Abstract
Angelman syndrome (AS) is a severe neurodevelopmental disorder caused by the loss of neuronal ubiquitin E3 ligase UBE3A, with no available treatment. Restoring UBE3A by downregulating the paternally cis-acting long noncoding antisense transcript (UBE3A-ATS) is a potentially disease modifying strategy. However, developing molecules targeting human UBE3A-ATS is challenging due to its selective expression in mature neurons and lack of sequence conservation across species. To overcome this, we screened a library of locked nucleic acid (LNA)-modified antisense oligonucleotides (ASOs) in AS patient-derived neurons. This let to the identification of rugonersen (RO7248824), which selectively and potently reduces UBE3A-ATS and upregulates UBE3A messenger RNA (mRNA) and protein in neurons derived from neurotypical humans, AS patients, and cynomolgus monkeys. In vivo studies with rugonersen or tool molecules in wild-type and AS mice, and cynomolgus monkeys revealed a steep relationship between Ube3a-ats knock-down and UBE3A mRNA/protein upregulation, requiring ∼90% knock-down for 50% upregulation. Two studies of up to three lumbar intrathecal (IT) rugonersen doses in monkeys showed no adverse effects and produced long-lasting paternal UBE3A mRNA/protein reactivation in key brain regions. In summary, we identified rugonersen, an ASO targeting UBE3A-ATS with excellent drug-like properties. Its sustained efficacy supports infrequent, IT dosing, and underlies its ongoing clinical development for AS.
Graphical Abstract
Graphical Abstract.
Introduction
Angelman syndrome (AS) is a rare neurodevelopmental disorder characterized by lack of speech, epilepsy, developmental and motor skills delays, sleep disturbances, and cognitive impairment [1]. AS is caused by the loss of function from the maternally inherited allele of the gene encoding Ubiquitin Protein Ligase E3A (UBE3A) [2–4]. UBE3A is expressed biallelically in all tissues of the body, except neurons. In neurons, UBE3A shows imprinted expression; it is expressed almost exclusively from the maternally inherited allele. UBE3A has functions in both neuronal development and physiology. Overexpression contributes to the pathophysiology of another neurodevelopmental disorder, chromosome 15q11–q13 duplication syndrome [5–8]. Furthermore, two expressed copies of UBE3A may underlie increased susceptibility and severity of psychiatric illness in Prader–Willi syndrome [8], and an extra copy of UBE3A led to developmental delay and neuropsychiatric phenotypes in a multigenerational family. The need for tight regulation of UBE3A may be due to the prescribed roles in fundamental neuronal and cellular function, including neuronal synaptic plasticity, proteasome and tRNA pathways and recently, in controlling virus-derived GAG proteins [9–12] (https://www.angelman-proteome-project.org/).
Imprinted expression of UBE3A is achieved via the expression of an antisense transcript (UBE3A-ATS) only in neurons, which was discovered over two decades ago [13]. Elegant work later demonstrated that Ube3a-ats represses Ube3a in cis on the paternally inherited chromosome, providing the rationale that reduction of human UBE3A-ATS could be a therapeutic strategy for AS [14–17]. Subsequently, it was demonstrated that inserting a transcriptional terminator into Ube3a-ats in mice could activate expression of Ube3a from the paternal chromosome and could revert disease-related symptoms in the AS mouse [18]. Multiple therapeutic approaches using small molecules to reduce transcription of Ube3a-ats or antisense oligonucleotides (ASOs) to recruit RNase H to cleave Ube3a-ats have demonstrated proof-of-concept in mice with established therapeutic modalities, and revealed that messenger RNA (mRNA) and protein levels comparable to maternal Ube3a can be achieved [11, 15, 18–20]. Furthermore, genetic reinstatement of Ube3a in AS mouse models at different timepoints during postnatal development using a tamoxifen-inducible Cre knock-in paradigm or via ASOs, has demonstrated partial rescue of phenotypes, further supporting this therapeutic approach [20–22]. These studies with AS animal models have been instrumental in supporting therapeutic strategies that focus on increasing neuronal UBE3A expression levels by targeting UBE3A-ATS in humans [22]. Notably, an ASO now in clinical development was reported to efficiently downregulate UBE3A-ATS, but unfortunately did not robustly upregulate UBE3A protein in both neurons and nonhuman primate (NHP) brains [23]. Therefore, molecules with the ability to potently unsilence the paternal locus resulting in a homogeneous increase in UBE3A protein across relevant neuron types and brain regions are very much needed to progress in the development of treatments for AS. We describe the discovery and characterization of rugonersen, identified by screening AS patient-derived neurons for disease relevant unsilencing, which is now in clinical development [24] (NCT04428281). Lack of sequence conservation of UBE3A-ATS between rodents versus humans has hampered translational research. To overcome this challenge, we leveraged induced pluripotent stem cell (iPSC)-derived neurons from a human AS patient in an in vitro screening campaign to identify potent locked nucleic acid (LNA) ASOs targeting UBE3A-ATS. Rugonersen was identified to have excellent in vitro and in vivo potency and selectivity and was shown to be well tolerated in mice and monkeys. Rugonersen produced a long-lasting reduction of UBE3A-ATS in monkey brains, concomitant with upregulation of paternal UBE3A mRNA and protein. The data were used to derive a translational pharmacokinetic/pharmacodynamic (PK/PD) model, which enabled human interventional studies. Our data highlight the utility of human patient iPSC-derived neurons for drug screening and report key preclinical data of the LNA ASO rugonersen for the treatment of AS with potential best in disease properties.
Materials and methods
Ethics statement
All animal experiments conducted for this study were reviewed and approved by the Institutional Animal Care and Use Committee of the testing facility where the study was performed. The care and use of animals were conducted in accordance with the USA National Research Council and the Canadian Council on Animal Care (CCAC) guidelines. All pivotal safety studies were conducted in compliance with Good Laboratory Practice regulations, and the toxicology programs for the compounds were conducted according to International Conference on Harmonization guidelines. The studies were performed at contract research organizations accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International.
The studies involving human subjects were conducted in accordance with the Declaration of Helsinki and the International Conference on Harmonisation guidelines for Good Clinical Practice. The protocols were approved by the ethics committee of the participating institution, and all subjects provided written informed consent before participation in any study procedures. The studies are registered with ClinicalTrials.gov under the identifiers NCT01893437, NCT02164266, and NCT02699372.
Human neuronal cell culture, human and cynomolgus monkey iPSC-derived neuronal progenitor cell culture, and LNA treatment
Neuronal progenitor cells (NPCs) derived from neurotypical hiPSC lines 1302 [1302 a-C (ND 140321, L500)] and 902 as well as AS patient iPSC lines 1501 [1501a-A(A) (EdS150402)] and 1301 were previously described and characterized [9, 11]. The neuronal differentiation protocol was previously described [11, 25, 26]. In brief, human iPSC-derived NPCs were dissociated with trypsin–ethylenediaminetetraacetic acid (EDTA) 0.05% (Thermo Scientific), plated on polyornithine-/laminin-coated (PL) flasks/dishes at 10k cells/cm2 in SFA medium, and cultured for 1 week with medium replacement after 3–4 days. The resulting progenitors were dissociated with trypsin–EDTA 0.05%, plated on PL flasks/dishes at 50k cells/cm2 in BGAA medium and differentiated for 6 weeks with medium replacement every 3–4 days. UBE3A-ATS- or nontargeting ASO was added directly to the cell culture media and incubated at indicated concentration immediately following the 6 weeks of differentiation, unless otherwise noted.
For the human embryonic stem cell (hESC)-derived AS neuronal model, the H9 hESC derivative, H9Δmat15q_1 cell line was further edited to include a CRISPR-mediated deletion of the chromatin boundary elements IPW and PWAR1 as previously described (Supplementary Fig. 1). After generating the boundary deletion, an inducible neurogenin 2 (NGN2) construct was knocked into the AAVS1 safe harbor locus using constructs previously described [27]. The resulting cells can be induced to become neurons with doxycycline and fully imprint the sole paternal allele of UBE3A upon conversion. ASO treatment was conducted by adding a single dose of ASO at the indicated concentration on day 14 post-neuronal conversion. Neurons were harvested one week later for RNA and quantitative reverse transcription real-time polymerase chain reaction (qRT-PCR) as described below.
NPCs derived from cynomolgus macaque (Cips_BC671A_NSC line, generated at Roche) were differentiated in NPC differentiation medium [28]. Differentiating neurons were incubated at indicated concentration from day 15 with ASO. Cells were lysed for qRT-PCR analysis on day 20.
Generation of human/cynomolgus monkey cross-specific ASO and screen
Based on previous characterization of the regulatory relationship between UBE3A-ATS and UBE3A mRNA [20] we designed ∼2500 RNase H-recruiting gapmer ASOs targeting different sites of the UBE3A-ATS transcript between the last annotated small nucleolar RNA (snoRNA), SNORD109B, and the 3′end of UBE3A-ATS (chr15:25 280 119–25 414 623; GRCh38/hg38) as previously described by [29]. To allow for later in vivo characterization of ASOs in NHPs all ASOs were targeting sequences conserved between human and cynomolgus monkey (Fig. 1A). Conservation was evaluated by aligning the target sites of all possible 20-mer oligonucleotides designed against the aforementioned human region to the corresponding regions in mouse and cynomolgus monkey (Fig. 1A). The region surrounding the target site of rugonersen was specifically extracted from human and cynomolgus monkey sequences and aligned with the msa package in R using the ClustalW method (Fig. 2A).
Figure 1.
Human AS patient derived neuron screen identifies rugonersen as a UBE3A-ATS targeting ASO that upregulates UBE3A. (A) Human UBE3A-ATS and UBE3A loci from SNORD109B to the transcriptional start site of UBE3A (chr15:25 270 000–25 439 008; GRCh38/hg38 build). Colored bars denote the search space for 20-mer ASOs designed to target this region in human (gray), and if their target site is present in the corresponding region of cynomolgus monkey (cyan), or mouse (purple). The cyan range indicates the search space used. (B) RT-qPCR analysis for UBE3A-ATS (left) or UBE3A sense transcript (right) normalized to housekeeping gene PPIA in control and AS iPSC-derived neuronal progenitor cells (NPCs), and at days 21, 42, and 56 of differentiation. Two controls and two AS deletion lines were used. Each data point represents an independent neuronal differentiation (n = 4). (C) Scheme for large screen of AS human iPSC-derived neurons with ASOs targeting UBE3A-ATS. (D) Results of the ASO screen correlating UBE3A mRNA (y-axis) and UBE3A-ATS (x-axis) in human iPSC-derived AS neurons after incubation with ASOs in the cell culture media. Levels at indicated concentrations measured by RT-qPCR relative to GAPDH and normalized to phosphate buffered saline (PBS) treated cells from two separate ASO screens in biological duplicates with a coefficient of variation < 30%. The effect of rugonersen is highlighted in red at 7.5 μM (red dot).
Figure 2.
Rugonersen is a highly potent and selective molecule targeting UBE3A-ATS and unsilencing UBE3A in human and cynomolgus monkey neurons. (A) Sequence alignment of the region surrounding the binding site of rugonersen (box, magenta) in human and cynomolgus monkey corresponding to the genomic positions chr15:25 308 946–25 309 225 (GRCh38/hg38 build) and chr7: 3314 713–3314 992 (macFas5). The nucleotide sequence and chemistry of rugonersen are shown at the bottom. Diamonds indicate LNA-modified bases, "E" indicate methyl cytosines, and small “x” indicate phosphorothioate (PS) linkages. (B) Immunostaining with anti-UBE3A (green) and anti-MAP2 (magenta) antibodies, in control 1302 and AS del. 1501 lines. After 56 days of differentiation, cell lines were left untreated (Untr.) or treated with 1 μM nontargeting ASO (NT) or 1 μM rugonersen for 7 days (scale bar: 50 μm). (C) Dose-response curves to assess pharmacological properties of rugonersen in vitro using human (derived from a neurotypical (Human NT) individual, boxes) and cynomolgus monkey (Cyno, wild-type animals, disks) iPSC-derived neurons (UBE3A-ATS, in red; UBE3A mRNA, in yellow; UBE3A protein, in blue). (D) Dose-response curves to assess pharmacological properties of rugonersen in vitro using iPSC-derived neurons derived from an AS individual (UBE3A-ATS RNA, disks; UBE3A mRNA, triangles; UBE3A protein, squares). (E) Single reads from the stranded RNAseq on the UBE3A gene locus and the up-stream loci in control and 30 μM rugonersen-treated iPSC-derived neurons. Coverage is represented in blue for the plus strand that encodes UBE3A-ATS and minus strand that encodes UBE3A mRNA. (F) Selective UBE3A mRNA upregulation in rugonersen-treated iPSC-derived neurons. Volcano plot showing the genes dysregulated by 30 μM rugonersen treatment for 48 h. The red lines show the thresholds applied for log2-fold-change and adjusted P-value to identify differentially expressed genes. A single transcript, UBE3A (red dot), was significantly upregulated upon treatment. All ASO treatment was carried out via free uptake in the cell culture media.
While optimizing for ASO screening (Fig. 1A–D), in an effort to make the neuronal cultures more homogenous and reproducible, AS patient 21 day differentiated cultures were replated. On day 21 medium was aspirated and cells were washed with PBS-/-, prior to addition of 6-ml pre-warmed Accutase/T75 flask and incubated for 5 min at room temperature (RT). Cells were resuspended in 6 ml of NDM + BGAA + Laminin medium and centrifuged in a 15-ml Falcon tube at 1100 rpm for 4 min. Supernatant was removed and cells were resuspended in 10 ml of NDM + BGAA + laminin medium, counted and seeded in the same medium with the addition of 10-μM ROCK inhibitor. Cells were seeded with 200 000 cells/cm2 in 96-well plates. On day 28 ASOs were added (final concentration of 0.2, 1, or 7.5 μM) and RNA was harvested (RNeasy mini kit [Qiagen, Cat#: 74106]) after 5 days of treatment. The RNA levels of UBE3A-ATS (Assay ID: Hs01372957_m1, FAM, Thermo Fisher). UBE3A mRNA (Assay ID: Hs00166580_m1, FAM, Thermo Fisher), and GAPDH (housekeeping reference gene; Assay ID: 1_Hu GAPDH VIC 4325792, Thermo Fisher) were evaluated in duplicates by qRT-PCR according to the manufacturer’s instructions. The expression of UBE3A mRNA and UBE3A-ATS were normalized to GAPDH expression from the same well and shown relative to the mean of the PBS samples represented at each qPCR plate. Values with a coefficient of variance below 30% for UBE3A were plotted in Fig. 1D.
Generation of mouse specific tool compound
ASOs (N = 192) were designed and synthesized as previously described [30]. All ASOs were complementary to the minus strand of chromosome 7 in a 34-kb region (chr7: 59 307 929–59 341 825) downstream of Ube3a which is partly overlapping with the annotated transcript, Snhg14 (Ube3a-ats). The ASOs were screened at a single concentration for their ability to knock-down Ube3-ats in embryonic (E14-15) mouse primary cortical cultures (data not shown). The potency of the most efficacious compounds was determined by generating concentration response curves in primary cortical cultures prepared from Ube3am-/p+ P1 pups measuring both Ube3a-ats and Ube3a mRNA expression by qPCR. This identified the ASO, RTR26183, AACTcatacaccaatTA and RTR26235, 5′-TTAcatccatactCCT-3′, both with fully modified PS backbone. Capital letters are LNA modifications and lower-case letters are DNA. All LNA-C nucleotides contain the 5-methylcytosine modification.
Neuronal cellular readouts: mRNA and protein
qPCR analysis for determination of UBE3A mRNA and UBE3A-ATSin vitro
RNA purification from both iPSC and primary mouse cortical cultures were done using RNeasy 96 Kit (Qiagen) according to the manufacturer’s protocol. The purified RNA was diluted in water to a concentration suitable for the one-step RT-PCR reaction (XLT-Onestep, Quanta). The RT-PCR reaction was carried out on a ViiA 7 384-well Real-Time PCR System (Thermo Fisher) running a standard 40 cycle program according to the manufacturer protocol. For quantification of UBE3A mRNA, UBE3A-ATS, and GAPDH in human iPSC-derived neurons, we used Taqman assays Hs00166580_m1 (UBE3A), Hs01372957_m1 (UBE3A-ATS), and Hs 4325792 (GAPDH) from Thermo Fisher. For quantification of Ube3a mRNA, Ube3a-ats, and Gapdh in mouse primary cortical cultures we used Mm02580987_m1 (Ube3a), Mm03455899_m1 (Ube3a-ats), and Mm99999915_g1 (Gapdh) all from Thermo Fisher.
All reactions were carried out in duplicates and on each qPCR plate a standard curve was included with two-fold serial dilution of RNA from PBS treated control wells. Using the QuantStudioTM software (Applied Biosystems) RNA quantities were calculated based on the standard curve. Quantities were then normalized to the calculated quantity for the housekeeping gene (GAPDH) in the same well. Hence, Relative Target Quantity = QUANTITY_target gene/QUANTITY_housekeeping gene. The RNA knock-down level was calculated for each well by dividing with the median Quantity of all PBS-treated wells (N = 14) on the same plate. Normalized Target Quantity = (Relative Target Quantity/[median] Relative Target quantity_PBS_Wells). This way expression level data of UBE3A mRNA and UBE3A-ATS are shown as percentage of PBS-treated wells.
For concentration response experiments, cultures and oligonucleotide addition were carried out as described above. Compounds were added to the cells in an 8 or 10-step, 
-log serial dilution. Concentration response curves were fitted using Graphpad Prism 7, using a four-parameter sigmoidal dose-response model. Expression values from PBS treated cells were included in the curve fitting, by setting the PBS treatment to fixed concentration of a full log10 below the lowest tested ASO concentration.
AlphaLISA assay for determination of UBE3A protein levels in vitro
hiPSC-derived neurons were cultured in 96-well plates. Medium was aspirated, wash carefully twice with PBS, add 50 μl of lysis buffer/well, incubate on ice for 5 min, mix the samples by pipetting to ensure good lysis of the cells, proceed with analysis or store at −80°C.
Lysis buffer: in 20 mM Tris–HCl, pH 8, 137-mM NaCl, 10% Glycerol Acros Ref. 184695000, 1% NP-40 Sigma Ref. 74385, 2-mM EDTA Gibco Ref. 15575–038, phosphatase inhibitors (PhosSTOP, Roche Ref. 04906837001), protease inhibitors cocktail (Complete, Roche Ref. 46931132001)
The standard curve was used for normalization of sample values and was included in each assay plate. The curve included 11 concentrations with 1:2 dilution steps ranging from 1.25 to 0.0012 μg/ml. For the standard curve the recombinant protein E6AP/UBE3A, Boston Biochem Cat.# E3-230 was used.
For each sample the assay reagent mix was prepared, containing: 5 μl of sample (cell lysate or standard curve), 10 μl of biotin conjugated antibody UBE3A-1/10a (in house AB) to a final concentration of 1 nM, 10 μl of AlphaLISA beads-conjugated antibody UBE3A-1/4b (in house ab) to a final concentration of 10 μl/ml, 25 μl of Streptavidin AlphaLISA donor beads to a final concentration of 40 μg/ml. Samples were prepared in a 384-well plate (OptiPlate, Perkin Elmer Cat. #6007290). After preparation, the plates were centrifuged at 1 200 rpm for 2 min at RT, the plates sealed and incubated for 30 min at RT. The plates were read using the instrument SpectraMax (Molecular Devices). Specific details of the stocks of conjugated antibodies used were reported in each original data file.
Whole transcriptome sequencing from hiPSC neurons
Total RNA was isolated from iPSC-derived neurons in two batches using the RNEasy Mini Kit (Qiagen) according to the manufacturer’s protocol. An on-column DNAse digestion was included to deplete genomic DNA. Three hundred and fifty nanograms of total RNA per sample was used as input for library preparation. Ribosomal RNA (rRNA) was depleted using the Ribo-Zero Gold rRNA Removal Kit (Illumina). rRNA-depleted RNA was further processed into sequencing libraries using the TruSeq Stranded Total RNA kit (Illumina) according to manufacturer’s instructions. Cluster generation was performed on the cBot instrument and paired-end sequencing reads were subsequently generated on a HiSeq4000 instrument.
To estimate gene expression levels, paired-end RNAseq reads were mapped onto the human genome (hg19) by using the short read aligner GSNAP version 2017-05-08 [Wu et al., (2005, 2010)] with default alignment parameters and allowing the program to detect novel splice. Mapped reads for all RefSeq transcript variants of a gene were combined into a single value, read counts per gene, by applying SAMtools version 1.5 [31] and customized in-house tools. Subsequently, read counts were normalized by sequencing library size and gene length according to [32], denoted as RPKMs (number of mapped reads per kilobase transcript per million sequenced reads). Prior to differential gene expression analysis (LNA versus vehicle, n = 4 each), a negative binomial regression model was derived to correct for potential confounding factors with the inclusion of the following covariates: lane, percent mapped reads, RNA integrity number, and an estimated iPSC differentiation to neuron score based on the derived Fantom5 primary cells genesets information. The implementation was conducted in R (version 3.4.0) using the DESeq2 package (version 1.16.0).
Mouse intra-cerebroventricular injections
In both wild-type (WT) and AS Ube3am-/p+ mice adult mice (10–12 weeks of age), a single 150 μg intra-cerebroventricular (ICV) dose of RTR26183 was administered to mice by ICV injection. Brain regions (hippocampus, striatum, and cortex) were harvested 2 weeks post-injection.
Ube3am-/p+ and Ube3am+/p+ animals were genotyped as previously described [30]. ICV injections were performed using a Hamilton microsyringe fitted with a 27 or 30-gauge needle, according to the method of Haley and McCormick [33]. The needle was equipped with a polyethylene guard at 2.5 mm from the tip in order to limit its penetration into the brain. Mice were anesthetized using isoflurane anesthetic (1.5%–4%). The needle tip was then inserted through the scalp and the skull into the right lateral ventricle, about 1-mm lateral and 1-mm caudal to bregma. Oligomer was given in a concentration of 20–40 mg/ml in a volume of 5 μl in 0.9% saline. The needle was left in place for 10 s before removal. This procedure required no surgery or incision. Animals were warmed on heating pads until they recovered from the procedure. Brain tissue (right, frontal cortical region) was collected on dry ice or RNAlater for drug concentration analysis, Ube3a-ats, Ube3a, and Gapdh qPCR and protein measurement.
NHP ASO dosing
Subjects were male cynomolgus monkeys weighing 2.5–3.4 kg at the start of the study. All animals got a polyurethane catheter implanted in the lumbar intrathecal (IT) space which was connected to a subcutaneous access port (Access Technologies) to allow for injection into the IT space. Catheter placement at the lumbar site was confirmed by fluoroscopy. A contrasting agent (Omnipaque 240®, up to 2 ml) was injected into the catheter, and a dorsal image of the catheter was taken. Patency checks involving both injection of 0.2 ml of saline into lumbar ports and collection of CSF were performed weekly until completion of the study, except when dosing or CSF collections are scheduled. The reference or test item/article was administered to the appropriate animals using a syringe and needle via the subcutaneous access port connected to a catheter implanted in the IT space at the lumbar level. Selected animals were dosed by a single dose on day 1 (all groups) or two doses on days 1 and 15 (groups 7 and 8 only). Animals were dosed in the prone position and maintained in the prone position for at least 30 min after the completion of dosing. Prior to dose administration, 1 ml of CSF was withdrawn; the first 0.2 ml of CSF was discarded. The dose volume for each animal was fixed at 1 ml and was given over 3 min. Following dose administration, the catheter was flushed with 0.5 ml/kg of artificial CSF given over a target of 1.5 min. Blood was collected by femoral venipuncture. Animals were fasted overnight before blood sampling. CSF samples were collected from the lumbar access port of all animals. The first 0.2 ml of CSF were discarded. After sample collection, the catheter was flushed with 0.2 ml of saline. Sample was centrifuged in a centrifuge set to maintain 4°C, for 5 min at 2000 × g within 1 h of collection, and the supernatant was harvested and stored at −80°C. All protocols and procedures involving the care and use of animals were reviewed and approved by the Institutional Animal Care and Use Committee of the testing facility where the study was conducted. The care and the use of animals were conducted according to the USA National Research Council and the CCAC.
Tissue readouts from in vivo experiments
Tissue exposure analysis
At scheduled necropsy, tissue samples were dissected on a chilled surface, weighed and snap frozen in liquid nitrogen, kept on dry ice and then stored at −70°C or below. Freshly frozen and weighed tissue samples (50–80 mg) were homogenized in 2 ml Precellys tubes (Bertin Instruments) containing ceramic beads and 600 μl MagNaPure LC RNA Isolation Tissue buffer (Roche Life Science). To measure tissue exposure of ASOs enzyme-linked immunosorbent assay (ELISA) was used for determination of ASO in tissue extract solutions using a biotinylated ASO oligonucleotide as capture and a digoxigenin conjugated ASO oligonucleotide as detection probe. The principle in this method is to hybridize the ASO probe to the capture and detection probes and bind the assembled complex to a streptavidin-coated ELISA plate. The hybridized complex bound on the ELISA plate was visualized by adding an anti-Digoxigenin-Alkaline Phosphatase (AP)-Fab fragment and subsequently Blue Phos Substrate for color development. The color development is measured spectrophotometrically at 615 nm. This assay can measure oligonucleotide concentrations from about 20.000 to below 10 ng/l with a color development time between 20 and 45 min. Usually, color development times of 30–45 min give the best results. Most optimized oligonucleotide ELISA assays will be able to detect oligonucleotide concentrations of 10 ng/l and will have a quantification limit (5× background signal) of 78–625 ng/l. The ELISA analyses were run as a manual ELISA procedure by the following procedure. The samples were diluted in a 5× saline-sodium citrate with tween 20 (SSCT) buffer. Dilution factors ranged from five-fold (low concentration plasma) to 10.000-fold dilutions for CSF early time points. Appropriate standards matching sample matrix and dilution factors were run on every plate. Add samples and standards to a dilution plate in the desired setup and make a dilution series. Three hundred-microliter sample/standard plus capture-detection solution is added to the first wells and 150 μl capture-detection solution in the remaining wells. Make a two-fold dilution series of standards and samples by transferring 150 μl liquid sequentially. Keep 2–4 wells for blanks (capture-detection solution only). A two-fold sample dilution series of at least six wells is recommended for optimal results. Incubate on the dilution plate for 30 min at RT. One hundred microliters of liquid is transferred from the dilution plate to a streptavidin plate. The plate is incubated for 1 h at RT with gentle agitation (plate shaker). The wells are aspirated and washed three times with 300 μl of 2× SSCT buffer. Hundred microliters of anti-DIG-AP diluted 1:4000 in phosphate-buffered saline with tween 20 (PBST, made on the same day) is added to each well and incubated for 1 h at RT under gentle agitation. The wells are aspirated and washed three times with 300 μl of 2× SSCT buffer. Add 100 μl of substrate (AP) solution (freshly prepared) to each well. The intensity of the color is measured spectrophotometrically at 615 nm after a 30 min incubation with gentle agitation. For the manual ELISA procedure plates are transferred directly to the reader and read every 5 min from t = O to t = 45, with shaking before each reading.
RNA analysis
Freshly frozen and weighed tissue samples (15–50 mg) were homogenized in 2 ml Precellys tube (Bertin Instruments) containing ceramic beads and 600 μl MagNaPure LC RNA Isolation Tissue buffer (Roche Life Science). The homogenate (400 μl) was then transferred to a MagNaPure 96 Processing Cartridge and RNA was purified using the MagNa Pure 96 with the kit Cellular RNA Large Volume Kit (05467535001) and using the protocol “RNA Tissue FF Standard LV 3.1″ (Roche Life Science). The remaining lysate was stored for later analysis of ASO contend (see below).
RNA concentration of all samples was determined using the Eon Microplate Spectrophotometer (BioTek Instruments). Based on these concentrations, samples were diluted to 50 ng/μl and a few random samples were checked for RNA quality using a Bianalyzer (Agilent). The purified RNA was further diluted between 1:10 and 1:100 (e.g. 5 μl RNA + 95 μl H2O) into a new RNA dilution plate (96-well PCR quality plate; Thermo Scientific #AB0900). The final in-well concentration of the diluted RNA was in the range of 0.2–5 ng/μl.
The RT-PCR reaction was carried out as a one-step RT-PCR reaction (XLT-Onestep, Quanta) on a ViiA 7 384-well Real-Time PCR System (Thermo Fisher) running a standard 40 cycle program according to the manufacturer protocol. All reactions were carried out in duplicates and all samples from the same tissue were kept on the same qPCR plate. On each plate a standard curve was included with two-fold serial dilution of RNA from the same tissue purified from saline treated control animals. Using the Quantstudio software (Applied Biosystems) RNA quantities were calculated based on the standard curve.
All reactions were carried out in duplicates and on each qPCR plate a standard curve was included with two-fold serial dilution of purified RNA from treated control wells. Using the Quantstudio software (Applied Biosystems) RNA quantities were calculated based on the standard curve.
For analysis of cynomolgus monkey tissue several housekeeping genes (GAPDH, POLR3F, UBC, and YWHAZ) were quantified from each sample and the three most stable genes were used for normalization in each tissue. The stability of HK genes was calculated using the method published by Vandesompele et al. [34]. For assays, primers and probes used for housekeeping genes and target genes we used the following assays: UBE3A mRNA (Hs001666580_ml, Invitrogen), GAPDH (Mf04392546_g1, Invitrogen), POLR3F (Mf + 2860939_m1, Invitrogen), YWHAZ (Mf029302410_m1, Invitrogen), UBC (Mf02798368_m1, Invitrogen), and for UBE3A-ATS a custom assay from Integrated DNA technologies, Forward primer: 5′-CCA TCT CTG ATA AGG ATG ATT GAG G-3′, Reverse primer: 5′-GTT CAC AGG AGA CCA AAC AGA TA-3′ and a Zen™ dual quenched 5′-labeled PCR probe with the sequence: 5′-/56-FAM/TTT GGC TTG/zen/TTG ACA CCA GCA CAG /3IABkFQ/-3′
For analysis of RNA expression levels in mice we used following Taqman assays: Ube3a mRNA (Mm02580987_m1), Ube3a-ats (Mm03455899_m1), and Gapdh (4351309) all from Thermo Fisher. Quantities of UBE3A mRNA and UBE3A-ATS were normalized to quantities of Gapdh in the same sample.
Allele specific digital qPCR assay
From the available cohort of cynomolgus monkeys at Charles River Laboratories, Montreal, blood samples were drawn, and DNA was extracted (Genewiz). Prior to paired end sequencing on an Illumina Hiseq platform, DNA from a 1.6-Mb region containing the UBE3A and UBE3A-ATS loci was enriched using Roche–Nimblegen gene enrichment library, all carried out at Genewiz. This identified a single nucleotide polymorphism (SNP) in an exonic region of UBE3A, chr7: 3.339.686C → A. A total of 12 heterozygous UBE3A SNP carriers were used and assigned to the saline group (N = 3) as well as d15 (N = 3) and d29 (N = 6).
A SNP digital PCR assay was set-up using Primetime probes, (Integrated DNA Technologies) covering the C→A SNP in position 3359 686 of chromosome 7. For detection of the C-allele we used a 5′FAM labeled probe (5′-CcgTCGTCtttTga-3′) (uppercase: locked nucleic acids, lower case: DNA) with a 3′ IowaBlack FQ quencher. For the A-allele we used a 5′HEX labeled probe (5′-atCCgTCtTctttTga-3′) with 3′ Iowa Black FQ quencher. Amplification primers were place in exon3 and -4 of the cynomolgus monkey transcript AB179374, using forward primer (5′-GCTGGTTGTGGAGGAAATCT-3′) and reverse primer (5′GTAAATAGCCAGACCCAGTACTAT-3′). Droplets were generated on a Bio-Rad Automated Droplet generator, using the manufacturer’s protocol. Following droplet generation, RT-PCR was run on a Bio-Rad C1000 touch thermal cycler PCR machine, using the following program: 45°C 10 min, 40 cycles of 94°C for 30 s and 55°C for 30 s, 98°C for 10 min. Finally, droplets were analyzed on a Bio-Rad QX200tm droplet reader and data analyzed using the Quantsoft Pro software. Linearity and specificity of the assay was confirmed by mixing different ratios of gBlocks/synthetic DNA (Integrated DNA Technologies) covering the PCR amplicon and carrying either the C- or the A-allele.
To ascertain the parental origin of the two alleles in each animal we used the allele specific quantification from the midbrain. In midbrain, one allele was clearly dominating over the other allele also in treated animals. Midbrain is known to be a low exposure tissue for ASOs, also confirmed from our exposure data (Fig. 4B); moreover the qPCR quantification of UBE3A-ATS (Fig. 4C) confirmed that low exposure resulted in limited knock-down of UBE3A-ATS. This is highly indicative that in midbrain, the maternal imprinting is largely preserved due to low exposure and poor UBE3A-ATS knock-down (Fig. 4B), and consequently the maternal allele remains the most dominant allele after treatment with rugonersen. Moreover, in the absolute quantification of UBE3A mRNA expression there was no indication of exaggerated pharmacology in the midbrain. Hence, we could use the samples from the midbrain with least UBE3A-ATS knock-down to determine which allele in a given animal was of maternal (most expressed allele) and paternal (least expressed allele) origin.
Figure 4.
Design and analysis of two NHP PK/PD studies. (A) Left panel shows the schematic for NHP study 1: Saline, a single dose of 24 mg or two doses of 16-mg 2 weeks apart were administered intrathecally. Sacrifice occurred 8, 15, 29, 57, and 85 days after the last dose. Nine brain regions were analyzed for UBE3A-ATS and UBE3A mRNA and UBE3A protein concomitantly with ASO concentration. Right panel shows the schematic for NHP study 2: Saline or three doses of 4, 14, or 30 mg 29 days apart were administered intrathecally. Sacrifice occurred 64 and 120 days after the first dose. Six brain regions were analyzed for UBE3A-ATS and UBE3A mRNA and UBE3A protein concomitantly with ASO concentration. (B) Exposure dependent reduction of UBE3A-ATS from NHP studies 1 and 2 are shown on left and right panels, respectively. Measured values are shown as dots that are color coded depending on the tissues. The plain line represents the median model fit with the variability depicted as a shadow area (90% confidence interval around the median). (C) The relationship between UBE3A-ATS and UBE3A mRNA for both NHP studies across multiple brain regions. Different brain regions are shown with different colors. Squares and triangles represent NHP study 1 and 2, respectively. (D) Steep relationship between brain concentration of rugonersen and UBE3A mRNA expression. The plain line represents the median prediction of UBE3A mRNA upregulation versus concentration, EC50 = 580 nM.
UBE3A protein analysis by western blot
All cynomolgus monkey tissues were placed in Precellys tubes after dissection and shock frozen. On the day of sample preparation, 200 μl of sodium deoxycholate (SDC) lysis buffer were added to the tissues. The composition of the SDC lysis buffer for 30 ml was as follows: 300 mg of SDC, 85.99 mg of TCEP [Tris(2-carboxyethyl) phosphine hydrochloride], 112.21 mg of CAA (2-chloroacetamide), 3 ml of Tris (pH = 8.5) and three tablets each of the protease/phosphatase inhibitors PhosStop and Complete Mini (Roche). The tissues were then lysed in Precellys Homogenizer Bertin (program 6 × 20 s pulse, 6000 rpm) and placed on ice for 30 min. The supernatant was taken out and placed into 0.5-ml Eppendorf Protein Lobind tubes and boiled at 95°C for 5 min and then placed on ice. The supernatant was taken out and sonicated in Bioruptor (15 × 15 s) and centrifuged. Five microliters of the supernatant was used for protein measurement by BCA Pierce and the rest was divided into aliquots and stored at −80°C. In preparation for the western blot (WB) analysis 10 μl Loading buffer 4× and 4 μl NuPage Sample Reducing Agent 10× were added to 26 μl of each tissue sample. Water was then added to adjust to a final concentration of around 2.5 μg/μl of protein.
Ten microliters of tissue samples (25 μg of protein per well) were placed into each well of the WB gel (Bio-Rad Criterion TGX Stain free 4%–15% 18 wells; Cat. #5678085) and run in running buffer Tris-Glycine SDS (Invitrogen, Cat. #LC2675) under the following conditions: 60 V for 20 min and then 100 V for 100 min. The WB gels were analyzed with the software “Image Lab” on the Bio-Rad imager and proteins transferred onto a nitrocellulose membrane with the device iBlot2 NC Regular Stacks (Invitrogen, Cat. #IB23001; program P0: step 1 – 20V/1 min, step 2 – 23 V/4 min, step 3 – 25 V/2 min). The membranes were blocked during 30 min in TBS/0.1% Tween20/3% milk at RT followed by incubation and shaking overnight at 4°C in 10 ml TBS/0.1% Tween20/3% milk containing 10 μl rabbit anti E6AP_Bethyl A300-352A. After washing three times for 5–10 min with TBS/0.1% Tween20/3% milk at RT the membranes were incubated and shaken for 90 min at RT in 10 ml TBS/0.1% Tween20/3% milk including 2 μl goat antirabbit IgG (Jackson Laboratories, Cat #115-035-045) and washed four times for 5–10 min each with TBS/0.1% Tween20/3% milk at RT. For detection the membranes were shaken in 10 ml Lumi-light WB substrate (Roche, Cat. #12015200001) spiked in 1 ml SuperSignalTM West Femto Maximum Sensitivity Substrate (Thermo Fischer, Cat. #34096) for 1 min and then analyzed on the Bio-Rad imager.
Images of gels and blots were analyzed with the software “Image Lab”, which allowed to get a numerical value of the amount of protein in the gel or on the membrane. First, squares were drawn around one band per sample of the gel to select the total amount of protein on the gel. A second square was drawn next to the samples for subtraction of the background. The total protein values were called “gel”. The same was done for the nitrocellulose membrane and the UBE3A protein values were called “blot”. By calculating the “blot/gel” ratio (value on the blot per sample divided by the value on the gel) we got the proportion of UBE3A/total protein in the samples.
Selective Reaction Monitoring assay for UBE3A protein from cynomolgus monkey and mouse
One hundred micrograms of tissue lysates prepared in SDC buffer for WB was used for digestion for selected reaction monitoring (SRM)-based liquid chromatography-mass spectrometry (LC-MS) analysis. Samples were digested according to [35] with slight modifications. Briefly, SDC buffer containing samples were diluted in 4× in ddH2O and digested with 10 µg trypsin overnight at RT. Digested peptides were cleaned up in Sep-Pak C18 96-well plates. Cleaned peptides were dried in a speed vac (Thermo) for 4 h.
One microgram of digested peptide was analyzed using LC-MS as described in [9]. Skyline v4.2 was used to generate a target peptide list based on shotgun MS analysis of UBE3A protein using a shared peptide between mice, cynomolgus monkey, and human. Isotope‐labeled peptide, containing either L‐[U‐13C, U‐15N]R or L‐[U‐13C, U‐15N]K, corresponding to the unique target peptide (VDPLETELGVK) was synthesized (Thermo) and their sequences confirmed by LC‐MS/MS.
For MS analysis, digests were diluted to 200 ng/μl with 0.1% (v/v) formic acid, 2% (v/v) acetonitrile (ACN) containing the pooled isotope‐labeled peptides at a final concentration of ∼3 fmole/μl. SRM analyses were performed on an Ultimate RSLCnano LC coupled to a TSQ Quantiva triple quadrupole MS (Thermo Scientific). Samples (5 μl) were loaded at 3 μl/min for 6 min onto a 2 cm × 75 μm C18 trap column (Acclaim Pepmap 100, 3 μm, 300 Å, Thermo Scientific) in loading buffer [0.5% (v/v) formic acid, 2% (v/v) ACN]. Peptides were then resolved on a 50 cm × 75 μm C18 analytical column with integrated electrospray emitter heated to 40 °C (Easy‐SPRAY, 2 μm, 100 Å, Thermo Scientific) using the following gradient at a flow rate of 250 nl/min: 6 min, 98% buffer A (2% ACN, 0.1% formic acid), 2% buffer B (ACN + 0.1% formic acid); 90 min, 30% buffer B; 96 min, 60% buffer B; 98 min, 80% buffer B; 114 min, 80% buffer B; 115 min, 2% buffer B; 138 min, 2% buffer B. The TSQ Vantage was operated in 2-min retention time windows; cycle time, 1.5 s; spray voltage, 2600 V; collision gas pressure, 1.5-mTorr; Q1 and Q3 resolution, 0.7 full width at half maximum (FWHM); capillary temperature 240°C.
Skyline version 4.2 was used for automated peak integration. In a small number of cases, the peak selection was manually corrected based on the isotope‐labeled peptide elution time and transition ratio as well as the expected transition ratios based on the chromatogram library. A transition was removed from a peptide if obvious interference was identified (through comparison to the chromatogram library and the isotope‐labeled peptides). Post data acquisition, total precursor peak areas were exported, and further analysis was performed on Spotfire (Tibco). The endogenous peptide peak areas were corrected for using spiked-in heavy labeled peptides by applying a correction factor (median heavy peptide intensity/heavy peptide area per sample).
Mouse PK/PD modeling
The relationship between Ube3a-ats knock-down and concentration was investigated, as well as the relationship between the three PD markers. All markers were measured on the various brain regions (hippocampus, striatum, and cortex) and expressed as percentage of expression in comparison to vehicle-treated animals. The relationship between the brain concentrations and the Ube3a-ats knock-down was assumed to be direct, i.e. without delay, as well as the relationship between the three PD markers. To note, the relationship between the Ube3a mRNA and the UBE3A protein was assumed to be compound independent. Therefore, data obtained from RTR26183 and RTR26235 were analyzed together to have enough information to derive a relationship between the two markers. All the analyses were performed in GraphPad Prism® 8.4.2 using the nonlinear fit after log transformation of the data.
Cynomolgus monkey PK/PD modeling
Similarly, as in mouse PK/PD modeling, the three PD markers and their relationship to brain concentrations were investigated. All PD markers were measured on various brain regions and expressed as percentage of expression in comparison to vehicle treated animals. To note, in cynomolgus monkeys the maternal allele is expressed, in contrast to AS patients. Therefore, in cynomolgus monkeys UBE3A mRNA and protein are expressed and at maximal rugonersen effect expression is doubled.
In the NHP study I, the UBE3A-ATS knock-down is maximal for the entire study duration, in most brain regions except in the brainstem (midbrain, medulla and pons). The first time point in the study was at 7 days and potential delay in UBE3A-ATS knock-down could not be observed. In the NHP study II, sampling times were at day 64 or 120 preventing an earlier observation. UBE3A-ATS knock-down was therefore fitted against the measured concentrations with a direct sigmoid Imax response model on the data from both NHP studies [36]. The analysis of the UBE3A mRNA data is conducted on qPCR data from the midbrain and cortex regions, as various imprinting levels in other regions led to lower UBE3A mRNA level; see the ‘Results’ section on paternal and maternal UBE3A expression. In cortices, the UBE3A mRNA elevation is maximal from the first to the last observation; delay could not be observed (first observation at day 7). Therefore, a direct response model is used to fit the UBE3A mRNA elevation against the UBE3A-ATS knock-down values. Parameter estimation is performed using Monolix 2018R1 (The Monolix software, Analysis of mixed effects models, LIXOFT and INRIA, http://www.lixoft.com/) software using the linear approximation for the estimation of the Fisher Information Matrix.
Results
Rugonersen: an ASO targeting UBE3A-ATS with human/cynomolgus monkey cross reactivity
Human AS iPSC-derived neurons have been previously reported to imprint paternal UBE3A following maturation in vitro. To model a physiologically relevant neuronal network which reliably achieves this maturation, we differentiated AS iPSCs into mixed cultures consisting of glutamatergic and GABAergic neurons, as well as astrocytes [11, 25, 26]. We had previously described and characterized AS iPSC lines harboring deletions of the 15q11–13 locus and a point mutation in UBE3A [9, 11]. Over the course of neuronal differentiation, we observed increased expression of UBE3A-ATS in both control and AS patient lines concomitant with a decrease of UBE3A mRNA (Fig. 1B). UBE3A mRNA expression in AS neurons plateaued at their lowest levels by day 42 in culture. UBE3A protein levels were confirmed to be very low in the nuclei of microtubule-associated protein 2 (MAP2) positive neurons by immunocytochemistry (Fig. 2B). We concluded that our AS iPSC-derived neuronal model recapitulates the loss of UBE3A in disease and chose the day 42 time point for subsequent screening.
Screening for UBE3A-ATS targeting ASOs
In order to scale AS iPSC-derived neurons for screening, NPCs were expanded, replated in 96-well plates for maturation, and matured for ∼5 weeks (Fig. 1C). We designed ∼2500 LNA/DNA gapmers composed of LNA-modified nucleosides flanking a DNA gap, with PS backbones throughout [37]. These gapmers, which recruit RNase H to cleave UBE3A-ATS in a sequence-specific manner, were tiled across a ∼60 kb region between SNORD109B and the annotated 3′ end of UBE3A (chr15:25 280 119–25 414 623; GRCh38/hg38 build) (Fig. 1A and D).
UBE3A-ATS is poorly conserved between rodents and humans, therefore we only designed ASOs targeting sequences in UBE3A-ATS that were conserved between human and cynomolgus monkey. As such, the search space was minimally constrained while ensuring that drug metabolism and PK studies could be performed in at least one responder species. ASOs in this tiling library had no perfect complementarity to any other human transcript and had fewer than 50 matches to any human transcript when allowing one mismatch (Fig. 1A).
These sequences were chemically designed to include different patterns of LNAs in the flanks and were subsequently subjected to predictive analyses using machine-learning algorithms trained on in house data from in vitro and in vivo assays for cytotoxicity, hepatotoxicity and acute neurotoxicity [38, 39]. ASOs passing in silico predicted safety and specificity criteria were screened for their potential to upregulate UBE3A mRNA using AS patient iPSC-derived neurons with the assumption that increasing UBE3A would represent unsilencing of the paternal allele. This effort led to the identification of multiple ASOs targeting UBE3A-ATS, in regions both upstream of and overlapping with the UBE3A gene body. We selected ASOs that passed a cutoff of two-fold upregulation of UBE3A mRNA for further study (Fig. 1D). These data demonstrated a good correlation between UBE3A-ATS knock-down and UBE3A upregulation (Fig. 1D). Based on UBE3A upregulation and the most favorable scores in terms of off-target and safety predictions, we selected the molecule that ultimately became rugonersen, depicted in red, for further characterization (Fig. 1D).
Rugonersen potently and selectively unsilences paternal UBE3A in human and cynomolgus monkey neurons
Rugonersen targets a sequence conserved between human and cynomolgus monkey (Fig. 2A), making it ideal for translational research and development. We next sought to determine in vitro pharmacology in both human and cynomolgus monkey neurons. Immunocytochemical analysis demonstrated that MAP2 positive AS iPSC-derived neurons show reduced UBE3A immunoreactivity, which is restored by rugonersen (Fig. 2B).
Dose-response treatment of AS neurons with rugonersen revealed nanomolar potency for UBE3A-ATS mRNA downregulation (IC50 = 26.3 nM), UBE3A mRNA upregulation (EC50 = 15.4 nM), and UBE3A protein upregulation (EC50 = 24.8 nM) (Fig. 2C and D). Rugonersen showed similar potencies in neurons derived from neurotypical human iPSCs and cynomolgus monkey iPSCs (Fig. 2C and D).
An additional AS hESC model (H9ASΔI-P [23, 40]) engineered to allow inducible conversion into glutamatergic neurons, was used to confirm potencies and to benchmark rugonersen to two ASOs similar to those in clinical development [23, 40, 41]. Rugonersen was generally more potent than the other two ASOs, while the difference was particularly pronounced for UBE3A mRNA (UBE3A-ATS: IC50 = 12 nM for rugonersen versus 78 or 50 nM for reference ASO 1 and ASO 2, respectively; UBE3A mRNA: EC50= 122 nM for rugonersen versus 1.4 and 2.7 μM for reference ASO 1 and ASO 2, respectively) (Supplementary Fig. 1). In summary, these cross-primate in vitro neuronal results demonstrate that rugonersen-mediated upregulation of UBE3A mRNA translates to increased UBE3A protein levels in neurons derived from multiple AS patient neuronal models, neurotypical human iPSCs and cynomolgus monkey iPSCs (Fig. 2C and D).
Off-target analysis of rugonersen
To identify direct off-target RNAs downregulated by rugonersen, we performed RNAseq in AS iPSC-derived neurons treated with 30-μM rugonersen, ∼1000 times the UBE3A-ATS IC50, for 48 h, thus using a brief, but high exposure of rugonersen to focus on primary over secondary effects. Differential gene expression analysis comparing rugonersen and vehicle-treated cells demonstrated only a single transcript, UBE3A, was significantly (adj. P < 0.05) upregulated or downregulated by two-fold or more (Fig. 2F).
Since UBE3A-ATS is located at the 3′-most end of the SNHG14 transcript, which also encodes multiple other small RNAs, we next examined the whole transcriptome data for downregulation of other RNAs within SNHG14 by rugonersen. Consistent with the proposed mechanism of action of rugonersen, targeting the nascent RNA of UBE3A-ATS, the expression of the distal portions of SNHG14 (Fig. 1A), including introns that host SNORD109B and the last copies SNORD115, were reduced upon treatment (Fig. 2E, top panel). However, the portions of SNHG14, such as SNRPN, PWAR-SN, PWAR5, the SNORD116 cluster, and the 5′ portion of the SNORD115 cluster, showed comparable levels of expression as seen in the control samples (Fig. 2E, top panel).
Tool ASO enables characterization of PK/PD relationship using AS model mice
We next performed studies in WT and AS model mice to confirm in vivo effects for LNA/DNA gapmers targeting Ube3a-ats, and to understand the relationship between Ube3a-ats reduction and Ube3a mRNA and UBE3A protein expression to support pivotal translational studies in NHPs. To that end, we screened LNA/DNA gapmers targeting murine Ube3a-ats in mouse primary cortical cultures to identify a tool molecule for in vivo studies as previously described [21]. The screen identified both RTR26183 and RTR26235 as potent unsilencers of paternal Ube3a mRNA in primary cortical neuron cultures prepared from Ube3am-/p+ pups [42] as well as WT littermates (EC50 = 23 nM for RTR26183 and 38 nM for RTR26183, respectively). The targeting sequence of RTR26183 with respect to mouse genomic position is depicted in Fig. 3H.
Figure 3.
Single dose of a potent tool molecule, RTR26183, for two weeks exhibits sustained unsilencing of paternal locus in both WT and AS mouse model. Quantification of Ube3a-ats (A), Ube3a mRNA (B), and UBE3A protein (C) in WT and AS mouse brains upon treatment with a single 150-μg dose of RTR26183 or vehicle, ICV. (D–F) PKPD relationships between the PK and the different PD markers in AS mouse. (D) Exposure-response curve of RTR26183 reduction of Ube3a-ats in the AS mouse model. The plain line represents the Emax model. Emax = 174 nM, IC50 = 76 nM (E) PD relationship between Ube3a-ats reduction and Ube3a mRNA unsilencing. The plain line represents the Emax model, and the dotted lines represent the 95% confidence interval with an EC50 of 18.5%. (F) PD relationship between Ube3a mRNA elevation and UBE3A protein. The plain line represents the Emax model with EC50 of 72%. Ube3a mRNA and UBE3A protein data from experiments conducted with RTR26183 and RTR26235 have been analyzed together since the relationship between these two pharmacological biomarkers are independent of the compound. In panels (D)–(F) we highlight the PK and PD markers values in purple to elevate the protein to 50%, also depicted in the scheme in panel (G). (G) Quantitative relationship between Ube3a-ats knock-down, Ube3a mRNA, and UBE3A protein expression in mice. To elevate the protein to a 50% level versus saline treated group, nearly 90% reduction of Ube3a-ats is required. (H) Sequence of murine Ube3a-ats adjacent to the binding site of RTR26183 (magenta), corresponding to the genomic position chr7: 58 989 211–58 989 527 (mm39). Median (m) and standard deviation (σ) values are indicated for each group, and stars above horizontal lines indicate significant levels of corresponding groups comparison: ***for adjusted P-value <.001 and *for adjusted P-value <.05 (t-tests on log-transformed data, multiple comparison with false discovery rate approach).
In vivo effects of tool ASO in mice
A single 150-μg ICV dose of RTR26183 was well-tolerated and sufficient to reduce Ube3a-ats by ∼90% in both WT and AS Ube3am-/p+ adult mice in several investigated brain regions (hippocampus, striatum, and cortex; Fig. 3A). This equated to an in vivo potency in the nanomolar range (IC50= 75 nM; Fig. 3D). Ube3a mRNA was upregulated by 55% in WT mice, while paternal Ube3a mRNA was unsilenced to 72% of WT levels in Ube3am-/p+ mice (Fig. 3B). To confirm that the treatment effect translated to protein changes, UBE3A protein levels were quantified in AS mouse tissue via WB and SRM. Both methods confirmed a significant increase of UBE3A protein expression in the treatment group, in line with previous results (Fig. 3C) [20, 21].
PK/PD relationships in mouse model
From these murine data, PK/PD relationships were derived to link ASO brain exposure to Ube3a-ats, Ube3a mRNA and UBE3A protein levels from different brain regions including hippocampus, striatum, and cortex at 14 days post-injection in Ube3am-/p+ mice (Fig. 3D–F). To investigate the relationship between Ube3a mRNA and its protein, additional data from in vivo experiments with RTR26235 were added to increase the sample size and coverage of the PK/PD range. Correlating the reduction of Ube3a-ats with paternal Ube3a mRNA and protein elevation, we found that almost 90% Ube3a-ats reduction was required to increase Ube3a mRNA by 72%, which translated to 50% upregulation of UBE3A protein (Fig. 3D–F). This steep relationship was not completely in agreement with in vitro data and revealed that a large knock-down of Ube3a-ats is needed to allow protein re-expression in mice. These data bolstered confidence in the in vivo mode-of-action and informed the design of the NHP study needed for human dose prediction and subsequent translation of rugonersen to clinical development.
Rugonersen mediates widespread and sustained unsilencing of UBE3A in NHPs
Rugonersen was both potent and selective in in vitro studies using AS, neurotypical human, and cynomolgus monkey neurons, and in vivo potency was demonstrated with tool molecules in mice. As a next step towards clinical development, rugonersen was evaluated in cynomolgus monkeys using IT administration to assess pharmacological properties and safety.
In a first study, three male cynomolgus monkeys per group and time point (terminal sampling) were dosed via IT administration (ported catheter) with either saline, a single dose of 24-mg rugonersen, or two doses of 16 mg separated by 2 weeks (Fig. 4A). The doses were selected based on considerations of prior experience with tolerability of ASOs upon IT dosing in monkeys and the estimated dose requirement for full activation of the paternal UBE3A allele in relevant brain regions. Subgroups were sacrificed at different time points spanning from days 8 to 85 (animals dosed twice on days 1 and 15 were sacrificed on day 29) (Fig. 4A, left panel) and brains were harvested. Rugonersen concentrations in the tissues were quantified, along with UBE3A-ATS mRNA, UBE3A mRNA, and UBE3A protein. From these measurements, relationships between brain ASO concentrations, UBE3A-ATS reduction, and UBE3A mRNA and protein unsilencing were derived (Fig. 4B, left panel). These data were analyzed for multiple brain regions thought to be neurologically relevant for AS clinical phenotypes including cortex (frontal, occipital, parietal, temporal), hippocampus, and cerebellum along with spinal cord regions (lumbar, thoracic, cervical) (Figs 4 and 5 and Supplementary Fig. 3).
Figure 5.
Single dose of rugonersen has a long duration of action on paternal UBE3A reactivation in NHP brains after IT delivery. (A) WB showing on the right side UBE3A protein (band at ∼100 kDa) expression in cortical regions 57 and 85 days after single 24 mg dose of rugonersen and 8 days after vehicle dosing in NHP study 1 and on the left side total protein loaded, (B) Quantification of UBE3A-ATS, UBE3A mRNA, and UBE3A protein in cortical regions following a single 24-mg dose of rugonersen in NHP study 1. Squares indicate median values (also indicated with m), and standard deviations (σ) are shown with vertical segments. For protein values, we show median values and standard deviations for three time points grouped together: days 29, 57, and 85 because the effect is delayed for protein (n = 3 at each timepoint). (C) Paternal gene expression relative to maternal gene expression in brain tissues following treatment with saline or rugonersen. (D) Correlation between UBE3A protein (SRM) and ratio between paternal/maternal UBE3A mRNA expression in cortex (magenta) and spinal cord (green) after saline or rugonersen treatment.
Brain uptake and PKs in NHPs
IT administration of 24 mg rugonersen showed a high brain uptake. Seven days post-injection, the highest exposure was observed in cortical regions (6382 ± 2621 nM) and the lowest in deep brain regions (3774 ± 3193 nM; Supplementary Table 1). The data also suggested tissue ASO half-life of ∼6–7 weeks in most brain regions. The high brain uptake and long half-life provide ideal properties to enable a tolerable IT dosing regimen in the clinic. As expected, rugonersen is also distributed in plasma and peripheral organs such as the kidney and liver. Seven days after dosing, liver exposure (9854 nM) was about two-fold higher than in hippocampal regions (4050 nM), but the elimination kinetics were faster. Exposure in the kidney (37114 nM) was approximately four times higher than in liver, highlighting the main elimination pathway with IT dosing of rugonersen, which is consistent with IT dosing of other ASOs to NHPs (Supplementary Table 1 [43, 44]). In this context, it was important to recognize the absence of any clinical or histopathological evidence for toxicity to the liver or kidney (Supplementary Fig. 2).
Safety and tolerability in NHPs (Study 1)
Transient lack of coordination was observed in most animals receiving rugonersen at 24 mg. These changes correlated with transient neurological observations noted at 2 and/or 6 h post-dosing, which included decreased proprioceptive hindlimb positioning, decreased or absent tactile and/or visual hindlimb placing reaction, decreased or absent hindlimb flexor reflex and/or decreased muscle tone. These types of reactions are rather typical for intrathecally dosed ASOs and are generally not adverse [45]. Histopathological examination revealed test item related findings of neuronal vacuolation in the nervous system (Supplementary Fig. 2 [46]), vacuolated macrophages in spinal cord, near the injection site, and nearby lymph nodes, as well as basophilic granules in the kidneys. None of these were considered adverse based on their low severity and lack of associated degenerative/inflammatory changes or clinical pathological changes (e.g. renal function for kidney). All these safety-directed observations were consistent with those observed previously after administration of single-stranded oligonucleotides [46, 47].
UBE3A-ATS reduction and UBE3A unsilencing in NHPs (Study 1)
This study explored two dose levels, close to each other. However, as the monkeys were sampled at different time points, brain tissue concentrations ranged from single digit nanomolar to above 3000 nM allowing us to explore the relationship between UBE3A-ATS reduction and rugonersen exposure. Remarkably, UBE3A-ATS was reduced >95% in cortical areas relative to the saline controls and was maintained up to day 85 (last sacrifice, Fig. 5A). In deeper brain structures, such as the midbrain, exposure and UBE3A-ATS reduction were lower and more variable, resulting in ∼75% reduction at day 7 (Fig. 4B, left panel). In all brain regions analyzed, rugonersen exposure correlated with reduction of UBE3A-ATS (Fig. 4B). The calculated in vivo potency was similar in all tissues, and we found that UBE3A-ATS was reduced by 80% at ∼400 nM.
In cortical areas, strong reduction of UBE3A-ATS translated to approximately two-fold upregulation of UBE3A mRNA expression versus saline and was sustained throughout the 85 days of the experiment (Fig. 5A, lower panel). These results are consistent with full unsilencing of the paternal allele in this brain region in the presence of the functional maternal allele. In the midbrain, the UBE3A-ATS reduction and UBE3A mRNA upregulation were less pronounced compared to cortical regions, likely due to lower exposure and/or faster clearance in this region (Fig. 4C). Surprisingly, in other deep brain regions such as medulla and pons, the upregulation of UBE3A mRNA was weaker and more variable despite high exposure and strong decreases in UBE3A-ATS, especially shortly after IT injection on days 8 and 15 (Supplementary Fig. 3). In cortex and midbrain, we observed that at least 84% reduction of UBE3A-ATS was needed to achieve a 50% increase of UBE3A mRNA (Fig. 4C), like what was observed in the mouse. By integrating UBE3A-ATS reduction, UBE3A upregulation, and rugonersen concentration, we determined the relationship between rugonersen concentration and UBE3A mRNA expression to be an EC50 of 580 nM (Fig. 4D).
UBE3A protein, as quantified with both WB and SRM analysis, revealed that after some delay, levels were doubled in cortical regions. These levels were maintained up to day 85 (last sacrifice) (Fig. 5A), correlating well with the UBE3A mRNA expression in cortical regions (Fig. 5A, lower panel). We concluded that due to the favorable PK properties (a high exposure and long half-life), a single administration of rugonersen can fully reinstate paternally-derived UBE3A protein in NHP cortical regions for at least 12 weeks.
NHP Study 2: dose range exploration and confirmatory findings
In a second NHP study, rugonersen PK and PD responses were explored across a larger dose range. Rugonersen was given three times, every four weeks at either 4, 14, or 30 mg using IT delivery. Monkeys were sacrificed and tissue sampled either 64 or 120 days after the first administration (Fig. 4A, right panel). We then compared these results with the results of the first NHP study to deduce the PK/PD relationships (reference model). UBE3A-ATS reduction was plotted against the reference model simulations. Most of the observed values (90%) were within the simulated range of UBE3A-ATS reduction (Fig. 4B). Similarly, the relationship between UBE3A-ATS and UBE3A mRNA expression follows the same trend in both studies (Fig. 4C). This second data set confirmed target engagement and the relationship and dynamics of rugonersen to paternal unsilencing in NHPs.
Safety and tolerability in NHPs (Study 2)
In this study, no adverse findings that would have hindered human dosing were found at any dose level. Consistent with the first study described above, transient nonadverse clinical signs of slight tremor, circling, or uncoordinated movements with origin in the lower extremities were seen occasionally at animals dosed with 30 mg. In terms of time-to-occurrence, these were seen in close connection to the dosing procedure [48] and were considered typical with IT injection of antisense molecules. A test item-related sporadic absence of patellar reflex was observed 4 h after dosing for all dose groups. Microscopically, vacuolated macrophages were present in the Virchow–Robins space of small vessels, choroid plexus, and meninges of the brain, in animals administered 14 or 30 mg/animal, and in the lymph nodes of all groups, which regressed slightly after the 8-week recovery phase (in line with bioanalytical data on tissue elimination), nonadverse mononuclear immune cell infiltrates in meninges of brain and spinal cord, and vacuolation of the CA1 region of the hippocampus. Again, all these safety-directed observations were consistent with those observed previously after administration of single-stranded oligonucleotides [46, 47]. Consistent with the first study mentioned above, no test item related clinical chemistry or histopathological effects were seen in the liver and kidneys of the animals at any dose.
In a subsequent 39-week chronic toxicity study with 6-month recovery using the same dose levels and with once monthly dosing, the immune cell infiltrates were more pronounced up to moderate severity and widely distributed throughout the nervous system but not associated with any evidence of tissue damage or neuronal necrosis. After 6 months recovery phase the immune cell infiltrates regressed substantially. Such findings were recently described for another antisense moiety in a 41-week monkey study [48]. For the clinical dosing regimen, this study supported a dosing regimen with less than monthly dosing frequency to minimize any risk for immune cell infiltrations mentioned above.
Rugonersen equalizes paternal and maternal UBE3A expression across most brain regions in NHPs.
Rugonersen increased UBE3A mRNA in deeper brain regions and spinal cord tissues to a lower and more variable extent than in cortical regions, beyond what could be simply explained by lower exposure (Fig. 4C). Potential explanations include the differential degree of paternal allele silencing (i.e. imprinting) as well as different fractions of neurons versus nonneuronal cells between brain regions. To elucidate the relationship between paternal silencing and UBE3A protein upregulation across brain regions, we utilized a subgroup of animals that are heterozygous UBE3A SNP carriers (chr7: 3.339.686C → A; n = 12) in our sample from both saline and rugonersen treated groups at days 15 and 29 post injections.
We developed an allele specific UBE3A mRNA digital droplet PCR assay to determine the contribution of the paternal versus maternal allele to overall UBE3A mRNA expression in respective brain tissues. First, this analysis showed differences in baseline imprinting levels between brain regions, ranging from paternal/maternal ratio of ∼0.05 in cortical regions, where paternal UBE3A is almost completely imprinted, to 0.5 in spinal cord tissues, where paternal UBE3A is expressed (Fig. 5B, light blue boxes). Second, brain regions with the lowest paternal/maternal ratio in the saline group (e.g. cortex) showed greater increases of UBE3A mRNA upon treatment with rugonersen (Fig. 5B, dark green boxes). In cortical brain regions, UBE3A expression levels in rugonersen-treated animals were twice those of the saline group and had equal expression of paternally versus maternally derived UBE3A (ratio ∼1). This indicated that the imprinted paternal UBE3A allele becomes almost fully unsilenced following treatment with rugonersen. The hippocampus showed similarly strong unsilencing of UBE3A following treatment.
In general, (with exception of Pons) the imprinting level in the saline treated animals dictates the level of upregulation that we see of UBE3A mRNA following rugonersen treatment. In the midbrain, and moderately in cerebellum, we observed a more modest increase in UBE3A mRNA following treatment, despite almost full imprinting of UBE3A. This can however be explained by the lower exposure, and consequently weaker UBE3A-ATS reduction, in this tissue compared to cortex regions. Finally, the medulla and spinal cord regions, which showed less imprinting of paternal UBE3A in the saline-treated condition, also had a relatively weak increase in UBE3A mRNA upon rugonersen treatment. Nonetheless, in spinal cord tissues (both thoracic and cervical), equal expression of paternal and maternal UBE3A alleles, was observed in following rugonersen treatment, indicating that imprinted UBE3A was fully unsilenced in these tissues, similar to cortex (Fig. 4B). Correlation analysis suggested that the degree of imprinting determined the extent of UBE3A protein upregulation (Fig. 5C).
PK/PD modeling of the UBE3A LNA ASO to support human dose selection
A PK/PD model was derived by considering the ASO tissue concentration and target engagement, as measured by UBE3A-ATS reduction, and relating that to UBE3A mRNA and protein in NHP brain tissue (Fig. 4C). Rugonersen showed brain exposure well above the in vivo IC50 for UBE3A-ATS and EC50 for UBE3A mRNA (>5 μM 7 days post injection) in most brain regions, with the highest exposure in cortical regions and lowest in midbrain (Fig. 4C). Assuming that the relationship between rugonersen brain concentration and UBE3A-ATS mRNA reduction is similar across brain regions, we found that a direct, i.e. immediate, response model best fits the data. Potency was measured in the nanomolar range, and the slope of the response is steep (hill coefficient > 1).
The UBE3A mRNA response is also best described by a direct response model, linking the remaining UBE3A-ATS mRNA to the UBE3A mRNA elevation. UBE3A-ATS mRNA needs to be reduced by ∼84% (i.e. 16% remaining) to produce a half maximal response in UBE3A mRNA expression (Fig. 4C). The finding is in line with the AS mouse model, where a knock-down of 90% of Ube3a-ats mRNA was needed to reinstate Ube3a mRNA to 70% (Fig. 3E). UBE3A protein expression followed the UBE3A mRNA expression but with some delay (Fig. 5B). Therefore, the protein response is described with a turnover model, where UBE3A mRNA and protein levels have a 1:1 relationship under steady-state conditions. The half-life of the protein in cynomolgus monkey was estimated at about 13 days which is in line with protein half-life previously described in the brain [49]. With a 24 mg dose level, a steady state protein level of at least 50% above vehicle-treated animals is predicted to be achieved with dosing intervals every 16 weeks in monkeys (median prediction in the cortical regions). The possibility to use such infrequent dosing while maintaining 50% unsilencing can be attributed to the combination of both the long ASO- and protein half-life. This model provided the justification for the human dose selection in the Phase 1 Tangelo trial in AS patients.
The established PK/PD relationships between the three PD markers and their relationship to brain concentrations were incorporated into a larger PK/PD model describing and linking the dose to cerebrospinal fluid PK, brain PK and finally to the dynamic response of the three PD markers. A paper detailing the NHP PK data, model, and its translation to human is forthcoming. Given the similar in vitro potency of the compound towards UBE3A-ATS in both human and cynomolgus neurons, no scaling of the PD model from cynomolgus to human was required. However, the baseline level of UBE3A sense RNA differs in patients (versus healthy NHP) due to the silencing of the paternal allele expression by UBE3A-ATS RNA. The PK/PD model provided the justification for the human dose selection in the Phase 1 Tangelo trial in AS patients.
Discussion
Therapeutic strategy for AS
AS patients carry an intact paternal copy of UBE3A, which is silenced by the long noncoding RNA, UBE3A-ATS, in almost all neurons. Multiple groups have demonstrated that unsilencing of paternal UBE3A can be achieved by reducing UBE3A-ATS by various mechanisms, including ASOs that cleave UBE3A-ATS [4, 15, 20, 21, 23]. Thus, this atypical neuronal silencing mechanism provides a therapeutic means to restore UBE3A in the neurons of AS individuals.
Development and characterization of rugonersen
Unfortunately, there is substantial evolutionary divergence between human and mouse UBE3A-ATS sequences. Therefore, we developed an AS patient-derived neuronal cell model to screen for gapmer ASOs targeting human UBE3A-ATS. Starting with ASOs targeting only the sequences conserved between human and cynomolgus monkey, iterative screening cycles led to the identification of potent and selective cross-reactive molecules, including our lead candidate, rugonersen. The activity of rugonersen was characterized in human iPSC-derived neurons from both neurotypical and AS individuals, cynomolgus monkey neurons, and in vivo, after IT in NHPs, enabling a detailed understanding of the relationship between ASO concentration, UBE3A-ATS reduction, and UBE3A mRNA and protein unsilencing in both in vitro and in vivo models.
In vivo efficacy and PK/PD modeling
Remarkably, rugonersen administration in juvenile to pubertal NHP resulted in sustained and almost complete neuronal unsilencing of the paternal UBE3A locus and a concomitant increase of UBE3A protein in the cortex, hippocampus, and cerebellum: brain regions likely involved in the impairment of cognition, language/speech development, and motor skills in AS [39, 40]. Even three months after a single injection, UBE3A mRNA and protein in various cortical regions were nearly doubled compared to vehicle-treated animals. These data from the NHP studies were used to derive a translational PK/PD model, suggesting an acceptable dosing frequency in patients. This work combined with a desirable safety profile, culminated in rugonersen entering a phase 1 clinical trial in AS (NCT04428281).
Comparison with other UBE3A-ATS ASOs and mechanism insights
We and others previously identified mouse reactive Ube3a-ats ASOs that led to both knock-down of the Ube3a-ats and a partial unsilencing of Ube3a mRNA and protein [20, 21]. These ASOs resulted in a reversal of deficits in an AS mouse model linked to the disease including synaptic plasticity and cognition [20, 21]. In addition, human specific ASOs were previously shown to revert functional and molecular changes in patient neurons [11, 19]. In a similar study to ours, human/cynomolgus monkey cross reactive ASOs targeting UBE3A-ATS knocked down UBE3A-ATS but only moderately upregulated UBE3A mRNA and protein levels in AS human neurons and NHP brains [23]. In contrast, we demonstrated that potent reduction of UBE3A-ATS enables almost complete unsilencing of paternal UBE3A expression with maximal UBE3A protein upregulation across multiple brain regions in healthy NHP brains. Head-to-head comparison of rugonersen and this previously published ASO molecule along with a more traditional ASO constructed with 2′methoxy-ethyl chemistry, suggested that rugonersen more potently unsilenced paternal UBE3A (Supplementary Fig. 1). The difference between these molecules might be explained, in part, by molecule properties (e.g. binding affinities) or alternatively the region targeted on the UBE3A-ATS locus. For example, rugonersen targets an intronic sequence which may be more effective in terminating transcription downstream of ASO-mediated cleavage of nascent transcripts, which is considered crucial for the proposed silencing mechanism of UBE3A-ATS [50]. Furthermore, the precise location relative to other genomic elements (i.e. splice sites, protein binding sites, and others) may influence the ability of an ASO to activate UBE3A, since these other elements may also influence transcriptional termination [19, 51]. Comparison of ASO-mediated UBE3A-ATS cleavage, transcriptional termination, and UBE3A transcription might give some answers and may reveal how different published ASOs may impact the so-called “RNA polymerase collision hypothesis” at this locus [14, 18, 19, 52].
In vitro to in vivo translation and species-specific responses
Another poorly understood mechanism, albeit very important for modeling and human dose predictions, is the different relationships between UBE3A-ATS lowering and UBE3A unsilencing when mouse and cynomolgus monkey brains are compared to human neurons. The amount of UBE3A-ATS reduction required to unsilence UBE3A is substantially higher in vivo compared to in vitro models. Additional work using ASOs that are cross-reactive between species or humanized mouse models are needed to better understand in vitro to in vivo translation of ASOs.
PK profile and brain distribution of rugonersen
Identification of molecules with long duration of action are greatly needed to reduce the frequency of administration for drugs delivered via invasive IT injections. Rugonersen showed a desirable PK profile. Notably, the ASO produced an extended duration of effect of paternal unsilencing in monkeys, up to 3 months after IT dosing, in key disease brain regions. While an almost two-fold overexpression of both RNA and protein in brains of both mice and NHPs, would indirectly suggest unsilencing and reinstatement of most neuronal subpopulations including excitatory and inhibitory neurons, future studies are warranted to determine whether there is biased unsilencing of UBE3A across different neuronal subtypes. Similar to other oligonucleotides tested in NHPs and humans, rugonersen did not distribute across tissues equally in NHP brains [53]. We found that the increase in UBE3A mRNA and protein levels upon rugonersen treatment was highest in cortical regions, where the drug exposure and imprinting levels were also the highest. Variations in imprinting levels across brain tissues may result from either ‘leaky’ imprinting in the neurons of some tissues or from increased populations of non-neuronal cells without imprinted UBE3A expression, contributing more to the total pool of UBE3A mRNA. Regardless of the cause, it was encouraging that rugonersen treatment led to strong unsilencing of the paternal UBE3A allele expression in all brain tissues. These results demonstrate that it is possible to fully unsilence paternal UBE3A, and thus an ASO strategy has the potential to fully compensate for the lack of functional maternal UBE3A protein in AS patients. Few data have been published on oligonucleotide brain PKs in patients after IT administration, but it seems that concentrations in the micromolar range, as have been demonstrated here in NHPs, are achievable [53, 54].
Clinical development challenges and future delivery technologies
Several challenges for clinical development of UBE3A-ATS ASO therapeutics remain. These include understanding and optimizing the brain distribution of IT-injected ASOs in humans, identification of biomarkers to measure PD responses on a molecular and circuit level, and identification of objective outcomes measures to assess patient benefits. Future innovations in technologies to deliver ASO payloads may provide a more homogeneous ASO brain concentration. For example, technologies that hijack transport systems such as the transferrin or CD98 receptors, or molecules conjugated with fatty acids or cholesterol moieties to improve distribution may facilitate crossing the blood brain barrier and broader tissue distribution [30, 55, 56]. While it may also be tempting to explore siRNAs targeting UBE3A-ATS, this is unlikely to be successful, as siRNAs have not been shown as of yet to engage nuclear RNAs.
Considerations for genetic subtypes and safety profile
In some genetic subtypes of AS, for example in individuals with AS due to uniparental paternal disomy or imprinting center defects, both alleles are silenced via UBE3A-ATS. In these cases, rugonersen and other ASOs could potentially lead to overexpression of UBE3A mRNA and protein. This may either be of concern since increased UBE3A gene dosage contributes to 15q11–q13 duplication syndrome, or of interest due to the increased potential to restore optimal UBE3A protein levels. Study of mouse-specific ASOs in a murine model of AS imprinting defects may help better predict the risk versus benefit. For individuals with these genetic subtypes, rugonersen and other similar ASOs will need extra attention in the clinic. Fortunately, in healthy NHPs, and with short term treatments every 4–8 weeks where rugonersen treatment led to two-fold increased expression of UBE3A protein, the ASO was well-tolerated with no histological findings or behaviors that were considered adverse. This is particularly reassuring, considering the relatively high doses of a very potent ASO with LNA chemistry that are required for meaningful reinstatement of UBE3A protein.
Biomarkers for monitoring treatment response
Whether UBE3A protein can be measured in human CSF, and more specifically if treatment response could be quantified using a fluid biomarker, remains to be determined. Proteins differentially expressed upon UBE3A protein reinstatement in neurons that are also measurable in body fluids may allow monitoring of proximal target engagement after restoring UBE3A. For example, we reported PEG10 and TKT as potential biomarkers of target engagement [9–11] (https://www.angelman-proteome-project.org/). In addition, EEG (excess delta-band power) is emerging as an AS-relevant biomarker of abnormal brain function in AS and rodent models that has been linked to symptom severity in natural history data [57–59]. It will be important to understand the relationship between these candidate biomarkers to clinical severity and outcome measures both in the natural history and in response to treatments.
Conclusion and clinical progress of rugonersen
Collectively, our data support the clinical development of rugonersen for AS. A phase 1 study of rugonersen is close to completing an optional open label extension (clinicaltrials.gov; NCT04428281). The data from the multiple ascending dose and long-term extension portions of the Phase 1 study found that rugonersen has an acceptable safety and tolerability profile, led to a dose-dependent partial normalization of the AS-associated EEG abnormality, and revealed signals of clinical improvement in core AS symptom domains beyond expectation from natural history data [24]. Rugonersen was licensed by Oak Hill Bio, Ltd., who subsequently announced plans to start a Phase 3 trial in 2026.
Supplementary Material
Acknowledgements
We would like to thank Children’s Hospital in Boston, under the leadership of Prof. Christopher Walsh, for recruitment of patients into the study for iPSC derivation; Harvard iPSC core facility and the team of Laurence Daheron for the reprogramming and quality control of hiPSC lines; the Roche postdoctoral fellowship program for funding N.J.P., C.W., V.C., and M.T., and the Roche RiSE internship program for funding M.T. We thank Silke Zimmermann, Laura Badi, Tony Kam-Thong, Roland Schmucki, Fabian Koechl, Nadia Anastasi, Hippolyte Gander, Muriel Schmitt, and Nicole Hauser for the excellent technical support. We thank Anirvan Ghosh, Omar Khwaja, Kelly Bales, and Martin Ebeling for their continuous support and critical contributions to the project.
Author contributions: Ravi Jagasia (Conceptualization [lead], Data curation [supporting], Formal analysis [equal], Investigation [equal], Methodology [supporting], Visualization [equal], Writing—original draft [equal], Writing—review & editing [equal]), Charlotte Bon (Data curation [lead], Formal analysis [equal], Investigation [equal], Methodology [lead], Project administration [equal], Visualization [equal], Writing—original draft [equal], Writing—review & editing [equal]), Soren V. Rasmussen (Conceptualization [equal], Data curation [equal], Formal analysis [equal], Investigation [equal], Methodology [equal], Project administration [equal], Visualization [equal], Writing—original draft [equal], Writing—review & editing [equal]), Solveig Badillo (Data curation [equal], Formal analysis [equal], Software [supporting], Visualization [equal]), Disa Tehler (Formal analysis [supporting], Investigation [supporting]), Danièle Buchy (Data curation [equal], Formal analysis [equal], Investigation [supporting], Methodology [equal], Writing—original draft [supporting]), Marco Berrera (Data curation [equal], Formal analysis [equal], Investigation [supporting], Methodology [supporting], Visualization [equal]), Megana Prasad (Data curation [equal], Formal analysis [equal], Writing—original draft [supporting]), Stormy Chamberlain (Data curation [equal], Formal analysis [equal], Investigation [equal], Writing—original draft [equal], Writing—review & editing [lead]), Marco Terrigno (Data curation [supporting], Formal analysis [supporting]), Nikhil Pandya (Data curation [equal], Formal analysis [equal], Investigation [supporting], Methodology [equal], Writing—original draft [supporting]), Veronica Costa (Data curation [equal], Formal analysis [equal], Methodology [supporting]), Congwei Wang (Data curation [supporting], Formal analysis [supporting]), Lykke Pedersen (Data curation [equal], Formal analysis [equal], Investigation [supporting], Methodology [supporting]), Meghan T. Miller (Investigation [supporting], Project administration [supporting]), Kamille Dumong Erichsen (Data curation [equal], Formal analysis [equal], Investigation [supporting]), Lars Joenson (Data curation [supporting], Formal analysis [supporting]), Christoph Patsch (Data curation [supporting], Formal analysis [supporting], Methodology [equal], Project administration [supporting]), Joerg F Hipp (Investigation [supporting], Project administration [supporting], Writing—original draft [supporting], Writing—review & editing [supporting]), Azad Bonni (Project administration [supporting]), Lutz Müller (Data curation [supporting], Formal analysis [supporting], Investigation [supporting], Project administration [supporting], Writing—original draft [supporting], Writing—review & editing [equal]), Thomas Kremer (Data curation [supporting], Formal analysis [supporting], Investigation [supporting], Project administration [supporting], Writing—original draft [supporting]), Erich Koller (Data curation [equal], Project administration [equal], Validation [supporting], Writing—original draft [supporting], Writing—review & editing [supporting]), Marius Christian Hoener (Conceptualization [lead], Data curation [lead], Formal analysis [lead], Investigation [lead], Methodology [equal], Project administration [lead], Writing—original draft [lead], Writing—review & editing [lead])
Contributor Information
Ravi Jagasia, Roche Pharma Research and Early Development, Neuroscience and Rare Disease discovery and translational area, Roche Innovation Center Basel, Basel 4070, Switzerland.
Charlotte Bon, Roche Pharma Research and Early Development, Pharmaceutical Sciences, Roche Innovation Center Basel, Basel 4070, Switzerland.
Soren V Rasmussen, Roche Pharma Research and Early Development, Therapeutic Modalities, Roche Innovation Center Basel, Basel 4070, Switzerland.
Solveig Badillo, Roche Pharma Research and Early Development, Pharmaceutical Sciences, Roche Innovation Center Basel, Basel 4070, Switzerland.
Disa Tehler, Roche Pharma Research and Early Development, Therapeutic Modalities, Roche Innovation Center Basel, Basel 4070, Switzerland.
Danièle Buchy, Roche Pharma Research and Early Development, Neuroscience and Rare Disease discovery and translational area, Roche Innovation Center Basel, Basel 4070, Switzerland.
Marco Berrera, Roche Pharma Research and Early Development, Pharmaceutical Sciences, Roche Innovation Center Basel, Basel 4070, Switzerland.
Megana Prasad, Roche Pharma Research and Early Development, Pharmaceutical Sciences, Roche Innovation Center Basel, Basel 4070, Switzerland.
Stormy Chamberlain, Roche Pharma Research and Early Development, Neuroscience and Rare Disease discovery and translational area, Roche Innovation Center Basel, Basel 4070, Switzerland.
Marco Terrigno, Roche Pharma Research and Early Development, Neuroscience and Rare Disease discovery and translational area, Roche Innovation Center Basel, Basel 4070, Switzerland.
Nikhil J Pandya, Roche Pharma Research and Early Development, Neuroscience and Rare Disease discovery and translational area, Roche Innovation Center Basel, Basel 4070, Switzerland.
Veronica Costa, Roche Pharma Research and Early Development, Neuroscience and Rare Disease discovery and translational area, Roche Innovation Center Basel, Basel 4070, Switzerland.
Congwei Wang, Roche Pharma Research and Early Development, Neuroscience and Rare Disease discovery and translational area, Roche Innovation Center Basel, Basel 4070, Switzerland.
Lykke Pedersen, Roche Pharma Research and Early Development, Therapeutic Modalities, Roche Innovation Center Basel, Basel 4070, Switzerland.
Meghan T Miller, Roche Pharma Research and Early Development, Neuroscience and Rare Disease discovery and translational area, Roche Innovation Center Basel, Basel 4070, Switzerland.
Kamille Dumong Erichsen, Roche Pharma Research and Early Development, Therapeutic Modalities, Roche Innovation Center Basel, Basel 4070, Switzerland.
Lars Joenson, Roche Pharma Research and Early Development, Therapeutic Modalities, Roche Innovation Center Basel, Basel 4070, Switzerland.
Christoph Patsch, Roche Pharma Research and Early Development, Therapeutic Modalities, Roche Innovation Center Basel, Basel 4070, Switzerland.
Joerg F Hipp, Roche Pharma Research and Early Development, Neuroscience and Rare Disease discovery and translational area, Roche Innovation Center Basel, Basel 4070, Switzerland.
Azad Bonni, Roche Pharma Research and Early Development, Neuroscience and Rare Disease discovery and translational area, Roche Innovation Center Basel, Basel 4070, Switzerland.
Lutz Müller, Roche Pharma Research and Early Development, Pharmaceutical Sciences, Roche Innovation Center Basel, Basel 4070, Switzerland.
Annamaria Braendli-Baiocco, Roche Pharma Research and Early Development, Pharmaceutical Sciences, Roche Innovation Center Basel, Basel 4070, Switzerland.
Thomas Kremer, Roche Pharma Research and Early Development, Neuroscience and Rare Disease discovery and translational area, Roche Innovation Center Basel, Basel 4070, Switzerland.
Erich Koller, Roche Pharma Research and Early Development, Pharmaceutical Sciences, Roche Innovation Center Basel, Basel 4070, Switzerland.
Marius C Hoener, Roche Pharma Research and Early Development, Neuroscience and Rare Disease discovery and translational area, Roche Innovation Center Basel, Basel 4070, Switzerland.
Supplementary data
Supplementary data is available at NAR online.
Conflict of interest
R.J., C.B., S.V.R, S.B., D.T., D.B., M.B., M.P., S.C., M.T., N.J.P., V.C., C.W., L.P., M.T.M., K.D.E., L.J., C.P., J.F.H., A.B., L.M., A.B.-B., T.K., E.K., and M.C.H., are or were employed by F. Hoffmann-La Roche. Parts of the work in this study have been filed in patent application WO2017/081223A1.
Funding
Funding to pay the Open Access publication charges for this article was provided by F. Hoffmann-La Roche Ltd.
Data availability
The sequencing data generated in this study have been deposited in the NCBI Sequence Read Archive (SRA) under the accession number PRJNA846765. All data are written up in reports used for regulatory submissions for our clinical trial(s) and are as such available in regulatory file depositories, IND reviewed. Additional data related to this study are available from the corresponding author upon reasonable request.
References
- 1. Buiting K, Williams C, Horsthemke B Angelman syndrome – insights into a rare neurogenetic disorder. 2016; 12:584–93. 10.1038/nrneurol.2016.133. [DOI] [PubMed] [Google Scholar]
- 2. Fang P, Lev-Lehman E, Tsai TF et al. The spectrum of mutations in UBE3A causing Angelman syndrome. Hum Mol Genet. 1999; 8:129–35. 10.1093/hmg/8.1.129. [DOI] [PubMed] [Google Scholar]
- 3. Cook EH, Lindgren V, Leventhal BL et al. Autism or atypical autism in maternally but not paternally derived proximal 15q duplication. Am J Hum Genet. 1997; 60:928–34. [PMC free article] [PubMed] [Google Scholar]
- 4. Matsuura T, Sutcliffe JS, Fang P et al. De novo truncating mutations in E6-AP ubiquitin-protein ligase gene (UBE3A) in Angelman syndrome. Nat Genet. 1997; 15:74–7. 10.1038/ng0197-74. [DOI] [PubMed] [Google Scholar]
- 5. Noor A, Dupuis L, Mittal K et al. 15q11.2 Duplication encompassing only the UBE3A gene is associated with developmental delay and neuropsychiatric phenotypes. Hum Mutat. 2015; 36:689–93. 10.1002/humu.22800. [DOI] [PubMed] [Google Scholar]
- 6. Chamberlain SJ, Lalande M Neurodevelopmental disorders involving genomic imprinting at human chromosome 15q11–q13. Neurobiol Dis. 2010; 39:13–20. 10.1016/j.nbd.2010.03.011. [DOI] [PubMed] [Google Scholar]
- 7. Mabb AM, Judson MC, Zylka MJ et al. Angelman syndrome: insights into genomic imprinting and neurodevelopmental phenotypes. Trends Neurosci. 2011; 34:293–303. 10.1016/j.tins.2011.04.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Soni S, Whittington J, Holland AJ et al. The course and outcome of psychiatric illness in people with Prader–Willi syndrome: implications for management and treatment. J Intellect Disabil Res. 2007; 51:32–42. 10.1111/j.1365-2788.2006.00895.x. [DOI] [PubMed] [Google Scholar]
- 9. Pandya NJ, Meier S, Tyanova S et al. A cross-species spatiotemporal proteomic analysis identifies UBE3A-dependent signaling pathways and targets. Mol Psychiatry. 2022; 27:2590–601. 10.1038/s41380-022-01484-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Scheiffele P, Beg AA Neuroscience: angelman syndrome connections. Nature. 2010; 468:907–8. 10.1038/468907a. [DOI] [PubMed] [Google Scholar]
- 11. Pandya NJ, Wang C, Costa V et al. Secreted retrovirus-like GAG-domain-containing protein PEG10 is regulated by UBE3A and is involved in Angelman syndrome pathophysiology. Cell Reports Medicine. 2021; 2:100360. 10.1016/j.xcrm.2021.100360. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Sun J, Zhu G, Liu Y et al. UBE3A Regulates synaptic plasticity and learning and memory by controlling SK2 channel endocytosis. Cell Rep. 2015; 12:449–61. 10.1016/j.celrep.2015.06.023. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Rougeulle C, Cardoso C, Fontes M et al. An imprinted antisense RNA overlaps UBE3A and a second maternally expressed transcript. Nat Genet. 1998; 19:15–6. 10.1038/ng0598-15. [DOI] [PubMed] [Google Scholar]
- 14. Meng L, Person RE, Beaudet AL Ube3a-ATS is an atypical RNA polymerase II transcript that represses the paternal expression of Ube3a. Hum Mol Genet. 2012; 21:3001–12. 10.1093/hmg/dds130. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Huang HS, Allen JA, Mabb AM et al. Topoisomerase inhibitors unsilence the dormant allele of Ube3a in neurons. Nature. 2012; 481:185–9. 10.1038/nature10726. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Runte M, Kroisel PM, Gillessen-Kaesbach G et al. SNURF-SNRPN and UBE3A transcript levels in patients with Angelman syndrome. Hum Genet. 2004; 114:553–61. 10.1007/s00439-004-1104-z. [DOI] [PubMed] [Google Scholar]
- 17. Chamberlain SJ, Brannan CI The Prader–Willi syndrome imprinting center activates the paternally expressed murine Ube3a antisense transcript but represses paternal Ube3a. Genomics. 2001; 73:316–22. 10.1006/geno.2001.6543. [DOI] [PubMed] [Google Scholar]
- 18. Meng L, Person RE, Huang W et al. Truncation of Ube3a-ATS unsilences paternal Ube3a and ameliorates behavioral defects in the Angelman syndrome mouse model. PLoS Genet. 2013; 9:e1004039. 10.1371/journal.pgen.1004039. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Germain ND, Gorka D, Drennan R et al. Antisense oligonucleotides targeting UBE3A-ATSrestore expression ofUBE3Aby relieving transcriptional interference. bioRxiv 10.1101/2021.07.09.451826. [DOI]
- 20. Meng L, Ward AJ, Chun S et al. Towards a therapy for Angelman syndrome by targeting a long non-coding RNA. Nature. 2015; 518:409–12. 10.1038/nature13975. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Milazzo C, Mientjes EJ, Wallaard I et al. Antisense oligonucleotide treatment rescues UBE3A expression and multiple phenotypes of an Angelman syndrome mouse model. JCI Insight. 2021; 6:e145991. 10.1172/jci.insight.145991. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Elgersma Y, Sonzogni M UBE3A reinstatement as a disease-modifying therapy for Angelman syndrome. Develop Med Child Neuro. 2021; 63:802–7. 10.1111/dmcn.14831. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Dindot SV, Christian S, Murphy WJ et al. An ASO therapy for Angelman syndrome that targets an evolutionarily conserved region at the start of the transcript. Sci Transl Med. 2023; 15:eabf4077. 10.1126/scitranslmed.abf4077. [DOI] [PubMed] [Google Scholar]
- 24. Hipp JF, Bacino CA, Bird LM et al. The UBE3A-ATS antisense oligonucleotide rugonersen in children with Angelman syndrome: a phase 1 trial. Nat Med. 2025; 10.1038/s41591-025-03784-7. [DOI] [PubMed] [Google Scholar]
- 25. Costa V, Aigner S, Vukcevic M et al. mTORC1 Inhibition corrects neurodevelopmental and synaptic alterations in a human stem cell model of tuberous sclerosis. Cell Rep. 2016; 15:86–95. 10.1016/j.celrep.2016.02.090. [DOI] [PubMed] [Google Scholar]
- 26. Dunkley T, Costa V, Friedlein A et al. Characterization of a human pluripotent stem cell-derived model of neuronal development using multiplexed targeted proteomics. Proteomics Clinical Apps. 2015; 9:684–94. 10.1002/prca.201400150. [DOI] [PubMed] [Google Scholar]
- 27. Gonzalez F, Zhu Z, Shi ZD et al. An iCRISPR platform for rapid, multiplexable, and inducible genome editing in human pluripotent stem cells. Cell Stem Cell. 2014; 15:215–26. 10.1016/j.stem.2014.05.018. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Thoma EC, Heckel T, Keller D et al. Establishment of a translational endothelial cell model using directed differentiation of induced pluripotent stem cells from Cynomolgus monkey. Sci Rep. 2016; 6:35830. 10.1038/srep35830. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Pedersen L, Hagedorn PH, Koch T Identifying suitable target regions and analyzing off-target effects of therapeutic oligonucleotides. Methods Mol Biol. 2019; 2036:261–82. 10.1007/978-1-4939-9670-4_16. [DOI] [PubMed] [Google Scholar]
- 30. Hammond SM, Abendroth F, Goli L et al. Antibody-oligonucleotide conjugate achieves CNS delivery in animal models for spinal muscular atrophy. JCI Insight. 2022; 7:e154142. 10.1172/jci.insight.154142. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Li H, Handsaker B, Wysoker A et al. The sequence alignment/map format and SAMtools. Bioinformatics. 2009; 25:2078–9. 10.1093/bioinformatics/btp352. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Mortazavi A, Williams BA, McCue K et al. Mapping and quantifying mammalian transcriptomes by RNA-seq. Nat Methods. 2008; 5:621–8. 10.1038/nmeth.1226. [DOI] [PubMed] [Google Scholar]
- 33. Haley TJ, McCormick WG Pharmacological effects produced by intracerebral injection of drugs in the conscious mouse. Br J Pharmacol Chemother. 1957; 12:12–5. 10.1111/j.1476-5381.1957.tb01354.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Vandesompele J, De Preter K, Pattyn F et al. Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes. Genome Biol. 2002; 3:0034. 10.1186/gb-2002-3-7-research0034. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Kulak NA, Pichler G, Paron I et al. Minimal, encapsulated proteomic-sample processing applied to copy-number estimation in eukaryotic cells. Nat Methods. 2014; 11:319–24. 10.1038/nmeth.2834. [DOI] [PubMed] [Google Scholar]
- 36. Derendorf H, Meibohm B Modeling of pharmacokinetic/pharmacodynamic (PK/PD) relationships: concepts and perspectives. Pharm Res. 16:176–85. 10.1023/A:1011907920641. [DOI] [PubMed] [Google Scholar]
- 37. Costa V, Hedtjärn M, Hoener M et al. Oligonucleotides for inducing paternal UBE3A expression. 2017; WO2017081223A1:Patent application. [Google Scholar]
- 38. Hagedorn PH, Brown JM, Easton A et al. Acute neurotoxicity of antisense oligonucleotides after intracerebroventricular injection into mouse brain can Be predicted from sequence features. Nucleic Acid Ther. 2022; 32:151–62. 10.1089/nat.2021.0071. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Dieckmann A, Hagedorn PH, Burki Y et al. A sensitive in vitro approach to assess the hybridization-dependent toxic potential of high affinity gapmer oligonucleotides. Mol Ther Nucleic Acids. 2018; 10:45–54. 10.1016/j.omtn.2017.11.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Gilmore RB, Gorka D, Stoddard CE et al. Generation of isogenic models of Angelman syndrome and Prader–Willi syndrome in CRISPR/Cas9-engineered human embryonic stem cells. PLoS One. 2024; 19:e0311565. 10.1371/journal.pone.0311565. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Hicks J, Navin N, Troge J et al. Varietal counting of nucleic acids for obtaining genomic copy number information. 2012; WO2012054873A2:Patent application. [Google Scholar]
- 42. Jiang YH, Armstrong D, Albrecht U et al. Mutation of the Angelman ubiquitin ligase in mice causes increased cytoplasmic p53 and deficits of contextual learning and long-term potentiation. Neuron. 1998; 21:799–811. 10.1016/S0896-6273(00)80596-6. [DOI] [PubMed] [Google Scholar]
- 43. Peters S, Wirkert E, Kuespert S et al. Safe and effective Cynomolgus monkey GLP-tox study with repetitive intrathecal application of a TGFBR2 targeting LNA-Gapmer antisense oligonucleotide as treatment candidate for neurodegenerative disorders. Pharmaceutics. 2022; 14:200. 10.3390/pharmaceutics14010200. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44. Monine M, Norris D, Wang Y et al. A physiologically-based pharmacokinetic model to describe antisense oligonucleotide distribution after intrathecal administration. J Pharmacokinet Pharmacodyn. 2021; 48:639–54. 10.1007/s10928-021-09761-0. [DOI] [PubMed] [Google Scholar]
- 45. Korte S, Runge F, Wozniak MM et al. Range of neurological signs in Cynomolgus monkeys after intrathecal Bolus administration of antisense oligonucleotides. Int J Toxicol. 2020; 39:505–9. 10.1177/1091581820948454. [DOI] [PubMed] [Google Scholar]
- 46. Lamb M, Engelhardt JA, Grubor B et al. Antisense oligonucleotide-related macrovesicular vacuolation of hippocampal neurons in nonhuman primates. Toxicol Pathol. 2022; 50:197–210. 10.1177/01926233211063831. [DOI] [PubMed] [Google Scholar]
- 47. Frazier KS Antisense oligonucleotide therapies: the promise and the challenges from a toxicologic pathologist’s perspective. Toxicol Pathol. 2015; 43:78–89. 10.1177/0192623314551840. [DOI] [PubMed] [Google Scholar]
- 48. Dubost V, Wuersch K, Penraat K et al. Characterizing antisense oligonucleotide-induced histopathology findings in spinal cord of Mauritius cynomolgus monkeys by molecular localization investigation. Toxicol Pathol. 2025; 53:95–106. 10.1177/01926233241311277. [DOI] [PubMed] [Google Scholar]
- 49. Alvarez-Castelao B, Schuman EM The regulation of synaptic protein turnover. J Biol Chem. 2015; 290:28623–30. 10.1074/jbc.R115.657130. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50. Lee JS, Mendell JT Antisense-mediated transcript knockdown triggers premature transcription termination. Mol Cell. 2020; 77:1044–54. 10.1016/j.molcel.2019.12.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51. Rigo F, Seth PP, Bennett CF Antisense oligonucleotide-based therapies for diseases caused by pre-mRNA processing defects. Adv Exp Med Biol. 2014; 825:303–52. 10.1007/978-1-4939-1221-6_9. [DOI] [PubMed] [Google Scholar]
- 52. Hsiao JS, Germain ND, Wilderman A et al. A bipartite boundary element restricts UBE3A imprinting to mature neurons. Proc Natl Acad Sci USA. 2019; 116:2181–6. 10.1073/pnas.1815279116. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 53. Finkel RS, Chiriboga CA, Vajsar J et al. Treatment of infantile-onset spinal muscular atrophy with nusinersen: final report of a phase 2, open-label, multicentre, dose-escalation study. Lancet Child Adolesc Health. 2021; 5:491–500. 10.1016/S2352-4642(21)00100-0. [DOI] [PubMed] [Google Scholar]
- 54. Ramos DM, Ydewalle C, Gabbeta V et al. Age-dependent SMN expression in disease-relevant tissue and implications for SMA treatment. J Clin Invest. 2019; 129:4817–31. 10.1172/JCI124120. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 55. Barker SJ, Thayer MB, Kim C et al. Targeting the transferrin receptor to transport antisense oligonucleotides across the mammalian blood-brain barrier. Sci Transl Med. 2024; 16:eadi2245. 10.1126/scitranslmed.adi2245. [DOI] [PubMed] [Google Scholar]
- 56. Nagata T, Dwyer CA, Yoshida-Tanaka K et al. Cholesterol-functionalized DNA/RNA heteroduplexes cross the blood–brain barrier and knock down genes in the rodent CNS. Nat Biotechnol. 2021; 39:1529–36. 10.1038/s41587-021-00972-x. [DOI] [PubMed] [Google Scholar]
- 57. Frohlich J, Miller MT, Bird LM et al. Electrophysiological phenotype in Angelman syndrome differs between genotypes. Biol Psychiatry. 2019; 85:752–9. 10.1016/j.biopsych.2019.01.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 58. Sidorov MS, Deck GM, Dolatshahi M et al. Delta rhythmicity is a reliable EEG biomarker in Angelman syndrome: a parallel mouse and human analysis. J Neurodev Disord. 2017; 9:17. 10.1186/s11689-017-9195-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 59. Hipp JF, Frohlich J, Keute M et al. Electrophysiological abnormalities in Angelman syndrome correlate with symptom severity. Biol Psychiatry Glob Open Sci. 2021; 1:201–9. 10.1016/j.bpsgos.2021.05.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The sequencing data generated in this study have been deposited in the NCBI Sequence Read Archive (SRA) under the accession number PRJNA846765. All data are written up in reports used for regulatory submissions for our clinical trial(s) and are as such available in regulatory file depositories, IND reviewed. Additional data related to this study are available from the corresponding author upon reasonable request.