Vitamin D and Its Analogues Decrease Amyloid-β (Aβ) Formation and Increase Aβ-Degradation

Alzheimer’s disease (AD) is characterized by extracellular plaques in the brain, mainly consisting of amyloid-β (Aβ), as derived from sequential cleavage of the amyloid precursor protein. Epidemiological studies suggest a tight link between hypovitaminosis of the secosteroid vitamin D and AD. Besides decreased vitamin D level in AD patients, an effect of vitamin D on Aβ-homeostasis is discussed. However, the exact underlying mechanisms remain to be elucidated and nothing is known about the potential effect of vitamin D analogues. Here we systematically investigate the effect of vitamin D and therapeutically used analogues (maxacalcitol, calcipotriol, alfacalcidol, paricalcitol, doxercalciferol) on AD-relevant mechanisms. D2 and D3 analogues decreased Aβ-production and increased Aβ-degradation in neuroblastoma cells or vitamin D deficient mouse brains. Effects were mediated by affecting the Aβ-producing enzymes BACE1 and γ-secretase. A reduced secretase activity was accompanied by a decreased BACE1 protein level and nicastrin expression, an essential component of the γ-secretase. Vitamin D and analogues decreased β-secretase activity, not only in mouse brains with mild vitamin D hypovitaminosis, but also in non-deficient mouse brains. Our results further strengthen the link between AD and vitamin D, suggesting that supplementation of vitamin D or vitamin D analogues might have beneficial effects in AD prevention.


Introduction
The most important pathological hallmarks of Alzheimer's disease (AD), which is a progressive neurodegenerative disorder, are extracellular plaques that are composed of aggregated Aβ peptides and intracellular neurofibrillary tangles, consisting of hyperphosphorylated Tau proteins [1][2][3][4]. Aβ-generation depends on initial cleavage of the amyloid precursor protein (APP) by β-secretase 1 (BACE1), followed by intramembrane cleavage of APP by γ-secretase, a heterotetrameric protein complex consisting of presenilin 1 or 2 (PS1, PS2), nicastrin, anterior-pharynx-defective 1a or 1b (Aph1a, Aph1b), and presenilin-enhancer 2 (PEN2) [5][6][7][8]. Besides amyloidogenic, Aβ-releasing processing of APP, APP can be shed by α-secretases in a non-amyloidogenic pathway [9][10][11][12]. The α-secretases cleave APP within the Aβ domain and preclude the formation of Aβ peptides. Total Aβ level is not only dependent on the proteolytic activities of the APP cleaving secretases, but also on Aβ-degradation, involving e.g., the Aβ degrading enzymes neprilysin (NEP) and insulin-degrading enzyme (IDE) [13,14]. In addition to several lipids that influence the generation, degradation, and aggregation of Aβ peptides [15][16][17][18][19][20][21][22][23][24][25], it has recently been shown that fat soluble vitamins affect molecular mechanisms that are involved in AD pathogenesis, e.g., Aβ-induced neurotoxicity, oxidative stress, inflammatory processes, as well as Aβ-generation, Aβ-degradation, and Aβ-clearance [26][27][28][29][30][31]. Several vitamins, including vitamin A, provitamin Aβ-carotene, vitamin D 3 , vitamin K, and vitamin E, have also been reported to be reduced in plasma/serum of AD patients [30,[32][33][34][35][36] and vitamin D hypovitaminosis affects up to 90% of the elderly population [37]. The biological activity of vitamin D can be attributed to binding interactions with the vitamin D receptor (VDR), which undergoes a conformational change, allowing for an interaction with the retinoid X receptor (RXR). The VDR-RXR heterodimer is considered to be the active complex that binds to vitamin D response elements in the DNA of target genes [38,39]. The VDR as well as the 1α-hydroxylase (CYP27B1), which converts 25(OH) vitamin D 3 (calcifediol) into its active form 1,25(OH) 2 vitamin D 3 (calcitriol), have been shown to be expressed in human brain [40], and vitamin D and vitamin D metabolites have been reported to cross the blood-brain-barrier [41]. Recently, we and others could show that vitamin D deficiency causes an increase in amyloidogenic β-secretase cleavage of APP and a decrease in Aβ-degradation, resulting in elevated Aβ level [29,42]. In line, 25(OH) vitamin D supplementation elevated Aβ-degradation due to increased NEP expression and activity [29], supporting vitamin D supplementation as a novel approach to treat AD. In the present study, we compare these effects as mediated by vitamin D with vitamin D analogues on their amyloidogenic potential by investigating the APP processing pathways, as well as Aβ-degradation. Experiments were performed in neuroblastoma cells, revealing a concentration of approximately 2.5 ng/mL 25(OH) vitamin D 3 [29]. However, it has to be pointed out that, although neuroblastoma cell lines have some neuronal properties, substantial differences when compared to neurons exist, which is a caveat of the study. Therefore, the main results were also validated ex vivo in homogenates of wildtype (wt) and hypovitaminosis D mouse brains. Further studies are needed to verify the potential positive effects of vitamin D and its analogues in vivo.
We selected 25(OH) vitamin D 3 (calcifediol), which is converted by CYP27B1 to 1,25-dihydroxyvitamin D 3 , the natural vitamin D hormone, and therapeutically used analogues of vitamin D 3 and D 2 (Figure 1). Vitamin D analogues are modified in the side-chain portion of the molecule and exert a lower calcemic activity than natural vitamin D 3 , but retain many therapeutic properties of 1,25-dihydroxyvitamin D 3 . The 1,25-hydroxylated vitamin D 3 analogue, maxacalcitol, is used to treat renal patients that are affected by secondary hyperparathyroidism, while the 1,24-hydroxylated vitamin D 3 analogue calcipotriol for treatment of psoriasis and 1-hydroxylated alfacalcidol is administered to treat osteoporosis and secondary hyperparathyroidism. Paricalcitol and doxercalciferol were selected as vitamin D 2 analogues. Identical to the vitamin D 3 analogue maxacalcitol, paricalcitol contains a hydroxyl group at C 1 and C 25 , but a vitamin D 2 instead of a vitamin D 3 side chain. 1-hydroxylated doxercalciferol is the vitamin D 2 analogue that is comparable to the vitamin D 3 analogue alfacalcidol in regard of the hydroxylation status. Both vitamin D 2 analogues are used to treat elevated serum parathyroid hormone levels that are associated with secondary hyperparathyroidism [43,44].

Vitamin D Analogues Decrease Total Aβ Level
In order to analyze whether analogues of vitamin D3 and vitamin D2 have similar anti-amyloidogenic properties when compared to 25(OH) vitamin D3 [29], we examined total secreted Aβ level in the human neuroblastoma cell line SH-SY5Y stably transfected with human APP695, the major isoform that is found in neurons [45], in the presence of calcifediol (25(OH) vitamin D3), maxacalcitol, calcipotriol, and alfacalcidol (vitamin D3 analogues), as well as the vitamin D2 analogues paricalcitol and doxercalciferol ( Figure 1). We decided to use 25(OH) vitamin D3 instead of active 1,25(OH)2 vitamin D3 as calcifediol has a long serum half-life of approximately three weeks when compared to the short serum half-life (4-6 h) of 1,25(OH)2 vitamin D3 [46]. Cells were treated for 24 h in presence of vitamin D or analogues and secreted Aβ level were examined by analyzing the conditioned media of treated cells or control cells, incubated with the solvent EtOH. Vitamin D and its analogues were incubated in a final concentration of 100 nM. The concentration that was used corresponds to physiological serum concentrations of approximately 75 nmol/L [47,48], and is frequently used for cell culture experiments [26,49,50]. Furthermore, no significant change in cell viability was found using 100 nM vitamin D as compared to the solvent control (supplement Figure S1). Moreover, it has to be mentioned that calcifediol is used as a therapeutical supplement to treat vitamin D hypovitaminosis [51].

Vitamin D Analogues Decrease Total Aβ Level
In order to analyze whether analogues of vitamin D 3 and vitamin D 2 have similar anti-amyloidogenic properties when compared to 25(OH) vitamin D 3 [29], we examined total secreted Aβ level in the human neuroblastoma cell line SH-SY5Y stably transfected with human APP695, the major isoform that is found in neurons [45], in the presence of calcifediol (25(OH) vitamin D 3 ), maxacalcitol, calcipotriol, and alfacalcidol (vitamin D 3 analogues), as well as the vitamin D 2 analogues paricalcitol and doxercalciferol ( Figure 1). We decided to use 25(OH) vitamin D 3 instead of active 1,25(OH) 2 vitamin D 3 as calcifediol has a long serum half-life of approximately three weeks when compared to the short serum half-life (4-6 h) of 1,25(OH) 2 vitamin D 3 [46]. Cells were treated for 24 h in presence of vitamin D or analogues and secreted Aβ level were examined by analyzing the conditioned media of treated cells or control cells, incubated with the solvent EtOH. Vitamin D and its analogues were incubated in a final concentration of 100 nM. The concentration that was used corresponds to physiological serum concentrations of approximately 75 nmol/L [47,48], and is frequently used for cell culture experiments [26,49,50]. Furthermore, no significant change in cell viability was found using 100 nM vitamin D as compared to the solvent control (supplement Figure S1). Moreover, it has to be mentioned that calcifediol is used as a therapeutical supplement to treat vitamin D hypovitaminosis [51].
To further examine the potential of vitamin D analogues to decrease Aβ level by reducing β-secretase activity in an ex vivo situation, we analyzed vitamin D deficient mouse brains showing a 23% reduction in the 25(OH) vitamin D level as compared to wt mouse brains [29], either supplemented with vitamin D or unsupplemented. Therefore, we used homogenates of vitamin D deficient and wt mouse brains, and measured β-secretase activity in the presence or absence of vitamin D analogues. As already shown in our previous study [29] we found a significant increase in β-secretase activity in vitamin D deficient mouse brains when compared to wt mouse brains ( Figure 2D). Supplementing of vitamin D deficient mouse brains with vitamin D 3 or vitamin D 3 and D 2 analogues revealed the following results: five out of six analyzed vitamins showed a decrease in β-secretase activity, resulting in a partial rescue of β-secretase activity obtained for unsupplemented wt mouse brains. However, only calcifediol and doxercalciferol reached a level of significance (calcifediol, p = 0.026; doxercalciferol: p = 0.040; Table 1). Four out of six vitamins/analogues tended to decrease β-secretase activity when supplemented on wt mouse brains; three vitamin D species, calcifediol, alfacalcidol and maxacalcitol, revealed a significant reduction in β-secretase activity (calcifediol, p = 0.028; alfacalcidol, p = 0.050; maxacalcitol, p = 0.015). Analyzing the effect of all of the analogues averaged on either wt mouse brains or vitamin D deficient mouse brains, we found a similar decrease in β-secretase activity ( Figure 2D), suggesting that both individuals with hypovitaminosis and individuals with normal vitamin D status, might profit from vitamin D supplementation in a similar range, which should be evaluated in further studies. A more detailed statistical analysis is given in Tables 1 and 2. Table 1. Comparison between β-secretase activity in supplemented (WT+; 100 nM calcifediol or its analogues) and unsupplemented (WT-) wildtype mouse brains and in supplemented (deficient+; 100 nM calcifediol or its analogues) and unsupplemented (deficient-) vitamin D deficient mouse brains.  As described above, we observed a more pronounced reduction in β-secretase activity in living cells in presence of vitamin D analogues when compared to the direct effect of these substances on β-secretase activity. To examine whether vitamin D analogues decrease β-secretase activity by affecting the gene expression of BACE1, we performed qPCR analysis of BACE1 and determined BACE1 protein level by WB analysis in SH-SY5Y wt cells. Except for paricalcitol, all vitamin D 3 and vitamin D 2 analogues, as well as calcifediol, tended to decrease BACE1 gene expression ( Figure 2E).
Significant results were only obtained for calcifediol, calcipotriol, and alfacalcidol (calcifediol: 80.8 ± 2.9%, p = 0.003; calcipotriol: 75.4 ± 8.1%, p = 0.040; alfacalcidol: 70.3 ± 5.4%, p = 0.006). The reason why paricalcitol showed no effect on BACE1 gene expression remains speculative. Paricalcitol, a third generation analogue of vitamin D 2 , is an activator of VDR and is used for secondary hyperparathyroidism. When compared to calcitriol, different effects were observed e.g., a reduced stimulation of the intestinal calcium transport proteins or a reduced effect on calbindin expression [52]. Similar mechanisms might be responsible for an absent effect on BACE1 expression in the presence of paricalcitol. However, the underlying mechanisms remain to be elucidated.

Discussion
Besides the central role of vitamin D in calcium and phosphate homeostasis, vitamin D is discussed to have neuroprotective and anti-inflammatory properties [55,56] and is important for brain development influencing proliferation, differentiation, neurite outgrowth, and neuronal density [57,58]. Several studies also observed an association between cognitive impairment and vitamin D hypovitaminosis [59][60][61][62][63][64]. Moreover, Sutherland et al. reported a link between vitamin D and AD, describing a reduction in VDR mRNA levels in hippocampal CA1 and CA2 pyramidal cells of AD patients when compared to patients that were suffering from Huntington disease [65]. Studies during the past years suggest that lower vitamin D concentrations are associated with a substantially increased risk of all-cause dementia and AD [66][67][68]. Recently, Mokry et al. identified four single nucleotide polymorphisms (SNPs) in the vitamin D pathway that are significantly linked to AD [69]. In our previous study, analyzing mild vitamin D deficiency, we found increased Aβ peptide level caused by increased β-secretase cleavage of APP and decreased Aβ-degradation in vitamin D deficient mouse brains [29]. The aim of the present study was to investigate whether therapeutically used analogues of vitamin D3 and vitamin D2 also shows anti-amyloidogenic properties, and whether differences exist between single vitamin D analogues. In agreement with increased Aβ

Discussion
Besides the central role of vitamin D in calcium and phosphate homeostasis, vitamin D is discussed to have neuroprotective and anti-inflammatory properties [55,56] and is important for brain development influencing proliferation, differentiation, neurite outgrowth, and neuronal density [57,58]. Several studies also observed an association between cognitive impairment and vitamin D hypovitaminosis [59][60][61][62][63][64]. Moreover, Sutherland et al. reported a link between vitamin D and AD, describing a reduction in VDR mRNA levels in hippocampal CA1 and CA2 pyramidal cells of AD patients when compared to patients that were suffering from Huntington disease [65]. Studies during the past years suggest that lower vitamin D concentrations are associated with a substantially increased risk of all-cause dementia and AD [66][67][68]. Recently, Mokry et al. identified four single nucleotide polymorphisms (SNPs) in the vitamin D pathway that are significantly linked to AD [69].
In our previous study, analyzing mild vitamin D deficiency, we found increased Aβ peptide level caused by increased β-secretase cleavage of APP and decreased Aβ-degradation in vitamin D deficient mouse brains [29]. The aim of the present study was to investigate whether therapeutically used analogues of vitamin D 3 and vitamin D 2 also shows anti-amyloidogenic properties, and whether differences exist between single vitamin D analogues. In agreement with increased Aβ peptide level in the case of mild 25(OH) vitamin D 3 deficiency, calcifediol, which is converted by 1α-hydroxylase CYP27B1 to active 1,25(OH) 2 vitamin D 3 , calcitriol, decreased total Aβ level to 55.1% in human neuroblastoma cells. Also, the vitamin D 3 analogues maxacalcitol, calcipotriol and alfacalcidol and the vitamin D 2 analogues paricalcitol and doxercalciferol significantly reduced Aβ level. No significant differences were obtained between single vitamin D analogues. In line with our results, a similar reduction of approximately 50% in Aβ 40 and Aβ 42 peptides and a decrease in the number of amyloid plaques has also been shown in APP transgenic mice that were fed for five months with a vitamin D enriched diet [70]. Beside the potential of vitamin D to reduce Aβ level it has been recently shown that vitamin D 3 protects against Aβ peptide cytotoxicity by reverting the Aβ 1-42 induced reduction in the sphingosine-1-phosphate/ceramide ratio [26]. We found that vitamin D 3 and analogues of vitamin D 3 and D 2 reduce secreted Aβ peptide level by decreasing amyloidogenic APP processing and by an elevation of Aβ-degradation. Impaired amyloidogenic APP processing in presence of vitamin D and vitamin D analogues is caused by a slight direct effect of these vitamins on β-secretase activity, and by a decrease in BACE1 gene expression, accompanied by reduced BACE1 protein level. Besides transcriptional effects, it cannot be excluded that altered BACE1 protein stability in presence of vitamin D or vitamin D analogues might also contribute to the reduction in BACE1 protein level. BACE1 can be degraded by the lysosomal and ubiquitin-proteasome system (UPS) [71,72]. Interestingly, 1,25-dihydroxyvitamin D 3 is involved in the regulation of numerous UPS genes, including ubiquitinating and deubiquitinating enzymes [73]. In this respect, is has to be mentioned that BACE1 degradation is dependent on ubiquitin carboxyl-terminal hydrolase L1, a deubiquitinating enzyme highly specific to neurons [74]. Furthermore, it could be recently shown that vitamin D deficiency down-regulates genes that are involved in protein catabolism [75]. However, so far it is unclear whether these mechanisms are also affected by vitamin D analogues, which has to be investigated in further studies.
In analogy to our observed effect on BACE1 protein level, Briones et al. found a 24% reduction in BACE1 protein level in old rats supplemented with vitamin D 3 as compared to control animals [76]. In accordance to decreased BACE1 protein level in presence of vitamin D, silencing VDR in E16 primary rat cortical neurons increased mRNA and protein level of BACE1 [42], resulting in an increased intracellular Aβ 1-42 level. The vitamin D induced reduction in β-secretase APP cleavage could be further substantiated ex vivo by supplementing vitamin D deficient and wt mouse brain homogenates with vitamin D 3 or analogues of vitamin D 3 and vitamin D 2 , indicating that both, patients with vitamin D hypovitaminosis and patients with a normal vitamin D status, might profit from vitamin D supplementation. Beside the effect of vitamin D and its analogues on β-secretase, we found significantly reduced γ-secretase activity in metabolically active cells in the presence of vitamin D or vitamin D analogues. In accordance to reduced γ-secretase activity, we found significantly reduced mRNA levels of nicastrin, indicating that altered gene expression of nicastrin, which is necessary for the maturation of the γ-secretase complex [77,78], contributes to the observed reduction in γ-secretase activity. Notably, the mRNA level of nicastrin have been shown to be increased in VDR silenced primary rat cortical neurons [42], thus supporting our results. In addition to the important role of βand γ-secretase processing of APP in Aβ anabolism, Aβ level can be also decreased by elevated non-amyloidogenic α-secretase processing of APP. However, the effect of vitamin D and its analogues on α-secretase shedding is not as congruent as found for amyloidogenic β-secretase cleavage. In our study, calcifediol and paricalcitol were the only vitamin D species elevating non-amyloidogenic APP processing, whereas the other vitamin D analogues showed no significant effect. Gezen-Ak et al. even found an increase in mRNA and protein level of α-secretase ADAM10 in VDR siRNA-treated cortical neurons [42]. However, as additional metalloproteinases of the ADAM family have been identified as α-secretases [9,12,79], further studies in vitamin D treated, vitamin D deficient or VDR deficient cells are necessary to clarify the role of vitamin D and vitamin D analogues in α-secretase APP processing. Beside Aβ anabolism impaired Aβ-degradation is discussed to contribute to sporadic AD. Vitamin D 3 , as well as all analyzed vitamin D 3 and vitamin D 2 analogues, significantly elevated Aβ-degradation in mouse neuroblastoma cells. Increased Aβ-degradation in presence of vitamin D 3 or its analogues was also obtained in vitamin D deficient mouse brains. Notably, we observed an increased gene expression of NEP and elevated NEP activity in the presence of vitamin D and its analogues, indicating that the Aβ-degrading enzyme NEP is affected by vitamin D and vitamin D 3 and D 2 analogues. These results are in line with the finding of Briones et al. reporting an increase in the protein level of NEP in vitamin D supplemented old rats [76]. Based on these findings, vitamin D and its analogues reduce total Aβ level by pleiotropic mechanisms affecting Aβ anabolism and Aβ catabolism ( Figure 6). Each of the individual effects of the vitamin D analogues seem to be rather small and in some cases a significant level is not reached. However, all of the observed mechanisms result in a reduction of total Aβ level, which is highly significant and much more pronounced. in the presence of vitamin D and its analogues, indicating that the Aβ-degrading enzyme NEP is affected by vitamin D and vitamin D3 and D2 analogues. These results are in line with the finding of Briones et al. reporting an increase in the protein level of NEP in vitamin D supplemented old rats [76]. Based on these findings, vitamin D and its analogues reduce total Aβ level by pleiotropic mechanisms affecting Aβ anabolism and Aβ catabolism ( Figure 6). Each of the individual effects of the vitamin D analogues seem to be rather small and in some cases a significant level is not reached. However, all of the observed mechanisms result in a reduction of total Aβ level, which is highly significant and much more pronounced. In summary, β-secretase cleavage is affected by a slight direct inhibitory effect on enzyme activity and a decrease in BACE1 expression and BACE1 protein level. γ-secretase activity is reduced by decreased gene expression of the γ-secretase component nicastrin. Furthermore, vitamin D and its analogues increase Aβ-degradation, substantiating the important role of vitamin D and its analogues in Aβ homeostasis, and that supplementation with vitamin D3 or analogues of vitamin D3 and D2 might be protective against biological processes that are associated with AD. Notably, we found a correlation between β-secretase activity, the rate-limiting step in Aβ-generation, Aβ-degradation, and total Aβ level in presence of vitamin D or its analogues, as determined by Pearson correlation (r = 0.699, p = 0.081). In line, BACE1 protein level and Aβ-degradation correlates with the Aβ level (r = 0.746, p = 0.054). However, the Pearson correlation did not reach significance. Regarding vitamin D supplementation to prevent or treat AD, it is important to note that vitamin D3 also interferes with inflammatory processes that are known to contribute to AD. A significant increase in cytokine IL-1β level has been found in post mortem samples from AD patients with a Figure 6. Model of the pleiotropic effects of vitamin D and analogues on Aβ-homeostasis. Vitamin D and analogues decrease amyloidogenic amyloid precursor protein (APP) processing by affecting βand γ-secretase activity. The reduction of β-secretase activity is caused by a direct effect of vitamin D and its analogues on β-secretase activity combined with indirect effects on BACE1 gene expression and total BACE1 protein level. The γ-secretase activity is reduced by decreased gene expression of nicastrin responsible for the maturation of the heterotetrameric γ-secretase complex. A stimulation of the non-amyloidogenic α-secretase processing of APP was found for 25(OH) vitamin D 3 and the vitamin D 2 analogue paricalcitol. Total Aβ level in presence of vitamin D and analogues are further reduced by increased Aβ-degradation.
In summary, β-secretase cleavage is affected by a slight direct inhibitory effect on enzyme activity and a decrease in BACE1 expression and BACE1 protein level. γ-secretase activity is reduced by decreased gene expression of the γ-secretase component nicastrin. Furthermore, vitamin D and its analogues increase Aβ-degradation, substantiating the important role of vitamin D and its analogues in Aβ homeostasis, and that supplementation with vitamin D 3 or analogues of vitamin D 3 and D 2 might be protective against biological processes that are associated with AD. Notably, we found a correlation between β-secretase activity, the rate-limiting step in Aβ-generation, Aβ-degradation, and total Aβ level in presence of vitamin D or its analogues, as determined by Pearson correlation (r = 0.699, p = 0.081). In line, BACE1 protein level and Aβ-degradation correlates with the Aβ level (r = 0.746, p = 0.054). However, the Pearson correlation did not reach significance. Regarding vitamin D supplementation to prevent or treat AD, it is important to note that vitamin D 3 also interferes with inflammatory processes that are known to contribute to AD. A significant increase in cytokine IL-1β level has been found in post mortem samples from AD patients with a maximum response in those brain regions, frontal cortex, and hippocampus, where AD neuropathology is most prominent [54]. Analyzing IL-1β, initiating inflammatory responses, we could show that vitamin D 3 and vitamin D 2 analogues, except paricalcitol, significantly reduced the pro-inflammatory IL-1β level. This finding is consistent with the recent findings by Raha et al. reporting that vitamin D 2 attenuates Aβ 25-35 induced pro-inflammatory cytokines, such as IL-1β, IL-6, and TNFα [27]. Furthermore, a recent study, describing the increased level of pro-inflammatory IL-1β and decreased level of anti-inflammatory IL-10 in old rats as compared to young animals, shows that this age-related change in inflammatory states was mitigated by vitamin D supplementation [76]. This effect seems to be not limited to pro-inflammatory cytokines, as it has recently been shown that 1,25-dihydroxyvitamin D 3 upregulates IL-34 known to provide strong neuroprotective and survival signs in brain injury and neurodegeneration [80].
In conclusion, our results substantiate vitamin D supplementation as an approach to prevent or treat AD by reducing Aβ anabolism, elevating Aβ catabolism and the reduction of pro-inflammatory cytokines. The analyzed vitamin D analogues reduce secreted Aβ level with a similar potency, but differ partially in the effect strength of the underlying mechanisms, illustrating that individual AD patients might profit in a different extend of vitamin D analogues. AD patients with reduced anti-amyloidogenic β-secretase activity might benefit the most from supplementation with species like calcifediol and paricalcitol, as these vitamin D species were the only vitamins increasing non-amyloidogenic APP processing. According to our results, individuals with impaired Aβ-degradation might have the highest benefits from vitamin D analogues, like calcipotriol and maxacalcitol, showing the strongest effect on Aβ-degradation. In respect to βand γ-secretase processing all of the vitamin D analogues revealed similar results, indicating that individuals with increased amyloidogenic secretase activities might benefit from the vitamin D analogues in a similar way.
However, it has to be emphasized that the differences in the effect strength are rather small, and all of the analogues have been shown in respect to Aβ level to be similarly beneficial. Additionally, the use of vitamin D analogues in respect to AD seems to have no therapeutical advantage when compared to the use of calcifediol (or calcitriol), as no significant differences were found between vitamin D analogues and vitamin D. Therefore, further medical indications or pharmacological aspects of the vitamin D analogues should be taken into consideration as to which vitamin D analogue should be applied. Related to this, factors like plasma half-life of individual analogues, the ability to pass the blood-brain-barrier, the affinity for VDR as well as resorption, compatibleness, availability and potential side-effects have to be considered [43,44]. Especially pharmacological factors and the ability to pass the blood-brain-barrier might be even more relevant than the mechanistical differences of the individual analogues, which has to be further evaluated in in vivo experiments and clinical studies.

Chemicals and Reagents
Calcifediol, and its analogues, alfacalcidol, calcipotriol, doxercalciferol, maxacalcitol, and paricalcitol were purchased from MedChem Express (Monmouth Junction, NJ, USA), and all other chemicals used in this study were acquired from Sigma-Aldrich (Taufkirchen, Germany), if not stated otherwise.
For the ex vivo experiments we used brains from female C57BL/6 wt mice (Charles River, Sulzfeld, Germany) and vitamin D deficient mice. All animal experiments were approved by the "Landesamt für Soziales, Gesundheit und Verbraucherschutz of the State of Saarland" (reference number 17/2011) following the national guidelines for animal treatment. To create the vitamin D deficit, C57BL/6 mice were fed as described, resulting in a 23% reduction in the 25(OH) vitamin D level [29]. After removing, the brains were washed in 0.9% sodium chloride and directly frozen in liquid nitrogen. To establish the homogenates, the mouse brains were slowly defrosted on ice, and afterwards were treated by Minilys (Peqlab, Erlangen, Germany) in HPLC-grade H 2 O.

Cell Culture
To minimize the influence of 25(OH) vitamin D 3 from serum, FCS in DMEM was reduced to 0.1%-2.5% 16 h before incubation, dependent on subsequent experiments. Incubation with 100 nM vitamin D or its analogues (dissolved in ethanol) was carried out for 24 h (8 + 16 h) in up to 2.5% FCS/DMEM. Controls were treated with ethanol in a final concentration of 1‰, corresponding to the concentration in the incubation media.

Mouse Brains or Purified Membranes
Equal amounts of mouse brain homogenates or postnuclear fractions, adjusted using bicinchoninic acid assay, were incubated with 100 nM of vitamin D, its analogues or ethanol for 15 min at 4 • C.

Determination of Protein Concentration
The determination of the protein concentration of the samples was performed by using the bicinchoninic acid assay (BCA), as described in [81].

Western Blot Experiments
Samples used for the WB experiments were adjusted to equal protein concentration in advance. For the determination of BACE1, cell lysates were prepared by lysing cells in 150 mM NaCl, 50 mM Tris/HCl pH 7.4, 2 mM EDTA, 0.1% NP-40, 0.1% Triton-X 100. For the determination of total secreted Aβ level and sAPPβ conditioned media were used.
Aβ levels were detected by performing immunoprecipitation of conditioned media before WB analysis. Therefore, 20 µL protein G-Sepharose and W02 antibody (5 µg/mL) were used.

Determination of β-Secretase Activity in Isolated SH-SY5Y Membranes
Measurements have been done as published in [23]. After homogenization of SH-SY5Y cells in sucrose buffer with Minilys (Peqlab, Erlangen, Germany) using ceramic beads, homogenates were adjusted to an equal protein amount by bicinchoninic acid assay and postnuclear fractions (PNFs) were isolated by sucrose density centrifugation. PNFs were incubated with calcifediol or its analogues for 15 min at 4 • C, and for pelleting membranes ultracentrifuged at 55,000 rpm for 75 min at 4 • C. Following, membranes were resuspended using glass beads in Minilys. For determination of β-secretase activity 20 µM specific β-secretase substrate (described above) was added to 125 µg of protein diluted 1:1 with 1 × PBS pH 4.5. Fluorescence was measured as described before.

Enzyme-Linked Immunosorbent Assay (ELISA)
The level of cytokine IL-1β was measured in the medium of incubated SH-SY5Y wt cells using the Human IL-1 beta ELISA Kit (Abcam, Cambridge, UK).

Lactate Dehydrogenase (LDH) Activity Assay
The Cytotoxicity Detection Kit (LDH) from Roche (Basel, Schweiz), a colorimetric assay for the measurement of lactate dehydrogenase (LDH) release from cells, was used according manufacturer's instructions to quantify cell death and lysis after incubation with different vitamin D concentrations.

Data Analysis
All quantified data represent an average of at least three independent experiments. Error bars represent standard error of the mean. Statistical significance was calculated using two-tailed Student's t test, ANOVA, and post hoc Tests for multiple comparison analysis. The normality of the data distribution was tested with Shapiro Wilk test. Significance was set at * p ≤ 0.05; ** p ≤ 0.01 and *** p ≤ 0.001.