Aβ may start it, but neurons keep it going—“it” being persistent microglial activation. According to a mouse study in the April 16 Nature Communications, Aβ triggers a reduction in sphingosine kinase 1, an enzyme that prompts neurons to secrete lipids that soothe inflammation and enhance microglial phagocytosis. Two of these anti-inflammatory molecules, resolvin E1 and lipoxin A4, were reduced in postmortem AD brain tissue, according to the paper. Led by Jae-Sung Bae and Hee Kyung Jin of Kyungpook National University in Daegu, South Korea, the authors drew a signaling pathway controlled by the kinase, and suggested activating it as a therapeutic strategy for AD. Oddly enough, echoing the modus operandi of aspirin, the strategy hinges on SphK1 acting as an acetylase of cyclooxygenase 2, and increasing production of resolvins. Gregory Cole of the University of California, Los Angeles, found that connection interesting. “This report raises questions about the potential of aspirin to reduce AD risk,” he wrote.

  • AD mice had too little SphK1, a kinase that promotes inflammatory resolution.
  • SphK1 promotes the secretion of lipids that enhance microglial phagocytosis and lessen damaging inflammation.
  • Boosting SphK1 expression reduced plaque load and memory problems.

“This new pathway can explain the increased neuroinflammation in AD and other neurodegenerative diseases,” commented Charles Serhan of Brigham and Women’s Hospital in Boston, who discovered these lipid mediators. “This pathway may also be operative in other diseases where inflammation is uncontrolled and resolution fails, leading to chronic inflammatory states.”

In their homeostatic state, microglia stretch out their processes and keep house, pruning withering synapses, gobbling up bits of detritus, and generally sensing the state of the brain. Once set off by threats such as neuropathology, these formidable cells round up and spew out pro-inflammatory cytokines. A failure of microglia to snap out of this posture has been observed in AD. It can lead to damaging neuroinflammation and prevent microglia from resuming critical functions such as phagocytosis (Apr 2018 news; Sep 2017 news). Enter the specialized proresolving mediators (SPMs). Derived from polyunsaturated fatty acids, this class of lipid molecules orchestrates the resolution phase of the microglial inflammatory response (Serhan et al., 2002; Buckley et al., 2014). Studies suggest SPMs are released by neurons, promote microglial phagocytosis of Aβ42, and are reduced in the AD brain (Wang et al., 2014; Zhu et al., 2016). 

Aβ Clearance. Microglia (red) surrounded Aβ plaques (green) in APP/PS1 mice (top and bottom), but engulfed more Aβ42 (yellow, merge) when SphK1 was overexpressed (bottom). [Courtesy of Lee et al., Nature Communications, 2018.]

For the current study, first author Ju Youn Lee and colleagues set out to investigate the role in AD of a different lipid molecule, sphingosine-1-phosphate (S1P). An integral part of membrane phospholipids, sphingosine also transforms into a potent bioactive signaling molecule once phosphorylated by sphingosine kinase (SphK). S1P has been implicated in myriad cellular processes, including inflammation, autophagy, and cellular survival, and reportedly promotes pro-inflammatory responses in microglia (Melendez, 2008; Nayak et al., 2010Lin et al., 2011). To investigate, the researchers first asked whether expression of either of two isoforms—SphK1 or SphK2—were altered in the brains of APP/PS1 mice. They found less mRNA for SphK1, but not SphK2, in APP/PS1 than wild-type mice. SphK1 expression was lower only in neurons, not in microglia or astrocytes. However, despite the drop in neuronal SphK1 expression, the researchers found normal levels of S1P in mouse brain, and further experiments revealed that SphK2 compensated, at least in terms of phosphorylating sphingosine.

SphK1 expression started dropping in APP/PS1 animals at six months of age, when plaques started to emerge. To investigate how SphK1 might affect Aβ pathology, the researchers generated APP/PS1 mice that overexpressed SphK1 in the brain, including in neurons, in microglia, and in astrocytes. Despite this overexpression, levels of S1P in the animals’ brains were normal, again suggesting that SphK1 was not the primary kinase phosphorylating sphingosine.

Strikingly, nine-month-old APP/PS1-SphK1-Tg mice had far fewer cortical plaques and less soluble brain Aβ than APP/PS1 controls. The SphK1 boost also spared them from losing synapses, and from faltering in the Morris water maze test of spatial memory. In all, something about SphK1 activity, other than phosphorylation of S1P, appeared to ease plaque clearance and protect memory.

An improved neuroinflammatory profile stood out. APP/PS1 mice overexpressing SphK1 had lower concentrations of pro-inflammatory cytokines in the brain, including TNF-α, IL1β, and IL-6, but more of the anti-inflammatory cytokines IL-4, TGF-β, and Arg1. The mice also had fewer activated microglia and astrocytes, judging by the markers Iba1 and GFAP. Notably, compared to plaque-associated microglia in APP/PS1 animals, those in mice overexpressing SphK1 had more phagolysosomes loaded with Aβ, an amoeboid morphology, and expressed the phagocytic marker CD36. Plaques in APP/PS1-SphK1-tg mice were smaller than in APP/PS1 mice, suggesting microglia had feasted on their outer rims (see image below).

Lipid Resolvers. Following an initial burst of inflammatory prostaglandins, proresolving mediators such as lipoxins (bottom of A) and resolvins (B) are synthesized by COX2. Acetylation of this enzyme by aspirin inhibits prostaglandin synthesis, but promotes synthesis of some lipoxins and resolvins. [Courtesy of Whittington et al., Frontiers in Immunology, 2017.]

Because previous studies reported that SPMs quell inflammation and stimulate phagocytosis, the researchers measured their secretion from neurons. Primary cortical neurons from APP/PS1 mice secreted less lipoxin A4 and resolvin E1 than did wild-type neurons, but neurons from APP/PS1 mice overexpressing SphK1 secreted normal levels of the two SPMs, suggesting the kinase somehow stimulated release of these inflammation dousers. Mass spectrometry to compare levels of SPMs in mouse brain extracts led to similar conclusions. A series of biochemical experiments connected the dots between SphK1 and secretion of SPMs.

Curiously, the kinase appears to moonlight as an acetyltransferase. According to the authors, it snatches an acetyl group from acetyl-CoA and transfers it onto cyclooxygenase 2 (COX2) at serine-565. This acetylation activates COX2, which helps convert polyunsaturated fatty acids into lipoxins and resolvins (see image above). Neurons in APP/PS1 animals had less acetylated COX2 than those in wild-type animals, or those in APP/PS1 mice overexpressing SphK1.

Interestingly, in a chemical reaction, aspirin also acetylates COX2, which inhibits the enzyme’s ability to convert arachidonic acid into pro-inflammatory mediators such as prostaglandins, while promoting the enzyme’s ability to generate some anti-inflammatory SPMs. How a kinase can double as an acetylase is unclear. The authors write that they are investigating, but did not respond to requests for an interview.

In support of their mouse data, the researchers report lower concentrations of SphK1, but not SphK2, in postmortem brain samples from AD patients and in neurons derived from induced pluripotent stem cells from multiple AD patients who carried ApoE4 or a presenilin mutation. Compared to neurons from controls, the AD neurons had less acetylated COX2 and secreted fewer SPMs. Both were restored to normal when the researchers treated the neurons with an activator of phosphoinositol-3-kinase, which spurs SphK1 expression.

Overall, the findings suggested that SphK1-mediated secretion of SPMs from neurons promotes microglial phagocytosis and thwarts damaging inflammation. Aβ somehow reduces the expression of neuronal SphK1, locking microglia into a persistent pro-inflammatory state. “These findings describe new properties and functions for SphK1, identify new mechanisms contributing to AD, and suggest a new therapeutic approach for AD and perhaps other inflammatory conditions by modulating SphK1 activity,” the authors concluded.

Serhan found the results exciting. He previously reported that acetylation of COX2 by aspirin promoted the production of SPMs, while blocking production of prostaglandins (Serhan et al., 2000). He thinks Lee and colleagues have discovered an endogenous pathway that does something similar. “The discovery of Lee et al. uncovers an endogenous enzyme system to produce these mediators that is diminished in AD,” he wrote.

Cole pointed out that the paper addresses important mechanisms governing the effective clearance of Aβ. “This paper offers a new pathway for neuronal regulation of microglia which is provocative and has various implications,” he wrote to Alzforum (see full comment below).—Jessica Shugart


  1. Given that anti-Aβ immunotherapy induces amyloid phagocytosis and clearance and is being tested in the clinic, the mechanisms that limit natural clearance via this pathway are obviously important. This paper is data-rich and makes a number of interesting and novel claims about the mechanisms limiting amyloid clearance by microglia. In brief, it argues that impaired microglial amyloid phagocytosis in AD is caused by a reduction in neuronal SphK1 that results in less acetylation of COX-2, which in turn leads to deficits in a proresolving lipid mediator, LXA4, a COX-2 product promoted by acetylation of COX-2. LXA4 was previously identified as pro-phagocytic and has been shown by Meideros and others to promote amyloid clearance in 3xAD Tg mice.

    First, this paper offers a new pathway for neuronal regulation of microglia that is provocative and has various implications. For example, it may help explain the anti-inflammatory activity of the second wave of COX-2 induction at 48 hours described by Gilroy and Willoughby (Gilroy et al., 1999). It may help explain how high dose continuous exposure to classical COX inhibitor NSAIDs may increase amyloid accumulation, as was observed previously (Sonnen et al., 2010). This ties in to the issue of NSAID dosing, which is typically intermittent in community epidemiology but chronically high in intervention trials. If chronic COX-2 inhibitors block protective LXA4 production, episodic inhibition with shorter half-life drugs like ibuprofen could avoid LXA4 deficits.

    Because aspirin can also promote acetylated COX-2 and LXA4 generation, this report also raises questions about the potential of aspirin to reduce AD risk. The literature is mixed on this issue but generally suggests that 80mg baby aspirin dosing doesn’t reduce AD risk, while the higher dosing that is required to promote COX-2 acetylation may be protective, notwithstanding some risk for hemorrhage. The report is also interesting in that it highlights the significant potential immunomodulatory protection from an n-6 anti-inflammatory lipid (LXA4) that may help explain the puzzling epidemiology of AD protection with high linoleic acid intake, which is allegedly pro-inflammatory, but not in our experiments with mice. In general, this study suggests an interesting new mechanistic pathway that deserves follow-up and independent efforts to confirm and extend.  


    . Inducible cyclooxygenase may have anti-inflammatory properties. Nat Med. 1999 Jun;5(6):698-701. PubMed.

    . Nonsteroidal anti-inflammatory drugs are associated with increased neuritic plaques. Neurology. 2010 Sep 28;75(13):1203-10. PubMed.

  2. The paper from Lee and collaborators is a great example of how much remains yet to discover with respect to nervous system cellular interactions. Microglia, the brain-resident macrophages, are highly tuned to local environmental cues. Yet, they don’t act in isolation. It is evolutionarily advantageous for other neural cell types to signal microglial responses using soluble factors. The present report investigated the ability of neurons to secrete specialized mediators through sphingosine kinase type 1 (Sk1) to signal microglia to re-establish homeostasis. The net positive effect is mitigation of Alzheimer-like pathology. Specifically, the authors posit that neuronal Sk1 promotes lipoxin A4 secretion, which increases microglial Aβ phagocytosis. Interestingly, APP/PS1 mice bearing transgenic neuronal Sk1 present with decreased expression of brain cytokines, including interleukin-10, which we have previously shown to restrict microglial phagocytosis of Aβ. This work nicely illustrates the importance of intercellular communication at the brain-immune interface.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Stuck in the Past? Microglial Memories Dictate Response to Aβ
  2. ApoE and Trem2 Flip a Microglial Switch in Neurodegenerative Disease

Paper Citations

  1. . Resolvins: a family of bioactive products of omega-3 fatty acid transformation circuits initiated by aspirin treatment that counter proinflammation signals. J Exp Med. 2002 Oct 21;196(8):1025-37. PubMed.
  2. . Proresolving lipid mediators and mechanisms in the resolution of acute inflammation. Immunity. 2014 Mar 20;40(3):315-27. PubMed.
  3. . Resolution of inflammation is altered in Alzheimer's disease. Alzheimers Dement. 2014 Feb 12; PubMed.
  4. . Pro-Resolving Lipid Mediators Improve Neuronal Survival and Increase Aβ42 Phagocytosis. Mol Neurobiol. 2016 May;53(4):2733-49. Epub 2015 Dec 9 PubMed.
  5. . Sphingosine kinase signalling in immune cells: potential as novel therapeutic targets. Biochim Biophys Acta. 2008 Jan;1784(1):66-75. Epub 2007 Aug 14 PubMed.
  6. . Sphingosine kinase 1 regulates the expression of proinflammatory cytokines and nitric oxide in activated microglia. Neuroscience. 2010 Mar 10;166(1):132-44. PubMed.
  7. . Expression of sphingosine kinase 1 in amoeboid microglial cells in the corpus callosum of postnatal rats. J Neuroinflammation. 2011 Feb 11;8:13. PubMed.
  8. . Novel functional sets of lipid-derived mediators with antiinflammatory actions generated from omega-3 fatty acids via cyclooxygenase 2-nonsteroidal antiinflammatory drugs and transcellular processing. J Exp Med. 2000 Oct 16;192(8):1197-204. PubMed.

Further Reading

Primary Papers

  1. . Neuronal SphK1 acetylates COX2 and contributes to pathogenesis in a model of Alzheimer's Disease. Nat Commun. 2018 Apr 16;9(1):1479. PubMed.