A rationally designed small molecule for identifying an in vivo link between metal–amyloid-β complexes and the pathogenesis of Alzheimer's disease

An in vivo chemical tool designed to target metal–Aβ complexes and modulate their activity was applied to the 5XFAD mouse model of Alzheimer’s disease (AD) demonstrating the involvement of metal–Aβ in AD pathology.


Introduction
Alzheimer's disease (AD), a progressive neurodegenerative disease, is the most common form of dementia afflicting 24 million people worldwide. 1 Despite AD being the sixth leading cause of death in the United States, there are currently no disease modifying treatments; approved therapies only offer symptomatic relief without having an effect on the underlying pathogenesis. 1,2 Development of effective therapeutics has been hindered by the fact that AD pathogenesis is still poorly understood. Pathologically, AD is characterized by the accumulation of aggregated, misfolded proteins, such as amyloid-b (Ab) peptides (two major forms exist, Ab 40 and Ab 42 ). 3,4 The amyloid cascade hypothesis suggests that Ab is the causative agent in AD; 5 however, the etiology of AD can be multifactorial; of particular interest is the role of Ab with other factors (i.e., metals) toward AD development. 4,[6][7][8][9][10][11] High concentrations of Fe, Cu, and Zn (ca. low mM) are found within Ab deposits in ex vivo tissues from the AD-afflicted brain. 12,13 These metal ions are observed to coordinate to Ab peptides in vitro forming metal-Ab complexes which could direct toxicity via two possible pathways: 4,6-11,14-19 (i) metals could inuence the Ab aggregation pathways leading to the generation and stabilization of toxic Ab oligomers; 4,7-9,14 (ii) redox active metal ions (i.e., Cu(I/II) and Fe(II/III)) associated with Ab are shown to produce reactive oxygen species (ROS) under physiological conditions through Fenton-like reactions. 4,[6][7][8][9][10][11][16][17][18][19] Overproduction of ROS by metal-Ab can result in oxidative stress and eventually neuronal death in the AD-affected brain. Although the reactivity of metal-Ab (i.e., (i) metal-Ab aggregation (toxic Ab oligomer formation) and (ii) redox active metal-Ab-triggered ROS generation, vide supra) has been indicated in vitro, 4,[6][7][8][9][10][11][14][15][16][17][18][19] the direct involvement of metal-Ab complexes in AD pathogenesis in vivo is uncertain.
Metal chelating agents have shown that the interference of metal-Ab interactions as well as the modulation of metal distribution in the brain could lead to an improvement in AD pathology. 4,19-24 8-Hydroxyquinoline derivatives have been employed to regulate metal-related neurotoxicity in AD; some small molecules, including clioquinol (CQ) and PBT2, have indicated promising results for possible AD treatment in clinical trials. 4,22,23 The effects of CQ and PBT2 are mainly from their ability to act as an ionophore to redistribute metal ions in the brain instead of directly disrupting metal-Ab complexes; 4,19,[23][24][25] thus, these compounds would not be able to directly probe the relation between metal-Ab complexes and AD pathogenesis. Therefore, chemical tools, termed as metamorphosizers, have been recently developed in order to (i) specically target metal-Ab complexes and (ii) alter the interaction between the metal and Ab, consequently (iii) redirecting the toxic aggregation pathway of metal-Ab into off-pathway, less toxic unstructured Ab forms and (iv) reducing metal-Abinduced ROS production, which eventually alleviates metal-Ablinked toxicity. 4,24 Herein, we demonstrate that a chemical tool (L2-b, Fig. 1a) stands out as being well suited in vivo for identifying the association of metal-Ab 40 /Ab 42 with AD pathogenesis, through in vitro biochemical/biophysical/cytotoxicity/metabolism investigations, as well as in vivo brain uptake studies. Our in vivo tool specically interacts with metal-Ab over metal-free Ab and generates a ternary L2-b-metal-Ab complex causing structural compaction, as validated by mass spectrometry (MS) and ion mobility-mass spectrometry (IM-MS). Most signicantly, we present the rst report that the control of metal-Ab interaction and reactivity by an in vivo chemical tool mitigates amyloid pathology and improves cognitive decits in the 5XFAD AD mouse model. This robust AD mouse model develops severe amyloid pathology and cognitive decline at an early age through high expression of three familial mutant types of human amyloid precursor protein (hAPP; Swedish, Florida, and London) and two mutant forms of presenilin (PSEN1; M146L and L286V). 26 Overall, our studies establish strong experimental evidence for an in vivo link between metal-Ab and AD development, implying that targeting metal-Ab complexes could be an effective strategy for the future development of new therapeutics.

Results and discussion
Design principle and characterization of a chemical tool for investigating metal-Ab complexes in vivo L2-b (Fig. 1a) was designed to target metal-Ab complexes and modulate their interaction/reactivity with subsequent reduction of toxicity, 27 in order to determine whether they are connected with AD pathology. For in vivo applications, rst, chemical tools for this purpose must have specicity toward metal-Ab complexes in order to limit the disruption of other metalloproteins. 4,24 This property can be imparted into small molecules by using inorganic chemistry concepts to allow specicity for disease-relevant metal ions (Fe(II/III), Cu(I/II), and Zn(II)), along with limiting the metal binding affinity (K d ) to $10 À10 M, and by including structural components for Ab interaction. 4,24 To satisfy this aspect, L2-b (a bidentate ligand; Fig. 1a) was constructed upon incorporation of two nitrogen donor atoms (for metal chelation) into the structure of an Ab aggregate imaging agent (stilbene derivative; for Ab interaction), 28 which could interact with metal-Ab complexes (Fig. 1a). 27 L2-b is shown to have apparent K d values of ca. 10 À10 and 10 À6 M for Cu(II) and Zn(II), respectively, and is relatively selective for Cu(II) over other biologically relevant bivalent ions. 27 Secondly, the blood-brain barrier (BBB) permeability of L2-b is critical for applications in the brain, which was rst predicted by considering Lipinski's rules of drug-likeness and observing calculated log BB values. 27 Employing CD1 mice, in vivo brain uptake studies of L2-b newly conrmed its BBB penetration. L2-b (ca. 250 ng g À1 ) is observed to be available in the brain when administered by oral gavage (10 mg kg À1 ) to the mice (Table  S1 †). Thirdly, the metabolic stability of L2-b for in vivo applications was also veried utilizing human liver microsomes. Susceptibility of L2-b to metabolism is between 30 min and 120 min indicating that this compound has moderate metabolic stability, suggesting its suitability for use in vivo. Lastly, L2-b acts as an antioxidant as well as an inhibitor of Cu(I/II)or Cu(I/II)-Ab-induced ROS production as presented in previous studies. 27,29 From our newly performed study using the Trolox equivalent antioxidant capacity in a cellular environment (i.e., murine neuroblastoma Neuro-2a (N2a) cell lysates), 30 L2-b exhibits a greater free radical scavenging capacity (2.3 AE 0.2) than Trolox (1.0 AE 0.1), a known antioxidant vitamin E analogue. Therefore, L2-b is clearly demonstrated to be viable for in vivo use as a chemical tool for exploring the association of metal-Ab complexes with AD pathogenesis.

Specic modulation of metal-induced over metal-free Ab aggregation pathways in vitro
To elucidate whether L2-b could redirect metal-Ab aggregation into off-pathway amorphous Ab aggregates, suggested to be less toxic or nontoxic, 31 while leaving metal-free Ab cases unaffected, inhibition ( Fig. 1b) and disaggregation (Fig. S1a †) experiments 30 were performed employing Ab 40 and Ab 42 , the two main Ab forms found in the AD-affected brain. The inuence of L2-b on both metal-free and metal-mediated Ab aggregation was monitored at short and long incubation time points. 32 Gel electrophoresis and Western blotting (gel/Western blot, utilizing an anti-Ab antibody, 6E10) 30 were conducted to determine the molecular weight (MW) distribution of the resulting Ab aggregates. Dot blot analysis with an anti-Ab oligomer antibody A11 33 and an anti-Ab bril antibody OC, 34 along with 6E10, was carried out to identify the type of Ab species produced. Moreover, transmission electron microscopy (TEM) images were taken to visualize the morphologies of the resultant Ab aggregates. 30 Both the inhibition and disaggregation experiments indicate that L2-b does not modulate the aggregation pathways of both Ab 40 and Ab 42 under metal-free conditions aer either short or long incubation periods. Nearly identical MW distributions of the Ab species in the absence and presence of L2-b were observed in the gel/Western blots (Fig. 1c, S1b, and S2a †). The dot blots of the inhibition samples indicated A11 (oligomer)and OC (bril)-positive aggregates for metal-free Ab 40 /Ab 42 even when treated with L2-b ( Fig. 1d and S2b †). TEM images revealed that Ab brils were mainly present in both the inhibition and disaggregation experiments of metal-free Ab 40 /Ab 42 with and without L2-b aer 24 h of incubation (Fig. 1e, S1c, and S2c †).
Thus, metal-free Ab aggregation is not noticeably inuenced upon treatment with L2-b.
In contrast to the metal-free conditions, signicantly noticeable changes in the metal [Cu(II) or Zn(II)]-induced Ab 40 and Ab 42 aggregation pathways by L2-b were observed compared to L2-b-untreated analogues. In both the inhibition and disaggregation experiments, aer 24 h of incubation of the Cu(II)-Ab species with L2-b, the resulting peptide species with a wide range of MWs were visualized by gel/Western blot (Fig. 1c, S1b, and S2a †). In the inhibition studies of both Ab 40 and Ab 42 , as well as in the disaggregation experiment of Ab 42 , Cu(II)-Ab samples treated with L2-b even for 4 h also exhibited the distinct MW distribution of Ab (Fig. 1c, S1b and S2a †). Distinguishably, L2-b was capable of limiting the formation of A11-and OCpositive Cu(II)-induced Ab 40 /Ab 42 aggregates at both short and longer incubation times ( Fig. 1d and S2b †). Morphologies of L2b-incubated Cu(II)-Ab, analyzed by TEM, displayed both narrower and shorter brils, as well as unstructured Ab aggregates in the inhibition experiments ( Fig. 1e and S2c †); while less dense, thinner brils were mainly observed in the disaggregation experiments (Fig. S1c †). In the case of Zn(II)-Ab, L2-b could also transform the aggregation pathways (Fig. 1c, S1b, and S2a †). The TEM studies revealed L2-b-triggered, smaller amorphous Zn(II)-Ab aggregates in both the inhibition and disaggregation experiments (Fig. 1e, S1c, and S2c †). Overall, L2-b is observed to redirect metal-Ab aggregation mainly into unstructured Ab aggregates that are generated via the offpathway aggregation and are known to be less toxic or nontoxic. 31 Thus, L2-b could be used as a chemical tool specic for such anti-amyloidogenic activity toward metal-Ab complexes over metal-free Ab in this manner.

Formation of structurally-compact complexes with metal-Ab not metal-free Ab in vitro
In order to explore the specic interaction of L2-b with metal-Ab over metal-free Ab, nanoelectrospray ionization-MS (nESI-MS) studies were employed (Fig. 2a). When metal-free Ab 40 was allowed to react with L2-b, no binding events were observed, even with a six fold excess of the ligand (Fig. 2a(ii)). In comparison, incubating a comparatively smaller concentration of L2-b with Ab 40 and Cu(II) promoted readily observed levels of complexes containing Ab 40 , Cu(II), and L2-b approximately in the ratio 1 : 2 : 1, supporting the metal specic nature of the interaction (Fig. 2a(iv)). The formation of a ternary complex between L2-b and Cu(II)-Ab 40 is supported by the previously reported NMR studies of L2-b with Zn(II)-Ab 40 in solution. 27 Additionally, another MS signal was observed. This signal corresponds to an intact molecular mass of 89.24 Da less than the full-length Ab 40 peptide in good agreement with ternary Ab 40 -Cu(II)-L2-b complex formation (gray, Fig. 2a(iv)). Tandem MS data (Fig. S3 †) and subsequent analysis of the fragment ions indicate that this new signal corresponds to a chemical modi-cation within the rst ve residues of Ab 40 (D 1 A 2 E 3 F 4 R 5 ). Given a mass measurement error of AE1 Da, and supporting L2-b binding experiments performed using an Ab 40 F4A sequence variant, as well as acetylated analogs (Fig. S4 †), we can eliminate alterations to F4 as a source of the product observed and show that free primary amines are critical for binding and subsequent Ab degradation. While no direct observations of the Cu(II)-L2-b-bound Ab 42 form were indicated by MS, the 89.24 Da mass loss product was detected (gray, Fig. 2a(v)) upon addition of both L2-b and Cu(II) to the samples, implying the generation of a transient ternary Ab 42 -Cu(II)-L2-b complex of unknown stoichiometry. These Ab 40 /Ab 42 fragmentation results also suggest that, as expected, Cu(II) likely binds to Ab proximal to the site of L2-b attachment. 4,7 In all cases, neither L2-b nor Cu(II) was detected in complex with the identied Ab degradation product. Detailed structures of these ternary complexes will be the subject of future studies.
To study the molecular level structural dynamics by which L2-b redirects metal-Ab aggregation pathways, IM-MS experiments of the complexes produced were performed. A comparison of the arrival time distributions of the metal-free Ab 40 form with the different ligated states supports an increasing level of structural compaction as additional components (i.e., Cu(II) and L2-b) associate with Ab 40 (Fig. 2b and Table S2 †). Analyzing the arrival time distributions for all complex states presented, along with the nESI-MS data, our IM-MS investigations demonstrate that L2-b is capable of specically interacting with Cu(II)-bound Ab over metal-free Ab, subsequently promoting a high level of structural compaction of the complex. This binding of L2-b to metal-Ab with increased structural compactness could be a key property for the distinguishable reorganization of metal-Ab aggregation pathways, similar to the previously suggested molecular level mode of action of EGCG toward metal-Ab  complexes which promotes the generation of nontoxic unstructured aggregates via off-pathway aggregation. 24,31 Targeting and reacting with metal-Ab complexes in living cells and in the brain of 5XFAD AD mice The effect of L2-b on metal-Ab 40 /Ab 42 -induced toxicity was rst examined using N2a cells, as an indication of its interaction with metal-Ab complexes. An increase (ca. 10-20%) in cell viability for both Ab 40 and Ab 42 was displayed upon treatment of cells incubated with Cu(II) or Zn(II), Ab, and L2-b (10 mM each; Fig. S5 †). Moving forward, the ability of L2-b to penetrate the BBB and interact with metal-Ab species in the brain was veried in the 5XFAD AD mouse model. Zn(II) found in Ab plaques was visualized in the brain tissue slices by a uorophore specic for Zn(II), 6-methoxy-(8-p-toluenesulfonamido)quinolone (TSQ; Fig. 3). 35 Administration of L2-b to 5XFAD AD mice intraperitoneally for three weeks on a daily basis starting at the age of three months resulted in drastically diminished uorescence of TSQ in the plaques (arrows shown in Fig. 3c). Additionally, in the hippocampal mossy ber terminals, a Zn(II)-rich region in the brain, 36 there was no difference in the uorescence of TSQ between wild type and 5XFAD AD mice treated daily with the vehicle, whereas L2-b reduced uorescence by ca. 13% (P < 0.05) in 5XFAD AD mice over the same time span (Fig. 3d). Thus, these in vivo studies suggest that L2-b is BBB permeable and can enter the brain to interact with intracerebral metals, including those found in Ab plaques.

Reduction of amyloid pathology in 5XFAD AD mice
To identify the direct involvement of metal-Ab complexes in amyloid pathology leading to improved cognition, the 5XFAD mouse model of AD 26 was chosen. L2-b (1 mg kg À1 ) was injected into nontransgenic littermates (wild type) and 5XFAD AD mice via the intraperitoneal route for three weeks on a daily basis starting at the age of three months. All mice survived the consecutive treatments, which rarely caused changes in body weight (Table S3 †). Necropsy of all major organs in L2-b-treated mice revealed no gross changes. Fig. 4 Effect of daily treatments with L2-b for three weeks on the amyloid deposits in the brains of 5XFAD male mice. Representative microscopic images of (a and b) 4G8-immunostained or (d and e) Congo red stained brain sections of 5XFAD mice, which were given daily (a and d) the vehicle or (b and e) L2-b (1 mg per kg per day) via intraperitoneal injection for three weeks starting at three months of age (magnification ¼ 40Â; scale bar ¼ 100 mm). Inset in (d) and (e): enlarged micrographs of congophilic amyloid plaques in the cortical area (magnification, 400Â; hip, hippocampus; ctx, cortex). To evaluate the amyloid pathology of the vehicle (black bars; n ¼ 5)-or L2-b (gray bars; n ¼ 7)-treated male 5XFAD mice, (c) the load of 4G8-immunoreactive amyloid deposits and (f) the number of congophilic amyloid plaques in the cortex were measured in five brain sections taken from each animal. *P < 0.05 by one-way ANOVA.  5 Levels of Ab in whole brain tissues of three-month-old male 5XFAD mice. The amounts of (a) total Ab 40 , (b) total Ab 42 , (c) PBSsoluble Ab oligomers, and (d) Ab fibrils were assessed using ELISA after three weeks of treatment with vehicle (black bars; n ¼ 5) or L2-b (1 mg per kg per day; gray bars; n ¼ 7). Bars denote the levels of Ab, which were calculated from three independent experiments and expressed as values per gram of tissue. *P < 0.05 or **P < 0.01 by one-way ANOVA. (e) 4-20% (lower panels) and 16.5% (upper panels) trisglycine gel/Western blot analyses were performed to visualize the Ab monomers and aggregates, respectively, in the brain tissue lysates of wild type (WT; left panels) and 5XFAD male mice (right panels).
The potential association of L2-b with amyloid pathology was investigated by rst observing the amyloid plaque load in the brain tissue of 5XFAD AD mice. When the brain tissue slices of L2-b-administered 5XFAD AD mice were stained with an APP/ Ab-specic antibody (4G8) or a compact core amyloid plaque indicator (Congo red), it was found that the amyloid plaque burden was ameliorated (Fig. 4). Reduction (ca. 15%) of both the area of 4G8-immunoreactive deposits and the number of congophilic amyloid plaques was revealed in the cortex of L2b-treated 5XFAD AD mice when compared to vehicle-treated 5XFAD AD mice (Fig. 4c and f). The changes in the amount of both Ab 40 and Ab 42 in the brain tissues of 5XFAD AD mice following L2-b administration were also assessed. Total amounts of Ab peptides were analyzed by an enzyme-linked immunosorbent assay (ELISA) in sodium dodecyl sulfate (SDS)and formic acid (FA)-soluble brain tissue lysates ( Fig. 5a and b and S6 †), as well as oligomeric and brillar Ab aggregates in the phosphate buffered saline (PBS)-soluble fraction ( Fig. 5c and  d). 37 Relative to vehicle-treated 5XFAD mice, the L2-b-treated 5XFAD mice showed diminished cerebral levels of both Ab 40 and Ab 42 in all fractions (ca. 15-20%, P < 0.05, Fig. 5a and b). Oligomeric and brillar Ab species in the PBS fraction were additionally abated by 27% and 15%, respectively (P < 0.05, Fig. 5c and d). Similarly, the overall reduction of Ab species was also indicated by gel/Western blot, where Ab monomers and oligomers were noticeably decreased in brain tissue lysates from L2-b-treated 5XFAD AD mice (Fig. 5e). Together, these studies demonstrate that daily administration of L2-b to the AD model mitigates amyloid pathology in AD, including the load of amyloid plaque deposits and the levels of a wide range of conformations from monomers to brils.

Cognitive improvement in 5XFAD AD mice
Investigation of behavioral performance was carried out by administering L2-b to 5XFAD AD mice which suffer from decits in learning and memory capabilities as amyloid pathology progresses. 26 The Morris water maze was used to evaluate different aspects of spatial learning and memory in threemonth-old 5XFAD AD mice. 26 The wild type mice, which were consecutively injected with vehicle during the experimental period, normally took shorter times upon repetition of the training trial to nd the escape platform, located in the northwest (NW) quadrant ( Fig. 6a and b). In contrast, vehicle-treated 5XFAD AD mice spent longer times searching for and reaching the platform indicating they had difficulties with learning and memory (Fig. 6). Administration of L2-b to 5XFAD AD mice led to a remarkable improvement in the performance of the task. L2-b-treated 5XFAD AD mice were capable of nding the target in a comparable time to the wild type mice displaying signicantly better memory and learning abilities than their untreated 5XFAD AD littermates (P < 0.05, Fig. 6a and b). Additionally, L2b-treated 5XFAD AD mice took a more direct and easier path than the vehicle-treated 5XFAD AD mice to search for the platform (P < 0.05, Fig. 6c). Therefore, L2-b, a chemical reagent specic for metal-Ab, ameliorates cognitive defects in the AD mouse model, along with the attenuation of amyloid pathology. Fig. 6 Learning and memory abilities of three-month-old male wild type (WT) and 5XFAD male mice treated with vehicle (black and white bars) and L2-b (gray), measured using the Morris water maze task. (a) The escape latency time was counted every day during the period of the 21 st -25 th daily treatments of either vehicle or L2-b and the probe trials were performed on the day of the final 25 th treatment to measure (b) how quickly the mice reach and (c) how long they spend in the target quadrant (NW, highlighted in gray; circles show images of the representative tracks of the mice in the water maze). *P < 0.05 or **P < 0.01 by one-way ANOVA (n ¼ 6, 13, and 14 for vehicle-treated WT and vehicle-/L2-b-treated 5XFAD mice, respectively).
These overall in vivo observations and results indicate that metal-Ab complexes could be directly linked to AD pathogenesis.

Conclusions
In summary, for the rst time, experimental evidence affirms that metal-Ab complexes can be directly associated with AD pathogenesis, by applying the rst in vivo chemical tool which specically targets metal-Ab complexes and ameliorates metal-Ab reactivity (i.e., metal-Ab aggregation, formation of toxic oligomers, and ROS production) in 5XFAD AD mice. Our ndings presented herein demonstrate the feasibility of developing small molecules as in vivo chemical tools for studying metal-Ab. In addition, our studies indicate that research efforts toward understanding metal-Ab-induced pathological pathways and identifying interrelated partners with metal-Ab in AD onset and progression at the molecular level should continue to be made. The current and future outcomes, obtained from metal-Abinvolved AD research, can open new directions for our longterm goal, the discovery of effective drugs for this fatal neurological disorder.