Designing antisense oligonucleotide delivery peptides with interpretable machine learning

Abstract

Provided herein are oligonucleotides, trimeric peptides, and peptide-oligonucleotide conjugates. Also provided herein are methods for treating muscle disorders in a subject in need thereof, comprising administering to the subject the oligonucleotides, trimeric peptides, and peptide-oligonucleotide conjugates described herein. The synthesis method provides a machine learning-based generator-predictor-optimizer loop for generating a library of cell-penetrating peptides conjugated to antisense oligonucleotides and generating novel peptide sequences capable of enhancing delivery of oligonucleotide cargo from the library of conjugates.

Description

[Technical Field]

[0001] Related Applications This application claims priority to U.S. Provisional Patent Application No. 62 / 965,555, filed January 24, 2020, and U.S. Provisional Patent Application No. 63 / 134,405, filed January 6, 2021, the contents of which are incorporated herein by reference in their entireties. [Background technology]

[0002] Antisense technology provides a means to regulate the expression of one or more specific gene products, including products of alternative splicing, and is uniquely useful in many therapeutic, diagnostic, and research applications. The principle behind antisense technology is that antisense compounds, e.g., oligonucleotides, that hybridize to a target nucleic acid regulate gene expression activities such as transcription, splicing, or translation through one of a number of antisense mechanisms. The sequence specificity of antisense compounds makes them attractive as tools for target validation and gene functioning, as well as therapeutic agents for selectively regulating the expression of genes involved in disease.

[0003] Although significant advances have been made in the field of antisense technology, there remains a need in the art for oligonucleotides and peptide-oligonucleotide conjugates with improved antisense or antigene performance. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] U.S. Patent No. 5,698,685 [Patent Document 2] U.S. Patent No. 5,217,866 [Patent Document 3] U.S. Patent No. 5,142,047 [Patent Document 4] U.S. Patent No. 5,034,506

Patent document 5

Patent document 6

Patent document 7

Patent document 8

Patent Document 9

Patent document 10

Patent document 11

Patent document 12

Non-licensed literature

[0005] [Non-licensed document 1] TW Greene, PGM Wuts, Protective Groups in Organic Synthesis, 3rd Edition, John Wiley & Sons (1999) [Non-licensed document 2] Chiu and Rana, RNA, 2003, 9, 1034–1048; Limbach et al., Nucleic Acids Research, 1994, 22, 2183–2196 [Non-licensed document 3] Revankar and Rao, Comprehensive Natural ProductsChemistry, Volume 7, Page 313

Non-licensed Document 4

Non-licensed Document 5

Non-licensed Document 6

[0006] Provided herein is a peptide-oligonucleotide conjugate, comprising the oligonucleotide covalently linked to peptide.Also provided herein is a method for treating the disease of a subject in need thereof, comprising administering to the subject the peptide-oligonucleotide conjugate described herein.Also provided herein is a method for identifying one or more cell-penetrating peptides with optimal activity using machine learning. [Means for solving the problem]

[0007] Thus, in one aspect, provided herein is a peptide-oligonucleotide conjugate of Formula I:

[0008] [ka]

[0009] or a pharmaceutically acceptable salt thereof, During the ceremony, A' is -N(H)CH2C(O)NH2, -N(C 1~6 -alkyl)CH2C(O)NH2,

[0010] [ka]

[0011] is selected from R 5 is —C(O)(O-alkyl)x-OH, where x is 3 to 10, and each alkyl group, independently at each occurrence, is C 2~6 -alkyl, Or R 5 is -C(O)C 1~6 -Alkyl, trityl, monomethoxytrityl, -(C 1~6 -alkyl)-R 6 , -(C 1~6 -heteroalkyl)-R 6 , aryl-R 6 , heteroaryl-R6 , -C(O)O-(C 1~6 -alkyl)-R 6 , -C(O)O-aryl-R 6 , —C(O)O-heteroaryl-R 6 , and

[0012] [ka]

[0013] Selected from; R 6 is selected from OH, SH, and NH2, or R 6 is O, S, or NH, each of which is covalently attached to a solid support; Each R 1 is OH and -N(R 3 )(R 4 ) are independently selected from each R 3 and R 4 independently for each occurrence, -C 1~6 - is alkyl; Each R 2 is independently selected at each occurrence from H, a nucleobase, and a nucleobase functionalized with a chemical protecting group, and the nucleobase is independently selected at each occurrence from pyridine, pyrimidine, triazinane, purine, and deazapurine. 3~6 -contains heterocyclic rings; z is 8 to 40; E' is H, -C 1~6 -Alkyl, -C(O)C 1~6 -Alkyl, benzoyl, stearoyl, trityl, monomethoxytrityl, dimethoxytrityl, trimethoxytrityl,

[0014] [ka]

[0015] Selected from; Q is —C(O)(CH)C(O)— or —C(O)(CH)S(CH)C(O)—; R 7 is -(CH2)2OC(O)N(R 8 )2 and R 8 is -(CH2)6NHC(=NH)NH2; L is -C(O)(CH2) 1~6 -C 7~15 -Heteroaromatic-(CH2) 1~6 C(O)—, and L is covalently linked to J by an amide bond; J is a carrier peptide; G is H, -C(O)C 1~6 - selected from alkyl, benzoyl, and stearoyl, and G is covalently linked to J; The following conditions: 1) A' is

[0016] [ka]

[0017] or 2) E' is

[0018] [ka]

[0019] Being At least one of the following is true: The carrier peptide J is selected from the following sequences:

[0020] [Table 1A]

[0021] [Table 1B]

[0022] [Table 1C]

[0023] wherein X is 6-aminohexanoic acid and B is β-alanine; C L 1 By another C is covalently bound to; L 1 teeth

[0024] [ka]

[0025] and; M is

[0026] [ka]

[0027] and; R 10 is independently at each occurrence H or halogen.

[0028] In another aspect, also provided herein are compounds of Formula II having the above definition, wherein the carrier peptide J is selected from the following sequences:

[0029] [Table 2]

[0030] In one embodiment, the peptide-oligonucleotide conjugate of Formula I is a peptide-oligonucleotide conjugate of Formula Ia:

[0031] [ka]

[0032] or a pharmaceutically acceptable salt thereof.

[0033] In another embodiment, the peptide-oligonucleotide conjugate of formula I is a peptide-oligonucleotide conjugate of formula Ib:

[0034] [ka]

[0035] or a pharmaceutically acceptable salt thereof.

[0036] In yet another aspect, provided herein is a method of treating a neuromuscular disease comprising administering to a subject a peptide-oligonucleotide conjugate of the present disclosure.

[0037] In another aspect, provided herein is a method for identifying one or more cell-penetrating peptides with optimal activity using machine learning, the method comprising: a.) synthesizing a library of training oligonucleotide-cell penetrating peptide conjugates; b.) generating seed peptide sequences by training a nested long short-term memory (LSTM) recurrent neural network model using the synthesized library; c.) predicting, from the generated seed peptide sequences, which peptide sequences have a predetermined structure-activity relationship of amino acid residues; and Activity predictor - Identifying the best of one or more of the predicted peptide sequences using a genetic algorithm optimizer loop. [Brief explanation of the drawings]

[0038] [Figure 1]A) Inverse design model: A modular PMO-CPP library that was tested for activity and used to train machine learning algorithms to design novel, highly active CPPs, which were then evaluated for activity and toxicity in vitro and in vivo. B) Four modules joined using orthogonal bioconjugation. [Figure 2] A) Amino acid residues represented as topological fingerprints. B) A series of sequence representations: Conv1D (linear arrangement of fingerprints, representing covalently bonded residues and local interactions), Conv2D (pairwise contact map of fingerprints, representing a fully connected molecular graph), Conv2D-macrocycle (pairwise contact map of fingerprints with explicit information about cyclic covalent bonds, representing a fully connected molecular graph with additional information), and DeConv2D (pairwise variational contact map with learned weights, representing 3D interactions captured by learning functionality values). C) Comparison of predicted MFI values ​​for the original Conv1D model with experimentally observed MFI values. D) Improvement of PMO for sequences from the training dataset (boxplots) and validation dataset (blue dots). E-G) Optimized key properties (length, percentage of arginine residues in the sequence, net charge of the sequence) are shown comparing the training and validation sequences for MFI. [Figure 3] A) Positive gradient map of Mach3. B) Positive substructures (green) at the most positive residues in Mach3. C, D) Positive gradient activation for the best sequences of specific lengths (30, 35, 40, 45, 50) averaged over C) residue position and D) fingerprint index from Figure 3. E) Clustering of amino acids in the best performing sequences based on residue position. F) Substructures of the most activated fingerprint index. [Figure 4]A) Dose-response curves for activity (corrected splicing in eGFP654 HeLa cells) and toxicity (LDH release in RIPTEC cells) are shown for PMO alone, the known active peptide Bpep-Bpep, and four Mach peptides. Activity was determined using an eGFP assay: HeLa 654 cells were incubated with PMO-Mach constructs for 22 hours before analysis by flow cytometry. Results are shown as percent increase compared to PMO alone; triplicate experiments were performed twice. Toxicity was determined using identically treated renal epithelial cells (RPTEC TH1) and analyzed using an LDH release assay (*p<0.01, two-tailed Student's t-test). B) Mach PMO-peptides do not induce the release of proinflammatory cytokines, as determined by an inflammatory cytokine panel in human macrophages. Human monocyte-derived macrophages were treated with various concentrations of each PMO-peptide for 3 hours, washed, and incubated for 12 hours. Released cytokines were detected by bead-based immunoassay and flow cytometry. [Figure 5] FIG. 1 shows specific peptide sequences and names for proof-of-concept experiments. [Figure 6] A) Heatmap disclosing the mean cell fluorescence of HeLa-654 cells treated with each modular construct (n=3 replicate wells). B) Heatmap disclosing the total cell number of HeLa-654 cells after treatment with each modular construct. Each experiment was capped at 5000 cells. Low cell numbers suggest cytotoxicity. C) Heatmap disclosing the mean fluorescence multiplied by cell number (FxC), giving a single metric capturing the two most important parameters of a modular construct. [Figure 7]Figure 1 shows a heatmap disclosing the FxC of 600 constructs tested in the HeLa-654 assay (n=1 replicate well). The most potent compound was PMO-DPV6-SV40-W / R, a peptide combination that was not predicted to be particularly noteworthy prior to testing. Boxes marked with an "X" are constructs for which zero cells were gated. [Figure 8] Figure 1 shows a heatmap showing the mean fluorescence intensity of 600 constructs tested in the HeLa-654 assay (n=1 replicate wells). Squares marked with an "X" are constructs for which the gated cell count was zero. [Figure 9] Figure 1 shows a heatmap disclosing the total cell number after treatment with 600 constructs (n=1 replicate well). The number of gated cells was limited to 5,000. [Figure 10] Jaro-Winkler self-similarity of training sequences. A) Shows the sequences used in training the generator (nested LSTM). B) Shows the sequences used in training the predictor (convolutional neural network-based model). [Figure 11] Figure 1 shows absolute intensity plots of predicted and experimental results for training (80% of the dataset), validation (20% of the dataset), and the model accuracy as a percentage within the range of training values ​​stated in the title. 128-bit fingerprints. Models obtained after hyperparameter optimization for various representations using A) Conv1D, B) Conv2D, C) Conv2D macrocycle, and D) DeConv2D. [Figure 12] A) Predicted sequence novelty versus experimental potency. B) Immunogenicity score based on an online tool (IEDB) that predicts T cell epitopes. [Figure 13]Gradient activity of the training set sequences is sorted in descending order of MFI, showing A) residue position from the C-terminus and B) positive activation averaged by fingerprint index, and C) residue position from the C-terminus and D) negative activation averaged by fingerprint index. The original amino acid chemical substructures and fingerprint indexes for E) arginine, F) lysine, G) histidine, and H) aminohexanoic acid are shown. [Figure 14] This figure shows that the Mach peptide enhances PMO delivery by 40-50 fold as measured in an in vitro exon skipping assay. Experimental activity (blue) is comparable to predicted activity (blue). [Figure 15] Mach CPP is half-toxic at 5 μM as measured by A) LDH release assay and B) MTT assay. Cytotoxicity is reported as the percentage of LDH release compared to cell lysate, and viability is reported as the percentage of untreated. [Figure 16] FIG. 1 shows the results of an inflammation panel of cytokines detected in human monocyte-derived macrophages. [Figure 17] FIG. 10 shows a Coomassie stained SDS-page gel of the ligation of Mach-LPSTGG peptide to G5-DTA. [Figure 18] Figure 1 shows the activity of PMO-peptide conjugates (eGFP assay) measured at a concentration of 5 μM in three different biological replicates for each PMO-peptide conjugate. eGFP fluorescence was normalized to cells treated with unconjugated PMO. [Figure 19] FIG. 1 shows the superior activity of PMO-P7 over its analogs (PMO-P8 to PMO-P12). [Figure 20]1 shows that the KXXC motif at the C-terminus of the peptide does not result in increased PMO delivery relative to analog PMO-peptide conjugates in the absence of KXXC. Activity (eGFP assay) of pairs of PMO-peptide conjugates in the absence and presence of a KXXC motif at the C-terminus of the peptide. [Figure 21] FIG. 1 shows the activity of PMO-P7 derivatives, PMO-P21, PMO-P22 and PMO-P23 at 5 μM. [Figure 22] A) Depiction of the dose-response curve (eGFP and LDH) for PMO-P7(acetate). B) Depiction of the dose-response curve (eGFP and LDH) for PMO-P21(acetate). C) Depiction of the dose-response curve (eGFP and LDH) for PMO-P23(acetate). [Figure 23] This figure shows that the polylysine backbone of peptide 6 is primarily responsible for its improved activity in PMO delivery. Within rectangle 2300 are the activities of PMO-peptide conjugates (PMO-8 to PMO-11) containing Ala substitutions in the KXXC motif. Within rectangle 2302 are the activities of PMO-peptide conjugates (PMO-12 to PMO-17) containing Ala substitutions in the polylysine backbone. Within dashed line 2304 are the activities of two PMO-peptide conjugates (PMO-8 and PMO-18) that lack a Cys residue at the C-terminus. A single asterisk (*) indicates a p-value less than 0.005 (p<0.005). A double asterisk (**) indicates a p-value less than 0.0005 (p<0.0005). Three asterisks (***) indicate a p-value less than 0.00005 (p<0.00005). Four asterisks (***) indicate a p-value less than 0.000005 (p<0.000005). [Figure 24]Figure 1 shows that P7 does not exhibit nephrotoxicity while enhancing GFP protein levels in the quadriceps, diaphragm, and heart. A) No significant change in BUN (blood urea nitrogen) levels after 7 days, B) No significant change in creatinine levels after 7 days, and C) No significant change in cystatin C levels after 7 days. D) GFP protein levels in the quadriceps (1300 pg GFP / μg protein at 30 mg / kg and 4000 pg GFP / μg protein at 60 mg / kg); saline, N=6, 10 mg / kg, N=6, 30 mg / kg, N=7, 60 mg / kg, N=4. E) GFP protein levels in the diaphragm (1100 pg GFP / μg protein at 30 mg / kg and 2500 pg GFP / μg protein at 60 mg / kg); saline, N=6, 10 mg / kg, N=6, 30 mg / kg, N=6, 60 mg / kg, N=4. F) GFP protein levels in the heart (2000 pg GFP / μg protein at 30 mg / kg and 2200 pg GFP / μg protein at 60 mg / kg); saline, N=6, 10 mg / kg, N=7, 30 mg / kg, N=8, 60 mg / kg, N=4. (N refers to the number of mice used). [Figure 25] FIG. 1 illustrates an example of a computing device that can be used to implement the techniques described herein. [Figure 26] FIG. 1 shows a block diagram of a modularized library synthesizer-generator-predictor-identifier system used in accordance with the methods described herein to identify one or more cell-penetrating peptides with optimal activity using machine learning. [Figure 27] A, B, C) A flowchart showing how to use the Library Synthesizer-Generator-Predictor-Identifier module of FIG. DETAILED DESCRIPTION OF THE INVENTION

[0039] Phosphorodiamidate morpholino oligonucleotides (PMOs) are attractive therapeutic molecules for genetic diseases. PMOs recognize targets through Watson-Crick base pairing and are designed to exhibit high specificity for their complementary nucleotide sequences. Depending on the type of target sequence, PMOs can mediate various effects, including blocking protein translation or altering gene splicing. Etepleursen, a PMO approved by the FDA to treat Duchenne muscular dystrophy, removes exons containing mutations in the pre-mRNA encoding dystrophin from the final protein transcript, restoring protein function.

[0040] Structurally, PMOs are neutral oligonucleotide analogs in which the ribosyl ring is replaced with a morpholino ring, and the negatively charged phosphodiester backbone is replaced with an uncharged phosphorodiamidate. The altered backbone structure prevents degradation in serum and by intracellular nucleases. However, the relatively large size and neutral charge of PMOs can lead to inefficient delivery to the cytosol and nucleus.

[0041] Cell-penetrating peptides (CPPs) are a promising strategy to improve the delivery of PMOs to the nucleus. CPPs are relatively short sequences of 5–40 amino acids that ideally can access the cytosol and facilitate intracellular delivery of their cargo. CPPs can be classified into various groups based on their physicochemical properties. One common CPP class is the R 12These oligoarginine peptides are often random coil structures. When conjugated to PMOs, oligoarginine peptides are some of the most effective peptides for promoting PMO delivery. Other CPPs, such as penetratin, pVEC, and melittin, are more amphipathic in nature. Although these sequences contain cationic residues, a well-defined separation of charged and hydrophobic residues can promote the formation of amphipathic helices. However, amphipathic CPPs have not been demonstrated to significantly improve the efficacy of PMOs.

[0042] There is no universal mechanism of cell entry for CPPs or CPP-PMO conjugates. The mechanism often depends heavily on the treatment concentration and the type of cargo attached. Above a certain threshold concentration (usually low micromolar), energy-independent uptake into the cytoplasm can be observed faster than the timescales of endocytosis and cell surface recycling. The rapid uptake rate provides evidence for a direct translocation mechanism similar to that observed with small molecules. However, at physiologically relevant low concentrations, uptake is primarily via endocytosis. Even within the endocytic category, CPPs and CPP-PMO conjugates can enter cells using one or more endocytic mechanisms. These endocytic mechanisms include micropinocytosis, clathrin-mediated endocytosis, caveolae-mediated endocytosis, and clathrin / caveolae-independent endocytosis. At low concentrations, CPP-PMO conjugates are primarily taken up by endocytosis, and CPPs that are insufficient for PMO delivery are likely trapped within endosomes or excluded from the nuclear compartment.

[0043] Provided herein are peptide-PMO conjugates for improving PMO delivery. Described herein are increased intracellular uptake of oligonucleotides, particularly compared to unconjugated PMOs and single CPP-PMO conjugates. Also provided herein are methods for identifying one or more cell-penetrating peptides with optimal activity using machine learning.

[0044] definition Listed below are definitions of various terms used to describe this disclosure. These definitions apply to the terms as they are used throughout this specification and claims, unless limited in specific instances, either individually or as part of a larger group.

[0045] The term "about" will be understood by one of ordinary skill in the art and will vary to some extent depending on the context in which it is used. As used herein, when referring to a measurable value such as an amount, a temporal duration, etc., the term "about" is intended to encompass a variation of ±20% or ±10%, including ±5%, ±1%, and ±0.1%, from the specified value, as variations are appropriate for practicing the disclosed methods.

[0046] The term "alkyl" refers, in certain embodiments, to a saturated straight or branched chain hydrocarbon moiety containing 1 to 6, or 1 to 8, carbon atoms, respectively. 1~6 Examples of -alkyl moieties include, but are not limited to, methyl, ethyl, propyl, isopropyl, n-butyl, tert-butyl, neopentyl, and n-hexyl moieties; and examples of C1-8-alkyl moieties include, but are not limited to, methyl, ethyl, propyl, isopropyl, n-butyl, tert-butyl, neopentyl, n-hexyl, heptyl, and octyl moieties.

[0047] The number of carbon atoms in an alkyl substituent is indicated by the prefix "C x~y", where x is the minimum number of carbon atoms in the substituent and y is the maximum number of carbon atoms. Similarly, a Cx chain refers to an alkyl chain containing x carbon atoms.

[0048] The term "heteroalkyl," by itself or in combination with other terms, means, unless otherwise stated, a stable linear or branched alkyl group consisting of the stated number of carbon atoms and one or two heteroatoms selected from the group consisting of O, N, and S, wherein the nitrogen and sulfur atoms may optionally be oxidized, and the nitrogen heteroatom may optionally be quaternized. The heteroatom may be placed at any position on the heteroalkyl group, including between the remainder of the heteroalkyl group and the fragment to which it is attached, or may be attached to the most distal carbon atom of the heteroalkyl group. Examples include: -O-CH-CH-CH, -CH-CH-CH-OH, -CH-CH-NH-CH, -CH-S-CH-CH, and -CH-CH-S(=O)-CH. Up to two heteroatoms may be consecutive, for example, -CH-NH-OCH, or -CH-CH-SS-CH.

[0049] The term "aryl," used alone or in combination with other terms, means, unless otherwise specified, a carbocyclic aromatic system containing one or more rings (typically one, two, or three rings), which may be joined together in a pendant fashion, such as in biphenyl, or fused, such as in naphthalene. Examples of aryl groups include phenyl, anthracyl, and naphthyl. In various embodiments, example aryl groups are phenyl (e.g., C-aryl) and biphenyl (e.g., C 12 In some embodiments, the aryl group has 6 to 16 carbon atoms. In some embodiments, the aryl group has 6 to 12 carbon atoms (e.g., C 6~12 -aryl). In some embodiments, the aryl group has 6 carbon atoms (e.g., C6-aryl).

[0050] The terms "heteroaryl" or "heteroaromatic" as used herein refer to heterocycles having aromatic character. Heteroaryl substituents can be defined by the number of carbon atoms, e.g., C 1~15 -heteroaryl indicates the number of carbon atoms in the heteroaryl group, not including the number of heteroatoms. For example, C 1~9 Heteroaryl will contain 1 to 4 additional heteroatoms. Polycyclic heteroaryls can contain one or more rings that are partially saturated. Non-limiting examples of heteroaryls include pyridyl, pyrazinyl, pyrimidinyl (including, for example, 2- and 4-pyrimidinyl), pyridazinyl, thienyl, furyl, pyrrolyl (including, for example, 2-pyrrolyl), imidazolyl, thiazolyl, oxazolyl, pyrazolyl (including, for example, 3- and 5-pyrazolyl), isothiazolyl, 1,2,3-triazolyl, 1,2,4-triazolyl, 1,3,4-triazolyl, tetrazolyl, 1,2,3-thiadiazolyl, 1,2,3-oxadiazolyl, 1,3,4-thiadiazolyl, and 1,3,4-oxadiazolyl.

[0051] Non-limiting examples of polycyclic heterocycles and heteroaryls include indolyl (including, for example, 3-, 4-, 5-, 6-, and 7-indolyl), indolinyl, quinolyl, tetrahydroquinolyl, isoquinolyl (including, for example, 1- and 5-isoquinolyl), 1,2,3,4-tetrahydroisoquinolyl, cinnolinyl, quinoxalinyl (including, for example, 2- and 5-quinoxalinyl), quinazolinyl, phthaldinyl, 1,8-naphthyridinyl, 1,4-benzodioxanyl, coumarin, dihydrocoumarin, 1,5-naphthyridinyl, benzofuryl (including, for example, Examples include 3-, 4-, 5-, 6- and 7-benzofuryl), 2,3-dihydrobenzofuryl, 1,2-benzisoxazolyl, benzothienyl (including, for example, 3-, 4-, 5-, 6- and 7-benzothienyl), benzoxazolyl, benzothiazolyl (including, for example, 2-benzothiazolyl and 5-benzothiazolyl), purinyl, benzimidazolyl (including, for example, 2-benzimidazolyl), benzotriazolyl, thioxanthinyl, carbazolyl, carbolinyl, acridinyl, pyrrolidinyl and quinolidinyl.

[0052] As used herein, the acronym DBCO refers to 8,9-dihydro-3H-dibenzo[b,f][1,2,3]triazolo[4,5-d]azocine.

[0053] The term "protecting group" or "chemical protecting group" refers to a chemical moiety that blocks some or all reactive moieties in a compound, preventing such moieties from participating in a chemical reaction until the protecting group is removed, such as those listed and described in T.W. Greene, P.G.M. Wuts, Protective Groups in Organic Synthesis, 3rd Edition, John Wiley & Sons (1999). When different protecting groups are used, it can be advantageous for each (different) protecting group to be removable by a different means. Protecting groups that are cleaved under completely different reaction conditions allow for different removal of such protecting groups from others. For example, protecting groups can be removed by acid, base, and hydrogenolysis. Groups such as trityl, monomethoxytrityl, dimethoxytrityl, acetal, and tert-butyldimethylsilyl can be used to protect carboxy and hydroxy reactive moieties, in the presence of amino groups protected with the acid-labile, hydrogenolysis-removable Cbz group and the base-labile Fmoc group. Carboxylic acid moieties may be blocked with base-labile groups such as, but not limited to, methyl or ethyl, and hydroxy-reactive moieties may be blocked with acid-labile groups such as tert-butyl carbamate, or with base-labile groups such as acetyl in the presence of an acid- and base-stable but hydrolytically removable carbamate-blocked amine.

[0054] Carboxylic acid and hydroxyl reactive moieties may be blocked with hydrolytically removable protecting groups such as benzyl groups, while amine groups may be blocked with base-labile groups such as Fmoc. A particularly useful amine protecting group for the synthesis of compounds of formula (I) is trifluoroacetamide. Carboxylic acid reactive moieties may be blocked with oxidatively removable protecting groups such as 2,4-dimethoxybenzyl, and coexisting amino groups may be blocked with fluoride-labile silyl carbamates.

[0055] Allyl blocking groups are useful in the presence of acid and base protecting groups because the former are stable and can be subsequently removed by metal or bis-acid catalysts. For example, an allyl-blocked carboxylic acid can be deprotected with a palladium(0)-catalyzed reaction in the presence of acid-labile t-butyl carbamate or base-labile acetate amine protecting groups. Yet another form of protecting group is a resin to which a compound or intermediate can be attached. As long as the residue is attached to the resin, the functional group is blocked and cannot react. Once released from the resin, the functional group becomes reactive.

[0056] The terms "nucleobase," "base-pairing moiety," "nucleobase-pairing moiety," or "base" refer to the heterocyclic portion of a nucleoside, nucleotide, and / or morpholino subunit. The nucleobase may be naturally occurring, modified, or an analog of these naturally occurring nucleobases, e.g., one or more nitrogen atoms of the nucleobase may be independently replaced with carbon at each occurrence. Exemplary analogs include hypoxanthine (the base component of the nucleoside inosine); 2,6-diaminopurine; 5-methylcytosine; C5-propynyl-modified pyrimidines; 10-(9-(aminoethoxy)phenoxazinyl) (G-clamp); and the like.

[0057] Further examples of base-pairing moieties include uracil, thymine, adenine, cytosine, guanine, and pyrimidine analogs such as hypoxanthine, 2-fluorouracil, 2-fluorocytosine, 5-bromouracil, 5-iodouracil, 2,6-diaminopurine, azacytosine, pseudoisocytosine, and pseudouracil, having their respective amino groups protected by acyl protecting groups, and modified nucleobases such as 8-substituted purines, xanthines, or hypoxanthines (the latter two being natural degradation products). Also contemplated are modified nucleobases disclosed in Chiu and Rana, RNA, 2003, 9, 1034-1048; Limbach et al., Nucleic Acids Research, 1994, 22, 2183-2196; and Revankar and Rao, Comprehensive Natural Products Chemistry, Vol. 7, 313, the contents of which are incorporated herein by reference.

[0058] Further examples of base pairing moieties include, but are not limited to, extended size nucleobases with one or more additional benzene rings. Nucleobase substitutions described in the Glen Research catalog (www.glenresearch.com); Krueger AT et al., Acc. Chem. Res., 2007, 40, 141-150; Kool, ET, Acc. Chem. Res., 2002, 35, 936-943; Benner SA et al., Nat. Rev. Genet., 2005, 6, 553-543; Romesberg, FE et al., Curr. Opin. Chem. Biol., 2003, 7, 723-733; Hirao, I., Curr. Opin. Chem. Biol., 2006, 10, 622-627 are believed to be useful in the synthesis of the oligomers described herein, the contents of which are incorporated herein by reference. Examples of extended-size nucleobases are shown below:

[0059] [ka]

[0060] The term "oligonucleotide" or "oligomer" refers to a compound comprising multiple linked nucleosides, nucleotides, or a combination of both nucleosides and nucleotides. In certain embodiments provided herein, the oligonucleotide is a morpholino oligonucleotide.

[0061] The term "morpholino oligonucleotide" or "PMO" refers to a modified oligonucleotide having morpholino subunits linked together by phosphoramidate or phosphorodiamidate bonds linking the morpholino nitrogen of one subunit to the 5' exocyclic carbon of an adjacent subunit. Each morpholino subunit contains a nucleobase-pairing moiety effective to bind to a nucleobase in a target by nucleobase-specific hydrogen bonding.

[0062] The terms "antisense oligomer," "antisense compound," and "antisense oligonucleotide" are used interchangeably and refer to a sequence of subunits joined by intersubunit bonds, each carrying a base-pairing moiety that allows the base-pairing moiety to hybridize to a target sequence in a nucleic acid (typically RNA) by Watson-Crick base pairing to form a nucleic acid:oligomer heteroduplex within the target sequence. The oligomer may have exact (perfect) or near (full) sequence complementarity to the target sequence; variation in sequence near the ends of the oligomer is generally preferred over variation within it.

[0063] Such antisense oligomers can be designed to block or inhibit mRNA translation or inhibit / alter splicing processing of natural or aberrant pre-mRNA and can be said to be "directed" or "targeted" to the target sequence to which they hybridize. The target sequence is typically the AUG start codon region of an mRNA, a translation suppressor oligomer, or a splice site region of a pre-processed mRNA, a splice suppressor oligomer (SSO). A splice site target sequence can include an mRNA sequence having its 5' end 1 to about 25 base pairs downstream of the normal splice acceptor junction in the pre-processed mRNA. In various embodiments, the target sequence can be any region of a pre-processed mRNA that includes a splice site, is contained entirely within an exon-coding sequence, or spans the splice acceptor or donor site. When an oligomer targets a target nucleic acid in the above manner, it is more commonly said to "target" a biologically relevant target, such as a protein, virus, or bacterium.

[0064] Antisense oligonucleotides and target RNAs are complementary to each other when a sufficient number of corresponding positions in each molecule are occupied by nucleotides that can hydrogen bond with each other, so that stable and specific binding occurs between the oligonucleotide and the target. Thus, "specifically hybridizable" and "complementary" are terms used to indicate a sufficient degree of complementarity or exact match so that stable and specific binding occurs between the oligonucleotide and the target. It is understood in the art that the sequence of an oligonucleotide does not need to be 100% complementary to the sequence of its target sequence to be specifically hybridizable. An oligonucleotide is specifically hybridizable when there is a sufficient degree of complementarity to avoid non-specific binding of the antisense oligonucleotide to non-target sequences under conditions where specific binding is desired, i.e., under physiological conditions in the case of in vivo assays or therapeutic treatments, or under the conditions under which the assay is performed in the case of in vitro assays.

[0065] Oligonucleotides may also contain modified or substituted nucleobases (often simply referred to in the art as "bases"). Oligonucleotides containing modified or substituted bases include oligonucleotides in which one or more of the most commonly found purine or pyrimidine bases in nucleic acids are replaced with less common or unnatural bases. In some embodiments, the nucleobase is covalently linked to the morpholine ring of the nucleotide or nucleoside at the N atom of the purine base or the N atom of the pyrimidine base.

[0066] Purine bases contain a pyrimidine ring fused to an imidazole ring, as represented by the general formula:

[0067] [ka]

[0068] Adenine and guanine are the two most commonly found purine nucleobases in nucleic acids, which can be substituted with other naturally occurring purines, including, but not limited to, N6-methyladenine, N2-methylguanine, hypoxanthine, and 7-methylguanine.

[0069] Pyrimidine bases contain a six-membered pyrimidine ring, as represented by the general formula:

[0070] [ka]

[0071] Cytosine, uracil, and thymine are the most commonly found pyrimidine bases in nucleic acids. They may be substituted with other naturally occurring pyrimidines, including, but not limited to, 5-methylcytosine, 5-hydroxymethylcytosine, pseudouracil, and 4-thiouracil. In one embodiment, the oligonucleotides described herein contain thymine bases instead of uracil.

[0072] Other modified or substituted bases include 2,6-diaminopurine, orotic acid, agmatidine, lysidine, 2-thiopyrimidines (e.g., 2-thiouracil, 2-thiothymine), G-clamps and their derivatives, 5-substituted pyrimidines (e.g., 5-halouracil, 5-propynyluracil, 5-propynylcytosine, 5-aminomethyluracil, 5-hydroxymethyluracil, 5-aminomethylcytosine, 5-hydroxymethylcytosine, super T), 7-deazaguanine, 7-deazaadenine, 7-aza-2,6-diaminopurine, 8-aza-7-deazaguanine, 8-aza-7-deazaadenine, 8-aza-7-de ... These bases include, but are not limited to, diaminopurine, super-G, super-A, and N4-ethylcytosine, or derivatives thereof; N2-cyclopentylguanine (cPent-G), N2-cyclopentyl-2-aminopurine (cPent-AP), N2-propyl-2-aminopurine (Pr-AP), pseudouracil, or derivatives thereof; and degenerate or universal bases such as 2,6-difluorotoluene, or absent bases such as abasic sites (e.g., 1-deoxyribose, 1,2-dideoxyribose, 1-deoxy-2-O-methylribose, or pyrrolidine derivatives in which the ring oxygen is replaced with nitrogen (azaribose)). Pseudouracil is an isomeric version of the naturally occurring uracil, with a C-glycoside rather than the usual N-glycoside as in uridine.

[0073] Certain modified or substituted nucleobases are particularly useful for increasing the binding affinity of the antisense oligonucleotides of the present disclosure. These include 5-substituted pyrimidines, 6-azapyrimidines, and N-2, N-6, and O-6 substituted purines, including 2-aminopropyladenine, 5-propynyluracil, and 5-propynylcytosine. In various embodiments, the nucleobase may contain 5-methylcytosine substitutions, which have been shown to increase nucleic acid duplex stability by 0.6 to 1.2°C.

[0074] In some embodiments, modified or substituted nucleobases are useful for facilitating the purification of antisense oligonucleotides. For example, in certain embodiments, antisense oligonucleotides may contain three or more (e.g., 3, 4, 5, 6, or more) consecutive guanine bases. In certain antisense oligonucleotides, a series of three or more consecutive guanine bases may cause aggregation of the oligonucleotide, complicating purification. In such antisense oligonucleotides, one or more consecutive guanines may be substituted with hypoxanthine. Substituting one or more guanines in a series of three or more consecutive guanine bases with hypoxanthine can reduce aggregation of the antisense oligonucleotide, thereby facilitating purification.

[0075] The oligonucleotides provided herein are synthetic and do not include biologically derived antisense compositions. The molecules of the present disclosure may also be mixed, encapsulated, conjugated, or otherwise associated with other molecules, molecular structures, or mixtures of compounds, such as, for example, liposomes, receptor-targeting molecules, oral, rectal, topical, or other formulations, to aid in uptake, distribution, or absorption, or a combination thereof.

[0076] The terms "complementary" and "complementarity" refer to oligonucleotides (i.e., sequences of nucleotides) related by the base-pairing rules. For example, the sequence "TGA(5'-3')" is complementary to the sequence "TCA(5'-3')." Complementarity can be "partial," in which only a portion of the nucleic acid bases match according to the base-pairing rules. Alternatively, there can be "complete," "total," or "perfect" (100%) complementarity between nucleic acids. The degree of complementarity between nucleic acid strands significantly affects the efficiency and strength of hybridization between nucleic acid strands. While complete complementarity is often desired, some embodiments can include one or more, preferably 6, 5, 4, 3, 2, or 1, mismatches with respect to the target RNA. Such hybridization can occur with "near" or "substantial" complementarity, as well as exact complementarity, of the antisense oligomer to the target sequence. In some embodiments, an oligomer may hybridize to a target sequence with about 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 99%, or 100% complementarity. Variation anywhere within the oligomer is included. In certain embodiments, sequence variation near the ends of the oligomer is generally preferred over internal variation, and, if present, is typically within about 6, 5, 4, 3, 2, or 1 nucleotide of the 5' end, 3' end, or both ends.

[0077] The term "peptide" refers to a compound comprising multiple linked amino acids. The peptides provided herein can be considered to be cell-penetrating peptides.

[0078] The terms "cell-penetrating peptide" and "CPP" are used interchangeably and refer to cationic cell-penetrating peptides, also known as transport peptides, carrier peptides, or peptide transduction domains. The peptides provided herein have the ability to induce cell penetration within 100% of the cells in a given cell culture population, allowing for the translocation of macromolecules into multiple tissues in vivo upon systemic administration. In various embodiments, CPP embodiments of the present disclosure can include arginine-rich peptides, as further described below.

[0079] As used herein, the term "chimeric peptide" refers to a polypeptide comprising a first portion, which is a first peptide or a fragment thereof, fused to a second portion, which is a different peptide or a fragment thereof. A chimeric peptide can comprise two or more covalently linked peptides. The peptides can be covalently linked via amino acid side chains, the N-terminus, the C-terminus, or any combination thereof. In certain embodiments, peptides are covalently linked via the N-terminus of one peptide to the C-terminus of the other. In certain embodiments, the covalent linker is an amide bond.

[0080] As used herein, the term "trimeric peptide" refers to a polypeptide comprising a first portion, which is a first peptide or a fragment thereof, fused to a second portion, which is a different peptide or a fragment thereof, and fused to a third portion, which is a different peptide or a fragment thereof. A trimeric peptide can comprise three or more covalently linked peptides. The peptides can be covalently linked via amino acid side chains, N-terminus, C-terminus, or any combination thereof. In certain embodiments, the peptides are covalently linked via the N-terminus of one peptide to the C-terminus of the other. In certain embodiments, the covalent linker is an amide bond.

[0081] As used herein, the term "MACH peptide" refers to a polypeptide comprising a cationic cell-penetrating peptide, also known as a transport peptide, carrier peptide, or peptide transduction domain. The peptides provided herein are capable of inducing cell penetration within 100% of cells in a given cell culture population, enabling macromolecules to be translocated into multiple tissues in vivo upon systemic administration. A MACH peptide can comprise three or more covalently linked peptides. The peptides can be covalently linked via amino acid side chains, the N-terminus, the C-terminus, or any combination thereof. In certain embodiments, peptides are covalently linked via the N-terminus of one peptide to the C-terminus of the other. In certain embodiments, the covalent linker is an amide bond. In certain embodiments, a MACH peptide is composed of peptides optimized for cellular delivery using machine learning methods. Examples of MACH peptides can be found in Table 4 provided herein.

[0082] As used herein, the term "amphipathic peptide" refers to a peptide having separate regions of essentially charged and essentially uncharged amino acids, known as hydrophilic and hydrophobic peptide segments, respectively.

[0083] As used herein, the term "oligoarginine peptide" refers to a peptide that is composed entirely of arginine or mostly arginine amino acid residues. In certain embodiments, the peptide is composed entirely of arginine amino acid residues. In certain embodiments, the peptide is composed of 50-99% arginine amino acid residues separated by amino acid linkers, such as, but not limited to, aminohexanoic acid or beta-alanine. In certain embodiments, the peptide is composed of 75% arginine amino acid residues separated by amino acid linkers, such as, but not limited to, aminohexanoic acid or beta-alanine.

[0084] As used herein, the term "nuclear targeting peptide" refers to a peptide that contains a nuclear localization sequence that enables the peptide to import a protein into the cell nucleus by nuclear transport. In certain embodiments, this sequence consists of one or more positively charged amino acids exposed on the protein surface.

[0085] As used herein, the term "endosomal disrupting peptide" refers to a peptide that can aid in the release of a drug into the cytoplasm of a cell. In certain embodiments, the sequence consists of one or more positively charged amino acids.

[0086] The term "treatment" refers to the application of one or more specific procedures used to ameliorate a disease. In certain embodiments, the specific procedure is the administration of one or more pharmaceutical agents. "Treatment" of an individual (e.g., a mammal, such as a human) or cell is any type of intervention used in an attempt to alter the natural course of the individual or cell. Treatment includes, but is not limited to, the administration of a pharmaceutical composition and may be administered prophylactically or after the initiation of a pathological event or contact with a pathogenic agent. Treatment includes any desired effect on the symptoms or pathology of a disease or condition and may include, for example, a minimal change or improvement in one or more measurable markers of the disease or condition being treated. Also included are "prophylactic" treatments, which may be directed at reducing the rate of progression of the disease or condition being treated, delaying the onset of the disease or condition, or reducing the severity of its onset.

[0087] An "effective amount" or "therapeutically effective amount" refers to an amount of a therapeutic compound, such as an antisense oligomer, administered to a mammalian subject, either in a single dose or as part of a series, effective to produce the desired therapeutic effect.

[0088] The term "amelioration" refers to a reduction in the severity of at least one indicator of a condition or disease. In certain embodiments, improvement includes a delay or slowing of the progression of one or more indicators of a condition or disease. The severity of an indicator may be determined by subjective or objective measures known to those skilled in the art.

[0089] As used herein, "pharmaceutically acceptable salt" refers to a derivative of the disclosed oligonucleotide, in which the parent oligonucleotide is modified by converting an existing acid or base moiety into its salt form. Lists of suitable salts can be found in Remington's Pharmaceutical Sciences, 17th Edition, Mack Publishing Company, Easton, Pa., 1985, p. 1418, and Journal of Pharmaceutical Science, 66, 2 (1977), each of which is incorporated herein by reference in its entirety.

[0090] Peptide-oligonucleotide conjugates Provided herein are oligonucleotides chemically conjugated to cell-penetrating peptides that enhance the activity, cellular distribution, or cellular uptake of the oligonucleotides.

[0091] In one embodiment, the cell penetrating peptide consists of a MACH peptide.

[0092] In one embodiment, the cell-penetrating peptide is a MACH peptide that has been optimized using machine learning methods.

[0093] The oligonucleotide may further be chemically conjugated to one or more heteroalkyl moieties (e.g., polyethylene glycol) that further enhance the activity, cellular distribution, or cellular uptake of the oligonucleotide. In one exemplary embodiment, a cell-penetrating peptide is covalently attached to either or both ends of the oligonucleotide at its N-terminal or C-terminal residue.

[0094] Thus, in one aspect, provided herein is a peptide-oligonucleotide conjugate of Formula I:

[0095] [ka]

[0096] or a pharmaceutically acceptable salt thereof, During the ceremony, A' is -N(H)CH2C(O)NH2, -N(C 1~6 -alkyl)CH2C(O)NH2,

[0097] [ka]

[0098] is selected from R 5 is —C(O)(O-alkyl)x-OH, where x is 3 to 10, and each alkyl group, independently at each occurrence, is C 2~6 -alkyl, Or R 5 is -C(O)C 1~6 -Alkyl, trityl, monomethoxytrityl, -(C 1~6 -alkyl)-R 6 , -(C 1~6 -heteroalkyl)-R 6 , aryl-R 6 , heteroaryl-R 6 , -C(O)O-(C 1~6 -alkyl)-R 6 , -C(O)O-aryl-R 6 , —C(O)O-heteroaryl-R 6 , and

[0099] [ka]

[0100] Selected from; R 6 is selected from OH, SH, and NH2, or R 6 is O, S, or NH, each of which is covalently attached to a solid support; Each R 1 is OH and -N(R 3 )(R 4 ) are independently selected from each R 3 and R4 independently for each occurrence, -C 1~6 - is alkyl; Each R 2 is independently, at each occurrence, selected from H, a nucleobase, and a nucleobase functionalized with a chemical protecting group, and the nucleobase is, at each occurrence, independently selected from pyridine, pyrimidine, triazinane, purine, and deazapurine. 3~6 -contains heterocyclic rings; z is 8 to 40; E' is H, -C 1~6 -Alkyl, -C(O)C 1~6 -Alkyl, benzoyl, stearoyl, trityl, monomethoxytrityl, dimethoxytrityl, trimethoxytrityl,

[0101] [ka]

[0102] Selected from; Q is —C(O)(CH)C(O)— or —C(O)(CH)S(CH)C(O)—; R 7 is -(CH2)2OC(O)N(R 8 )2 and R 8 is -(CH2)6NHC(=NH)NH2; L is -C(O)(CH2) 1~6 -C 7~15 -Heteroaromatic-(CH2) 1~6 C(O)—, and L is covalently linked to J by an amide bond; J is a carrier peptide; G is H, -C(O)C 1~6 - selected from alkyl, benzoyl, and stearoyl, and G is covalently linked to J; The following conditions: 1) A' is

[0103] [ka]

[0104] or 2) E' is

[0105] [ka]

[0106] Being At least one of the following is true: The carrier peptide J is selected from the following sequences:

[0107] [Table 3A]

[0108] [Table 3B]

[0109] [Table 3C]

[0110] wherein X is 6-aminohexanoic acid and B is β-alanine; C L 1 By another C is covalently bound to; L 1 teeth,

[0111] [ka]

[0112] and; M is

[0113] [ka]

[0114] and; R10 is independently at each occurrence H or halogen.

[0115] In one embodiment, z is 8 to 30. In another embodiment, z is 10 to 30. In a further embodiment, z is 15 to 25. In another embodiment, z is 20 to 25. In one embodiment, z is 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, or 30.

[0116] In yet another embodiment, E' is H, -C 1~6 -Alkyl, -C(O)C 1~6 -alkyl, benzoyl, stearoyl, trityl, monomethoxytrityl, dimethoxytrityl, trimethoxytrityl, and

[0117] [ka]

[0118] is selected from.

[0119] In another embodiment, A' is -N(C 1~6 -alkyl)CH2C(O)NH2,

[0120] [ka]

[0121] is selected from.

[0122] In yet another embodiment, E' is H, -C(O)CH3, benzoyl, stearoyl, trityl, 4-methoxytrityl, and

[0123] [ka]

[0124] is selected from.

[0125] In yet another embodiment, A' is -N(C 1~6 -alkyl)CH2C(O)NH2,

[0126] [ka]

[0127] is selected from; and E' is

[0128] [ka]

[0129] is.

[0130] In another embodiment, A' is

[0131] [ka]

[0132] and E' is selected from H, -C(O)CH3, trityl, 4-methoxytrityl, benzoyl, and stearoyl.

[0133] In one embodiment, the peptide-oligonucleotide conjugate of Formula I is a peptide-oligonucleotide conjugate of Formula Ia.

[0134] [ka]

[0135] In one embodiment, the peptide-oligonucleotide conjugate of Formula I is a peptide-oligonucleotide conjugate of Formula Ib.

[0136] [ka]

[0137] In the formula, E' is H, C 1~6 -alkyl, -C(O)CH3, benzoyl, and stearoyl.

[0138] In embodiments of Formulas I, Ia, and Ib, each R 1 is N(CH3)2.

[0139] In still other embodiments of Formulas I, Ia, and Ib, each R 2 is a nucleobase, wherein the nucleobase is independently selected at each occurrence from pyridine, pyrimidine, triazinane, purine, and deazapurine. 4~6 Contains heterocycles.

[0140] In another embodiment of Formulas I, Ia, and Ib, each R 2 is a nucleobase, wherein the nucleobase is independently selected at each occurrence from pyrimidine, purine, and deazapurine. 4~6 Contains heterocycles.

[0141] In still other embodiments of Formulas I, Ia, and Ib, each R 2 is a nucleobase selected independently at each occurrence from adenine, 2,6-diaminopurine, 7-deaza-adenine, guanine, 7-deaza-guanine, hypoxanthine, cytosine, 5-methyl-cytosine, thymine, uracil, and hypoxanthine.

[0142] In still other embodiments of Formulas I, Ia, and Ib, each R 2 is, independently at each occurrence, a nucleobase selected from adenine, guanine, cytosine, 5-methyl-cytosine, thymine, uracil, and hypoxanthine.

[0143] In another embodiment of Formula I, Ia, and Ib, L is -C(O)(CH) 1~6 -DBCO-(CH2) 1~6 It is C(O)-.

[0144] In another embodiment of Formulas I, Ia, and Ib, L is

[0145] [ka]

[0146] is.

[0147] In another embodiment of Formulas I, Ia, and Ib, M is

[0148] [ka]

[0149] is.

[0150] In still other embodiments of Formulas I, Ia, and Ib, M is

[0151] [ka]

[0152] is.

[0153] In another embodiment of Formulas I, Ia, and Ib, L 1 is P 1 and P 2 covalently attached to the side chain of the terminal cysteine ​​above to form the following structure:

[0154] [ka]

[0155] In another embodiment of Formulas I, Ia, and Ib, G is selected from H, C(O)CH3, benzoyl, and stearoyl.

[0156] In still other embodiments of Formulas I, Ia, and Ib, G is H or —C(O)CH 3 .

[0157] In still other embodiments of Formulas I, Ia, and Ib, G is H.

[0158] In still other embodiments of Formulas I, Ia, and Ib, G is —C(O)CH 3 .

[0159] In yet another embodiment of Formulas I, Ia, and Ib, the oligonucleotide-peptide conjugate exhibits at least a 40-fold improvement in uptake compared to the unconjugated oligonucleotide.

[0160] In further embodiments of Formulas I, Ia, and Ib, the oligonucleotide-peptide conjugate exhibits at least a 5-fold improvement in uptake compared to the unconjugated oligonucleotide.

[0161] In one embodiment, the oligonucleotide-peptide conjugate is non-toxic.

[0162] In another embodiment, the oligonucleotide-peptide conjugate is non-immunogenic.

[0163] In another aspect, provided herein is a peptide-oligonucleotide conjugate of formula II:

[0164] [ka]

[0165] or a pharmaceutically acceptable salt thereof; During the ceremony, A' is -N(H)CH2C(O)NH2, -N(C 1~6 -alkyl)CH2C(O)NH2,

[0166] [ka]

[0167] is selected from R 5 is —C(O)(O-alkyl)x-OH, where x is 3 to 10, and each alkyl group, independently at each occurrence, is C 2~6 -alkyl, Or R 5 is -C(O)C 1~6 -Alkyl, trityl, monomethoxytrityl, -(C 1~6 -alkyl)-R 6 , -(C 1~6 -heteroalkyl)-R 6 , aryl-R 6 , heteroaryl-R 6 , -C(O)O-(C 1~6 -alkyl)-R 6 , -C(O)O-aryl-R 6 , —C(O)O-heteroaryl-R 6 , and

[0168] [ka]

[0169] Selected from; R 6 is selected from OH, SH, and NH2, or R 6 is O, S, or NH, each of which is covalently attached to a solid support; Each R 1 is OH and -N(R 3 )(R 4 ) are independently selected from each R 3 and R 4 independently for each occurrence, -C 1~6 - is alkyl; Each R 2 is independently, at each occurrence, selected from H, a nucleobase, and a nucleobase functionalized with a chemical protecting group, and the nucleobase is, at each occurrence, independently selected from pyridine, pyrimidine, triazinane, purine, and deazapurine. 3~6 -contains heterocyclic rings; z is 8 to 40; E' is H, -C 1~6 -Alkyl, -C(O)C 1~6 -Alkyl, benzoyl, stearoyl, trityl, monomethoxytrityl, dimethoxytrityl, trimethoxytrityl,

[0170] [ka]

[0171] Selected from; Q is —C(O)(CH)C(O)— or —C(O)(CH)S(CH)C(O)—; R 7 is -(CH2)2OC(O)N(R 8 )2 and R 8 is -(CH2)6NHC(=NH)NH2; L is -C(O)(CH2) 1~6 -C 7~15 -Heteroaromatic-(CH2) 1~6 C(O)—, and L is covalently linked to J by an amide bond; J is a carrier peptide; G is H, -C(O)C 1~6 - selected from alkyl, benzoyl, and stearoyl, and G is covalently linked to J; The following conditions: 1) A' is

[0172] [ka]

[0173] or 2) E' is

[0174] [ka]

[0175] Being At least one of the following is true: The carrier peptide J is selected from the following sequences:

[0176] [Table 4A]

[0177] [Table 4B]

[0178] [Table 4C]

[0179] [Table 4D]

[0180] wherein X is 6-aminohexanoic acid and B is β-alanine; C L 1 By another C is covalently bound to; L 1 teeth

[0181] [ka]

[0182] and; M is

[0183] [ka]

[0184] and; R 10 is independently at each occurrence H or halogen.

[0185] In one embodiment, z is 8 to 30. In another embodiment, z is 10 to 30. In a further embodiment, z is 15 to 25. In another embodiment, z is 20 to 25. In one embodiment, z is 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, or 30.

[0186] In yet another embodiment, E' is H, -C 1~6 -Alkyl, -C(O)C 1~6 -alkyl, benzoyl, stearoyl, trityl, monomethoxytrityl, dimethoxytrityl, trimethoxytrityl, and

[0187] [ka]

[0188] is selected from.

[0189] In another embodiment, A' is -N(C 1~6 -alkyl)CH2C(O)NH2,

[0190] [ka]

[0191] is selected from.

[0192] In yet another embodiment, E' is H, -C(O)CH3, benzoyl, stearoyl, trityl, 4-methoxytrityl, and

[0193] [ka]

[0194] is selected from.

[0195] In yet another embodiment, A' is -N(C 1~6-alkyl)CH2C(O)NH2,

[0196] [ka]

[0197] is selected from; and E' is

[0198] [ka]

[0199] is.

[0200] In another embodiment, A' is

[0201] [ka]

[0202] and E' is selected from H, -C(O)CH3, trityl, 4-methoxytrityl, benzoyl, and stearoyl.

[0203] In one embodiment, the peptide-oligonucleotide conjugate of formula IA is a peptide-oligonucleotide conjugate of formula Ia.

[0204] [ka]

[0205] In one embodiment, the peptide-oligonucleotide conjugate of formula IA is a peptide-oligonucleotide conjugate of formula Ib:

[0206] [ka]

[0207] In the formula, E' is H, C 1~6 -alkyl, -C(O)CH3, benzoyl, and stearoyl.

[0208] In embodiments of Formulas II, I, Ia, and Ib, each R 1 is N(CH3)2.

[0209] In still other embodiments of Formulas II, I, Ia, and Ib, each R 2 is a nucleobase, wherein the nucleobase is independently selected at each occurrence from pyridine, pyrimidine, triazinane, purine, and deazapurine. 4~6 Contains heterocycles.

[0210] In another embodiment of Formulas II, I, Ia, and Ib, each R 2 is a nucleobase, wherein the nucleobase is independently selected at each occurrence from pyrimidine, purine, and deazapurine. 4~6 Contains heterocycles.

[0211] In still other embodiments of Formulas II, I, Ia, and Ib, each R 2 is a nucleobase selected independently at each occurrence from adenine, 2,6-diaminopurine, 7-deaza-adenine, guanine, 7-deaza-guanine, hypoxanthine, cytosine, 5-methyl-cytosine, thymine, uracil, and hypoxanthine.

[0212] In still other embodiments of Formulas II, I, Ia, and Ib, each R 2 is, independently at each occurrence, a nucleobase selected from adenine, guanine, cytosine, 5-methyl-cytosine, thymine, uracil, and hypoxanthine.

[0213] In another embodiment of Formula II, I, Ia, and Ib, L is -C(O)(CH) 1~6 -DBCO-(CH2) 1~6 It is C(O)-.

[0214] In another embodiment of Formulas II, I, Ia, and Ib, L is

[0215] [ka]

[0216] is.

[0217] In another embodiment of Formulas II, I, Ia, and Ib, M is

[0218] [ka]

[0219] is.

[0220] In still other embodiments of Formulas II, I, Ia, and Ib, M is

[0221] [ka]

[0222] is.

[0223] In another embodiment of Formulas II, I, Ia, and Ib, L 1 is P 1 and P 2 covalently attached to the side chain of the terminal cysteine ​​above to form the following structure:

[0224] [ka]

[0225] In other embodiments of Formulas II, I, Ia, and Ib, G is selected from H, C(O)CH3, benzoyl, and stearoyl.

[0226] In still other embodiments of Formulas II, I, Ia, and Ib, G is H or —C(O)CH 3 .

[0227] In still other embodiments of Formulas II, I, Ia, and Ib, G is H.

[0228] In still other embodiments of Formulas II, I, Ia, and Ib, G is -C(O)CH3.

[0229] In yet other embodiments of Formulas II, I, Ia, and Ib, the oligonucleotide-peptide conjugates exhibit at least a 40-fold improvement in uptake compared to the unconjugated oligonucleotide.

[0230] In further embodiments of Formulas II, I, Ia, and Ib, the oligonucleotide-peptide conjugate exhibits at least a 5-fold improvement in uptake compared to the unconjugated oligonucleotide.

[0231] In one embodiment, the oligonucleotide-peptide conjugate is non-toxic.

[0232] In another embodiment, the oligonucleotide-peptide conjugate is non-immunogenic.

[0233] Trimeric peptides In certain embodiments, the trimeric peptides are useful for generating libraries of oligonucleotide cell-penetrating peptide conjugates for training.

[0234] A non-limiting representation of such a trimeric peptide is shown below:

[0235] [ka]

[0236] wherein the C-terminus is covalently attached to the oligonucleotide.

[0237] In one embodiment, each trimeric peptide is three covalently linked cell-penetrating peptides, and the cell-penetrating peptides are independently an amphipathic peptide, a nuclear targeting peptide, an endosome-disrupting peptide, a chimeric peptide, a cyclic peptide, a bicyclic peptide, or an oligoarginine peptide.

[0238] In another embodiment, each trimeric peptide is three covalently linked cell-penetrating peptides, one of the cell-penetrating peptides is an amphipathic peptide, one of the cell-penetrating peptides is a nuclear targeting peptide, and one of the peptides is an additional cell-penetrating peptide.

[0239] In yet another embodiment, each trimeric peptide is three covalently linked cell-penetrating peptides, the three cell-penetrating peptides comprising one amphipathic peptide, one nuclear-targeting peptide, and one additional cell-penetrating peptide, the amphipathic peptide being at the N-terminus of the trimeric peptide, the nuclear-targeting peptide being a middle peptide, and the additional cell-penetrating peptide being at the C-terminus of the trimeric peptide.

[0240] In yet another embodiment, the amphipathic peptide comprises a hydrophobic peptide segment and a hydrophilic peptide segment, wherein the hydrophobic peptide segment comprises a sequence of 2 to 10 amino acids independently selected from glycine, isoleucine, alanine, valine, leucine, phenylalanine, tyrosine, or tryptophan, the hydrophilic peptide segment comprises a sequence of 2 to 20 amino acids independently selected from charged amino acids, uncharged polar amino acids, or hydrophobic amino acids, and the hydrophilic peptidyl segment comprises at least one non-hydrophobic amino acid.

[0241] In one embodiment, the hydrophilic peptide segment comprises a sequence of 2 to 20 amino acids independently selected from arginine, lysine, glutamine, asparagine, histidine, serine, threonine, tryptophan, alanine, isoleucine, leucine, methionine, phenylalanine, valine, proline, or glycine, and the hydrophilic peptide segment comprises at least one non-hydrophobic amino acid.

[0242] Various non-limiting embodiments of peptides for the trimer peptide are provided in Table 2:

[0243] [Table 5A]

[0244] [Table 5B]

[0245] The bolded cysteine ​​is conjugated to decafluorobiphenyl. The italicized cysteine ​​is conjugated to 1,3,5-trisbromomethylbenzene.

[0246] Representative peptide-oligonucleotide conjugates of the present disclosure include, among others, trimeric peptide-oligonucleotide conjugates of the following structure:

[0247] [ka]

[0248] or a pharmaceutically acceptable salt thereof, wherein: G is H or -C(O)CH; R 2 is, independently at each occurrence, a nucleobase selected from adenine, guanine, cytosine, 5-methyl-cytosine, thymine, uracil, and hypoxanthine; K is -C(O)(CH2) 1~6 -C 7~15 -Heteroaromatic-(CH2) 1~6 C(O)-; M is

[0249] [ka]

[0250] and And R 10 is, independently at each occurrence, H or halogen; L 1 is P 1 and P 2 covalently attached to the side chain of a terminal or internal cysteine ​​on z is 8 to 40; P 1 , P 2 , and P 3 are each independently a cell-penetrating peptide, and P 1 and P 2 each contain at least one cysteine ​​amino acid residue, and each of the cell-penetrating peptides is independently an amphipathic peptide, a nuclear-targeting peptide, an endosome-disrupting peptide, a chimeric peptide, a cyclic peptide, a bicyclic peptide, or an oligoarginine peptide.

[0251] In one embodiment, the structure of formula (IV) is formula (IVa):

[0252] [ka]

[0253] In one embodiment of the trimeric peptide-oligonucleotide conjugate of the present disclosure, G is H.

[0254] In another embodiment of the trimeric peptide-oligonucleotide conjugate of the present disclosure, G is —C(O)CH 3 .

[0255] In some embodiments, the trimeric peptide-oligonucleotide conjugates described herein are not solvated. In other embodiments, one or more of the trimeric peptide-oligonucleotide conjugates are in a solvated form. As known in the art, solvates can be any pharmaceutically acceptable solvent, such as water, ethanol, etc.

[0256] Although the peptide-oligonucleotide conjugates of Formulas I, II, Ia, Ib, IV, and IVa are shown in their neutral form, in some embodiments, these peptide-oligonucleotide conjugates are used in the form of pharmaceutically acceptable salts.

[0257] Oligonucleotides Important properties of morpholino-based subunits include: 1) the ability to associate in oligomeric form through stable, uncharged or positively charged backbone linkages; 2) the ability to support nucleotide bases (e.g., adenine, cytosine, guanine, thymidine, uracil, 5-methylcytosine, and hypoxanthine) so that the formed polymer can hybridize with target nucleic acids of complementary bases, including target RNA, with a T of about 45°C or higher for relatively short oligonucleotides (e.g., 10-15 bases). M 3) the ability of the oligonucleotide to be actively or passively transported into mammalian cells; 4) the ability of the oligonucleotide and oligonucleotide:RNA heteroduplex to resist RNAse and RNase H degradation, respectively.

[0258] The stability of the duplex formed between the oligomer and the target sequence is determined by the binding T M and the susceptibility of the duplex to enzymatic cleavage in the cell. M can be measured by conventional methods, such as those described in Hames et al., Nucleic Acid Hybridization, IRL Press, 1985, pp. 107-108, or Miyada C. G. and Wallace RB, 1987, Oligomer Hybridization Techniques, Methods Enzymol., vol. 154, pp. 94-107. In certain embodiments, the antisense oligomer has a binding T with respect to the complementary sequence RNA that is above body temperature, and in some embodiments, above about 45°C or 50°C. M T in the range of 60 to 80°C or more MAccording to well-known principles, the T M can be increased for complementary-based RNA hybrids by increasing the ratio of C:G paired bases in the duplex, or by increasing the length (in base pairs) of the heteroduplex, or both. At the same time, it may be advantageous to limit the size of the oligomer to optimize cellular uptake. For this reason, compounds of the present disclosure have high T M This includes compounds that exhibit a temperature of 45 to 50°C or higher.

[0259] The length of an oligonucleotide can vary so long as it is capable of selectively binding to the intended location within a pre-mRNA molecule. The length of such a sequence can be determined according to the selection procedures described herein. Generally, an oligonucleotide will be from about 8 nucleotides in length to about 50 nucleotides in length. For example, the length (z) of an oligonucleotide can be 8-38, 8-25, 15-25, 17-21, or about 18 nucleotides. However, it will be understood that any length within this range of nucleotides can be used in the methods described herein.

[0260] In some embodiments, antisense oligonucleotides contain base modifications or substitutions. For example, specific nucleobases can be selected to enhance the binding affinity of the antisense oligonucleotides described herein. These include 5-substituted pyrimidines, 6-azapyrimidines, and N-2, N-6, and O-6 substituted purines, including 2-aminopropyladenine, 5-propynyluracil, 5-propynylcytosine, and 2,6-diaminopurine. 5-methylcytosine substitutions have been shown to increase nucleic acid duplex stability by 0.6-1.2°C and can be incorporated into the antisense oligonucleotides described herein. In one embodiment, at least one pyrimidine base of the oligonucleotide comprises a 5-substituted pyrimidine base, the pyrimidine base being selected from the group consisting of cytosine, thymine, and uracil. In one embodiment, the 5-substituted pyrimidine base is 5-methylcytosine. In another embodiment, at least one purine base of the oligonucleotide comprises an N-2, N-6 substituted purine base. In one embodiment, the N-2,N-6 substituted purine base is 2,6-diaminopurine.

[0261] Morpholino-based oligomers (including antisense oligomers) are described in detail in, for example, U.S. Patent Nos. 5,698,685; and 5,217,866; 5,142,047; 5,034,506; 5,166,315; 5,185,444; 5,521,063; 5,506,337 and pending U.S. patent applications Ser. Nos. 12 / 271,036; 12 / 271,040; and PCT International Publication Nos. WO / 2009 / 064471 and WO / 2012 / 043730, and in Summerton et al., 1997, Antisense and Nucleic Acid Drug Development, 7, 187-195, which are incorporated herein by reference in their entireties.

[0262] In embodiments of Formulas I, II, Ia, Ib, IV, and IVa, R 2are, independently for each occurrence, adenine, 2,6-diaminopurine, guanine, hypoxanthine, cytosine, 5-methyl-cytosine, thymine, uracil, and hypoxanthine; Each R 1 is -N(CH3)2.

[0263] Various embodiments of the nucleotide moieties described herein are provided in Table 1.

[0264] [Table 6]

[0265] In a particular embodiment, the sequence listing for the oligonucleotide is GCTATTACCTTAACCCAG (SEQ ID NO: 56).

[0266] In one embodiment, provided herein is a compound having the following structure:

[0267] [ka]

[0268] where z is 18 and R 2 is a sequence of nucleobases having the sequence GCTATTACCTTAACCCAG (SEQ ID NO: 56). This compound is also referred to herein as "PMO IVS2-654."

[0269] In some embodiments, the oligonucleotides described herein are not solvated. In other embodiments, one or more of the oligonucleotides are in solvated form. As known in the art, solvates can be any pharmaceutically acceptable solvent such as water, ethanol, etc.

[0270] Another aspect of the invention relates to fluorescent dye, spin label, heavy metal or radiolabeled compounds of the invention that may be useful not only for imaging, but also for both in vitro and in vivo assays for localizing and quantifying targets in tissue samples, including humans, and for identifying target regions by inhibitory binding of the labeled compounds.

[0271] The present invention further includes isotopically labeled peptides of the conjugates of the invention. An "isotopic" or "radiolabeled" conjugate is a conjugate of the invention in which one or more atoms have been replaced or substituted by an atom having an atomic mass or mass number different from that typically found in nature (i.e., naturally occurring). Suitable radionuclides that can be incorporated into the compounds of the invention include, but are not limited to, 2H (also written as D for deuterium), 3H (also written as T for tritium), 11C, 13C, 14C, 13N, 15N, 15O, 17O, 18O, 18F, 35S, 36Cl, 82Br, 75Br, 76Br, 77Br, 123I, 124I, 125I, and 131I. The radionuclide incorporated into the present radiolabeled compounds will depend on the specific use of the radiolabeled compound. For example, in in vitro IDO enzyme labeling and competition assays, compounds incorporating 3H, 14C, 82Br, 125I, 131I, or 35S will generally be most useful. For radioimaging applications, 11C, 18F, 125I, 123I, 124I, 131I, 75Br, 76Br, or 77Br will generally be most useful.

[0272] A "radiolabel" or "labeled compound" is understood to be a compound that incorporates at least one radionuclide. In some embodiments, the radionuclide is selected from the group consisting of H, C, I, S, and Br.

[0273] Synthetic methods for incorporating radioisotopes into organic compounds are applicable to the compounds of the present invention and are well known in the art.

[0274] The radiolabeled compounds of the invention can be used in screening assays to identify / evaluate compounds such that the ability of a test compound to compete with the radiolabeled compound for binding directly correlates to its binding affinity.

[0275] Although the oligonucleotides of Formula I, II, Ia, Ib, IV, and IVa are shown in their neutral form, in some embodiments, these oligonucleotides are used in a pharmaceutically acceptable salt form.

[0276] Machine learning methods In one aspect, provided herein are systems and methods for identifying one or more cell-penetrating peptides with optimal activity using machine learning, the methods including: a.) synthesizing a library of training oligonucleotide-cell penetrating peptide conjugates; b.) generating seed peptide sequences by training a nested long short-term memory (LSTM) recurrent neural network model using the synthesized library; c.) predicting, from the generated seed peptide sequences, which peptide sequences have a predetermined structure-activity relationship of amino acid residues; and d.) Identifying the best of one or more of the predicted peptide sequences using an activity predictor-genetic algorithm optimizer loop.

[0277] A functional system embodying this method is shown in FIG. 26 and includes a library synthesizer module 2602, a generator network module 2604, a predictor network module 2606, and an optimization tool module 2608, each performing a respective function described herein.

[0278] The output gate of an LSTM encodes the intuition that a memory that is not relevant to the current process may be worth remembering. Nested LSTMs use this intuition to create a temporal hierarchy of memories. Access to inner memories is gated in exactly the same way, allowing selective access to long-term information that is only relevant in certain situations. In alternative embodiments, the generation process may be performed by alternative recurrent neural network (RNN) structures with other feedback connections to make predictions based on time-series data, such as stacked LSTMs and gated recurrent unit (GRU) architectures.

[0279] In one embodiment, the prediction comprises comparing the seed sequence to a chemical fingerprint of amino acid residues.

[0280] In a further embodiment, the predicting step comprises expressing the activity of the topological fingerprint as Conv1D, Conv2D, Conv2D macrocycle, and DeConv2D convolutions.

[0281] In another embodiment, the activity is the mean fluorescence intensity.

[0282] In one embodiment, the Conv1D convolution is trained on a row matrix of amino acid fingerprints in a one-dimensional representation of the peptide sequence.

[0283] In a further embodiment, the Conv2D convolution is trained on an OR operation between individual fingerprints in a two-dimensional representation of the peptide sequence.

[0284] In another embodiment, the Conv2D macrocycle convolution is trained with explicit linker fingerprints in the off-diagonal index in a two-dimensional representation of the peptide sequence.

[0285] In a further embodiment, the DeConv2D convolution is trained with weights of off-diagonal interactions determined by the correlation of each off-diagonal index in the two-dimensional variational representation.

[0286] In another embodiment, the predicting step comprises training a seed peptide sequence against the mean fluorescence intensity using a convolutional neural network model.

[0287] In yet another embodiment, the identifying step comprises an objective function of an activity predictor-genetic algorithm optimizer loop that maximizes the mean fluorescence intensity predicted by the convolutional neural network model.

[0288] In one embodiment, the identifying step comprises an objective function of an active predictor-genetic algorithm optimizer loop that minimizes sequence length and arginine content.

[0289] In certain embodiments, the minimized arginine content is a single arginine residue.

[0290] In another particular embodiment, the minimized sequence length of the peptide is 20 residues or less.

[0291] In another embodiment, the genetic algorithm involves mutation of a single residue with insertion or deletion and replacement, or mutation of multiple residues with insertion and / or deletion and replacement.

[0292] In one embodiment, the genetic algorithm solves the following objective function:

[0293]

number

[0294] (In the formula, Intensity = mean fluorescence intensity R count = number of arginine residues Length = length of the array Net Charge = Net charge of the target sequence) Implement.

[0295] In one embodiment, the library of training oligonucleotide-cell penetrating peptide conjugates consists of:

[0296] (a) a compound of formula (III)

[0297] [ka]

[0298] to formula (IV)

[0299] [ka]

[0300] to form a compound of formula (V)

[0301] [ka]

[0302] The process of forming (b) a compound of formula (VI)

[0303] [ka]

[0304] with a compound of formula (VII)

[0305] [ka]

[0306] in the presence of a copper catalyst to form a compound of formula (VIII)

[0307] [ka]

[0308] The process of forming (c) a compound of formula (V)

[0309] [ka]

[0310] with a compound of formula (VIII)

[0311] [ka]

[0312] in the presence of a coupling reagent to form a compound of formula (II)

[0313] [ka]

[0314] A process of forming

[0315] In one embodiment, peptide 1 (P 1 ), peptide 2 (P 2 ), and peptide 3 (P 3 ) are each independently a cell-penetrating peptide.

[0316] In another embodiment, P 1 , P 2 , and P 3 are cell-penetrating peptides, and the cell-penetrating peptides are independently an amphipathic peptide, a nuclear-targeting peptide, an endosome-disrupting peptide, a chimeric peptide, a cyclic peptide, a bicyclic peptide, a cysteine-linked macrocyclic peptide, a peptide containing at least one unnatural amino acid residue, or an oligoarginine peptide.

[0317] In one embodiment, the acid in step (a) is trifluoroacetic acid.

[0318] In another embodiment, the copper catalyst in step (b) is copper(I) bromide.

[0319] In yet another embodiment, the coupling reagent in step (c) is tris(2-carboxyethyl)phosphine hydrochloride (TCEP).

[0320] In a further embodiment, the solvent in step (a) is water, the solvent in step (b) is water / DMSO, and the solvent in step (c) is water / DMSO.

[0321] In another embodiment, the product of steps (a) and (b) is inert to the reaction conditions of step (c).

[0322] In another embodiment, the products of steps (a) and (b) may be used in step (c) without purification.

[0323] In a further embodiment, the final product is useful for immediate in vitro testing.

[0324] 25 illustrates an example of a generalized computing device 2500 that may be used to implement the machine learning methodologies described herein. The generalized computing device 2500 is intended to represent various forms of digital computers, such as laptops, desktops, workstations, servers, mainframes, and other suitable computers. The components shown here, their connections and relationships, and their functions are merely examples and are not intended to be limiting.

[0325] Included within computing device 2500 are processor 2502, memory 2504, storage device 2506, a high-speed interface 2508 connecting memory 2504 and multiple high-speed expansion ports 2510, and a low-speed interface 2512 connecting low-speed expansion port 2514 and storage device 2506, interconnected using various buses. Processor 2502 can process instructions for execution within computing device 2500, including instructions stored in memory 2504 or storage device 2506, to display graphical information for a graphical user interface (GUI) on an external input / output device, such as a display (not shown), coupled to high-speed interface 2508. In other implementations, multiple processors and / or multiple buses may be used, along with multiple memories and types of memory, as needed. Additionally, multiple computing devices may be connected, each providing a portion of the required operations (e.g., as a server bank or multiprocessor system).

[0326] Memory 2504 may be one or more volatile memory units or may be comprised of one or more non-volatile memory units. Storage device 2506 may be capable of providing mass storage for computing device 2500. For example, storage device 2506 may be or include a computer-readable medium such as a hard disk drive, an optical disk drive, a flash memory or other similar solid-state memory device, or an array of devices, including devices in a storage area network or other configuration. Instructions stored in storage device 2506, when executed by one or more processing units, such as processor 2502, perform one or more methods as described herein. Instructions may also be stored by memory 2504, storage device 2506, or a memory coupled to processor 2502.

[0327] The high-speed interface 2508 manages bandwidth-intensive operations of the computing device 2500, while the low-speed interface 2512 manages lower-bandwidth intensive operations. The high-speed interface 2508 may be coupled to memory 2504, a display (not shown), and a high-speed expansion port 2510 that can accept various expansion cards (not shown). The low-speed interface 2512 may be coupled to a storage device 2506 and a low-speed expansion port 2514, the latter of which may include various communication ports such as USB, Bluetooth, and / or Ethernet, which may be coupled to one or more input / output devices.

[0328] Computing device 2500 may be implemented in many different forms, such as a standard server or a collection of such servers. It may also be implemented in a personal computer, such as a laptop computer, or as part of a rack server system. Alternatively, components of computing device 2500 may be combined with other components in a mobile device (not shown), such as a mobile computing device.

[0329] method Provided herein is a method of treating a neuromuscular disorder, a muscular disorder, a viral infection, or a bacterial infection in a subject in need thereof, comprising administering to the subject a peptide-oligonucleotide conjugate of Formula I, II, Ia, Ib, IV, or IVa.

[0330] Thus, in one aspect, provided herein is a method of treating a muscular disease, a viral infection, a neuromuscular disease, or a bacterial infection in a subject in need thereof, comprising administering to the subject a chimeric peptide-oligonucleotide conjugate of the present disclosure.

[0331] In one embodiment, the neuromuscular disease is Duchenne muscular dystrophy.

[0332] In another embodiment, the viral infection is caused by a virus selected from the group consisting of Marburg virus, Ebola virus, influenza virus, and dengue virus.

[0333] In another embodiment, the bacterial infection is caused by Mycobacterium tuberculosis.

[0334] The subjects contemplated herein are typically humans. However, the subject may be any mammal for which treatment is desired. Thus, the methods described herein may be applied to both human and veterinary applications.

[0335] Administration / Dosage The formulation of therapeutic compositions and their subsequent administration (dosing) are within the skill of those in the art. Dosing depends on the severity and responsiveness of the disease state being treated, with the course of treatment lasting from several days to several months, or until a sufficient diminution of the disease state is achieved. Optimal dosing schedules can be calculated from measurements of drug accumulation in the patient's body.

[0336] Those skilled in the art can readily determine optimal dosages, administration methods, and repetition rates. Optimal dosages may vary depending on the relative potency of individual oligomers, but generally are those EC50 / EC60 / EC70 / EC80 / EC9 ... 50 The dosage can be estimated based on the following: Generally, dosage is 0.01 μg to 100 g per kg of body weight and is administered daily, weekly, monthly, or once or more per year, and can be administered once every two to 20 years. One of skill in the art can readily estimate the repetition rate of dosing based on the measured residence time and drug concentration in body fluids or tissues. After successful treatment, it may be desirable to subject the patient to maintenance therapy to prevent recurrence of the disease state, in which case the oligomer is administered in a maintenance dose ranging from 0.01 μg to 100 g per kg of body weight once or more per day up to once every 20 years.

[0337] In some embodiments, the conjugate of Formula I, II, Ia, Ib, IV, or IVa is administered alone.

[0338] In some embodiments, the conjugate of Formula I, II, Ia, Ib, IV, or IVa is administered in a therapeutically effective amount or dosage. A "therapeutically effective amount" is the amount of the conjugate of Formula I, II, Ia, Ib, IV, or IVa that, when administered to a patient, effectively treats a muscular disease, a viral infection, or a bacterial infection. Although an amount that proves to be a "therapeutically effective amount" for a particular subject in a given example may not be effective for 100% of subjects treated for the disease or condition under consideration, such an amount is still considered a "therapeutically effective amount" by those skilled in the art. The amount of oligonucleotide corresponding to a therapeutically effective amount strongly depends on the type of disease, the stage of the disease, the age of the patient being treated, and other factors.

[0339] In different embodiments, depending on the conjugate of Formula I, II, Ia, Ib, IV, or IVa and the effective amount used, the oligonucleotide can modulate the expression of genes involved in muscle diseases, viral infections, or bacterial infections.

[0340] The amount of the conjugate of Formula I, II, Ia, Ib, IV, or IVa should provide effective treatment of the muscular disease, viral infection, or bacterial infection, but preferably is not excessively toxic to the patient (i.e., the amount is preferably within toxicity limits established by medical guidelines). In some embodiments, a total dose limit is provided to prevent excessive toxicity of the muscular disease, viral infection, or bacterial infection, or to provide more effective treatment, or both. Typically, the amounts considered herein are per day; however, half-day and two- or three-day cycles are also considered herein.

[0341] Different dosing regimens may be used to treat muscle disorders, viral infections, or bacterial infections. In some embodiments, a daily dose, such as any of the exemplary doses described above, is administered once, twice, three times, or four times daily for 3, 4, 5, 6, 7, 8, 9, or 10 days. Depending on the stage and severity of the disease being treated, higher doses may be used for shorter treatment periods (e.g., up to 5 days), or lower doses may be used for longer treatment periods (e.g., 10 days or more, or several weeks, or a month or more). In some embodiments, one or two doses per day are administered every other day.

[0342] The conjugates of Formula I, II, Ia, Ib, IV, or IVa, or their pharmaceutically acceptable salts or solvates, can be administered in pure form or in a suitable pharmaceutical composition via any of the accepted modes of administration or agents known in the art. Oligonucleotides can be administered, for example, orally, intranasally, parenterally (intravenously, intramuscularly, or subcutaneously), topically, transdermally, intravaginally, intravesically, intracisternally, or rectally. Dosage forms can be, for example, solid, semisolid, lyophilized powder, or liquid dosage forms, such as tablets, pills, soft elastic or hard gelatin capsules, powders, solutions, suspensions, suppositories, aerosols, etc., e.g., unit dosage forms suitable for convenient administration of precise dosages. In one embodiment, the oligomer is a phosphorodiamidate morpholino oligomer in a pharmaceutically acceptable carrier and is delivered orally. In another embodiment, the oligomer is a phosphorodiamidate morpholino oligomer conjugated to a peptide, in a pharmaceutically acceptable carrier, and delivered orally.

[0343] In another embodiment, the oligomer is a phosphorodiamidate morpholino oligomer and is delivered intravenously (iv) in a pharmaceutically acceptable carrier. In another embodiment, the oligomer is a phosphorodiamidate morpholino oligomer conjugated to a peptide and is delivered intravenously in a pharmaceutically acceptable carrier.

[0344] Additional routes of administration, such as subcutaneous, intraperitoneal, and pulmonary, are also contemplated by the present disclosure.

[0345] Auxiliaries and adjuvants may include, for example, preservatives, wetting agents, suspending agents, sweeteners, flavorings, perfuming agents, emulsifying agents, and dispensing agents. Prevention of microbial action is generally provided by various antibacterial and antifungal agents, such as parabens, chlorobutanol, phenol, and sorbic acid. Isotonic agents, such as sugars and sodium chloride, may also be included. Prolonged absorption of injectable pharmaceutical forms can be achieved by the use of agents delaying absorption, such as aluminum monostearate and gelatin. Auxiliaries may also include wetting agents, emulsifying agents, pH buffering agents, and antioxidants, such as citric acid, sorbitan monolaurate, triethanolamine oleate, and butylated hydroxytoluene.

[0346] Solid dosage forms can be prepared with coatings and shells, such as enteric coatings and others known in the art.They can contain a substance that satisfies hunger, and can be a composition that releases active oligonucleotide in a delayed manner in a specific part of the intestinal tract.Examples of embedding compositions that can be used are polymeric substances and waxes.Active oligonucleotide can also be in microencapsulated form, if appropriate, with one or more of the above-mentioned excipients.

[0347] Liquid dosage forms for oral administration include pharmaceutically acceptable emulsions, solutions, suspensions, syrups, and elixirs. Such dosage forms are prepared by dissolving, dispersing, or the like, a conjugate or a pharmaceutically acceptable salt thereof described herein and an optional pharmaceutical adjuvant in a carrier such as water, saline, aqueous dextrose, glycerol, ethanol, or the like; a solubilizing agent and emulsifier such as ethyl alcohol, isopropyl alcohol, ethyl carbonate, ethyl acetate, benzyl alcohol, benzyl benzoate, propylene glycol, 1,3-butylene glycol, dimethylformamide; an oil, particularly cottonseed oil, peanut oil, corn germ oil, olive oil, castor oil, and sesame oil, glycerol, tetrahydrofurfuryl alcohol, polyethylene glycol, and fatty acid esters of sorbitan; or a mixture of these substances to form a solution or suspension.

[0348] Generally, depending on the intended mode of administration, a pharmaceutically acceptable composition will contain from about 1% to about 99% by weight of an oligonucleotide described herein, or a pharmaceutically acceptable salt thereof, and from 99% to 1% by weight of a pharmaceutically acceptable excipient. In one example, the composition will be from about 5% to about 75% by weight of an oligonucleotide described herein, or a pharmaceutically acceptable salt thereof, with the remainder being suitable pharmaceutical excipients.

[0349] Actual methods for preparing such dosage forms are known, or will be apparent, to those skilled in the art. See, for example, Remington's Pharmaceutical Sciences, 18th Edition (Mack Publishing Company, Easton, PA, 1990).

[0350] kit In other embodiments, a kit is provided. The kit according to the present disclosure includes a package containing an oligonucleotide, peptide, peptide-oligonucleotide conjugate, or composition of the present disclosure. In some embodiments, the kit includes a peptide-oligonucleotide conjugate according to Formula I, II, Ia, Ib, IV, or IVa, or a pharmaceutically acceptable salt thereof.

[0351] The term "package" refers to any container containing the oligonucleotides or compositions presented herein. In some embodiments, the package can be a box or a package. Packaging materials used to package pharmaceuticals are well known to those skilled in the art. Examples of pharmaceutical packaging materials include, but are not limited to, bottles, tubes, inhalers, pumps, bags, vials, containers, syringes, bottles, and any packaging material suitable for the selected formulation and intended mode of administration and treatment.

[0352] The kit may also include items that are not contained within the package but are attached to the outside of the package, for example, pipettes.

[0353] The kit may further include instructions for administering the oligonucleotide or composition of the present disclosure to a patient. The kit may also include instructions regarding the approved uses of the oligonucleotides described herein by regulatory authorities, such as the U.S. Food and Drug Administration. The kit may also include labeling or inserts for the oligonucleotides. The packaging, product inserts, or both may themselves be approved by regulatory authorities. The kit may include the oligonucleotides in a solid phase or in a liquid phase (such as a provided buffer) in the package. The kit may also include buffers for preparing solutions for carrying out the method, and pipettes for transferring liquids from one container to another. [Example]

[0354] Examples are provided below for illustrative purposes to illustrate certain specific embodiments of the present disclosure. However, the scope of the claims is in no way limited by the examples described herein. Various changes and modifications to the disclosed embodiments will be apparent to those skilled in the art, and such changes and modifications, including but not limited to those related to the chemical structures, substituents, derivatives, formulations, or methods of the present disclosure, can be made without departing from the spirit of the present disclosure and the scope of the appended claims. The definitions of variables in the structures in the diagrams herein are equivalent to the definitions of the corresponding positions in the formulas shown herein.

[0355] Library synthesis For Figure 26, element 2602 and Figure 27A, the focus is on a construct containing four modules: one for an oligonucleotide and three for distinct peptide sequences. It was envisioned that the modules would need to contain various functional peptides, such as nuclear targeting peptides or endosome-disrupting peptides. To synthesize the construct, a convergent approach was chosen in which module 1 was linked to module 2 and separately module 3 was linked to module 4. The two dimers could then be conjugated to provide a four-module construct (Figure lb and Figure 27A, step 2700).

[0356] The choice of bioconjugation reaction was important because each reaction had to be optimized to tolerate specific functional groups, be compatible with common solvents and conditions, and be suitable for peptide substrates. Several reactions were investigated in the context of peptide conjugation, and various limitations of specific reactions were encountered.

[0357] [Table 7]

[0358] For example, tetrazines can be incorporated into peptides on the resin but have been found to be reduced during peptide cleavage and side-chain deprotection. Similarly, the tertiary amides present in commercially available DBCO reagents are cleaved in trifluoroacetic acid, necessitating the incorporation of DBCO into the substrate off the resin. Furthermore, maleimides and azides will react when present on the same peptide.

[0359] After investigating several potential reactions, the final synthetic scheme involved two azide-alkyne cycloadditions to one S N This is coupled with an Ar reaction (Figure 1b). In reaction 1, PMO-DBCO is conjugated with an azido peptide to link modules 1 and 2. The azido peptide will also contain a free thiol that will not react with DBCO under neutral conditions. Separately, in reaction 2, modules 3 and 4 are linked via copper-catalyzed azide-alkyne cycloaddition. Module 3 will contain an N-terminal cysteine ​​residue and a C-terminal azido lysine linked to a decafluorobiphenyl. The perfluoroarene also serves to enable reaction 3 and prevent free thiols from interfering with the azide / alkyne cycloaddition. Module 4 contains only alkynes, which are stable to most reactions, such as peptide macrocyclization. Finally, in reaction 3, modules 1-2 and 3-4 are combined with thiol-perfluoroarene S. N Conjugation can be achieved by the Ar reaction. Because the azide has already reacted with the alkyne, TCEP can be used to prevent disulfide formation without worrying about unintended reduction of the azide.

[0360] The selected synthetic scheme offers numerous advantages for the synthesis of combinatorial libraries. First, all reactions have been previously used in biological assays to generate stable, irreversible linkages. Second, the reactions do not produce by-products and are theoretically quantitative, reducing the need for purification. Third, all reagents are relatively benign and should not adversely affect cell culture experiments. While copper will be present, low micromolar concentrations of copper have been shown not to affect cell viability for the purposes of this screening. Finally, all reactions can be performed on a very small scale (e.g., volumes less than 5 μL). In particular, the combination of high yield and small volume suggests that reactions can be run at high concentrations and immediately diluted into medium for cell culture treatment without the need to purify each reaction individually.

[0361] After optimizing individual reaction conditions, a set of 36 proof-of-concept constructs was synthesized for the modular library. For module 1, the oligonucleotide PMOIVS2-654 (SEQ ID NO: 56) was used, which induces eGFP fluorescence upon successful delivery to the nucleus of engineered HeLa cell lines. Module 2 contained a set of four different CPPs: penetratin, pVEC, TP10, and DPV6. Module 3 contained the KRVK and SV40 nuclear localization sequences (NLSs) and the peptide PHP.eB, a sequence recently reported to improve viral delivery to the brain. Module 4 contained three CPPs: Bpep, DPV6, and PPC3 (Figure 5).

[0362] The synthesis began with the conjugation of modules 1 and 2 using an azide-strained alkyne cycloaddition reaction. 5 mM PMO-DBCO was incubated with 5 mM azide-module 2 peptide-cys in water. After 1 h, the reactions were flash-frozen in liquid nitrogen, and the solvent was removed by lyophilization. For each reaction, LC-MS analysis showed nearly complete conversion to product, indicating that the reactions proceeded cleanly without the need for purification.

[0363] Modules 3 and 4 were conjugated using copper-catalyzed azide-alkyne cycloaddition. The decafluorobiphenyl-module 3 peptide-azide and alkyne-module 4 peptide were dissolved in water to create a 10 mM stock solution of each module. Separately, copper(I) bromide was dissolved in DMSO under an inert atmosphere. The peptides were combined (final concentration 3.3 mM each), and the reaction was initiated by the addition of copper bromide solution (final concentration 6.7 mM). After 2 hours, the reaction was quenched by the addition of 100 mM disodium phosphate in water. The solvent was removed under vacuum in preparation for reaction 3.

[0364] Finally, modules 1-2 and 3-4 were combined. Modules 1-2 (final concentration 0.63 mM) were mixed with modules 3-4 (final concentration 1.25 mM, 2 equivalents) in DMSO containing 5 mM TCEP. Module 1 was used as the limiting reagent because it was the active component in the cell assay. After 2 hours, the reaction was flash frozen and stored at -80°C until dilution and cell processing. Individual testing of the reaction components suggested that the presence of copper interfered with the reaction, and despite considerable optimization efforts, reaction conversion never exceeded approximately 70%.

[0365] The 36 synthesized constructs were used to test their ability to modulate PMO activity using a modified HeLa cell assay. HeLa-654 cells were stably transfected to express non-fluorescent eGFP protein. The eGFP gene is interrupted by a mutated intron from the human β-globin gene (IVS2-654). This insertion alters pre-mRNA splicing, causing retention of the fragment in the mature mRNA, resulting in the production of a non-fluorescent protein. The PMO IVS2-654 base-pairs with the β-globin insert, altering mRNA splicing and thereby leading to the expression of fluorescent eGFP.

[0366] For treatment, the crude reaction mixture was diluted to 5 μM with medium. The concentration of the modular constructs was calculated based on the original concentration of the module 1-2 conjugates mixed in the reaction. Cells were treated with each construct for 22 hours using medium containing 10% fetal bovine serum (FBS), after which cell fluorescence was measured by flow cytometry.

[0367] Different modules induced several notable trends in cellular fluorescence levels (Figure 6). When module 2 was DPV6, the entire construct consistently exhibited high fluorescence, regardless of the peptides placed in modules 3 and 4. However, minor cellular fluorescence was observed when either pVEC or TP10 was placed in module 2. Gated cell counts were used during flow cytometry as an indirect readout to control for compound toxicity. Highly toxic compounds resulted in a reduction in overall cell number, and nonviable cells were gated out based on propidium iodide staining. In this experiment, consistently low cell numbers were observed when the module 3 peptide was the nuclear localization sequence KRVK. Preferred compounds are highly active and nontoxic, so the cellular fluorescence and cell number readouts were multiplied to obtain a measure of overall compound efficacy (FxC).

[0368] Following the success of the proof-of-concept experiments, we synthesized a library of 600 conjugates for testing in HeLa-654 cells. We chose to increase the number of peptides in module 4 from 3 to 50. A mixture of chimeric, cyclic, and bicyclic peptides was included to increase the diversity of peptide types in the library and highlight the possibility of incorporating modified peptides and unique functional groups. Cyclic peptides included R12, Bpep, and a scalloped-edge variant in which two cysteine ​​residues were linked to form a stable peptide macrocycle compatible with modular reactions. Bicyclic variants included a double macrocyclic R12 and an alternative R12 sequence in which three side chains were linked by 1,3,5-trisbromomethylbenzene. Other peptides included several previously reported CPPs, peptides computationally predicted to be effective PMO carriers (PPCs), and peptides with appended NLS sequences (see Table 2).

[0369] Reactions 1 and 2 were carried out as described above, except that additional compounds were used, and reaction 2 now contained 150 different products. For reaction 3, synthesis was carried out over two days in a 384-well plate using the conditions described above to accommodate a larger number of compounds. After synthesis, compounds were diluted to 100 μM in PBS and then to 5 μM in medium containing 10% FBS. HeLa-654 cells were again treated with the constructs for 22 hours, and cell fluorescence was analyzed by flow cytometry (Figure 7).

[0370] Machine learning models As defined with respect to step 2700 in Figure 27A, a series of interpretable machine learning models were trained to predict novel, more effective sequences using sequence and activity information from a modular library. The models can be implemented by a generalized computer system, such as that shown in Figure 25, or on a custom-configured computing platform. A key consideration in machine learning is the appropriate representation of input features and output parameters. Given the lack of predefined quantitative sequence-activity relationships correlating amino acid chemical structure and sequence position with cell permeability, previous heuristic research in this field has met with limited success. Furthermore, limitations of computational approaches often stem from non-standardized datasets and the use of peptide physicochemical descriptors as machine learning features relative to unrelated functional parameters. To overcome these limitations, an inverse design model was developed that uses a topological representation of peptide sequences to extract information from homogeneous datasets, such as those proposed above.

[0371] This inverse design model can also be referred to as a generator-predictor-optimizer machine learning model. The generator network generated realistic peptides, the predictor network addressed sequence-activity relationships using a topological representation of the molecule, and the optimizer tool maximized activity while minimizing length and arginine content. This machine learning model is summarized in functional block format in Figure 26. The combination of bioactivity and other design constraints resulted in optimized synthetic peptides that were nontoxic and nonimmunogenic and significantly improved PMO delivery.

[0372] To extract chemical information from each peptide, rather than placing peptides into bins of various physicochemical properties, we characterized the atomic connectivity of each amino acid sequence. First, as shown in Figure 2a, we represented amino acids and their modular linkers as fingerprints that encoded a topological exploration of neighboring atoms and bonds. Next, we developed a series of 1D and 2D peptide sequence representations that treated the sequence as a linear (1D) and complete graph (2D) representation. The 1D representation captured the linear interactions of covalent bonds along the peptide backbone, while the 2D representation introduced off-diagonal elements to represent the folding of macrocyclic sequences, interactions through space, and covalent bonds. The training dataset, developed as described in the library synthesis section above, consisted of a modular library containing 600 peptides and other sequences previously tested in the eGFP assay. Sequences that resulted in low cell numbers due to toxicity were eliminated. The output from this assay was the mean fluorescence intensity (MFI) linked to each graph representation.

[0373] For the inverse design, a machine learning-based generator-predictor-optimizer loop was developed, as described above. The generator was based on a recurrent neural network using a nested long short-term memory (RNN-Nested LSTM) architecture that captures the grammatical intuition for describing cell-penetrating peptide sequences (Figure 27A, step 2702). This enabled the generation of novel cell-penetrating peptide sequences with similar appearances. For the predictor, a convolutional neural network (CNN) model was used to train the sequence representation on MFI (Figure 27B, step 2704). Finally, optimization was performed using a genetic algorithm (GA), whose objective functions included maximizing the MFI predicted by the CNN model and minimizing the length and arginine content while maintaining water solubility (Figure 27C, step 2706). The design of these algorithms evolved through several iterations, eventually incorporating variational pixel maps and interactions via a weighted space including cysteine ​​bonds (DeConv2D). The optimized predicted peptide sequence is provided (Figure 27C, step 2708).

[0374] The original machine learning model based on the Conv1D architecture was able to predict MFI with an accuracy rate of 89% when the predicted value was within the range of the training values ​​(0.32–19.5). After hyperparameter optimization and model development, the accuracy rate improved to 92%.

[0375] To predict new sequences, we generated seed sequences using a trained nested LSTM model and optimized them with an activity predictor genetic algorithm optimizer loop against the objective functions of maximizing MFI and minimizing sequence length and arginine content. Furthermore, to observe the model's ability to predict peptides with specific activity, we intentionally predicted low-activity sequences for negative control validation, resulting in a class of synthetic peptides called "Mach11."

[0376] The predicted sequences were determined to share no significant sequence similarity with previously reported CPPs or naturally occurring peptides and proteins, as determined by a homology search using the protein-protein basic local alignment search tool (BLASTp). Finally, an online tool (IEDB) was used to predict the probability that the predicted sequences were immunogenic T cell epitopes, and the predicted immunogenicity was found to be low according to a set probability score.

[0377] To interpret the design principles discovered by the algorithm, we investigated the chemical features that activate the learning of the Conv1D predictor. To accomplish this, we examined the positive gradient activation of the input features (sequences) relative to the output (MFI) of the first convolutional layer. More activated regions within the layer point to the specific features that guided the neural network to make a particular prediction. We observed that the predictor was primarily activated by amino acids toward the C-terminus of the peptide, and a preference for cationic residues was also noted. Detailed analysis of a highly active predicted sequence, Mach3, revealed a preference for guanidinium substructures within arginines. These motifs are consistent with previous empirical findings regarding cell penetration.

[0378] To better understand how the model generated predictions, we selected five random sequences of different lengths, fed them into the optimizer, and visualized the best predictions. Positive activations were averaged by residue position (Figure 3c) and chemical fingerprint (Figure 3d). Furthermore, amino acid types were analyzed based on residue position in the sequence (Figure 3e) and on substructures highlighted in the inter-residue fingerprint (Figure 3f). Similar trends to Mach3 were observed for the better-performing training sequences.

[0379] Mach peptides enhance delivery of PMOs From a list of hundreds of predicted peptide sequences, 20 candidates varying in length, charge, and predicted activity were selected, synthesized, and tested. Similar to the PMO-peptide library, the PMO-Mach constructs were first tested for PMO delivery in a HeLa 654 assay at 5 μM in complete medium for 22 h and analyzed by flow cytometry (Figure 14). Experiments were performed using three technical triplicates and two or three biological replicates. Comparison of the resulting activities with the training dataset revealed that nearly all sequences outperformed the best-performing library peptide (Figure 2d).

[0380] We purposefully selected predicted peptides with diverse lengths, charges, and structures. Mach1 through Mach11 were linear PMO-peptide constructs. Mach12 and Mach13 contained two cysteines linked by a decafluorobiphenyl group, forming an internal macrocycle. Sequence lengths ranged from 33 to 80 amino acids, with net charges ranging from +11 to +22. Furthermore, to confirm that the algorithm grasped the design principles, we asked it to predict sequences with reduced performance. To this end, we rearranged the sequence of the high-performing peptide, Mach1, to decrease its predicted activity, resulting in Mach7. However, the experimental activity of the two constructs was nearly identical. Next, the algorithm designed unique peptides predicted to have lower activity, resulting in Mach11. Indeed, Mach11 did not significantly improve PMO delivery. Notably, Mach5 did not significantly increase PMO activity, but was predicted to have similar activity to Mach2 through Mach4.

[0381] Dose-response experiments were performed using several highly active Mach peptides (Figure 4a). In the same format as above, HeLa 654 cells were treated with various concentrations of Mach2, 3, 4, and 7 for 22 hours and analyzed by flow cytometry. The EC 50 The values ​​were close to 1 μM and did not show cytotoxicity at the concentrations tested as determined by cell counts and PI staining.

[0382] Example 1 General methods for peptide preparation and purification Fast-Flow Peptide Synthesis Peptides were synthesized on a 0.1-mmol scale using an automated flow peptide synthesizer. 200 mg of ChemMatrix Rink Amide HYR resin was loaded into a reactor maintained at 90 °C. All reagents were pumped at 80 mL / min with an HPLC pump through a stainless steel loop maintained at 90 °C before being introduced into the reactor. For each coupling, 10 mL of a solution containing 0.2 M amino acid and 0.17 M HATU in DMF was mixed with 200 μL of diisopropylethylamine and delivered to the reactor. Fmoc removal was achieved using 10.4 mL of 20% (v / v) piperidine. Between each step, the reactor was flushed with 15 mL of DMF. Special coupling conditions were used for arginine: the flow rate was reduced to 40 mL / min, and 10 mL of a DMF solution containing 0.2 M Fmoc-L-Arg(Pbf)-OH and 0.17 M PyAOP was mixed with 200 μL of diisopropylethylamine and delivered to the reactor. To couple unnatural amino acids or cap peptides (e.g., with 4-pentynoic acid), the resin was incubated for 30 min at room temperature with 4-pentynoic acid (1 mMol) dissolved in 2.5 mL of 0.4 M HATU in DMF containing 500 μL of diisopropylethylamine. After completion of the synthesis, the resin was washed three times with DCM and dried under vacuum.

[0383] Peptide cleavage and deprotection Each peptide was treated with 5 mL of 94% trifluoroacetic acid (TFA), 2.5% 1,2-ethanedithiol (EDT), 2.5% water, and 1% triisopropylsilane (TIPS) (v / v) for 7 min at 60 °C to simultaneously deprotect the overall side chains and cleave them from the resin. For arginine-rich sequences, the resin was treated with a cleavage cocktail consisting of 82.5% TFA, 5% phenol, 5% thioanisole, 5% water, and 2.5% EDT (v / v) for 14 h at room temperature. The TFA was evaporated by bubbling N2 through the mixture. The peptide was then precipitated and washed with approximately 40 mL of cold ether (cooled to -80 °C). The crude product was pelleted by centrifugation at 4,000 rpm for 3 min, and the ether was decanted. The ether precipitation and centrifugation were repeated two more times. After the third wash, the pellet was redissolved in 50% water and 50% acetonitrile containing 0.1% TFA, filtered through a fritted syringe to remove the resin, and lyophilized.

[0384] Peptide purification The peptides were redissolved in water and acetonitrile containing 0.1% TFA, filtered through a 0.22 μm nylon filter, and purified by mass-coupled semi-preparative reverse-phase HPLC. Solvent A was water with 0.1% TFA additive, and solvent B was acetonitrile with 0.1% TFA additive. A linear gradient was used, varying at a rate of 0.5% / min. Most peptides were purified on an Agilent Zorbax SBC3 column (9.4 × 250 mm, 5 μm). Highly hydrophilic peptides, such as arginine-rich sequences, were purified on an Agilent Zorbax SBC18 column (9.4 × 250 mm, 5 μm). Using the instrument's mass data for each fraction, only pure fractions were pooled and lyophilized. The purity of the fraction pool was confirmed by LC-MS.

[0385] The peptides in Table 2 were synthesized using the protocol in Example 1.

[0386] Example 2 PMO-DBCO synthesis PMO IVS-654 (50 mg, 8 μmol) was dissolved in 150 μL of DMSO. To this solution was added a solution containing two equivalents of dibenzocyclooctynoic acid (5.3 mg, 16 μmol) activated with HBTU (37.5 μL of 0.4 M HBTU in DMF, 15 μmol) and DIEA (2.8 μL, 16 μmol) in 40 μL of DMF (final reaction volume = 0.23 mL). The reaction was allowed to proceed for 25 min and then quenched with 1 mL of water and 2 mL of ammonium hydroxide. The ammonium hydroxide will hydrolyze the ester formed during the reaction. After 1 h, the solution was diluted to 40 mL and purified using reverse-phase HPLC (Agilent Zorbax SBC3 column: 21.2 × 100 mm, 5 μm) using a linear gradient of 2 to 60% B (solvent A: water; solvent B: acetonitrile) over 58 min (1% B / min). Using the mass data from the instrument for each fraction, only the pure fractions were pooled and lyophilized. The purity of the fraction pool was confirmed by LC-MS.

[0387] Example 3 Library synthesis conditions Reaction 1 PMO-DBCO was dissolved in water at a concentration of 10 mM (determined gravimetrically). Module 2 peptide was dissolved in water containing 0.1% trifluoroacetic acid at a concentration of 10 mM (determined gravimetrically; molecular weight was calculated to contain 0.5 trifluoroacetic acid counterions per lysine, arginine, and histidine residue). In a microcentrifuge tube, 50 μL of PMO-DBCO solution was mixed with 50 μL of module 2 peptide. The solution was mixed, and the reaction was allowed to proceed for 1 h. The product was then analyzed by LC-MS, and the solvent was removed by lyophilization. Finally, the product was resuspended in 100 μL of DMSO to prepare a 5 mM solution and stored at -20 °C.

[0388] Reaction 2 Stock solutions were prepared by dissolving module 3 and module 4 peptides in water to a concentration of 10 mM (determined gravimetrically). For each reaction, 4 μL of module 3 peptide was mixed with 4 μL of module 4 peptide in a PCR tube. Separately, a copper bromide solution was prepared by mixing 1 mL of degassed DMSO with 2.8 mg of copper(I) bromide under N2 to obtain a 20 mM solution. 4 μL of CuBr solution was added to the mixture of module peptides 3 and 4 under ambient conditions. The reaction was capped and allowed to proceed for 2 hours; the small amount of O2 present in the reaction setup does not substantially impede the reaction's progress. After 2 hours, 2 μL of a 100 mM solution of Na2HPO4 was added. The PCR tube was then sonicated, vortexed, and centrifuged. To remove the solvent, the PCR tube was centrifuged under vacuum for 2 hours using a Savant SPD121P Speed-Vac set at 35°C. Finally, the product was resuspended in 16 μL of DMSO to prepare a 5 mM solution and stored at −80° C. The product was analyzed by LC-MS.

[0389] Reaction 3 The final modular construct was synthesized by combining modules 1-2 and 3-4. First, 1.6 μL of reaction 2 was added to a 384-well plate. Separately, 30 μL of reaction 1 was mixed with 15 μL of TCEP solution (100 mM TCEP-HCl in 50 / 50 water / DMSO containing 400 mM NaOH) and 75 μL of DMSO. Next, 1.6 μL of reaction 1 solution was added to reaction 2 in a 384-well plate. Each individual reaction ultimately contained 0.4 μL of reaction 1 (5 mM in DMSO), 1.6 μL of reaction 2 (5 mM in DMSO), 0.2 μL of TCEP solution (100 mM in water / DMSO), and 1 μL of DMSO. Excess reaction 2 was used to force the reaction to completion; the presence of copper hinders the efficiency of this conjugation. Reaction 1 was used as the limiting reagent to avoid excess PMO, the active component in the cell culture assay. The reaction proceeded for 2 hours, after which the plate was stored at -80°C. The reaction was analyzed by LC-MS.

[0390] Example 4 HeLa-654 eGFP assay HeLa 654 cells were maintained at 37°C and 5% CO2 in MEM supplemented with 10% (v / v) fetal bovine serum (FBS) and 1% (v / v) penicillin-streptomycin. Eighteen hours before treatment, cells were seeded at a density of 5,000 cells per well of a 96-well plate in MEM supplemented with 10% FBS and 1% penicillin-streptomycin. On the day of the experiment, a 384-well plate containing the crude reaction mixture in DMSO was diluted to 100 μM by adding 16.8 μL of PBS to 3.2 μL of the reaction mixture. Each construct was then diluted to 5 μM in MEM supplemented with 10% FBS and 1% penicillin-streptomycin. Cells were incubated with each conjugate at a concentration of 5 μM for 22 hours at 37°C and 5% CO2. The treatment medium was then aspirated, and the cells were incubated with 0.25% trypsin-EDTA for 15 minutes at 37°C and 5% CO2, washed once with PBS, and resuspended in PBS containing 2% FBS and 2 μg / mL propidium iodide. Flow cytometry analysis was performed on a BD LSRII flow cytometer. Gates were applied to the data to exclude cells that were highly positive for propidium iodide or that displayed forward / side scatter measurements sufficiently different from the main cell population. Each sample was limited to an upper limit of 5,000 gated events.

[0391] Analysis was performed using Graphpad Prism 7. For each sample, the mean fluorescence intensity and the number of gated cells were measured (Figures 8 and 9), and the intensity multiplied by the number of cells was calculated (Figure 7).

[0392] Furthermore, analysis of the exon skipping activity of PMO-P1 through PMO-P7, measured in eGFP assays performed using three different biological replicates (Figure 18), indicated that PMO-P7 was the most active conjugate, exhibiting a 14-fold increase over unconjugated PMO, comparable to the activity of PMO-Bpep. Among these seven predicted peptide-PMO conjugates, the second and third best performers were PMO-P2 and PMO-P4, with 9- and 7-fold increases in activity, respectively, compared to unconjugated PMO. The remaining conjugates, PMO-P1, PMO-P3, PMO-P5, and PMO-P6, showed 4-fold increases or even lower activity. PMO-P7 exhibited superior activity to analogs PMO-P8 through PMO-P12 (Figure 19). The KXXC motif at the C-terminus of the peptide did not result in increased PMO delivery (Figure 20). PMO-P21 to PMO-P23 (Figure 21) and P30 to P40 (Figure 23) were also tested.

[0393] Example 5 MTT assay Cell viability after treatment was determined using MTT (see Figure 15b). HeLa 654 cells were treated with various concentrations of PMO-peptide constructs for 22 hours at 37°C and 5% CO2. Two wells containing medium alone were used as blanks, two wells containing untreated cells were used as negative controls, and two wells containing SDS-treated cells were used as positive controls. The supernatant was transferred to a new 96-well plate and replaced with complete medium without phenol red. 10 μL of MTT stock solution was added to each well and incubated for 4 hours. 100 μL of SDS-HCl was added to each well, mixed thoroughly, and incubated for 4 hours. Each sample was mixed again, and its absorbance was read at 570 nm. The blank measurement was subtracted from each measurement, and cell viability was calculated as follows: % viability = 100 × experimental (OD570) / no treatment (OD570).

[0394] Example 6 LDH release assay Cytotoxicity assays were performed on both HeLa 654 cells and human RPTECs (human renal proximal tubule epithelial cells, ECH001, Kerafast; see Figures 4a and 15a). RPTECs were maintained in high-glucose DMEM supplemented with 10% (v / v) fetal bovine serum (FBS) and 1% (v / v) penicillin-streptomycin at 37°C and 5% CO2. RPTECs were treated in the same manner as HeLa 654 cells. After treatment, the supernatants were transferred to new 96-well plates.

[0395] CytoTox 96 Reagent (Promega) was added to each well of the 96-well plate containing the supernatant. The plate was protected from light and incubated at room temperature for 30 minutes. An equal volume of stop solution was added to each well and mixed, and the absorbance of each well was measured at 490 nm. Blank measurements were subtracted from each measurement, and % LDH release was calculated as follows: % cytotoxicity = 100 × experimental LDH release (OD490) / maximum LDH release (OD490).

[0396] Example 7 Inflammation Panel Assay The inflammatory response induced by the PMO-peptide conjugates was assayed by profiling the release of inflammatory cytokines after treatment of THP-1-derived macrophages (see Figure 4b and Figure 16). THP-1 cells (ATCC TIB-202) were grown in RPMI 1640 medium supplemented with 10% (v / v) FBS, 1% (v / v) penicillin-streptomycin, L-glutamine, non-essential amino acids, and sodium pyruvate at 37°C and 5% CO2. Two days before the experiment, THP-1 cells (450 kJ / mL) were treated with 25 nM phorbol 12-myristate 13-acetate (PMA) for 24 hours at 37°C and 5% CO2 to induce differentiation into macrophages. The medium was then replaced with fresh RPMI medium, and the cells were incubated for an additional 24 hours. At this time, the phenotype changed from floating cells to strongly adherent cells. On the morning of the experiment, the supernatant was removed and macrophages were lifted by incubating in enzyme-free cell dissociation buffer (Thermo) for 5 minutes. Cells were then collected, spun down, and expanded in complete RPMI medium to a cell density of 500k / mL. 100k cells were seeded into each well of a 96-well plate, leaving the first two rows empty. Cells were allowed to reattach before treatment. Duplicate wells were treated with various concentrations of PMO-peptide conjugates for 2 hours at 37°C and 5% CO2. Medium alone and untreated wells served as negative controls, while 10 μg / mL bacterial lipopolysaccharide (LPS) treatment served as a positive control. After treatment, each well was washed three times, refilled with fresh medium, and incubated for 12 hours. The supernatant was transferred to a V-bottom plate and spun down at 4000 rcf to remove debris. Inflammatory cytokines in the supernatants were assayed using the LEGENDplex Human Inflammation panel (BioLegend), a fluorescent bead-based assay. The cytokines assayed were IL-1 beta, IFN-alpha2, IFN-gamma, IFN-alpha, MCP-1, IL-6, IL-8, IL-10, IL-12p70, IL-17A, IL-18, IL-23, and IL-33. Analysis was performed on a BD LSRII flow cytometer, and data were analyzed using BioLegend's accompanying software.

[0397] Example 8 Recombinant expression His6-SUMO-G5-DTA(C186S) was overexpressed in E. coli BL21(DE3) cells (see Figure 17). Approximately 10 g of cell pellet was lysed by sonication in 50 mL of 20 mM Tris, 150 mM NaCl, pH 7.5 buffer containing 30 mg of lysozyme, 2 mg of DNAase I, and one tablet of Complete Protease Inhibitor Cocktail. The suspension was centrifuged at 16,000 rpm for 30 minutes to remove cell debris. The supernatant was loaded onto a 5 mL HisTrap FF Ni-NTA column (GE Healthcare, UK) and washed with 30 mL of 100 mM imidazole in 20 mM Tris, 150 mM NaCl, pH 8.5. The protein was eluted from the column with a buffer containing 300 mM imidazole in 20 mM Tris, 150 mM NaCl, pH 8.5. Imidazole was removed from the protein by centrifugation through a Millipore centrifugal filter unit (10K). The His6-SUMO tag was then cleaved from the protein using SUMO protease (previously recombinantly expressed) by overnight incubation at 4°C in a protease:protein ratio of 1:1000 in 20 mM Tris, 150 mM NaCl, pH 7.5. The desired protein was separated from the His6-SUMO tag by loading the mixture onto a 5 mL HisTrap FF Ni-NTA column. The final purified protein was separated by size exclusion chromatography using a HiLoad 26 / 600 Superdex 200 prep-grade size-exclusion chromatography column (GE Healthcare, UK) in 20 mM Tris, 150 mM NaCl, pH 7.5 buffer.

[0398] Proteins were analyzed using SDS-PAGE gels. Proteins were further analyzed by ESI-QTOF LCMS to confirm molecular weight and purity. Protein charge-state envelopes were deconvoluted using maximum entropy using an Agilent Mass Hunter Bioconfirm (Agilent Zorbax 300SBC3 column: 150 × 2.1 mM ID, 5 μM, 1% B 0–2 min, 1% to 91% B 2–11 min, 91% to 9% B 11–12 min; flow rate: 0.8 mL / min stepwise increase).

[0399] Example 9 immunogenicity The immunogenicity of the sequences (see Figure 12) was calculated using an online server. The score is an arbitrary numerical value, with more positive values ​​indicating a higher probability that the peptide is immunogenic, and vice versa. For non-natural residues, B (β-alanine) and X (6-aminohexanoic acid) were replaced with a (alanine) and L (leucine), respectively, for the search procedure. It was found that none of the peptides were predicted to elicit an immune response.

[0400] Example 10 Reverse Design Model Generator - Recurrent Neural Network. The generator is a data-driven tool for generating new peptide sequences that conform to the "Cell-Penetrating Peptide Ontology." A recurrent neural network (nested LSTM-based model) was trained to capture the basic rules (see Figures 1a and 10). The training dataset consisted of 1,150 sequences, including the unique (non-modular) sequences used to create the library and sequences from CPPSite2.0. (See also element 2604 in Figure 26 and step 2702 in Figure 27A.)

[0401] Predictor - a convolutional neural network. The predictor estimates the fluorescence intensity from PMO delivery by a given peptide sequence, as measured in the HeLa 654 assay. An initial model (original: Conv1D) was trained on a 1D representation of the peptide sequence using a row matrix of amino acid fingerprints (see Figures 2, 3, and 13). Next, we developed a series of 2D representations to capture long-range interactions: (i) Conv2D - a 2D representation based on OR operations between individual fingerprints; (ii) Conv2D macrocycle - a 2D representation using explicit linker fingerprints in the off-diagonal index; and (iii) DeConv2D - a 2D variational representation using off-diagonal interaction weights determined by a function of each off-diagonal index (see Figure 11). All fingerprints were generated using the RDKit. We compiled 640 PMO peptide sequences for training by combining the CPP library from this study with a collection of CPPs from previous studies. (See also element 2606 in Figure 26 and step 2704 in Figure 27B).

[0402] Optimizer. Optimization was performed using a genetic algorithm (GA). Single residue mutations included insertion, deletion, and exchange, while multiple residue mutations were performed using hybridization. Single residue mutations involved selecting the index of a residue to delete or, in the case of insertion / exchange, add another residue, and all steps were performed randomly. For hybridization, the length and position of the hybridizing sequence, as well as the hybridized sequence (from the list of all CPPs), were all selected randomly. For all LSTM-generated sequences, a GA was implemented with 1000 evolutionary steps, with the following objective function:

[0403]

number

[0404] (See element 2608 in Figure 26 and step 2708 in Figure 27C).

[0405] Benchmark Models. Using fingerprinting and one-hot encoding, we trained benchmark models: support vector regression, Gaussian process regression, kernel ridge regression, k-nearest neighbor regression, and XGBoost regression.

[0406] Hyperparameter optimization. All hyperparameters of the generator and predictor models were optimized using SigOpt.

[0407] Using this model, a list of 13 peptides was generated.

[0408] [Table 8A]

[0409] [Table 8B]

[0410] [Table 8C]

[0411] [Table 8D]

[0412] wherein X is 6-aminohexanoic acid and B is β-alanine; C L 1 By another C is covalently bound to; L 1 teeth,

[0413] [ka]

[0414] and; M is

[0415] [ka]

[0416] and; R 10 is independently at each occurrence H or halogen.

[0417] Example 11 Activity and toxicity of PMO-P7, dose-response curve The half-maximal effective concentration (EC50) of PMO-P7 was calculated by measuring the eGFP fluorescence of this conjugate (using HeLa654 cells) over a range of concentrations (0.1-100 μM). The obtained EC50 value was 4 μM, a maximal effective concentration that showed a 45-fold increase over unconjugated PMO.

[0418] For the LDH assay, TH1 cells were maintained in DMEM-high glucose supplemented with 10% (v / v) FBS and 1% (v / v) Pen-Strep at 37°C and 5% CO2. 18 hours before treatment, TH1 cells were seeded into 96-well plates at a density of 8,000 cells per well. The next day, fresh 10 mM stocks of each PMO-peptide conjugate were prepared in PBS (1X). The concentration of the stocks was determined by measuring absorbance at 260 nm to be 168,700 L mol -1 Cm -1The absorbance was determined using an extinction coefficient of 1 μM. Growth medium was aspirated from the cells, and different concentrations (1–200 μM) of each conjugate were added to the treatment medium in DMEM-high glucose supplemented with 10% FBS and 1% Pen Strep. The cells were incubated with the treatment medium at 37°C and 5% CO2 for 22 hours. The supernatant treatment medium was then transferred to a separate clear-bottom 96-well plate for assay. The assay was performed using the CytoTox 96® Non-Radioactive Cytotoxicity Assay (Promega) according to the accompanying technical information, except that half the specified volume (25 μL of each supernatant, 25 μL of LDH reagent, and 25 μL of stop solution) was used. Absorbance was measured at 490 nm using a BioTek Epoch Microplate Spectrophotometer. Positive and negative controls correspond to maximum cell lysis and untreated cells, respectively. Data were processed by subtracting the absorbance of untreated cells from all treatment conditions, including cell lysis, and dividing by the corrected lysis value. The % cytotoxicity was calculated as follows:

[0419]

number

[0420] Cell viability was assessed by measuring the amount of lactate dehydrogenase (LDH) released into the cell culture supernatant by injured cells. The conversion of lactate to pyruvate generates NADH, which then reduces a yellow tetrazolium salt (iodonitrotetrazolium violet; INT) to a red formazan color that absorbs at 490 and 492 nm. Consequently, the amount of LDH in the supernatant is proportional to the amount of formazan and provides information about the number of lysed (dead or injured) cells.

[0421] LDH release was assessed using TH1 cells and measured between 1 and 200 μM of PMO-P7, PMO-P21, and PMO-P23 (FIG. 22).

[0422] Example 12 Endotoxin assay The endotoxin content of PMO-P7 was measured using a chromogenic Limulus Amebocyte Lysate (LAL) assay for the detection and quantification of bacterial endotoxins. LAL is an extract of blood cells (amoeba cells) from Atlantic horseshoe crabs. This assay is based on the reaction of LAL with the bacterial endotoxin lipopolysaccharide (LPS), a membrane component of Gram-negative bacteria. In this method, the LAL reagent is mixed with a chromogenic reagent (a peptide linked to p-nitroaniline, a yellow colorant) to generate a synthetic chromogenic substrate. Before incubation, the sample was added to this chromogenic substrate. If endotoxin is present in the sample, a series of enzymatic reactions occur in the LAL reagent, cleaving the peptide bond and releasing p-nitroaniline molecules, thus producing a yellow color. Endotoxin concentration is quantified by measuring absorbance at 405–410 nm. Using the PMO-P7 lot for animal testing, a PMO-P7 solution was prepared using 0.5 mg of PMO-P7 as the acetate salt dissolved in 1 mL of PBS (1X). The cartridge used was a Charles River Endosafe nexgen-PTS 0.01. 25 μL of sample was placed in each of the cartridge's four sample reservoirs. The reader mixed the sample with Limulus Amebocyte Lysate (LAL) reagent. The sample was combined with a chromogenic substrate and then incubated. After mixing, the optical density of the wells was measured and analyzed by comparison with an internal archived standard curve. The reading was 0.0471 EU / mg (EU: endotoxin units).

[0423] The molecular weight of PMO-P7 as its trifluoroacetate salt is 10,069 g / mol and as its acetate salt is 9,529 g / mol.

[0424] Example 13 Animal testing The mice used in this study contain a transgene similar to that used in Example 4 for HeLa654 cells. This mouse model ubiquitously expresses the EGFP-654 transgene throughout the body under the control of the chicken β-actin promoter. A mutant nucleotide 654 in intron 2 of the human β-globin gene interrupts the EGFP-654 coding sequence, preventing proper translation of the EGFP protein. The antisense activity of PMO blocked the aberrant splicing and resulted in EGFP expression similar to that observed in the HeLa654 assay. This study used 6- to 8-week-old male EGFP-654 mice maintained at Charles River Laboratory. These mice were housed in groups with free access to food and water.

[0425] Prior to injection, PMO-peptides were confirmed to have minimal endotoxin levels. For endotoxin assays, 0.5 mg of PMO-P7 (as acetate) was dissolved in 1 mL of PBS (1X) using the PMO-P7 lot used in animal studies. The cartridges used were Charles River Endosafe Nexgen-PTS 0.01. A 25 μL sample was placed in each of the cartridge's four sample reservoirs. The PMO-P7 lot used in animal studies (63 mg as acetate) showed 0.0471 EU / mg (EU refers to endotoxin units).

[0426] After 3 days of acclimation, mice were randomly assigned to receive a single intravenous injection of either saline or PMO-P7 at the indicated dose (5, 10, or 30 mg / kg) via the tail vein. Seven days after injection, mice were euthanized and serum and tissue samples were collected. Quadriceps, diaphragms, and hearts were rapidly dissected, flash-frozen in liquid nitrogen, and stored at -80°C until analysis.

[0427] Serum from all groups was collected 7 days post-injection and tested for markers of renal injury using a Vet Axcel clinical chemistry system (Alfa Wassermann Diagnostic Technologies, LLC). Specifically, serum BUN, creatinine, and cystatin C levels were measured using ACE® Creatinine Reagent (Alfa Wassermann, Cat. No. SA1012), ACE® Blood Urea Nitrogen Reagent (Alfa Wassermann, Cat. No. SA2024), and Diazyme Cystatin C immunoassay (Diazyme Laboratories, Cat. No. DX133C-K), each according to the manufacturer's recommendations (see Figures 24A-C).

[0428] Using a Fast PreP24-5G instrument (MP Biomedical), 20–25 mg of mouse tissue was homogenized in RIPA buffer (Thermo Fisher, catalog no. 89900) containing a protease inhibitor cocktail (Roche, 04693124001). The homogenate was centrifuged at 12,000 g for 10 minutes at 4°C. The resulting supernatant lysate was quantified using the Pierce BCA Protein Assay Kit (Thermo Fisher, catalog no. 23225) and stored for EGFP expression measurement. Specifically, 80 μg of lysate was dispensed into each well of a 96-well black-walled, clear-bottom microplate (Corning). The EGFP fluorescence intensity of each sample was measured in duplicate using a SpectraMAx i3x microplate reader (Molecular devices) with default settings. The average EGFP fluorescence intensity of each sample was then plotted against a standard curve constructed by recombinant EGFP protein (Origen, catalog no. TP790050) to quantify the EGFP protein levels per µg of protein lysate (see Figure 24D-F).

[0429] Incorporation by Reference The contents of all documents cited in this application (including literature references, issued patents, published patent applications, and co-pending patent applications) are expressly incorporated herein in their entirety. All technical and scientific terms used herein shall be given the meaning commonly known to those skilled in the art unless otherwise defined.

[0430] equivalent Those skilled in the art will recognize, or be able to ascertain using no more than routine experimentation, many equivalents to the specific embodiments of the disclosure described herein which equivalents are intended to be encompassed by the following claims.

Claims

1. A peptide-oligonucleotide conjugate comprising a compound of formula II: 【Chemical 1】 or a pharmaceutically acceptable salt thereof (In the formula: A' is -N(H)CH 2 C(O)NH 2 , -N(C 1～6 -alkyl)CH 2 C(O)NH 2 , 【Chemistry 2】 is selected from R 5 is —C(O)(O-alkyl)x-OH, where x is 3 to 10, and each alkyl group is independently selected from the group consisting of C 2～6 -alkyl, Or R 5 is -C(O)C 1～6 -Alkyl, trityl, monomethoxytrityl, -(C 1～6 -alkyl)-R 6 , -(C 1～6 -heteroalkyl)-R 6 , aryl-R 6 , heteroaryl-R 6 , -C(O)O-(C 1～6 -alkyl)-R 6 , -C(O)O-aryl-R 6 , —C(O)O-heteroaryl-R 6 , and 【Chemistry 3】 Selected from; R 6 OH, SH, and NH 2 or R 6 is O, S, or NH, each of which is covalently attached to a solid support; Each R 1 is OH and -N(R 3 )(R 4 ) are independently selected from each R 3 and R 4 independently for each occurrence, -C 1～6 - is alkyl; Each R 2 is independently, at each occurrence, selected from H, a nucleobase, and a nucleobase functionalized with a chemical protecting group, and the nucleobase is, at each occurrence, independently selected from pyridine, pyrimidine, triazinane, purine, and deazapurine. 3～6 -contains heterocyclic rings; z is 8 to 40; E' is H, -C 1～6 -Alkyl, -C(O)C 1～6 -Alkyl, benzoyl, stearoyl, trityl, monomethoxytrityl, dimethoxytrityl, trimethoxytrityl, 【Chemistry 4】 Selected from; Q is -C(O)(CH 2 ) 6 C(O)- or -C(O)(CH 2 ) 2 S 2 (CH 2 ) 2 C(O)-; R 7 is -(CH 2 ) 2 OC(O)N(R 8 ) 2 and R 8 is -(CH 2 ) 6 NHC(=NH)NH 2 and; L is -C(O)(CH 2 ) 1～6 -C 7～15 -Heteroaromatic-(CH 2 ) 1～6 C(O)—, and L is covalently linked to J by an amide bond; J is a carrier peptide; G is H, -C(O)C 1～6 - selected from alkyl, benzoyl, and stearoyl, and G is covalently linked to J; The following conditions: 1) A' is 【Chemistry 5】 being; or 2) E' is 【Chemistry 6】 Being At least one of the following is true: The carrier peptide J is selected from the following sequences: Table 1A Table 1B Table 1C Table 1D wherein X is 6-aminohexanoic acid, B is β-alanine, and C is L 1 is covalently bonded to another C by; L 1 teeth 【Chemistry 7】 and; M is 【Chemistry 8】 and; R 10 is independently at each occurrence H or halogen).

2. E' is H, -C 1～6 -Alkyl, -C(O)C 1～6 -alkyl, benzoyl, stearoyl, trityl, monomethoxytrityl, dimethoxytrityl, trimethoxytrityl, and 【Chemistry 9】 2. The conjugate of claim 1, selected from:

3. A' is -N(C 1～6 -alkyl)CH 2 C(O)NH 2 , 【Chemistry 10】 2. The conjugate of claim 1, selected from:

4. E' is H, -C(O)CH 3 , benzoyl, stearoyl, trityl, 4-methoxytrityl, and 【Chemistry 11】 2. The conjugate of claim 1, selected from:

5. A' is -N(C 1～6 -alkyl)CH 2 C(O)NH 2 , 【Chemistry 12】 is selected from; and E' is 【Chemistry 13】 2. The conjugate of claim 1, wherein:

6. A' is 【Chemistry 14】 and E' is H, -C(O)CH 3 , trityl, 4-methoxytrityl, benzoyl, and stearoyl.

7. The peptide-oligonucleotide conjugate of formula IA is 【Chemistry 15】 and 【Chemistry 16】 is a peptide-oligonucleotide conjugate selected from In the formula, E' is H, C 1～6 -Alkyl, -C(O)CH 3 , benzoyl, and stearoyl.

8. 8. The conjugate of claim 1 or claim 7, wherein the peptide-oligonucleotide conjugate is of formula (Ia):

9. 8. The conjugate of claim 1 or 7, wherein the peptide-oligonucleotide conjugate is of formula (Ib):

10. Each R 1 But N(CH 3 ) 2 The conjugate of any one of claims 1 to 9, wherein

11. Each R 2 11. The conjugate of any one of claims 1 to 10, wherein, independently at each occurrence, is a nucleobase selected from adenine, guanine, cytosine, 5-methyl-cytosine, thymine, uracil, and hypoxanthine.

12. L is -C(O)(CH 2 ) 1～6 -DBCO-(CH 2 ) 1～6 The conjugate of any one of claims 1 to 11, which is C(O)-.

13. L, 【Chemistry 17】 The conjugate of any one of claims 1 to 12, wherein

14. L 1 but, 【Chemistry 18】 and; M, 【Chemistry 19】 The conjugate of any one of claims 1 to 13, wherein

15. L 1 but, 【Chemistry 20】 and; M, 【Chemical 21】 The conjugate of any one of claims 1 to 14, wherein

16. L 1 is covalently attached to the side chains of two cysteines, giving the structure: 【Chemical 22】 The conjugate according to any one of claims 1 to 15, which forms

17. G is H, C(O)CH 3 17. The conjugate of any one of claims 1 to 16, wherein the alkyl group is selected from the group consisting of benzoyl, benzoyl, and stearoyl.

18. G is H or -C(O)CH 3 The conjugate of any one of claims 1 to 17, wherein

19. The conjugate of any one of claims 1 to 18, wherein G is H.

20. G is -C(O)CH 3 The conjugate of any one of claims 1 to 19, wherein

21. 10. The conjugate of claim 1, wherein the peptide-oligonucleotide conjugate exhibits at least a 40-fold improvement in uptake compared to the unconjugated oligonucleotide.

22. 2. The conjugate of claim 1, wherein the peptide-oligonucleotide conjugate exhibits at least a 5-fold improvement in uptake compared to the unconjugated oligonucleotide.

23. 10. The conjugate of claim 1, wherein the peptide-oligonucleotide conjugate is non-toxic.

24. 10. The conjugate of claim 1, wherein the peptide-oligonucleotide conjugate is non-immunogenic.

25. A pharmaceutical composition comprising the conjugate of any one of claims 1 to 24 and at least one pharmaceutically acceptable carrier.

26. A pharmaceutical composition for treating a disease in a subject in need thereof, comprising a therapeutically effective amount of the conjugate described in claim 1, wherein the oligonucleotide is an antisense oligonucleotide.

27. 27. The pharmaceutical composition of claim 26, wherein the disease is a neuromuscular disease.

28. 28. The pharmaceutical composition of claim 27, wherein the neuromuscular disease is Duchenne muscular dystrophy.

Images