Multi-target-directed phenol–triazole ligands as therapeutic agents for Alzheimer's disease† †Electronic supplementary information (ESI) available. See DOI: 10.1039/c7sc01269a Click here for additional data file.

A series of multi-target-directed ligands are described that bind Cu, act as antioxidants, modulate Aβ peptide aggregation, and abolish Aβ toxicity in primary neurons.


Introduction
Healthcare advances across the globe have increased life expectancy, facilitating increases in the prevalence of neurodegenerative diseases such as Alzheimer's disease (AD). 1,2 In Canada for example, ca. 5% of the population over 65 years of age have AD, increasing to 25% over the age of 85 years old. 3 Currently, there are no approved disease-modifying therapeutic interventions other than symptomatic treatments such as cholinesterase inhibitors. 4 AD is formally characterized by two distinct neuropathological features: extracellular amyloid-b (Ab) plaques and intracellular neurobrillary tangles (NFTs). 5,6 The major constituent of Ab plaques is the Ab peptide of ca. 38-43 amino acid residues, with Ab 1-40 and Ab 1-42 being the most abundant forms. NFTs result from the hyperphosphorylation of tau proteins. Oxidative stress is linked to the formation of both of these pathological features, 7 and recent reports suggest an inter-relationship between Ab and NFTs. 8 AD is a multifactorial disease where oxidative stress, altered Ab clearance mechanisms, Ab and tau aggregation, and dysregulated metal ions play a role in disease etiology. 9 The Nterminus of the Ab peptide exhibits a relatively high affinity for Cu (K d ¼ 10 À11 to 10 À7 ) and Zn (K d ¼ 10 À9 to 10 À6 ) where residues His6, His13, and His14 are located. [10][11][12][13] The concentration of metal ions in Ab plaque deposits is 3-5 fold higher in comparison to age-matched healthy parenchyma, suggesting that Ab plaques act as metal reservoirs, as overall metal concentrations in the brain are not altered. 14 The interaction between metal ions and the Ab peptide drastically alters the Ab aggregation prole; interaction with Zn affords amorphous high molecular weight species, while interaction with Cu affords neurotoxic oligomers. 15 In the case of CuAb, peptide binding promotes Cu II /Cu I redox cycling, and generation of reactive oxygen species (ROS), potentially implicating these metalated species in oxidative stress, and neuronal toxicity. [16][17][18] Small molecule chemical agents that can address multiple factors associated with AD etiology may play a key role in the development of new effective therapeutic strategies. Specically, developing multifunctional agents that can act on multiple disease pathways could provide key information on the interrelationship between these pathways, enhancing our overall understanding of AD etiology. The rational design of multifunctional metal binding agents has become a promising therapeutic strategy. [19][20][21][22][23][24][25][26][27] Previously, we developed several multifunctional frameworks based on pyridine-triazole and quinoline-triazole scaffolds (Fig. 1). The pyridine-triazole frameworks were shown to alter metal-Ab interactions and associated Ab aggregation. 28 The quinoline-triazole analogues were developed to provide an enhanced interaction with the hydrophobic region of the Ab peptide. 29 In a signicant advance, we report herein a series of multi-target directed ligands possessing a phenol-triazole framework (Fig. 1). These ligands exhibit an enhanced affinity for Cu in comparison to the pyridine/quinoline analogues, and in addition, contain an antioxidant phenolic group. The phenol-triazoles were shown to modulate the Ab aggregation prole in the presence and absence of added Cu based on gel electrophoresis and transmission electron microscopy. 2-D NMR and molecular modelling studies were employed to provide further insight into the ligand-peptide interaction. Finally, one compound in the series (POH) was shown to reverse Ab 1-42 neurotoxicity in primary neurons.

Results and discussion
Ligand properties AD drug development has targeted several different pathophysiological pathways, including Ab peptide aggregation, tau protein hyperphosporylation, oxidative stress, cholinesterase inhibition, and metal ion dyshomeostasis. 30 The lack of a single dened target suggests that an effective therapeutic intervention may require the ability to address several different factors associated with the disease. Herein, a series of phenol-triazoles (POH, PMorph and PTMorph, Fig. 1) were designed to bind Cu, modulate metal-peptide interactions, and limit the generation of reactive oxygen species (ROS) via the phenol moiety. The phenol-triazole series were synthesized in a modular fashion using Huisgen's 1,3-dipolar cycloaddition, also known as click chemistry. 31,32 Reaction of an alkyne substituted phenol and azides with different peripheral R-groups affords a bidentate metal-binding site (Scheme S1 †).
We next determined the ligand acidity constants via a combination of variable pH UV-Vis and 1 H NMR spectroscopies ( Fig. S1-S3 †). The phenol pK a values were thus determined to be within the range of 9.54-9.55 (Table S1 †), comparable to the deprotonation of free phenol (pK a ¼ 9.98 (ref. 33)). The additional pK a values for the peripheral morpholine (PMorph) and thiomorpholine (PTMorph) functions were determined to be 5.5 and 5.6, respectively. These values are lower than free morpholine and thiomorpholine (pK a morpholine ¼ 8.36, thiomorpholine ¼ 9.0) as has been previously reported in similar systems. 29,34-36 Overall, each ligand was found to be neutral at physiological pH 7.4, which is optimal for passive diffusion across the BBB. To further investigate the drug-like properties of the phenol-triazoles, several physicochemical parameters were calculated. Each phenol-triazole ligand was determined to comply with Lipinski's rules and log BB for drug-likeness and BBB penetration (Table S2 †). 37,38 Overall, the phenol-triazole series exhibit promising physicochemical properties warranting further development.

Metal binding affinity
We next investigated the Cu-binding affinity of the bidentate phenol-triazole ligands in solution using variable pH UV-vis titrations. The Ab peptide exhibits a high affinity for Cu(II) (K d ca. 10 À11 to 10 À7 ) 12,13 and thus in order to compete with Ab for Cu, either via de-metallation or ternary complex formation, ligands should exhibit a Cu K d value in the 10 À12 to 10 À8 range. 12 This range is appropriate to disrupt Cu-Ab interactions, while limiting sequestration of essential metal ions from metalloproteins. 39 The stoichiometry of the Cu(II)phenol-triazole complexes in PBS (pH 7.4) was initially determined using Jobs plot analysis (Fig. S4 †). 40 For the example ligand PMorph, a broad maxima is observed between ca. 0.3-0.6 mole fraction of Cu(II), suggesting the formation of both 1 : 1 and 1 : 2 PMorph-Cu(II) complexes in solution (vide infra). 41,42 Further, both 1 : 1 and 1 : 2 binding stoichiometries were observed for the three phenol-triazoles by both Jobs plot and ESI-MS analysis (data not shown). Measurement of the binding affinity of the phenol-triazole ligands for Cu(II) was completed using variable pH spectrophotometric titrations and speciation modelling using HypSpec (Fig. 2). 43 The ligand pK a values, as well as the hydrolysis reactions of free Cu(II), were included as constants in the calculations. 44 Modelling of the data for POH shows signicant log K values for both 1 : 1 and 1 : 2 Cu : L ratios (where L is deprotonated) as shown in Table 1. As expected, the calculated log K values are similar across the phenol-triazole series (Table 1), reecting a common metal-binding motif. A representative speciation diagram for POH (Fig. 2) shows that at physiological pH 1 : 1 and 1 : 2 Cu : L species predominate (in agreement with Jobs plot analysis) with negligible free Cu. At higher pH, the [CuL 2 (OH)] À species becomes relevant. Similar speciation diagrams for PMorph and PTMorph are shown in the ESI ( Fig. S5 and S6 †). Using the solution speciation diagrams, the concentration of unchelated Cu(II) (pCu ¼ Àlog([Cu unchelated ])) at pH 7.4 and total Cu concentration can be calculated (Table 1). The pCu value is a direct estimate of the ligand-Cu affinity by taking into account all relevant equilibria, and can therefore be used to compare the metal-binding affinity among the various ligands, including the Ab peptide. The pCu values are similar across the phenol-triazole series (6.6-6.9), and represent approximate dissociation constants (nM to mM range) that compare favourably with the K d values reported for Cu-Ab species. 12,13 The Cu-binding affinity for the phenol-triazole ligands are thus signicantly stronger than those reported for the quinoline-triazole analogues. 29 A Cu-competition assay was then performed in which 2 eq. of each phenol-triazole ligand was added to pre-formed Cu-Ab 1-16 and Cu-Ab 1-42 species. The Cu complex of the phenol-triazole ligands displays an absorption peak at ca. 320 nm (Fig. S7-S9 †), and upon addition of ligand to solutions containing Cu-Ab 1-42/1-16 , an increase at 320 nm is observed. This data suggests an interaction of the phenol-triazole ligands with Cu. A baseline increase in the Ab 1-42 experiments is likely due to aggregate formation and associated light scattering. On the basis of this data we expect the phenol-triazole ligands to have the appropriate Cu-binding affinity to interact with Cu in the presence of the Ab peptide.

Anti-oxidant assays
To further evaluate the multifunctional nature of the phenoltriazoles, we investigated their antioxidant capability using several different assays. [46][47][48] In the rst assay, the antioxidant activity of the phenolic functions were evaluated using a Trolox-Equivalent Antioxidant Capacity (TEAC) assay. The TEAC assay has been used to quantify the antioxidant activity of biological uids, extracts, and pure compounds by measuring the disappearance of the ABTS + c radical cation via UV-vis spectroscopy. 49 Trolox, a water-soluble vitamin-E analogue, is used as a standard against which each ligand is compared. In addition, each phenol-triazole ligand was compared with glutathione and PBT2, an 8-hydroxyquinoline derivative, that has shown promise as an Alzheimer's disease therapeutic. 50,51 Each ligand exhibited TEAC values comparable to Trolox and slightly enhanced in comparison to PBT2 (Fig. 3).
In two other complimentary antioxidant assays, we measured the ability of the phenol-triazole ligands to bind Cu and limit hydroxyl radical formation from aqueous Cu in the presence of dioxygen. Firstly, using a uorescent coumarin carboxylic acid (CCA) assay, in which the highly reactive hydroxyl radical specically hydroxylates CCA at the 7-position to form a uorescent product, 52,53 aerobic aqueous Cu solutions in the presence of a physiologically-relevant concentration of ascorbate 54 exhibit signicant uorescence over a short period (Fig. 3). The addition of 2 eq. POH, PMorph, or PTMorph to the Cu/ascorbate/CCA solutions signicantly reduces the observed uorescence (Fig. 3), consistent with either the high affinity of these ligands for Cu (vide supra), or direct reaction of the ligands with the generated cOH. A complimentary ascorbate (Asc) reduction assay 55,56 was employed to discern if Cu chelation was responsible for the observed response in the CCA experiments. In the absence of ligands, Asc consumption in the presence of Cu is rapid (Fig. S10 †), however, upon addition of 3 eq. of the high affinity chelator DTPA, Asc consumption is arrested. We observed only a small decrease in Asc consumption upon   addition of 10 eq. of phenol-triazole ligands to a Cu/Asc system (Fig. S10 †), signifying that the compounds act primarily as antioxidants, and not via Cu redox silencing, in the CCA assay. Ascorbate consumption was also evaluated in the presence of Ab 1-16 , which contains the metal binding region for Cu (Fig. S11 †). When comparing Cu-Ab 1-16 vs. Cu-Ab 1-16 + 10 eq. of phenol-triazole ligand, no change in the rate of ascorbate consumption was observed. This suggests that the ligands do not limit Cu 2+/+ redox cycling and therefore, act primarily as antioxidants, reacting with Cu-generated ROS.

Ligand-Ab peptide interactions
The direct interaction of the phenol-triazole ligands with monomeric Ab 1-40 was investigated using 2-D 1 H-15 N SOFAST-HMQC NMR experiments. 29,57 The less aggregation prone Ab 1-40 peptide length was used here to limit aggregation during data collection and ensure all shis are solely the result of peptideligand interactions and not peptide aggregation. Incubation with one of POH, PMorph, or PTMorph resulted in small but detectable chemical shi changes for specic peptide residues (see ESI for experimental details †). Interestingly, POH was determined to exhibit signicant chemical shi changes distributed across the entire peptide length, including Ab residues E3, D7, Y10, V18, F20, G33, V36, and G38 (Fig. 4). In the case of PMorph, signicant chemical shi changes were observed for Ab residues D7, F19, D23, and N27 (Fig. S12 †). The hydrophobic region of the Ab peptide, encompassing residues 17-21, and in particular F19, is considered to play a critical role in the initial stages of peptide aggregation, and thus the interaction of PMorph with F19 may play a signicant role in mediating the Ab aggregation process (vide infra). [58][59][60][61][62][63] PTMorph demonstrates similar interactions to those observed with PMorph, including signicant chemical shi changes of Ab residues D7, V18, F19, N27, and G33 (Fig. S13 †). The similar interaction of PMorph and PTMorph with Ab is not surprising as the only structural difference is the O for S heteroatom substitution.
Although limited to three ligands, the 2-D NMR results suggest that the triazole R-group plays a signicant role in dictating Ab peptide interactions (Fig. S14 †), with the propanol group of POH affording non-specic interactions, while the morpholine and thiomorpholine heterocycles confer a higher degree of selectivity for Ab peptide residues. The direct interaction of the phenol-triazoles with the Ab peptide, as supported by the 2-D NMR studies, suggest that these ligands may modulate the Ab aggregation prole in solution, even in the absence of Cu (vide infra).

Ab peptide aggregation experiments
Modulation of the Ab peptide aggregation pathway offers a signicant opportunity for drug development. Shiing the Ab aggregation pathway away from toxic intermediates, such as soluble oligomeric species, 64-68 may lead to a decrease in neuronal cell death.
To further explore the multifunctional nature of the phenoltriazole ligands we investigated the effects of these ligands on Ab peptide aggregation in the presence and absence of Cu ions (Fig. 5A). Using native gel electrophoresis/western blotting to visualize the size distribution of Ab species, in conjunction with transmission electron microscopy (TEM) to examine Ab aggregate morphology, we assessed the ability of the phenol-triazole ligands to modulate Ab peptide aggregation.
Incubation of the phenol-triazole ligands with the Ab peptide results in signicant changes to the size distribution of peptide aggregates (Fig. 5B), highlighting the importance of the interactions observed in the 2-D NMR experiments. Incubation of Ab 1-42 alone over 24 h (lane 1) affords both high and low molecular weight species, and signicant bril formation as observed by TEM (Fig. 5C), in line with previous reports. 42,69,70 Addition of the phenol-triazole ligands signicantly alters the aggregation prole towards high molecular weight species (Fig. 5B, lanes, 3, 4 and 5), and in addition the aggregate morphology is now amorphous as ascertained by TEM (Fig. 5C). These results show that the presence of the phenol-triazole ligands limits the formation of oligomeric species (10-100 kDa range), which have been implicated to play a major role in Ab-associated toxicity. 71,72 We next investigated the effect of the phenol-triazole ligands on Ab peptide aggregation in the presence of Cu ions. The metal competition assays (Fig. S7-S9 †) suggest that the phenol-triazole ligands only partially demetallate Cu-Ab and thus we investigated the effect of the ligands on aggregation of metalated peptide species. Incubation of Ab 1-42 with 1 eq. of Cu affords primarily oligomeric species in comparison to Ab 1-42 alone at 24 h (Fig. 5B, lane 1 vs. 2). 15 In addition, aggregate morphology has changed from brillar to amorphous (Fig. 5C). Incubation of Cu-Ab peptide solutions with the phenol-triazole ligands leads to signicant changes in the aggregation prole (Fig. 5B, lane 2, vs. lanes 6, 7, and 8). For POH, a signicant shi towards high molecular weight species is observed, with a similar soluble aggregate prole to that observed in the peptide only experiment (Fig. 5B, lane 3 vs. lane 6). Interestingly, TEM analysis shows primarily brillar aggregates for Cu-Ab in the presence of POH (Fig. 5C). While this data is consistent with Cu sequestration by the ligand and aggregation of the peptide, 73 it is likely that ligand-peptide interactions are the dominant factor that controls the aggregation process of both Ab 1-42 and Cu-Ab 1-42 .
For PMorph and PTMorph, similar Ab-aggregation proles are observed in the presence of Cu (Fig. 5B, lanes 7 and 8), which differ signicantly from the Cu-Ab experiment (Fig. 5B, lane 2), and also from the POH experiment (Fig. 5, lane 6). Interestingly, PMorph and PTMorph shi Ab aggregation towards distinctly different high molecular weight species in comparison to POH (ca. 100 kDa vs. 250 kDa). TEM analysis of the Cu-Ab aggregation experiment in the presence of PMorph shows the formation of brils (Fig. 5C), while for PTMorph both brils and amorphous aggregates are observed (Fig. 5C). The three phenol-triazole ligands exhibit similar Cu affinity constants and thus the different aggregation proles observed for the ligands in the presence of Cu-Ab may result from distinct interactions of the triazole R-groups with the Ab peptide, leading to ternary complex formation, 19,74,75 and/or interactions with specic aggregates, such as oligomeric Ab.

Attenuation of Ab-induced neurotoxicity in human neuronal culture
To investigate whether the phenol-triazole ligands protect against Ab 1-42 -mediated neurotoxicity, human fetal neurons (HFN) were treated with POH, PTMorph and PMorph for 1 hour prior to Ab 1-42 application. Treatment with monomeric Ab 1-42 induced signicant cell death (Fig. 6, 24% fewer MAP-2 positive neurons compared to the control condition, p < 0.0001). Upon treatment with 75 mM POH, complete prevention of Ab 1-42 -induced neurotoxicity was achieved (Fig. 6, p < 0.001). PMorph was also able to protect against Ab 1-42 -induced neurotoxicity (p < 0.05), while PTMorph demonstrated no protective effect in comparison to Ab 1-42 . When 75 mM of ligand was incubated with HFN for 24 hours in the absence of any Ab 1-42 , similar cell viability was observed in comparison to the control, indicating that these ligands are not toxic at the experimental concentration used.
We next assessed the neurotoxicity of Ab 1-42 premixed with Cu (1 : 1) at 5, 10 and 25 mM (Fig. S15 †). While Cu alone was not toxic at these concentrations, Cu-Ab 1-42 exhibited signicant toxicity at all concentrations studied. Pre-incubation of the HFN with the phenol-triazole ligands before addition of Cu-Ab 1-42 did not have a neuroprotective effect at the 24 h timepoint (Fig. S15 †). Overall, these results are consistent with the previously reported toxicity of Cu-Ab 1-42 species, 15,76,77 and show that under the HFN assay conditions the phenol-triazole ligands cannot prevent Cu-Ab 1-42 toxicity. While not effective in limiting the toxicity of Cu-Ab 1-42 in neurons, the phenol-triazole ligands are effective under metal-free conditions. Overall, POH demonstrates the best neuroprotective properties when in the presence of Ab 1-42 , conferring complete protection, followed by PMorph, with PTMorph being ineffective in this assay.

Summary
In this report, we describe the development of a series of phenol-triazole ligands that target multiple factors associated with AD etiology. The modular synthetic strategy afforded ligands that exhibit favourable physicochemical properties via both experiments (pK a values) and calculations (drug-likeness), while also displaying antioxidant activity. The three phenoltriazole ligands were shown to interact with the Ab peptide via 2-D NMR studies, with the PMorph and PTMorph derivatives displaying similar interactions with specic peptide residues, including F19. Interestingly, POH displays a larger number of interactions distributed across the length of the Ab peptide. The three ligands alter the Ab peptide aggregation prole in a similar manner via limiting oligomer formation in favour of amorphous high molecular weight species. The Cu-binding affinity of the phenol-triazoles was determined to be of appropriate strength to compete with the Ab peptide, and as expected the ligands altered the Cu-Ab aggregation prole. In the presence of Cu, only POH signicantly reduced Ab oligomer formation promoting the formation of large molecular weight aggregates. These results highlight that the triazole R-group offers a signicant opportunity to tune the biological properties of the ligand scaffold. Encouragingly, POH, and to a lesser extent PMorph, exhibited a protective effect against Ab 1-42induced neurotoxicity in human neuronal culture, while these ligands were not able to rescue the neurotoxicity associated with Cu-Ab 1-42 under our conditions. Altogether, these promising results strongly suggest that the multifunctional phenol-triazole ligand scaffold, and in particular POH, warrants further investigation in an animal model to interrogate mechanisms of action and efficacy.  . One hour pre-incubation with 75 mM POH prevented Ab 1-42induced neurotoxicity (p < 0.001, (D)). PMorph was also shown to be neuroprotective (p < 0.05, (E)), whereas PTMorph had no effect (F). Treatment of HFN with ligands in the absence of Ab 1-42 did not induce cell death (A). Pictures are shown at 20Â original magnification, scale bar ¼ 100 mm. Significance is shown compared to HFN with Ab 1-42 . *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.