Following simvastatin treatment, most of the observed changes in secretory profile were reversed by directly supplementing microglia with cholesterol. The increased release of both TNFα and BDNF from LPS-activated, simvastatin treated microglia was reversed by treatment with hpβcd-cholesterol, returning mean cellular cholesterol to levels at or above that of controls (Figure
1A, B). TNFα release from simvastatin-treated microglia was also attenuated by treatment with MEV, which is expected as mevalonic acid is the immediate product of HMG-CoA reductase and acts as the precursor to both sterol and isoprenoid biosynthesis. These changes strongly suggest that simvastatin affects microglial secretion through a cholesterol-dependent mechanism, rather than through a mechanism mediated by isoprenoids. Two prior studies offer conflicting interpretations regarding this mechanism: in primate microglial cultures TNFα release was found to be cholesterol-dependent [
42]; in rat hippocampal slices TNFα levels reverted to control after MEV treatment or treatment with geranylgeranyl pyrophosphate, the first committed intermediate in isoprenoid biosynthesis [
43], though a role for cholesterol was not directly tested, the
ex vivo system precludes resolution of cell-specific effects. In a comparable study on peripheral immune cells, an observed increase in TNFα release from simvastatin treated human blood derived macrophages was reversed by MEV treatment, but as in the aforementioned study, a direct role of cholesterol was not tested [
24]. As release of NO from LPS-activated microglia was unaffected by any of the treatments used in this study (Figure
1C and further data not shown), we can speculate that neither simvastatin treatment nor cholesterol-depletion with mβcd affect activation of microglia through Toll-like receptor 4 (TLR4) mediated activation of NFκB, as such a mechanism would be expected to affect the expression of inducible nitric oxide synthetase (iNOS).
To our knowledge, this is the first report to demonstrate changes in Il1β secretion from primary microglia following direct manipulation of cholesterol, though a comparable increase in Il1β secretion was observed after mβcd treatment of the immortalized microglial BV-2 cell line [
48]. In our assays an inverse correlation was observed between directly manipulated cholesterol levels and Il1β release: decreasing cholesterol resulted in increased Il1β release, while increasing cholesterol markedly decreased Il1β release even after stimulation with the potent TLR4 agonist, LPS (Figures
1D and
4A). Curiously, while decreasing cholesterol through statin treatment resulted in lower Il1β levels (Figure
1D), directly decreasing cellular cholesterol content through mβcd treatment resulted in an increase in Il1β with or without LPS-activation (Figure
4A). This discrepancy may be explained by one of two models. First, Il1β release may be affected by two distinct mechanisms – one sensitive to cholesterol in the plasma membrane, and one sensitive to intracellular cholesterol. Simvastatin acts on HMG-CoA reductase, an enzyme localized primarily to the endoplasmic reticulum [
49], and so reduces cholesterol from the ‘inside-out’ , likely resulting in changes in cholesterol in intracellular compartments followed by later changes in plasma membrane cholesterol. Conversely, mβcd acts directly on the extracellular leaflet of the plasma membrane (PM), reducing cholesterol from the ‘outside-in’ , rapidly affecting PM cholesterol, which subsequently equilibrates across membrane leaflets rapidly, but is then transported into intracellular compartments by relatively slow trafficking through vesicles or carrier proteins. A second interpretation of these data would require a distinct isoprenoid-dependent mechanism affecting Il1β release from statin-treated microglia. In either scenario the observed decreased release of Il1β from simvastatin-treated, LPS-activated microglia after hpβcd-cholesterol treatment may represent the PM cholesterol-dependent mechanism masking the contribution of subtler cholesterol- or isoprenoid-dependent mechanisms. The sensitivity to PM cholesterol is consistent with observations that, unlike TNFα and BDNF, Il1β secretion is dependent on the non-canonical pathway of cytokine secretion. TNFα and BDNF are trafficked in conventional ER- and Golgi-derived vesicles (as a membrane bound pro-cytokine in the case of TNFα and in dense-core vesicles for BDNF) while Il1β is synthesized in the cytoplasm and is secreted either though budding of microvesicles from the PM or possible transport through specific PM resident transporters (eloquently reviewed in [
50]). This non-canonical secretory pathway is likely to be differentially sensitive to the lipid composition of the PM than conventional Golgi trafficking, as cholesterol affects membrane fluidity and curvature (as required for vesicular budding) [
51]–[
53] as well as regulating protein activity through lipid microdomains (as may affect membrane resident transporters) [
54]–[
56]. As such, we propose that the observed sensitivity of Il1β release to cholesterol, in particular the marked decrease in Il1β release from LPS-activated microglia treated with hpβcd-cholesterol (Figure
1D,
4A), may result from a blockade of the late release steps of Il1β. However, as a detailed mechanistic analysis of Il1β secretion was not the objective of this study further analysis will be explored in future works. The increase in basal secretion in the simvastatin control groups can be attributed to the use of an ethanolic vehicle for delivery of simvastatin. As ethanol is itself a TLR4 agonist [
57],[
58] even at the very low ethanol concentrations used (<0.005%), the vehicle resulted in slightly higher Il1β secretion as compared to an untreated control (compare Figures
1D,
2,
3 and Figure
4A). We do not consider this to be a significant confound as the vehicle effect was slight, relative to potent TLR4 activation by LPS, and was not found to affect any of the other measures reported (TNFα, BDNF, NO, viability, or phagocytosis).