If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Aptamers are oligonucleotides selected from large pools of random sequences based on their affinity for bioactive molecules and are used in similar ways to antibodies. Aptamers provide several advantages over antibodies, including their small size, facile, large-scale chemical synthesis, high stability, and low immunogenicity. Amyloidogenic proteins, whose aggregation is relevant to neurodegenerative diseases, such as Alzheimer’s, Parkinson’s, and prion diseases, are among the most challenging targets for aptamer development due to their conformational instability and heterogeneity, the same characteristics that make drug development against amyloidogenic proteins difficult. Recently, chemical tethering of aptagens (equivalent to antigens) and advances in high-throughput sequencing-based analysis have been used to overcome some of these challenges. In addition, internalization technologies using fusion to cellular receptors and extracellular vesicles have facilitated central nervous system (CNS) aptamer delivery. In view of the development of these techniques and resources, here we review antiamyloid aptamers, highlighting preclinical application to CNS therapy.
Neurodegenerative diseases are characterized by a progressive loss of neuronal function. Most of these diseases are age-related, and self-assembly of amyloidogenic proteins is thought to be a cause or a major deleterious mechanism in many of them. Examples include Alzheimer’s disease (AD) and various other tauopathies, Parkinson’s diseases (PD) and other synucleinopathies, prion diseases, and many other sporadic or genetic proteinopathies. The amyloidogenic proteins involved in these diseases are prone to self-association into neurotoxic oligomers and amyloid fibrils (Fig. 1). In many cases, the oligomers, which have metastable structures, have been shown to play pivotal roles in the pathogenesis of the associated diseases and to be more toxic than the structurally stable fibrils (
). The abnormal protein assemblies cause neurotoxicity by a variety of mechanisms, including apoptosis, oxidative stress, inflammation, and disruption of proteostasis through blockage of proteasomal and lysosomal protein degradation. Given the significance of oligomers in the pathogenesis of many neurodegenerative diseases (
), they have been the focus of attention as molecular targets of both diagnostic and therapeutic research and development. Therefore, generation of specific oligomer-binding reagents is a promising approach for development of early detection tools and redirecting the self-assembly process into dissociation back to nontoxic monomers, formation of nontoxic and nonamyloidogenic assemblies amenable to degradation, or in some cases, accelerating the aggregation process to reduce steady-state levels of oligomers in favor of less-toxic fibrils.
Aptamers are molecular-recognition agents comprising single-stranded DNA or RNA oligonucleotides that similar to antibodies, bind specifically to diverse targets, including small molecules, peptides, proteins, and nucleic acids (
) making it difficult to select aptamers specific for just one protein and even more so for distinct assembly states of these proteins. Nonetheless, aptamers offer several advantages compared with antibodies including their small size, facile chemical synthesis, including in large scale, high stability, and low immunogenicity, making them attractive for researchers aiming at developing molecular recognition tools for amyloidogenic proteins. Aptamers typically are obtained by selection from a random-sequence oligonucleotide library based on their affinity for the target of interest using a method called systematic evolution of ligands by exponential enrichment (SELEX, Fig. 2). The oligonucleotides usually span 30 to 100 nucleotides in length and their dissociation constants in complexes with their targets range from pM to mM. The sequence also contains constant regions required for enzymatic manipulation, such as PCR-primer binding and in vitro transcription.
A systematic survey by Dumontier, DeRosa, and their colleagues (
) based on 492 published aptamer-related papers has found that the use of DNA aptamers is increasing compared with RNA aptamers. Each nucleic acid has its own advantages and limitations. DNA oligonucleotides are more stable than their RNA counterparts to enzymatic and chemical degradation and are therefore easier to work with. On the other hand, the presence of the 2′-OH in ribose, as opposed to deoxyribose, and the absence of the 5′-methyl group in uracil compared with thymine allow higher conformational stability of RNA, potentially increasing their affinity for the target (
), yet to our knowledge, there are no systematic reviews on aptamers targeting amyloidogenic proteins. The metastable nature of intrinsically disordered or misfolded proteins, which are prone to form toxic oligomers and eventually amyloid, makes these proteins one of the most challenging targets for aptamer generation. In the case of oligomers, the aptagens (equivalent to antigens) presented to the oligonucleotide library constantly change, whereas in the amyloid fibrils, the conformation of the aptagens can be both variable for the same protein due to formation of different strains and similar for different proteins in the core cross-β structure shared by most amyloid fibrils.
Aptamers selected against amyloidogenic proteins can have various applications, including sensitive detection of biomarkers, as selective inhibitors of the self-assembly process, and as tools for probing molecular mechanisms. The first application—using aptamers as probes for biomarkers—has been developing rapidly and is too large to include in this review. Therefore, we focus here on the challenges and potential solutions in this field and on preclinical therapeutic applications of aptamers specific for amyloidogenic proteins.
To overcome the difficulty in selecting aptamers against metastable protein assemblies, approaches using stable mimics of these assemblies could be useful if the stabilized molecules represent accurately the metastable target. Here, we provide an update on the development of specific aptamers against amyloidogenic proteins, including amyloid β-protein (Aβ), tau, α-synuclein (αSyn), and prion protein (PrP). These proteins represent a range of sizes spanning an order of magnitude—from 40- to 441-amino acid residues, and the two main mechanisms of initial misfolding and aggregation—partial folding of an unstructured protein, such as Aβ, αSyn, or tau, and partial unfolding of a structured protein—PrP (Fig. 1) (
). We discuss how bioinformatics-assisted approaches or artificial-intelligence-based technologies are used to assist the aptamer selection and optimization processes. We also examine approaches for analysis of the secondary and tertiary structures of aptamers and aptamer–target interaction and strategies for CNS-targeting delivery of aptamers for future development of therapeutics in neurodegenerative diseases.
Development of aptamers against amyloidogenic proteins
Several dozen amyloidogenic proteins play major deleterious roles in over 50 proteinopathies. However, studies of aptamers against these proteins have concentrated primarily on four proteins, Aβ, tau, αSyn, and PrP, and therefore, these proteins are also the focus of this review.
Aptamers against Aβ
AD is the most common neurodegenerative disease. Amyloid fibrils in senile plaques in the AD brain consist mainly of the 42-amino acid residue form of Aβ, Aβ42, whereas vascular deposits comprise predominantly the 40-residue form, Aβ40. Both forms are generated from the Aβ-protein precursor by the somewhat promiscuous protease, γ-secretase (
), making Aβ42 oligomers a primary target for therapy development. In this context, it is important to consider, however, that fibrils might sequester the more toxic Aβ oligomers and possibly are a way cells attempt to reduce the damage caused by the oligomers (
). Thus, if strategies targeting fibril dissociation are considered, inadvertent increase in the concentration of the toxic oligomers, leading to exacerbation, rather than amelioration, of the disease might occur, and one must ensure that this is not the case.
A complicating factor is that “oligomer” is a loosely defined term used for anything from a dimer to large assemblies consisting of hundreds of monomers (
) as long as these assemblies are soluble in aqueous solutions, as opposed to fibrils and other insoluble aggregates. Many different types of Aβ oligomers have been reported, including paranuclei (5,6-mers) (
), though due to the metastable nature of the targets, in all cases the specificity of the antibody was lower than what is typically expected for antibodies against stable antigens. Thus, stabilization of the antigens, for example, by attachment of Aβ to gold particles (
), has been an important strategy for generation of antioligomer antibodies. However, an important consideration is that the relevance of the stabilized antigens to the pathological species in the AD brain must be established in each case. Of the many clinical trials using anti-Aβ antibodies, more recent ones have examined antibodies selective against Aβ oligomers, including Biogen’s aducanumab, which recently was approved by the Food and Drug Administration (FDA), and BAN2401, a humanized version of mAb158.
The first aptamers targeting Aβ were reported in 2002 by Ylera et al. The aptamers were RNA oligonucleotides screened against Aβ40 monomers tethered to a Sepharose support. All the selected aptamers recognized Aβ fibrils but not monomers (
To select aptamers against Aβ oligomers, several chemical tethering approaches have been used in later studies. Rahimi et al. prepared covalently cross-linked Aβ40 oligomers using Photo-Induced Cross-linking of Unmodified Proteins (PICUP) (
). Of several Aβ40 oligomers stabilized by this technique, they then isolated the most abundant type, trimers, using SDS-PAGE and used the cross-linked trimers to isolate RNA aptamers out of a 77-nucleotide library including 49 randomized nucleotides (A:U:G:C at equal ratios). This size library contains 3.2 × 1029 unique sequences theoretically, yet in reality, if every sequence indeed was included in the library, the mass of such a library would be too high, and the actual number of unique sequences is lower.
Although the selection of the RNA aptamers was against Aβ40 trimers, the final aptamers, KM33 and KM41, were not selective for trimers or other oligomers, but bound to Aβ40 fibrils and did not display a higher affinity for the fibrils than the naive oligonucleotide library used for the selection before enrichment (
). Part of the explanation of these results was that the PICUP-immobilized Aβ40 trimers themselves might have aggregated during the selection process to form fibrillar structures. Further analysis suggested that oligonucleotides have a high, nonspecific affinity for amyloid fibrils, as KM33 and KM41 also recognized fibrils of other amyloidogenic proteins, including calcitonin, islet amyloid polypeptide, insulin, lysozyme, and PrP106–126 (
) demonstrated that the aptamers could be used to monitor fibril formation in a similar manner to thioflavin T (ThT) fluorescence.
Interestingly, a DNA aptamer called T-SO517, originally developed by Ikebukuro and colleagues against oligomeric αSyn, a key protein involved in PD and other synucleinopathies, bound not only to their intended target, αSyn oligomers, but also to Aβ40 oligomers (
), demonstrating the difficulty in obtaining highly specific aptamers against assemblies of amyloidogenic proteins. Considering the high affinity of the KM aptamers mentioned above for fibrils, these findings likely reflect the presence of common structures in such assemblies, i.e., not only the common cross-β structure of the fibrils, but also in the oligomers, as has been suggested by binding of antioligomer antibodies, such as A11, to oligomers of multiple amyloidogenic proteins (
) for preparation of oligomer-specific antibodies. The RNA aptamers, N2 and E2, bound to Aβ40 monomers and inhibited Aβ40 fibril formation assessed using the ThT fluorescence assay and transmission electron micrography. However, as was found by Rahimi et al. (
) demonstrated later by using atomic force microscopy that these aptamers also bound to Aβ40 fibrils. Binding to Aβ- or other protein oligomers was not described in either study. A DNA aptamer named RNV95, selected using column-immobilized Aβ40 by Chakravarthy et al. with the goal of detecting low-molecular-weight Aβ40, recognized human brain Aβ oligomers at both ∼75 kDa and ∼150 kDa. However, it was not tested for binding to Aβ fibrils or fibrils of other amyloidogenic proteins so its selectivity for oligomers was not established (
). The first characteristic, substantially faster aggregation, makes working with Aβ42 particularly difficult because the preparation of the protein changes during the experiment making obtaining reproducible data highly challenging.
The first example of a successful generation of aptamers against Aβ oligomers we are aware of is by Murakami et al., who developed RNA aptamers termed E22P-AbD4, -AbD31, and -AbD43 against Aβ42 protofibrils (
). To select the aptamers, a dimer of E22P-Aβ42 was used in which the monomers were tethered covalently by a bivalent amino acid linker in place of Val40, within the C-terminal hydrophobic region of Aβ42 (Fig. 3B) (
), formed protofibrils, the morphology of which was confirmed by transmission electron microscopy. The protofibrils then were used in SELEX to obtain the aptamers from a 77-nucleotide RNA library including a 49-nucleutide random sequence. When used in histological experiments, all three aptamers stained diffuse oligomeric aggregates in two mouse models of AD, Tg2576/PS2 (
), suggesting that the Aβ assemblies formed in the brain of these two mouse models contained similar protofibril-derived aptatopes. Incubation of the E22P-Aβ42 dimer in the presence of aptamer E22P-AbD43 showed that the aptamer inhibited the nucleation phase of the protofibril formation. The aptamer also inhibited dose-dependently the neurotoxicity of both the E22P-Aβ42 dimer and Aβ42 in the neuroblastoma cell line SH-SY5Y cells. Computational and two-dimensional structure analysis of E22P-AbD43 suggested that preferential binding of the aptamer to Aβ42 protofibrils compared with fibrils might be related to formation of a G-quadruplex structure, implying the presence of a common structure in protofibrils made of either the synthetic dimer or native Aβ42 protofibrils (
For the purpose of developing inhibitors of Aβ42 aggregation, Zheng et al. selected an anti-Aβ42 DNA aptamer (Aβ7-92-1H1) by incubating a library of DNA-oligonucleotide-coated beads with nonaggregated Aβ42. This library consisted of 100-nucleotide-long sequences including two separate 18-nucleotide random sequences. Aβ7-92-1H1 bound Aβ42 and neither Aβ40 nor other amyloids. The affinity of the aptamer for Aβ42 oligomers was slightly higher than for Aβ42 monomers, measured by surface plasmon resonance (SPR). The aptamer inhibited Aβ42 fibril formation as evidenced by atomic force microscopy and β-sheet formation measured using CD spectra (
Tauopathies are neurodegenerative diseases caused by formation of toxic oligomers and aggregates of the microtubule-associated protein tau. Tau is produced as six different isoforms in humans, due to alternative splicing of exons 2, 3, and 10 leading to polypeptides ranging from 352 to 441 amino acid residues. Translation of neither, one, or both exons 2 and 3 is marked as 0N, 1N, or 2N isoforms of tau, whereas the alternative splicing of exon 10, which encodes part of the microtubule binding, repeat domain of tau leads to isoforms containing three (3R) or four (4R) repeats.
). Other examples of tauopathies include frontotemporal lobar degeneration, progressive supranuclear palsy, and chronic traumatic encephalopathy. Aggregation and deposition of tau, e.g., as neurofibrillary tangles in AD, are associated with hyperphosphorylation and other aberrant posttranslational modifications of the protein.
In the course of testing whether tau binds single-strand DNA, Krylova et al. (
) first found DNA sequences binding recombinant 1N3R-tau or 2N3R-tau using nonequilibrium capillary electrophoresis of equilibrium mixtures, in which the gel shift caused by binding of the DNA oligonucleotides to tau was evaluated. In follow-up studies, these DNA sequences were used as aptamers for detection of 1N3R-tau in human plasma using SPR and were shown to reach femtomolar level sensitivity (
). However, this is not a standard method in the amyloid field. Lisi et al. applied this method coupled with capillary electrophoresis for partitioning of bound DNA from unbound DNA to isolate DNA aptamers targeting several tau isoforms. The aptamer 3146 was obtained from a 5 × 1012 DNA oligonucleotide library in only three rounds within 1 day. 3146 bound tau isoforms in the order 2N4R (KD = 13 nM) > 0N4R (49 nM) > 0N3R (84 nM) > 1N3R (116 nM) (
An RNA aptamer called tau-1 was obtained against 2N4R-tau, by applying SELEX to a 90-nucleotide RNA library containing a 40-nucleotide random region. This aptamer prevented the formation of 2N4R-tau dimers and trimers assessed by SDS-PAGE/western blotting (
). It is important to note, however, that SDS-PAGE is not a reliable method for assessing oligomer formation by amyloidogenic proteins, as SDS perturbs protein conformation and can both dissociate existing assemblies and induce formation of different assemblies (
αSyn, a causative agent of synucleinopathies, such as PD, dementia with Lewy bodies, and multiple system atrophy, oligomerizes and aggregates in the brain of patients leading to neurotoxicity and neurodegeneration (
). αSyn is a 140-amino-acid long protein mainly located at presynaptic terminals. The amphipathic N-terminal region (residues 1–60), which includes four 11-residue imperfect repeats, and the hydrophobic middle region (residues 61–95) are more important for aggregation than the acidic and proline-rich C-terminal region (residues 96–140) of αSyn (
), yet most antibodies bind to this region and therefore may miss αSyn aggregates in pathological analysis of patient brains or animal models. This problem is particularly important because the most common form of αSyn used to detect pathological aggregates is phosphorylated at Ser129 (pS129-αSyn), yet some cleavage sites are N-terminal to position 129 and eliminate the epitope for antibodies against pS129-αSyn.
Similar to Aβ and tau, αSyn oligomers, rather than fibrils, are thought to be the primary neurotoxic form of the protein. αSyn oligomers cause neurotoxicity, synaptic impairment, mitochondrial dysfunction, endoplasmic reticulum stress, neuroinflammation, proteostasis dysregulation, and apoptosis, culminating in neuronal death (
). The details of αSyn oligomer structures are not known, yet β-sheet structure has been reported in rigid regions of toxic αSyn oligomers, whereas in nontoxic oligomers, these regions are unstructured (
). In follow-up studies, the same group applied counterselection of monomers and fibrils, leading to a more selective aptamer, T-SO530, against αSyn oligomers, whose selectivity was confirmed by dot blots (
). CD analysis suggested that T-SO530 formed a G-quadruplex structure, which might have contributed to its affinity. Indeed, G-quadruplex stabilizers, such as L1H1-7OTD and TmPyP4 (Fig. 4), enhanced the binding of T-SO530 to αSyn oligomers (
) developed 58-nucleotide DNA aptamers they termed F5R1 and F5R2, starting with a 40-nucleotide random single-strand DNA library, which bound αSyn with a high affinity and inhibited αSyn aggregation. To enhance cell membrane permeability of the aptamers, they were modified by attachment of a peptide carrier, CADY, which had been reported previously to form stable complexes with nucleic acids, leading to improvement of their delivery into cultured cells (
). The modified aptamers reduced intracellular aggregation of αSyn, synaptic protein loss, and neuronal death caused by αSyn overexpression.
Aptamers against prion protein
The accumulation of misfolded PrP characterizes prionoses, such as Creutzfeldt-Jakob disease, Kuru, Gerstmann–Sträussler–Scheinker Syndrome, and fatal familial insomnia in humans, bovine spongiform encephalopathy in cattle, scrapie in sheep, and chronic wasting disease in deer and elk (
), yet this form can misfold due to genetic, environmental, or yet unknown causes into a toxic and infectious form called PrPSc (scrapie form). The term PrPSc does not describe one particular structure but refers to many conformational strains that are self-propagating, transmissible from cell to cell, and are infectious within the same species and sometimes across species. PrPSc infectivity to other organisms depends on the specific strain and species barrier. PrPC is rich in α-helix, whereas PrPSc contains the typical cross-β structure of amyloid fibrils. The aggregation process, including oligomerization of PrPSc, plays a central role in the prion’s propagation and neurotoxicity (
). Thus, both PrPC and PrPSc are potential therapeutic targets for development of various drug modalities, including aptamers.
Weiss et al. reported the first RNA aptamer, Ap1, against a recombinant Syrian golden hamster prion protein PrP23–231, a common model of full-length PrPC. The aptamer recognized PrPC in brain extracts of scrapie-infected mice, hamsters, and cattle, though its binding affinity was not measured (
) developed an RNA aptamer, RM312, against sheep PrP23–234, which bound with 15 nM affinity to PrPC. In another study, Ikebukuro and colleagues developed a DNA aptamer, termed “4-9”, against mouse PrP23–230 that bound to both PrPC and PrPSc equally (
) suggesting that shortening the target length in this case improved the affinity.
To pursue aptamers selective for PrPScversus PrPC, an RNA aptamer, SAF-93, was developed against scrapie-associated fibrils from the brains of infected hamsters. In a competitive binding assay using bovine PrPSc fibrils, SAF-93 showed >10-fold higher affinity for PrPSc than for PrPC (
). Uniquely, SAF-93 was obtained by SELEX using 2′-fluoro-modified pyrimidine triphosphate nucleotides together with unmodified purine nucleotides. Enzymatic probing and gel footprinting, in addition to computer-assisted secondary structure analysis, identified multiple binding sites of SAF-93 on PrPSc (
Several aptamers also have been selected against human prion proteins. An RNA aptamer termed DP7 was prepared by Proske et al. against human PrP90–129, an important domain for the conversion of PrPC into PrPSc. The aptamer also bound PrP derived from hamster and mouse. The utility of this aptamer for reducing the ratio of PrPSc to PrPC was demonstrated in prion-infected mouse neuroblastoma N2a cells (
) reported a DNA aptamer called SSAP3-10 against human PrP23–231, which bound other mammalian PrPC (sheep, calf, piglet, and deer) in addition to human PrPC, but not to PrPSc.
Structural studies of prion aptamers have been advanced compared with other amyloidogenic proteins. Murakami et al. developed an RNA aptamer, apt#1, selected against bovine PrP25–241. The binding was deduced to be due to a G-quadruplex structure using circular dichroism (CD) spectroscopy. An extracted sequence from apt#1 (R14 aptamer, GGAGGUUUUGGAGG) was identified using mutagenesis and showed improved binding affinity and ∼30-fold selectivity for PrPC compared with PrPSc (
). Katahira and colleagues further optimized the sequence to deduce a tandem repeat sequence (R12 aptamer, (GGA)4), and found using NMR measurements that it formed an intramolecular parallel G-quadruplex (
Computer-assisted development and structure prediction of aptamers
Selection and clustering of aptamers by high-throughput sequencing
Despite the large potential of aptamers as therapeutic agents, biosensors, and research tools, technical issues can require major time and money investments, limiting development and progress. One such technical issue is the use of classic Sanger sequencing of the selected oligonucleotides. Although the selection process removes the vast majority of the sequences, the final pool still may contain thousands of distinct oligonucleotide sequences, making sequencing of this pool and identifying the sequences possessing the best affinity and/or specificity labor-intensive and time-consuming.
In silico approaches, previously reviewed by Hamada (
), have been useful for aptamer design. Artificial intelligence (AI) coupled with machine learning algorithms assists in identifying potential aptamer candidates from the selected sequences expeditiously, leading to improvement compared with classical prediction tools (
). As only the oligonucleotide pool of the last round of selection typically is cloned and sequenced in conventional SELEX, superior aptamers may be overlooked. In this section, we discuss the role of bioinformatics in aptamer development, including in sequencing, secondary-structure prediction, and simulation of aptamer–target interaction, and highlight the use of these advances in selection of aptamers against amyloidogenic proteins.
With the advent of new sequencing technologies, such as Next-Generation sequencing, exhaustive parallel sequencing of all the selection rounds has facilitated high-throughput SELEX (HT-SELEX) (
). The meta-Z-score is a statistical scoring method for high prediction accuracy of binding potential. This method allows gleaning insight into whether sequence enrichment occurs, rather than whether the affinity for the target is enhanced at substantially earlier stages of the aptamer-selection process, resulting in considerable time saving. Additionally, a decreased number of selection rounds helps reduce PCR bias (nonspecific amplification) caused by over selection or excessive cycle numbers in PCR (
). Comprehensive analysis of very large sequence datasets by such bioinformatics methods has enabled not only accurate characterization of aptamers, data alignment, and clustering, but also improved prediction of aptamer structure and aptamer–target interaction mode.
After collecting and ranking datasets of sequences, data clustering typically is a next step, which can be achieved using free software tools, such as AptaCluster (
) (https://burkelab.missouri.edu/fastaptamer.html), both of which use RNA sequence, but not structural information. Ikebukuro and colleagues proposed a genetic methodology named evolution-mimicking algorithm, to identify optimized aptamer sequences in rugged sequence spaces. Using this methodology, by combining several in vitro assays, e.g., binding affinity, inhibitory activity, specificity, and 3D structure, in silico maturation that included selection and duplication, recombination, and point mutation has facilitated identifying several aptamers against PrP and αSyn (
Analysis of secondary and tertiary structure of aptamers by motif finding and structure optimization
When conducting an aptamer discovery campaign, the three-dimensional structures of the selected aptamers and aptamer–target complexes are important considerations. The relatively simpler conformational space of nucleic acids compared with proteins makes computer-aided calculation and modeling of secondary and tertiary structures particularly useful for aptamers. The secondary structure of oligonucleic acids is determined by canonical Watson–Crick base-paring interactions or noncanonical Hoogsteen base pairs in single or double strands (Fig. 6). Due to the additional hydroxyl group in ribose compared with deoxyribose, RNA single strands are believed to form more complex and diverse structures than DNA, such as stem loops and pseudoknots.
The program Mfold, originally developed by Zuker (
), uses minimal free energy optimization and is one of the most popular tools for determination of nucleic acids’ secondary structure. Typically used after Mfold, MEMERIS is a general motif-finding algorithm for integration of secondary structures (
). AptaMotif is a program designed for identifying binding motifs in aptamers identified through SELEX and is based on structural processing, including suboptimal secondary structures, for prediction of sequence motifs in the loop regions (
). A putative flow from an identified sequence to computer-aided structural optimization of aptamers is shown in Figure 7. To allow convenient selection of the appropriate tool for each task, Rtools (http://rtools.cbrc.jp/) was developed as an integrated web-server hosting existing prediction algorithms for RNA secondary structure analysis (
Structural prediction programs have begun being applied in the amyloid field. Mfold likely is most used program in the field due to its free availability. Mfold analysis of the anti-Aβ40 DNA aptamer RNV95 suggested a stem-loop structure, which was validated by CD spectroscopy measurements (
) utilized Mfold for the prediction of RNA aptamer structure against β2-microglobulin, a protein whose deposition as amyloid in joints is associated with symptoms of dialysis-related amyloidosis, to reveal an enzyme cleavage site in a stem-loop structure, facilitating the identification of the aptamer’s aptatope.
G-quadruplex is a noncanonical structure in both RNA and DNA aptamers, which, as mentioned in previous sections, has been found in several cases to play key roles in aptamer binding to amyloid–protein targets. The G-quadruplex structure includes a stable planar core comprising four guanine bases in the same plane forming G-tetrads (Fig. 4A) stabilized by π–π interactions (
), which are specific to prediction of G-quadruplexes in RNA and DNA sequences. For example, the formation of G-quadruplex in an RNA aptamer against Aβ42 protofibrils predicted by QGRS Mapper was validated by detection of a negative peak at ∼240 nm and a positive peak at ∼265 nm in the CD spectrum (
) can provide clues for the potential effects of point mutations on affinity and structure. To enhance the resistance of RNA aptamers to chemical and enzymatic degradation, inclusion of modified nucleotides, such as 2′-O-methyl, 2′-fluoro, or other unnatural nucleotides (
), who incorporated modified nucleotides originally designed to form unnatural base pairs for synthetic xenobiology, into aptamers, increasing the resistance of the oligonucleotides to degradation (Fig. 8).
Among the challenges in translating aptamers against amyloidogenic proteins into therapeutic applications is also their potential toxicity, similar to other nucleic-acid-based drugs. For example, in the development of antisense therapy, acute hepatotoxicity of 2′-fluoro-modified antisense oligonucleotides was found in 6- to 8-week-old wild-type mice, in which the oligonucleotides were administered subcutaneously at 20 mg/kg three times a week (
). In contrast, in other studies, a 2′-fluoro nucleotide-stabilized siRNA was administered subcutaneously as a single, 30-mg/kg bolus for 2 years in a rat or at daily 7.5-mg/kg injections for 5 days in healthy volunteers without apparent toxicity (
). Similarly, antisense oligonucleotides containing 2′-O-methyl nucleotides were administered to cynomolgus macaques (Macaca fascicularis) at 20 mg/kg by subcutaneous injections for 9 months and to patients with hereditary transthyretin amyloidosis in a phase 2/3 clinical study at 300 mg for 15 months without cytotoxic effects (
) and disrupt their function, leading to cytotoxicity. Thus, in addition to characterization of their binding and efficacy, aptamers intended for in vivo use must be evaluated for their safety and be applied only if they have a sufficiently large therapeutic index. In this context, an advantage of aptamers as chemotypes for drug development is the ease with which they can be modified and subjected to structure–activity relationship studies compared with small molecules or protein biologics.
Prediction of aptamer–protein interaction by machine learning and molecular docking
After determination of the secondary and tertiary structures of new aptamers, deciphering the aptamer–protein complex structure often is the next goal. Molecular docking of aptamers and target proteins on each other is a forecasting method that typically consists of searching all the potential binding modes between the pair of molecules in silico and scoring each complex based on its thermodynamic stability. As an example, a docking simulation of multiple aptamer–protein pairs was carried out to optimize the aptamer sequence using different proteins in combination either with RNA aptamers or with ribosomal RNA by Zhang et al. (
). This system of molecular docking allowed obtaining high-affinity aptamers with low false-positive and false-negative rates. This new system coupled with modeling of the tertiary structure has high promise for optimization of the interaction mode between aptamers and amyloid–protein targets. However, although these approaches have helped optimizing aptamers in different cases, such as for human embryonic stem cells, epithelial cell adhesion molecule, and cell-surface vimentin (
), they have not yet been used in the amyloid field.
Brain-targeting delivery of aptamers by exosomes and nanoliposomes
Oligonucleic acids are too large and negatively charged to pass through the blood–brain barrier (BBB). To exert the functions of amyloid-targeting aptamers in the CNS, the development of effective delivery systems is an important challenge. Since bare nucleic acid delivered systemically also would be subject to enzymatic degradation and may not even reach the BBB in sufficient amounts (
), innovative engineering of CNS-targeting delivery methods of aptamers using nanocarriers, such as exosomes or nanoliposome, has been a major focus in recent years.
Exosomes are nanovesicles, 30 to 200 nm in diameter, secreted by virtually all cell types and are thought to mediate intercellular and interorgan communication, as well as serve a mechanism for removal of cellular stressors (
). To investigate a therapeutic potential of F5R1, it was encapsuled, following polyethylenimine-assisted transfection, in exosomes isolated from the culture medium of HEK293 cells. It was then delivered into the brain of wild-type mice, which received intrastriatal injection of preformed recombinant αSyn fibrils prepared from recombinant αSyn, by taking advantage of a virus-transmission system (Fig. 9A) (
). The cells also expressed rabies virus glycoprotein (RVG) fused with lamp2B, which positions the RVG on the outer membrane of exosomes secreted by these cells. Intraperitoneal administration of the RVG-decorated exosomes containing F5R1 led to retrograde transport and transsynaptic transmission into the CNS through the axons and synapses of peripheral neurons, circumventing the BBB. The treatment decreased αSyn aggregation in the substantia nigra and ameliorated motor dysfunction in the treated mice (
). This pioneering work provided proof of concept for application of DNA–aptamer therapy delivered into the CNS using an innovative delivery system that bypasses the BBB, resulting in reduction of the aggregation, transmission, and toxicity of αSyn in a preclinical model of synucleinopathy.
A different aptamer-delivery method into the CNS using nanoliposomes was reported by McConnell et al. (Fig. 9B) (
). Like exosomes, nanoliposomes are nanovesicles, yet they are prepared synthetically, as opposed to the natural origin of exosomes, allowing a greater degree of control over their composition and cargo. Similar to exosomes, the nanoliposomes are enclosed by a lipid bilayer surrounding trapped water in which hydrophilic compounds can be transported. McConnell et al. encapsuled a dopamine-specific DNA aptamer, Da20m, into nanoliposomes conjugated on their surface by a different aptamer, specific for the transferrin receptor (TfR). TfR is expressed on endothelial cells in the luminal surface of the BBB. Binding of the anti-TfR aptamer to TfR followed by clathrin-mediated transcytosis of TfR allowed delivery of the nanoliposomes across the BBB and release of the antidopamine aptamer cargo on the CNS-parenchymal side. Compared with unencapsulated Da20m, which had little or no effect, systemic administration of Da20m via the functionalized nanoliposomes potently attenuated abnormal behavior of wild-type mice triggered by aberrantly increased extracellular dopamine following exposure to cocaine (
). Encapsulation of RNA (and DNA) in liposomes promotes the biosafety and stability of the exogenous nucleic acids inside the body. Both exosomes and nanoliposomes are attractive delivery systems for aptamers. Exosome cargo and surface markers are easier to manipulate genetically, compared with those of nanoliposomes, yet nanoliposomes can be produced more easily on large scales (
). In addition, because exosomes are produced in cells and may contain unintended biological material, it is crucial to establish ways of controlling and monitoring their clinical safety if they are going to be approved for therapeutic applications (
). Although exosomes are less immunogenic than cell-based therapies, a concern in using them as drug carriers is that they may be recognized as nonself by the immune system and induce an immune response, potentially causing damaging inflammation. Should this become a problem, it could be overcome by extracting the patient’s own cells and preparing the exosomes using the patient-derived cells. Though more time-consuming and potentially expensive, such a procedure could be justified for life-saving treatments.
Conclusions and future perspectives
In 2004, the first aptamer drug, the RNA aptamer Macugen (pegaptanib sodium), which acts as an antagonist of vascular endothelial growth factor, was approved by FDA for treatment of age-related macular degeneration. Several additional aptamers have been tested in clinical trials for cancer, cardiovascular disease, and eye diseases, yet to date, despite the growing number of reports on antiamyloid aptamers and the advances in their design and delivery discussed above, aptamers have not yet advanced to clinical trials in the neurodegenerative disease field.
To advance the development of aptamers targeting amyloidogenic protein for biomedical applications, several issues should be addressed. First, the conformational metastability and heterogeneity of the targets, particularly oligomers of amyloidogenic proteins, are major impediments for aptamer selection. This challenge can be addressed by covalent tethering of the protein monomers to create a more stable aptagen that mimics the metastable oligomers, as was demonstrated in the case of Aβ42 (
Second, isolating aptamers with high affinity and specificity for the desired target is a labor-intensive and time-consuming process that is always accompanied by a level of uncertainty. Next-Generation sequencing and structural prediction programs coupled with AI and machine learning algorithms are promising approaches for increasing both the speed of the process and the likelihood of success. Application of these modern techniques has been limited in the amyloid field but is beginning to catch up. Additional advances are expected when detailed structures of aptamers and their complexes with their respective targets become available and support structure prediction algorithms. To our knowledge, to date, only two X-ray crystallography studies reporting such structures have been published (
). However, the increasing popularity of high-resolution structural determination by cryo-electron microscopy suggests that additional data will be forthcoming. High-resolution structures and improved structural prediction technology not only will facilitate obtaining improved aptamers, but also will help clarify the molecular basis of target recognition by aptamers, including in the amyloid field.
Finally, the inability of aptamers to cross the BBB on their own can now be addressed by at least two delivery systems, including exosomes and nanoliposomes, as discussed above. Each approach has its advantages and limitations, yet we expect that future research will address the current shortcomings and lead to successful application of these techniques in delivering aptamers targeting amyloidogenic proteins into the CNS, as has been demonstrated recently for aptamer F5R1 in a mouse model of synucleinopathy (
). Recent advances in both computational and experimental approaches suggest that using aptamers as antiamyloid therapeutics is achievable in the near future.
Conflicts of interest
The authors declare that they have no conflicts of interest with the contents of this article.
K. M. and G. B. conceptualization; K. M. supervision; K. M. and N. I. writing—original draft; G. B. writing—review and editing.
Funding and additional information
This study was supported in part by the JSPS KAKENHI, grant number 20KK0126 to K. M. and N. I, and by NIH / NIA grant RF1AG054000 , The Alzheimer's Association , The Michael J. Fox Foundation , Weston Brain Institute , and Alzheimer’s Research UK Biomarkers Across Neurodegenerative Diseases (BAND 3) grant 17990 , and CurePSP grant 665-2019-07 to G. B.
A new era for understanding amyloid structures and disease.