2025 年 73 巻 10 号 p. 938-943
Guanine quadruplexes (G4s) are non-canonical nucleic acid structures that have emerged as attractive therapeutic targets owing to their involvement in diverse biological processes. Additionally, peptides derived from G4-binding proteins provide promising platforms for selective G4 recognition. In this study, we explored the G4-binding capacity of arginine–glycine–glycine (RGG)-rich sequences derived from the RGG3 domain translocated in liposarcoma/fused in sarcoma (TLS/FUS), a known G4 RNA binding protein. In this study, we synthesized a library of overlapping 15-mer peptides and evaluated their G4-binding affinities. Among the 10 evaluated native sequences, several peptides demonstrated measurable affinities toward G4 RNA structures, with STK5-1 exhibiting the highest G4-binding affinity. Furthermore, to investigate the impact of conformational constraints on G4 recognition, we introduced (E)-methylalkene dipeptide isosteres (MADIs) into selected Gly–Gly motifs, generating a series of RGG peptidomimetics. Subsequent binding assays revealed that some of these MADI peptidomimetics exhibited enhanced affinity and selectivity compared with their unmodified counterparts. Our findings offer new insights into the sequence and structural features governing G4-binding, establishing a foundation for the further development of peptide-based G4 ligands.
Guanine quadruplexes (G-quadruplexes, G4s) are non-canonical (secondary) nucleic acid structures that are formed by guanine-rich sequences through Hoogsteen hydrogen bonding and characterized by stacked planar G-quartets.1,2) These structures are widely distributed in functionally essential regions of the human genome and transcriptome, including telomeres, gene promoters, and untranslated regions. According to the findings of numerous studies, G4s participate in various biological processes, including telomere maintenance,3,4) DNA replication,5,6) transcriptional regulation,7,8) RNA processing, and translation.9,10) Given their functional significance, G4s have attracted significant attention as therapeutic targets and as molecular tools for elucidating gene regulation and disease mechanisms, particularly in cancer and neurological disorders.
Although small-molecule G4 ligands exhibited high binding affinities, they often lack sequence and structural selectivities because of their limited molecular surface areas. This constraint mostly restricts their interactions primarily to the terminal G-tetrads of G-quadruplexes, thus complicating discrimination among the diverse G4 topologies present in the genome.11,12) The structural polymorphism of G4s, including variations in their strand orientations, loop configurations, and groove dimensions, further complicates the design of small molecules that can selectively target specific G4 conformations. Overall, these limitations can induce off-target effects and lower the therapeutic efficacy of small-molecule G4 ligands.
G4-binding proteins exhibit high specificity and affinity for G4 structures owing to their expansive molecular surfaces, which facilitate multivalent interactions with diverse G4 topologies.13) However, their large sizes and structural complexities pose significant challenges for therapeutic applications. These challenges include poor cell permeability, as well as susceptibility to proteolytic degradation and challenging in large-scale synthesis and formulation. These limitations underscore the need for alternative G4-targeting strategies that integrate protein specificity with enhanced pharmacokinetic properties.
The intermediate molecular sizes and structural diversities of peptides makes them promising alternatives for targeting G-quadruplex (G4) structures.14,15) Peptide-based G4 ligands can circumvent several limitations of small molecules and proteins, thereby offering advantages such as chemical accessibility, modifiability, and cell penetration potential.16–21) Among such peptides for developing G4 ligands, arginine–glycine–glycine (RGG) motifs are particularly remarkable. These motifs represent the second most abundant RNA-binding domains in the human genome, after the RNA-recognition motif, and are crucial for G4 interactions owing to their intrinsic affinities for these structures.22,23) For instance, the RGG3 domain of Ewing’s sarcoma (EWS), comprising 112 amino acids, has been shown to bind and stabilize both DNA and RNA G4s.24,25) Similarly, the RGG3 domain of translocated in liposarcoma/fused in sarcoma (TLS/FUS), comprising 78 amino acids, exhibits comparable binding properties.26)
Notably, shorter RGG peptides also demonstrate functional versatilities. For instance, an 18-mer peptide from a fragile X mental retardation protein (FMRP) binds the duplex–quadruplex junction by adopting a β-turn structure.16,17) Additionally, a 25-mer peptide derived from a cold-inducible RNA-binding protein (CIRBP) interacts with a single G-quartet plane of a G4 structure.18) These findings highlight the versatility of RGG motifs as modular elements for targeting G4, offering valuable insights into the design of peptide-based G4 ligands.
Here, we report a peptide-based approach for targeting RNA G4s. This strategy harnesses the intrinsic G4-binding ability of the RGG3 domain of TLS/FUS. Guided by the modular architecture of this domain, we designed a focused set of peptides to probe the sequence elements driving G4 recognition.
To evaluate the impact of conformational rigidity on G4-binding, we further incorporated (E)-methylalkene dipeptide isosteres (MADIs) in place of flexible Gly–Gly dipeptide motifs to promote the formation of β-turn, a conventional structural feature of RGG-rich sequences. This strategy provided a platform for investigating the influence of backbone modifications on the affinity and selectivity of peptide–G4 interactions.
To identify G4-binding peptides from the RGG3 domain of TLS/FUS, we designed an overlapping peptide library spanning residues 449–508 of the protein (Fig. 1). Studies revealed that this domain selectively binds to RNA G4 structures, such as those in telomeric repeat-containing RNAs (TERRA). The library divided the sequence into 15-residue peptides, each with a 10 amino acid overlap, to preserve the potential secondary structures. This approach yielded 10 distinct RGG peptides. All the peptides were synthesized using standard 9-fluorenylmethyloxycarbonyl (Fmoc)-based solid-phase peptide synthesis (Fmoc-SPPS) and purified by reverse-phase HPLC.
Filter-binding assays were performed to determine the binding affinity of the native RGG peptides for G4 RNA and single-stranded RNA (ssRNA). Among the 10 peptides evaluated, STK5-1 exhibited the highest G4-binding affinity, with an equilibrium dissociation constant (Kd) of 1.47 μM (Table 1). Additionally, peptides STK8-1 and STK9-1 exhibited measurable binding affinities to the G4 structures, with Kd values of 7.29 and 7.78 μM, respectively. Conversely, peptide STK6-1 exhibited weak but detectable G4-binding affinity (Kd = 28.5 μM), whereas the remaining peptides displayed negligible affinities at concentrations of up to 30 μM.
| Peptide | Peptide sequence | Kd (μM)a) | Selectivity Index (SI) |
Amino acid features | |||||
|---|---|---|---|---|---|---|---|---|---|
| G4 RNA | ssRNA | Aromatic | RGGs | Base | Acid | Total charge | |||
| STK1-1 | H-APKPDGPGGGPGGSH-NH2 | >30 | 18.8 | – | 1 | 0 | 2 | 1 | ++ |
| STK2-1 | H-GPGGGPGGSHMGGNY-NH2 | >30 | >30 | – | 2 | 0 | 1 | 0 | ++ |
| STK3-1 | H-PGGSHMGGNYGDDRR-NH2 | >30 | >30 | – | 2 | 0 | 2 | 2 | ++ |
| STK4-1 | H-MGGNYGDDRRGGRGG-NH2 | >30 | >30 | – | 1 | 2 | 3 | 2 | ++ |
| STK5-1 | H-GDDRRGGRGGYDRGG-NH2 | 1.47 | 0.82 | 0.56 | 1 | 3 | 4 | 3 | ++ |
| STK6-1 | H-GGRGGYDRGGYRGRG-NH2 | 28.5 | >30 | – | 2 | 2 | 4 | 1 | ++++ |
| STK7-1 | H-YDRGGYRGRGGDRGG-NH2 | >30 | >30 | – | 2 | 3 | 4 | 2 | +++ |
| STK8-1 | H-YRGRGGDRGGFRGGR-NH2 | 7.29 | 2.14 | 0.29 | 2 | 3 | 5 | 1 | +++++ |
| STK9-1 | H-GDRGGFRGGRGGGDR-NH2 | 7.78 | 10.9 | 1.4 | 1 | 3 | 4 | 2 | +++ |
| STK10-1 | H-FRGGRGGGDRGGFGP-NH2 | >30 | >30 | – | 2 | 3 | 3 | 1 | +++ |
a) The Kd values were determined using the filter-binding assay.
To gain insight into the sequence features driving G4 binding, we first examined the distribution of RGG repeats across the peptide library. The strongest binders—STK5-1, STK8-1, and STK9-1—contained multiple RGG motifs and exhibited higher overall cationic characters compared with the others. The arginine residues likely facilitated binding through electrostatic interactions with the RNA phosphate backbone, as well as cation–π and hydrogen-bonding interactions with nucleobases. However, several peptides containing RGG motifs, such as STK6-1, STK7-1, and STK10-1, exhibited weak or negligible binding affinities, indicating that the mere presence of RGG repeats was not sufficient. Thus, the positioning of these motifs and the nature of the intervening residues might have influenced G4 recognition.
Next, we investigated the role of the negatively charged residues. As previously reported, acidic amino acids can interfere with G4 binding by disrupting favorable electrostatic interactions.18) For example, STK7-1 contains aspartic acids, which may diminish the binding efficiency, at both termini. Although STK5-1 also contains acidic residues, it retains strong affinity, indicating that the overall binding efficiency was modulated by an equilibrium between the acidic and basic side chains.
Furthermore, aromatic residues, such as tyrosine and phenylalanine, in the RGG3 domain of TLS/FUS, may facilitate G4 recognition through hydrogen bonding and π–π interactions, as confirmed by their presence in STK5-1 to STK9-1. In STK10-1, the C-terminal positioning of phenylalanine may limit its effective interaction.
These observations are further supported by the annotations provided in Table 1. Peptides such as STK5-1 and STK8-1, which contain a relatively high number of basic residues (total charge = +5), exhibited strong G4-binding affinities but limited selectivity over single-stranded RNA. One possible explanation is that the excess positive charge promotes nonspecific electrostatic interactions with both structured and unstructured RNAs. By contrast, peptides with fewer basic residues, such as STK6-1 and STK10-1, displayed lower G4 affinity, suggesting that a certain threshold of positive charge is necessary but not solely sufficient for strong and selective G4 binding.
In addition to charge contributions, the presence and positioning of aromatic residues such as Tyr and Phe also appear to play important roles in G4 recognition. Previous reports have highlighted the significance of Tyr479, Tyr484, and Phe494 in the RGG3 domain of TLS/FUS for TERRA G4 binding.27–29) Consistent with these findings, peptides containing these residues showed measurable G4-binding affinity in our library. These aromatic side chains may facilitate G4 interaction through hydrogen bonding and π–π stacking with the G-quartet planes. Together, these results suggest that both the distribution of positive charges and the inclusion of key aromatic residues are essential for the effective recognition and stabilization of G4 structures by RGG peptides.
To elucidate the interaction mechanism of a series of RGG peptides with G4 RNA, we investigated the secondary structures of these peptides by circular dichroism (CD) in a 10 mM Tris–HCl buffer solution (pH 7.5) (Fig. 2). Our CD analysis revealed that the 10 peptides adopted largely unstructured conformations under the implemented experimental conditions. These findings indicate that a defined secondary structure was not relevant for G4 recognition; rather, the spatial arrangements of key residues may be the dominant factor. Overall, these results highlight the importance of both the motif content and the sequence context in tuning G4-binding affinity.
(A) CD spectra of Peptides STK1-1–STK5-1, represented by brown, red, gray, green, and yellow lines, respectively. (B) CD spectra of Peptides STK6-1–STK10-1, represented by black, blue, orange, purple, and light-blue lines, respectively.
Although several native peptides exhibited measurable G4 binding affinities, their largely unstructured nature and sequence-dependent variabilities indicated that their binding efficiencies might be sensitive to conformational dynamics. Given the prevalence of β-turn motifs in the RGG-rich sequences, as well as their potential roles in G4 recognition, we sought to introduce conformational constraints to stabilize favorable backbone geometries. We mainly explored the Gly–Gly dipeptide segments, which are highly flexible and may disrupt G4-binding conformations. To address this, we incorporated MADIs, a synthetically accessible β-turn mimetic,30,31) into selected Gly–Gly positions to promote the formation of the β-turn structure. This MADI-driven modification allowed us to assess the impact of backbone preorganization on the G4-binding affinity and selectivity of peptides.
Each peptidomimetic contained a single Gly–Gly-type MADI, replacing one flexible Gly–Gly segment within the original peptide sequence. Overall, 32 peptidomimetics were designed using the 10 native RGG peptides (Fig. 3). The Gly–Gly-type MADI unit was synthesized, following our reported method,32) and all the peptidomimetics were assembled by standard Fmoc-SPPS and purified by reversed-phase HPLC.
To assess the influence of β-turn-inducing backbone modifications on G4-binding affinity, we performed filter-binding assays to evaluate the MADI-type RGG peptidomimetics. Among the 32 modified peptides, five exhibited measurable binding affinities to G4 RNA with Kd of less than 30 μM (Table 2).
 |
The peptide exhibiting most potent binding, STK3-2, realized a Kd of 3.7 μM, representing an 18-fold improvement over its native counterpart STK3-1 (Kd = 54.1 μM). Furthermore, peptides STK1-3 and STK4-3 also demonstrated enhanced binding affinities, with Kd values of 9.19 and 18.0 μM, respectively, corresponding to 3.5- and 3.1-fold improvements over their parent peptides, respectively. Additionally, STK5-3 and STK5-4 maintained moderate G4-binding affinities (Kd = 17.7 and 25.4 μM, respectively), although these affinities were lower than that of their parent STK5-1 peptide. These results indicate that Gly–Gly-to-MADI substitutions can enhance or modulate G4-binding affinity depending on the sequence context and modification position. Notably, although STK1-3 only exhibited moderate G4-binding affinity (Kd = 9.19 μM), it exhibited the highest selectivity index (SI = 3.0) among the tested peptides. This result suggests that conformational constraints not only modulate binding affinity but also improve discrimination between G4 structures and single-stranded RNA. By contrast, MADI-substituted peptides such as STK5-2, STK5-3, and STK8-3 exhibited moderate-to-strong G4-binding affinities but poor selectivity. One possible explanation is that their native counterparts, STK5-1 and STK8-1, contain four or five basic residues, respectively, which may promote nonspecific interactions with ssRNA even after MADI incorporation. On the other hand, STK4-1—having fewer basic residues—shows low ssRNA affinity, and its MADI-modified form STK4-3 achieves both higher affinity and greater selectivity. These findings suggest that the selectivity of MADI-modified peptides depends not only on conformational constraints but also on the sequence-inherent base composition that determines ssRNA binding propensity.
In summary, we developed a series of RGG peptides and their MADI-type peptidomimetics derived from the RGG3 domain of the G4-binding protein, TLS/FUS. Thereafter, their binding affinities to G4 RNA structures were evaluated. Among the native peptides, STK5-1—located in the central region of the RGG3 domain and containing multiple RGG motifs—exhibited the highest G4-binding affinity, highlighting the role of RGG repeats to G4 recognition.
To further improve the binding efficiencies of these peptides through conformational stabilization, we substituted the flexible Gly–Gly segments with MADIs. The results revealed that several MADI-type peptidomimetics exhibited enhanced G4-binding affinity, with STK3-2 exhibiting the highest (demonstrating an 18-fold improvement over its native peptide). These findings underscore the potential of backbone engineering in modulating G4 recognition, providing a basis for further structural optimization of RGG-based G4 ligands.
This work was supported by JSPS KAKENHI Grant Numbers: JP20H03363 and JP23H02601, partially by JST SPRING, Grant Numbers: JPMJFS2119 and JPMJSP2167, as well as the Naito Foundation and the Takeda Science Foundation.
The authors declare no conflict of interest.
This article contains supplementary materials.
