M EMBRANE ORDER CONTROLS TAU SECRETION

If tau secretion occurs via translocation through the plasma membrane, then the modulation of plasma membrane properties, such as membrane order and fluidity, would be expected to affect tau secretion. Membrane fluidity is a parameter that measures the ability of molecules to undergo rotational and diffusional motions within the membrane. Membrane fluidity depends on external (such as temperature, pressure, pH) and internal (membrane composition) factors [686]. Factors that affect lipid order in the membrane also affect fluidity, as the higher the lipid order, the more restricted are the motions of molecules. Generally, cholesterol, sphingolipids, and longer fatty acids with a lower degree of unsaturation decrease lipid order and increase fluidity of the membrane [686].

Thus, first, we targeted cholesterol that can increase membrane order, therefore making the membrane less fluid, and organize lipid microdomains within the membrane [687, 688]. When N2A/tau were depleted from cholesterol with 1 mM methyl-β-cyclodextrin, tau secretion decreased by 47% ± 7% (III, Figure 2A). When we loaded these cells with cholesterol using preformed cholesterol:mβCD, this enhanced tau secretion by 76% ± 8% (III, Figure 2B). Additionally, we replicated these two experiments in mature rat primary cortical neurons (21 days in vitro) and obtained similar results for the secretion of endogenous tau (III, Figure 4A-B). When these cells were depleted from cholesterol, the amount of tau in the media decreased in a concentration-dependent manner, with the maximum effect at 1 mM mβCD (−48% ± 1%; III, Figure 4A), as measured by tau ELISA. Cholesterol loading, in turn,

66

roughly doubled the amount of tau in the media (+105% ± 16%; III, Figure 4B). These results suggest that cholesterol, via the organization of lipid microdomains, or membrane fluidity affects the secretion of tau in models of both pathological (overexpressed tau) and physiological (endogenous tau) conditions.

Second, we targeted sphingolipids, a group of membrane-ordering lipids that also play an essential role in the organization of lipid microdomains [688]. Thus, we depleted N2A/tau cells either nonspecifically from all sphingolipids by inhibiting sphingolipid synthesis with myriocin or specifically by depleting sphingomyelin at the extracellular leaflet of the plasma membrane by adding exogenous sphingomyelinase (SMase) to the media. Both treatments reduced tau secretion by approximately the same magnitude (50 µM myriocin: −36% ± 4%; 3 U/ml SMase: −36% ± 3%; III, Figure 2C-D), although the myriocin showed some toxicity at the highest concentration, as determined with an LDH assay.

Third, N2A/tau were loaded with docosahexaenoic acid (DHA), a n-3 polyunsaturated fatty acid (n-3 PUFA) that decreases the order of non-raft domains, remodels the lipidome of the plasma membrane and may promote the exclusion of the raft-resident proteins from the microdomains [689-691]. High consumption of DHA correlates epidemiologically with a lower risk of AD, and its oral intake in animal studies reduced Alzheimer-like brain pathology [692-696]. DHA loading reduced the secretion of tau in a concentration-dependent manner, with maximal inhibition at 50 μM DHA (−94% ± 1%; III, Figure 2E), although according to the LDH assay, this concentration of DHA was mildly toxic for N2A cells, probably due to DHA anticancer properties [697, 698]. Tau secretion, however, was reduced by 63% ± 3%

already at 1 µM, which is close to the concentration of DHA found in normal human CSF [699]. In conclusion, these results suggest that tau secretion proceeds more efficiently in the highly ordered plasma membrane enriched in cholesterol and sphingomyelin.

1.18.1 Heparan sulfate proteoglycans positively regulate tau secretion The unconventional secretion of FGF2 via translocation through the plasma membrane requires HSPGs, which capture FGF2 and assist its translocation to the extracellular side of the plasma membrane [532, 539, 540]. As tau binds GAGs in a sulfation-dependent manner, it seemed plausible that tau binding to HSPGs might play a role in tau secretion as well [349, 494, 553]. Thus, first, we inhibited the synthesis of the general sulfate donor in GAG biosynthesis (3′-phosphoadenosine 5′-phosphosulfate) in N2A/tau with NaClO3, which reduced tau secretion in a concentration-dependent manner, with the maximal decrease at 50 mM NaClO3

(−38% ± 5%; III, Figure 3A) [700, 701]. The treatment of primary cortical neurons with NaClO3 resulted in a similar decrease of tau secretion (−45% ± 5%; III, Figure 4C).

As NaClO3 treatment inhibits all sulfation reactions in the cell, next, we reduced the amount of binding sites of HSPG more specifically with heparinase I and heparinase III, enzymes that cleave sulfated glycan side chains of proteoglycans with different specificities [701, 702]. Although treatment with heparinase I seemed to decrease tau secretion, the effect was insignificant; treatment with heparinase III, on the other hand, consistently reduced tau secretion by 25% ± 3% (III, Figure 3B). Such

67

difference in the effects of the two heparinases likely originates from their different substrate specificity. Both enzymes can cleave heparin and heparan sulfates, but while heparinase III is more active towards ubiquitous heparan sulfate, heparinase I is more active towards mast cell-specific heparin sulfate [703, 704]. As only heparan sulfates are present in the cell surface proteoglycans in N2A cells, this may explain why only heparinase III treatment reduced tau secretion. To control for the specificity of these effects for HSPG, we reduced the cell surface levels of chondroitin and dermatan sulfate GAGs with a chondroitinase ABC treatment. This treatment, however, failed to decrease tau secretion (III, Figure 3C), suggesting a specific role of HSPGs in tau secretion, which is in line with a recent study showing similar results [526].

1.18.2 Tau secretion requires the formation of soluble oligomers

During secretion, FGF2 localizes to the inner leaflet of the plasma membrane, where it undergoes phosphorylation and forms a membrane-spanning oligomer [537, 705]. The majority of FGF2 species at the inner leaflet of the plasma membrane are dimers, but it is unclear if dimerization occurs before or after the association with the plasma membrane [705]. Thus, phosphorylation and oligomerization play a vital role in FGF2 secretion; both of these events are also involved in the tau pathology in AD and other tauopathies. Therefore, it was necessary to learn whether the phosphorylation and oligomerization of tau play a role in tau secretion. Indeed, as shown in Section 5.3.2, at least some tau localizes to the plasma membrane in a hyperphosphorylated oligomerized form. To further specify which tau species undergo secretion, N2A/tau received four tau aggregation inhibitors (TAIs), which inhibit tau oligomerization at different stages: (1) emodin and (2) BSc3094 interfere with the formation of β-sheet structure; (3) phthalocyanine tetrasulfonate (PcTS) stabilizes soluble off-pathway oligomers; and (4) epigallocatechin gallate (EGCG) prevents the formation of dimers, oligomers, and aggregates [706-712].

As three out of four TAIs are brightly colored, they are incompatible with a luminescence-based PCA; thus, we used a dot blot instead. With the dot blot, only the BSc3094 effect was insignificant in N2A/tau, while emodin, PcTS, and EGCG reduced tau secretion, with EGCG having the most potent effect (−94% ± 2%; III, Figure 3D-E), suggesting that tau secretion requires the formation of oligomeric aggregation intermediates. As EGCG is not a colored compound, we also tested it in the tau secretion PCA, where it reduced the secretion of tau oligomers by 63% ± 4%

and tau dimerization inside the cell by 32% ± 2% (III, Figure 3F). Furthermore, when naïve N2A acceptor cells were subjected to conditioned media produced by N2A/tau donor cells treated with EGCG, it resulted in a significantly lower amount of internalized tau in the acceptor cells compared to the acceptor cells subjected to the media from N2A/tau control donor cells (III, Figure 3G-I). Importantly, EGCG failed to change the uptake efficiency when added to the cell-free tau-conditioned media during the internalization stage (III, Figure 3J). Notably, in primary cortical neurons, EGCG also reduced tau secretion (maximum effect at 25 µM EGCG: −36% ± 7%; III, Figure 4D), suggesting that the secretion of endogenous tau in healthy neurons also depends on oligomerization.

68