The Role of Flexibility in the Rational Design of Modularly Assembled Ligands Targeting the RNAs that Cause the Myotonic Dystrophies

Module Identification and Modular Assembly

In a previous 2DCS study, it was determined that the RNA internal loops that preferred to bind 6″-azido-kanamycin A (Kaz, Figure 1) contain pyrimidines. A subset of the selected loops is shown in Figure 1.[7] Comparison of these selected structures to toxic RNAs found commonalities with the repeating RNAs that cause myotonic dystrophy types 1 and 2 (DM1 and DM2, respectively). Because DM1 and DM2 RNAs display regularly repeating pyrimidine-rich internal loops, we assembled the Kaz ligands onto a peptoid scaffold (Scheme 1) to bind multiple loops simultaneously and thus inhibit formation of the RNA-MBNL1 complexes (Figure 2).

Previously, several important studies have reported small molecule ligands that bind a single 5′CUG/3′GUC repeat.[23, 24] These monomeric ligands bind the RNA repeats with dissociation constants ranging from 450 nM to 10 μM and inhibit the RNA-MBNL1 interaction with K_I’s in the low micromolar range.[23, 24] Modularly assembled ligands have been utilized by our group to target the DM1 and the DM2 RNAs.[8, 22] The assembled compounds that were used displayed multiple copies of either 6′-N-5-hexynoate kanamycin A[8] or a bis-benzimidazole[22] module conjugated onto a peptoid backbone. These ligands bind toxic RNAs with nanomolar affinities and inhibit the formation of the toxic RNA-protein interactions with IC₅₀‘s in low nanomolar range. The enhanced potency of the assembled ligands can be attributed to the larger surface area coverage of modularly assembled ligands than monomeric ones.[8, 22] In this study, Kaz modules, which have been shown to prefer DM-like RNA internal loops via 2DCS (Figure 1)[7], were displayed on a peptoid backbone to study how the identity of the ligand, ligand spacing, and flexibility of the scaffold affect RNA binding affinity and potency. Such studies are critical to further understand how to develop highly potent and specific ligands that target the DM RNAs as well as others in genomic sequence.

Peptoids that display propargylamine were synthesized and conjugated to Kaz via a Huisgen dipolar cycloaddition reaction [25] (HDCR, Scheme 1), which is the same anchoring chemistry used to immobilize Kaz onto microarrays used in 2DCS (Figure 1).[7] Valency and spacing were controlled by the number of propargylamine couplings (a + 1, Scheme 1) and the number of propylamine spacers (b, Scheme 1) inserted between the propargylamine submonomers, respectively. The nomenclature used for the assembled compounds has the general format (a + 1)M−b where (a + 1) is the valency or number of ligand modules displayed, M is the ligand module displayed, and b is the number of propylamine spacing modules (Scheme 1). The ligand modules are Kaz which was conjugated to alkyne-displaying peptoids, and 6′-N-5-hexynoate neamine (N) or proparglyamine (P) which were conjugated to azide-displaying peptoids.[8] Thus, 3Kaz-2 is a peptoid in which three Kaz modules are displayed on a peptoid, each separated by two propylamine spacing modules. Modularly assembled compounds that display N or P were used to control for non-specific inhibition of the RNA-MBNL1 complexes; 3N-4 and 3P-4 are weak inhibitors of the DM1 RNA- and DM2 RNA-MBNL1 complexes with IC₅₀’s > 1 μM (Table 1).

Potency of Assembled Ligands for Inhibition of the DM1 RNA and DM2 RNA-MBNL1 complexes

A series of dimeric ligands with different spacings between the RNA-binding modules (the value of b, Scheme 1) were synthesized. Their potencies for inhibiting the DM1 RNA- and the DM2 RNA-MBNL1 complexes were determined using a microtiter plate assay as previously described.[8] Model RNA constructs (Figure 3) were used to mimic the DM1 RNA (RNA2) and the DM2 RNA (RNA3) by inserting 12 copies of 5′CUG/3′GUC or 5′CCUG/3′GUCC, respectively, into a hairpin cassette (RNA1). Results show that the IC₅₀’s of the most potent dimers were 500 nM for DM2 RNA3 and 3 μM for DM1 RNA2 (Table 1).[26] The optimal dimer for the DM1 RNA has Kaz ligand modules separated by two propylamines while the optimal dimer for the DM2 RNA has four propylamines (Table 1).[26] These were the same optimal spacings identified for the 6′-N-5-hexynoate kanamycin A ligand module.[8] Surprisingly, modularly assembled compounds with higher valencies were less potent inhibitors. For example, 2Kaz-2, 3Kaz-2, and, 4Kaz-2 had IC₅₀’s of 3, 19, and >25μM for inhibition of the DM1 RNA2-MBNL1 complex while 2Kaz-4, 3Kaz-4, and, 4Kaz-4 2 had IC₅₀’s of 0.5, 1.7, and 4.0 μM for inhibition of the DM2 RNA3-MBNL1 complex (Table 1). Since previous studies showed that potency increased as a function of valency, we set out to determine the cause of the decreased potency observed with the Kaz modules.[8, 22]

Once conjugated, the 6′-N-5-hexynoate kanamycin A ligand module used previously has ten atoms between the aminoglycoside and the peptoid backbone.[8] In contrast, when Kaz is conjugated, there are only four atoms between the ligand and the peptoid. Therefore, second-generation peptoids were synthesized to increase the distance between the peptoid backbone and the aminoglycoside (Scheme 1). In order to do so, a flexible ethylene glycol linker was used to conjugate Kaz to the peptoid. These compounds have the general format (a + 1)KazL-b where KazL denotes conjugation of Kaz to the flexible linker, L. These peptoids showed enhanced potency as valency increased. For example, 2KazL-2 and 4KazL-2 had IC₅₀’s of 1.8 and 0.05 μM, respectively, for inhibition of the DM1 RNA2-MBNL1 complex while 2KazL-4 and 4KazL-4 had IC₅₀’s of 1.0 and 0.02 μM for inhibition of the DM2 RNA3-MBNL1 complex (Table 1). Thus, potencies for the tetrameric ligands are enhanced by >500-fold for the DM1 model system and 200-fold for the DM2 system by increasing the flexibility of the ligand module.

Potencies for each assembled compound were used to calculate the multivalent effect (MV), which quantifies the bonus in potency that an assembled ligand provides over a monomeric one (Table 1). The multivalent effect is calculated by first multiplying the IC₅₀ value by the valency, affording the normalized IC₅₀ or NIC₅₀. The IC₅₀ for the Kaz monomer is then divided by the NIC₅₀ of the compound of interest to afford the multivalent effect. The MV values range from 1 for Kaz to 2,500-fold for 4KazL-4. Not surprisingly, the increased flexibility of the KazL compounds provides enhancements in the magnitude of the multivalent effect. For example, 4Kaz-2 has a MV of <1.9-fold for inhibition of the DM1-MBNL1 complex while 4KazL-2 has a MV of 950-fold. Analogously, 4Kaz-4 has a MV of 12.5-fold for inhibition of the DM2-MBNL1 complex while 4KazL-4 has a MV of 2,500-fold.

In order to investigate the selectivity of the optimal modularly assembled compounds, the potencies of the ligands were determined in the presence of about 15-fold molar excess in nucleotides of yeast bulk tRNAs (Table 1). The IC₅₀’s for the 4KazL-2 and 4KazL-4 ligands inhibiting the DM1- and DM2 RNA-MBNL1 complexes are mid nanomolar (~200 nM) in the presence of these competitor nucleic acids.

Affinities and Specificities of Modularly Assembled Ligands

The affinities and specificities of the RNA-ligand complexes were also determined as previously described.[8, 27, 28] Six different RNAs were tested (Figure 3) including DM1 RNA2, DM2 RNA3, and RNA4 which contains 12 copies of the 5′CAG/3′GAC motif present in RNAs that cause spinocerebellar ataxia type 3 (SCA3). Other control RNAs include RNA5 and RNA6 that have 12 copies of 2×2 nucleotide internal loops, 5′UU/3′UU and 5′CC/3′CC, respectively, and fully base-paired RNA7. Results show that the optimally spaced ligands are all higher affinity DM1 and DM2 binders than MBNL1. The ligands with two propylamine spacing modules are DM1-specific while the ligands with four spacing modules are DM2-specific. Importantly, ligand flexibility can drastically affect RNA binding affinity (Table 2).

For the DM1 RNA2, the Kaz monomeric ligand binds with a dissociation constant of 920 nM while it binds to RNA1 (the cassette in which the RNAs were embedded), RNA4, RNA6, and RNA7 with dissociation constants >5 μM. The 2Kaz-2 dimer binds weakly to RNA2 while the corresponding compound with the ethylene glycol linker, 2KazL-2, is higher affinity with a K_d of 75 nM. A similar trend is observed with the tetramers, 4Kaz-2 and 4KazL-2, which have dissociation constants of >500 nM and 20 nM, respectively. The 4KazL-2 ligand binds RNA2 at least ~12.5 times more tightly than RNA3 (the DM2 RNA), RNA4 (the SCA3 RNA), RNA6, and fully paired RNA7. It should be noted that RNAs with long stretches of periodically repeating internal loops are rare in genomic sequences while stretches of base pairs are more common.[29] Fortuitously, the 4KazL-2 ligand binds fully paired RNA very weakly. In addition, each ligand binds to the DM1 RNA with the expected stoichiometry (Table 2).[28]

Optimal ligands that bind the DM2 RNA3 are also high affinity and specific. The Kaz monomer binds more tightly to the DM2 RNA than the DM1 RNA with a K_d of 680 nM. The dimers 2Kaz-4 and 2KazL-4 bind RNA3 with similar affinities (145 nM and 80 nM, respectively) in good agreement with their similar potencies for disrupting the RNA3-MBNL1 complex (Table 1). Consistent with the observed potencies, 4Kaz-4 binds weakly to the DM2 RNA (K_d >250 nM) while the corresponding ligand with the extended, flexible linker, 4KazL-4, binds with a dissociation constant of 30 nM. The compound also binds weakly to RNA4-RNA7. As was the case with the optimal DM1 ligands, the optimal DM2 ligands bind RNA3 with the expected stoichiometries (Table 2).[28]

The modularly assembled Kaz compounds are slightly less potent and bind with slightly less affinity than the modularly assembled compounds that display 6′-N-5-hexynoate kanamycin A.[8] This indicates that the manner in which the ligand is displayed can affect affinity and potency. Interestingly, the position of functionalization does not appear to affect cellular uptake or toxicity.[8, 26] However, the distance between the ligand module and the peptoid backbone, and hence flexibility of the ligand modules, is an important consideration when modularly assembling compounds, at least for the aminoglycoside derivative Kaz.

Uptake of Assembled Ligands into Mouse Myoblasts

The modularly asembled ligands were tesed for uptake into mouse myoblasts using fluorescence microscopy. Results show that all modularly assembled compounds permeate cells when 5 μM of the ligand is added to a cell culture. Comparison of the uptake patterns of the dimeric and tetrameric ligands show that there is no signifcant difference in their uptake or cellular localization (Figure 4). This is interesting because one would expect that as the size of the assembled ligand increases that uptake would decrease. In fact, earlier studies have shown that a minomeric aminoglycoside is taken up by C2C12 cells only in small amounts compared to the modularly assembled ligands.[8, 9]

Uptake and toxicity of the assembled ligands were quantified using flow cytometry (Table 3). After incubating the cells with the ligand of interest for 14 h, they were treated with propidium iodide, which stains cells with compromised cell membranes. All modularly assembled ligands are more cell permeable than Kaz-FITC, the monomeric ligand. At least five times more cells are postiive for uptake. By comparing the normalized fluorescence intensities for the ligands, the modularly assembled ligands are present in ~2- to ~10-fold higher concentrations. The results also show that the ligands are non-toxic; at worst, only ~5% of the cell population that has taken up fluorescently labeled modularly assembled ligand are stained with propidium iodide after a 14 h incubation. Similar results were also obtained after the cells were incubated with the ligand of interest for 24 or 48 h. These data are available in the Supporting Information.

Chemical Synthesis

Details of chemical syntheses and analytical data are in the Supporting Information.

MBNL-1 Displacement Assays

Displacement assays were completed as previously described[8] in black 384-well plates coated with Streptavidin (Nunc). In these experiments, 2.5 pmoles of RNA2 or RNA3 was delivered to each well. Resorufin-β-D-galactopyranoside was used as a substrate to quantify the amount of β-galactosidase, and hence MBNL1, present. Fluorescence intensity was measured using a BioTek FLX-800 plate reader. By comparing the fluorescence intensities to wells containing no inhibitor (maximum response) and no RNA (minimum response), the percentage of MBNL1 bound can be determined. The percentage of MBNL1 bound was plotted versus ligand concentration and the resulting curve fit to SigmaPlot’s 4-parameter logistic function in order to determine the IC₅₀ (Equation 1):

where y is the percentage of MBNL1 bound, D is the minimum response plateau, A is the maximum response plateau, and x is the concentration of ligand. A and D are typically 100% and 0%, respectively. Each IC₅₀ is the average of at least two measurements.

In order to determine the multivalent effect, the IC₅₀’s were normalized for the number of ligand modules conjugated to the peptoid backbone. These normalized IC₅₀’s, or NIC₅₀’s, were calculated by multiplying the IC₅₀ by the number of ligand modules displayed on the peptoid. Multivalent effects were calculated by dividing the IC₅₀ for Kaz (monomer) by the NIC₅₀ of the compound of interest. The number of moles of RNA immobilized in each well was determined using SYBR Green II as described.[22] Approximately 20% of the moles of RNA delivered to a well are immobilized.

RNA Binding Assays

The affinities of RNA-ligand complexes were determined as described using a fluorescence emission-based assay.[8] Briefly, RNA was folded in 1X MBNL Buffer (50 mM Tris HCl, pH 8.0, 50 mM NaCl, 50 mM KCl, 1 mM MgCl₂) without MgCl₂ by incubating at 60 °C for 5 min followed by slowly cooling to room temperature. Then, MgCl₂, BSA, and ligand of interest were added to final concentrations of 1 mM, 40 μg/mL, and 75 nM, respectively. The RNA was serially diluted in 1X MBNL buffer containing 40 μg/mL BSA and 75 nM ligand and incubated for 1 h at room temperature. Fluorescence intensity was determined using a BioTek FLX-800 plate reader. Stoichiometries were determined by constructing plots of [Ligand]/[RNA] vs. change in fluorescence. The resulting curves were fit to two lines and the equations solved simultaneously to afford the stoichiometry as previously desribed.[27, 32] Scatchard analyses were completed to determine dissociation constants, accounting for statistical effects by using a functional form of the Scatchard equation for large ligands binding to a lattice (Equation 2) [28, 33]:

where v is the moles of ligand per moles of RNA lattice, [L] is the concentration of ligand, N is the number of repeating units on the RNA, l is the number of consecutive lattice units occupied by the ligand, and k is the microscopic dissociation constant. This equation simplifies to the commonly used form of the Scatchard equation for simple systems.[28, 33]