreading from cell to cell^1,2. This contributes to the progression of pathology and neurotoxicity, eventually culminating in widespread neuronal dysfunction and degeneration³. At the molecular level, tau aggregates replicate by templating the conversion of native tau into an amyloid conformation, promoting its accumulation into fibrillar aggregates. For this to occur at the cellular level, seeding-competent tau species (or tau “seeds”) must be released from a donor cell and taken up into the cytosol of the neighboring receiving cell in order to come into direct contact with the cytosolic native tau protein^4–6.

In this process, the autophagy-lysosomal pathway (ALP), which is an important clearance route for tau⁷, appears to play a central role. Studies have demonstrated impaired ALP function in the brains of tauopathy patients as well as in animal and cell models, showing that the accumulation of abnormal autolysosomal and endolysosomal vesicles correlating with neuronal toxicity^8–11. Inhibition of autophagic clearance of tau increases its secretion and spreading¹². In addition, the accumulation of misfolded tau within endolysosomes leads to a destabilization and rupture of these vesicles^13–18. While intact endolysosomes normally restrict tau seeds from reaching cytosolic monomers, membrane rupture allows their escape. Recent studies highlight this escape as a critical rate-limiting step in seeded tau propagation^13–18. Endolysosomal damage not only promotes the propagation of pathological aggregates, but also causes the release of hydrolytic enzymes into the cytosol, leading to cellular damage and death¹⁹. Although recent efforts to identify pathways involved in endolysosomal damage and repair have intensified, many aspects of these processes remain unknown.

To identify cellular factors that are critical for the integrity of endolysosomal vesicles, we recently performed an unbiased genome-wide RNAi screen in C. elegans²⁰. One of the pathways identified was sphingolipid (SL) metabolism. SLs constitute a large and diverse class of lipids that are involved in various physiological processes^21,22. They consist of two main building blocks, a long-chain base with a serine backbone and an attached long acyl chain. This characteristic chemical structure mediates unique biophysical properties. SL biosynthesis and degradation rely on a specialized enzymatic machinery distinct from that of other lipids²³ (Figure 1A). Functionally, SLs, along with glycerolipids and sterols, serve as structural components of cell membranes, contributing to the stability and fluidity of membranes and to the organization of microdomains such as lipid rafts. Beyond their structural role, SLs serve as bioactive molecules participating in cellular signaling pathways that control cell growth, differentiation, apoptosis, and intercellular communication²⁴. Abnormalities in SL metabolism have been observed during aging and in neurodegenerative conditions, including AD, underscoring their potential significance in the pathogenesis of these disorders^25,26. Interestingly, recent evidence suggests that sphingolipid accumulation disturbs the endolysosomal pathway and induces or potentiates endolysosomal membrane rupture^27,28; however, the mechanisms underlying membrane destabilization remain unclear.

Figure 1.

Here we investigated the role of SL metabolism in the maintenance of endolysosomal vesicle integrity. Our results revealed that silencing of SL metabolism genes led to a marked reduction in membrane fluidity, particularly in endolysosomal vesicles, thereby increasing their fragility and rupture. The accumulation of aggregated tau had an additive effect and led to a further reduction in endomembrane fluidity, which aggravated the damage to the endolysosomal compartment. Increased membrane rigidity also facilitated seeded tau aggregation. Improvement of membrane fluidity through supplementation of polyunsaturated fatty acids (PUFAs) counteracted tau propagation in cells and tau-associated neurotoxicity in C. elegans.

This study highlights the interplay between lipid metabolism and proteostasis and provides insights into how disturbances in sphingolipid homeostasis contribute to endolysosomal dysfunction and tau pathology. In addition, it provides a possible mechanistic explanation for the beneficial effect of PUFA supplementation that has been observed in AD patients and suggests that restoring membrane fluidity may help mitigate tau toxicity and disease progression.

Results

Knockdown of sphingolipid metabolism genes induces lysosomal rupture

Several of the strongest hits identified in the unbiased genome-wide RNAi screen we performed to identify modulators of endolysosomal vesicles integrity were genes involved in SL metabolism (Figure 1A). Knockdown (KD) of these genes (the serine incorporator R11H6.2, the 3-ketodihydrosphingosine reductase Y37E11AM.3, the hydroxyacyl-CoA dehydratase hpo-8, the cytochrome b5 reductases hpo-19 and T05H4.4, the saposin spp-10, the sphingosine kinase sphk-1, and the sphingosine-1-phosphate lyase spl-1) via RNA interference (RNAi) induced endolysosomal membrane damage as detected by a galectin puncta assay²⁹ (Figure 1B-D). Under steady-state conditions, human galectin-3 fused to superfolder-GFP (sfGFP::LGALS3) exhibits an even distribution within the cytosol of hypodermal cells. Upon endolysosomal vesicle damage and exposure of their lumen to the cytosol, sfGFP::LGALS3 binds exposed β-galactoside and re-localizes, yielding visible puncta²⁹.

The screen identified distinct steps in sphingolipid metabolism catalyzed by non-redundant genes or targeted by a single RNAi clone, as in the case of hpo-19/T05H4.4 which are both depleted by hpo-19 RNAi. However, other steps in this pathway, which are mediated by two or more proteins may have been missed in the screen. Indeed, simultaneous KD of selective redundant genes using co-RNAi revealed additional genes functioning in de novo sphingolipid biosynthesis that also induced endolysosomal rupture, including the serine palmitoyltransferases sptl-1 and –3 and the ceramide synthases hyl-1 and –2 (Figure S1A). In addition, the co-KD of all four glucocerebrosidases gba-1-4 or only gba-2-4, also triggered endolysosomal rupture (Figure S1B). Hence, these findings suggest that the KD of genes involved in both biosynthesis and degradation of SLs contributes to endolysosomal rupture, highlighting the tightly regulated nature of sphingolipid metabolism. Any disruption in this balance poses challenges to endolysosomal integrity.

KD of sphingolipid metabolism genes particularly affects membrane fluidity of endolysosomal vesicles

SLs are important components of eukaryotic cell membranes and genetic modulation of SL metabolism likely affects the lipid composition and biophysical properties of membranes. Similar to their mammalian isoforms, the N-acyl chains of C. elegans SLs contain predominantly long and saturated fatty acid chains^30,31, which tend to pack tightly within the lipid bilayer. As such, a higher proportion of SLs generally reduces membrane fluidity, leading to increased order, rigidity, or viscosity³². Consequently, KD of the hits involved in SL degradation should lead to SL accumulation and reduced membrane fluidity, while KD of the hits involved in SL biosynthesis should have the opposite effect.

To evaluate the membrane fluidity of endolysosomal membranes, fluorescence recovery after photobleaching (FRAP) was employed on C. elegans that express the Lysine/Arginine Transporter 1 (LAAT-1) tagged with mCherry, which localizes to the lysosomal membrane³³. FRAP enables the assessment of the lateral mobility of LAAT-1 within the lysosomal membrane, thereby indirectly measuring its fluidity. We focused our analysis on the hypodermis, where we expressed the sfGFP::LGALS3 reporter and observed endolysosomal rupture following KD of the SL metabolism genes (Figure 1C, D). Interestingly, KD of all of our hits significantly increased the time required to recover half of the maximum fluorescence intensity (t_half), indicating less fluid lysosomal membrane in the hypodermis (Figure 2A-C, E, F). With some gene KDs, we also observed a decrease in the maximal recovered signal after bleaching (Figure 2D, G). The decrease in lysosomal membrane fluidity was not specific to the hypodermis as the KD of spl-1 also increased the t_half of LAAT-1::mCherry in intestinal lysosomes (Figure S2A-D). To discern whether the detrimental effect of KD on membrane fluidity is confined to lysosomal membranes or extends to other cellular membranes, we utilized a C. elegans strain expressing a plasma membrane-anchored GFP³⁴. Intriguingly, this reporter showed no change in the t_half with any KD and only spl-1 KD led to decrease in the maximal recovered signal (Figure 2H-K). Hence, the endolysosomal membrane seems to be particularly sensitive to a disruption of SL metabolism in contrast to the plasma membrane. Dysregulation of both, synthesis and degradation of SLs reduces the fluidity of the endolysosomal membrane and makes it more susceptible to rupture.

Figure 2.

KD of sphingolipid metabolism genes affects membrane fluidity in human cells

Next, we asked whether dysregulation of SL metabolism exerts similar effects in human cells. To this end, we employed HEK293T cells and C-Laurdan dye to evaluate membrane fluidity (Figure 3A). This approach leverages the unique properties of C-Laurdan, which emits fluorescent light of varying wavelengths in response to changes within the phospholipid bilayer, particularly reflective of membrane fluidity^35,36. The shift in emission profile between liquid-disordered and liquid-ordered phases allows a quantitative assessment of the membrane order by calculating the ratiometric relationship of the fluorescence intensity recorded in two spectral channels, known as a generalized polarization (GP) value³⁵ (Figure 3A). Decreasing GP values are indicative of increasingly fluid membrane while increasing GP values denote increased rigidity. Of note, we exposed the cells to C-Laurdan for an extended time of 2 hours (instead of 30 min as recommended in the original protocol), to promote internalization of the dye and thus increase the staining of endolysosomal membranes relative to the plasma membrane³⁵. Upon KD of SPHK2, one of the human homologs of the C. elegans sphk-1 gene, the C-Laurdan fluorescence GP ratio augmented significantly, indicating increased membrane rigidity (Figure 3B, D and Figure S3A). Of note, Lipofectamine 2000 was used to deliver siRNA into the cells and lipid-based transfection reagents have been shown to impact endolysosomal membranes³⁷. Indeed, Lipofectamine alone slightly increased membrane fluidity (Figure 3C). Taken together, FRAP and C-Laurdan analyses revealed that disrupted SL homeostasis led to more ordered and rigid endolysosomal membranes, in both C. elegans and human cells.

Figure 3.

Fibrillar tau increases membrane rigidity and exacerbates endolysosomal damage

To investigate whether disruption of SL homeostasis also affects endolysosomal integrity in human cells, we generated a HEK293T cell line stably expressing the same sfGFP-LGALS3 reporter used for the galectin puncta assay in C. elegans. To further mimic the C. elegans model, in which aggregation-prone tau fragments are chronically transferred from the touch receptor neurons into the hypodermis where the sfGFP-LGALS3 reporter is expressed²⁰, recombinant 1N4R tau fibrils were added to sfGFP-LGALS3 expressing HEK293T cells on top of the KD of SPHK2. Interestingly, both the addition of tau fibrils alone and the KD of SPHK2 alone already caused rupture of endolysosomes in sfGFP-LGALS3 cells (Figure 3D, E and Figure S3B). The combination of both had an additive effect (Figure 3D, E and Figure S3B).

We therefore asked whether the KD of SLs alone would also already rupture hypodermal endolysosomes in C. elegans and whether accumulation of transmitted tau would exacerbate this process. To answer this question, we determined the amount of hypodermal sfGFP-LGALS3 foci in F3^ΔK281::mCherry C. elegans compared to control animals expressing only the mCherry tag. KD of Y37E11AM.3 alone did not increase lysosomal rupture compared to the EV control RNAi, only the presence of tau spreading resulted in more lysosomal rupture relative to the control (Figure 3F). KD of the remaining hits resulted in near 100% animals with hypodermal sfGFP-LGALS3 foci in both the F3^ΔK281::mCherry and mCherry control strains. To test whether the additional negative impact of the F3 tau fragment might be masked by the strength of the RNAi clones, different dilutions of the RNAi bacteria were tested. Under conditions where the RNAi KD did not saturate the readout of the assay, KD of spl-1, hpo-8, and hpo-19 alone already affected lysosomal integrity (Figure 3G). Moreover, F3^ΔK281::mCherry transgenic animals showed a significantly higher level of endolysosomal rupture upon KD of spl-1, hpo-8, and hpo-19 compared to the mCherry control (Figure 3G). Diluting the R11H6.2 RNAi clone showed no significant differences, and KD of sphk-1 still resulted in 100% endolysosomal rupture-positive animals in both genetic backgrounds even after dilution of the RNAi bacteria to below 5% and could therefore not be assessed. In conclusion, chronic transmission of the F3^ΔK281::mCherry fragment from the touch receptor neurons to the hypodermis impaired the endolysosomal system exacerbating the rupture due to perturbation of SL homeostasis.

Since disruption of SLs metabolism led to an increase in membrane rigidity resulting in endolysosomal rupture, and fibrillar tau also ruptured endolysosomal membranes (Figure 3D, E)¹⁴, we asked whether tau itself could also affect membrane fluidity. To test this, membrane fluidity was assessed for somas of the posterior (PLM) touch receptor neurons of C. elegans using C-Laurdan dye. Indeed, expression of the tau fragment led to a decrease in membrane fluidity compared to expression of the mCherry tag alone (Figure 3H, I). Also, HEK293T cells incubated with tau fibrils for 5 hours exhibited an increased GP index (Figure 3J, K). Importantly, this was specific to fibrillar tau, as monomeric tau had no effect on membrane fluidity (Figure 3L and Figure S3C). Thus, aggregated forms of tau can cause rigidification of the endolysosomal membrane, resulting in increased fragility and rupture.

Membrane rigidification exacerbates seeded tau aggregation

The packing of fatty acids within the membrane influences membrane fluidity. Saturated fatty acids (SFAs) have straight hydrocarbon chains forming a tightly packed, highly ordered and rigid membrane. To investigate the role endomembrane lipid homeostasis plays in tau propagation, we introduced an SFA to rigidify membranes. Incubating HEK293T cells with the SFA palmitic acid (PA) conjugated to BSA increased membrane rigidity (Figure 4A, B). In addition, cells incubated with PA and tau fibrils had an even more rigid membrane than cells incubated with tau fibrils alone (Figure 4C, D).

Figure 4.

An important step in tau propagation is tau seeds escape from the endolysosomal compartment, which can then propagate themself by recruiting native monomeric tau. Therefore, we asked if the PA-induced loss of membrane fluidity impairs endolysosomal integrity. Worms grown on plates supplemented with PA displayed an increase in endolysosomal rupture (Figure 4E). In cells, the addition of PA alone did not lead to endolysosomal rupture; however due to compounding impact of PA and tau fibrils on membrane fluidity, the addition of PA increased fibrillar tau-induced endolysosomal rupture (Figure 4F, G). Hence, loss of membrane fluidity mediated by fibrillar tau and/or SFAs negatively affects endolysosomal integrity, resulting in increased endolysosomal rupture.

Impairment of endolysosomal integrity may enhance tau propagation as tau could escape from the endolysosomal compartments, reach the cytosol and recruit native monomeric tau, promoting seeded tau aggregation. Given the observed membrane rigidification after pre-loading with SFAs, we investigated whether membrane lipid composition influences tau propagation using a tau biosensor cell line expressing Venus-tagged full-length P301S mutant 0N4R tau (tau-Venus)^38,39. PA-induced membrane rigidification led to an increase in tau-Venus foci upon seeding with tau fibrils (Figure 4H, I). Moreover, KD of SPHK2 also increased the formation of tau-Venus foci due to SPHK2 KD-mediated membrane rigidification (Figure S4C, D). This highlights a positive correlation between membrane rigidity and seeded tau aggregation. Changes in membrane fluidity, which compromise endolysosomal integrity may promote the entry of seeds into the cytosol. This would consequently foster the seeded propagation of tau aggregation.

Increasing membrane fluidity by PUFAs rescues neurotoxic phenotypes

Given that membrane stiffening exacerbates tau propagation, we wondered whether we could prevent this by improving membrane viscosity. While SFA allow for tightly packed lipids, unsaturated fatty acids have one or more double bonds in their hydrocarbon chain that introduce kinks in their structure, resulting in loosely packed lipids and increased membrane fluidity. Therefore, we investigated whether we could rescue tau pathology using polyunsaturated fatty acids (PUFAs). Incubating HEK293T cells with the ω-3 PUFA α-linolenic acid (ALA) increased membrane fluidity (Figure 5A and Figure S5A). Moreover, pre-loading with ALA prevented tau-induced membrane rigidification and decreased the GP-index of membranes even after the addition of tau fibrils (Figure 5B, C). Importantly, pre-loading cells with ALA also counteracted fibrillar tau-mediated endolysosomal rupture (Figure 5D, E and Figure S5B).

Figure 5.

Since ALA prevented endolysosomal rupture, we expected it to also prevent seeded tau aggregation. Indeed, tau-Venus cells pre-loaded with ALA exhibited significantly fewer foci when exposed to tau fibrils (Figure 5F and Figure S5C). These data show that ALA supplementation counteracts tau-mediated decrease in membrane fluidity, thereby preventing endolysosomal rupture and decreasing seeded tau aggregation.

Finally, we asked whether ALA would also reduce the toxicity associated with aggregated tau. Expression of F3^ΔK281::mCherry in touch receptor neurons resulted in an age-dependent decline in the response to gentle touch from day 4, corresponding to the first day of adulthood to day 8, corresponding to day 5 of adulthood (Figure 5H, and Figure S5D)²⁰. Addition of ALA to the growth medium mitigated this loss of function (Figure 5H and Figure S5D) and also reduced neurotoxicity (Figure 5I, J). Thus, supplementation with ALA not only reduced endolysosomal damage but also attenuated neurotoxic phenotypes associated with aggregated tau.

Discussion

In a previous genome-wide screen in C. elegans, we identified genes involved in SL metabolism as crucial for maintaining endolysosomal integrity²⁰. In this study we explored the mechanisms by which disruption of SL homeostasis leads to endolysosomal rupture and how it affects tau pathology. We found that imbalanced SL metabolism leads to decreased membrane fluidity, and subsequent endolysosomal rupture. This promoted seeded tau aggregation in a biosensor cell line exposed to fibrillar tau, implying that dysregulation of SL metabolism might boost disease progression by enhancing propagation of tau aggregation.

The central nervous system (CNS) is particularly rich in lipids. Approximately 50% of the brain’s dry weight consists of lipids with high concentrations of cholesterol, phospholipids and sphingolipids ^40,41. Alterations in lipid and SL metabolism have been observed not only during the aging process, but also in various late-onset neurodegenerative diseases including AD^26,42. While aging is the major risk factor for AD, one of the most significant genetic risk factors for AD is the apolipoprotein E (ApoE) polymorphism, specifically the ApoE4 allele⁴³. ApoE is the primary lipoprotein in the brain, crucial for lipid transport and metabolism. Besides cholesterol, the ApoE4 variant has been shown to influence SL metabolism, thereby affecting the pathogenesis of AD from its early stages^26,42,44. While disturbances in SL homeostasis are increasingly recognized as a contributing factor to the onset and progression of the disease^26,42, the underlying mechanisms are poorly understood.

Our results show that reducing the expression of genes involved in both, biosynthesis and degradation of SLs significantly reduced the fluidity of endolysosomal membranes, making them more fragile and prone to rupture. KD of the saposin spp-10, the sphingosine kinase sphk-1, and the sphingosine-1-phosphate lyase spl-1 blocks the degradation of complex SLs and should lead to an increase in their relative abundance, and therefore to a decrease in membrane fluidity. KD of the very-long-chain (3R)-3-hydroxyacyl-CoA dehydratase hpo-8, which is involved in fatty acid elongation, has been demonstrated to promote the incorporation of long-chain UFAs into phospholipids, thereby restoring plasma membrane fluidity⁴⁵. The absence of hpo-8 leads to membrane rigidification⁴⁵. Conversely, KD of the serine incorporator R11H6.2 and the 3-ketodihydrosphingosine reductase Y37E11AM.3, which function in the SL de novo synthesis pathway, should theoretically decrease total SL content and consequently increase membrane fluidity. The cytochrome b5 reductases hpo-19 and T05H4.4, required for lipid desaturation, are involved in the conversion of dihydroceramide to ceramide and the biosynthesis of PUFAs⁴⁶; KD of these genes is expected to reduce the levels of SLs and PUFAs, which could theoretically either increase or decrease membrane fluidity, depending on which metabolic pathway is predominantly affected.

Our observation that KD of all these genes results in decreased membrane fluidity is seemingly counterintuitive. However, SLs form a complex metabolic network and can interconvert, allowing a dynamic adaptation of SL levels. In addition, the cell likely compensates for the loss of certain SLs by increasing the biosynthesis of other lipid molecules. This dynamic interplay makes it difficult to predict how individual genes affect the endolysosomal membrane lipid composition. Genes that are involved in apparently opposing steps of SL metabolism could therefore still lead to an accumulation of similar SL species.

Future lipidomic analyses are necessary to determine the SL and lipid content of endolysosomal membranes in our model systems. Such studies will provide a clearer understanding of how genetic manipulation of SL metabolism affects endomembrane lipid composition and could pinpoint the exact species likely responsible for the observed membrane rigidification.

AD brains exhibit elevated levels of sphingosine, ceramide 1-phosphate, and ceramide compared to control subjects⁴². Multivariate analysis and machine learning also identified these SL species as key contributors to AD⁴². These and other analyses were performed in different brain regions but without cell type or subcellular resolution. We found that disruption of C. elegans SL homeostasis primarily affected endolysosomal membranes and had little effect on plasma membrane fluidity. The lysosomal membrane is particularly rich in sphingolipids⁴⁷, which seems to make it more susceptible to disruption of SL homeostasis. Studies in yeast have shown that increased membrane lipid saturation affects the vacuole —the functional counterpart of mammalian lysosomes—as well as the nuclear envelope, whereas the plasma membrane and mitochondria remained relatively unaffected⁴⁸. These findings indicate that cellular membranes vary in their resilience to lipid saturation stress, potentially due to differences in lipid composition and protective mechanisms^48,49. To better understand the implications of SL alterations in AD, it would be important to examine whether the observed changes primarily originate from endolysosomal membranes or also from other organelles, such as the endoplasmic reticulum and the nucleus, or from the plasma membrane.

Intriguingly, expression of the aggregation-prone F3 tau fragment also enhanced membrane rigidity with SL homeostasis disruption having an additive effect, exacerbating endomembrane damage. Thus, alterations in SL metabolism could be not only a cause but also a consequence of AD related pathology^26,42. The accumulation of misfolded tau in lysosomes could impair the lysosomal clearance capacity. This dysfunction may reduce the activity of acid hydrolases, leading to an accumulation of biomolecules, including membrane lipids. Impairment of membrane lipid metabolism could in turn further compromise lysosomal integrity and function. Both processes – accumulation of SLs and protein aggregates – may create a vicious self-promoting cycle of lysosomal dysfunction that eventually leads to neurodegeneration.

Supplementation of SFAs exacerbated the negative impact of the F3 tau fragment on endomembrane fluidity. Given the high levels of saturated and trans fats in the Western diet⁵⁰, it is plausible that dietary factors also contribute to the dysregulation of lipid homeostasis in the brain. The Mediterranean UFA-rich diet has been associated with healthy aging, while reduced UFA brain levels have been correlated with increased tau and Aβ pathology and decreased cognitive function in AD^51,52. Restoring membrane fluidity through the supplementation of PUFAs mitigated tau propagation in cells and reduced tau-associated neurotoxicity in C. elegans. These results suggest a potential mechanism for the beneficial effects of PUFA supplementation observed in AD patients^53,54. Future studies are needed to evaluate whether these benefits are related to an increase in the fluidity of endolysosomal membranes and the consequent prevention of their damage by protein aggregates and/or impaired SL homeostasis.

In sum, this study revealed an important role of endolysosomal membrane homeostasis in disease progression providing a possible mechanism for the detrimental effect of SL imbalance and beneficial effect of PUFA supplementation that have been observed in AD patients.

Methods

Maintenance of C. elegans and age synchronization

All animals were cultured using standard methods⁵⁵. If not otherwise indicated, worms were grown on nematode growth medium (NGM) plates seeded with E. coli strain OP50 at 20 °C. For RNAi, NGM medium were supplemented with Ampicillin, Tetracycline, and IPTG, seeded with the respective HT115 E. coli RNAi clones and grown at 20 °C. Animals were age-synchronized by bleaching. Briefly, gravid adults were dissolved in 20% sodium hypochlorite solution. The surviving C. elegans embryos were hatched overnight in M9 buffer with gentle rocking at 20 °C. The next day, appropriate amounts of L1 larvae were added to the plates.

C. elegans fatty acid supplementation

Polyunsaturated fatty acids were dissolved in EtOH and stored in the dark at –20 °C under a N₂ atmosphere to prevent oxidation. NGM was cooled after autoclaving to 55 °C. To enhance fatty acid distribution, NP-40 substitute (0.001% v/v) was added to the media and subsequently either 0.3 mM of the desired fatty acid solution or the ethanol solvent as control and stirred for 5 min before plate pouring ⁵⁶. Palmitic acid (PA) was supplemented by enriching the OP50 bacterial food source. Overnight cultures were grown in the presence of 2 mM of PA dissolved in EtOH or EtOH. The next day, OP50 bacteria were pelleted by centrifugation (10 min at 4500x g) and washed 2x with M9 buffer⁵⁷. 10x concentrated bacteria (in M9) were then seeded on NGM plates.

Quantification of lysosomal rupture (galectin puncta assay) in C. elegans

Quantification of animals showing hypodermal lysosomal rupture was done as previously described^29,58. Briefly, age-synchronized animals expressing the sfGFP::LGALS3 construct in the hypodermis were seeded on OP50 or RNAi NGM plates. Scoring was done using a Leica M205 FA widefield binocular microscope. 20-50 worms were analyzed per replicate.

Mounting of live animals for imaging

Synchronized animals were mounted on 8-10% agarose in M9 buffer pads with a drop of mounting mix (2% (w/v) levamisole and 50% (v/v) nanosphere size standards (Thermo Fisher)) and covered with a coverslip.

Fluorescence recovery after photobleaching (FRAP)

FRAP measurements to estimate membrane fluidity were done following a published protocol⁵⁹. Animals were synchronized by bleaching and grown until the second day of adulthood on the indicated RNAi or empty vector (EV) control plates. Directly before the FRAP experiments, animals were mounted at the Zeiss LSM 780 confocal microscope. FRAP measurements of either GFP enriched in the intestinal plasma membrane, or LAAT-1::mCherry located in intestinal or hypodermal lysosomal membranes were taken with 40X Water immersion objective. For the lysosomal membrane, mCherry-positive membranes were photobleached over a circular area (seven pixel radius) using 10 iterations of the 561 nm DPSS laser with 100% laser power transmission. Images were collected at a 12-bit intensity resolution over 128 × 128 pixels (digital zoom 6X) using a pixel dwell time of 3.15 μsec with 2% laser power transmission. For the plasma membrane, GFP-positive membranes were photobleached over a circular area (10 pixel radius) using 20 iterations of the 488 nm Argon laser with 100% laser power transmission. Images were collected at a 12-bit intensity resolution over 128 × 128 pixels (digital zoom 6X) using a pixel dwell time of 3.15 μsec with 10% laser power transmission. At least 6 pre-bleaching images were collected before the region of interest was bleached. The recovery of fluorescence was traced for 25s. Fluorescence recovery and T_half were calculated as previously described in Svensk et al.⁶⁰.

Touch sensory assay and neurotoxicity in touch receptor neurons

Touch sensory assay was done as previously described⁶¹. Briefly, touch sensitivity was tested by gently stroking the tip of an eyebrow hair attached to a toothpick transversely across the anterior or posterior half of an animal. A touch-sensitive animal was one that stopped/moved away from the stimulus. 10 worms were examined for each strain in 3 biological replicates with 5 strokes in the anterior and posterior. For neurotoxicity scoring of the PLM neurons, live animals were imaged using an Olympus IXplore SpinSR Confocal microscope with an UplanS Apo 60x/1.30 Silicon oil objective. The neurotoxicity score was determined by adding a score of 1 for each of the following phenotypes: soma outgrowth, process branching, process break, wavy process. 15 animals were examined at each day of age in 3 independent biological replicates.

Cell culture

Cells were cultured in DMEM containing high glucose, GlutaMAX Supplement, pyruvate (Gibco), and 10% FBS (Gibco) at 37 °C and 5% CO₂. Regular mycoplasma tests were performed (GATC Biotech). HEK293T cells expressing Venus-tagged full-length P301S mutant 0N4R tau (0N4R tauP301S-Venus) were kindly provided by Dr. William A. McEwan, Cambridge University³⁸.

Cloning of mammalian expression constructs and generation of sfGFP-LGALS3 HEK293T cells

HEK293T cells were cultured in DMEM containing high glucose, GlutaMAX supplement, pyruvate (Gibco), and 10% FBS (Gibco) at 37 °C and 5% CO₂. Cells expressing sfGFP-LGALS3 were generated as follows. sfGFP::LGALS3 was amplified from the C. elegans expression plasmid pPD49.26 by PCR and cloned into the pLenti PGK Puro DEST (w529-2) vector backbone (Addgene plasmid #19068) by MultiSite Gateway cloning according to the manufacturers protocol. pLenti PGK Puro DEST (w529-2) was a gift from Eric Campeau & Paul Kaufman⁶². After each cloning step plasmids isolated from bacterial colonies grown on selective media were extracted with the GenEluteTM Five-Minute Plasmid Miniprep Kit (Qiagen) and validated by sequencing. HEK293T sfGFP-LGALS3 were generated using a 3^rd generation lentiviral system. Lentivirus was produced by transiently transfecting HEK293T cells with the pLenti PGK Puro DEST, MDL-gagpol, RSV-REV and IVS-VSVG at a ratio of 5:3:1:1, respectively. Cell media was changed 24 hours after transfection. Viral supernatant was collected 72 h post-transfection. After centrifugation at 2000 x g for 5 min, the supernatant containing the virus was concentrated by adding the PEG-it Virus Precipitation Solution (Systems Biosciences) followed by overnight incubation at 4 °C. The precipitated virus was pelleted by centrifugation (1500 x g, 30 min, 4°C), resuspended in DMEM containing 25 mM HEPES buffer and stored at –80 °C. The lentivirus was added to HEK293T followed by treatment with Puromycin to select successfully transduced HEK293T cells stably expressing sfGFP-LGALS3.

Quantification of lysosomal rupture (galectin puncta assay) in cells

Cells were imaged on an Olympus IXplore SpinSR Confocal microscope with an UplanS Apo 60x/1.30 Silicon oil objective. To detect the number of foci in an image, a difference of gaussian blur was applied to max projections to isolate foci from cytsosolic sfGFP signal. Thresholding was applied using ‘moments’ threshold on FIJI. Number of foci was counted using the ‘analyze particles’ function. Cell number was counted manually to obtain number of foci/cell for each image.

C-Laurdan dye measurement of membrane fluidity in worms and cells

Live worms were stained with 10 mM C-Laurdan dye (6-dodecanoyl-2-dimethylaminonaphthalene) (Thermo Fisher Scientific) as previously described⁶³. Cells were also stained with C-Laurdan at 15 µM for 2 hours then fixed in 4% PFA in PBS. Images were acquired with an DMI6000 confocal microscope and Leica software with a 63x oil-immersion objective, with a zoom factor of 4. Samples were excited with a 405 nm laser and the emission recorded between 400 and 460 nm (ordered phase) and between 470 and 530 nm (disordered phase). Quantitative assessment of the membrane order was achieved by calculating the ratiometric relationship of the fluorescence intensity recorded in two spectral channels (GP value) by using an automated ImageJ macro, according to published guidelines³⁵.

Knockdown of target genes by RNAi in cells

SPHK2 or a scrambled control siRNA was purchased from Thermo Scientific Dharmacon (ONTarget plus siRNA and Control Pool). Cells were seeded in DMEM with 10% FBS. siRNA was diluted in siRNA buffer (Final concentration 20 nM) before mixing with Lipofectmine2000 for 20 min prior to being added to cells. To minimize lipofectamine impact on the endolysosomal system, media was exchanged after 6 hours.

SDS-Page and immunoblotting

Cells were pelleted by centrifugation followed by lysis in lysis buffer (10mM Tris, 100mM NaCl, 0.2% Triton X-100, 10mM EDTA) on ice for 20 min. The lysates were transferred into fresh Eppendorf tubes and centrifuged (1000 g for 1 min at 4 °C) in a tabletop centrifuge to remove cellular debris. The protein concentration was determined using protein assay dye reagent concentrate (Bio-Rad). Proteins were separated under denaturing conditions by SDS–PAGE and transferred onto a PVDF membrane (Carl Roth) by standard wet blotting protocols. Samples were probed with rabbit polyclonal anti-SPHK2 (1:5000, 9C5E1, Santa Cruz) primary antibody. Anti-GAPDH antibody (1:5000, clone GAPDH-71.1, Sigma-Aldrich) was used as loading control.

HRP-conjugated anti-mouse and anti-rabbit IgG secondary antibodies (Bio-Rad) were used for subsequent ECL-based detection (Bio-Rad).

Fatty acid treatment in cells

For α-linolenic acid (ALA), a stock solution was made in EtOH and stored in the dark at –20°C under a N₂ atmosphere to prevent oxidation⁶⁴. Palmitic acid (PA) was dissolved in 0.1 M NaOH according to Cousin et al.⁶⁵. ALA and PA were then conjugated to BSA in 10% fatty acid free solution by shaking for 1 hour at 37°C to a concentration of 6 mM for ALA and 5 mM for PA⁶⁶. Prior to preloaded cells with BSA-PA, the solution was heated to 65 °C for 15 min. For assays with C-Laurdan, cells were preloaded with FAs for 3 hours before the addition of C-Laurdan for another 2 hours. If preloading was followed by seeding with tau, preloading with FAs was done for 3 hours before fibrillar tau was added for an additional 5 hours. C-Laurdan was spiked in for the last 2 hours. No media changes were done between steps. For experiments with HEK293T tau-Venus cells and HEK293T sfGFP-LGALS3 cells, preloading with FAs was done for 3 hours before seeding with 1N4R tau fibrils. Cells were then fixed after 48 hours.

Monomeric and fibrillar tau

Full-length human 1N4R tau was expressed and purified as previously described⁶⁷. The purified monomeric tau was dialyzed against PBS buffer containing 1 mM DTT, flash frozen in liquid nitrogen and stored at –80 °C. The concentration of monomeric tau was determined spectrophotometrically using an extinction coefficient at 280 nm of 7575 M⁻¹ cm⁻¹. Monomeric 1N4R tau was assembled into fibrils at 40 μM in the presence of 10 μM heparin, in PBS buffer containing 1 mM DTT under continuous shaking (600 rpm) for 5 days at 37 °C in an Eppendorf Thermomixer. The quality of 1N4R tau fibrils produced after 5 days was assessed by ThioflavinT binding, sedimentation, and transmission electron microscopy (TEM). The resulting 1N4R tau fibrils were spun at 25 °C and 75 000 g for 30 min. The pelleted fibrils were resuspended in PBS buffer with 1 mM DTT at 50 μM equivalent monomeric tau concentration. 1N4R tau fibrils were labeled by incubation with the 2 molar equivalents of lysine-reactive ATTO 550 (ATTO-TEC, GMBH) for 1 h at room temperature. The unreacted fluorophore was removed by two cycles of centrifugation at 75 000 g for 10 min and resuspension of the pellet in PBS. 1N4R tau fibrils were fragmented by sonication for 5 min in 2-ml Eppendorf tubes in a Vial Tweeter powered by an ultrasonic processor UIS250v (250 W, 2.4 kHz; Hielscher Ultrasonic, Teltow, Germany) to generate fibrillar particles with an average size 50 nm as assessed by TEM analysis. For seeding experiments, either monomeric or fibrillar tau was added directly into the cell culture media to a final concentration of 400 nM. Cells were then fixed after 48 hours.

Statistical analysis

Statistical analysis was performed with GraphPad Prism (GraphPad Software, Version 6h and 10.1.1). Data was tested for normal distribution by D’Agostino and Pearson omnibus normality test. Data presentation, sample size (number of biological repeats and sample size per repeat and condition), and the applied statistical tests are indicated in the figure legends for each experiment. For all cell culture data, data represents 3 independent replicates with 10 images each replicate. Significance levels: non-significant (n.s.) p > 0.05, ∗p ≤ 0.05, ∗∗p ≤ 0.01, and ∗∗∗p ≤ 0.001.

Data availability

All data supporting the findings of this study are available within the paper and its supplementary information. C. elegans strains and plasmids generated in this study are available upon request.

Acknowledgements

We gratefully acknowledge the technical support of Sarah Wübbel. We extend our gratitude to Dr. Sunil Yeruva for his assistance with the use of the DMI6000 confocal microscope. We appreciate the help of Dr. Hana Nuskova and Dr. Aurelio Teleman (DKFZ, Heidelberg) with the fatty acid conjugation protocol. The tauP301S-Venus expressing HEK293T cell line was kindly provided by Dr. William A. McEwan (Cambridge University)³⁸. The pLenti PGK Puro DEST (w529-2) vector was a gift from Dr. Eric Campeau and Dr. Paul Kaufman⁶². We also thank Dr. Bin Liu and Dr. Marja Jäättelä (University of Copenhagen) for sharing the strain BIJ34 and a pPD49.26 expression plasmid coding for sfGFP::LGALS3 and Dr. Xiaochen Wang (Chinese Academy of Sciences, Beijing) for nematodes expressing the qxIs352 transgene. Some strains were provided by the Caenorhabditis Genetics Center (CGC), which is funded by the NIH Office of Research Infrastructure Programs (P40 OD010440).

Additional information

Author contributions

Conceptualization J.T., C.A.S., C.N.K.

Formal Analysis J.T., C.A.S., N.M.

Funding acquisition R.M., C.N.K.

Investigation J.T., C.A.S., N.M., D.E.

Methodology J.T., C.A.S., N.M.

Project administration C.N.K.

Resources S.B.N.N.

Supervision R.M., C.N.K.

Visualization J.T., C.N.K.

Writing – original draft J.T., C.A.S., C.N.K.

Writing – review & editing J.T., C.A.S., N.M., D.E., R.M., C.N.K.

Additional files

Supplemental File