Small molecule inhibitors of amyloid β peptide aggregation as a potential therapeutic strategy for Alzheimer's disease

Abstract

Amyloid β (Aβ) peptides have long been viewed as a potential target for Alzheimer's disease (AD). Aggregation of Aβ peptides in the brain tissue is believed to be an exclusively pathological process. Therefore, blocking the initial stages of Aβ peptide aggregation with small molecules could hold considerable promise as the starting point for the development of new therapies for AD. Recent rapid progresses in our understanding of toxic amyloid assembly provide a fresh impetus for this interesting approach. Here, we discuss the problems, challenges and new concepts in targeting Aβ peptides.

Introduction

More than 100 years have passed since Alzheimer's disease (AD) was first characterized. However, due to the lack of effective treatment, AD remains pandemic in the 21st century, imposing enormous social, and economic burdens on patients and their families¹. Modern demographic trends compound the problem; our aging global population has led to a steep increase in the number of individuals with AD. In the United States alone, more than 13 million individuals are predicted to be afflicted with AD by the year 2050², leading to an overburden of scarce healthcare resources. Some studies estimate that the present availability of a treatment that can delay disease onset by 6.7 years would decrease the prevalence of AD 38% by 2050³. Such disease-modifying treatments would lower the annual cost of individual patient care by up to $24 000, thereby reducing the national cost of AD by trillions of dollars through the year 2050^{4, 5}.

Unfortunately, currently available treatments, eg, Aricept and Memantine, usually provide at best only temporary and incomplete symptomatic relief. The marginal benefits provided by current therapies emphasize the urgent need to develop effective disease-modifying AD treatments.

Amyloid β–targeting strategies

Disease-modifying strategies currently being pursued for AD mainly focus on two AD-related proteins, amyloid β (Aβ) and Tau. Of these, Aβ has attracted the most attention by far.

Substantial data derived from genetics, animal modeling, and biochemical studies support the idea that Aβ, the major component of senile plaques, plays a central role in AD pathophysiology^{6, 7}. Paradoxically, Aβ peptides — the seeds of AD — are present at birth, and these neurotoxic peptides are continuously produced throughout life. Thus, the most direct target in anti-Aβ therapy is reduction of Aβ production, which has led naturally to a focus on β- and γ-secretase inhibitors^{8, 9, 10, 11, 12, 13}. However, despite considerable effort on this front, few novel chemical compounds based on this strategy have reached clinic trials. Safety issues are the overriding reason; because γ-secretase has many physiologically essential substrates, undesirable side effects are inevitable. One such substrate is the Notch signaling protein, which plays an intermediate and essential role in the differentiation and proliferation of many cell types. The consequences of γ-secretase inhibition include impaired lymphocyte differentiation and altered intestinal goblet cell structure^{14, 15, 16}. Thus, although the use of small-molecule γ-secretase modulators is receiving increasing attention as a promising therapeutic approach¹⁷, many companies have abandoned γ-secretase as a potential target.

In theory, inhibiting β-secretase should not carry the same risk of toxicity as γ-secretase inhibition. However, designing β-secretase inhibitors has been challenging. The β-secretase protein contains a large catalytic pocket; thus, the β-secretase inhibitors that have been developed to date are too large to penetrate the blood-brain barrier^{18, 19, 20}. There may also be conceptual drawbacks to targeting β-secretase. The development of Aβ secretase inhibitors is based on the hypothesis that Aβ deposition in the brain is due to overproduction or poor clearance of Aβ. However, this is likely only true for cases of familial AD caused by genetic mutation. Non-genetic AD cases, which represent the majority of AD patients, do not carry mutations and do not necessarily have overproduction of amyloid precursor protein (APP)²¹. More importantly, a deeper understanding of Aβ has revealed that Aβ isoforms also serve as endogenous positive regulators of release probability at hippocampal synapses, with some studies suggesting that monomeric Aβ is beneficial for neurons^{22, 23}. Thus, because Aβ production may be important for physiological health, inhibiting Aβ generation may not necessarily be a sound strategy. An alternative tactic is to focus on Aβ clearance. One approach is to enhance the peripheral Aβ “sink action” by sequestering plasma Aβ, for example using Aβ immunotherapy. The AN1792 Aβ vaccine was designed for this purpose; however, it failed in Phase II clinical trials due to the development of meningoencephalitis²⁴. Nevertheless, it might be possible to circumvent such complications by targeting the proteolytic machinery in the brain. However, from a therapeutic point of view, it remains to be seen whether any of the relevant proteases will serve as a viable AD drug target²⁵.

Aβ can aggregate into fibrils and oligomers (Figure 1). It was originally believed that only the large fibrils constituting mature neuritic amyloid plaques are toxic. However, in recent years research focus has shifted towards Aβ oligomers. Currently, Aβ peptide aggregation into toxic, prefibrillar oligomers is considered the key pathogenic event in the onset of AD^{26, 27}. Consistent with this view, Aβ oligomers can: i) directly induce synaptic dysfunction and neuronal death, both which are responsible for AD initiation and progression^{28, 29, 30}; and ii) trigger events such as oxidative damage and inflammation, which contribute to the progression of AD^{31, 32, 33}. Although generation of Aβ monomers is, in and of itself, a physiologically relevant event, their aggregation into oligomers and fibrils is pathogenic. Thus, blocking aggregation (while sparing Aβ generation) should not lead to mechanism-based toxicity. Hence, the strategy of inhibiting aggregation of Aβ, more specifically the Aβ42 isoform, has emerged as a valid disease-modifying therapy for AD³⁴.

Figure 1.

The plaque formation.

Problems associated with inhibiting Aβ aggregation

Tremendous effort has been expended in recent years on developing small molecules that are capable of inhibiting Aβ aggregation. However, to date, no compounds have entered into clinical use. This is because the inhibition of Aβ aggregation requires blocking interactions between Aβ monomers, and historically, protein–protein interaction interfaces are particularly difficult drug targets. Thus, inhibiting Aβ aggregation has been deemed intractable^{35, 36}. The challenges associated with targeting Aβ aggregation are substantial, with the most obvious obstacle being the sheer size and geometry of the protein interaction surface. First, the regions of protein–protein interactions are approximately 1500–3000 Å^{237, 38, 39, 40}, whereas protein–small molecule interaction regions are only about 300–1000 Å^{2 41, 42}. Thus, small molecules are often unable to produce sufficient steric hindrance to inhibit Aβ aggregation⁴³. Second, the regions of protein–protein interactions are often relatively featureless, devoid of grooves or pockets into which a small molecule could dock in an energetically favorable manner^{43, 44, 45, 46, 47}. The binding energy that drives protein–protein contacts is typically distributed over a large area that lacks a defined “hotspot” for pharmacological intervention⁴⁸. Third, the highly plastic nature of protein surfaces can accommodate small molecules, thereby thwarting inhibition⁴⁸. Together, these problems place severe restrictions on the development of Aβ aggregation inhibitors.

Designing solutions to these problems

To tackle these challenges, researchers have envisioned a Trojan horse strategy in which a small bifunctional molecule gains access to the relevant biological compartment, binds tightly to a chaperone, and thereby provides the necessary steric bulk to disrupt protein–protein interactions⁴⁸. A second approach is to directly target the important chaperones, such as metals, in the brain. A growing body of evidence suggests that the presence of excess zinc, iron, copper and/or aluminum within senile plaques exacerbates Aβ-mediated oxidative damage and acts as a catalyst for Aβ aggregation in AD⁴⁹. Thus, chelation therapies aimed at disrupting aberrant interactions between Aβ and metals have attracted considerable attention; one such example is PBT2, which is currently in Phase IIa clinical trials⁵⁰. Unfortunately, to date, no new compounds employing this strategy have entered into clinical use.

The discovery that Aβ fibril formation is controlled by specific amino acids within the Aβ peptide itself has given renewed life to the idea that small molecules might occlude Aβ aggregation. Studies have demonstrated that various Aβ peptide regions contribute differently to Aβ aggregation and have identified important interactions among specific peptide regions that control this process⁵¹. For instance, the N-terminus^{52, 53}, hydrophobic core^{54, 55}, hinge or turn regions^{51, 56, 57} and C-terminus^{58, 59} of Aβ41-42 are all crucial for the peptide's ability to aggregate and promote neurotoxicity.

The importance of the His13–Lys16 (HHQK) region in oligomerization, fibril propagation and neurotoxicity is well known^{60, 61, 62}. This four-residue region is also an essential component of the heparin-binding site for glycosaminoglycans (GAGs), which assist the HHQK motif in effecting a change in the Aβ secondary structure from a soluble, unordered α-helix to a stable β-sheet–rich conformation. These β-sheet–rich structures, in turn, have an affinity for associating with other monomers to form oligomers, protofibrils and fibrils that aggregate into plaques⁶⁰. The β-hairpin structure serves to nucleate the folding of Aβ monomers, which is the rate-limiting step in fibril formation^{56, 57, 63}. The formation of hinge or turn regions brings together two hydrophobic segments in space and produces the β-strand structure. In this region, Lys28 forms a salt bridge with Asp23/Glu22, which further stabilizes the structure^{63, 64, 65}. Electrostatic and hydrophobic interactions between these non-native side chains bring the peptides together, allowing native backbone−backbone interactions to initiate the assembly of β-sheets, albeit in non-native β-sheet registries⁶⁴. The hydrophobic segments are the main core of Aβ aggregates, and hydrophobic interactions are an important driving force in Aβ aggregate formation^{66, 67}. Based on these observations, a β-sheet breaker was first investigated as a potential inhibitor of Aβ aggregation⁶⁸. A substantial body of evidence suggests that the C-terminus of Aβ42 plays a key role in controlling oligomerization; indeed, several studies of prefibrillar Aβ have suggested that the C-terminus of Aβ42 is more rigid than the C-terminus of the more abundant but less toxic Aβ40^{69, 70, 71}. This increased rigidity has been attributed to interactions involving C-terminal residues Ile41 and Ala42, which stabilize a putative turn conformation⁷¹. Although these subregions contribute differently to structure formation, each is important for Aβ aggregation.

By targeting these subregions for binding, a small molecule will be more likely to directly occlude Aβ aggregation. In fact, most inhibitors that have entered clinical trials target these specific subregions within the Aβ sequence (Figure 2). Tramiprosate (3-aminopropanesulfonic acid), a mimic of GAGs, targets the HHQK subregion at the N-terminus. Although the structural simplicity of tramiprosate makes it highly specific to Aβ, it ultimately failed in the late stages of a Phase III clinical trial⁷². Despite this setback, the data obtained provided evidence that inhibitors targeting specific Aβ subregions represent a viable approach for AD treatment⁷³. The agents targeting C-terminus feature one scaffold, the cyclohexanehexol stereoisomers. Among these, scyllo-inositol, which interferes with Aβ fibril assembly by competing with endogenous phosphatidylinositol for binding to Aβ41-42⁵⁹, has shown promise in current Phase II trials. The hydrophobic central region is yet another promising target for Aβ fibrillation inhibitors, which include RS-0406⁷⁴.

Figure 2.

The subregion-targets in Aβ.

The quest for small molecules that target Aβ

As the strategy of inhibiting Aβ aggregation has increasingly gained acceptance, greater numbers of inhibitors have been developed and the structure-activity relationships of potent inhibitors have been systematically explored⁷⁵. These studies revealed that typical Aβ aggregation inhibitors such as Congo red (CR), chrysamine G (CG) and curcumin share a similar chemical scaffold. These molecules contain two aromatic groups or inositol groups (with a suitable substituted group) separated by a backbone of the appropriate length⁷⁵. We speculate that the two terminal groups interact with Aβ protein residues to provide the binding affinity, whereas the linker facilitates binding of inhibitors to specific subregions. Determining which subregion the inhibitors target will be key to future research efforts. A molecule that targets a specific subregion should satisfy the following design criteria: (1) it should contain terminal groups suitable for interacting with residues in, or adjacent to, the subregion; and (2) the linker should be of an appropriate length to join the two terminal groups, span the subregion, and provide sufficient steric hindrance. Molecules designed according to these specifications would therefore be capable of inhibiting Aβ aggregation by binding to a specific subregion.

The handful of inhibitors that have been developed to date do not possess this chemical scaffold. Scyllo-inositol contains only one terminal group and can therefore target only a very small region. Fortunately, scyllo-inositol targets the C-terminus, which is important for Aβ aggregation. However, the presence of a single terminal group restricts the area of the targeted subregion to the C-terminus, and the lack of a linker prevents scyllo-inositol from modulating other targeted subregions. Additionally, a single terminal group results in a weaker binding avidity. In the case of tramiprosate, its simple structure provides specificity for Aβ, yet, for unknown reasons it failed in Phase III clinical trials⁷². Thus, although these atypical molecules possess inhibitory activity, they are severely limited by their weak binding affinity and small occupied region.

To our knowledge, most inhibitors target only one subregion. In fact, if inhibitors could target multiple specific subregions in concert, they might act synergistically to effectively inhibit Aβ aggregation. Thus, in our view, the ideal Aβ-targeting inhibitor should be of appropriate length and suitable flexibility, and should contain multiple groups that interact with Aβ residues. These multiple groups would both convey tighter binding avidity and permit targeting of multiple subregions. In addition, having sufficient length and flexibility would enable the inhibitor to capture the entire Aβ peptide and accommodate conformational changes in the target. Unfortunately, our current lack of the Aβ peptide's detailed molecular structure precludes designing such inhibitors. The successful creation of these new inhibitors will require gaining a better understanding of the modules necessary for activity as well as the key elements in the amyloid surface required for aggregation and toxicity. Alternatively, it might also be possible to employ a combination of multiple molecules.

Future perspectives

Compounds such as peptide-based inhibitors, antibodies and small molecules that target specific Aβ subregions represent the first generation of amyloid-based therapeutics with the potential to demonstrate disease-modifying activity. Although the results of ongoing clinical trials are inconclusive, these compounds hold the promise of a new day in the development of disease-modifying therapies for AD; some Aβ aggregation inhibitors are listed in Table 1. As we gain additional insights into amyloid biology and AD itself, this will likely guide the development of the next generation of inhibitors.

Table 1. Some chemical inhibitors of amyloid beta aggregation.

Chemical compound	Structure	Current trial state
Bapineuzumab	Monoclonal antibody	Phase III
Tramiprosate		Failed in phase III
PTB2		Phase IIa
Scyllo-inositol		Phase II
PPI1019		Completed phase II
RS0406		Phase II
SP-233		Phase I
EGCG		Phase II
Exebryl-1	No structural details are available	Phase I
SEN606		Preclinical