The average human brain contains nearly a kilogram of fat. That’s right. The seat of your consciousness—that very thing that might be registering surprise right now—comprises about 60 percent lipid. What does it all do and how might it figure into Alzheimer’s and other neurodegenerative disorders? Scientists who gathered for the 2nd Symposium on Lipids in Brain Disease, shared some answers. From fats that drive proteinopathies and inflammation to others that might bring relief, learn about some of the newest concepts in Tom Fagan’s meeting series.
Cutting (or Slippery?) Edge: Lipids in Neurodegeneration Science
Readers, ignore lipids at your peril. These oily molecules make up 60 percent of the human brain. Long considered mere wrapping for more interesting cellular contents, these fats hold secrets worth exploring. At the 2nd Symposium on Lipids in Brain Diseases, held September 13-15 in Leiden, The Netherlands, about 100 experts on lipids mingled with nonexperts to share data and ideas in discussions of how lipids both grease and prevent neurological disease.
“It’s the new frontier,” said Adrian Isaacs, University College London. Saranna Fanning, Brigham and Women’s Hospital, Boston, said lipids have potential as biomarkers, drug targets, and therapeutic strategies. “We are at the point where the study of brain lipids is worth multiple meetings a year,” she told Alzforum.
Herman Boerhaave would have approved. Renowned for his hands-on teaching style, and sometimes called the "Dutch Hippocrates", this physician, chemist, botanist, and humanist attracted scholars from all over the world to his town in southern Holland starting in the late 17th century. Rijksmuseum Boerhaave is built around his theatrum anatomicum (image below), which stands today as it did almost 400 years ago. It was the venue for this lipid conference.
Theatrum Anatomicum. The theater stands today, just as it was when Herman Boerhaave and other scholars probed the workings of the human body in its center circle. [Courtesy of Tom Fagan.]
The organizers were Birol Cabukusta and Martin Giera from Leiden University Medical Center; Jerome Hendriks, Hasselt University, Hasselt, Belgium; Gijs Kooij, Amsterdam University Medical Center; and Rik van der Kant, Vrije University, Amsterdam.
Featuring nine keynote and nine shorter talks, the meeting was well-suited for this hot topic. Attendees learned about how different lipids operate in proteinopathies, how cells regulate myelin in health and disease, and how diet affects lipid metabolism. They also considered how the fine balance between lipids causing and resolving inflammation might open new windows for therapy.
In this four-part series, Alzforum summarizes highlights from the meeting, which will reconvene in 2025. The stories focus on imbalances in long-chain fatty acid metabolism being both a cause and a consequence of proteinopathy, on the role of microglia in maintaining the cholesterol-laden myelin sheaths, on sterols driving amyloid and tau pathology, and on how mysterious specialized lipids can help resolve inflammation.
Lipids and Proteinopathies
Scientists have known for decades that neurodegeneration and problems with lipids can go hand in hand. In lipid storage disorders, such as Nieman Pick’s disease, cholesterol and sphingolipids clog lysosomes, preventing cells from degrading and recycling unwanted protein and leading to the death of neurons For a good synopsis, see Pfrieger 2023.
By now, lipids have come under intense scrutiny in Alzheimer’s, Parkinson’s, and other neurodegenerative disorders. Genome-wide association studies have linked lipid metabolism genes to AD. Lipid droplets accumulate in microglia in the AD brain and mouse models (Aug 2019 news; Claes et al., 2021; Sep 2023 news).
In Parkinson’s, Fanning, when working in Dennis Selkoe’s lab at Brigham and Women’s, had previously linked changes in the lipid bilayer of cellular membranes to aggregation of α-synuclein, the principal component of Lewy bodies found in PD and other synucleinopathies, including dementia with Lewy bodies.
Specifically, Fanning reported an exacerbation of α-synuclein pathology, thanks to the enzyme stearoyl CoA desaturase. SCD introduces a double bond into stearic acid, an 18-carbon saturated fatty acid, to create oleic acid, a monounsaturated 18C fatty acid. This cis double bond, between carbon 9 and 10, forces a 30° kink in the fatty acid chain. This loosens the packing of the fatty acid and its derivatives, such as phosphatidylcholine, in lipid membranes. As a result, monomers of α-synuclein can linger in the membrane a little longer than usual, giving them more opportunity to aggregate, Fanning found. Blocking SCD attenuated α-synuclein aggregation and toxicity in yeast, roundworms, and human neurons (Dec 2018 news).
Tri This. Three fatty acid chains attach to the 3-carbon glycerol backbone to form triglycerides. In this case the fatty acids are palmitic (top), oleic (middle), and α-linolenic. In mammals, fatty acids are stored and transported as triglycerides.
If oleic acid spells trouble, then would it be wise to rid cells of excess, including any oleic acid stored in triglycerides? Not exactly. In Leiden, Fanning reported that blocking metabolism of these fatty acid troikas (image at right) can also alleviate synuclein toxicity. When she blocked lipase E (LIPE) in neuroblastoma cells expressing mutant α-synuclein, fewer aggregates formed and less of the protein was phosphorylated on serine 129, a marker of α-synuclein toxicity.
LIPE hydrolyses neutral lipids, specifically the di- and triglycerides that accumulate in lipid droplets. In so doing, this enzyme releases free fatty acids for further metabolism. Knocking down its expression with a short hairpin RNA also protected the cells from α-synuclein toxicity. Likewise, knocking down the lipase protected dopaminergic neurons from degenerating in a C. elegans model of α-synuclein toxicity.
As You LIPE It. Cells expressing mutant -synuclein (green) accumulate aggregates of the protein (left), but not if a lipase is blocked (right). [Courtesy of Fanning et al., 2022.]
How does silencing LIPE help? The free fatty acids it liberates from lipid droplets and triglycerides include oleic acid—that same one made by the enzyme SCD (image at right). Fanning thinks cells cannot metabolize this fatty acid fast enough once triglycerides have been hydrolyzed, leaving it free to slither into cell membranes where it promotes aggregation of α-synuclein. In short, it might be better to leave oleic acid sequestered away in the lipid droplets, where it does less harm.
What might be the better approach—stopping the synthesis of oleic acid, or its release from triglycerides? Perhaps the answer is both. “Ideally, you would prevent the synthesis,” Fanning said, “but by the time a patient enters the physician's office, the synthesis pathway may have been cranking out oleic acid for over a decade, and lipases may be releasing any stored in lipid droplets. So if we reduce LIPE activity at that stage, it may be helpful.”
Indeed, in Leiden Fanning reported that, at least in neuroblastoma cells, inhibiting both the making and the breaking enzymes, i.e. SCD and LIPE, beat blocking either alone at reducing levels of oleic acid, p-Syn129, and α-synuclein aggregates.
Lipid Balance. Healthy cells (top left) have few di- and triglycerides, lipid droplets, and α-synuclein aggregates. Not so when mutant α-synuclein is around (bottom left). Blocking SCD to limit monosaturated fatty acids (top right), blocking LIPE to prevent degradation of di- and triglycerides (center right), or both (bottom right) might hold off synuclein toxicity. [Courtesy of Fanning et al., 2022.]
Blocking LIPE could have downsides. For one, lipid droplets might accumulate. “When manipulating any lipid in cells, you want to be sure to return membranes to the fatty acid composition associated with health,” said Fanning. One way to do this might be to counter monounsaturated fatty acids, such as oleic acid, with saturated ones.
Ulf Dettmer at Brigham and Women’s Hospital, who collaborates with Fanning, has done just that. He loaded neuroblastoma cells with 14-carbon myristic acid, 16-carbon palmitic acid, and 18-carbon stearic acid. All three tempered phosphorylation of α-synuclein at serine 129, while myristic acid increased the multimeric-to-monomeric ratio of α-synuclein and decreased aggregates (Imberdis et al., 2019). α-Synuclein tetramers are believed to be more stable and less likely to misfold into fibrils (Aug 2011 news). Together, Dettmer and Fanning are currently exploring how myristic and other fatty acids affect α-synuclein toxicity.
Others in Leiden called the work impressive. “This is a good example of how scientists can delve into lipidomics and come up with potential drug targets,” van der Kant said. “When we go to clinical trial meetings we never hear people talking about lipids, but this is an example of how they may help tackle pathology,” he said.
Femke Feringa, also from Vrije University, noted that lipid research in the past has been held back by the difficulties of understanding how the fluidity of membranes affects biological processes. “Finding the best systems to study this relationship is a big challenge” she said. “Now that we have iPSC models, we can hopefully find better experimental approaches.” For more on that, see below.
What about other proteinopathies? In Leiden, UCL’s Isaacs described quite a different instance of lipid dysregulation, though still of the fatty acid kind. Isaacs studies models of amyotrophic lateral sclerosis/frontotemporal dementia.
Scientists in his lab developed a fly model of ALS/FTD based on expression of the dipeptide repeats that are translated from the CCGGG hexanucleotide expansion in the C9ORF72 gene (Mizielinska et al., 2014).
Since then, Ashling Giblin in the lab has found that, of the top 10 diminished processes in these flies, three had to do with lipid synthesis. Downregulated genes included acetyl-coenzyme A synthetase, which makes the metabolite acetyl CoA, and acetyl CoA carboxylase, which turns acetyl CoA into malonyl CoA. Malonyl CoA is the rate-limiting substrate in the formation of longer-chain fatty acids.
Expression of FASN1 and FASN2 genes, which encode enzymes that make those long-chain fatty acids from malonyl CoA, were also downregulated. So was SCD1, the desaturase that turns stearic acid into oleic acid. C9ORF72 flies made half as much FASN1 and SCD1 RNA as did controls. This jibes with prior transcriptomic analysis of postmortem ALS spinal cord. Among 154 people with the disease, FASN1 expression was almost halved, and SCD1 expression was down about 10 percent, compared to control tissue (Humphrey et al., 2023).
What about human cells? Alex Cammack in Isaacs' lab found that there is a dearth of highly unsaturated phospholipids in neurons generated from iPSC cells carrying a C9ORF72 expansion. Expressing a 92-repeat hexanucleotide—the length that gives a person disease—in normal neurons tanked the same phospholipids, whereas adding antisense nucleotides to silence the expansions restored lipid balance.
Conceptually, could restoring healthy lipid profiles protect against the C9ORF72 repeat? In the fly model, overexpressing FASN1 or SCD1 extended lifespan, while in the human C9ORF72 neurons, overexpressing enzymes that generate polyunsaturated fatty acids increased cell viability “We see a very specific change in phospholipids that contain highly saturated fatty acids; if we can reverse that therapeutically, it may have great potential,” Isaacs told the audience.
The literature hints that he might be right. In a prospective study of a million people in the U.S., those who ate more ω3 polyunsaturated fatty acids, such as the 18-carbon, were less likely to develop ALS in the next two decades (Fitzgerald et al., 2014). The ω3 PUFAs are those whose first double bond occurs at the third carbon from the methyl end of the chain. Two case-control studies had also linked high PUFA intake to reduced risk for the disease (Veldink et al., 2007; Okamoto et al., 2007). More direct evidence has come from analysis of plasma. In a Phase 3 clinical trial of dexpramipexole in ALS, people who had more of the ω3 eicosapentaenoic acid and the ω6 linoleic acid at the beginning of the trial were likelier to still live 12 months later (Bjornevik et al., 2023).
Manipulating brain lipids through the diet may be a tall order, since many of these lipids bounce off the blood-brain barrier. Indeed, Isaacs found that while feeding PUFAs to the C9ORF72 flies made them live longer, this was less effective than expressing the enzymes that make unsaturated fatty acids.
Attendees in Leiden found Isaac’s data fascinating. Tiago Gil Oliveira, Universidade do Minho, Portugal, asked if the lipid imbalance affects the fluidity of cell membranes. Isaacs said this question was top of his mind these days. Others wondered how the dipeptide repeats made from the expanded C9ORF72 RNA change lipid dynamics. Isaacs said this needs to be figured out.—Tom Fagan
Does the Brain Use Microglia to Maintain Its Myelin?
You might well think that myelin, the fatty insulation that speeds action potentials between neurons, falls solely under the purview of oligodendrocytes. After all, these specialized cells wrap cholesterol-laden myelin around axons in the brain's white matter beginning early in development, and throughout life they sustain and repair this insulation. Well, it turns out they need microglia to help them do that. At the 2nd Symposium on Lipids in Brain Diseases, held September 13-15 in Leiden, The Netherlands, scientists reported that microglia are essential for regular maintenance of myelin.
They showed that in demyelinating conditions, microglia can scupper remyelination. In diseases such as multiple sclerosis, Charcot-Marie-Tooth disease, and adult-onset leukoencephalopathy with axonal spheroids and pigmented glia, microglia are summoned to mop up debris. In so doing, they become both phagocytic and inflammatory. If they get stuck in that state, persistent inflammation hinders oligodendrocytes' repairing the insulation. The findings could have implications not only for MS, but for other neurodegenerative diseases that degrade myelin, including Alzheimer’s.
Herman Boerhaave, 1668-1738, is sometimes called the Father of Physiology. An esophageal rupture syndrome, caused by forceful vomiting, is named after him. The symposium was held at the Rijksmuseum Boerhaave. [Courtesy of Tom Fagan.]
Niamh McNamara, Netherlands Institute for Neuroscience, Amsterdam, explained how microglia are crucial for the normal growth and maintenance of myelin. Without them, the myelin begins to unravel and develops bulges, also called outfolds, McNamara found. In the absence of microglia, oligodendrocyte lipid and cholesterol metabolism falters, and the cells soon struggle to sustain the insulation. McNamara carried out most of this research while in Veronique Miron’s lab at the University of Edinburgh (McNamara et al., 2023).
The effect is subtle and required sleuthing to figure out. McNamara and colleagues knew that when they ablated microglia, or macrophages, myelination crumbled in mice. But which of these immune cells were to blame? To find out, she turned to mice lacking the fms-intronic regulatory element (FIRE) in the gene for macrophage colony stimulating factor 1 receptor. CSF1R is essential for microglial/macrophage proliferation and survival. FIRE is a super enhancer for transcription; without it, macrophages and monocytes cannot make CSF1R, and die. Curiously, only certain cells depend on FIRE to produce this receptor. These include microglia, but not border-associated and perivascular macrophages in the brain (Rojo et al., 2019). Would FIREΔ/Δ mice have normal myelin, then?
They did—at first. In 1-month-old mice, it was business as usual for oligodendrocytes. They insulated axons of all diameters in the white matter; myelin basic protein levels seemed normal.
Even so, on closer inspection, McNamara found that all was not well with the myelin around FIREΔ/Δ axons. Under the electron microscope, it appeared to be coming apart in places and beginning to blister. Myelin's “inner tongue”—the area where oligodendrocytes start laying it down—was abnormally thick.
Deformities persisted as the mice aged, suggesting it was not a developmental phenomenon. By 3 to 4 months of age, axons became hypermyelinated; by 6 months the myelin had begun to break down, endangering axon insulation. This waning of myelin also happened in wild-type mice fed PLX-5622, the CSF1R inhibitor that stymies both microglial and macrophage proliferation.
FIRE Damage. Myelin around axons in FIREΔ/Δ mice blebs (left), comes undone (center), and its inner tongue (brown, right) thickens. [Courtesy McNamara et al., 2022.]
These structural changes had consequences. In the Barnes circular maze, FIREΔ/Δ mice made more errors than wild-type when McNamara moved their escape hole 180 degrees. This switcheroo forces mice to adapt, hence the FIREΔ/Δ mice's shortcomings imply loss of cognitive flexibility. There may be parallels in a rare disease that goes by the mouthful adult-onset leukoencephalopathy with axonal spheroids and pigmented glia. In ALSP, heterozygous mutations in CSF1R make microglia scarce, and neurons in white matter degenerate (Oosterhof et al., 2018). People with ALSP have trouble with memory and executive function.
Why do FIREΔ/Δ mice fail at myelin upkeep? They do have normal numbers of oligodendrocytes. To see if these cells might be compromised in other ways, McNamara teamed up with the labs of Josef Priller at Charité-Universitätsmedizin Berlin and U Edinburgh and Anna Williams at U Edinburgh. Profiling transcriptomes via single-nucleus RNA-Seq, the scientists found a preponderance of oligodendrocytes expressing Serpina3n.
This serine protease inhibitor marks an inflammatory type of oligodendrocyte found in mouse amyloidosis models (Jan 2020 news). Serpina3n also ticks up in plasma in early stages of AD, and in the cerebrospinal fluid of people with multiple sclerosis (Aug 2023 news; Fissolo et al., 2021).
Oligodendrocytes on Fire. Single-nucleus RNA-Seq identifies an inflammatory subset of oligodendrocytes (pink, left) that emerge in FIREΔ/Δ mice. These predominantly express Serpina3n (right). [Courtesy of McNamara et al., 2022.]
To figure out what these Serpina3n cells do, the scientists surveyed their transcriptomes for hints of altered biological pathways. This showed that these oligodendrocytes poorly regulate their own cholesterol and lipid metabolism. Lipidomic analysis of white matter backed this up. In collaboration with Jerome Hendricks at Hasselt University, Hasselt, Belgium, McNamara found that white-matter triglyceride levels were lower, and cholesteryl ester levels higher in FIREΔ/Δ mice than in wild-type.
What might connect the lipid metabolism of oligodendrocytes to a dearth of microglia? McNamara suspects TGF-β. Gene pathway analysis of the Serpina3n oligodendrocytes placed this growth factor upstream of the dysregulated lipid metabolism genes. In FIRED/D mice, twice as many oligodendrocytes lacked the TGF-β receptor. Conditionally knocking out this receptor in mouse oligodendrocytes had the same effect as ablating microglia, i.e., myelin started unraveling and the myelin tongue swelled. The plot thickened further when McNamara treated FIREΔ/Δ mice with SRI-011381, a TGF-β agonist that bypasses the receptor to trigger SMAD2–SMAD3 transcription factors downstream. Injected into the peritoneum three times a week beginning when mice were 2 months old, this agonist rescued myelination deficits in FIREΔ/Δ mice a month later.
All told, the findings suggest that microglia, by releasing TGF-β, help oligodendrocytes keep their lipid metabolism on an even keel.
In Leiden, Edorardo Marcora, Icahn School of Medicine at Mount Sinai, New York, asked if TREM2 loss-of-function variants linked to Alzheimer’s affect myelination. McNamara considers this question important but doesn’t think anyone has investigated.
Anil Cashikar, Washington University, St. Louis, wondered what happens to the excess cholesteryl esters made by the Serpina3n oligodendrocytes. “Do they form cholesterol crystals?” he asked. Such crystals have been linked to faulty myelin repair in old mice (Cantuti-Castelvetri et al., 2018). McNamara saw no such crystals in the FIRE mice, but noted that low ApoE expression by oligodendrocytes might mean the cells do not export cholesterol. “We have not yet nailed down the link between the pathology and lipid metabolism,” she said.
Pointers toward that link may come from work on myelin repair done in Gesine Saher’s lab at the Max-Planck-Institute of Experimental Medicine, Gottingen, Germany. In Leiden, Saher talked about how cholesterol helps microglia corral amyloid and how sterols made by microglia can help repair myelin lesions in mouse models of multiple sclerosis.
Myelin holds 70 percent of the cholesterol in the brain. When axons lose this insulation, oligodendrocytes trying to repair it churn out the enzymes needed to make cholesterol and other sterols. They include squalene synthase (SQS) and HMG-CoA reductase, the target of statins. Remyelination typically follows in two steps—an acute phase and a chronic phase. To their surprise, Stefan Berghoff and colleagues in Saher’s lab found that oligodendrocytes boosted sterol synthesis only in the chronic phase of remyelination. In the acute phase, it was microglia that churned out these enzymes. Furthermore, conditionally knocking out SQS in microglia severely limited acute remyelination (Berghoff et al., 2021).
Covered in Flames. Spinal cords (green) in a mouse mode of MS contend with many more inflammatory microglia (red) when microglial squalene synthase is knocked out (right). [Courtesy of Berghoff et al., 2021.]
How do microglia support remyelination? Once again, their transcriptomes might point to an answer. In MS, and related animal models, degenerating myelin nudges microglia and macrophages into a phagocytic, inflammatory state so they can clear myelin debris. This state must resolve before remyelination can occur, but microglia unable to make sterols never reverted to their “normal” selves, reports Saher (image below). Instead, they kept making nitric oxide, interleukin-1β, Cxcl10, and other inflammatory mediators.
These mediators had something in common. Most are activated by liver X receptor signaling. Originally found in the liver, LXR receptors are expressed throughout the body. They regulate cholesterol, lipid, and glucose metabolism. Could microglial LXR signaling matter to remyelination?
Pursuing this idea, Berghoff found that phagocytic cells in three different models of MS upregulated LXR-dependent genes, including ApoE and ABCA1. But what set off this transcriptional change? Given that sterols activate LXR, and that remyelination requires microglial sterol synthesis, he looked for a sterol that might do the trick. He found that 24-dehydrocholesterol reductase (Dhcr24) was downregulated in the three models of MS—an important clue. If this enzyme were to stop working, then the LXR agonist desmosterol would accumulate provided the rest of the sterol synthesis pathway remained operational (image below).
Desperate for Desmosterol? In MS, remyelination requires not only upregulation of cholesterol synthesis enzymes in microglia and macrophages, i.e., squalene synthase (coded by Fdft1), but also downregulation of 24-dehydrocholesterol reductase (Dhcr24). Upshot: Desmosterol accumulates, activates LXR signaling pathways, and suppresses inflammation. [Courtesy of Berghoff et al., 2021.]
Indeed, Saher said that’s exactly what happens. Using mass spectrometry, Berghoff found that desmosterol levels rose in mouse models of MS, but only when microglia synthesized squalene, a desmosterol precursor. In fact, adding squalene to mouse chow calmed microglial inflammation in MS mice, as did the Dhcr24 inhibitor, SH42. Squalene also reduced paralysis scores in the MS mice. So did the LXR agonist N,N-dimethyl-3β-hydroxycholenamide. It and squalene worked even better when given together.
All told, Saher explained, it is sterols, the very thing myelin is made of, that ultimately regulate microglial response to myelin injury. Curiously, even though microglia are packed with cholesterol they have phagocytosed from damaged myelin, the cells cannot simply recycle sterols from it. If they are to revert to homeostasis, they also need to boost sterol synthesis, because desmosterol cannot be made from cholesterol.
This work was done in mice and in cells. That said, it could be relevant to MS, and Alzheimer’s and other neurodegenerative disorders that feature demyelination (Roher et al., 2002; Aug 2021 news). If microglia in MS lesions cannot switch back to a non-inflammatory state, a vicious cycle of inflammation and demyelination may ensue, said Saher. Indeed, studies have found that Dhcr24 is down- and LXR target genes are upregulated in MS lesions. Might the same anti-inflammatory mechanisms be at play (Boven et al., 2006; Hendrickx et al., 2017; Mailleux et al., 2018)?—Tom Fagan
McNamara NB, Munro DA, Bestard-Cuche N, Uyeda A, Bogie JF, Hoffmann A, Holloway RK, Molina-Gonzalez I, Askew KE, Mitchell S, Mungall W, Dodds M, Dittmayer C, Moss J, Rose J, Szymkowiak S, Amann L, McColl BW, Prinz M, Spires-Jones TL, Stenzel W, Horsburgh K, Hendriks JJ, Pridans C, Muramatsu R, Williams A, Priller J, Miron VE.
Microglia regulate central nervous system myelin growth and integrity.
Nature. 2023 Jan;613(7942):120-129. Epub 2022 Dec 14
PubMed.
Correction.
Rojo R, Raper A, Ozdemir DD, Lefevre L, Grabert K, Wollscheid-Lengeling E, Bradford B, Caruso M, Gazova I, Sánchez A, Lisowski ZM, Alves J, Molina-Gonzalez I, Davtyan H, Lodge RJ, Glover JD, Wallace R, Munro DA, David E, Amit I, Miron VE, Priller J, Jenkins SJ, Hardingham GE, Blurton-Jones M, Mabbott NA, Summers KM, Hohenstein P, Hume DA, Pridans C.
Deletion of a Csf1r enhancer selectively impacts CSF1R expression and development of tissue macrophage populations.
Nat Commun. 2019 Jul 19;10(1):3215.
PubMed.
Fissolo N, Matute-Blanch C, Osman M, Costa C, Pinteac R, Miró B, Sanchez A, Brito V, Dujmovic I, Voortman M, Khalil M, Borràs E, Sabidó E, Issazadeh-Navikas S, Montalban X, Comabella Lopez M.
CSF SERPINA3 Levels Are Elevated in Patients With Progressive MS.
Neurol Neuroimmunol Neuroinflamm. 2021 Mar;8(2) Print 2021 Mar
PubMed.
Cracking the Cholesterol-AD Code: Metabolites and Cell Type
Since the 1990s, scientists have known cholesterol is important in Alzheimer's, but they gained little traction in their efforts to understand the relationship between the lipid and the disease. Though amyloid plaques, where the sterol rubs shoulders with Aβ and ApoE, might be a place to begin, cholesterol's omnipresence has hampered progress. Every cell in the brain makes and stores it. Some even pass it around to other cells. But as seen at the 2nd Symposium on Lipids in Brain Diseases, held September 13-15 in Leiden, The Netherlands, scientists are finally getting a grip on this slippery fellow by focusing on specific metabolites and specific cells.
Anil Cashikar, from St Louis, reported that ablating 25-hydroxycholesterol (25-HC) prevents aggregation of tau in mice, while Gesine Saher from Göttingen described how mice end up with more plaques if their microglia cannot make cholesterol. From Amsterdam, next door to the historic host city, Femke Feringa reported that not only do astrocytes pass cholesterol to neurons, but that neurons hand it back—perhaps to dump toxic lipids.
On display at the meeting venue, Rijkmuseum Boerhaave: microscopes of 18th century human biology research.
Cholesterol Metabolite and Tau
Cashikar, Washington University, St. Louis, knew from previous work with Steven Paul at WashU and Wenjie Luo at Weill Cornell Medicine, New York, that cholesterol 25-hydroxylase, the enzyme that makes 25-hydroxycholesterol, ticks up in AD and in mouse models of amyloidosis and neurofibrillary tangles (Wong et al., 2020). Cashikar was curious what exactly this metabolite was doing in neurodegenerative disorders because it had been reported to promote production of interleukin-1β from microglia, and because TREM2 and ApoE, two genes upregulated in activated microglia in AD and other tauopathies, promote 25-hydroxylase activity (Wong et al., 2020). Could this enzyme, and the cholesterol derivative, contribute to microglial responses in disease?
Saved by a Knockout. Neurons in hippocampal layer CA1 are lost in tauopathy mice (T) relative to wild-type (WT). Crossing in a 25-hydroxylase knockout (CKO) preserved neurons (bottom right). [Courtesy of Toral-Rios et al., 2023.]
To find out, Danira Toral-Rios in Cashikar’s lab turned to PS19 mice, which overexpress a human tau gene carrying the P301S mutation that causes frontotemporal dementia. She found that PS19 microglia express five times as much cholesterol 25-hydroxylase as do cells in wild-type mice. To see if this affects pathology, Toral-Rios mated PS19s with cholesterol 25-hydroxylase knockout (ChKO) mice. She found that 9.5-month-old crosses had much less atrophy of the piriform/entorhinal cortex than did PS19 parents. They had normal numbers of cells in the CA1 layer of the hippocampus (image at right), and of synapses in the CA3 layer. The crosses produced less phospho-tau and accumulated fewer tangles in the hippocampus and cortex, said Cashikar.
How could reducing 25-hydroxycholesterol cause such profound change? Toral-Rios found little astrogliosis and microgliosis in the crosses, suggesting inflammatory responses were dialed down. Measuring microglial markers clec7a, ApoE, and TREM2 indicated fewer disease-associated microglia in the PS19/ChKO mice than PS19 controls. Even so, the crosses had fewer homeostatic microglia than did wild-type mice, an indication that merely cutting 25-HC out of the picture does not fully restore cell homeostasis in PS19 mice.
25-HC Ramification. Inflammatory microglia (green) abound in PS19 brain (left). They ball up (bottom left), a morphological sign of cell activation. When the mice lack cholesterol 25-hydroxlyase (right), few microglia shift to inflammation (top), more retain their ramified homeostatic shape (bottom). [Courtesy of Toral-Rios et al., 2023.]
More evidence for tempered immune responses came from transcriptomics. As per network analysis of downregulated genes, inflammatory pathways were off in the PS19/ChKO mice, including those controlled by NF-kB and Jak/Stat signaling, the latter of which is implicated in neurodegeneration in PS19 mice (Litvinchuk et al., 2018).
How does this intersect with tau? In vitro, Cashikar’s group found that tau fibrils induce IL-1β and Cxcl10, a chemokine, in wild-type primary microglia but not microglia from the hydroxylase knockouts. IL-1β is a well-known inflammatory cytokine, while Cxcl10 is suspected of beckoning T cells into the brain (Mar 2023 news; Sep 2023 news).
Indeed, Toral-Rios found much less of the T cell CD3 in the brains of 9.5-month-old PS19/ChKO knockouts than PS19 mice. Microglia from the knockouts made more of the anti-inflammatory cytokine IL-10, whether they were challenged with tau fibrils or not. That was a surprise, said Cashikar. “It suggests that without the hydroxylase, or 25-hydroxycholesterol, microglia are generally less inflammatory,” he said.
All told, the data suggest that when microglia contact tau fibrils, they fire up inflammatory signaling in a way that involves 25-HC.
How this metabolite figures here is unclear, said Cashikar. That said, tamping it down restored normal lipid profiles. The PS19 cortex contains abnormally high levels of cerebrosides, sphingomyelins, ceramides, and phospholipids such as phosphatidylserines and phosphatidylinositols; some of these might worsen inflammation. Ceramides and sphingomyelins, for example, are known to do as much (Alessenko and Albi, 2020). Cashikar thinks blocking 25-hydroxylase activity or expression therapeutically might bring down inflammation in tauopathies.
Cholesterol and Alzheimer's
Saher’s group, at the Max-Planck Institute of Experimental Medicine, Gottingen, Germany, took a different tack. They asked what happens when they mess with cholesterol itself.
Previous hints that cholesterol acts at the cellular level had come from meeting co-organizer Rik van der Kant's postdoc in Larry Goldstein’s lab at the University of California, San Diego. Van der Kant had found that, in neurons derived from familial AD stem cells, cholesteryl esters drove release of Aβ and accumulation of p-tau231 (Feb 2019 news). Others had reported that membrane cholesterol renders hippocampal neurons susceptible to Aβ and tau toxicity (Apr 2009 news). This implied that lowering cholesterol might slow AD, but subsequent clinical trials were negative (e.g., atorvastatin, simvastatin).
Statins may have failed because the brain synthesizes its own cholesterol. All cells in the CNS make it. Still, Saher wondered if there is a source of cholesterol that is particularly relevant to amyloidosis. To test this, Lena Spieth in the lab crossed 5xFAD mice, which develop plaques at 2 months old, with mice that had cholesterol synthesis enzymes knocked out in specific cell lineages. She examined plaques with light-sheet microscopy, in which a thin “sheet” of light excites fluorescent tags in one focal plane, reducing photobleaching and improving resolution.
Spieth found that knocking out cholesterol synthesis in 5xFAD neurons barely affected amyloid load. Amyloid precursor protein processing also proceeded apace. Might astrocytes compensate for the loss of neuronal cholesterol? Spieth looked at them next. Deleting cholesterol synthesis in astrocytes did reduce plaque load, suggesting that these glia supply neurons with the cholesterol they need to produce Aβ.
What about amyloid clearance? Scientist believe microglia phagocytose Aβ and package any they can’t degrade into dense-core plaques, compacting the overall amyloid load (May 2016 news). When Spieth shut off microglial cholesterol synthesis, amyloidosis increased and the microglia no longer rallied around plaques.
What's more, microglia failed to upregulate many of the genes they usually call upon in response to amyloid, such as Cst7, Axl, and Clec7a, suggesting that microglia need local synthesis of cholesterol to gear up. These genes are turned up in DAM, aka disease-associated microglia, known from mouse models of AD.
Scientists at the meeting liked Saher’s data. They asked her how the DAM signal was suppressed. Saher had also reported that the cholesterol precursor desmosterol tamed inflammatory microglia in models of multiple sclerosis by activating liver X receptor signaling (see Part 2 of this series), raising the question if cholesterol might do the same in AD. She thinks it is possible, but likely more complicated than in myelin disease.
Others inquired about ApoE4. “That’s a question we’ve been struggling with for 50 years,” quipped Saher. ApoE4 binds cholesterol less well than do ApoE2 or E3, but ApoE4s effects are multifacted, including changes to peripheral lipid metabolism. Two recent studies concluded that ApoE4 prevents microglia from transitioning to disease-associated states (Yin et al., 2023; Liu et al., 2023).
The Cholesterol Shuffle
Saher's data place cholesterol transport between astrocytes and neurons at the center of amyloid deposition. In her talk, Feringa, from Vrije University, Amsterdam, went a step further, reporting that transport between astrocytes and neurons is a two-way street. Not only do astrocytes pass cholesterol to neurons, as is well known, but in cell culture experiments, she found that neurons transport cholesterol to astrocytes, and also to microglia. This depends on ApoE.
Feringa used human isogenic iPSC lines engineered to carry ApoE2, 3, or 4 alleles (Schmid et al., 2021). She first compared the lipidomes of neurons, astrocytes, and microglia derived from these cells, and then asked how APOE genotype affected them. She found the three cells had substantially different lipid profiles. For example, phosphatidyl choline and ceramide predominated in neurons, cholesteryl esters and diacylglycerol in astrocytes, and lysophospholipids and ceramide derivatives in microglia. The latter had more lipids that are involved in cell signaling pathways than in metabolism or membrane structure, Feringa found. Next, ApoE genotype. APOE4 neurons contained smaller quantities of cholesteryl esters and triglycerides than did APOE3 neurons. The opposite was true of astrocytes.
Since ApoE carries lipids, Feringa thought it might help transport them in and out of these cells. To test this, she used a layer co-culture system. She grew neurons or glia in dishes, then after the cells had matured, layered cells from one culture on top of the other and assessed lipid movement a few days later. Lo and behold, when neurons were grown at the bottom with fluorescently tagged cholesterol or fatty acid, the tags ended up in astrocytes above. Neurons also transferred cholesterol to microglia. This worked only half as efficiently in ApoE4 cells.
The transport from neurons to glia surprised Feringa. “We know that once the brain has matured, astrocytes transport cholesterol to neurons via ApoE, but this data suggests the transport is less unidirectional than we thought,” Feringa told Alzforum. Because ApoE4 seems to suppress the process, she thinks it may be relevant to disease. Indeed, she ventured that the transfer may be a way to protect neurons. Some recent studies showed that if neurons accumulate reactive oxygen species, e.g., when they are hyperactive, these ROS can generate peroxidized lipids, which are toxic. “This might be a way to get rid of those lipids,” Feringa suggested.
Scientists at the meeting noted that Feringa’s data came from healthy cells, asking what might happen in a disease setting. Feringa is testing the effect of Aβ and hypoxia on the lipid transport. Others thought it important to find out how this transport happens, suggesting she look to extracellular vesicles. As for the consequences of this reverse transport, Feringa did report that when astrocytes take up cholesterol from neurons, they secrete interleukin-6. How this inflammatory cytokine affects surrounding cells remains to be seen.—Tom Fagan
Inflammation can be a helpful response to the common cold or the bite of an insect, but dangerous if it persists. In neurodegeneration, chronic micro- and astrogliosis spell trouble for the brain. Does it have to be that way? Over the past 20 years, scientists have discovered “specialized pro-resolving mediators” that take the sting out of inflammation. Derived from long-chain fatty acids, these SPMs are potent, yet hard to come by in the body.
As described at the 2nd Symposium on Lipids in Brain Diseases, there may be ways to gin them up. Oliver Werz, Friedrich Schiller University in Jena, Germany, described small molecules that fool lipid oxygenases into producing SPMs instead of their inflammatory cousins, such as prostaglandins. Amsterdam University's Julia Konings reported ways to shift the microglial lipidome toward a “pro-resolution” phenotype. Whether these strategies could help reduce inflammatory responses in the brain needs to be studied, but some of these SPMs wane in the AD brain.
Before we delve into the, ahem, “meat” of this fat story, a bit of background on the subtle differences between the inflammatory and pro-resolving lipids. The former mostly come from arachidonic acid, a 20-carbon fatty acid with four double bonds. Cyclooxygenases, the target of vioxx, celecoxib, and some other non-steroidal anti-inflammatories, generate a stew of pro-inflammatory prostaglandins and thromboxanes from this fatty acid by forming an aliphatic ring in the middle of the carbon chain. 5-Lipoxygenase (5-LOX) creates another group of inflammatories, the leukotrienes, by introducing a hydroxyl group on carbon number 5, hence the name 5-lipoxygenase. Here’s where it starts to get complicated. 5-LOX also makes some of the SPMs, namely the lipoxins (image below). The lipoxins can also be made from another 20-carbon fatty acid, eicosapentaenoic acid. The penta here signifies that this fatty acid has five double bonds, one more than in arachidonic acid, which is a less technical name for eicosatetraenoic acid.
There’s more. Wait till you hear about resolvins and protectins. (If you can deal with multi-omic datasets, you can handle this biochemistry.) These SPMs form when yet another enzyme, 15-LOX, oxidizes eicosapentaenoic acid or eicosatetraenoic acid. See the nomenclature pattern here—15-LOX adds a hydroxyl group to the carbon 15 of the fatty acids. Other enzymes jump in to create varieties of lipoxins, resolvins, and protectins, but the 5- and 15-LOXs are the main drivers. Yet another set of SPMs, the maresins, form when 12-LOX oxidizes docosahexaenoic acid—a 22 carbon fatty acid with, you guessed it, six double bonds.
Lives of Lipids. Long-chain fatty acids give rise to a plethora of inflammatory lipids (red) and pro-resolving mediators (blue). [Courtesy of Oliver Werz.]
Could this web of relationships be tweaked to shift the balance from pro-inflammatory to pro-resolving? Werz believes so. However, a major obstacle has stymied this line of research. 15-LOX has been notoriously difficult to activate. Scientists have a multitude of tools at hand to induce prostaglandins and leukotrienes, including lipopolysaccharide from bacteria, zymosan from fungi, complement 5a, calcium ionophores, even cell stressors such as hyperosmosis; alas, they could not figure out how to stimulate SPMs until researchers in Charles Serhan’s lab at Brigham and Women’s Hospital, Boston, discovered that bacteria do the trick (Chiang et al., 2012). While in Serhan’s lab, Werz found that E. coli and S. aureus, counterintuitively perhaps, did so by stimulating 15-LOX in pro-resolving macrophages (Werz et al., 2018).
It now appears that some noninfectious agents mimic this effect. In Leiden, Werz reminded the audience that α-hemolysin, a toxin produced by S. aureus, bumps up production of maresins and resolvins when injected into the peritoneum of mice (Jordan et al., 2020). Better than using the bacteria, but still a toxin. What about other approaches?
Blocking cyclooxygenase to limit production of prostaglandins may seem an option, but Werz emphasized that this only tempers acute inflammation. Cox inhibitors do poorly in the long run because they prevent formation of prostaglandin E2, one of the few anti-inflammatory prostaglandins, and because with no cyclooxygenase to act on arachidonic acid, 5-LOX begins to churn out the inflammatory leukotrienes. Indeed, macrophages treated with celecoxib ramp up production of leukotriene B4 up to fourfold. The key to flipping the “lipid class switch,” i.e., producing more SPMs at the expense of inflammatory lipids, lies in 5-LOX, said Werz.
In his talk, Werz showed how he could turn 5-LOX into 15-LOX using an allosteric modulator called acetyl-keto-boswellic acid. No, not named after the Scottish author, AKBA comes from plants in the genus Boswellia. (It includes frankincense). Unlike 5-LOX inhibitors that sit in the active site of the oxygenase, AKBA sits at a distal site. From there, it breaks hydrophobic bonds between amino acid side chains and causes a conformational ripple through the enzyme right down to where arachidonic acid docks (Gilbert et al., 2020). The upshot: The fatty acid repositions in the active site in such a way that carbons 12 or 15 get attacked and oxidized instead of carbon 5.
Lipoxygenase Switch. In 5-lipoxygenase (left), AKBA (yellow) sits between the amino (gray) and catalytic domains (cyan). By disturbing that interface (right) it restructures the active site (red circle). [Courtesy of Gilbert et al., 2020.]
Werz’s group found that in cell-free systems, in HEK293 cells, and in human macrophages, monocytes and leukocytes, AKBA shifted lipid profiles toward the pro-resolving mediators (Gilbert et al., 2020; Börner et al., 2023). As a bonus, AKBA also increased 15-LOX activity, reported Werz, boosting SPMs even more. Injected into the peritonea of mice, the allosteric modulator evoked a huge increase in SPMs when the animals were injected with zymosan (image below).
Allosteric Approach. AKBA, a modulator of 5-Lipoxygenase from frankincense and other Boswellia species, boosts 12- and 15-lipoxygenase activity and production of SPMs in macrophages, monocytes, and leukocytes. [Courtesy of Oliver Werz.]
Could AKBA resolve inflammation in people? A hint comes from an eight-month, open-label clinical trial of a frankincense extract in 28 people with relapsing-remitting multiple sclerosis. Within five months of starting treatment, the number of myelin lesions in the brain was reported to go down, as judged by MRI, and remain low until month eight (Stürner et al., 2018). Of 44 lipids tested in the plasma during that time, the concentrations of eight fell; seven of those were products of 5-LOX (Stürner et al., 2020).
Participants who continued the treatment had no or few relapses up to month 36, though no further MRI scans were taken during that time. Werz said he does not know of any follow-up trial of this extract in MS, and he thinks other compounds would be worth testing. Screening a variety of natural anti-inflammatories, scientists in his lab found that celastrol and cannabidiol act similarly to AKBA, inhibiting pro-inflammatory and promoting pro-resolving mediators in mice (Pace et al., 2021; Peltner et al., 2023).
A different strategy to tackle inflammation in MS—though still focused on 5-LOX—was presented by Konings, who works in Gijs Kooij’s lab in Amsterdam. Last June, Jelle Broos and colleagues in the same lab reported that all was not well with arachidonic acid metabolism in MS (Broos et al., 2023). Specifically, in people with the progressive form of the disease, plasma levels of lipid metabolites, including leukotrienes produced by 5-LOX, correlated with disease severity. Konings postulated that 5-LOX activating protein (FLAP), which transfers lipids to 5-LOX, might be involved, but little was known about FLAP in MS.
Why Hello, Microglia
Konings looked to see where 5-LOX and FLAP are made in the brain. While MS and control brains contained similar levels of 5-LOX, MS brains had more FLAP, mostly around active lesions. Co-staining with IBA1 and TMEM119 identified microglia as the source. To investigate, Konings used microglia derived from human induced pluripotent stem cells. When she challenged them with inflammatory molecules, such as lipopolysaccharide or interferon-γ, they churned out five times as much FLAP as did control cells.
Could this be modulated? When Konings treated microglia with experimental FLAP inhibitors such as Fiboflapon or AZX5718, the cells' lipidome shifted from a pro-inflammatory to a pro-resolving profile, suppressing leukotrienes and other 5-LOX products. How this affects microglial function needs to be worked out, but preliminary data suggest that they reduce expression of the inflammatory cytokine interleukin 1β.
At the Leiden conference, scientists were excited at the potential for shifting lipidomes in this way. “The concept is very promising,” Kooij told Alzforum. “If you are not just targeting one molecule, but shifting a whole suite of molecules from inflammation to protection, that could be extremely potent.”
Even so, they had many questions. Are SPMs perturbed in diseases besides MS? What triggers their production? How do they work? “We don’t have the answers yet,” said Werz. Kooij told Alzforum that these lipid mediators have been extremely difficult to study because they are so hard to detect. “We believe they have direct effects only on nearby cells and, as such, they are only produced at really low concentrations,” he said.
There are signs that these molecules are important in other diseases. Previous studies reported that resolvins and lipoxins drop in the Alzheimer's brain. These same compounds tempered inflammation in mouse models of the disease (Zhu et al., 2016; Apr 2018 news; Kantarci et al., 2018).
As for physiological triggers of SPMs, these might include certain cytokines, said Kooij, adding that this remains to be studied. Werz thinks specific receptors activate 15-Lox. Curiously, one might be ADAM10, the α-secretase that processes amyloid-precursor protein. ADAM10 binds α-hemolysin.
Downstream, SPMs mainly target immune cells, stimulating pro-resolving macrophages and blocking neutrophil infiltration and migration. How they do this needs to be understood. They appear to start out by binding G-protein coupled receptors. “At least six receptors have been proposed, but it is not entirely clear which might mediate effects of SPMs,” said Werz. It’s a buzzing area of investigation. “I think we will hear a lot more about this in the coming year,” Kooij predicted.—Tom Fagan
Comments
No Available Comments
Make a Comment
To make a comment you must login or register.