Conjugation to a transferrin receptor 1-binding Bicycle peptide enhances ASO and siRNA potency in skeletal and cardiac muscles

Abstract

Improving the delivery of antisense oligonucleotides (ASOs) and small interfering RNAs (siRNAs) to skeletal and cardiac muscles remains a pivotal task toward the broader application of oligonucleotide therapeutics. The targeting of myofibers and cardiomyocytes via conjugation of ASOs and siRNAs to ligands that bind the human transferrin receptor 1 (TfR1) has gathered significant interest in recent years. However, the selection of ligands with low molecular weight and optimal biophysical and binding properties is crucial to maximize the potential of the TfR1 ligand-conjugated antisense (LICA) technology. Here, through effective combination of phage display and peptide medicinal chemistry, we identified and characterized a bicyclic peptide (Bicycle® molecule BCY17901), with a molecular weight of ∼2 kDa, that binds human TfR1 with high affinity and specificity. Conjugation to BCY17901 improved ASO and siRNA potency in skeletal and cardiac muscles of human TfR1 knock-in mice, after either intravenous or subcutaneous administration. Furthermore, single-nucleus RNA sequencing showed that conjugation to BCY17901 enhanced ASO activity in myonuclei of different muscle fiber types. Importantly, we demonstrated good translatability of our TfR1-targeting platform in skeletal and cardiac muscles of nonhuman primates. Our results offer great promise toward potential future applications of low-molecular-weight Bicycle LICA therapeutics for the treatment of diseases affecting skeletal muscle and heart.

Graphical Abstract

Figure 1.

Use of non-natural substitutions and guidance from crystal structure to enhance the binding affinity of lead Bicycle peptides for human TfR1. Non-natural substitutions (shown in red in the table) were screened using surface plasmon resonance (SPR) against human TfR1. Aib, 2-aminoisobutyric acid; Aze, azetidine-2-carboxylic acid; HyP, hydroxyproline; tBuAla, tert-butylalanine; Cba, cyclobutylalanine; tBuGly, tert-butylglycine; EPA, 2-amino-3-ethylpentanoic acid; 2Nal, 3-(2-naphthyl)-alanine; 1Nal, 3-(1-naphthyl)-alanine.

Figure 2.

Twelve different Bicycle–Dmpk–ASO conjugates bind human TfR1 with different affinities. Bicycle peptide sequences and inhibition constants (K_i) of Dmpk–ASO conjugates measured using bioluminescence resonance energy transfer (BRET). Modified positions are shown in red in the table.

Figure 3.

In vivo screening of Bicycle peptides that bind to human TfR1 with different affinities identifies BCY17901 as the lead ligand. Bicycle peptides were conjugated to mouse Dmpk ASO and dosed intravenously in human TfR1 heterozygote KI mice at 3.5 mg/kg ASO equivalent dose. A group of mice were dosed with the unconjugated Dmpk ASO at 35 mg/kg. Another group of mice was dosed with Dmpk ASO conjugated to the benchmark human TfR1 ligand OKT9 Fab′. The graphs report the mouse Dmpk (mDmpk) mRNA levels measured by RT-qPCR in (A) the quadriceps muscle and (B) the heart, expressed as percentage of the PBS-treated animals (vehicle control), after normalization using mouse Gapdh as the housekeeping gene. Error bars indicate standard deviation.

Figure 4.

Conjugation of Dmpk ASO, Malat1 ASO, and Hprt siRNA to BCY17901 improves ON potency in skeletal muscle and heart of human TfR1 heterozygote KI mice. Dose-dependent target knockdown was measured by RT-qPCR in various skeletal muscle groups (quadriceps and gastrocnemius), heart, and liver of human TfR1 KI mice after intravenous injection of (A) Dmpk ASO, (B) Malat1 ASO, and (C) Hprt siRNA conjugated to BCY17901. Unconjugated ASOs and a lipid (palmitate, C16)-conjugated Hprt siRNA (orange) were included in the study to compare their potency versus the Bicycle-conjugated counterparts (green). Data are expressed as percentage target RNA level compared to PBS-treated (vehicle control) mice, after normalization using mouse Gapdh as housekeeping gene. Each open dot represents one animal. Doses refer to the ASO or siRNA component of the LICA molecules (ASO or siRNA equivalents).

Figure 5.

ASO conjugation to BCY17901 improves activity in myonuclei of all muscle fiber subtypes. (A) Top left panel: annotation of cell types identified in the snRNAseq analysis of gastrocnemius muscle plotted in a uniform manifold approximation and projection embedding. Myonuclei subtypes are annotated in distinct colors. Top right panel: violin plot of mouse Malat1 expression levels in myonuclei of gastrocnemius muscle from mice treated with PBS (vehicle, blue), unconjugated Malat1 ASO (ASO, red), and BCY17901-conjugated Malat1 ASO (green). Bottom panel: violin plot showing mouse Malat1 expression levels (normalized counts) in different myofiber subtypes (i.e. types IIb, IIa, and I). Data from n = 8 mice (2 mice dosed with PBS vehicle control, 3 mice with unconjugated Malat1 ASO, and 3 mice with BCY17901-conjugated Malat1 ASO). Asterisks (***) indicate adjusted P-value <.001 in differential expression test comparing unconjugated and BCY17901-conjugated Malat1 ASOs (using MAST with Bonferroni correction). (B) Representative images of ASO IHC (labels ASO in brown) in histological sections of quadriceps muscle from human TfR1 KI mice dosed with PBS (vehicle control), unconjugated Malat1 ASO, or BCY17901-conjugated Malat1 ASO. The top panels report the results for the 3 mg/kg ASO equivalent dose level, whereas the bottom panels report the results for the 10 mg/kg dose level. Scale bars: 50 μm. (C) Representative images of Malat1 ISH (labels Malat1 RNA in brown) in histological sections of quadriceps muscle. Scale bars: 50 μm. The PBS images in top and bottom panels of (B) are identical, as they were intentionally duplicated for ease of comparison versus both the 3 and 10 mg/kg treatment groups, respectively. The PBS images represent the same experimental conditions used across all treatment groups. Similarly, the PBS images in top and bottom panels of (C) are identical.

Figure 6.

BCY17901-conjugated MALAT1 ASO and HPRT siRNA achieve robust target knockdown in skeletal muscle and heart of NHPs. Determination of Bicycle–ASO conjugate affinity to (A) cynomolgus monkey TfR1 and (B) human TfR1 using BRET. (C) Cynomolgus monkey MALAT1 (cMALAT1) RNA and HPRT (cHPRT1) mRNA levels measured by RT-qPCR in quadriceps muscle, heart, and liver of NHPs dosed by intravenous infusion with 25 mg/kg BCY17901-conjugated MALAT1 ASO and HPRT siRNA. The expression data were normalized using RiboGreen and are reported as percentage of control-treated animals (NHPs dosed with non-MALAT1 or non-HPRT targeting compounds). Error bars indicate standard deviation.

Figure 7.

Subcutaneous administration of Bicycle–ASO conjugate results in robust target knockdown and improved potency compared to unconjugated ASO in skeletal muscle and heart of human TfR1 heterozygote KI mice. (A) Dose-dependent target knockdown of mouse Dmpk (mDmpk) mRNA was measured by RT-qPCR in various skeletal muscle groups (quadriceps and gastrocnemius) and heart of human TfR1 KI mice after both intravenous (light blue) and subcutaneous injection (purple) of Dmpk ASO conjugated to BCY17901. Unconjugated ASO (orange) dosed subcutaneously was included in the study to compare its potency versus the Bicycle-conjugated counterpart. Data are expressed as percentage target RNA level compared to PBS-treated (vehicle control) mice, after normalization using mouse Gapdh as housekeeping gene. Each open dot represents one animal. (B) ED₅₀ values extrapolated from the graphs in panel (A) were calculated in GraphPad Prism software using the following constraints: top = 100, bottom = 0, Hill slope <−1. Doses refer to the ASO component of the LICA molecules (ASO equivalents).