Nucleobase Amino Acids in the Development of Amide‐Linked Oligonucleotide Mimics

Abstract

Although peptide nucleic acids (PNAs) have been extensively used in antisense strategies, the lack of any side chain in the PNA backbone deprives them of having any protein‐like tertiary structures, limiting their use in nucleic acid–protein interactions. To overcome this limitation, scientists have developed hybrid monomeric structures like nucleobase amino acids (NBAs) where canonical nucleobases are covalently attached to natural or unnatural amino acid side chains, making them amenable to incorporation into peptides, proteins, and PNAs using solid‐ or solution‐phase peptide synthesis. The resulting assemblies are capable of recognizing their complementary oligonucleotides as well as having the ability to form well‐defined secondary/tertiary protein‐like structures for enhanced nucleic acid–protein interactions. This perspective article chronicles the developments in this area of research over the years.

This perspective summarizes the synthesis of various nucleobase‐functionalized amino acids, termed as nucleobase amino acids (NBAs), which are utilized as building blocks for amide‐linked oligonucleotide mimics. The resulting oligomers combine enhanced backbone stability with complementary nucleic acid recognition properties, as well as the ability to form well‐defined secondary/tertiary protein‐like structures for enhanced nucleic acid–protein interactions.

1. Introduction

Targeting cellular RNAs for therapeutic interventions is being vigorously pursued worldwide in recent years [1]. One such approach involving the antisense principle has emerged as an asset in the development of potential therapeutic agents by controlling protein synthesis at the nucleic acid level [2]. The short (12–24‐mers) chemically modified antisense oligonucleotides (ASOs) modulate the function of cognate mRNA in cells by Watson–Crick base‐pairing. ASO molecules typically consist of short single DNA and/or RNA sequences designed to modify protein expression [3]. An ideal ASO must be sufficiently stable under physiological conditions and should possess a high binding affinity to the target RNA, high nuclease resistance, binding selectivity to transport proteins, and cell permeability in vivo [4].

From the early pioneering works of Zamecnik and Stephenson who designed a tridecamer oligo d(A‐A‐T‐G‐G‐T‐A‐A‐A‐A‐T‐G‐G), complementary to 13 nucleotides of the 3′‐ and 5′‐reiterated terminal sequences of Rous sarcoma virus 35S RNA, inhibiting its replication in chicken embryo fibroblasts [5, 6], the field has witnessed exponential growth in the last five decades [7]. Currently, there are over a dozen ASO‐based drugs approved by the Food and Drug Administration (FDA), or the European Medicines Agency (EMA), for wide‐ranging therapeutic applications against many genetic disorders and infectious diseases [8, 9].

Naturally occurring oligonucleotides are readily degraded by nucleases in vivo; therefore, chemical modifications become necessary to address these shortcomings [10]. The most common of them, backbone modifications wherein the natural phosphodiester linkage is replaced with a phosphorothioate (PS, 1 and 2, Figure 1) linkage, provide high resistance to nuclease degradation, increasing serum protein binding, and are used in nearly all approved oligonucleotide drugs [11].

FIGURE 1.

Some examples of backbone and sugar modifications of naturally occurring oligonucleotides.

Other backbone modifications include those where the nonbridging oxygen is replaced by a borane group or a methyl group, resulting in a charge‐neutral backbone that enhances cellular penetration. Sugar modifications include 2′‐OMe (3), 2′‐O‐methoxyethyl (2′‐MOE, 4), 2′‐F (5), locked nucleic acids (LNA, 6 and cEt, 7) with a methylene bridge connecting the 2′‐oxygen and 4′‐carbon in a rigid sugar conformation, etc., that confer higher binding affinity and increased stability to the ASOs [12].

There are also many nucleobase modifications, for example, adenine methylation, cytosine methylation, uridine methylation, guanidine oxidation, etc., that enhance binding affinity and decrease immune recognition (8 and 9, Figure 2).

FIGURE 2.

Some examples of nucleobase modifications.

Another important backbone alteration is the replacement of the phosphodiester linkages with amide bonds in many ribose‐based constructs (10–12, Figure 3) [13]. Synthesis of fully amide‐linked RNA analogs of short RNA oligos carrying all four canonical nucleobases has been synthesized using standard solid‐phase Fmoc‐chemistry [14]. Although the glycosidic bond strength in these amide‐linked mimics was found to be slightly weaker compared to natural ribonucleosides under acidic conditions, the high predilection toward N‐type conformations of the sugar rings in these molecules [15] could have a positive leverage toward further development of these fully amide‐linked short oligos, especially for seed sequences.

FIGURE 3.

Some examples of amide‐linked mimics of phosphodiester linkages in DNA and RNA.

Amide‐linked mimics of phosphate linkages within A‐type RNA helices have shown good results in maintaining proper hydrogen bonding and base stacking, forming stable A‐form duplexes. They are also highly resistant to nuclease and protease digestion, thus enhancing their stability in serum and cell extracts compared to natural RNA. When used in siRNAs, the amide linkages are well‐tolerated in the seed and central regions [16]. They can increase RNA interference (RNAi) activity and specificity, especially when introduced near the 5′‐end of the passenger strand [17].

Complete replacement of the ribose ring is seen in peptide nucleic acids (13, PNAs) that are amide‐linked oligomers of N‐(2‐aminoethyl)‐glycine carrying nucleobases through Gly N‐acylation (Figure 4) [18, 19]. The semi‐rigid PNA backbone attaches itself to complementary base pairs on DNA or RNA in a sequence‐dependent manner. PNAs and their analogs are endowed with exceptionally high affinity and specificity for receptor sites, essentially due to the uncharged and flexible polyamide backbone.

FIGURE 4.

PNAs and their synthesis.

PNAs have been extensively studied for potential therapeutic applications involving the antisense strategy. Due to their stability toward nucleases, ability to recognize their complementary oligonucleotides, high affinity due to the uncharged nature of the peptide backbone, and their ease of synthesis, PNAs serve as very attractive candidates for antisense technology as well as other areas of research [20].

However, the lack of any side chain in the PNA backbone deprives them of having any protein‐like tertiary structures. Several studies reported structurally modified PNA analogs having α, β and γ‐substituents (14, Figure 5) to study their mode of binding and binding affinity for natural nucleic acids and their use in medicinal chemistry as potential miRNA binders [21]. As an example, a PNA analog with γ‐backbone modification preorganized PNA into a helical structure [22].

FIGURE 5.

PNA monomers and side‐chain‐containing modified PNA monomers.

2. Nucleobase Amino Acids (NBAs)

To have molecules capable of recognizing their complementary oligonucleotides as well as having the ability to form well‐defined secondary/tertiary protein‐like structures for an enhanced nucleic acid–protein interactions, scientists have developed hybrid structures like NBAs [23, 24, 25, 26] in which nucleobases are covalently attached to the side chain of natural or unnatural amino acids that can easily be incorporated into PNAs, proteins or peptides using standard solid‐ or solution‐phase methods (15, Figure 6).

FIGURE 6.

PNA and NBA‐PNA hybrids.

The resulting NBA‐containing PNAs, proteins, or peptides interact with their nucleic acid substrates using Watson–Crick and other base–pairing interactions. Several NBAs have been reported in the literature; among them, the prominent ones are alanyl‐based NBAs (16–19, Figure 7) synthesized from serine lactone following multistep synthetic routes. Alanyl NBAs have been widely used for developing polypeptides for controlling and enhancing DNA–protein interactions. Mihara's group has done extensive amount of work on NBAs, prepared Ala and hAla NBAs and incorporated them into peptides by chemical synthesis to enable RNA binding studies [23, 24, 25]. The NBA‐containing peptides were also studied for their 3D structures and binding with cognate DNA/RNA. For example, to study the applicability of an NBA moiety in peptides, a nucleobase‐conjugated peptide, derived from HIV‐1 Rev, was designed and synthesized. The α‐helix conformation of the Arg‐rich domain (34–50) of the Rev protein (Rev _34–50 , Figure 7) binds specifically to the RRE RNA stem‐loop IIB region. The α‐helix conformation of the Rev _34–50 peptide contributes to its binding affinity and specificity to RRE IIB RNA. NMR structural studies revealed that the Gln36 residue in the Rev _34–50 peptide comes close to the guanine‐48 base in the internal loop region of the RRE IIB RNA, enabling the binding. That is why the NBA‐conjugated analog of Rev _34–50 was prepared by replacing the Gln36 with C_NBA (Q36C _NBA in Figure 7), and to compare this analog with its PNA‐based congener, analog Q36C _PNA was also prepared by replacing two amino acids Q36A37 of Rev _34–50 with C _PNA , as the PNA is a dipeptide isostere. Circular dichroism studies were carried out to determine the structures of these peptides in different solvents. Percent helicity of the 3 compounds in trifluoroethanol (TFE) is shown in Figure 7. The binding properties of the peptides with RRE IIB RNA were evaluated by competition assay using the Rev peptide modified with 5‐carboxytetramethylrhodamine at the N‐terminal (Rhod‐Rev) as a fluorescence tracer. Even though Q36C _NBA had almost the same α‐helix potential as Rev _34–50 , its binding to RRE IIB RNA was 2.0‐fold stronger than the native peptide, suggesting a possible H‐bonding interaction between the cytosine moiety of the NBA‐containing peptide analog and the complementary G‐48 of the RNA. On the contrary, the α‐helicity of the PNA‐containing analog Q36C _PNA was decreased to 41% in the same TFE solution, resulting in much decreased binding with the RNA. These results indicate that the flexible and achiral PNA disturbed the α‐helix structure, an essential factor for having better RNA‐binding, giving an edge to the NBA‐conjugated peptides [23].

FIGURE 7.

Mihara's work on NBA‐containing coiled‐coil peptides.

To explore the applicability of nucleobase complementarity in a peptide structure, Mihara's group carried out a study wherein A–T nucleobase pairs were incorporated in coiled‐coil peptides (T _NBA  = 17, A _NBA  = 18, Figure 7) that were found to make an effective contribution to the formation of stable α‐helical structures (bottom, Figure 7) [24]. The CD spectra of the designed peptides (TA‐1, TA‐2, and TA‐1.2) in a buffer (pH 7.4) at 25 °C showed negative maxima near 208 and 222 nm and a positive maximum at 195 nm, characteristic of α‐helical conformation. The ratio of the molar ellipticities at 222 and 208 nm was reduced in a solution containing trifluoroethanol, known as a solvent enhancing an α‐helix form but isolating a helix‐dimer, indicating two‐stranded α‐helical formation in the buffer. In another interesting study, the same group incorporated NBAs as additional recognition elements in coiled‐coil peptides for enhancing interstrand recogniotions [25]. The effect of the base–pair interaction on intermolecular recognition between peptides was evaluated through a self‐replication reaction. Peptides with complementary base pairs such as T‐A or G‐C at the g–g′ heptad positions showed accelerated self‐replication in comparison with those having mismatched base pairs or without NBAs.

Diederichsen's group published MMT/acyl‐protected NBAs (20–23) for the solid‐phase synthesis of DNA/alanyl‐PNA chimeras under conditions fully compatible with the standard phosphoramidite DNA synthesis strategy (Figure 8) [26].

FIGURE 8.

MMT/acyl‐protected alanyl‐NBAs.

Several other such studies were published. In 1995, Taddei's group reported the synthesis of homoalanyl (hAla) NBAs by nucleophilic substitution reaction of a Boc‐Glu‐OMe‐derived sidechain brominated precursor with nucleobases to give the hAla NBAs carrying all 4 nucleobases (24, Figure 9) [27].

FIGURE 9.

Taddei's work on the synthesis of NBAs.

For the synthesis of prolyl NBAs, G. Lowe's group started from (4R)‐4‐hydroxyproline (4R‐Hyp), the tosylate of which was used to make the prolyl NBAs by nucleophilic substitution as well as by a Mitsunobu reaction of the Boc‐Hyp‐OMe (25, Figure 10) [28].

FIGURE 10.

Lowe's work on the synthesis of Boc‐Pro‐OMe substituted at the 4‐position with all four canonical nucleobases.

Besides the α‐amino acid‐derived NBAs, nucleobases have also been incorporated into the side chains of β‐amino acids, and oligomers of these β‐amino acid‐derived NBAs have been prepared [29]. A novel β‐amino acid having (thymine‐1‐yl)methyl functionality at the α‐position (26, Figure 11), a useful component for making α‐substituted β‐homoalanyl peptide nucleic acids (β²‐PNAs), was synthesized via an enzymatic desymmetrization catalyzed by lipase PS as a key step [29]. For the synthesis of pyrimidinyl‐nucleo‐β³‐amino acids (27, Figure 11), besides the Mitsunobu reaction for the attachment of the nucleobases to the γ‐position of β‐homoalanine, nucleophilic substitution of a methanesulfonate was also used by Diederichsen's group [30]. For the preparation of the guaninyl‐β³‐amino acid (28, Figure 11), an efficient nucleo‐β‐lactam route was established. Using these β³‐amino acid‐derived NBAs, β³‐homoalanyl‐PNA G‐C hexamers (GGCGCC and GCGCGC) were synthesized that formed quite stable Watson–Crick base‐pairing with a double‐stranded helical topology [30]. This was extended later in another study where a common bromide precursor, derived from Boc‐L‐aspartic acid 4‐benzyl ester, was used to prepare the pyrimidinyl‐ and purinyl‐β³‐amino acid NBAs with all four canonical nucleobases linked by an ethylene spacer to the β³‐amino acid backbone by a nucleophilic substitution method. Using these pyrimidinyl‐ and purinyl‐β³‐amino acid NBAs, complementary oligomers 29 and 30 were synthesized by solid‐phase peptide synthesis and were studied by temperature‐dependent CD and UV spectroscopy for interstrand bindings [31]. The base‐pair recognition units were carefully chosen and incorporated in specific sites along with other β³‐ and β‐amino acids to facilitate the formation of very stable triply hydrogen‐bonded G‐C Watson−Crick base pairs. The resulting β‐peptides were shown to adopt a 14‐helical conformation in aqueous solution, having approximately three residues per turn. This led to the required spacing among nucleo‐β³‐amino acid residues in these oligomers to have an alignment of the bases along one side of the 14‐helical conformations to facilitate interstrand bindings. However, the self‐pairing and the 1:1 pairing of these β‐peptides were found to be too low.

FIGURE 11.

Various β‐amino acid NBAs, their synthesis and uses in the synthesis of oligomers.

Diezemann's group prepared norvalyl NBAs by the alkylation method, similar to that reported earlier by Taddei's group (Figure 9) [27], from glutamic acid‐derived bromides, which were subjected to nucleophilic substitution using the nucleobases [32]. They prepared NBA‐based short peptide sequences (Figure 12), both homooligomers and heterooligomers, using monomeric alanyl‐NBA (AlaG, AlaC, etc.), homoalanyl‐NBA, norvalyl‐NBA, and also using similar NBAs but with α‐D‐configuration (shown in italics). Side chain homology in combination with a linear double‐strand topology turned out to be valuable for selective manipulation of pairing selectivity (pairing mode) and base pair stacking leading to rigid and structurally well‐defined double strands, a model system comparable with DNA nucleobase stacking. The stabilities of these double‐stranded NBA‐based oligos were determined in aqueous buffer solution (pH 7.0) by temperature‐dependent UV spectroscopy. Since the duplex separation into single strands is a cooperative process, de‐stacking could be detected by an increase in absorption. The temperature at the turning point of the sigmoidal curve indicated the stability of the complex. The pairing selectivity depended on the types of the nucleobase and their orientation, which in turn was governed by the conformation of the backbone and the side chain linker length. The peptide backbone of alanyl‐NBA‐based peptide duplexes showed a β‐sheet conformation forced by double‐strand formation (Figure 12). The best stability was found in the enantiomeric oligomers 31 and 32 pairing in an antiparallel double strand in the Watson–Crick mode (T _m = 58 °C), which was more stable than the self‐pairing antiparallel double strand of oligomer 31 having reverse Watson–Crick recognition with two hydrogen bonds (T _m = 40 °C). The consecutive side chains with alternating configuration with a distance of 3.6 Å favored stacking distance of nucleobase pairs. The structures and stabilities of other NBA‐based oligos controlled by their base pairing abilities for both homooligomers and mixed‐base oligomers are discussed in detail in the paper. The NBA‐based peptides with longer side chains were less preorganized, entropically lowering the double‐strand stability. Overall, the study offers very valuable model systems and tools for investigating various biochemical applications of base stack‐mediated processes like intercalation, electron transfer, and interactions with the base stack.

FIGURE 12.

Diezemann's work on alanyl (D‐isomer in italics), homoalanyl and norvalyl NBA‐based short oligomers.

The asymmetric syntheses of alanyl NBAs were achieved by Hecht's group in 2016 by nucleophilic ring opening of N‐Boc‐L‐serine β‐lactone on treatment with suitably protected nucleobases in the presence of DBU using DMSO as solvent, like the synthesis of 28 (Figure 11) [33]. They synthesized five alanyl NBAs 33–37 (Figure 13) containing all canonical DNA‐RNA nucleobases, cytosine, uracil, thymine, adenine, and guanine, protected in a fashion which enabled their attachment to a suppressor tRNA, and their incorporation into E. coli dihydrofolate reductase (DHFR) and also into the Klenow fragment of E. coli DNA polymerase I, a DNA‐binding protein. The pdCpA derivatives of alanyl NBAs were prepared to synthesize the requisite aminoacylated tRNAs like 38 for protein synthesis. The protein sites modified were position 10 of E. coli DHFR and 484 of the Klenow fragment of E. coli DNA polymerase I. Pyrimidine‐based alanyl NBAs exhibited better suppression efficiencies than did purine‐based alanyl NBAs.

FIGURE 13.

The pdCpA derivatives of alanyl NBAs 33–37 and a suppressor tRNA_CUA (38) activated with an NBA.

Pentulate's group reported NBA‐containing combinatorial libraries with up to 100 million biohybrid molecules [34]. This biohybrid material showed a higher bulk affinity to oligonucleotides than peptides composed exclusively of canonical amino acids. From the 100 million‐membered NBA‐containing peptide library, they enriched a nanomolar binder (39) to RNA hairpins by MS/MS decoding (Figure 14), with selectivity over DNA. The affinity selection mass spectrometry (AS‐MS) workflows developed by them enabled high‐throughput screening of synthetic combinatorial peptide libraries having the flexibility to incorporate virtually any noncanonical amino acid. The AS‐MS strategy, essentially a “tag‐free” approach, required efficient synthesis and high‐fidelity MS/MS decoding. Twelve NBA monomers, three of them shown in Figure 14 (40–42) with different linkers connecting each of the four nucleobases to the amino acid moiety, were synthesized by already reported procedures and incorporated into peptides by SPPS.

FIGURE 14.

Pentulate's work on the synthesis of NBA‐containing biohybrid libraries by combinatorial method and RNA binding analysis.

Sanjayan's group has developed a novel triple G‐C‐T NBA building block featuring three recognition faces: DDA (G mimic), DAA (C mimic), and ADA (T mimic) (43 and 43′, Figure 15). Crystal structure studies with a model system confirmed the hypothesis that the molecule can indeed exist in both G and C forms, presumably owing to a prototropy shift, for an eventual G‐C duplex [35].

FIGURE 15.

Triple G‐C‐T base‐coded nucleobase amino acid NBA 43. Tautomerism leading to “G‐C inversion” is shown in 43′.

3. Summary and Outlook

Since their discovery by Peter E. Nielsen and colleagues in 1991, PNAs have emerged as a significant advancement in medical science. Among the numerous gene‐editing technologies explored over the years, PNAs are considered a safe, versatile, and clinically translatable platform with considerable therapeutic promise. The isolation and characterization of the earliest naturally occurring amino acid derivatives of nucleobases inspired researchers to incorporate NBAs into polypeptide or polyamide backbones and investigate their intrinsic capacity for the selective and specific recognition of natural nucleic acids. Owing to their ability to recognize double‐stranded DNA and RNA, NBA‐PNA mimics have become valuable tools in the detection and diagnosis of various disorders. Furthermore, it is anticipated that such “nucleoproteins”, whose biological functions normally involve DNA or RNA binding, can be selectively directed toward specific nucleic acid regions through nucleobase interactions that were previously considered exclusive to natural nucleic acids. The versatile nature of PNA chemistry not only enables structural optimization of PNAs but also facilitates the design and incorporation of modified NBAs at selected sites for targeted applications involving both PNA backbones and natural DNA. These modified NBA‐PNA mimics exhibit a broad spectrum of biomedical and diagnostic applications. Moreover, they possess significant potential to function as peptide mimics due to their unique intrinsic capability to encode information within amino acid sequences. This perspective chronicles the diverse modifications undertaken to develop NBA‐containing novel and structurally modified PNA mimics, as well as synthetic oligos of choice with enhanced functionalities and the dual ability to mimic both peptides and nucleic acids simultaneously. PNAs with backbone modifications by insertion of NBAs are expected to contribute to the development of advanced PNA surrogates with improved activity, selectivity, and substantial biomedical relevance. The NBA oligos also have tremendous potential in developing novel gene‐editing technologies. However, further extensive investigations and research efforts are necessary to fully realize their wide‐ranging potential.

Disclaimer

The opinions expressed in this publication are the view of the author(s) and do not necessarily reflect the opinions or views of ChemMedChem, the Publisher, ChemistryEurope, or the affiliated editors.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

Data sharing not applicable to this article as no datasets were generated or analyzed during the current study.