Role of Metal Cations of Copper, Iron, and Aluminum and Multifunctional Ligands in Alzheimer’s Disease: Experimental and Computational Insights

Abstract

Alzheimer’s disease (AD) is the most common form of dementia, affecting millions of people around the world. Even though the causes of AD are not completely understood due to its multifactorial nature, some neuropathological hallmarks of its development have been related to the high concentration of some metal cations. These roles include the participation of these metal cations in the production of reactive oxygen species, which have been involved in neuronal damage. In order to avoid the increment in the oxidative stress, multifunctional ligands used to coordinate these metal cations have been proposed as a possible treatment to AD. In this review, we present the recent advances in experimental and computational works aiming to understand the role of two redox active and essential transition-metal cations (Cu and Fe) and one nonbiological metal (Al) and the recent proposals on the development of multifunctional ligands to stop or revert the damaging effects promoted by these metal cations.

1. Introduction

Nearly a century has passed since Alzheimer’s disease (AD) has been recognized as a type of dementia. Since then, the incidence of the disease has been increasing and has become the most common form of dementia nowadays, covering from 60% to 70% of all global cases.¹ In its 2022 report the Alzheimer’s Association presented an evaluation of the impact of this disease in the United States, showing that, while cardiovascular diseases or stroke-associated deaths decreased from 2000 to 2019, the reported deaths due to AD increased by 145%. In 2019 121,499 deaths related to AD were registered, which places it as the fifth leading cause of death for adults aged 65 years and older. On the other hand, the estimated economic impact on 2022 was about $321 billion USD meant for the treatment and care of this dementia. This amount is expected to increase to about $1.0 trillion USD by 2050.²

Only 5% of reported AD cases are caused by familial-inherited genetic mutations, generally found on genes codifying for the amyloid precursor protein (APP), presenilin-1 (PS1), presenilin-2 (PS2), and apolipoprotein E (APOE); it is identified as familial AD.³ Remaining cases, which account for about 95% of the total, occur sporadically, with the older adults being the highest risk population. It is worth mentioning that AD’s related massive neuronal damage, as well as other of its pathological characteristics, do not take part in natural aging.

Even though different drugs have been tested to stop the progression of the disease, currently there is not an effective treatment available. On a review of 413 clinical trials conducted between 2002 and 2012 it was found that there was only success in 0.4% of them,⁴ showing that the probability of failure of the potential pharmacological agents in the testing phase is very high. In this way, an even greater economic and investigative effort is necessary.

AD is a multifactorial disease; the risk factors are both genetic and environmental, and it is still not possible to point out among its characteristics which is a cause or consequence, giving rise to many hypotheses looking forward to clarify the root cause of the disease but without certainty about which of them is the most determinant. A variety of diagnostic imaging methods and therapeutic strategies for many AD targets were reviewed by Ramesh et al.⁵

The pathological hallmarks involved in AD are extensive, but among the most recognized and studied are those related to the aberrant processing of brain proteins and its consequences. However, throughout the years the research field has been evolving to the design of pharmacological agents able to tackle multiple targets of the disease. In this sense, many complementary hypotheses to protein misfolding have arisen around it as alternatives to give a better understanding of AD and as valuable options to elucidate new biological targets.^6,7 Herein, we discuss two of them: metallic ion dysregulation and oxidative stress, from a beta amyloid peptide (Aβ) point of view. This review will focus on two essential metal cations (copper and iron) and one exogenous metal (aluminum) and their role in the increase of oxidative stress related to AD from experimental and computational means and in the multifunctional ligands that have been proposed in order to stop some causes of the disease. Although zinc has also been found in senile plaques its activity is more related with the promotion of peptide aggregation and not in the increase of oxidative stress.⁸

2. Amyloid Cascade Hypothesis

The amyloidoses are disorders characterized by an extracellular accumulation of proteins in various tissues, developing insoluble fibrillar structures that adopt mostly parallel β-sheet conformations. Examples of proteins involved in amyloidosis are the immunoglobulins in primary systemic amyloidosis, protein fragments of amyloid A in secondary systemic amyloidosis, fragments of apolipoprotein A-1 in familial amyloid polyneuropathy, and amylin in the pancreas on diabetes’ patients.⁹

One of the most notable features found in the brains of AD patients is the accumulation of deposits of senile plaques. These are composed by the Aβ peptide, whose most common sequences consist of 40 to 42 amino acids and contain a hydrophobic fragment (G29-V40 or A42) that reduces its solubility, inducing its aggregation, as well as a fragment with high affinity to transition-metal cations (D1-K16).¹⁰ The Aβ derives from the amyloid precursor protein (APP), an integral membrane protein that may be degraded by two protease-mediated mechanisms (Figure 1).

Figure 1.

Degradation pathways for the APP. The amyloidogenic route produces the Aβ peptide.

The first step in the process is an ectodomain cleavage of the protein. The pathway known as nonamyloidogenic occurs through α-secretase and generates an APPsα fragment, while the amyloidogenic pathway begins with β-secretase and generates an APPsβ fragment. The second step is executed by γ-secretase, which proceeds to cut through the transmembrane domain of the APP. This releases the peptide p3 (not amyloidogenic pathway) or Aβ (amyloidogenic pathway) toward the cerebral medium.¹¹ The p3 peptide is harmless, but Aβ monomers aggregate to form oligomers, fibrils, and finally, senile plaques. Among these structures, the oligomeric forms have been pointed out to be the most toxic, even more than the senile plaques featured on advanced stages of the disease, since this conformation may be responsible for a detriment in the neuronal synapse.^12,13 Interestingly, this is contrary to the beneficial effects found for Aβ monomers, including neuronal growth, synaptic function modulation, and protection against oxidative stress, toxins, and pathogens.¹⁴ However, when the monomeric form interacts with transition-metal cations, the ions are able to promote the catalytic production of hydrogen peroxide promoting an increase in the oxidative stress.¹⁵

Although the specific role of the APP is not clear, a study with APP knockout mice has shown that there is an increase of copper concentration in the cerebral cortex relative to normal mice.¹⁶ This suggests that there could be a copper regulation function performed by the APP on the brain. A feature article by Rajasekhar and co-workers covers various aspects of the Aβ peptide in depth; these include the aggregation processes, beneficial behaviors, toxicity, and a variety of oriented therapies.¹⁷

The main idea of the amyloid cascade hypothesis holds that the senile plaques’ accumulation is responsible for triggering the disease. This hypothesis has been the most recognized and studied since it was first suggested in 1992, due to the clear example seen in familial AD, where genetic mutations are associated in one way or another with the APP. In accordance with this, drug development has been directed toward the inhibition of the Aβ production through inhibition of β-secretase, its decrease in the brain medium, and its deposition mechanisms. Unfortunately, the list of drugs focusing on this hypothesis that have failed over more than 25 years of research is extensive, and many of these have resulted in cognitive deterioration and/or a null efficiency.¹⁸ Immunological monoclonal antibody-type drugs, such as solanezumab, crenezumab, and aducanumab, that are focused on disaggregation or inhibition of the production of Aβ, did not produce substantial improvement in any of the clinical stages of AD, i.e., early, prodromal, or established.^19,20

On the other hand, an autopsy performed on 59 individuals from advanced age and normal cognitive conditions revealed that the accumulation of Aβ deposits can also be found in healthy individuals. In fact, several examinees would qualify as AD patients under criteria based on the deposition of plaques of Aβ and neurofibrillary tangles.²¹ Furthermore, only 10 subjects presented no significant degenerative brain damage. In addition to this, a low correlation has been found between the amount of senile plaques formed, the degree of cognitive impairment, and both neuronal and synaptic loss.²² This could suggest that Aβ depositions may not be responsible for triggering the AD but rather a consequence of an underlying cause or even a compensatory response that arises to lessen the consequences of other homeostatic dysregulations, such as those presented in the following sections.

3. Oxidative Stress in AD

Oxidative stress occurs when a high concentration of reactive oxygen species (ROS) exceeds the biological load that antioxidant species can handle, triggering not only neurodegenerative diseases such as Parkinson’s disease, Huntington’s disease, and AD but also psychiatric disorders such as schizophrenia, anxiety, depression, and bipolar disorder.²³ This is, in part, because the brain is especially vulnerable to oxidative stress due to its high contents of oxidizable unsaturated fats and its high oxygen consumption rate. ROS can arise naturally in the mitochondrial metabolism, through an electron leak to oxygen in the electron transport chain and subsequent superoxide anion (O₂·^–) generation, which is known to be the precursor of many other ROS.^24,25 The control of oxidative stress in biological systems is of high importance. Even though at low concentrations the ROS participate in signaling processes,^26,27 it has been shown that plants increase their ROS under stressful conditions and heavy metal-associated toxicity, inducing cell damage that must be controlled by antioxidant mechanisms.^28,29 In the same way, the human body possesses a variety of defense mechanisms against ROS,³⁰ but when these species get out of control they can cause severe damage to important biomolecules like lipids, proteins, and DNA by several mechanisms.³¹

There is sufficient evidence of the role of transition-metal cations in ROS production and the increment in the oxidative stress.^32,33 For example, free iron and copper cations can produce ROS through Fenton and Haber-Weiss reactions³⁴⁻³⁷ and also decrease the available biological amounts of ascorbate.^38,39 Additionally, intracellular iron can stimulate calcium signaling pathways, resulting in the activation of kinases involved in synaptic plasticity through ROS mechanisms. In healthy brains this is a beneficial interaction since iron-mediated ROS are generated at low levels and improve the already present calcium signaling at the synapse; nevertheless, when pathological amounts of iron are present, it can cause an accumulation of calcium signals within the cell leading to mitochondrial dysfunction. Likewise, high levels of calcium in the cell increase the redox active iron in the medium, which creates a toxic cycle for the cell and could be an important neurodegeneration pathway related to AD.⁴⁰ Calcium itself is also capable of generating oxidative stress within the mitochondria, increasing its membrane permeability and releasing pro-apoptotic factors to the cytoplasm, such as cytochrome C and the apoptosis-inducing factor, causing neuronal death.⁴¹

Regarding copper, its complexes with Aβ can catalyze the activation of O₂ into the superoxide anion,^42,43 and computational studies for different Cu-Aβ complexes have shown that their respective reduction potentials are below the value for the O₂/H₂O₂ pair under physiological conditions (0.295 V⁴⁴), favoring its catalytic reduction. According to this mechanism, the first step is the reduction of the copper complex by a natural reducing agent. Then the Cu¹⁺ complex reduces O₂ to O₂·^–, and finally a second Cu¹⁺ complex reduces O₂·^– to H₂O₂ (Figure 2).¹⁵

Figure 2.

Proposed mechanism for the oxidative stress mediated by metal cations and Aβ peptide.

This shows the potential oxidative stress that it can provoke; in fact, it has been established that the Cu-Aβ complex has a higher efficiency in catalyzing the formation of H₂O₂ than the respective complex with iron,⁴⁵ which emphasizes the high redox activity of copper. Between the two physiological copper species Cu¹⁺ and Cu²⁺, the reduced specie is considered the most toxic state of copper, and then, the inhibition of its redox activity in the free state is also of research interest.⁴⁶ All this activity suggests a considerable oxidative brain damage due to copper, which is why there is a special need to develop chelating agents for copper as potential pharmacological agents in the early stages of AD.

Other in vitro studies indicate that mutations associated with familial AD on genes encoding the APP, PS1, and PS2 proteins involve a considerable increase of ROS that stimulate different apoptotic pathways with a wide variety of caspases and JNK kinases;^47,48 this oxidative stress can be controlled by using antioxidants such as α-tocopherol or vitamin E, which results in an inhibition of the apoptosis.⁴⁹ In vivo studies with mice also highlight the presence of ROS in the development of AD associated with these genes, which, in addition to inducing apoptosis, causes synaptosomal protein oxidation and lipid peroxidation.⁵⁰⁻⁵² Brain samples from familial AD patients also show a high oxidative stress in conjunction with higher oxidative deterioration of the temporal cortex in contrast to sporadic AD patients.⁵³ However, oxidative stress is not only present in familial AD, as it is well-known that the most associated risk factors for sporadic AD are correlated with oxidative stress, either by ROS production, or by a decrease of endogenous antioxidants: aging, diabetes, traumatic brain injury, high cholesterol levels, hypertension, and stroke,⁵⁴⁻⁵⁷ whereas antioxidant intake is a protective factor.⁵⁸⁻⁶¹

The oxidative stress hypothesis states that a dysregulation of ROS is responsible for triggering the disease. Evidence supporting this hypothesis is the fact that the effects of oxidative stress precede the formation of Aβ deposits. Post-mortem studies revealed that oxidative damage of nucleic acids and proteins in the brain is worse in early stages of AD and decreases as Aβ plaque deposition progresses.⁶² The same trend is observed in oxidative damage of nucleic acids in cerebrospinal fluid.⁶³ It was also found that RNA oxidation is prominent in the brain regions affected by AD, while other regions remain with no substantial RNA oxidation increase.⁶⁴ An examination of familial AD patients’ brains revealed that the oxidative stress is greater on the frontal cortices of brains exhibiting less amount of Aβ deposition.⁶⁵ The idea is supported by studies in mice, where lipid and protein oxidation and the decrease of antioxidant activity are more pronounced in cases with less fibrillar deposition of the peptide.⁶⁶ In fact, it has been proven that the Aβ peptide at low physiological concentrations has an antioxidant effect, which protects lipoproteins from oxidation in cerebrospinal fluid,⁶⁷ and that Aβ monomers can inhibit neuronal death caused by oxidative damage of Fe²⁺, Fe³⁺, and Cu²⁺ species by means of coordination and subsequent disruption of their redox activity with O₂.⁶⁸ This strengthens the idea that plaque formation is a biological response, rather than a cause. According to this evidence, it is suggested that early oxidative stress in AD has effects such as mitochondrial dysfunction, metallic dysregulation, and metabolic imbalances, typical characteristics of early stages of AD.⁶⁹

Along with the formation of Aβ deposits, the high levels of metallic ion concentration, and oxidative stress incidence, an additional neuropathologic mark is the neurotransmitter alteration in the absence of acetylcholine due to hydrolysis reactions produced by the acetylcholinesterase enzyme (AChE). The Food and Drug Administration of the United States (FDA) approved four drugs for the treatment of AD; all of them are inhibitors of AChE (i.e., tacrine, donepezil, rivastigmine, and galanthamine) capable of regulating the neurotransmitters for a period from one to two years in moderate cases of the disease. However, after this period it is not possible to change the course of the disease, and still there is no definitive cure.⁷⁰ All these drugs present serious side effects such as vomiting and diarrhea; for this reason, the use of tacrine was discontinued in the United States since 2013. The scarce therapeutic benefit observed with these treatments produces serial doubts about the cost-benefit relation of these drug treatments. The only drug approved by the FDA that is not an AChE inhibitor is memantine; its form is fused with donepezil (namzaric), and it is capable of improving intracellular regulation of calcium ions.⁷¹ Unfortunately, the therapeutic effects of drugs last between 6 and 12 months; after this period, the AD continues its progress.

4. Metal Ions in AD

Metal ions have significant functions in biological systems such as catalysts, electron-transfer reactions, or activation and transport of small molecules. Numerous body proteins and enzymes have metals in their structure as prosthetic groups, which operate as active sites and help them to accomplish their biological functions. The properties of metal ions in biological environments are mainly caused by the geometry of the metal complex due to the nature of the ligands coordinated to the metal center and the metal complex coordination environment. Evidence supports that the presence of unregulated metals can cause severe health conditions. Abnormal distributions of certain metals in the brain have been attributed in the diagnosis of diverse diseases of the central nervous system, such as AD, Parkinson’s disease, dementia with Lewy bodies, bipolar disorders, and depression.³⁹

Several investigations have been focused on the biological regulations of copper, iron, and zinc, and there is sufficient in vitro and in vivo evidence linking them to the development of various diseases, including AD.⁷²⁻⁷⁴ In addition, there is strong evidence suggesting that neurochemical reactions other than Aβ production must contribute to the development of AD.⁷⁵ For instance, through a post-mortem study, the concentration of these metal ions in the amygdalas of AD patients was found to be higher compared to that of control subjects (Cu²⁺: 5.7, Fe³⁺: 2.8, and Zn²⁺: 3.1 times higher).⁷⁶ The coordination spheres of these metal ions in their respective Aβ complexes and their affinities for the peptides and aggregates have also been reviewed by del Barrio et al.⁷⁷ Here we present some important characteristics of these metals whose action in the brain has been found to be related to AD development and the computational works aimed to determine their structures and molecular properties.

4.1. Copper

This metal can be found in metalloenzymes such as superoxide dismutase (SOD), cytochrome c oxidase, ascorbate oxidase, and ceruloplasmin. Its controlled release from synaptic vesicles is an integral part of neurotransmission^78,79 and accomplishes important functions related to neuropeptide synthesis and the proper functioning of the immune system,⁸⁰ so it is essential to maintain its consumption in traces. Among its main transporters are ATP7A, ATP7B, Atox1, and the copper transporter protein (CTR1), for transport through the cell membrane, whereas ceruloplasmin helps to transport throughout the body. A direct correlation has been found between the location of CTR1 and Atox transporters and the copper levels in different regions of the brain.⁸¹ Copper is found in higher concentrations in regions like substantia nigra, locus coeruleus, and hippocampus, areas that are intimately related to memory. The locus coeruleus is a brainstem structure that performs functions related to sleep, attention, learning, stress, and memory; it is recognized as the main source of norepinephrine (neurotransmitter produced by the copper-dependent enzyme dopamine-β-hydroxylase from dopamine); therefore, it contains a higher concentration of copper than other brain regions, demonstrating the importance of this metal in the circadian cycle.⁸²

Although the biological amounts of copper are lower than those of iron and zinc, it has been found that very low levels of intracellular free copper, on the order of 1 × 10^–18 M, can cause oxidative damage.⁸³ The toxicity of Cu²⁺ in the brain is attributed to an inhibition of the interaction between nerve growth factor and ubiquitin,⁸⁴ to a decrease in the available levels of glutathione,^85,86 and to the induced formation of oligomers from already formed fibrils, whose interactions with the cell membrane increase its permeability.⁸⁷

Aβ could form a metallopeptide in combination with Cu²⁺ and Zn²⁺ (and possibly Fe³⁺) ions, which are intrinsically present in the brain, mediating peptide toxicity (through the production of radicals and hydrogen peroxide) and peptide aggregation.⁷⁵ Studies in mice show that consumption of trace amounts of copper from water generates senile plaques and considerable cognitive impairment.⁸⁸ In AD patients, it has been shown that there is a 52.8–70.2% decrease in total brain copper levels in regions such as the hippocampus and the entorhinal, motor, and sensory cortexes, presenting severe damage.⁸⁹ This decrease takes similar values to those found in Menkes’ disease, where copper deficiency is recognized as the main cause. This evidence is in agreement with the regulatory effect of copper over the APP, since a high intracellular concentration of this metal promotes the nonamyloidogenic degradation, while a low concentration promotes the amyloidogenic pathway, generating senile plaques,^90,91 suggesting that intraneuronal copper deficiency is key in the pathogenesis of AD.⁹²

Computational Modeling of Cu-Aβ Systems

The computational treatment of copper in the context of AD has been mainly focused on the structural determination of its complexes with Aβ peptide and its participation in the production of reactive oxygen species. However, the study of Cu-Aβ complexes has been hampered by the fact that there are no reports of Cu-Aβ structures obtained by crystal or NMR studies, to the best of our knowledge. In this context, computational methods are a valuable tool to elucidate the structure and reactivity of these complexes. Nevertheless, the modeling of these complexes is challenging due to the open-shell nature of the Cu²⁺ systems and the high number of coordinating atoms from Aβ peptide. In addition, relevant properties to understand the participation of these complexes in ROS production (e.g., the standard reduction potential) depend on subtle electronic and solvent effects that could be incorrectly described by some computational methodologies.

In this sense, the initial Cu-Aβ models were built considering copper cations and its coordinating ligands, since a complete description of Cu-Aβ systems is computationally demanding.^93,94 Besides, to understand the geometrical changes that occur during the reduction or oxidation of the metal center, molecular dynamics (MD) simulations are appropriate due to the considerable size that these kinds of systems have. Nevertheless, due to the electronic effects that occur in this process and in the reaction with oxygen, classical molecular dynamics simulations are not adequate, as long as bond-breaking and bond-formation processes cannot be correctly modeled by these means. Instead, ab initio molecular dynamics (AIMD) simulations should be considered even if they are computationally demanding. A recent review by Strodel and Coskuner-Weber summarizes the computational studies on transition metals and Aβ.⁹⁵ In this work, we are going to highlight the computational studies intended to elucidate the structure and redox properties of copper, iron, and aluminum complexes with Aβ peptide.

Model Systems

Experimental studies on the determination of the Cu-Aβ structure have been performed by using electron paramagnetic resonance (EPR), NMR, and other techniques. The main conclusion of these experiments is that copper coordination to Aβ is highly dependent on the surrounding pH. Thus, two main components have been considered as plausible for the coordination of copper to Aβ peptide (Figure 3).^96,97

Figure 3.

Coordination spheres for Cu-Aβ complexes at low (6.4) and high (>7.4) pH.

Cu-Aβ model systems that consider copper cation and its coordination sphere have been used to study copper preferences in the Aβ sequence and to propose plausible mechanisms to the reactions that increase oxidative stress.^98,99 Recently, Bertini et al. have used density functional theory (DFT) methods to study the mechanism of the O₂ reduction to OH^– + ·OH radical mediated by Cu-Aβ model systems and ascorbate.¹⁰⁰

As has been pointed out, the use of small model systems has allowed the understanding of copper coordination and reactivity of Cu-Aβ complexes. However, the use of small models does not allow us to study the effect of the peptide on copper coordination and the role of the peptide chain in the stability of the Cu-Aβ complexes. To consider these effects it is necessary to model the Cu-Aβ complexes by considering a larger fragment of the Aβ peptide.

Cu-Aβ_1–16 Models

Modeling the 1–16 region is important to understand the effect of the peptide chain on the stability of Cu-Aβ complexes. This region is considered since, as stated before, it corresponds to the metal affinity region. Initially, larger fragments of the peptide were considered to build the models, and finally the whole region was considered. Rauk et al. proposed a model by using classical MD to provide insight into the flexibility and most frequent conformations of the peptide chain. However, due to its classical treatment of the copper center it has not considered its coordinating preferences.¹⁰¹ La Penna et al. proposed models considering key fragments of the copper affinity regions of Aβ using first-principles molecular dynamics simulations in a Car–Parrinello scheme for Cu¹⁺ and Cu²⁺ and Aβ complexes.¹⁰² In this work they were able to explain the coordinating preferences of copper to Aβ. Al-Torres et al. also proposed a computational protocol for building Cu-Aβ_1–16 models considering the full copper affinity region by combining homology modeling and DFT calculations.¹⁰³ A schematic representation of this protocol is presented in Scheme 1. The model building starts from the possible metal coordination spheres, and the peptide moiety is included by simulated annealing. Finally, full DFT calculations on the most plausible models are carried out to properly describe the stabilizing interactions.

Scheme 1. Computational Protocol for the Construction of Metal-Peptide Complexes.

Adapted with permission from ref (104). Copyright 2015 AIP Publishing.

The use of these larger models allowed us to understand the impact of the peptide organization on copper coordination preferences. In addition, it allowed us to study the reactivity of these Cu-Aβ complexes. In this sense, Mirats et al. reported superoxide as an intermediate in the water peroxide production confirming the results that Hewitt et al. obtained using model systems.^42,98 The presence of superoxide was later confirmed by experimental means using superoxide dismutase in the reaction media.¹⁰⁵ To explore the conformational and coordination changes on the Cu-Aβ1–16 complexes it was necessary to employ ab initio MD simulations. The results show that the Cu⁺-Aβ1–16 systems change from a square planar coordination sphere to a T-shaped structure, which is suitable for facilitating the production of hydrogen peroxide from molecular oxygen.⁴³

The modeling of these Cu-Aβ1–16 complexes has provided molecular insights to understand the electronic and geometrical phenomena that drive the coordination of copper to Aβ and to propose plausible mechanisms to explain the oxidative damage observed in the development of AD.

4.2. Iron

Iron is the most abundant transition metal in the human body. This ion is present in hemoglobin, myoglobin, cytochromes, ferritin, hemosiderin, and other enzymes.¹⁰⁶ Iron its known to be of great importance in fetal and postfetal neurodevelopment,¹⁰⁷ and as it is part of the heme prosthetic group, it is closely related to diseases such as anemia and diabetes.¹⁰⁸ Its transport to the interior of the neurons is carried out by the transferrin, DMT1, and lactoferrin transporters; its transport to the exterior is facilitated by ferroportin, and to be transported throughout the organism requires the ferritin fragment. Thus, any imbalance of these transporters leads to an erroneous distribution of this metal in the brain. Multiple neuronal disorders can be associated with iron dysregulation, and many of them keep their origin in mutations in different genes, e.g., the C19orf12 gene, whose mutations are related to hereditary spastic paraplegia and pallidopyramidal syndrome.¹⁰⁹ Oral administration of iron in mice showed long-term alterations in biomolecules related to the synapse and mitochondrial function, causing memory loss.¹¹⁰

Increasing iron concentration in regions such as the neocortex is attributed to normal aging, but this is not the case in areas such as the hippocampus and amygdala.^111,112 APP influences the stability of ferroportin in the cell membrane, and therefore it has a regulatory effect that aids iron transport out of the neuron¹¹³ and maintains control of transferrin and ferritin levels, which prevents iron accumulation in the brain.¹¹⁴ It is also known that iron overload can cause low levels of ferroportin and hepcidin, a fact that could be related to the iron accumulation found in this region in patients.^115,116 Additionally, considerable low levels of ferroportin and hepcidin have been found in AD patients’ brain hippocampus, a region highly associated with memory and consciousness.¹¹⁷

The presence of Fe³⁺ ions in the brain environment induces APP production through multiple mechanisms, higher β-secretase activity (through a possible mechanism involving reactive oxygen species¹¹⁸), and lower α-secretase activity (through furin modulation¹¹⁹), with subsequent release of Aβ to the medium. In addition, iron slows down the aggregation process by preventing the orderly arrangement of the peptide, which is thought to increase its toxicity,¹²⁰ and by its deposition along with the plaques, it creates new redox active sites that increase oxidative stress.¹²¹ This increase is supported by computational studies, which show that the complexes formed with Aβ have a reduction potential about 0.5 V lower than that of free iron under physiological conditions, increasing its reducing activity.^122,123 On the other hand, other studies suggest that this mechanism could be beneficial. An in vitro study showed that Pb²⁺ and Pb⁴⁺ intoxication inhibits APP translation in neurons, which produced a toxic accumulation of iron in the cytosol, but when iron was added to the medium, APP production was stimulated and then toxicity decreased. This could mean that Aβ deposition can be interpreted as a compensatory mechanism, as the concentration of copper, iron, and zinc metal ions has been found to increase within senile plaques compared to the surrounding tissue.¹²⁴

Fe²⁺ cation is related to the production of neurofibrillary tangles through activation of CDK5 and GSK-3β kinases.¹²⁵ This cation neurotoxicity led us to think for a while that magnetite (Fe²⁺O·Fe³⁺₂O₃) played an important role in AD. However, this hypothesis was discarded when it was shown that it is, in fact, very stable under biological conditions, and therefore it does not exhibit oxidative catalytic activity with peroxidase substrates.¹²⁶ Furthermore, no significant interaction was found with Aβ peptides in in vitro assays, and it was discarded along with iron oxides and hydroxides.

Computational Modeling of Fe-Aβ systems

Fe-Aβ systems have been less studied by computational means; in part, this is due to the fact that, even when iron is more abundant than copper in the human body, most of the biologically available iron forms the heme group and is not likely to facilitate the redox reactions involving in the development of AD.¹²⁷ Another reason is the poor solubility found in the Fe-Aβ complexes, which has made it difficult to conduct experiments in vitro. This lack of experimental information makes modeling Fe-Aβ systems a challenging task. Nevertheless, model systems have been explored computationally considering the information on the coordination sphere obtained by Raman experiments and the plausible coordinating ligands in an Aβ sequence.¹²⁸

Alí-Torres et al. used a combination of DFT and MP2 calculations to explore the different coordination spheres of Fe-Aβ complexes and to calculate their standard reduction potential, which could explain its participation in oxidative stress.¹²² The computational modeling of Fe-Aβ complexes increases its difficulty with respect to copper by the fact that both reduced and oxidized forms of the complexes are open-shell systems. In addition, three different spin states must be considered for all models, which increments the computational cost significantly. The description of the electronic structure has to be accurate since small changes in the electronic properties could change the calculated properties. In this sense, Orjuela et al. proposed a computational protocol to calculate the standard reduction potential (SRP) of iron complexes, and this protocol was applied to the calculation of the SRP of Fe-Aβ model systems (Figure 4). This work showed that a careful evaluation of density functional and basis sets as well as solvent models and thermodynamics cycles are needed for a proper description of these systems. The inclusion of explicit solvent models has been shown to improve the calculation of the solvation free energy. However, this increases the computational cost considerably.¹²³ According to the calculated SRP values, Fe-Aβ complexes could promote the catalytic production of hydrogen peroxide leading to an increase in the oxidative damage observed in AD brains.

Figure 4.

Fe-Aβ model systems and calculated SRP values. Adapted from ref (123). Copyright 2022 Royal Society of Chemistry.

4.3. Aluminum

Although it is the most abundant metal in earth’s crust, there is no known biological function related to it, as there is no enzyme or protein to make use of this metal.¹²⁹ For this reason, it has no associated transporters or chaperones, and negative effects related to its consumption have been reported, such as kidney dysfunction, anemia, and neurological disorders.¹³⁰

The role of aluminum in dementia began to be suspected in 1965, when after an injection of aluminum salts on mice, neuronal degeneration similar to the neurofibrillary tangles of AD occurred.¹³¹ In the 1970s, it was found that patients with chronic renal failure could develop an aluminum-associated encephalopathy, whose defining symptoms were: speech impairment, epileptic episodes, and movement disorders.¹³² Subsequent studies focused on relating this metal to AD. McDermott analyzed the aluminum content in 19 brain samples by atomic spectroscopy and then concluded that, even though aluminum concentration increases with age, no significant differences are found between AD and healthy brains.¹³³ In a clinical study of patients with this dialysis-associated encephalopathy it was found that there is no correlation between the amount of aluminum ingested and the morphology of AD in the brain. In contrast, the associated dementia disappeared when dialysis fluids were deionized, and aluminum’s role on AD was finally discarded.^134,135 However, recently the pro-oxidant ability of aluminum has been related to the formation of a stable aluminum-superoxide complex.¹³⁶

Computational Modeling of Al-Aβ Systems

Characterizing the coordination preferences of aluminum with ligands of biological interest is difficult due to the larger number of coordinating atoms available in the cellular environment. In this sense, computational chemistry could help by decreasing the number of possibilities and identifying the most stable ligands and coordination modes of this metal. Aluminum has been related to form stable complexes with the Aβ peptide, but there are no experimental reports on its coordination spheres and their role on the production of radical species. Nevertheless, there are computational reports dealing with these two aspects: modeling Al-Aβ complexes and mechanistic studies on the possible role of aluminum complexes in Fenton reactions related to the development of AD and other metal-promoted neurodegenerative disorders.

As aluminum has been used on a regular basis, its interactions with the Aβ peptide have been studied by computational means. The study of this complex pretends to provide more insight into the role of nonbiological metals in the development of AD. Mujika et al. has proposed a computational protocol to build Al-Aβ complexes. This methodology considered the possible coordination spheres of aluminum and explored all the possible binding sites using a preorganized form of the peptide that allowed them to build plausible Al-Aβ models (Scheme 2).¹³⁷ The main difference between this protocol and the one used for the Cu-Aβ model construction is that, in this protocol, there is no need of a metal-Aβ complex as a template.

Scheme 2. Computational Methodology for the Construction and Evaluation of Al-Aβ Models.

Adapted with permission from ref (137). Copyright 2017 Royal Society of Chemistry.

The reactivity of Al-Aβ model systems was also studied by computational methods. Mujika et al. have found that the complex of aluminum with citrate, the main low-molecular-mass chelator biologically available, favors the Fenton reaction by reducing Fe³⁺ to Fe²⁺. This is due in part to the formation of a stable aluminum-citrate-superoxide complex.¹³⁸

In summary, the metal ion hypothesis establishes that maintaining the homeostasis of metal ions at the brain level is key to healthy functioning and that, rather than an accumulation or deficiency of metals, there is an imbalance between the different brain regions¹³⁹ as well as changes in their bioavailability causing alterations at the synapse. The understanding of the role and coordination modes of metal cations relevant to Alzheimer’s disease has inspired the design of molecules able to coordinate these metal cations and avoid the reactions promoting ROS production. Besides the chelating properties, ligands including other multiple properties to target other AD pathological factors have been designed in the last years. Some examples of these multifunctional ligands will be presented in the next section.

5. Multifunctional Ligands

The multifactorial nature of AD and the scarcity of effective treatments have promoted the development of multifunctional agents capable of preventing or reverting some of the most important neuropathological hallmarks in AD. In the previous sections the role of three important metals involved in AD were reviewed. However, in addition to metallic ion dysregulation, other key factors with strong relationships among them could potentially trigger the disease. This is the case of Aβ deposits, high ROS levels, AChE inhibition, and monoamine oxidase (MAO) dysregulation. MAO is an enzyme located in the mitochondrial membrane that exhibits two isomeric structures A and B, upon which some neurotransmitter regulation is dependent. The catalyzed reaction of MAO isomers involves the release of hydrogen peroxide as a product, and, hence, an alteration of its enzymatic activity could lead to oxidative stress.¹⁴⁰ This multifactorial nature inspired the synthesis and testing of molecular scaffolds to obtain derived chelators with additional features.

One of last century’s paradigms in drug design consisted in the development of single-purpose drugs, i.e., one molecule as a drug active principle for each therapeutic target (metal, protein, enzyme, etc.). Over the years, results have shown that these kinds of compounds present undesirable side effects in the treatment of diseases characterized by having multiple pathological factors, such as neurodegenerative disorders.¹⁴¹ Nowadays it is more desirable to develop molecules with multiple therapeutical functions included within the same chemical structure. Sampietro et al. reviewed through a bibliographic search how in vitro and in vivo studies of multitarget compounds have been shown to be more effective compared to the use of single-target molecules in AD. In the same work they show that combinations of AChE/BChE-related oxidative stress, Aβ aggregation, and metal chelation-induced oxidative stress are among the top ten combinations used for the design of multitarget compounds.¹⁴²

The design of molecules that can carry out multiple functions has led to the proposal of candidates with chelating, intercalating, and antioxidant features. To ensure their safe application in AD treatment, these ligands need to exhibit low toxicity, good metabolic stability, and moderate affinity toward metals that are essential to metalloproteins and also possess the ability to cross the blood-brain barrier (BBB).^143,144 The design of multifunctional chelators is a complex task, and the development of new drug candidates is associated with high costs and large time periods. Currently, rational drug design is supported by computational tools that help to identify chemical structures with adequate drug-like properties and desirable activity toward therapeutical targets.^145−147 Diverse examples of multifunctional compounds have been reported in the literature throughout the years.^148−152

The implementation of different functions into a single structure for drug design follows three principal procedures.¹⁵³ The functionalization approach relies on organic chemistry reactions to generate chemical modifications in a compound with well-known drug-like properties, thus modulating its reactivity and mode of action. In contrast, the main goal of the attachment strategy is to obtain hybrid molecules by merging two or more known chemical structures using a linking bridge. This method has the advantage of leaving the active groups from the starting structures intact, building up a compound with mixed biological functions.¹⁵⁴ The attachment procedure is used to prevent undesired interactions for the metal chelator when crossing the BBB in order to reach the brain; this can be seen when glycoside fragments or nanoparticles are attached to the starting structure.^155,156 Lastly, the incorporation technique employs a structure whose function has to be optimized by the insertion of diverse substituents.^157,158 However, factors such as molecular weight, charge, and lipophilicity are important to achieve a proper absorption of the ligand into the organism and should be checked during the incorporation method.¹⁵⁹

One of the most common drug design filters is the well-known rule of five established by Lipinski, which consists in a set of structural constraints for an appropriate absorption into the organism: the ligand should have less than 500 amu in molecular weight, less than five hydrogen-bond donor groups (O–H or N–H), less than ten hydrogen-bond acceptor atoms (O or N), and an octanol/water partition coefficient lower than five.¹⁴⁵ These criteria potentially apply to water-soluble structures that display good permeability for biological barriers. According to these rules, highly polar compounds would be too hydrophilic to cross the BBB, and this would likely translate to a poor brain diffusion, making them innocuous or even toxic candidates, whereas more hydrophobic structures are desired for this purpose. In this sense, Quantitative Structure Activity Relationship (QSAR) models are very useful tools to filter initial candidates, as they make it possible to perform the fast calculation of drug-like related properties for large sets of compounds, such as BBB permeability, a key factor for drug candidates in the context of AD. However, it must be noted that the accuracy of these methods is usually moderate for these kinds of approaches.¹⁴⁷

The huge amount of possible structure modifications and functional diversity is one of the main challenges during the drug design process. Computational methodologies are essential for the design and evaluation of potential multifunctional chelators in AD and other metal-promoted neurodegenerative disorders. These tools allow us to identify and design compounds with potential therapeutic effects using diverse techniques, such as a database search, algorithms for the comparison of substructures, fast calculation of molecular descriptors through QSAR schemes, and virtual screening, among others. In addition, emerging tools such as artificial intelligence and machine learning models have been used to speed up and improve the screening process in computer-aided drug design.¹⁶⁰

A proposed computational protocol is presented in Scheme 3.^146,161 The strategy starts with the rational drug design of a structure that should present desirable properties, in addition to metal chelation, which will be used as a reference scaffold. Then, a virtual screening analysis is performed using various databases to find structures similar to this scaffold by checking chemical likeness, and so, the obtained list is expected to have a similar chemical structure and reactivity. At this point, a series of filters must be imposed to further detect the molecules with the best chances to become drug candidates; this can be done by executing a fast evaluation of the pharmacological properties over the entire molecular set. The last part of the protocol is to study the chelating and other desired properties, as antioxidant activity, for the reduced set of compounds by means of quantum-chemical calculations.

Scheme 3. Example of a Computational Protocol with the Essential Steps for the Design and Evaluation of Multifunctional Chelators.

Evidence supporting the role of metal ions in Aβ aggregation and the increase of oxidative stress has rendered metal ion chelation as a promising treatment in anti-AD drug design. Nevertheless, after more than 20 years of research, chelating therapy is in continuous debate. Recently, Drew suggested the possibility of abandoning therapeutic chelation, especially focusing on copper ions due to the lack of evidence of clinical benefits.¹⁶² However, Siotto et al. reviewed a meta-analysis that showed the relation between copper levels and AD development.¹⁶³

The initial proposals regarding metal chelators were focused mainly on the coordination of metal cations, and high metal affinities were desirable.¹²¹ However, chelators should not sequester metal ions that are part of essential proteins to avoid side effects and toxicity issues, but at the same time they must hold a competitive affinity compared to Aβ-metal complexes to induce therapeutical effects.¹⁶⁴ These agents are also called metal-protein attenuating compounds (MPACs), and, in addition to owning drug-like qualities, they possess regulated affinities toward metal ions. Many studies have been performed for in vitro and in vivo models and also in small groups of AD patients to assess Aβ-metal-related toxicity in the brain using MPACs.^165,166

Compounds with demonstrated metal chelation activity have served as inspiration in the design of new drug candidates with improved qualities by means of the functionalization, attachment, and incorporation techniques mentioned previously. Among these scaffolds clioquinol is a remarkable example.¹⁶⁷ After this, a whole generation of chelating agents were developed by including fluorescence activity through marker fragment incorporation, which makes them able to track the temporal evolution of Aβ senile plaques in the brain upon drug treatment.¹⁶¹ Some examples consist in thioflavin and p-I-stilbene structure-like compounds, where chelating properties are induced during drug design.^161,168 Most recently, chelating agents with appropriate antioxidative activity have been proposed to diminish the oxidative stress and neurotoxicity in some critical stages of the disease. Natural products and their derivatives are representative of this group of compounds with antioxidant properties; some examples are neoflavanoids, resveratrol derivatives, and curcuminoids.^169−171

The design of these multifunctional agents is typically carried out by empirical knowledge of the antioxidant activity given by functional groups or molecular structures (e.g., natural products). Nevertheless, the hydrogen peroxide cycle shown in Figure 2 involves multiple steps that are SRP-dependent, highlighting it as a potential key aspect to be considered throughout the drug design process. Chaparro et al. proposed a computational protocol for the design of promissory candidates with appropriate affinity toward the metallic ions of interest, drug-like characteristics, and controlled antioxidant properties by calculating the SRP values of the involved metal complexes.¹⁷² This methodology is presented in Figure 5, where it was applied to the search for copper multifunctional agents. It started with the selection of a set of 64 copper complexes with reported experimental SRP values in aqueous media, which were then modeled by DFT calculations in their oxidized and reduced states. The latter allowed us to estimate the metal affinity toward both Cu²⁺ and Cu¹⁺ ions, selecting a first set of compounds by their proper chelating ability. Then, further filtering was accomplished by applying Lipinski’s rule of five and calculating the BBB permeation. The remaining set of compounds was classified into two groups according to their SRP: antioxidants, whose SRP values are higher than the corresponding to produce hydrogen peroxide from oxygen in biological conditions (0.295 V),⁴⁴ and redistributors, whose SRP values are in the range from −0.32 to 0.30 V (the lower limit being defined by the most abundant natural reducing agents present in neurons).

Figure 5.

Computational protocol for the selection of third-generation chelating agents. Adapted with permission from ref (172). Copyright 2021 IOS Press.

In these terms, antioxidant molecules would be able to disrupt ROS generation thanks to their larger associated SRP value, whereas redistributor-like compounds would exhibit the characteristics needed to accomplish a mechanism for metal redistribution into neuronal cells. These SRP calculations were performed by using an adapted methodology for isodesmic reactions that allows for an accurate prediction, as it was evaluated with a wide variety of copper complexes with different types of coordination spheres.¹⁷³ Finally, molecular scaffolds with the desired properties were obtained and used to build up a set of derivatives that fulfill the previous requirements.

Extensive research has been carried out in recent times regarding promissory multifunctional chelating agents; some examples of these compounds and their most important characteristics are presented in Table 1. Here, the most relevant metal chelating affinities, experimental tests, computational studies, and additional features are listed for some representative derivative compounds obtained from commonly known scaffolds in AD drug design. Most of the derivatives contain N and O donor atoms, which can coordinate the metal ions and form 1:1 or 1:2 metal–ligand complexes depending on the geometrical constraints of the ligand. Frequently, the computational chemical studies of candidates include a QSAR analysis to assess their drug-like properties and filter a large set of structures, DFT calculations of chelators and its corresponding metal complexes, and molecular docking simulations to explore the interactions of the promissory candidates with Aβ monomers and fibrils. It is important to emphasize that, in many cases, the chelators described in Table 1 need further investigation in animal models before they can be tested in clinical trials.

Table 1. Relevant Chelating and Multifunctional Agents Proposed as Possible AD Drugs^a.

^aThe chemical names for the compounds are presented as footnotes. Deferiprone, DFP: 3-hydroxy-1,2-dimethyl-4(1H)-pyridone. Clioquinol, CQ: 5-chloro-7-iodo-8-hydroxyquinoline. PBT2:5,7-dichloro-2-[(dimethylamino)methyl]-8-hydroxyquinoline. INHHQ: 8-hydroxyquinoline-2-carboxaldehyde isonicotinoyl hydrazone. INHHQ: 8-hydroxyquinoline-2-carboxaldehyde isonicotinoyl hydrazone. Pic-stilbene: N¹,N¹-dimethyl-N⁴-(pyridin-2-ylmethyl)benzene-1,4-diamine. ML: 4-(dimethylamino)-2-(((2-(hydroxymethyl)quinolin-8-yl)-amino)-methyl)phenol. Curcumin: (1E,6E)-1,7-bis(4-hydroxy-3-methoxyphenyl)-1,6-heptadiene-3,5-dione. XH1: ([(4-benzothiazol-2-yl-phenylcarbamoyl)-methyl]-{2-[(2-{[(4 benzothiazol-2-yl phenylcarbamoyl)methyl]-carboxymethylamino}-ethyl)-carboxymethylamino]-ethyl}-amino)-acetic acid. HBXI: 2-(Benzo[d]oxazol-2-yl)-4-iodophenol. MAOI-8: (E)-5-(3-hydroxy-4-(piperidin-1-ylmethyl)styryl)-4-(hydroxymethyl)-2-methylpyridin-3-ol. TAC-BIM: Tacrine-hydroxyphenylbenzimidazole.

The increasing number of multifunctional agents has turned this area into an intense field of research. The number of molecules tested at several levels has allowed us to understanding the role of metal cations and to propose new and more efficient candidates to prevent or revert the neuronal damages observed in AD patients. This review attempts to provide a state of the art on the role of these metal cations in the increase of the oxidative stress and the multifunctional ligands proposed to fight the effects of the disease.

6. Conclusions

Alzheimer’s disease is currently one of the major challenges in human health. The projection about people that will become AD patients is a concerning situation for the upcoming years. This highlights the urgency of further research toward a full understanding of the disease and the development of alternative and effective treatments. Based on experimental observations, some pathological hallmarks in AD have been well-established, and several hypotheses have been formulated as an effort to explain them. The present review focused on the role of the main metals involved in AD such as the formation of highly stable Aβ-metal complexes or the production of ROS that cause oxidative damage in neurons. This is also related to the multifactorial nature of the disease, which increases the difficulty to find a cure due to the existence of different therapeutical targets.

Accordingly, rational drug design in AD aims to develop multifunctional chelating compounds capable of having multiple desirable features. Here, we reviewed the current state of multifunctional chelators, and a classification based on their therapeutical functions is provided. A list of remarkable chelating compounds along with their principal characteristics and related works is given. Furthermore, we also present the impact of computational chemistry in AD drug design during the last years.

Glossary

Abbreviations

AD

Alzheimer’s disease

APP

amyloid precursor protein

PS

presenilin

APOE

apolipoprotein E

Aβ

β-amyloid

ROS

reactive oxygen species

AChE

acetylcholinesterase

FDA

Food and Drug Administration of the United States

SOD

superoxide dismutase

CTR1

copper transporter protein

MD

molecular dynamics

AIMD

ab initio molecular dynamics

DFT

density functional theory

SRP

standard reduction potential

MAO

monoamine oxidase

BBB

blood-brain barrier

QSAR

quantitative structure activity relationship

MPACs

metal-protein attenuating compounds

BChE

butyrylcholinesterase.

Author Present Address

^∥ Department of Chemical Engineering, The University of Melbourne, Parkville, Victoria 3010, Australia

Author Contributions

^⊥ These authors contributed equally. The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.

This work was supported in part by Project INV-CIAS-3408 from Universidad Militar Nueva Granada.

The authors declare no competing financial interest.