Targeted tissue delivery of RNA therapeutics using antibody–oligonucleotide conjugates (AOCs)

Abstract Although targeting TfR1 to deliver oligonucleotides to skeletal muscle has been demonstrated in rodents, effectiveness and pharmacokinetic/pharmacodynamic (PKPD) properties remained unknown in higher species. We developed antibody–oligonucleotide conjugates (AOCs) towards mice or monkeys utilizing anti-TfR1 monoclonal antibodies (αTfR1) conjugated to various classes of oligonucleotides (siRNA, ASOs and PMOs). αTfR1 AOCs delivered oligonucleotides to muscle tissue in both species. In mice, αTfR1 AOCs achieved a > 15-fold higher concentration to muscle tissue than unconjugated siRNA. A single dose of an αTfR1 conjugated to an siRNA against Ssb mRNA produced > 75% Ssb mRNA reduction in mice and monkeys, and mRNA silencing was greatest in skeletal and cardiac (striated) muscle with minimal to no activity in other major organs. In mice the EC50 for Ssb mRNA reduction in skeletal muscle was >75-fold less than in systemic tissues. Oligonucleotides conjugated to control antibodies or cholesterol produced no mRNA reduction or were 10-fold less potent, respectively. Tissue PKPD of AOCs demonstrated mRNA silencing activity primarily driven by receptor-mediated delivery in striated muscle for siRNA oligonucleotides. In mice, we show that AOC-mediated delivery is operable across various oligonucleotide modalities. AOC PKPD properties translated to higher species, providing promise for a new class of oligonucleotide therapeutics.

pharmacological activity ( 2 , 5 ). This is partly due to the relati v ely higher rate of blood perfusion, and discontinuous sinusoidal endothelial ar chitectur e of the li v er, affor ding oligonucleotides to be more amenable for deli v ery and uptake. Although unmodified oligonucleotides have poor drug-like properties, decades of r esear ch investigating the structur e-activity r elationships for this broad class of drugs have identified chemical modifications that improve their phar macokinetic and phar macodynamic properties. Despite advances in improved stability and efficient delivery to the li v er, uptake and pharmacological acti vity in e xtrahepatic tissues have limited the broad application of oligonucleotides as therapeutics ( 6 ).
Stra tegies tha t have been employed to producti v ely deli v er oligonucleotide-based drugs include complex nanoparticle formulations and conjugation to lipids, peptides, or ligands. The de v elopment of lipid nanoparticle (LNP) f ormulations f or deli v ery of siRN A and mRN A enabled clinical testing of these nucleic acid-based classes of thera peutics ( 7 ). LNP form ulations were successful in deli v ering oligonucleotides to the li v er; howe v er, deli v ery to other organ systems was negligible, and their safety profile was not conduci v e for broad applicability as treatment options for chronic diseases ( 8 ). Although significant advances have been made to LNP formulations, alternati v e approaches hav e prov en superior for deli v ery to the li v er, namely conjugation of asialo gl ycoprotein receptor (ASGR) ligands to oligonucleotides. N -Acetylgalactosamine (GalNAc) is a high-affinity ligand for ASGR that is predominately expressed on hepatic parenchymal cells. A trimer of GalNAc linked to an oligonucleotide has demonstrated up to a 10-fold improvement in potency for liver-targeting oligonucleotide therapeutics in preclinical and clinical testing ( 9 ). Although this approach does not afford extrahepatic delivery, it demonstra ted tha t specific targeting of cell surface receptors that undergo internalization is a successful strategy for oligonucleotide deli v ery.
Monoclonal antibodies (mAbs) are well-established single-agent therapeutics, gi v en their specificity and sensitivity for extracellular targets. mAbs have also been utilized for the targeted deli v ery of small-molecule drugs (antibodydrug conjugates, [ADC]), particularly in oncology. ADCs provide specific targeting of a chemotherapeutic to tumor cells while reducing toxicity to normal cells and tissues ( 10 ). The successful application of ADCs in oncology further supports receptor-media ted internaliza tion as a means for effecti v e drug deli v ery. In this regar d, the ability of mAbs to enable specific targeting of oligonucleotides to tissues and cells to expand the therapeutic potential of this class of drugs beyond the li v er has been described ( 11 ). Such molecules consist of an antibody and an oligonucleotide conjugated through a linker. The conjugation of oligonucleotides to mAbs offers a fle xib le and tailored approach where the antibody targets a cell type of interest to deli v er an oligonucleotide for the desired disease-specific gene target. Indeed, Cuellar et al. demonstrated the feasibility of conjugating siRNA oligonucleotides to mAbs and producing conjuga tes tha t are acti v e and stab le in vitro for several mAb receptor and siRNA gene targets. In brief, the mAb-siRNA conjugates maintained comparable binding affinity and gene silencing relati v e to the mAb or siRNA alone in vitro ( 12 ). Howe v er, the in vivo activity of the mAb-siRNA constructs in tumor tissue was modest ( 12 ). Nonetheless, this and other early r esear ch provided evidence for the antibody-siRNA constructs to target and enter cell types outside of the li v er, including skeletal muscle ( 13 ). Sugo et al. described the utility of targeting the transferrin receptor 1 (TfR1) to deli v er oligonucleotides to skeletal muscle and heart tissue. They demonstrated in rodents a reduction of Mstn mRNA expression in muscle with an ␣TfR1-antibod y fragment (Fab)-linker-siRNA conjuga te.
Gi v en this progress, we investiga ted the transla tion of these observations to higher species in non-human primates using monoclonal antibod y-oligonucleotide conjuga tes (AOC). We engineered AOCs targeting mouse ASGR ( ␣ASGR) and mouse or human TfR1 ( ␣TfR1) receptors to facilitate the functional deli v ery of oligonucleotides to li v er and muscle cells, respecti v el y, across various mRN A targets with siRNA, ASO or PMO oligonucleotide modalities to further investigate the therapeutic potential and PKPD properties of the AOC platform.
We demonstra te tha t ␣TfR1-targeting AOCs are pharmacolo gicall y acti v e in non-human primates. Furthermore, the PKPD properties of the AOC platform demonstrate the dependence of receptor-mediated deli v ery for acti vity in skeletal and cardiac muscle with apparent selecti v e acti vity in striated muscle as compared to other tissues. This potential br oad appr oach to enable pr oducti v e oligonucleotide deli v ery to tissues beyond the li v er provides promise for a new class of oligonucleotide therapeutics for the treatment of muscle diseases, among others, with high unmet need.

Oligonucleotide synthesis
The oligonucleotide (siRNA and ASO) sequences were assembled on solid phase using well-described and standard phosphoramidite methodology and purified by highperformance liquid chromato gra phy. The sugar modifications (2 -fluoro, 2 -O -methyl and LNA), and backbone modifica tions (phosphorothioa tes) tha t are well described in the fields of RN A interference (RN Ai) and antisense were used to optimize the potency of the siRNA and ASO classes of oligonucleotides ( 14 , 15 ). The siRNA passenger strand and the ASO used for conjugation were synthesized with a C6 amino linker at the 5 -end. A trivalent GalNAc (purchased from AxoLabs, Kulmback, Germany) was conjugated to the siRNA at the 5 -end of the passenger strand. PMOs were purchased from Gene Tools LLC (Philomath, OR).

Antibody generation
A rat IgG2a anti-mouse TfR1 antibody ( ␣TfR1; Bio X Cell, Lebannon, NH) or a rabbit IgG anti-mouse ASGR1 antibody ( ␣ASGR; SinoBiological, Wayne, PA) was utilized for AOC synthesis in rodent experiments. A mouse IgG2 anti-human TfR1 that is cross-reacti v e in cynomolgus monkeys was de v eloped ( ␣hTfR1, Supplementary Table S1) for experiments in monkeys. A rabbit isotype control antibody ( ␣IgG) was utilized as a non-receptor-targeting Nucleic Acids Research, 2023, Vol. 51, No. 12 5903 contr ol AOC (Cr ownBio, Santa Clara, CA). For the TfR1 and ASGR antibodies, the binding affinity to the respecti v e receptor target was 42 and 4.8 pM, respecti v ely. Importantly, conjugation of an oligonucleotide had no impact on antibody binding to the target receptor (Supplementary Figure S1).

AOC synthesis
AOCs were generated using a standard random cysteine conjugation method ( 16 ). The interchain disulfide bonds of the antibody were partially reduced with tris(2carboxyethyl)phosphine (TCEP) prior to conjugation with a maleimide linker-oligonucleotide. The ASO and siRNA conjugate reaction mixtures were purified using strong anion exchange chromato gra phy to ensure an average drug-toantibod y ra tio (DAR) of 1 f or siRNA or 2.5 f or ASO AOCs. PMO AOCs were purified by hydrophobic interaction chromato gra phy with an average DAR of 1.5. The composition of all test articles is listed in Supplementary Table S2.

Animal treatment
All animal studies were conducted following protocols approved by the Institutional Animal Care and Use Committee (IACUC), in accordance with the regulations outlined in the USDA Animal Welfare Act as well as the 'Guide for the Care and Use of Laboratory Animals' ( 17 ).
Male or female CD-1 mice (Envigo, Indianapolis, IN), a pproximatel y 6-8 weeks old, were allowed to acclimate for at least 1 week prior to dosing. AOCs wer e administer ed by intravenous (IV) bolus injection to the tail vein (5 mL / kg) at the described dose. At the specified timepoints post a single dose, blood was collected and processed to plasma for assessment of AOC drug concentrations. At designated timepoints, animals were euthanized with carbon dioxide followed by cervical dislocation for collection of tissues (skeletal muscles, heart, li v er, kidney, among others) that were frozen in dry ice for assessment of oligonucleotide tissue concentrations or pharmacodynamic endpoints.
Na ïve male cynomolgus monkeys (a pproximatel y 2-4 years old, 2-3 kg body weight, Cambodia-sourced) were administered a single IV infusion (30 min at 5 ml / kg) at 6 mg / kg. Animals were euthanized using Euthasol (Virbac AH Inc., Fort Worth, TX) on Day 28 post dose and skeletal muscle as well as systemic tissues (25-40 mg) were collected. Samples were frozen in liquid nitrogen for evaluation of pharmacodynamic endpoints and oligonucleotide tissue concentrations. The in-life portion of the cynomolgus monkey experiments were conducted at Altasciences (Everett, WA).
The dose is expressed as the oligonucleotide component for all test molecules evaluated for all in vivo studies described, to allow for appropriate comparison across studies and classes of oligonucleotide chemistries.

Pharmacodynamic evaluation in tissue
Tissue was processed for RNA extraction using the Zymo Direct-zol-96 RNA purification kit (Zymo Research Corporation, Irvine, CA). Once RNA quality was confirmed, cDNA was generated from purified RNA and used in quantitati v e re v erse transcription PCR (RT-qPCR) analysis for gene expression for each of the targets of interests ( 18 ). The r efer ence mRNA transcripts for Ppib or AHSA1 were used in Ct calculations for mouse and cynomolgus monkey tissue samples, respecti v ely ( 19 ). For e xon skipping, nested RT-PCR was utilized as previously described ( 20 ). Briefly, primers in exon 20 (5 -A CCCAGTCTA CCA CCCTATC) and exon 25 (5 -CT CTTTAT CTT CTGCCCACCTT) wer e used to amplify both skipped and unskipped mRNA transcripts. PCR products were analyzed in 4% TAE agarose gel electrophoresis to resolve skipped and unskipped transcripts. Bands were quantified using image densitometry.
To evalua te siRNA concentra tions in plasma or tissues, custom stem-loop-RT-qPCR assays were de v eloped as described previously ( 21 ). For each siRNA, a specific stem-loop-RT-qPCR assay was designed to quantify the guide strand, using custom DNA forwar d, re v erse, and reverse transcription primers (Integrated DN A Technolo gies, Coralville, IA) and a custom Taqman probe (ThermoFisher Scientific, Waltham, MA). To evaluate ASO concentrations in tissues, a custom oligonucleotide probe hybridization Meso Scale Discovery (MSD) electrochemiluminescent (ECL) assay was de v eloped. The probe was complementary to the ASO sequence and contained flanking locked nucleic acid residues with internal DNA residues. The assay followed a similar format to previously described methods ( 22 ) but was converted to MSD format with a ruthenium-labeled detector.

Statistical analysis
GraphP ad Prism softwar e was utilized for all descripti v e and statistical analyses. As appropriate, one-or two-way ANOVA or paired t-tests were perf ormed f or the datasets analyzed. In the case of ANOVA, an appropriate post hoc test was utilized to determine the differences among the trea tment groups. Significant dif ferences were defined as P < 0.05.

AOC-mediated tissue delivery
The selecti v e deli v ery of oligonucleotide therapeutics to target tissues of interest was evaluated for the AOC platform in mice. For these experiments, an siRNA-targeting catenin beta 1 ( Ctnnb1 ) mRNA (si Ctnnb1 ) was conjugated to an antibody targeting mouse ASGR ( ␣ASGR-si Ctnnb1 ) or mouse TfR1 ( ␣TfR1-si Ctnnb1 ) for li v er or skeletal muscle siRNA deli v ery, respecti v ely. The tissue concentrations of si Ctnnb1 in gastrocnemius (GA) muscle were ∼9-fold greater with the ␣TfR1-si Ctnnb1 as compared with the ␣ASGR-si Ctnnb1 ; in contrast, ∼6-fold greater li v er concentrations were deli v ered with the ␣ASGR-si Ctnnb1 ( Figure  1 ). These r esults ar e consistent with expr ession of TfR1 in muscle and ASGR primarily in the li v er ( 23 , 24 ). Although TfR1 is minimally expressed in the li v er, meaningful concentrations of si Ctnnb1 are present in this tissue, which is suggesti v e of non-r eceptor-mediated uptake. Non-r eceptormediated uptake may also be evident in the heart, given that meaningful concentrations of siRNA were observed with Female CD-1 mice were treated with a single IV dose of antibody-siRNA conjugates; each composed of an antibody targeting either mouse ASGR or mouse TfR1 and an siRNA targeting Ctnnb1 mRNA at 3 mg / kg. GA, heart, and li v er samples were collected 4 days post dose, and siRNA concentration was determined using stem-loop qPCR (normalized to tissue weight, mean ± SEM; N = 4). Statistical analysis was performed using two-way ANOVA with Bonferroni post hoc test. *Indicates statistical difference at P < 0.05. the ␣ASGR-si Ctnnb1 , despite a lack of ASGR expression in the heart. An ∼1.5-fold increase in siRNA concentration in the heart with the ␣TfR1-si Ctnnb1 indicates receptormediated deli v ery to the heart ma y also be aff orded by TfR1 targeting (Figure 1 ). Oligonucleotides are known to bind non-specifically to proteins and cell surface receptors, and, in part, may account for some of the uptake of si Ctnnb1 in tissues that may not be dri v en by the antibody binding to its respecti v e target, particularly the li v er. Nonetheless, nonreceptor-mediated uptake of the antibody (e.g. neonatal Fc r eceptor r ecycling) may also be a contributory factor.

AOC-mediated activity in the liver in the mouse
Conjugation of the GalNAc ligand to oligonucleotides is an effecti v e deli v ery mechanism of siRNAs to the li v er, gi v en the selecti v e e xpression of ASGR on hepatocytes ( 9 ). We compared the siRNA-mediated potency of ␣ASGR AOCs across various mRNA targets relati v e to GalNAc-mediated deli v ery in mice. The AOC molecules targeting three li v erexpressing tar gets (hypo xanthine-guanine phosphoribosyltr ansfer ase [si Hprt ], si Ctnnb1 and factor VII [si FVII ]) produced dose-dependent reduction in hepatic mRNA expression of each transcript with comparable activity ( Figure  2 ). Furthermor e, mRNA r eduction for these molecules was receptor-dependent, indicated by a lack of activity when conjuga ting the si Hpr t to an isotype control antibody imm uno globulin G1 ( ␣IgG1), ␣IgG1-si Hprt . A scrambled siRNA sequence (siScr) with no kno wn mRNA tar get also did not modulate the expression of any of the target genes evaluated. The potency for the ␣ASGR AOCs appeared to be comparable with that of GalNAc-mediated deli v ery gi v en that ␣ASGR-si FVII had an ED 50 of 0.1 mg / kg while a GalNAc conjugate of the identical siRNA had an ED 50 of 0.2 mg / kg.

AOC-mediated activity in muscle tissue in the mouse
Oligonucleotide pharmacological mechanisms of action may address a broad range of muscle diseases; howe v er, Figure 2. ␣ASGR AOCs produce mRNA reduction in li v er across multiple gene targets. Female CD-1 mice were treated with a single IV dose of AOCs composed of an antibody targeting mouse ASGR and an siRNA targeting either Hprt, Ctnnb1, or FVII mRNAs a t indica ted doses. A GalNAc conjugated to an siRNA targeting FVII mRNA was also evaluated following a single IV dose. Li v er was collected 4 days post dose, and mRNA expression was analyzed by RT-qPCR. Hprt, Ctnnb1, and FVII mRNA expression was normalized to that of a reference gene, Ppib . Data are represented as percent of vehicle control (mean ± SEM; N = 4 for treated groups, N = 5 for vehicle groups). Statistical analysis was performed using one-way ANOVA with Dunnett's post hoc test. Statistical difference relati v e to vehicle control at P < 0.05 was observed for the ␣ASGR-si Hprt, ␣ASGR-siFVII, GalNAc-siFVII, and ␣ASGR-siCtnnb1 at all doses evaluated.
oligonucleotide deli v ery to striated muscle tissue is a key limitation for this class of molecules. Thus, we de v eloped an AOC to mediate oligonucleotide deli v ery to muscle tissue using ␣TfR1. To determine if the deli v ered siRNA into muscle is pharmacologically acti v e, se v eral ␣TfR1 AOCs conjugated to siRNAs targeting muscle-expressed genes m yostatin (si Mstn ), m yotonic dystr ophy pr otein kinase (si Dmpk ), or small RNA binding exonuclease protection factor La (si Ssb ) were tested in vivo in mice. All AOCs evaluated ( ␣TfR1-si Mstn , ␣TfR1-si Dmpk , and ␣TfR1-si Ssb ) produced dose-dependent reduction of mRNA expression for each target gene in GA muscle (Figure 3 ). Reduction of mRNA expression was > 80% for ␣TfR1-si Mstn and ␣TfR1-si Dmpk siRNAs, demonstra ting tha t marked pharmacological activity could be afforded by ␣TfR1-mediated deli v ery to muscle. Moreover, ␣TfR1-si Mstn was at least 10fold more potent than conjugation of cholesterol (Chol) ligand to the siRNA (Chol-si Mstn ), demonstrating the higher efficiency of the AOC platform relati v e to conjugation of a non-targeted ligand.
Gi v en the relati v ely broad protein expression of TfR1 ( https://www.proteinatlas.org/ ENSG00000072274-TFRC / tissue), we evaluated the uptake and activity of ␣TfR1-si Mstn and ␣TfR1-si Ssb across a broad panel of skeletal muscles and systemic tissues. Indeed, ␣TfR1si Mstn produced dose-dependent siRNA deli v ery in all skeletal muscles evaluated (Figure 4 ). siRNA concentrations ranged from 10 to 100 nM in muscle tissue at the highest dose tested (3 mg / kg). Mstn mRNA expression was similarly reduced in a concentration-dependent manner in m uscle tissue, w here acti vity was e vident in the diaphragm as low as 0.1 mg / kg. At doses ≥1 mg / kg, near-maximal reductions in Mstn mRNA expression ( ≥80% relati v e to vehicle control) were produced in all skeletal muscles evalua ted. To demonstra te the tissue specificity for the ␣TfR1 AOC, a gene target with ubiquitous expression, Ssb mRNA, was evaluated. The ␣TfR1-si Ssb was the most potent in muscle for Ssb mRNA reduction with an EC 50 of 0.4 nM in GA muscle (Table 1 ). In contrast, mRNA silencing activity was ∼10-fold less potent in the heart and ∼100-fold less potent in the li v er, relati v e to that achie v ed in skeletal muscle. Minimal to no activity was evident in other systemic tissues following ␣TfR1-si Ssb , including in the kidney, spleen, and gastrointestinal tract (GI [colon]).

AOC-mediated oligonucleotide delivery and activity in muscle tissue in cynomolgus monkeys
Based on the promising pharmacological activity of the AOC platform in mice, testing in non-human primates (cynomolgus monkeys) was performed to determine the translation of the pharmacokinetic and pharmacodynamic properties of this new oligonucleotide delivery platform in higher species. A human, cynomolgus monkey crossreacti v e, antibod y ( ␣hTfR1) conjuga ted to an siRNA targeting monkey SSB mRNA ( ␣hTfR1-si SSB ) was tested at a single dose of 6 mg / kg. ␣hTfR1-si SSB producti v ely deli v ered siRNA to skeletal muscle tissue, resulting in tissue concentra tion tha t ranged from 2 to 6.6 nM a t Day 28 post single dose ( Figure 5 ). The deli v ery of si SSB to muscle led to reductions of up to 75% (relati v e to vehicle-treated animals) in SSB mRNA expression in GA and quadriceps muscle ( Figure 5 ). ␣hTfR1-si SSB demonstrated pharmacological activity in a broad panel of muscles, including the heart, at Day 28 following a single dose (Supplementary Figure  S2). As observed from mouse studies, no meaningful activity was noted in li v er, kidney, lung, or spleen, as Ssb mRNA expression was within the normal range of control in these tissues after the AOC treatment.

AOC-mediated activity is amenable to various types of oligonucleotide chemistry
Gi v en that oligonucleotide therapeutics comprise se v eral chemical classes and utilize multiple mechanisms of pharmacological action, we evaluated the applicability of the AOC platform to other chemical classes, including PMO (exon splice switching) and phosphorothioate ASO (ribonuclease H1-mediated mRNA degradation) oligonucleotides. A PMO tar geting ex on 11 skipping of phenylalanine hydroxylase (pmo Pah ) pre-mRNA was conjugated to ␣ASGR to assess activity in the li v er. In mice, ␣ASGR-pmo Pah produced measurable skipping of exon 11 of Pah , with no evidence of skipping with a scrambled PMO sequence ( ␣ASGR-pmoScr) as control (Supplementary Figure S3). An ASO targeting Dmpk mRNA (aso Dmpk ) was conjugated to ␣TfR1 to assess activity in skeletal muscle. A single dose of ␣TfR1-aso Dmpk produced dose-dependent activity with > 75% mRNA reduction of Dmpk mRNA in GA muscle at the highest dose in mice (Figure 6 A). Importantly, the unconjugated ASO (aso Dmpk ) r equir ed a ∼25-fold greater cumulati v e dose to achie v e comparab le acti vity that further substantiates the efficiency of receptor-mediated deli v ery for pharmacological activity (Supplementary Figure S4).
We note that ␣TfR1-mediated deli v ery to muscle appeared to be independent of the oligonucleotide chemical class, gi v en that ␣TfR1-si Dmpk and ␣TfR1-aso Dmpk achie v ed comparab le oligonucleotide GA concentrations at similar dose le v els. Surprisingly, up to 10-fold greater aso Dmpk was present in li v er as compared with si Dmpk, which was associated with > 80% Dmpk mRNA reduction at 5.4 mg / kg (Figure 6 B). ␣TfR1-si Dmpk had negligib le acti vity in li v er with no clear dose dependency at the same dose tested (5.4 mg / kg). Collecti v ely, these data suggest tha t non-TfR1-media ted mechanisms may be accountable for the uptake and activity of ␣TfR1-aso Dmpk in the li v er, and these properties appear to be dependent on the oligonucleotide chemical class employed.

DISCUSSION
Advancements to improve the metabolic stability and safety profile, economize large-scale manufacturing, and identify therapeutic applications, among others, have propelled oligonucleotides as mainstream investigational therapeutics across a spectrum of diseases. A key limitation has been producti v e deli v ery to tissues, gi v en that the li v er has been the primary target tissue for systemically administered oligonucleotide therapeutics ( 25 ). Nonetheless, there has been r ecent progr ess with local administration to effecti v ely deli v er siRNAs to the central nervous system (intrathecal injection of divalent siRNA or 2 -O -hexadecyl [C16]-siRNA conjugates, and intravitreal injection of C16-siRNA conjugates) or lung (C16-siRNA conjugates via intranasal  injection) ( 26 , 27 ). We report the use of monoclonal antibody-based AOCs that are administered intravenously to deli v er oligonucleotides to tissues via specific receptormediated endocytosis, with a focus on skeletal muscle and heart. These findings, including the translation to nonhuman primates with ␣TfR1, constitute a potential breakthrough in oligonucleotide deli v ery.

AOC activity is translated from rodents to non-human primates
Although prior r esear ch demonstrated the utility of TfR1 targeting for deli v ery of siRN As to m uscle in rodents, the translation of these findings to larger species was unknown ( 13 ). Gi v en the successful utility of monoclonal antibodies to facilitate deli v ery of small-molecule cytotoxic drugs in oncology ( 28 ), we postulated that this would be a more favorable approach relative to other reported modalities (e.g. antibody fragments and peptides) ( 13 , 29 ). For example, the specificity and selectivity for the intended target afford biologics minimal potential for non-specific interactions relati v e to peptides, and neonatal Fc r eceptor r ecy cling affor ds improv ed plasma pharmacokinetic properties for full-length antibodies that are generally absent for antibody fragments. In this regard, we engineered human ␣hTfR1 that is cynomolgus monkey cross-reacti v e with low pM binding affinity to TfR1 to determine the translation of utilizing TfR1 targeting to deli v er oligonucleotides to mus- Figure 5. ␣TfR1-si SSB producti v ely deli v ers siRNA to muscle and produces mRNA reduction in cynomolgus monkeys. Male cynomolgus monkeys of Cambodian origin were administered a single IV dose of SSB siRNA conjugated to an antibody targeting TfR1 at 6 mg / kg. Tissues were collected at 28 days post dose, and mRNA expression was analyzed by RT-qPCR, w hile siRN A tissue concentrations were determined using stem-loop RT-qPCR. SSB expression was normalized to reference gene activator of HSP90 ATPase activity 1 ( AHSA1 ). mRNA expression (percent of vehicle control) and siRNA tissue concentration (normalized to tissue weight) data ar e r epr esented as mean ± SEM ( N = 3). Statistical analysis was performed using one-way ANOVA and Tukey's post hoc test. *Indicates significant difference relati v e to vehicle control at P < 0.05.
cle for pharmacological activity. Importantly, ␣TfR1 conjugated to the linker-oligonucleotide components did not impact binding affinity to TfR1 (Supplementary Figure S1). We show that anti-TfR1 AOCs targeting Ssb mRNA in muscle are acti v e in both mice and cynomolgus monkeys following a single dose. In the monkey, ␣hTfR1-si SSB achie v ed up to 70% reduction of SSB mRNA in multiple skeletal muscles that was comparable to that in the mouse.
Nucleic Acids Research, 2023, Vol. 51, No. 12 5907 Figure 6. ␣TfR1 AOCs conjugated to siRNA or ASO oligonucleotides produce mRNA silencing activity in muscle. Male C57BL / 6 mice were treated with a single IV dose of AOCs each composed of an antibody targeting mouse TfR1 conjugated to either an siRNA ( ␣TfR1-si Dmpk ; DAR = 1) or an ASO ( ␣TfR1-aso Dmpk ; DAR = average 2.5) targeting Dmpk mRNA at indicated doses. Tissue samples were isolated 14 days post dose, and mRNA expression was analyzed by RT-qPCR. siRNA concentration was assessed by stem-loop qPCR, and ASO concentration was assessed by oligonucleotide probe hybridization electrochemiluminescent (ECL) assay (data normalized to tissue weight). Dmpk mRNA expression was normalized to that of a r efer ence gene, Ppib . Data are represented as percent of vehicle control (mean ± SEM; N = 4 for treated groups, N = 5 for vehicle group). Statistical analysis for KD was performed using one-way ANOVA and Dunnett's post hoc test. Statistical significance relati v e to v ehicle control at P < 0.05 was noted for ␣TfR1si Dmpk in muscle and for ␣TfR1-aso Dmpk in muscle and li v er at all doses evaluated. To compare the ␣TfR1-si Dmpk -and ␣TfR1-aso Dmpk groups, a two-way ANOVA with Sidak's post hoc test was performed, and statistical significance was observed at the 0.6 and 1.8 mg / kg dose groups with P < 0.05. ␣TfR1-aso Dmpk (TC) in gastrocnemius at 0.6 mg / kg dose was below the limit of quantification. KD: mRNA knockdown; TC: oligonucleotide tissue concentration.
Importantly, ␣hTfR1 deli v ery resulted in siRN A m uscle tissue concentrations 10-fold above the EC 50 , as determined in mice. In addition to skeletal muscle tissue, systemic dosing of AOCs in mice and cynomolgus monkeys resulted in sufficient deli v ery of oligonucleotides to the heart, leading to marked Ssb mRNA reduction. This is the first report demonstrating the producti v e deli v ery and pharmacological activity of an AOC in skeletal muscle and the heart in non-human primates. Although targeting SSB has no immediate thera peutic a pplication, this w ork pro vides proof of concept that the AOC platform producti v ely deli v ers siRN A thera peutics to m uscle.

PKPD characterization of AOCs demonstrates that receptormediated delivery leads to tissue-selective pharmacological activity
We show that the oligonucleotide tissue deli v ery mediated by antibodies to either TfR1 or ASGR is highly efficient and producti v e for skeletal muscle and li v er, respecti v ely. An siRNA targeting Ctnnb1 mRNA conjugated to ␣TfR1 or ␣ASGR demonstrated tissue selectivity for oligonucleotide deli v ery and mRNA reduction. ␣TfR1-si Ctnnb1 led to a 10-fold greater uptake of si Ctnnb1 to skeletal muscle relati v e to li v er, while similar specificity was produced with the ␣ASGR-si Ctnnb1 for li v er deli v ery. To further support the dependence on receptor-mediated activity, we demonstra ted tha t mRN A reduction was not evident w hen conjugating an siRNA to a control non-specific antibody imm uno globulin G1 ( ␣IgG1-si Hprt ). Moreover, siRN A deli v ered via the AOC platform was a pproximatel y 10-fold more potent for Mstn mRNA r eduction, r elati v e to conjugation of oligonucleotides via a non-receptor-mediated deli v ery mechanism (conjugation to cholesterol). As antic-ipa ted, administra tion of unconjugated Mstn siRNA led to minimal uptake into skeletal muscle tissue, as most of the drug was ra pidl y cleared from systemic circulation (plasma half-life [ t 1 / 2 ] < 0.5 hours, Supplementary Figure S5). In contrast, ␣TfR1-conjugated si Mstn ( ␣TfR1-si Mstn ) had much lower plasma clearance ( t 1 / 2 of ∼6 hours) with a plasma concentration-time profile comparable to the antibod y alone, indica ting tha t the AOC assumed the antibody pharmacokinetic profile. This also demonstrates the stability of the antibody-siRNA comple x, gi v en that the assays measuring total antibody or siRNA components exhibited comparable plasma exposure. The efficiency of receptormediated activity was also demonstrated with an ␣TfR1aso Dmpk AOC, gi v en that a 25-fold greater cumulati v e dose of the unconjugated aso Dmpk was r equir ed to achie v e comparable Dmpk mRNA reduction in muscle tissue relati v e to a single dose of the AOC (Supplementary Figure S4).
We also noted presence of oligonucleotide in tissues that would not be anticipated based on the relati v ely low expression of the target receptor. For example, meaningful concentrations of si Ctnnb1 were present in the li v er and heart with the ␣TfR1or ␣ASGR-conjugated AOCs, respecti v ely, despite relati v ely minimal to low e xpression of the target receptors in these tissues. This was more evident with the ␣TfR1-aso Dmpk AOC, with which 10-fold greater tissue concentrations of the ASO were noted in the li v er compared with the siRNA AOC, ␣TfR1-si Dmpk (Figure 6 B). We postula te tha t non-receptor-media ted uptake exists for AOCs and the extent of uptake is dependent on the conjugated oligonucleotide.
To evaluate the impact of receptor-versus non-receptormediated uptake on pharmacological activity, we characterized the PKPD properties of an AOC w hose mRN A target expression is ubiquitous across a broad panel of tissues. In this regard, the Ssb mRNA target was selected, in large part, due to its relati v ely uniform e xpression across tissues, with slightly greater expression in the spleen relati v e to other organs evaluated ( 30 ). Indeed, in mice the EC 50 of si Ssb was 0.3 nM in skeletal muscle tissue for ␣TfR1-si Ssb , while 10-to 100-fold higher concentrations were required for activity in heart and li v er, respecti v ely (Tab le 1 ). Furthermor e, no meaningful r eductions in the target mRNA Ssb were evident in other organ systems evaluated, such as the GI tract, spleen, and kidney. Our findings in mice were translated to non-human primates where no meaningful reductions in SSB mRNA were evident with ␣TfR1-si SSB in the non-striated muscle tissues evaluated (li v er, kidney, lung and spleen). It is unlikely that the differences in mRNA silencing activity were due to differences in SSB expression among the tissues evaluated. Although we did not evaluate the relati v e mRNA or protein e xpression of TfR1 among various cell types within a tissue, the higher whole tissue expression of TfR1 in skeletal muscle may, in part, contribute to the tissue selectivity of pharmacological activity ( 12 , 30 ). Nonetheless, it does not fully explain our observations given the substantial 10-fold greater activity in skeletal muscle relati v e to heart at much lower siRNA concentrations. We postula te tha t receptor-media ted uptake leads to more producti v e oligonucleotide deli v ery and acti vity; howe v er, skeletal myofibers appear to be more sensiti v e to acti vity relati v e to other TfR1-expressing cells. More work is needed to fully understand the potential mechanism(s) of tissue-selecti v e pharmacological activity of AOCs, as this may include differences in endosomal processing, Ago-2-mediated mRNA cleavage, cell-specific expression of TfR1, among others, across cells / tissues. Our findings provide the foundation for further r esear ch to study and identify tissue-specific mechanisms that are critical for the activity of oligonucleotide therapeutics afforded by receptor-mediated internalization processes.
␣TfR1 AOCs support delivery and activity across multiple chemical classes of oligonucleotide therapeutics Because oligonucleotide therapeutics include numerous mechanisms of pharmacological action across various chemical classes, we evaluated whether the AOC platform was applicable to single-stranded ASO and PMO, as was demonstrated with a double-stranded siRNA. Indeed, AOCs comprised of ␣TfR1 conjugated to ASO for Dmpk mRNA ( ␣TfR1-aso Dmpk ) achie v ed > 80% knockdown of mRNA expression in mouse skeletal muscle following a single 10 mg / kg dose. As proof of concept for modulation of pre-mRNA splicing, an antibody designed for li v er deli v ery ( ␣ASGR) conjugated to a PMO targeting Pah ( ␣ASGR-pmo Pah ) was evaluated in vi vo . Measur able skipping of exon 11 of the Pah pre-mRNA was evident in the li v er following a single dose of ␣ASGR-pmo Pah (Supplementary Figure S3). This demonstrates the potential utility of the AOC platform to producti v ely deli v er another chemical class of oligon ucleotide, PMO , in vivo . Gi v en that modulation of pre-mRNA splicing by oligonucleotides has been demonstrated to be thera peuticall y beneficial for se v eral disorders, especially in muscle disease , we , along with others, have reported potent and durable exon skipping activity for a PMO in an animal model of Duchenne muscular dystrophy ( 29 , 31 ).
Inter estingly, ther e wer e notable differences in activity between the AOCs with siRNA and ASO. The siRNA was ∼3-fold more potent than the ASO in skeletal muscle, despite the concentrations of the siRNA and ASO being comparable ( Figure 6 ). This finding suggests that the difference in potency is not likely due to differential muscle deli v ery of the oligonucleotide gi v en that the same linker and ␣TfR1 antibody were utilized. In part, the inherent efficiency of RNA-induced silencing complex for siRNAs versus ribonuclease H1-mediated mRNA degradation by ASOs may account for the increased activity of the siRNA in skeletal muscle ( 32 ). Howe v er, in the li v er, the ASO produced substantial Dmpk mRN A reduction, w hile the siRNA was devoid of meaningful activity. The potency of ␣TfR1-aso Dmpk for mRNA reduction in skeletal muscle and in the li v er appears to be comparable, while the ␣TfR1si Dmpk demonstrated apparent selectivity for productive deli v ery and target-mRNA silencing in muscle tissue. We did not evaluate the cellular distribution of the oligonucleotides within the li v er so the target cell type(s) of the ASO is (are) uncertain. Dmpk mRNA expression appears higher in Kupffer cells relati v e to hepatocytes or endothelial cells ( 33 ), so it is reasonable to assume that Kupffer cells may be the primary cell target for the ASO within the li v er. Gi v en that ASOs have also been shown to bind to plasma membrane proteins ( 34 ), non-receptor-mediated uptake in the li v er appears to be greater for this chemical class. The findings are consistent with the majority of the activity in the li v er for the ␣TfR1-aso Dmpk being independent of the antibody portion of the AOC.

Therapeutic potential of antibody-mediated oligonucleotide delivery
A key advantage of the AOC platform is that a single antibody may be utilized across multiple diseases within a tissue type where the oligonucleotide component would be interchangeable based on the therapeutic target. Thus, the safety profile of interacting and modulating TfR1 would be important to characterize as it may have implications across se v eral potential drug candidates, especially gi v en the broad tissue expression of TfR1 and its main function of transporting iron via transferrin into cells critical for normal hematopoiesis ( 35 ). In this regard, we engineered an ␣hTfR1 antibody that would not interfere with binding of transferrin to TfR1 (Supplementary Figure S6A). Nonetheless, the transient internalization of TfR1 following antibody binding may impact the potential availability of transferrin-mediated iron uptake immediately follo wing AOC administration. Ho we v er, gi v en the relati v ely short plasma half-life t 1 / 2 ( < 24 h), the impact of this on hematopoiesis may be transient. Importantly, we modified the antibody to lack binding to Fc gamma receptors to elimina te ef fector function. Indeed, we demonstra ted tha t the ␣hTfR1 antibody was devoid of effector function in an in vitr o antibod y-dependent cell cytotoxicity assay (Supplementary Figure S6B).
In summary, we report a versatile approach using monoclonal AOCs for producti v e deli v ery of oligonucleotides Nucleic Acids Research, 2023, Vol. 51, No. 12 5909 to target muscle tissue that has been generally inaccessible to RN A thera peutics. We demonstrate the translation of the PKPD properties of the AOC platform across species from rodents to non-human prima tes, illustra ting the consistency and potential utility of this approach in humans. In fact, the first A OC therapeutic, A OC 1001, is in early clinical testing for the treatment of myotonic dystrophy type 1 (MARINA; https://clinicaltrials.gov/ct2/show/ NCT05027269 ). AOC 1001 is a humanized mAb targeting human TfR1 conjugated to an siRNA oligonucleotide against DMPK mRN A, w here m utations on this gene are the cause and genetic basis of the disease. The results of clinical testing to characterize the safety and efficacy of investigational AOC therapeutics are likely to springboard the discov ery and de v elopment of AOCs designed for deli v ery to other organs and tissue types to further extend the utility of RNA therapeutics.

DA T A A V AILABILITY
The data underlying this article are available in the article and in its online supplementary material.