WikiAdultCognitiveDisease.com

The Janus Lipid

Abstract

Ceramide is a single class of membrane lipid that the clinic reads in two opposite directions. In Alzheimer's disease, ceramide accumulates in vulnerable brain regions, in cerebrospinal fluid, and in plasma; it stabilises the amyloidogenic machinery, executes the post-mitotic neuron by way of oxidative and mitochondrial injury, and helps package the seeds of tau and amyloid into the exosomes that carry pathology from cell to cell. The therapeutic instinct there is to lower it. In cancer, the same lipid is a tumour suppressor: it is the death signal that radiation, chemotherapy, and the death receptors recruit to kill dividing cells, and the malignant phenotype is defined in large part by the machinery a tumour evolves to escape it — glucosylceramide synthase, the ceramidases, and sphingosine kinase 1, which together drain ceramide and convert it into the pro-survival mediator sphingosine-1-phosphate. The therapeutic instinct there is to raise it. This dissertation asks how one molecule can carry two verdicts, traces the mechanisms on each side to the level of the enzyme and the membrane, and argues that the divergence is not a contradiction but a consequence of a single organising variable: the replaceability of the cell in which ceramide acts. It further argues that both stories are staged on the same physical object — the sphingolipid- and cholesterol-ordered membrane microdomain, the lipid raft, and the ceramide-enriched platform that forms within it. The raft is where β-secretase cleaves the amyloid precursor protein in the neuron and where the CD95 death receptor is clustered in the tumour cell; it is the common stage on which ceramide's opposite dramas are performed. Throughout, the strength of each proposed connection is graded explicitly. The cancer link is the firmest — mechanistic, interventional, and already partly in the clinic. The Alzheimer's association is robust but its causal direction is only partly resolved, complicated by reverse causation from membrane breakdown and by the mixed verdict of Mendelian-randomisation studies. The raft construct itself, on which much of the argument rests, carries a real methodological caveat. The synthesis offered here — that ceramide is a conserved membrane death signal whose pathological sign is set by whether the tissue can afford the death it commands — is presented as a heuristic that generates falsifiable predictions and a specific warning: a drug that helps one disease by moving ceramide may, systemically, injure the other.Keywords: ceramide; sphingolipids; lipid rafts; Alzheimer's disease; cancer; sphingomyelinase; the sphingolipid rheostat; ceramide-enriched platforms; BACE1; sphingosine-1-phosphate.

1. Introduction: One Molecule, Two Verdicts

There are not many molecules that two large fields of medicine want to move in opposite directions. Ceramide is one of them. Oncology has spent three decades learning how to restore ceramide to cancer cells that have learned to evade it, because ceramide is the lipid through which so many cytotoxic therapies actually kill. Neurology, over roughly the same period, has watched ceramide accumulate in the aging and Alzheimer's brain and has come to regard that accumulation as part of what is killing the neuron. The oncologist's problem is too little ceramide where death is wanted; the neurologist's problem is too much ceramide where survival is wanted. Same lipid, inverted sign.

This is not a superficial paradox to be dissolved by noting that context matters — the phrase with which biology absolves most of its contradictions. It is a specific and tractable question. Ceramide's chemistry is the same in a pyramidal neuron of the entorhinal cortex and in a squamous carcinoma of the head and neck. The enzymes that make and clear it are largely the same. The membranes in which it acts are built on the same physical principles. If the outcome inverts, the inversion must be produced somewhere definite — in which ceramide species is generated, in which compartment, on which receptor, and above all in what the cell is able to do about the death signal once it arrives. The purpose of this dissertation is to locate that "somewhere definite," and to do so without letting the elegance of the paradox outrun the evidence.

The argument proceeds in five movements. First, a primer on the sphingolipid apparatus — what ceramide is, the three roads by which cells make it, the rheostat that converts it into its pro-survival antagonist, and the under-appreciated fact that "ceramide" is not one molecule but a family whose members can act against one another. Second, the lipid raft: the membrane microdomain that is the physical stage for everything that follows, together with an honest account of why the raft concept remains partly contested. Third and fourth, the two clinical stories in mechanistic detail — accumulation as catastrophe in Alzheimer's disease, and death as the desired outcome in cancer. Fifth, the synthesis: why a single signal inverts its meaning, what the two fields can borrow from each other, and — the section the reader should weigh most sceptically — how firm each of these connections actually is. The dissertation closes with the therapeutic implications of taking the inversion seriously, including the warning that a ceramide-lowering strategy aimed at the brain and a ceramide-raising strategy aimed at a tumour are, at the level of systemic lipid biology, pointed at each other.

A note on method and voice. This is a synthesis, not a primary experimental report. Where a claim rests on a single study, that is stated. Where a mechanism is established in cell culture but unproven in the human brain, that gap is named rather than smoothed over. Each of the major connections is assigned an explicit validity grade in Section 7, and the reader is invited to disagree with those grades on the evidence rather than on the prose.

2. The Sphingolipid Apparatus

2.1 A Lipid That Sits at the Centre

Ceramide is the structural and metabolic hub of the sphingolipids. Its skeleton is a sphingoid base — most often sphingosine, an eighteen-carbon amino alcohol — bearing a fatty acid joined through an amide bond to the second carbon. That amide linkage is the "ceramide" (from cera, wax; the lipid is genuinely waxy and hydrophobic). Almost every complex sphingolipid the cell builds passes through ceramide: sphingomyelin is ceramide with a phosphocholine head group; glucosylceramide and the glycosphingolipids and gangliosides are ceramide with sugars attached; ceramide-1-phosphate is ceramide phosphorylated. To make any of these, the cell first makes ceramide; to dismantle any of them, it passes back through ceramide. This central position is why ceramide behaves less like a structural brick and more like a crossroads at which the traffic of the whole sphingolipid economy can be sensed and redirected. It is also why small, localised changes in the enzymes around ceramide can have consequences out of proportion to the amount of lipid involved.

Two structural facts do most of the biological work. First, ceramide is extraordinarily hydrophobic and has a very small head group, so it does not sit comfortably as a lone molecule among the bulky phospholipids of a bilayer; it prefers its own company, self-associating into tightly packed microdomains. Second, the length and saturation of the N-linked fatty acid varies — from roughly sixteen to twenty-six carbons — and that variation is not cosmetic. It changes how the molecule packs, where it goes, and, as Section 2.4 argues, sometimes whether it kills or protects.

2.2 The Three Roads to Ceramide

Cells generate ceramide by three distinct routes, and which route is used carries information about the cell's situation.

The de novo pathway builds ceramide from scratch on the cytosolic face of the endoplasmic reticulum. Serine palmitoyltransferase condenses the amino acid serine with palmitoyl-coenzyme A to give 3-ketodihydrosphingosine; this is reduced to dihydrosphingosine (sphinganine), N-acylated by one of six ceramide synthases to dihydroceramide, and finally desaturated by dihydroceramide desaturase to ceramide. This is the slow, constitutive, biosynthetic road, engaged when the cell is under sustained metabolic or nutrient stress and by certain chemotherapeutics.

The sphingomyelinase pathway is the fast one, and for both diseases considered here it is the more important. Here ceramide is not built but liberated — sphingomyelin already present in the membrane is hydrolysed to ceramide plus phosphocholine by a sphingomyelinase. There is a family of these enzymes distinguished by their pH optima and locations: acid sphingomyelinase, encoded by SMPD1, acting in lysosomes and, in a secreted form, at the outer leaflet; and neutral sphingomyelinase 2, encoded by SMPD3, resident at the plasma membrane and central to stress signalling. Because sphingomyelin is abundant and pre-positioned in the membrane, this route can raise local ceramide within seconds to minutes of a stress — a receptor engaging, radiation striking, reactive oxygen species rising. It is a signalling route, not a housekeeping one.

The salvage pathway recycles. Complex sphingolipids delivered to the acidic late endosome and lysosome are broken down to sphingosine, which is re-acylated by ceramide synthases to regenerate ceramide. A substantial fraction of cellular ceramide is made this way, and because it runs through the same ceramide synthases as the de novo route, it too determines the chain-length distribution of the product.

The point of cataloguing three roads is that they are not interchangeable. A ceramide pool generated in seconds by neutral sphingomyelinase at the plasma membrane, in a raft, next to a death receptor, means something entirely different from a ceramide pool accumulated over hours in the endoplasmic reticulum by the de novo pathway. Both diseases exploit this: Alzheimer's pathology leans heavily on the sphingomyelinase route at the membrane, while much of cancer therapeutics turns on de novo generation and on preventing the tumour from clearing whatever ceramide is made.

2.3 The Rheostat

Ceramide does not act alone; it acts as one pole of a balance. Ceramidases hydrolyse ceramide to sphingosine, and sphingosine kinases — sphingosine kinase 1 and 2 — phosphorylate sphingosine to sphingosine-1-phosphate. These three lipids form a gradient of meaning. Ceramide and sphingosine are broadly pro-apoptotic and growth-arresting; sphingosine-1-phosphate, acting largely through a family of cell-surface G-protein-coupled receptors, is pro-survival, pro-proliferative, pro-migratory, and pro-angiogenic. The relative amounts of ceramide and sphingosine-1-phosphate therefore behave like a dial setting cell fate — the "sphingolipid rheostat," a concept introduced in the 1990s and elaborated since (Maceyka et al., 2002). A cell rich in ceramide leans toward death and senescence; a cell that has converted its ceramide to sphingosine-1-phosphate leans toward growth and survival.

This single idea does much of the explanatory work of the whole dissertation. Cancer, as Section 5 shows, is in large part a disease of a rheostat stuck toward sphingosine-1-phosphate: tumours drain ceramide and amplify sphingosine kinase 1. Alzheimer's disease, by contrast, is a state in which the rheostat has been pushed hard toward ceramide in a cell that cannot survive being told to die. The antithetical roles of the two lipids have been documented across tissues, from cancer to the heart (Cirillo et al., 2021), and the generality of the opposition is part of what makes it trustworthy.

2.4 The Chain-Length Code: "Ceramide" Is Not One Molecule

The most common error in reasoning about ceramide — and the one that most often makes the literature look contradictory — is to treat "ceramide" as a single entity. It is a family. The six mammalian ceramide synthases each prefer fatty acids of particular lengths, and so each generates ceramides of characteristic chain length: ceramide synthase 1 makes principally C18:0 ceramide and is enriched in neurons; ceramide synthases 5 and 6 make C16:0 ceramide; ceramide synthase 2 makes the very-long-chain C22–C24 species. The brain is unusual in its heavy reliance on C18:0 ceramide, which matters for any neuron-specific argument.

These species are not merely different labels on the same activity. C16:0 ceramide, generated by ceramide synthase 6, has in several cancer settings been found to be anti-apoptotic and pro-survival, acting through a selective arm of the endoplasmic-reticulum stress response (Senkal et al., 2009; Senkal et al., 2011), whereas C18:0 ceramide, generated by ceramide synthase 1, is tumour-suppressive and is characteristically lost in head-and-neck squamous carcinoma. In other words, two ceramides differing by six carbons on one chain can pull cell fate in opposite directions within the same cell. Any lipidomic study that reports "total ceramide" therefore aggregates over a variable whose components may partly cancel. This "chain-length code" is one of the deepest reasons the field's findings sometimes conflict, and it is a standing caution against reading a single ceramide number — in a tumour or in a brain — as a single verdict.

3. Ceramide and the Lipid Raft: The Physical Stage

3.1 What a Raft Is — and the Caveat That Belongs Here

The plasma membrane is not a uniform sea. Cholesterol, sphingomyelin, and glycosphingolipids can pack together into a more ordered, more tightly packed phase — the "liquid-ordered" phase — that floats within the looser, more fluid bulk membrane. These ordered assemblies are the lipid rafts. Because sphingomyelin's largely saturated acyl chains stack neatly and cholesterol slots between them, rafts are relatively rigid, relatively thick, and selectively enrich certain proteins: glycosylphosphatidylinositol-anchored proteins, particular receptors, and signalling enzymes that carry the right lipid modifications. A raft concentrates the right molecules in the right neighbourhood, and thereby turns a diffuse membrane into a set of organised signalling platforms.

Intellectual honesty requires flagging, at the outset and not in a footnote, that the raft concept has been genuinely contested. Much early evidence came from detergent-resistant membrane fractions — the material left when a membrane is treated with cold non-ionic detergent — and it was fairly objected that a detergent-defined fraction need not correspond to a structure that exists in the living membrane. For years the size, lifetime, and even the existence of rafts in vivo were disputed. The modern position, supported by super-resolution microscopy and single-molecule methods, is that nanoscale, dynamic, cholesterol- and sphingolipid-dependent assemblies do exist but are smaller and more transient than the original picture implied. The reader should therefore treat "raft" throughout as shorthand for a real but fluid nanoscale organisation, not for a fixed platform one could purify intact. Since a good deal of the mechanistic argument in both diseases runs through rafts, this caveat is carried forward explicitly into the validity assessment of Section 7.

3.2 The Ceramide-Enriched Platform

Ceramide's relationship to the raft is where its chemistry becomes physiology. When a sphingomyelinase acts within a raft, it converts sphingomyelin — a cone-shaped lipid with a large phosphocholine head — into ceramide, which has almost no head group at all. The newly generated ceramide, intensely hydrophobic and self-attracting, does two things: it displaces cholesterol from the ordered domain, and it drives small rafts to coalesce into much larger structures. The result is a ceramide-enriched membrane platform, a macrodomain that can be orders of magnitude larger than the rafts it grew from (Gulbins & Kolesnick, 2003; Bollinger et al., 2005). This is not a slow structural remodelling; it is a rapid, signal-triggered reorganisation of the membrane's surface.

The functional consequence is amplification. Whatever receptors and signalling molecules are caught in the coalescing platform are brought into forced proximity and high local concentration. A signal that would be too weak to fire from scattered receptors becomes decisive once those receptors are clustered on a ceramide platform. The membrane, in effect, uses ceramide to convert a whisper into a shout. Both diseases turn on exactly this amplification — in opposite directions, which is the whole point of the dissertation.

3.3 The Platform as Amplifier

In the canonical case, engagement of the CD95 (Fas) death receptor triggers acid sphingomyelinase to generate ceramide in the outer leaflet; the resulting platform clusters CD95, stabilises the death-inducing signalling complex, and commits the cell to apoptosis. Ceramide-enriched platforms have since been shown to cluster a wide range of receptors — other death receptors, CD40, certain receptor tyrosine kinases, and integrins — and to be exploited by pathogens as entry portals. The platform is therefore a general-purpose amplifier of membrane signalling, and its default cargo is often, though not always, a death signal.

Hold this image, because Sections 4 and 5 are the same image read twice. In the tumour cell, the ceramide platform that clusters CD95 is a mechanism of rescue: it is how radiation and death-receptor agonists get a malignant cell to kill itself, and the cancer's job is to prevent the platform from forming. In the neuron, the raft is where the amyloid precursor protein meets β-secretase, and ceramide's remodelling of that membrane promotes the very processing that seeds the disease; here the platform is a mechanism of injury, and the therapeutic job is to keep it from forming. One physical object, two clinical readings.

4. Ceramide in Alzheimer's Disease: Accumulation as Catastrophe

4.1 The Evidence of Accumulation

That ceramide and its generating enzymes are elevated in the Alzheimer's brain is one of the better-replicated observations in the lipid biology of the disease. Ceramides and sphingomyelinase activity are increased in vulnerable regions, and ceramides have been detected directly within senile plaques (Panchal et al., 2014), placing the lipid at the anatomical heart of the pathology rather than merely in its vicinity. Reviews of the disordered sphingolipid metabolism of Alzheimer's have long argued that this is not incidental but mechanistically upstream of neuronal injury (Haughey et al., 2010; Jazvinšćak Jembrek et al., 2015).

The signal is not confined to autopsy tissue. In cerebrospinal fluid, sphingolipids including ceramides are altered across the disease course, and — importantly for the causal question of Section 7 — a careful study found that glycerophospholipids and sphingolipids accumulate in cognitively healthy people who already carry Alzheimer's biomarkers, with frank lipolysis appearing only later, in the dementia stage (Fonteh et al., 2020). The temporal ordering matters: accumulation appears to precede clinical disease rather than merely accompany its end. In plasma, higher ceramides have been associated with the risk of developing Alzheimer's disease and with the rate of cognitive and hippocampal decline, though the association is not uniform — in the Baltimore Longitudinal Study of Aging it depended on sex and on APOE genotype (Mielke et al., 2017), an early sign that ceramide's effect in the brain is modified by the same variables that dominate Alzheimer's risk generally.

4.2 The Amyloidogenic Loop: Rafts, BACE1, and the Feed-Forward with Amyloid-β

The most mechanistically satisfying part of the Alzheimer's story is a loop in which ceramide and amyloid-β drive each other, and the loop is staged in the lipid raft.

The amyloid precursor protein can be cut two ways. The non-amyloidogenic route, led by α-secretase, cleaves within the amyloid-β sequence and forecloses its production; this happens largely outside rafts. The amyloidogenic route, led by β-secretase (BACE1) and then γ-secretase, liberates intact amyloid-β, and this route is a raft phenomenon. Amyloidogenic processing of the precursor depends on lipid rafts (Ehehalt et al., 2003); forcing BACE1 into rafts by adding a lipid anchor increases β-cleavage (Cordy et al., 2003); and the precursor is preferentially internalised from cholesterol-enriched raft microdomains into the endosomal compartments where cleavage occurs (Cho et al., 2020). Whatever changes raft composition therefore changes the balance between the two cuts — and ceramide changes raft composition profoundly.

Ceramide also acts on the amyloidogenic machinery more directly. In a foundational experiment, ceramide was shown to stabilise BACE1 post-translationally, extending the enzyme's half-life and thereby increasing amyloid-β production (Puglielli et al., 2003). The same laboratory tied this to an aging-associated pathway governing the neurotrophin-receptor switch and amyloid-β generation (Costantini et al., 2006). More recently, interrupting sphingolipid supply from the other direction — inhibiting sphingomyelin synthase 1 — was found to promote lysosomal degradation of BACE1 and to ameliorate Alzheimer-like pathology in a transgenic model (Lu et al., 2018), confirming from the reverse angle that BACE1's fate is coupled to the sphingolipid economy.

Now close the loop. Amyloid-β, in turn, activates sphingomyelinases — partly through the oxidative stress it generates — driving ceramide up. That ceramide stabilises BACE1 and remodels the raft to favour amyloidogenic cleavage, producing more amyloid-β, which activates more sphingomyelinase. It is a feed-forward spiral in which a membrane lipid and a peptide amplify one another, and it offers a mechanistic reason why ceramide accumulation is not merely a marker but a plausible accelerant of the core pathology.

4.3 The Executioner and the Irreplaceable Cell

Whatever ceramide does to amyloid processing, it also does what ceramide does everywhere: it kills. In the neuron, ceramide drives oxidative stress, impairs mitochondrial function, and engages the intrinsic apoptotic programme; it can self-assemble into channels in the outer mitochondrial membrane large enough to release the apoptogenic proteins that commit a cell to death, and it feeds forward with the reactive oxygen species that both provoke sphingomyelinase and are provoked by ceramide (Jazvinšćak Jembrek et al., 2015). Reviews of the disordered sphingolipid metabolism of Alzheimer's converge on ceramide as a mediator of the neuronal apoptosis induced by oxidative stress and amyloid-β (Haughey et al., 2010).

Here lies the single most important asymmetry between this disease and cancer, and it is worth stating plainly. The neuron is post-mitotic. It cannot be replaced. When ceramide executes a pyramidal cell of the entorhinal cortex, the circuit loses a component it will never rebuild, and the loss is summed, irreversibly, over years. Ceramide's apoptotic competence — the very property oncology spends fortunes trying to restore to cancer cells — is, in the brain, a mechanism of permanent subtraction. The same molecular event, the same channel in the same mitochondrial membrane, reads as therapy in one tissue and as disease in the other, and the difference is almost entirely a difference in what the tissue can afford to lose. Section 6 builds the synthesis on this asymmetry.

4.4 Propagation: The Exosome Route

Alzheimer's disease spreads. Tau pathology in particular moves through the brain along connected circuits in a manner often described as prion-like, and part of that movement is vesicular — pathological proteins packaged into small extracellular vesicles and delivered to downstream cells. Ceramide is not incidental to this; it is part of the packaging machinery. Neutral sphingomyelinase 2, one of the enzymes that generates ceramide at the membrane, is required for the inward budding that forms exosomes within multivesicular endosomes: blocking it reduces exosome release, and ceramide itself was shown to trigger the budding of these vesicles (Trajkovic et al., 2008). The consequence for the disease is direct — inhibiting neutral sphingomyelinase 2 reduces the propagation of tau in vivo (Tallon et al., 2022). Ceramide, in other words, participates not only in making and executing the pathology within a cell but in exporting its seeds to the next one. This gives the lipid a third role, alongside amyloid facilitation and neuronal execution: vector of spread.

4.5 APOE4 and the Loading of the Axis

No account of Alzheimer's is complete without the ε4 allele of APOE, the strongest common genetic risk factor, and ceramide biology intersects it. APOE governs lipid transport in the brain, and the ε4 variant reshapes the lipid landscape: APOE4 allelic dosage alters the lipidome of the entorhinal cortex — the region where Alzheimer's characteristically begins — in aged animals (Miranda et al., 2022). That the association between plasma ceramides and Alzheimer's risk is itself modified by APOE genotype (Mielke et al., 2017) suggests that the ceramide axis and the APOE axis are not parallel but coupled: the genetic background that most raises risk does so partly by setting the sphingolipid dial. This coupling is mechanistically incomplete — it is a coupling documented more than explained — and it is graded accordingly in Section 7. But it means that ceramide is unlikely to be a free-standing pathway; it is entangled with the disease's dominant genetics.

5. Ceramide in Cancer: Death as the Desired Outcome

5.1 The Tumour-Suppressor Logic

Turn now to the tissue that can afford to die, and the entire valence of the lipid flips. In cancer, ceramide is a tumour suppressor. It arrests proliferation, drives senescence and autophagy, and — most consequentially for therapy — executes the apoptotic programme that many cytotoxic treatments depend on. A large fraction of what radiation, several classes of chemotherapy, and the death-receptor ligands actually do, at the level of the lipid, is to raise ceramide, whether by activating sphingomyelinases at the membrane or by driving de novo synthesis. The comprehensive modern account of this is Ogretmen's synthesis of sphingolipid metabolism in cancer signalling and therapy (Ogretmen, 2018), which frames ceramide accumulation as pro-death and the tumour's survival as contingent on preventing it.

The reframing this forces is worth dwelling on. In the neuron, we described ceramide's apoptotic competence as the disease. In the tumour, that same competence is the cure — or at least the mechanism of the cure. A cancer that could not be made to accumulate ceramide would be a cancer resistant to much of oncology's arsenal. The clinical problem in cancer is therefore very often not that ceramide is present but that the tumour has learned to get rid of it.

5.2 How Tumours Escape Ceramide

Malignant cells evade the ceramide death signal by three convergent strategies, each of which drains ceramide or redirects the rheostat toward survival.

  • Glycosylation. Glucosylceramide synthase attaches a glucose to ceramide, converting the pro-death lipid into glucosylceramide and downstream glycosphingolipids. This removes ceramide from the apoptotic pool, and its upregulation is repeatedly associated with resistance to chemotherapy and with expression of the P-glycoprotein drug-efflux pump — the two hallmarks of the multidrug-resistant phenotype travelling together. High glucosylceramide synthase and P-glycoprotein carry adverse prognostic weight in oral-cavity cancer (Kim et al., 2016), and experimentally, resistance to a pro-apoptotic short-chain ceramide is accompanied by rising glucosylceramide synthase, P-glycoprotein, and the multidrug-resistance gene (Gutiérrez-Iglesias et al., 2014).

  • Hydrolysis and the shift to sphingosine-1-phosphate. Ceramidases hydrolyse ceramide to sphingosine, and sphingosine kinase 1 phosphorylates it to sphingosine-1-phosphate — moving the rheostat decisively from the death pole to the survival pole. Sphingosine kinase 1 is frequently overexpressed in tumours and drives proliferation, angiogenesis, inflammation, and metastasis; it is an established oncology target (Wang et al., 2020). A tumour that amplifies this enzyme not only removes a death signal but manufactures its pro-growth antagonist from the same carbon skeleton.

  • Suppression of generation. Tumours can also down-tune the synthases and desaturases that would otherwise raise ceramide under stress, and — through the chain-length code — bias the ceramide they do make toward the species that happen to be less lethal, or even protective, in their context (Section 5.3).

The signature of a chemoresistant cancer, then, is a sphingolipid rheostat welded toward survival: low ceramide, high glucosylceramide, high sphingosine-1-phosphate. Each of the three escape routes is also, encouragingly, a drug target, which is why oncology's sphingolipid pharmacology is comparatively mature.

5.3 The Chain-Length Paradox, in Cancer

The chain-length code introduced in Section 2.4 does its most consequential work here, and it is the reason cancer sphingolipidomics can look self-contradictory. It is not universally true that raising ceramide kills a tumour, because not every ceramide is a death signal. C16:0 ceramide, the product of ceramide synthase 6, has been shown in several settings to be anti-apoptotic and pro-survival, acting through a selective, protective arm of the endoplasmic-reticulum stress response (Senkal et al., 2009; Senkal et al., 2011). C18:0 ceramide, the product of ceramide synthase 1, is by contrast tumour-suppressive and is characteristically depleted in head-and-neck squamous carcinoma, where restoring it promotes death. A therapy that crudely raised "total ceramide" might, depending on which synthase it engaged, feed the protective C16:0 pool as much as the lethal C18:0 pool. The lesson for drug design is that the target is not ceramide in bulk but a particular ceramide made by a particular synthase — and the lesson for interpreting the literature is that a bulk-ceramide measurement is a blunt instrument on both sides of the paradox.

5.4 Raft Platforms and the Reactivation of Death

The ceramide-enriched platform of Section 3 is, in oncology, a therapeutic asset. Radiation and death-receptor agonists kill in part by triggering acid sphingomyelinase to build ceramide platforms that cluster CD95 and assemble the death-inducing complex; tumours with impaired platform formation are correspondingly harder to kill. Sphingolipid-directed drugs exploit this. The sphingosine analogue FTY720 (fingolimod), beyond its sphingosine-kinase and receptor effects, exerts anticancer activity by inhibiting a cellular inhibitor of the tumour-suppressor phosphatase PP2A and thereby driving a regulated necrotic death in lung tumours (Saddoughi et al., 2012) — a reminder that sphingolipid drugs often act through several nodes of the pathway at once. The general therapeutic logic of cancer is thus the mirror image of Alzheimer's: build the ceramide platform, restore the death signal, reset the rheostat toward ceramide. Everything neurology wants to prevent, oncology wants to provoke.

6. The Divergence Explained

6.1 The Axis of Replaceability

Why does one lipid carry two verdicts? The synthesis this dissertation offers is that the sign of ceramide's pathology is set, more than by any biochemical difference, by a single property of the cell in which it acts: whether the tissue can afford the death ceramide commands.

Ceramide is a conserved membrane death-and-stress signal. That is its job, and it does the same job in a neuron and in a carcinoma: sense stress at the membrane, amplify it into a platform, and, if the stress is sufficient, execute the cell. What differs is the meaning of that execution to the surrounding tissue. The neuron is post-mitotic and irreplaceable; its death is a permanent subtraction from a circuit, and a signal that reliably kills neurons is, at the scale of the organ, a disease. The cancer cell is defined by proliferation; its death is exactly the brake the organism has otherwise lost, and a signal that reliably kills it is, at the scale of the organism, protection. The molecule has not changed its behaviour. The tissue has changed the value it places on that behaviour.

This reframes the paradox as an apparent, not a real, contradiction. Ceramide is neither "good" nor "bad." It is an executioner, and whether one wants the executioner busy or idle depends entirely on who is standing at the block. In the brain, the condemned is irreplaceable, and one wants the executioner restrained. In the tumour, the condemned is exactly who the organism has been trying to reach, and one wants the executioner armed.

6.2 Same Mechanism, Inverted Vector

Placing the two diseases side by side, mechanism for mechanism, the inversion is exact rather than approximate.

  • Direction of the pathological change. Alzheimer's: ceramide too high. Cancer: ceramide effectively too low (drained or diverted). The therapeutic vectors point opposite ways — lower it in the brain, raise it in the tumour.

  • The enzyme to target. Alzheimer's: inhibit the ceramide generators — neutral sphingomyelinase 2, acid sphingomyelinase, sphingomyelin synthase feeding BACE1. Cancer: inhibit the escape routes — glucosylceramide synthase and sphingosine kinase 1 — so that ceramide is allowed to accumulate.

  • The rheostat setting one wants. Alzheimer's: push toward sphingosine-1-phosphate and survival. Cancer: push toward ceramide and death.

  • What the raft platform does. Alzheimer's: the platform hosts amyloidogenic cleavage and neuronal death — prevent it. Cancer: the platform clusters death receptors and kills the tumour — build it.

  • The cell's replaceability. Alzheimer's: post-mitotic, irreplaceable. Cancer: proliferative, over-abundant. This is the variable that sets every sign above.

Read down that list and the two diseases are not two problems but one axis seen from its two ends. The biochemistry is shared; the replaceability of the cell flips each entry.

6.3 The Shared Stage

The unification is not merely conceptual, because the two stories are literally staged on the same physical object. The lipid raft and the ceramide-enriched platform are where β-secretase meets the amyloid precursor protein in the neuron and where CD95 is clustered in the tumour cell. The same biophysics — sphingomyelin and cholesterol ordering the membrane, a sphingomyelinase converting sphingomyelin to ceramide, ceramide displacing cholesterol and coalescing the domain, the platform amplifying whatever receptor it captures — runs both dramas. This is why lipid rafts are the connective tissue of the whole argument, and why the methodological caveat about rafts (Section 3.1) is not a pedantic aside but a load-bearing uncertainty carried into the validity assessment.

6.4 What Each Field Can Borrow

Taking the inversion seriously has a practical payoff: each field holds tools the other needs, precisely because they face the same pathway from opposite sides.

Neurology can borrow oncology's mature sphingolipid pharmacology — the glucosylceramide-synthase inhibitors, sphingosine-kinase inhibitors, and ceramide analogues developed to move the rheostat — and run them in reverse, using the pro-survival tools oncology discarded to protect the neuron. Oncology can borrow neurology's hard-won appreciation of chain-length specificity and of the extreme vulnerability of post-mitotic cells, both of which bear on the neurotoxicity of ceramide-raising cancer drugs. And both can borrow a single warning, developed next.

7. Assessment of Validity

The user's request was explicit that these connections be weighed, not merely asserted. This section grades them. The rubric is deliberately coarse and stated up front so the reader can contest any grade on its evidence.

7.1 The Grading Rubric

  • Grade A — Established. Convergent evidence across models, including interventional or causal data; a mechanism specified to the level of enzyme and effect; broadly reproduced.
  • Grade B — Probable. Strong and reproduced association with a plausible, partly demonstrated mechanism, but a material gap in causal or in-vivo human evidence.
  • Grade C — Plausible. A coherent mechanism supported mainly by cell-culture or model data, or by association vulnerable to confounding or reverse causation.
  • Grade D — Speculative/Heuristic. An organising interpretation that fits the evidence and generates predictions but has not itself been tested.

That ceramide is elevated and associated with Alzheimer's disease is Grade A: it is seen in vulnerable tissue and plaques (Panchal et al., 2014), in cerebrospinal fluid (Fonteh et al., 2020), and in plasma as a risk and progression marker (Mielke et al., 2017), and it is mechanistically embedded through the BACE1 loop (Puglielli et al., 2003; Lu et al., 2018) and the exosomal spread of tau (Trajkovic et al., 2008; Tallon et al., 2022). The causal claim — that ceramide accumulation drives the disease rather than reporting it — is weaker, and honesty requires three specific reservations.

First, reverse causation is not a hypothetical here. Neurodegeneration is, among other things, the physical breakdown of membranes, and membrane breakdown releases sphingolipids; some of the ceramide seen in a degenerating brain is a product of the degeneration, not its cause. The Fonteh finding that sphingolipids accumulate before the dementia stage and give way to lipolysis later (Fonteh et al., 2020) is encouraging for an upstream role but also shows that the temporal relationship between ceramide and neuronal death is genuinely bidirectional across the disease course. Second, the human causal evidence from genetics is mixed. Mendelian-randomisation studies — which use inherited genetic variants to probe causality and are relatively robust to reverse causation — have returned a heterogeneous verdict: a bidirectional analysis of sphingomyelin and Alzheimer's found a relationship but with the usual caveats of instrument strength and the blood-versus-brain problem (Zhu et al., 2023), and a broad metabolome-wide causal screen across neurodegenerative and psychiatric disorders did not elevate ceramide to a confident causal factor (Gilchrist et al., 2025). These studies mostly interrogate circulating sphingolipids, which need not track brain sphingolipids, so their null or weak results neither confirm nor refute a brain-tissue mechanism. Third, most of the strongest mechanistic data — the BACE1 stabilisation, the sphingomyelin-synthase rescue, the tau-propagation effect — come from cell and mouse models, and the translational gap from a transgenic mouse's sphingolipids to a human patient's is real. The mechanism is compelling and the association is firm; the causal arrow in humans is probable but not proven, and it is graded accordingly.

7.3 The Lipid Raft Construct — Grade B, with an explicit caveat

Much of the Alzheimer's mechanism (raft-dependent amyloidogenic cleavage; Ehehalt et al., 2003; Cordy et al., 2003; Cho et al., 2020) and much of the cancer mechanism (ceramide platforms clustering death receptors; Gulbins & Kolesnick, 2003; Bollinger et al., 2005) rests on the reality of membrane microdomains. As Section 3.1 conceded, the raft concept has a contested history rooted in the detergent-resistant-membrane artifact problem, and the modern consensus supports nanoscale, transient domains rather than the large stable platforms of the early literature. The ceramide-enriched platform is on firmer ground than the resting raft, because it is a large, induced, imageable structure. But any argument that depends on rafts inherits a residual biophysical uncertainty, and it would be dishonest to grade the raft-dependent steps as highly as the enzyme-level steps. This is a caveat, not a refutation: the functional consequences (cluster BACE1, get more amyloid-β; cluster CD95, get apoptosis) are reproducible regardless of exactly how one draws the domain.

7.4 Ceramide as a Tumour Suppressor in Cancer — Grade A

This is the firmest connection in the dissertation. The tumour-suppressor role of ceramide, the tumour's escape by glycosylation and by the sphingosine-1-phosphate shift, and the therapeutic value of restoring ceramide are supported by convergent mechanistic and, crucially, interventional evidence, and are consolidated in an authoritative synthesis (Ogretmen, 2018). The escape machinery is documented at the level of individual enzymes with prognostic and functional data (Kim et al., 2016; Gutiérrez-Iglesias et al., 2014; Wang et al., 2020), and sphingolipid-directed agents have measurable anticancer effects with defined mechanisms (Saddoughi et al., 2012). One does not have to resolve the causal-direction problem that dogs the Alzheimer's side, because in cancer the manipulation has been done and the outcome observed.

7.5 The Chain-Length Code — Grade B, tending to A

That different ceramide synthases generate different-chain-length ceramides with distinct, sometimes opposing, biological effects is well established (Senkal et al., 2009; Senkal et al., 2011; Ogretmen, 2018). What is less settled is the sign of any particular species in any particular context, which is model-dependent. The general principle is strong; the specific prediction for a given cell requires that cell's data. The practical corollary — that bulk-ceramide measurements are inherently ambiguous — is itself a robust and important conclusion.

7.6 The Unifying Replaceability Thesis — Grade D (heuristic)

The organising claim of Section 6 — that ceramide's pathological sign is set by cellular replaceability — is an interpretation, not a datum. It has not been tested as a hypothesis, and it could not be tested in the way an enzyme mechanism is; it is a frame that renders the biochemistry coherent and generates the predictions of Section 8. It should be held as a heuristic of Grade D: useful if it predicts, discardable if it does not, and not to be mistaken for the kind of established mechanism that the cancer story enjoys. Presenting it at any higher grade would be exactly the kind of overreach this section exists to prevent.

7.7 Standing Confounds

Three confounds cut across all of the above and deserve a collective statement. The aggregation problem: "ceramide," measured in bulk, sums over species that can oppose one another (Section 2.4), so any single-number association is intrinsically noisy. The compartment problem: plasma, cerebrospinal-fluid, and tissue ceramides need not move together, and most human data are from the most accessible and least relevant compartment, blood. The pleiotropy problem: the sphingolipid enzymes have many substrates and many effects, so a drug or a variant that changes ceramide changes much else, and clean attribution to ceramide alone is rarely possible. None of these dissolves the connections; all of them should temper confidence, and they are the reason the causal claims sit at B and C rather than A.

8. Therapeutic Implications and Testable Predictions

8.1 Two Pharmacologies Pointed at Each Other

The synthesis yields an unusually concrete therapeutic map, because the two diseases want the sphingolipid pathway moved in opposite directions and, largely, with the same drugs run in reverse.

For Alzheimer's disease, the strategy is restraint of ceramide generation and of its downstream traffic. Inhibitors of neutral sphingomyelinase 2 are the most mechanistically motivated candidates: the enzyme sits at the intersection of ceramide generation and exosomal spread, and its inhibition reduces tau propagation in vivo (Tallon et al., 2022). Reducing the sphingomyelin supply to BACE1 (Lu et al., 2018) and the functional inhibition of acid sphingomyelinase are adjacent strategies. The goal is to move the neuronal rheostat away from ceramide and toward survival — precisely the setting oncology fights.

For cancer, the strategy is the restoration of ceramide and the dismantling of the escape machinery: glucosylceramide-synthase inhibitors to stop the glycosylation drain, sphingosine-kinase-1 inhibitors to stop the diversion to sphingosine-1-phosphate, exogenous or nanoparticle-delivered ceramide analogues to reload the death pool, and platform-building cytotoxics (Ogretmen, 2018; Wang et al., 2020; Saddoughi et al., 2012). The goal is the neuron's nightmare: a rheostat welded toward ceramide and death.

8.2 The Inversion Warning

Here the synthesis pays a debt it owes to safety. If a systemic drug lowers ceramide to protect the brain, the same drug, reaching every proliferating tissue, removes a tumour-suppressor signal — a theoretical pro-oncogenic liability that any chronic ceramide-lowering Alzheimer's therapy must be watched for. Conversely, if a systemic drug raises ceramide to kill a tumour, the same drug, crossing into the nervous system, delivers a pro-apoptotic signal to post-mitotic neurons — a theoretical neurotoxic liability, and one consistent with the peripheral neuropathies seen with several sphingolipid-active chemotherapeutics. The two pharmacologies are, at the level of systemic lipid biology, pointed at each other. This does not forbid either; it prescribes that each be delivered with tissue selectivity — brain-restricted or neuron-restricted for the Alzheimer's agents, tumour-targeted or tumour-activated for the oncology agents — and that each be monitored for the other's disease as an off-target risk. That warning is, in the author's view, the most immediately useful product of reading the two literatures as one.

8.3 Falsifiable Predictions

A synthesis earns its keep by predicting. The replaceability frame and the shared-raft mechanism generate the following, each stated so it could be shown wrong.

  • Chain-length prediction. Interventions that specifically raise C18:0 ceramide (via ceramide synthase 1) should be more neurotoxic and more tumour-suppressive than interventions raising C16:0 ceramide (via ceramide synthase 6); a therapy that fails to distinguish them will show inconsistent effects on both sides.
  • Compartment prediction. Brain and cerebrospinal-fluid ceramide should track Alzheimer's progression more tightly than plasma ceramide does; where the request is a causal test, the accessible compartment will underperform.
  • Inversion prediction. Chronic ceramide-lowering agents effective in the brain should, in long exposure and adequate cohorts, show a measurable shift in cancer incidence or progression; chronic ceramide-raising oncology agents should show a measurable signal of neuronal or peripheral-nerve injury.
  • Raft prediction. Interventions that stabilise raft cholesterol against ceramide-driven displacement should reduce amyloidogenic processing in the neuron and, in the tumour, impair death-receptor-triggered apoptosis — the same manipulation helping one disease and hindering the treatment of the other.
  • Causal-genetic prediction. Mendelian-randomisation instruments built specifically on brain-expressed ceramide-pathway variants (rather than on circulating sphingolipid levels) should recover a stronger causal signal for Alzheimer's than the blood-based instruments have so far (cf. Zhu et al., 2023; Gilchrist et al., 2025).

Each prediction is a place the frame could break. That is the intended standard.

9. Conclusion: The Lipid Between Two Deaths

Ceramide is a small, waxy, ancient molecule that the cell keeps at the crossroads of its membrane chemistry and uses, when stressed, to decide whether to die. That decision is the same computation in every cell. What differs — between the entorhinal neuron losing itself to Alzheimer's disease and the carcinoma cell that oncology is trying to reach — is not the computation but its consequence. In the neuron, ceramide's competence to execute is a tragedy, because the executed cell is irreplaceable and the loss accrues without repair. In the tumour, that same competence is the mechanism of rescue, and the disease is the cell's talent for switching the executioner off. Both dramas are staged on the same membrane platform, run by the same enzymes, tuned by the same rheostat; the tissue's replaceability supplies the sign.

Read this way, the two diseases stop being separate specialties that happen to share a lipid and become the two ends of one axis. The reading is not free of uncertainty, and this dissertation has tried to be exact about where the uncertainty lives: the cancer story is established, the Alzheimer's association firm but its causal arrow only probable, the raft substrate real but blurred at its edges, and the unifying frame a heuristic to be judged by its predictions rather than believed on its elegance. But the reading pays. It hands each field the other's pharmacology, run in reverse; it explains why a single lipidomic number can mean opposite things; and it issues a warning that neither field, reading alone, would think to make — that a drug good for one of these diseases may, by the plainest logic of the pathway, be bad for the other. Ceramide is the lipid between two deaths. Which death we want, and which we must prevent, depends on nothing more mysterious, and nothing less consequential, than whether the cell it is killing can ever be replaced.

References

1. Bollinger, C. R., Teichgräber, V., & Gulbins, E. (2005). Ceramide-enriched membrane domains. Biochimica et Biophysica Acta (Molecular Cell Research), 1746(3), 284–294. doi:10.1016/j.bbamcr.2005.09.001

2. Cho, Y. Y., Kwon, O. H., & Chung, S. (2020). Preferred endocytosis of amyloid precursor protein from cholesterol-enriched lipid raft microdomains. Molecules, 25(23), 5490. doi:10.3390/molecules25235490

3. Cirillo, F., Piccoli, M., Ghiroldi, A., Monasky, M. M., Rota, P., La Rocca, P., et al. (2021). The antithetic role of ceramide and sphingosine-1-phosphate in cardiac dysfunction. Journal of Cellular Physiology, 236(7), 4857–4873. doi:10.1002/jcp.30235

4. Cordy, J. M., Hussain, I., Dingwall, C., Hooper, N. M., & Turner, A. J. (2003). Exclusively targeting β-secretase to lipid rafts by GPI-anchor addition up-regulates β-site processing of the amyloid precursor protein. Proceedings of the National Academy of Sciences USA, 100(20), 11735–11740. doi:10.1073/pnas.1635130100

5. Costantini, C., Scrable, H., & Puglielli, L. (2006). An aging pathway controls the TrkA to p75NTR receptor switch and amyloid β-peptide generation. The EMBO Journal, 25(9), 1997–2006. doi:10.1038/sj.emboj.7601062

6. Ehehalt, R., Keller, P., Haass, C., Thiele, C., & Simons, K. (2003). Amyloidogenic processing of the Alzheimer β-amyloid precursor protein depends on lipid rafts. The Journal of Cell Biology, 160(1), 113–123. doi:10.1083/jcb.200207113

7. Fonteh, A. N., Chiang, A. J., Arakaki, X., Edminster, S. P., & Harrington, M. G. (2020). Accumulation of cerebrospinal fluid glycerophospholipids and sphingolipids in cognitively healthy participants with Alzheimer's biomarkers precedes lipolysis in the dementia stage. Frontiers in Neuroscience, 14, 611393. doi:10.3389/fnins.2020.611393

8. Gilchrist, L., Mutz, J., Hysi, P., Legido-Quigley, C., Kõks, S., Lewis, C. M., et al. (2025). Evaluating metabolome-wide causal effects on risk for psychiatric and neurodegenerative disorders. BMC Medicine, 23, 318. doi:10.1186/s12916-025-04129-4

9. Gulbins, E., & Kolesnick, R. (2003). Raft ceramide in molecular medicine. Oncogene, 22(45), 7070–7077. doi:10.1038/sj.onc.1207146

10. Gutiérrez-Iglesias, G., Hurtado, Y., Palma-Lara, I., & López-Marure, R. (2014). Resistance to the antiproliferative effect induced by a short-chain ceramide is associated with an increase of glucosylceramide synthase, P-glycoprotein, and multidrug-resistance gene-1 in cervical cancer cells. Cancer Chemotherapy and Pharmacology, 74(4), 809–817. doi:10.1007/s00280-014-2552-3

11. Haughey, N. J., Bandaru, V. V. R., Bae, M., & Mattson, M. P. (2010). Roles for dysfunctional sphingolipid metabolism in Alzheimer's disease neuropathogenesis. Biochimica et Biophysica Acta (Molecular and Cell Biology of Lipids), 1801(8), 878–886. doi:10.1016/j.bbalip.2010.05.003

12. Jazvinšćak Jembrek, M., Hof, P. R., & Šimić, G. (2015). Ceramides in Alzheimer's disease: key mediators of neuronal apoptosis induced by oxidative stress and Aβ accumulation. Oxidative Medicine and Cellular Longevity, 2015, 346783. doi:10.1155/2015/346783

13. Kim, J. W., Park, Y., Roh, J. L., Cho, K. J., Choi, S. H., Nam, S. Y., et al. (2016). Prognostic value of glucosylceramide synthase and P-glycoprotein expression in oral cavity cancer. International Journal of Clinical Oncology, 21(5), 883–889. doi:10.1007/s10147-016-0973-1

14. Lu, M. H., Ji, W. L., Xu, D. E., Yao, P. P., Zhao, X. Y., Wang, Z. T., et al. (2018). Inhibition of sphingomyelin synthase 1 ameliorates Alzheimer-like pathology in APP/PS1 transgenic mice through promoting lysosomal degradation of BACE1. Experimental Neurology, 311, 67–79. doi:10.1016/j.expneurol.2018.09.012

15. Maceyka, M., Payne, S. G., Milstien, S., & Spiegel, S. (2002). Sphingosine kinase, sphingosine-1-phosphate, and apoptosis. Biochimica et Biophysica Acta (Molecular and Cell Biology of Lipids), 1585(2–3), 193–201. doi:10.1016/s1388-1981(02)00341-4

16. Mielke, M. M., Haughey, N. J., Han, D., An, Y., Bandaru, V. V. R., Lyketsos, C. G., et al. (2017). The association between plasma ceramides and sphingomyelins and risk of Alzheimer's disease differs by sex and APOE in the Baltimore Longitudinal Study of Aging. Journal of Alzheimer's Disease, 60(3), 819–828. doi:10.3233/JAD-160925

17. Miranda, A. M., Ashok, A., Chan, R. B., Zhou, B., Xu, Y., McIntire, L. B., et al. (2022). Effects of APOE4 allelic dosage on lipidomic signatures in the entorhinal cortex of aged mice. Translational Psychiatry, 12, 129. doi:10.1038/s41398-022-01881-6

18. Ogretmen, B. (2018). Sphingolipid metabolism in cancer signalling and therapy. Nature Reviews Cancer, 18(1), 33–50. doi:10.1038/nrc.2017.96

19. Panchal, M., Gaudin, M., Lazar, A. N., Salvati, E., Rivals, I., Ayciriex, S., et al. (2014). Ceramides and sphingomyelinases in senile plaques. Neurobiology of Disease, 65, 193–201. doi:10.1016/j.nbd.2014.01.010

20. Puglielli, L., Ellis, B. C., Saunders, A. J., & Kovacs, D. M. (2003). Ceramide stabilizes β-site amyloid precursor protein-cleaving enzyme 1 and promotes amyloid β-peptide biogenesis. The Journal of Biological Chemistry, 278(22), 19777–19783. doi:10.1074/jbc.M300466200

21. Saddoughi, S. A., Gencer, S., Peterson, Y. K., Ward, K. E., Mukhopadhyay, A., Oaks, J., et al. (2013). Sphingosine analogue drug FTY720 targets I2PP2A/SET and mediates lung tumour suppression via activation of PP2A-RIPK1-dependent necroptosis. EMBO Molecular Medicine, 5(1), 105–121. doi:10.1002/emmm.201201283

22. Senkal, C. E., Ponnusamy, S., Bielawski, J., Hannun, Y. A., & Ogretmen, B. (2010). Antiapoptotic roles of ceramide-synthase-6-generated C16-ceramide via selective regulation of the ATF6/CHOP arm of ER-stress-response pathways. The FASEB Journal, 24(1), 296–308. doi:10.1096/fj.09-135087

23. Senkal, C. E., Ponnusamy, S., Manevich, Y., Meyers-Needham, M., Saddoughi, S. A., Mukhopadyay, A., et al. (2011). Alteration of ceramide synthase 6/C16-ceramide induces activating transcription factor 6-mediated endoplasmic reticulum stress and apoptosis via perturbation of cellular Ca²⁺ and ER/Golgi membrane network. The Journal of Biological Chemistry, 286(49), 42446–42458. doi:10.1074/jbc.M111.287383

24. Tallon, C., Bell, B. J., Sharma, A., Pal, A., Malvankar, M. M., Thomas, A. G., et al. (2022). Dendrimer-conjugated nSMase2 inhibitor reduces tau propagation in mice. Pharmaceutics, 14(10), 2066. doi:10.3390/pharmaceutics14102066

25. Trajkovic, K., Hsu, C., Chiantia, S., Rajendran, L., Wenzel, D., Wieland, F., et al. (2008). Ceramide triggers budding of exosome vesicles into multivesicular endosomes. Science, 319(5867), 1244–1247. doi:10.1126/science.1153124

26. Wang, X., Sun, Y., Peng, X., Naqvi, S. M. A. S., Yang, Y., Zhang, J., et al. (2020). The tumorigenic effect of sphingosine kinase 1 and its potential therapeutic target. Cancer Control, 27(1), 1073274820976664. doi:10.1177/1073274820976664

27. Zhu, H., Lu, R., Zhou, Q., Du, Z., & Jiang, Y. (2023). Relationship between sphingomyelin and risk of Alzheimer's disease: a bidirectional Mendelian randomization study. Journal of Alzheimer's Disease Reports, 7(1), 1327–1336. doi:10.3233/ADR-230126

Source: research/ceramide-janus/PhD_Thesis_Janus_Lipid.md