6 May 2011. Kids will often resist when explicitly asked to quiet down, but acquiesce if coaxed indirectly. Could β-secretase (BACE1) be similar? Blocking this membrane-bound protease could halt production of amyloid-β, one of the pathological peptides lurking in the brains of Alzheimer’s disease patients. However, some compounds that directly inhibit BACE1 have flopped in early AD clinical trials, forcing researchers to seek other ways to put a lid on the enzyme. In this week’s Journal of Neuroscience, researchers led by Taisuke Tomita and Takeshi Iwatsubo at the University of Tokyo, Japan, unveil a new way to hush BACE1. In their study, reducing levels of sphingosine-1-phosphate (S1P), a lipid that modulates extracellular signals, caused a drop in BACE1 activity and, correspondingly, lower Aβ production in vitro and in AD transgenic mice. Plus, the scientists found that S1P binds BACE1’s transmembrane region, and that AD patients have elevated brain levels of a kinase that generates S1P, suggesting the pathway may have clinical relevance.
“This paper is very well done,” said Philip Wong of Johns Hopkins University in Baltimore, Maryland. “The various approaches taken were multifaceted and demonstrate clearly that affecting BACE1 through these lipid modulators could be promising.” However, slashing S1P levels could also produce undesirable side effects, Wong noted, given the lipid’s key roles in the immune and vascular systems.
Early attempts to calm BACE1 activity relied primarily on peptide-based inhibitors directed against the protease’s catalytic domain. Those compounds work well in vitro, but have not gotten the job done where it counts—in dementia patients—in large part because they barely reach the brain. To overcome this problem, Iwatsubo and colleagues hunted for smaller, non-peptidic molecules that reduce Aβ production in neurons. That screen, done in collaboration with scientists at Takeda Pharmaceutical Company, Osaka, yielded a lipophilic compound (TAK-070) that binds the membrane-spanning region of BACE1 and lowers Aβ pathology when given orally to Tg2576 AD mice (ARF related news story on Fukumoto et al., 2010).
The new report details another compound from that Aβ-lowering screen. The molecule piqued the interest of the researchers because it acted on neuronal BACE1 but did not slow the enzyme in HEK293 embryonic kidney cells, suggesting the compound might be neuron-specific, Tomita told ARF. As it turns out, the compound (SKI II) inhibits SphK1 and SphK2, the kinases that phosphorylate sphingosine to generate S1P, and had been previously identified for its anti-tumor properties (French et al., 2003; French et al., 2006). In this study, first author Nobumasa Takasugi and colleagues showed that SKI II, as well as two structurally diverse SphK inhibitors, dampened Aβ production in a dose-dependent manner by reducing β-cleavage of amyloid precursor protein (APP) in mouse primary cortical neurons, and in mouse neuroblastoma N2a cells. The researchers saw similar results when they knocked down SphK1 or 2 with RNA interference, or overexpressed S1P degrading enzymes, in N2a cells. Conversely, overexpression of SphK2 drove up Aβ production in APP-expressing N2a cells.
S1P levels correlated with BACE1 activity measured in vitro from N2a membrane fractions. However, SKI II had no effect on BACE1 catalytic activity, suggesting that the compound modulates the secretase indirectly, through SphK suppression.
S1P binds to G protein-coupled receptor (GPCR)-type receptors on the cell surface, but also seems to have intracellular targets. The secretase activity assays likely reflect S1P’s effect on intracellular Aβ, the scientists note, since S1P failed to restore Aβ secretion when applied extracellularly to N2a cells treated with SKI II or with SphK2 knocked down. Ongoing experiments with a different compound—an S1P receptor antagonist—suggest that S1P also acts from outside the cell, Tomita told ARF. This compound, FTY720 (aka Fingolimod), has immunosuppressive properties and recently gained approval in the U.S. and Europe for treating multiple sclerosis.
The current study also confirmed, using pulldown experiments, direct binding between S1P and membrane-bound BACE1. Coupled with the earlier report on TAK-070 (Fukumoto et al., 2010), the research further suggests that “the transmembrane domain of BACE1 is a druggable target for BACE1 activity,” Tomita said in an interview with ARF. This reflects the general direction of the field, which is now beginning to focus on BACE modulation, rather than inhibition, as a potentially more viable therapeutic approach (see ARF conference story).
To explore effects of reducing S1P in an AD context, the University of Tokyo team injected SKI II into the brains of eight-week-old wild-type mice. Eight hours later, the mice were sacrificed. Treated animals showed a modest decrease in hippocampal Aβ levels, compared with vehicle-injected controls. The researchers did another set of experiments with six-month-old A7 transgenic mice overexpressing human APP with Swedish and Austrian mutations (Yamada et al., 2009). Here, they found reduced levels of brain Aβ after a six-day oral treatment with the SphK inhibitor. Based on its chemical properties, the scientists believe the compound has good brain penetration, but have not yet formally measured this in the AD mice, Tomita said. The lab is also doing a longer-term treatment study in this A7 strain.
In an interesting twist, the scientists found that treating N2a cells with fibrils made from synthetic Aβ1-42 heightened the cells’ SphK2 activity. Furthermore, they found that postmortem brain samples from AD patients had less SphK2 protein—yet greater SphK2 activity—than control specimens from non-demented elderly. These data hint at a possible vicious cycle whereby fibrillar Aβ increases production of cell-associated S1P, which in turn stimulates BACE1 activity, causing neurons to make even more Aβ. If this is true, “we think S1P may be a novel biomarker for AD,” Tomita told ARF. “We may be able to measure BACE activity in the brain using S1P activity.”
An important concern is whether long-term treatment with SphK inhibitors would stymie other important pathways. Recent studies show that S1P binds and inhibits the enzymatic activity of histone deacetylases (Hait et al., 2009), and can stimulate E3 ligase activity (Alvarez et al., 2010). Plus, S1P has important functions in the immune system (see Chi, 2011 for a review). Further studies will need to explore these possible side effects, Tomita said.—Esther Landhuis.
Takasugi N, Sasaki T, Suzuki K, Osawa S, Isshiki H, Hori Y, Shimada N, Higo T, Yokoshima S, Fukuyama T, Lee V, Trojanowski JQ, Tomita T, Iwatsubo T. BACE1 activity is modulated by cell-associated sphingosine-1-phosphate. J. Neurosci. 2011 May 4;31(18):6850-6857. Abstract