Synthesis and Characterization of Andrographolide Derivatives as Regulators of βAPP Processing in Human Cells

Alzheimer’s disease (AD) is a devastating neurodegenerative disorder, one of the main characteristics of which is the abnormal accumulation of amyloid peptide (Aβ) in the brain. Whereas β-secretase supports Aβ formation along the amyloidogenic processing of the β-amyloid precursor protein (βAPP), α-secretase counterbalances this pathway by both preventing Aβ production and triggering the release of the neuroprotective sAPPα metabolite. Therefore, stimulating α-secretase and/or inhibiting β-secretase can be considered a promising anti-AD therapeutic track. In this context, we tested andrographolide, a labdane diterpene derived from the plant Andrographis paniculata, as well as 24 synthesized derivatives, for their ability to induce sAPPα production in cultured SH-SY5Y human neuroblastoma cells. Following several rounds of screening, we identified three hits that were subjected to full characterization. Interestingly, andrographolide (8,17-olefinic) and its close derivative 14α-(5′,7′-dichloro-8′-quinolyloxy)-3,19-acetonylidene (compound 9) behave as moderate α-secretase activators, while 14α-(2′-methyl-5′,7′-dichloro-8′-quinolyloxy)-8,9-olefinic compounds 31 (3,19-acetonylidene) and 37 (3,19-diol), whose two structures are quite similar although distant from that of andrographolide and 9, stand as β-secretase inhibitors. Importantly, these results were confirmed in human HEK293 cells and these compounds do not trigger toxicity in either cell line. Altogether, these findings may represent an encouraging starting point for the future development of andrographolide-based compounds aimed at both activating α-secretase and inhibiting β-secretase that could prove useful in our quest for the therapeutic treatment of AD.


Introduction
Alzheimer's disease (AD) is the most prevalent neurodegenerative disorder and the main cause of dementia worldwide. It is characterized by a progressive loss of memory and cognitive function, which ultimately lead to dementia and death. Pathologically, there is an accumulation of extracellular β-amyloid peptide (Aβ) in senile plaques and of intracellular hyper-phosphorylated tau-containing neurofibrillary tangles (NTFs) in the hippocampus and cerebral cortex. This is accompanied by a large panel of molecular events related to the progression of the disease including oxidative stress, neuroinflammation, mitochondrial dysfunction, altered calcium homeostasis, and apoptosis [1].
The Aβ peptides are produced from the β-amyloid precursor protein (βAPP) along the amyloidogenic pathway through the sequential cleavages by β-secretase (BACE1) and the heterotetrameric γ-secretase complex that also gives rise to the production of the sAPPβ, The Aβ peptides are produced from the β-amyloid precursor protein (βAPP) the amyloidogenic pathway through the sequential cleavages by β-secretase (BACE1 the heterotetrameric γ-secretase complex that also gives rise to the production o sAPPβ, C99, and AICD metabolites [2]. At the same time, a major alternative non-am dogenic route involves α-secretase activity. It not only hampers the production of th yloid peptide since cleavage occurs in the middle of the Aβ sequence, but it also lea the secretion of the metabolite sAPPα with neuroprotective and neurotrophic powe Therefore, α-secretase activation appears as a promising therapeutic strategy aim preventing AD [4,5]. Unfortunately, sustained efforts to inhibit and/or modulate β γ-secretases or to activate α-secretases are still unsuccessful, this being mainly expl by the fact that these enzymes cleave numerous other substrates involved in vital ph logical functions [6][7][8]. Hence, the use of plant-derived active compounds has increas been considered in recent years as an alternative to pharmacotherapies [9] because effects for most medicinally used natural products are considerably low or inexisten Andrographolide (Scheme 1) is a bicyclic diterpenoid lactone present in the stem leaves of Andrographis paniculata. As the major bioactive constituent of this tradit Asian medicinal herb, it supports its antioxidant, anti-microbial, anti-inflammatory anti-cancer properties [10]. Interestingly, several beneficial effects of andrographoli physiological functions of the central nervous system have been reported in recent as shown by the ability of the compound to stimulate adult neurogenesis in the m hippocampus [11] and the capacity of andrographolide analogues to promote neurit growth in rat PC-12 cells [12]. Furthermore, several studies have very recently evide some positive effects of andrographolide on AD pathology. Firstly, it can strongly a uate Aβ-induced microglial activation [13,14] and autophagy-associated cell death [ vitro. Secondly and most importantly, andrographolide administration has been sh to alleviate AD-associated phenotypes, including cognitive deficits, observed in transgenic [16][17][18] and non-transgenic [19,20] models of the disease. 14α, R 1 = Me, R 2 = H, 13 [33] from 7 14β, R 1 = Me, R 2 = H, 14 [33] from 8 14α, R 1 = H, R 2 = Cl, 15 [33] from 9 14β, R 1 = H, R 2 = Cl, 16 [33] from 10 14α, R 1 = Me, R 2 = Cl, 17 from 11 14β, R 1 = Me, R 2 = Cl, 18 from 12 14α, R 1 = Me, R 2 = H, 7 [32] from 3 and 4 14β, R 1 = Me, R 2 = H, 8 [32] from 2 and 4 14α, R 1 = H, R 2 = Cl, 9 [32] from 3 and 5 14β, R 1 = H, R 2 = Cl, 10 [32] from 2 and 5 14α, R 1 = Me, R 2 = Cl, 11 from 3 and 6 14β, R 1 = Me, R 2 = Cl, 12 [33] [34] (a) [32][33][34] (b) [32][33][34] Mechanistically, it has been shown that andrographolide activates the canonica signaling pathway via an inhibition of GSK-3β in primary neurons [21] and in the rodent Octodon degus [22], and that it increases glucose uptake and utilization in a dependent manner in the J20 transgenic mouse model of AD [23]. Interestingly, the ists a close relationship between Wnt loss of function and AD-associated neurodege tion [24] mostly because of the ability of Wnt to repress the transcription of the β-secr BACE1 [25]. Indeed, it has been established that andrographolide can shift the metab of βAPP towards the non-amyloidogenic pathway as shown by an increased produ of C83 and a concomitant decrease in C99 and Aβ42/Aβ40 ratio in epithelial cells ov pressing βAPP [26].
Mechanistically, it has been shown that andrographolide activates the canonical Wnt signaling pathway via an inhibition of GSK-3β in primary neurons [21] and in the aged rodent Octodon degus [22], and that it increases glucose uptake and utilization in a Wntdependent manner in the J20 transgenic mouse model of AD [23]. Interestingly, there exists a close relationship between Wnt loss of function and AD-associated neurodegeneration [24] mostly because of the ability of Wnt to repress the transcription of the β-secretase BACE1 [25]. Indeed, it has been established that andrographolide can shift the metabolism of βAPP towards the non-amyloidogenic pathway as shown by an increased production of C83 and a concomitant decrease in C99 and Aβ42/Aβ40 ratio in epithelial cells overexpressing βAPP [26].
Because its simplistic structural nature brings amenability for semi-synthetic modifications, andrographolide has given rise to many derivatives with potent therapeutic effects in diverse fields [27]. In this context, we have undertaken to design, synthesize, and test twenty-four andrographolide derivatives, classified as two series, for their ability to favorably modulate βAPP processing through the stimulation of α-secretase and/or the inhibition of β-secretase catalytic activities/expression in human neuroblastoma SH-SY5Y cells. A first screening allowed us to identify andrographolide as well as the three derivatives 9, 31, and 37 as potent sAPPα production-enhancing compounds without altering cell viability. Their subsequent full characterization indicated that while andrographolide and 9 could moderately stimulate the α-secretase catalytic activity, 31 and 37 behaved as potent β-secretase inhibitors. These results thus established andrographolide and some of its derivatives as a promising basis for the future development of anti-amyloidogenic factors, the next generation of which hopefully leading to the setup of druggable α-secretase activator/β-secretase inhibitor compounds.
Because its simplistic structural nature brings amenability for semi-synthetic modifications, andrographolide has given rise to many derivatives with potent therapeutic effects in diverse fields [27]. In this context, we have undertaken to design, synthesize, and test twenty-four andrographolide derivatives, classified as two series, for their ability to favorably modulate βAPP processing through the stimulation of α-secretase and/or the inhibition of β-secretase catalytic activities/expression in human neuroblastoma SH-SY5Y cells. A first screening allowed us to identify andrographolide as well as the three derivatives 9, 31, and 37 as potent sAPPα production-enhancing compounds without altering cell viability. Their subsequent full characterization indicated that while andrographolide and 9 could moderately stimulate the α-secretase catalytic activity, 31 and 37 behaved as potent β-secretase inhibitors. These results thus established andrographolide and some of its derivatives as a promising basis for the future development of anti-amyloidogenic factors, the next generation of which hopefully leading to the setup of druggable α-secretase activator/β-secretase inhibitor compounds.

Screening of Andrographolide Derivatives for sAPPα Production and sAPPα/βAPP Ratio
We then investigated the effect of andrographolide and its 24 derivatives for their ability to promote the secretion of the βAPP-derived sAPPα metabolite in cultured naive SH-SY5Y human neuroblastoma cells. As a first step, we chose to treat the cells for 24 h with 1 µM of the compounds (andrographolide, the 12 derivatives of the 8,17-olefinic series (8,17-double bond) and the 12 derivatives of the 9-dehydro-17-hydro series (8,9-olefen/double bond)) and we used both sAPPα production and the sAPPα/βAPP/β-actin ratio as a read out for comparison with controls (duplicate).
Overall, these results suggest that an optimal combination of 8,9-double bond or 8,17double bond, substitution at quinolone, 14α or 14β, and 3,19-free diol or protection will benefit the enhancement of sAPPα production. As a whole, based on the screening results and structure-activity consideration, we undertook to focus on 9, 28, 31, 34, and 37 for further characterization, using andrographolide as the reference compound.

Further Characterization of Andrographolide Derivatives 9, 28, 34, 31, and 37
The results obtained following a consistent number of independent experiments (n ≥ 6 when compared with n = 2 for the initial screening step) first showed that 31 and 37 (1 μM) significantly induce sAPPα secretion (Figure 2A,B). Secondly, none of the compounds significantly altered βAPP immunoreactivity (Figure 2A,C), thereby ruling out an effect on βAPP expression or maturation and rather suggesting that they genuinely control βAPP processing. The additional measurement of the sAPPα/βAPP/β-actin ratio further indicated that all the selected derivatives including andrographolide itself were able to significantly increase this ratio although to different degrees ( Figure 2D).  In light of these results, we decided to reduce our field of investigation to 9, 31, and 37. At this stage, it was important to demonstrate that these compounds are not inherently toxic. For this purpose, we measured the survival rate of SH-SY5Y cells following a 24 h treatment at concentrations ranging from 100 nM to 10 µM with the MTT assay. In fact, no In light of these results, we decided to reduce our field of investigation to 9, 31, and 37. At this stage, it was important to demonstrate that these compounds are not inherently toxic. For this purpose, we measured the survival rate of SH-SY5Y cells following a 24 h treatment at concentrations ranging from 100 nM to 10 μM with the MTT assay. In fact, no notable changes were observed whatever the compound and the concentrations considered (Figure 3), clearly indicating that none of them are toxic under our experimental conditions. Following the screening phase carried out at one single concentration (1 μM), we then examined the ability of the selected compounds to stimulate sAPPα at lower concentrations and in a dose-dependent manner in SH-SY5Y cells. Although an increasing trend was observed for both andrographolide and 9, we could not establish statistically significant differences with controls ( Figure 4A). Nevertheless, in addition to the fact that 31 and 37 were triggering a significant increase in sAPPα production at 1 μM ( Figure 4A) as previously observed (see Figure 2), we showed that 37 is also effective at 10 nM and 100 nM concentrations ( Figure 4A). Parallel analysis of βAPP protein levels showed no significant variation in βAPP immunoreactivity under any conditions ( Figure 4B). The concomitant measurement of the sAPPα/βAPP/β-actin ratio showed in addition that the four compounds increased it significantly at the highest concentrations (100 nM to 1 μM) ( Figure  4C). Following the demonstration of the superior efficiency of compound 37 in inducing the production of sAPPα and in order to show that these effects are not restricted to a cell type but rather represent a ubiquitous phenomenon, we conducted the same experiments in human cells HEK293. The results showed that 37 produced effects similar and even Following the screening phase carried out at one single concentration (1 µM), we then examined the ability of the selected compounds to stimulate sAPPα at lower concentrations and in a dose-dependent manner in SH-SY5Y cells. Although an increasing trend was observed for both andrographolide and 9, we could not establish statistically significant differences with controls ( Figure 4A). Nevertheless, in addition to the fact that 31 and 37 were triggering a significant increase in sAPPα production at 1 µM ( Figure 4A) as previously observed (see Figure 2), we showed that 37 is also effective at 10 nM and 100 nM concentrations ( Figure 4A). Parallel analysis of βAPP protein levels showed no significant variation in βAPP immunoreactivity under any conditions ( Figure 4B). The concomitant measurement of the sAPPα/βAPP/β-actin ratio showed in addition that the four compounds increased it significantly at the highest concentrations (100 nM to 1 µM) ( Figure 4C). Following the demonstration of the superior efficiency of compound 37 in inducing the production of sAPPα and in order to show that these effects are not restricted to a cell type but rather represent a ubiquitous phenomenon, we conducted the same experiments in human cells HEK293. The results showed that 37 produced effects similar and even superior to those observed in SH-SY5Y cells, namely an increase in sAPPα production and in the sAPPα/βAPP/β-actin ratio at all concentrations tested ( Figure 4D, upper and lower panels, respectively). The additional observation that 37 also significantly reduced βAPP protein levels at the same concentrations in HEK293 cells ( Figure 4D, middle panel) most likely reflects some depletion of the substrate due to higher metabolic activity when compared to the SH-SY5Y cell line. superior to those observed in SH-SY5Y cells, namely an increase in sAPPα production and in the sAPPα/βAPP/β-actin ratio at all concentrations tested ( Figure 4D, upper and lower panels, respectively). The additional observation that 37 also significantly reduced βAPP protein levels at the same concentrations in HEK293 cells ( Figure 4D, middle panel) most likely reflects some depletion of the substrate due to higher metabolic activity when compared to the SH-SY5Y cell line. Based on these results, we then wanted to determine whether these compounds were capable of influencing the expression of the main α-secretase activity ADAM10 and of the β-secretase BACE1. Based on these results, we then wanted to determine whether these compounds were capable of influencing the expression of the main α-secretase activity ADAM10 and of the β-secretase BACE1.
Firstly, the Western blot analyses of ADAM10 ( Figure 5A) and BACE1 ( Figure 5B) in SH-SY5Y cells did not detect any significant changes between the control conditions and those where the cells were treated with the four compounds at concentrations ranging from 1 nM to 1 µM. These results were then confirmed for 37 in HEK293 cells ( Figure 5C). Because protein level measurement results from transcriptional, translational, and posttranslational events, we undertook to examine the genuine transcriptional effect of the four compounds (1 µM) by real time qPCR in both SH-SY5Y and HEK293 cells. The results indicated a slight but significant increase in ADAM10 mRNA levels when cells were treated with 1 µM of 31 and 37 (SH-SY5Y) or 9 (HEK293) ( Figure 5D), while no significant change in BACE1 mRNA levels was detected ( Figure 5E).
Firstly, the Western blot analyses of ADAM10 ( Figure 5A) and BACE1 ( Figure 5B) in SH-SY5Y cells did not detect any significant changes between the control conditions and those where the cells were treated with the four compounds at concentrations ranging from 1 nM to 1 μM. These results were then confirmed for 37 in HEK293 cells ( Figure 5C). Because protein level measurement results from transcriptional, translational, and posttranslational events, we undertook to examine the genuine transcriptional effect of the four compounds (1 μM) by real time qPCR in both SH-SY5Y and HEK293 cells. The results indicated a slight but significant increase in ADAM10 mRNA levels when cells were treated with 1 μM of 31 and 37 (SH-SY5Y) or 9 (HEK293) ( Figure 5D), while no significant change in BACE1 mRNA levels was detected ( Figure 5E).  This set of data suggests transcriptional up-regulation of the α-secretase ADAM10 as a minor although possibly involved mechanism in the observed beneficial effect of these compounds on βAPP metabolism.
We finally subjected andrographolide and the three derivatives to a thorough characterization aimed at evaluating their effect on the catalytic activities of αand β-secretases, which compete for βAPP processing, thereby tightly controlling the balancing between the amyloidogenic and the non-amyloidogenic pathways.

Effect of Derivatives 9, 31, and 37 on α-Secretase Catalytic Activity
In a first set of experiments, we examined the impact of increasing concentrations (10 nM up to 10 µM for andrographolide and 1 nM to 1 µM for compounds 9, 31, and 37) of the four compounds on the α-secretase activity by measuring the phenanthroline-sensitive hydrolysis of the fluorimetric JMV2770 substrate by cultured SH-SY5Y cells.
Our results indicated that andrographolide slightly and dose-dependently enhances the JMV2770-hydrolyzing activity, 9 displaying such capability only at 10 nM while 31 and 37 remain inert in this paradigm ( Figure 6A). Andrographolide and 9 were subsequently submitted to the same assay in HEK293 where they also significantly contributed to a moderate stimulation of the α-secretase activity, although showing a slightly different pattern when compared to SH-SY5Y cells ( Figure 6B). This set of data suggests transcriptional up-regulation of the α-secretase ADAM10 as a minor although possibly involved mechanism in the observed beneficial effect of these compounds on βAPP metabolism.
We finally subjected andrographolide and the three derivatives to a thorough characterization aimed at evaluating their effect on the catalytic activities of α-and β-secretases, which compete for βAPP processing, thereby tightly controlling the balancing between the amyloidogenic and the non-amyloidogenic pathways.

Effect of Derivatives 9, 31, and 37 on α-Secretase Catalytic Activity
In a first set of experiments, we examined the impact of increasing concentrations (10 nM up to 10 μM for andrographolide and 1 nM to 1 μM for compounds 9, 31, and 37) of the four compounds on the α-secretase activity by measuring the phenanthroline-sensitive hydrolysis of the fluorimetric JMV2770 substrate by cultured SH-SY5Y cells.
Our results indicated that andrographolide slightly and dose-dependently enhances the JMV2770-hydrolyzing activity, 9 displaying such capability only at 10 nM while 31 and 37 remain inert in this paradigm ( Figure 6A). Andrographolide and 9 were subsequently submitted to the same assay in HEK293 where they also significantly contributed to a moderate stimulation of the α-secretase activity, although showing a slightly different pattern when compared to SH-SY5Y cells ( Figure 6B).   for compounds 9, 31, and 37). (B) Experiments carried out with andrographolide and 9 under the same conditions on cultured HEK293 cells. The curves represent the mean specific fluorescence (from 2 to 3 independent experiments including two controls each) while bars in histograms are expressed as a percentage of control (white bars, non-treated cells) calculated from the linear parts of the curves (initial velocity) and are the means ±SE of 8 to 16 independent determinations. * p < 0.05; ** p < 0.03; *** p < 0.02; # p < 0.01;~p < 0.003; π p < 0.0001 (gray bars); ns, no statistical difference (black bars).

Effect of Derivatives 9, 31, and 37 on β-Secretase Catalytic Activity
Another important aspect of this study was to determine if these compounds could behave as inhibitors of the amyloidogenic β-secretase catalytic activity. Taking advantage of a well-characterized BACE1-selective fluorimetric assay, we have first measured the effect of the four molecules, at the same concentrations used for the α-secretase assay, on the JMV1197-sensitive hydrolysis of the fluorimetric JMV2236 substrate in SH-SY5Y cell extracts at acidic pH. We showed a capability of all the compounds to reduce BACE1 activity, 31 and 37 operating the most efficiently and in a dose-dependent manner ( Figure 7A). with andrographolide and 9 under the same conditions on cultured HEK293 cells. The curv resent the mean specific fluorescence (from 2 to 3 independent experiments including two co each) while bars in histograms are expressed as a percentage of control (white bars, non-t cells) calculated from the linear parts of the curves (initial velocity) and are the means ±SE of independent determinations. * p < 0.05; ** p < 0.03; *** p < 0.02; # p < 0.01; ~ p < 0.003; π p < 0.0001 bars); ns, no statistical difference (black bars).

Effect of Derivatives 9, 31, and 37 on β-Secretase Catalytic Activity
Another important aspect of this study was to determine if these compounds behave as inhibitors of the amyloidogenic β-secretase catalytic activity. Taking adva of a well-characterized BACE1-selective fluorimetric assay, we have first measure effect of the four molecules, at the same concentrations used for the α-secretase assa the JMV1197-sensitive hydrolysis of the fluorimetric JMV2236 substrate in SH-SY5 extracts at acidic pH. We showed a capability of all the compounds to reduce BAC tivity, 31 and 37 operating the most efficiently and in a dose-dependent manner (F 7A).   for compounds 9, 31, and 37). (B) Experiments carried out with 31 and 37 under the same conditions in HEK293 cell extracts. The curves represent the mean specific fluorescence (from 2 to 3 independent experiments including two controls each) while bars in histograms are expressed as a percentage of control (white bars, non-treated cells) calculated from the linear parts of the curves (initial velocity) and are the means ±SE of 6 to 15 independent determinations. * p < 0.05; ** p < 0.03; *** p < 0.02; π p < 0.0001 (gray bars); ns, no statistical difference (black bars).
The data obtained with 31 and 37 were then reproduced with HEK293 cell extracts, thereby confirming the genuine ability of these two andrographolide analogues to potently block the β-secretase catalytic activity ( Figure 7B). It should be noted here that the inhibitory effects of the compounds on the β-secretase activity seem to be more pronounced than their capability to stimulate the α-secretase activity. From the fact that compound 9 of the 8,17-olefinic series possesses anti-BACE1 activity, it is suggested that the 5 ,7 -dichloro-8 -hydroxyquinolyloxy moiety is important for BACE1 inhibition. As the 2 -methyl-5 ,7dichloro-8 -hydroxyquinolyloxy derivatives 31 and 37 are potent BACE1 inhibitors, this particular structure might interact in a more efficient way with the BACE1 catalytic site. Finally, the observation that 31 and 37 are the most potent β-secretase inhibiting factors in this study, suggests that the 8,9-double bond in the 9-dehydro-17-hydro series is an important feature for a proper inhibition of this activity.

Discussion
AD is a yet incurable neurodegenerative disorder characterized by loss of memory and cognition. The reason why available medical treatments are still incapable of curing AD symptoms efficiently mostly resides in the fact that AD is a complex and multifactorial pathology. Over the past decades, a huge effort, although in vain, has been made to develop novel synthetic drugs with disease-modifying properties and few side effects [35]. Hence, compounds extracted from natural sources are constantly gaining popularity in AD treatment with the notion of preventive rather than curative intervention against the disease being increasingly considered.
Regarding AD and on a mechanistic point of view, andrographolide most likely conveys some anti-AD effects via its well-established antioxidant [45] and NFκB inhibitory and anti-inflammatory [46] properties as illustrated for instance by the fact that andrographolide inhibits Aβ 1-42 -induced production of neuroinflammatory mediators in microglia [13,14]. However, whether it could regulate the processing of βAPP through the control of βAPP-cleaving secretases was still an unanswered question. Here, we first identified andrographolide as well as some chemically modified andrographolide analogues as regulators of βAPP processing in cultured human cell lines using sAPPα production as a read-out. This metabolite with beneficial properties arises from the cleavage of βAPP by the non-amyloidogenic α-secretase activity. Because α-secretase and β-secretase, the amyloidogenic rate-limiting initiator of amyloid peptide production, work competitively regarding βAPP processing as evidenced by the inverse correlation between sAPPα and Aβ productions under both α-secretase activation or β-secretase inhibition [47,48], any increase in sAPPα production can result from either an activation of α-secretase or an inhibition of β-secretase (that disrupts sAPPα integrity by cleaving inside its sequence), or both. We therefore undertook to study the effect of andrographolide, 9, 31, and 37, all initially identified on the basis of their ability to induce the production of sAPPα, on the catalytic activities of αand β-secretase by means of specific fluorimetric assays. This allowed us to establish that the four compounds indeed regulate these activities although to different degrees and that 31 and 37 behave as potent β-secretase inhibitors. It has to be underlined here that their efficiency at submicromolar concentrations lays the groundwork for the future production of highly potent derivatives that could serve as a basis for their therapeutic use. Moreover, the confirmation of the results in HEK293 cells suggests that the effects observed are probably ubiquitous and not restricted to one cell type.
Importantly, studies carried out in animals have shown that andrographolide does not trigger toxicity in the liver and the kidney of rat [49] and does not alter body and organ weight, inflammatory responses, hematological parameters, and mortality in mice [50]. These data therefore established andrographolide as a relatively safe compound in respect to toxicological side effects. Moreover, andrographolide easily passes the blood-brain barrier and distributes into different brain regions [51]. However, restricted bioavailability due to its poor solubility and relatively short half-life obviously limits its clinical application and numerous semi-synthetic transformations were performed in order to improve its physiochemical properties and stability [52,53].
Considering that quinoline is also a pharmacophore group for neuroprotection [28,30,31], we envisaged that our published active 14-quinolyloxy derivatives of andrographolide against Zika and dengue viruses [32,33] and bacteria [54] possibly have anti-AD activity. Moreover, in addition to andrographolide itself [26,29,36,55], some of its 9-dehydro-17hydro analogues have increased neuroprotective properties [12] and are more efficient against angiogenesis [56,57] than its 8,17-olefinic counterpart. These results led us to explore whether 14-aryloxy-9-dehydro-17-hydro analogues also possess a higher capability to inhibit BACE1 activity than the 8,17-olefinic ones. Our results confirmed the concept that 14-quinolyloxy modification or combination of 14-quinolyloxy and 9-dehydro-17hydro modifications on andrographolide benefits for neuroprotection and will lead to the discovery of more potent and druggable anti-AD compounds.
Overall, our original findings that andrographolide derivatives display both proα-secretase and anti-β-secretase properties open the way to the possible development, hitherto not explored, of molecules of natural origin capable of acting in a doubly beneficial manner on the metabolism of βAPP and representing a new class of factors to be developed as a therapeutic tool aimed at combating Alzheimer's disease. The design, synthesis and testing of new chemically modified andrographolide-derived compounds aimed at obtaining highly potent αand β-secretases regulating molecules is currently being carried out in our laboratories.

General Information for Chemistry
1 H and 13 C NMR spectra (Supplementary Materials) were recorded on a Bruker AV-400 spectrometer at 400 and 100 MHz, respectively, in CDCl 3 , CD 3 OD, (CD 3 ) 2 SO, and C 6 D 6 as indicated. Coupling constants (J) are expressed in hertz (Hz). Chemical shifts (δ) of NMR are reported in parts per million (ppm) units relative to the solvent. The high resolution of ESI-MS was recorded on an Applied Biosystems Q-STAR Elite ESI-LC-MS/MS mass spectrometer, respectively. Unless otherwise noted, materials were obtained from commercial suppliers and used without further purification. Melting points were measured using an YRT-3 melting point apparatus (Shanghai Yice Apparatus & Equipment Co., Ltd., Shanghai, China) and were uncorrected. The synthesis of compounds 7 to 10 and 12 to 16 (Scheme 1) was previously described [32,33].

Synthesis of Compounds 19 to 26
The key intermediates 25 and 26 of the 9-dehydro-17-hydro series were synthesized as shown in Scheme 2.
Synthesis of compound 19: a solution of compound 2 (5.0 g, 12.8 mmol) in 10 mL of dry dichloromethane (50 mL) was cooled in ice-water bath and then triethylamine (4.5 mL, 32.0 mmol) was added, followed by p-nitrobenzoyl chloride (2.85 g, 15.4 mmol) in 20 mL of dry dichloromethane. The reaction mixture was stirred in an ice-water bath for 5 h and volatile solvents were removed by Rotavapor. The residue was dissolved in ethyl acetate and treated with sat. NaHCO 3 aqueous solution. The organic phase was washed with brine twice and then dried over anhydrous Na 2 SO 4 . The filtered organic solution was evaporated to dryness and the residue was purified by silica gel column chromatography (petroleum ether/ethyl acetate 3/1) to afford 5.9 g of titled compound 19. Synthesis of compound 20: to a solution of compound 2 (5.0 g, 12.8 mmol), pnitrobenzoic acid (2.57 g, 15.4 mmol) and triphenylphosphine (5.0 g, 19.2 mmol) in anhydrous THF (50 mL) placed under N 2 and at 0 • C, diisopropyl azodicarboxylate (DIAD) (3.76 mL, 19.2 mmol) was added. The reaction was stirred at 0 • C for 1 h and at room temperature overnight. After the reaction was complete as established by TLC monitoring, the volatile solvents were distilled off, the residue was dissolved in ethyl acetate and the organic phase was washed with brine twice, dried over anhydrous Na 2 SO 4 , filtered, and evaporated to dryness. Titled compound 20 was purified by silica gel column chromatography (petroleum ether/ethyl acetate 8/1) to yield 5.3 g. Synthesis of compound 21: compound 19 (5.9 g, 10.9 mmol) was added to 85% phosphoric acid (40.0 mmol) with fast stirring and the solid was dissolved gradually. The reaction was monitored by TLC and complete in about 3 h before being diluted carefully with distilled water followed by extraction with ethyl acetate. The organic phase was washed with a sat. NaHCO 3 aqueous solution and brine and then dried over anhydrous Na 2 SO 4 . Filtered organic phase was evaporated and the residue was purified by silica gel column chromatography (petroleum ether/ethyl acetate 3/1) to yield 3.6 g of compound 21. (14α)-(4 -nitrobenzoyl)-9-dehydro-17-hydro andrographolide (21) Synthesis of compound 23: compound 21 (3.6 g, 7.2 mmol) was dissolved in 2,2dimethoxypropane (7.5 mL, 50.4 mmol) and 2.5 mL dry dichloromethane and pyridinium 4-toluenesulfonate (88 mg, 0.36 mmol) was added. The reaction was stirred at 45 • C and complete in 3 h as monitored by TLC. After volatile solvents were distilled off, the residue was taken off with ethyl acetate and the organic phase was washed with sat. CuSO 4 aqueous solution, sat. NaHCO 3 solution, and brine. The organic phase was dried over anhydrous Na 2 SO 4 , filtered, and evaporated in vacuo, and the residue was purified by silica gel column chromatography (petroleum ether/ethyl acetate 3/1) to give 3.1 g of compound 23. 7 mmol) in 20 mL methanol, lithium carbonate (794 mg, 11.5 mmol) was added, and the mixture was stirred at room temperature for 2 h (TLC monitoring). After removal of volatile solvents by rotavapor, the residue was treated with ethyl acetate and the organic phase was washed with brine twice, dried over anhydrous Na 2 SO 4 , filtered, and distilled off to dryness. The residue was purified by silica gel column chromatography (petroleum ether/ethyl acetate 3/1) to give 1.6 g of titled compound 25. (14α)-9-dehydro-17-hydro-3,19-isopropylideneoxy andrographolide (25) 13

Synthesis of Compounds 27 to 32
Titled 9-dehydro-17-hydro series compounds 27 to 32 were prepared as shown in Scheme 2 according to previously described procedures [32][33][34]. Under N 2 atmosphere, 1.0 mmol of compound 25 or 26, 1.5 mmol of PPh 3 , and 1.5 mmol of 8-hydroxyquinoline derivative 4, 5, or 6 were dissolved in 10.0 mL of anhydrous THF. The solution was cooled to 0 • C and then treated with 1.5 mmol of DIAD in 2.0 mL of anhydrous THF. The reaction was stirred overnight at room temperature after being stirred at 0 • C for 1 h. After distilling off the volatile solvents, the residue was dissolved in ethyl acetate and the organic phase was washed with brine about 5 times and dried over anhydrous Na 2 SO 4 . The filtered organic solution was evaporated to dryness and the residue was purified by silica gel column chromatography to afford compounds 27 to 32.

Statistical Analysis
Statistical analyses were performed with the Prism software (GraphPad, San Diego, USA) using an unpaired t-test for pairwise comparisons. All results are expressed as means ± SEM and p values equal to or less than 0.05 were considered significant.
Supplementary Materials: The following are available online. NMR spectra for Scheme 1; NMR spectra for Scheme 2.