RNAi-Induced Expression of Paternal UBE3A

TL;DR

This study explores using RNA interference to activate the paternal UBE3A gene in neurons, which could help treat Angelman syndrome.

Contribution

The study demonstrates that RNAi targeting SNORD115 can activate paternal UBE3A in both mouse and human neurons.

Findings

RNAi targeting SNORD115 reduced UBE3A-ATS and activated paternal UBE3A in mouse neurons.

Similar activation was observed in human iPSC-derived neurons using LV-shSNORD115.

AAV-based delivery of shRNA shows potential but requires further in vivo evaluation.

Abstract

Background/Objectives: Angelman syndrome is a neurodevelopmental disorder resulting from a deficiency of the maternally inherited UBE3A gene. In mature neurons, UBE3A expression is restricted to the maternal allele due to tissue-specific genomic imprinting, while the paternal allele is silenced in cis by the UBE3A antisense transcript (UBE3A-ATS). To date, numerous strategies have been employed to activate paternal UBE3A expression. In this study, we utilized RNA interference (RNAi) to investigate the downregulation of UBE3A-ATS in mouse primary neurons and human induced pluripotent stem cell (iPSC)-derived neurons. Methods: To induce paternal UBE3A expression, we employed small interfering RNA (siRNA) oligonucleotides (20 mouse candidates and 47 human candidates) and lentiviral short hairpin RNA (LV-shRNA) targeting SNORD115 to suppress UBE3A-ATS expression in both mouse primary neurons and iPSCs. Subsequently, we assessed the expression levels of Angelman syndrome-related neighboring and target genes at the transcript and, where applicable, protein levels. Results: Following treatment with siSnord115 or LV-shSnord115, we observed a reduction in Ube3a-ATS and a corresponding activation of paternal Ube3a RNA and protein expression in both Ube3aP-YFP/m+ and Ube3ap+/m− mouse primary neurons. A similar effect was observed upon treatment with LV-shSNORD115s in human iPSC-derived neurons. Conclusions: shRNA-mediated inhibition of Ube3a-ATS by targeting Snord115 effectively restores Ube3a/UBE3A expression in both mouse neurons and human iPSCs. While promising, the mild reduction in Snord116 raises concerns about potential off-target effects. AAV-based delivery of shRNA shows potential, but its translational applicability remains to be evaluated in vivo.

Figures

Full text

1. Introduction

Angelman syndrome (AS) is a severe neurodevelopmental disorder (approximately 1:12,000 to 1:24,000 individuals) caused by a deficiency of the maternal Ubiquitin Protein Ligase E3A (UBE3A) gene, resulting in severe intellectual disability, developmental delay, profound speech impairment, motor dysfunction, microcephaly, sleep disturbance, seizures, and abnormal electroencephalogram (EEG). AS is diagnosed based on a combination of characteristic clinical features and molecular genetic testing, confirming loss of function of the maternally inherited UBE3A allele [1,2,3,4,5]. Currently, only symptomatic treatments are available for AS; therefore, the development of a disease-modifying therapy remains a critical unmet need.

UBE3A (E6-AP) is a HECT-type E3 ubiquitin ligase that catalyzes the transfer of ubiquitin from an E2 enzyme to substrate proteins, thereby regulating their degradation via the ubiquitin–proteasome system. Through this activity, UBE3A controls the stability of several neuronal proteins, including Rho Guanine Nucleotide Exchange Factor 15 (Ephexin5), Activity-Regulated Cytoskeleton-associated protein (Arc), and Tumor protein p53 (p53), influencing synaptic development and plasticity [6,7,8,9,10,11].

UBE3A expression is regulated by genetic imprinting, wherein the maternal allele is expressed while the paternal allele is silenced in a neuron-specific manner. Paternal silencing is mediated by a segment of a large, noncoding antisense transcript, UBE3A-ATS, which is transcribed from the paternally inherited chromosome and localized within the nucleus. Two primary models have been proposed for paternal UBE3A silencing: (1) the collision model and (2) the R-loop stabilization model. In the collision model, antisense transcription from UBE3A-ATS causes head-on collisions between RNA polymerase II complexes, leading to premature termination of sense-strand transcription. In the R-loop stabilization model, antisense RNA forms RNA: DNA hybrids that recruit repressive chromatin modifiers and promote heterochromatin formation across the UBE3A locus. These processes likely function sequentially, whereby transcriptional interference initiates silencing and R-loop-mediated chromatin remodeling stabilizes long-term repression of the paternal UBE3A allele [12,13,14,15]. In both humans and mice, the UBE3A-ATS transcript is derived from the same precursor that encodes small nuclear ribonucleoprotein N (SNRPN). This precursor contains multiple upstream promoters located at the Prader–Willi Syndrome imprinting center (PWS-IC), as well as upstream of the U exons, from where transcription initiates and ultimately overlaps with the UBE3A gene [12]. Approximately 70% of individuals with AS carry a large deletion on chromosome 15 that encompasses the UBE3A locus. The combination of paternal UBE3A silencing and loss of maternal UBE3A expression due to mutations results in deficient neuronal UBE3A protein in AS [1]. If the paternal UBE3A gene remains genetically intact in these cases, a promising therapeutic strategy involves activating the silenced paternal allele by targeting and suppressing UBE3A-ATS expression.

Previous studies have demonstrated that targeting the Ube3a-ATS transcript can effectively induce expression of paternal Ube3a expression by using pharmaceutical reagents (Topotecan, (S)-PHA533533), genetic manipulations (deletion of the genomic area around its promoter or insertion of a transcriptional stop cassette), Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR) and CRISPR-associated protein 9 (CRISPR/Cas9), RNA-targeting CRISPR/Cas13, and Antisense Oligonucleotides (ASOs) [12,13,14,16,17,18,19,20,21]. In detail, Topotecan (50 nmol in 5 μL) treatment with intrathecal (IT) delivery showed a significantly increased number of paternal UBE3A-YFP-positive cells in spinal cord neurons, and it remained up to 12 weeks in Ube3a^P-YFP/m+^ mice (Paternal UBE3A-YFP mice) [18]. The Topotecan study demonstrates limited persistence of therapeutic effect after a single administration, showing paternal UBE3A reactivation for up to 12 weeks in subsets of neurons but lacking evidence of long-term durability or sustained efficacy beyond this period. After injection of CRISPR/Cas9 with gRNA, at E15.5 and P1, the UBE3A protein was restored to 37–40% across multiple brain regions at P90, and treated mice showed partial phenotypic rescue [16]. Another CRISPR/Cas9 study showed that newborn intracerebroventricular (ICV) injection with AAV increased UBE3A protein expression up to 25–31% in total brain lysates in AS mice and partially rescued behavior deficit at 8 weeks of age [17]. Delivery of Dead Neisseria meningitidis Cas9 (dNmCas9/Nmg15-AAV) targeting the Snord115 region also showed elevated paternal Ube3a expression and partially rescued the hindlimb clasp at 4 to 12 weeks of age [21]. Delivery of CRISPR/Cas13 and gRNA targeting UBE3A-ATS with AAV (ICV or intravenous, IV) at P1 in AS mice showed that the paternal UBE3A protein expression reached 35.9% in the cortex tissues and 41.6% in the hippocampus tissues of AS mice at 4 weeks and alleviated AS-related symptoms, such as motor function impairment at 13–15 weeks of age [20]. CRISPR/Cas9 and Cas13 strategies have shown promising efficacy in unsilencing the paternal Ube3a allele and offer potential for long-term or even permanent correction. However, these approaches have thus far achieved only partial improvement in behavioral deficits and raise concerns regarding possible Cas9 genome integration events and immune responses against the Cas proteins. After ICV injection, ASO targeting the 3’UTR of UBE3A-treated AS mice showed reduced Ube3a-ATS and partial restoration of the UBE3A protein 35–47% across brain regions and recovery of contextual freezing comparable to wild-type mice at 4 weeks after injection [13]. Another ASO study showed UBE3A reinstatement in the brain up to 74% of WT levels and full rescue of sensitivity to audiogenic seizures after ICV injection of ASO at P1 or P21 in AS mice [19]. Although the precise mechanisms by which disruption of Ube3a-ATS activates the paternal Ube3a allele are poorly understood, preliminary results from an ongoing ASO clinical trial (KIK-AS trial, ClinicalTrials.gov Identifier: NCT04259281) have reported positive findings for safety, tolerability, and clinical benefit, including cognition, communication, behavior, and sleep, according to a press release. The Aspire trial (ClinicalTrials.gov Identifier: NCT04259281) is a pivotal Phase 3 study that was launched following the positive results from the KIK-AS trial. The treatment under clinical investigation requires quarterly intrathecal re-administration of ASOs.

Short hairpin RNA (shRNA) is an engineered RNA molecule that mediates gene silencing through the RNAi pathway. shRNAs are typically delivered into cells using viral vectors, such as lentivirus or AAV, where they are transcribed from DNA templates under the control of RNA polymerase II or III promoters. In the cytoplasm, the enzyme Dicer cleaves the hairpin into a double-stranded siRNA, of which the guide strand is incorporated into the RNA-induced silencing complex (RISC) [22]. The RISC, guided by sequence complementarity, targets and degrades the corresponding messenger RNA (mRNA), effectively silencing gene expression [23,24,25]. Therapeutically, shRNA offers several advantages. Unlike synthetic siRNAs, which are transient and require repeated administration, shRNAs provide stable and long-lasting gene knockdown due to their continuous intracellular expression [26,27]. This makes them particularly suited for chronic diseases or genetic disorders where sustained suppression of a pathogenic gene is necessary [28,29,30]. Additionally, shRNA vectors can be tissue-specific or inducible, allowing greater control over gene silencing. However, shRNA-based therapies have been reported to exhibit several limitations, including saturation of the endogenous RNA interference machinery, induction of inflammatory responses, dorsal root ganglion (DRG) toxicity, and tissue injury, as well as the generation of heterogeneous small RNA species [31,32,33,34]. Despite these concerns, advancements in vector design and promoter selection have significantly improved the specificity and safety of shRNA-based therapies [35,36]. As such, shRNA holds promise for the long-term modulation of disease-associated genes in areas such as cancer, neurogenetic disorders, and viral infections [37,38,39,40,41]. To date, no shRNA-based therapy has been approved by the FDA. However, there is notable progress in clinical research involving shRNA (Clinicaltrials.gov Identifier: NCT04613557, NCT03208556) and miRNA (Clinicaltrials.gov Identifier: NCT0543017, NCT04120493). In theory, viral administration of an anti-UBE3A-ATS construct could result in permanent paternal expression of UBE3A and amelioration of AS symptoms from a single dose.

In this study, we screened 20 mouse and 47 human siRNA candidates to evaluate their ability to downregulate UBE3A-ATS in mouse primary neurons and human SH-SY5Y neuroblastoma cells differentiated with retinoic acid. From these, we selected the top candidates targeting mouse neurons and human cells (one mouse and two human). The selected RNAi sequences, incorporated into lentiviral vectors for shRNA expression, were then used to examine UBE3A-ATS suppression and UBE3A transcript and protein induction, in comparison with known controls—Topotecan and ASO.

2. Materials and Methods

2.1. Animal Procedures

All animal studies were performed in accordance with the Institutional Animal Care and Use Committee (IACUC) guidelines at UT Southwestern Medical Center. All the animals of wild-type, B6.129S7-Ube3a^tm2Alb^/J (Paternal YFP-tagged UBE3a mouse model, Ube3a^p-YFP/m+^) [10], and B6.129S7-Ube3atm1Alb/J (Jackson Laboratory, #016590, AS disease mouse model, Ube3^p+/m−^) were maintained on a C57BL/6 background. The animals were housed under standard conditions in a pathogen-free facility. All animal procedures adhered to National Institutes of Health guidelines and were approved by the IACUC of UT Southwestern Medical Center (approved animal protocol number: 2018-102619).

2.2. Cell Culture

SH-SY5Y cells were used for screening of human siRNA candidates. SH-SY5Y cells were maintained with DMEM/F12 including 10% FBS (Fisher, 10-082-147, Grand Island, NY, USA) and penicillin–streptomycin (Fisher, 15140122, USA). Retinoic acid (RA) (10 μM) (Sigma, R2625, St. Louis, MO, USA) was used to induce differentiation of human SH-SY5Y neuroblastoma cells into neuron-like cells [42,43]. Cells treated with RA were maintained with Neurobasal media (Fisher, 21-103-049, Waltham, MA, USA) including B27 (Fisher, 17504044, Waltham, MA, USA), GlutaMAX (Fisher, 35-050-061, Grand Island, NY, USA), and penicillin–streptomycin. SH-SY5Y cells were activated for 6 days before transfection, and cells were harvested at the indicated time.

2.2.1. iPSC Lines

The human induced pluripotent stem cell (iPSC) lines utilized in this study were obtained from the laboratory of Dr. Yong-Hui Jiang at Yale University. The iPSCs used in this study were derived from a healthy control line, VJ9612 #5 (Clone No. 5). This line was generated and characterized by the Yale Stem Cell Center and qualified by the Yale Pathology Laboratory. All experiments were conducted using cells within passage 9 to 12 windows. Mycoplasma testing was performed by the Yale Stem Cell Center Core Facility and independently verified by the Yale Pathology Laboratory. All results were negative. The VJ9612 #5 iPSC line carries a normal karyotype and exhibits no known imprinting-related abnormalities. Human iPSC lines were established from Yale IRB-approved human material. Upon receipt, the iPSC lines were transitioned to feeder-free culture conditions, following the manufacturer’s protocol (StemCell Technologies, Cambridge, MA, USA), and passaged an additional two times to ensure complete removal of feeder cells. The feeder-free iPSCs were maintained on Matrigel-coated plates with mTeSR™ Plus medium (StemCell Technologies, Cambridge, MA, USA) and passaged every 3–4 days at a ratio of 1:6–1:10 using Versene (Invitrogen, Waltham, MA, USA). The cells were cultured in a humidified incubator at 37 °C with 5% CO_2_ atmosphere.

2.2.2. iPSC-Derived Neuronal Induction

iPSCs were differentiated into neural progenitor cells (NPCs) following a monolayer dual SMAD inhibition protocol utilizing a STEMdiff™ SMADi Neural Induction Kit (StemCell Technologies, Cambridge, MA, USA). iPSCs were dissociated with Accutase to obtain 2 × 10^6^ single cells per well, which were plated onto Matrigel-coated 6-well plates. Neural induction media were refreshed daily for 8 days prior to each passage. After three passages, the NPCs were detached with Accutase and plated onto 6-well plates coated with 15 µg/mL poly-L-ornithine (PLO; Sigma-Aldrich, St. Louis, MO, USA, P3655) and 10 µg/mL Laminin (Sigma-Aldrich, L2020, USA) at a density of 105,000 cells per cm^2^ to induce differentiation into neural precursor cells (NPreCs). STEMdiff Forebrain Neuron Differentiation Medium (StemCell Technologies, 08600, Cambridge, MA, USA) was replaced daily for up to 7 days until the cultures reached 80–90% confluency. To generate iPSC-derived neurons, the NPreCs were dissociated with Accutase and plated at 126,000 per cm^2^ in STEMdiff Forebrain Neuron Maturation Medium (StemCell Technologies, 08605, Cambridge, MA, USA). Neuronal maturation was further promoted by media changes every other day until the experiment was carried out.

2.2.3. Mouse Primary Neurons

Primary neurons were isolated from the hippocampus and cortex of postnatal day 0–2 (P0–P2) male and female pups. Each cell preparation was performed from individual pups. The cell isolation procedure was adapted from the method described by Beaudoin et al. [44] with minor modifications, as detailed below. Cells from the mouse hippocampus and cortex were enzymatically dissociated using trypsin, and the dissociated tissue was carefully layered over a 4% bovine serum albumin (BSA) cushion, followed by centrifugation at 300× g for 7 min at room temperature [45]. During cell preparation, genotyping was performed. Cells with the same genotype were then pooled and plated for subsequent experiments. The cell pellet was resuspended in a plating medium and seeded onto plates with 0.5 mg/mL Poly L-Lysine (Sciencell Research, Carlsbad, CA, USA). Four hours post-plating, the plating medium was replaced with maintenance medium. The cultures were maintained until day in vitro 4 (DIV4), with half of the maintenance medium refreshed every other day prior to transfection.

2.3. siRNA Design

Multiple 21-nucleotide siRNA duplexes, targeting both mouse Ube3a-ATS and human UBE3A-ATS, were designed using the Genetic Perturbation Platform developed by the Broad Institute. To minimize off-target effects, candidate sequences were evaluated using the siRNA Sequence Probability-of-Off-Targeting Reduction (sispotr) algorithm (sispotr.icts.uiowa.edu; accessed March 2021). The most suitable siRNAs were selected based on their lowest Probability-of-Off-Targeting Sites (POTS) values against human and mouse transcriptomes (Table 1). Off-target information is available in the Supplementary Information. Universal control (UNC) siRNA was purchased from Millipore Sigma (SIC 001, St. Louis, MO, USA). The sequence of UNC siRNA is proprietary and not disclosed by the manufacturer.

2.4. Cell Transfection with siRNA

Mouse primary neurons were transfected with siRNAs using Lipofectamine RNAiMAX (Invitrogen, 13778-075, Carlsbad, CA, USA), as described. Cells (250,000 cells) were seeded in 24-well plates 24 h prior to transfection. On Day 4, the siRNA duplex was diluted in 50 µL/100 µL Opti-MEM^®^ I Reduced Serum Medium and combined with 1 µL Lipofectamine™ RNAiMAX diluted in 50 µL/100 µL Opti-MEM. After 10–20 min incubation at room temperature, the complexes were added to the cells to a final volume of 600 µL/1.2 mL and an RNA concentration of 5 nM or 10 nM, depending on the experiment. The cells were incubated at 37 °C, 5% CO_2_.

2.5. Transduction of iPSC-Derived Neurons with Lentivirus

Non-integrating LVs (NILVs) or integrating lentiviral vectors (ILVs) were custom-designed and manufactured by VectorBuilder (Chicago, IL, USA). These U6-based shRNA knockdown constructs included the EGFP reporter driven by the JeT promoter. The shRNA target sequences for both mouse and human candidates are listed in Table 2. Viral transduction was performed using a range of multiplicity of infection (MOI). The culture medium was replaced after 24 h post-transduction, and the cells were maintained until harvest at the indicated time points.

2.6. Quantitative Real-Time Reverse Transcription (qRT)-PCR

Total RNA extraction and cDNA synthesis were carried out using a TaqMan^®^ Gene Expression Cells-to-CT ™ Kit (Thermo, AM1728, Waltham, MA, USA). qPCR was performed with listed primers; Snrpn (F: 5′-TGTGATTGTGATGAGTTCAGGAAGA-3′, R: 5′-ACCAGACCCAAAACCCGTTT-3′, Probe: 5′-CAAGCCAAAGAATGCAAAACAGCCAGAA-3′), Snord116 (F: 5′-TGTTGCTGACTTGCCCTAG-3′, R: 5′-GTTCGATGGAGACTCAGTTGG-3′, Probe: 5′-AAACATGCAGAGGAAATGGCCCC-3′), Snord115 (F: 5′-CAGCAATCCCTCTCCAGTTC-3′, R: 5′-AAGGTGGCATGTGAGATGAC-3′, Probe: 5′-TGTGACCATTCCTACTCTGAGCCAGTT-3′), Ube3a-ATS (Mm03455899_m1/4426961), Ube3a (Mm00839910_m1/4331182), Eif4a2 (Mm01730183_gH/4448484), UBE3A-ATS (Hs01372957_m1/4426961), UBE3A (Hs00166580_m1/#4351370), and EIF4A2 (Hs00756996_g1/4448484). Data are presented as the relative quantity of targets, normalized with respect to internal control, and relative to the calibrator control sample.

2.7. Western Blot Analysis

The cells were lysed in RIPA buffer (25 mM Tris-HCl–pH 7.6, 150 mM NaCl, 1% NP-40, 1% sodium deoxycholate, 0.1% sodium dodecyl sulfate) supplemented with 1X Protease Inhibitor Cocktail (Cell Signaling Technologies, Danvers, MA, USA) (150 μL in 5 × 10^5^ cells). The cell lysate was clarified by centrifugation (12,000× g) for 10 min at 4 °C. Then, 30 μL of 4X sample buffer (including β-mercaptoethanol) was added to 90 μL of the cell lysates. An equal volume (15–20 μL) of cell lysates was separated by SDS-PAGE and transferred onto polyvinylidene difluoride (PVDF) membranes using Trans-Blot Turbo Mini 0.2 µm PVDF Transfer Packs (BioRad, Hercules, CA, USA). The membranes were blocked for 30 min at room temperature in EveryBlot Blocking Buffer (BioRad, Hercules, CA, USA). Primary antibody incubation was performed overnight at 4 °C. Following incubation, the membranes were washed three times with Tris-buffered saline with 0.1% TWEEN 20 (TBS-T) and subsequently incubated with the appropriate secondary antibody for 1 h at RT: Anti-GFP (Millipore-Sigma, 06-896, USA), anti-UBE3A (Sigma, SAB1404508, USA), hFAB™ Rhodamine Anti-Tubulin antibody (Biorad, 12004165, Hercules, CA, USA), Goat Anti-Mouse IgG StarBright Blue 700 (Biorad, 12004158, Hercules, CA, USA), or Goat Anti-Rabbit IgG StarBright Blue 700 (Biorad, 12004161, Hercules, CA, USA). Following secondary antibody incubation, the membranes were washed three times with TBS-T, and immunoreactive bands were visualized using the ChemiDoc™ MP Imaging System (Bio-Rad). The exposure time was set with the optimal autoexposure setting for each fluorescence. Band intensities were quantified using ImageJ software version 1.54j (National Institutes of Health, Bethesda, MD, USA) (protein levels were normalized against Tubulin).

2.8. Immunofluorescence

The cells were fixed with 4% paraformaldehyde for 10 min at room temperature. The cells were washed with ice-cold 1X PBS and blocked with 10% normal goat serum (NGS) (Jackson Immuno Research Laboratories, 005-000-121, West Grove, PA, USA) in 1X PBS with 0.3% Triton X-100. Sections were incubated with primary antibodies in 1X PBS with 0.3% Triton X-100 at 4C overnight. The cells were washed with 1X PBS with 0.3% Triton X-100 and incubated with secondary antibody in 5% NGS 1X PBS with 0.3% Triton X-100 for 1 h at room temperature. The cells were washed with 1X PBS with 0.3% Triton X-100 thoroughly, incubated with DAPI (Sigma, D9542, USA), and mounted (Vector Laboratories, H-1900-10, USA). GFP (Millipore-Sigma, 06-896, USA), UBE3A (Sigma-Aldrich, SAB1404508, USA), β-tubulin III (Stem Cell Technologies, 60092.1, Cambridge, MA, USA), Alexa Fluor 594 (Thermo, A21135, USA), and Alexa 647 (Invitrogen A21240, USA) were used. For siRNA screening data, the intensity of the YFP was measured using ImageJ software (NIH, Bethesda, MD, USA).

2.9. Statistics

GraphPad Prism 10 (Boston, MA, USA) was used to calculate significance. The threshold for significance was defined as p ≤ 0.05. Throughout, the following convention is used to convey significance: * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001, and **** p ≤ 0.0001. Post hoc adjusted p-values are reported when ANOVAs are used. Two-way ANOVA with repeated measures was used for the analyses of all qPCR data, followed by Šidák’s multiple-comparison test using adjusted p-values and 95% confidence intervals of the differences, with α = 0.05. Ordinary one-way ANOVA with repeated measures was used for the analyses of Western blot data, followed by Tukey’s multiple-comparison test using adjusted p-values and 95% confidence intervals of the differences, with α = 0.05. The size of biological replicates (N) is specified for each dataset in the corresponding figure legends.

3. Results

3.1. Inhibition of Ube3a-ATS by Transfection of siSnord115 Restores Ube3a Expression in Mouse Primary Neurons

Overall, 20 siRNA oligonucleotides were designed using the GPP Web Portal (the TRC shRNA design process) from the Broad Institute to target murine Ube3a-ATS, which is the Snord115 region. The 20 siRNAs were converted to shRNA sequences for screening the candidate to activate paternal Ube3a-YFP expression. Universal control (UNC) siRNA was used as a negative control, and ASO A (5 uM) was used as a positive control [13]. Primary neurons from Ube3a^P-YFP/m+^ mice were cultured and transfected with siRNAs and siUNC (5 nM, 48 h), and we determined the increased paternal YFP-tagged Ube3a (Ube3a-YFP) signal through YFP intensity (Figure S1). From the screening, a top candidate (siSnord115, intrinsic score 13.2 out of 15 from the Broad Institute GPP Web Portal) was identified (Table 1), which has multiple binding sites (102 binding sites are indicated in Figure 1a) to increase Ube3a-YFP expression (Figure 1b). Next, we investigated RNA expression from Snrpn, Snord116, and Snord115, which are processed from the same precursor transcripts as Ube3a-ATS and are Prader–Willi Syndrome (PWS) critical genes [46]. Topotecan, which was used as a positive control, inhibits the transcription of Ube3a-ATS by stabilizing R-loops at the paternal Snord116 locus [15]. Upon the treatment with Topotecan, Snrpn, Snord116, Snord115, and Ube3a-ATS were significantly inhibited compared to vehicle treatment. ASO-A targeting Ube3a-ATS downstream of the Snord115 efficiently inhibited Ube3a-ATS but did not affect Snrpn and Snord116 transcript levels compared to ASO-ctrl. siSnord115 significantly reduced target RNA Snord115 (folds 0.49 ± 0.09) and reduced Ube3a-ATS (folds 0.65 ± 0.02). Its efficacy was lower compared to Topotecan (folds 0.008 ± 0.004) or ASO-A (folds 0.16 ± 0.004) (Figure 1c, Table S1). Furthermore, we confirmed Ube3a-YFP protein expression increased with siSnord115 transfection compared to siUNC (Figure 1d, Figure S2). The efficacy of siSnord115 was further confirmed in disease-relevant Ube3^p+/m−^ mouse primary neurons. The downregulation pattern of RNAs was comparable across cell lines. Compared to siUNC, siSnord115 reduced Ube3a-ATS (folds 0.68 ± 0.02). Ube3a RNA levels also showed statistically increased levels (folds 1.17 ± 0.05) (Figure 1f, Table S2). However, analysis of protein expression revealed a significant elevation in UBE3A levels 72 h after siSnord115 treatment (Figure 1g, Figure S3).

3.2. Inhibition of Ube3a-ATS by Transduction of LV-shSnord115 Restores Ube3a Expression in Ube3ap+/m− Mouse Primary Neurons

Lentiviral vectors are a tool for gene transfer and can be used to transduce primary neurons. To test the efficacy of the shRNA system in primary neurons, lentiviral vector-shSnord115 (LV-shSnord115) and LV-shscramble were generated with the U6 promoter, and EGFP (derived from the JeT promoter) was inserted as a transduction marker. LV-shSnord115 was transduced at the indicated Multiplicity of Infection (MOI) in Ube3^p+/m−^ mouse primary neurons. The expression levels of Snord116, Snord115, and Ube3a-ATS were significantly decreased (>MOI = 0.2), whereas Ube3a expression was increased in a dose-dependent manner (Figure 2a, Table S3). Increased Ube3a protein expression was confirmed in both Ube3^p-YFP/m+^ and Ube3^p+/m−^ mouse primary neurons by Western blot (Figure 2b,c, Figures S4 and S5) and immunofluorescence staining (Figure 2d). Taken together, targeting Snord115 with shRNA can induce paternal Ube3a expression by suppressing Ube3a-ATS in mouse primary neurons.

3.3. Inhibition of UBE3A-ATS by Transduction of LV-shSNORD115 Restores UBE3A Expression in Healthy Human iPSCs

Based on the successful proof-of-concept results in mice using shRNA, we next sought to identify shRNAs that would be effective in a human neuronal model. Overall, 47 shRNA oligonucleotides were designed using the GPP Web Portal (The TRC shRNA design process) from the Broad Institute to target the downstream of Ube3a, which includes human SNORD115. For screening the candidates, we used SH-SY5Y, which can be differentiated into more neuron-like phenotype cells by treatment with retinoic acid (RA) [42,43]. Before initiating the screening, we examined the effect of retinoic acid (RA) treatment on UBE3A-ATS expression in SH-SY5Y cells. Quantitative RT–PCR analysis showed that at 6 days after RA treatment, the ΔΔCt value for UBE3A-ATS decreased from −1.06 (untreated) to −6.83 (RA-treated), indicating a greater than 50-fold increase in UBE3A-ATS transcript levels following RA-induced neuronal differentiation (Table S4).

Forty-seven synthesized siRNAs (10 nM each) were transfected into SH-SY5Y cells 6 days after RA induction, and then the UBE3A and UBE3A-ATS expression was measured 24 h post-transfection (Figure 3a). We selected 2 out of the original 47 candidates targeting the SNORD115 region for further assessment, which shows consistent downregulation in three independent experiments (Figure 3b). LV-shSNORD115 #1, #2 (intrinsic score 13.2 out of 15 from the Broad Institute GPP Web Portal), and LV-shscramble vectors were designed with the U6 promoter driving shRNA expression and inclusion of a JeT-EGFP expression cassette to label transduced cells. Human iPSCs from healthy donors were differentiated into neurons; LV addition occurred 8 days after starting maturation, and then, the cells were harvested 3 days after transduction (Figure 3c). Transduction efficiency was confirmed with GFP expression, and matured neurons were confirmed with β-tubulin III labeling by IF (Figure 3f). LV-shSNORD115 #1 has 1 binding site, and LV-shSNORD115 #2 has 11 multiple binding sites in the SNORD115 region. Both LV-shSNORD115 #1 (at MOI = 10) and #2 (>MOI = 2) significantly decreased UBE3A-ATS comparable to Topotecan (3uM) treatment (Figure 3d,e, Table S5). LV-shSNORD115 #1 showed a significant increase in UBE3A expression (>MOI = 5) (Figure 3d,e, Table S5). LV-shSNORD115 #1 increased SNORD116 expression (>MOI = 5) while LV-shSNORD115 #2 decreased SNORD116 expression at the MOI of two (Figure S6). There was no significant SNORD116 reduction after LV transduction. Taken together, we confirmed that the shRNA targeting SNORD115 can reduce UBE3A-ATS expression in healthy human iPSC-derived neurons.

4. Discussion

In this study, shRNA-mediated inhibition of Ube3a-ATS by targeting Snord115 effectively restored Ube3a expression. siSnord115 was identified as a lead candidate that robustly suppresses Ube3a-ATS and induces paternal UBE3A in mouse primary neurons. Consistently, LV-shSnord115 reduced Ube3a-ATS in a dose-dependent manner and increased UBE3A RNA and protein levels. Screening in human iPSC-derived neurons further identified LV-shSNORD115 #1 and #2 as effective candidates that lower UBE3A-ATS, with potency comparable to Topotecan. Taken together, these findings support siRNA/shRNA-mediated suppression of UBE3A-ATS as a viable strategy to reactivate paternal UBE3A expression.

RNAi-mediated therapy using shRNA has been used and evaluated in various mouse disease models, including heart disease, neurodegeneration, cancer, and metabolic disorders [22]. shRNA-based RNAi therapy has shown strong potential for long-term gene silencing. Delivered via plasmids or viral vectors, shRNA offers sustained and sequence-specific RNA silencing, targeting both mRNAs and long noncoding RNAs (lncRNAs). Like ASOs, shRNAs harness the cell’s RNA silencing machinery but differ in their processing and delivery methods.

While shRNA therapy faces challenges—including off-target effects, immune responses, disruption of endogenous miRNA function, and delivery barriers—it remains a promising strategy for regulating disease-associated gene expression, particularly in genetic disorders [17,36,37,38,39]. In this study, we used the U6 promoter to express shRNA expression. It has been reported that excessive shRNA expression driven by the U6 promoter carries several potential limitations, including saturation of endogenous RNAi machinery, induction of inflammatory responses, DRG toxicity and tissue injury, and generation of heterogeneous small RNAs [31,32,33,34]. To mitigate these risks, safety can be enhanced by embedding shRNAs in natural miRNA scaffolds for accurate processing, employing weaker or tissue-specific promoters (Pol II promoters; hSyn, CamKIIα, PGK, EF1α) to prevent RNAi machinery saturation, and optimizing guide design to minimize off-target seed effects [36,47,48,49,50,51]. As we show the efficacy of LV-shSnord115/shSNORD115 in vitro, this can be applied to AAV vectors, which are widely used for delivering shRNA in vivo due to their high transduction efficiency, tissue specificity, and long-term expression [52]. AAV-mediated shRNA delivery has been successfully applied in preclinical models of neurodegeneration, retinal disease, metabolic disorders, and cancer [39,53,54,55,56]. We anticipate that AAV-mediated shRNA delivery will be largely irreversible, as AAV vectors persist in target cells predominantly as episomal DNA. Because the primary target cells in this study are post-mitotic neurons, vector dilution through cell division is minimal. Consequently, AAV-mediated shRNA expression is expected to be long-lasting and potentially sustained for the lifetime of the treated neurons.

Angelman syndrome (AS) is caused by maternal loss of gene function at 15q11–q13, leading to deficient neuronal UBE3A expression, and comprises five molecular classes. The most common subtype is a maternal 15q11–q13 deletion (Class I; ~70–85%), followed by intragenic UBE3A mutations (Class IV; ~10–30%). Less frequent causes include paternal uniparental disomy (pUPD) (Class II; ~2–5%) and imprinting defects (Class III; ~3–5%). A small group of patients (Class V) shows typical AS features without an identifiable genetic cause [57]. This marked genetic heterogeneity underscores the importance of genotype-based patient stratification when evaluating therapeutic approaches such as paternal UBE3A unsilencing. In this context, UBE3A-ATS-targeting shRNA therapy is most suitable for patients with an intact but epigenetically silenced paternal UBE3A allele, particularly those with pUPD (Class II) or imprinting defects (Class III), with individualized consideration for UBE3A mutation carriers (Class IV). Although paternal UBE3A can be technically reactivated in Class I (maternal deletion) AS, the anticipated therapeutic benefit is limited, as pathology in this group reflects loss of multiple genes within 15q11–q13, rather than UBE3A deficiency alone.

SNRPN, SNORD115, and SNORD116 are important genes in Prader–Willi Syndrome (PWS), which arise primarily due to the loss of paternally expressed genes in this region. Deletion of SNORD116 alone has been shown to cause PWS-like symptoms [58,59,60,61,62]. In this study, siSnord115 and LV-shSnord115 show a significant reduction in Ube3a-ATS along with a reduction in Snord115 and Snord116, though not Srnpn (Figure 1d,f and Figure 2b). Although our candidate shows promising results in increasing UBE3A expression, a mild reduction in Snord116 in neuron cells may still carry potential risks, which may introduce unwanted side effects. One potential mechanistic explanation for the partial reduction in SNORD116 observed following shRNA targeting of SNORD115 is that both snoRNAs are processed from the same SNHG14 precursor transcript, which has been reported to form R-loops within the SNORD116 region. However, in this study, we did not directly assess downstream phenotypic consequences or transcriptomic alterations resulting from reduced SNORD116 levels. In particular, the transcriptomic effects associated with SNORD116 reduction were not evaluated, representing an important limitation of the current work, especially in the context of translational applicability. Addressing these potential consequences will be a critical next step and will be pursued in future studies using animal models treated with this therapeutic approach. RNA interference directed against SNORD115 may potentially influence the structural stability of the precursor transcript or interfere with RNA:DNA hybrid formation that is thought to support UBE3A-ATS elongation. In this hypothetical model, perturbation of snoRNA processing could impair antisense transcript elongation and thereby facilitate reactivation of the paternal UBE3A allele. This interpretation is conceptually consistent with the R-loop stabilization mechanism proposed in the context of topoisomerase inhibition, but it is not directly tested in the present study and should therefore be considered speculative. Although SNORD115 is not considered a critical gene for PWS, it nevertheless plays an important role in neuronal differentiation and regulates serotonin receptor (5-HT2C) expression and signaling [58,63,64,65,66,67]. A previous study demonstrated that CRISPR–Cas9/or dCas9-mediated disruption of Snord115 led to paternal Ube3a unsilencing throughout the nervous system, thereby rescuing some behavioral phenotypes in the Angelman syndrome (AS) mouse model at 4–28 weeks of age [16,21]. Therefore, further studies are warranted to carefully evaluate the safety and specificity of shRNA-mediated targeting of SNORD115 in the Angelman syndrome model, given their complex regulatory roles and potential off-target effects on imprinted gene networks.

In our siRNA screening using mouse primary neurons, only siRNAs targeting SNORD115 effectively reduced Ube3a-ATS expression, whereas candidates directed against the 3′UTR of UBE3A—an area commonly targeted in ASO-based therapeutic strategies—did not produce significant knockdown. In contrast, shRNAs directed against SNORD115 demonstrated robust efficacy, likely due to the presence of multiple repeated sequence elements that provide numerous potential binding sites, unlike the relatively unique sequence structure of the 3′UTR. These findings suggest that effective activation of paternal UBE3A expression via shRNA-mediated suppression may require multiple accessible binding sites within the target transcript to achieve sufficient knockdown efficiency. Although the 3′UTR region between Ube3a and Snord115 is considered an optimal target to minimize off-target effects on other PWS-related genes, our findings highlight the practical efficacy of targeting Snord115. Notably, several studies targeting the Snord115 region have not reported adverse effects in behavioral assessments, supporting its potential as a promising therapeutic target despite theoretical safety concerns [16,20,21,68]. Unlike the mouse study, we demonstrate that human shSNORD115 candidates effectively downregulate UBE3A-ATS expression in healthy human iPSC-derived neurons. We were unable to evaluate these candidates in AS patient-derived iPSCs, as neuronal differentiation under the same conditions used for healthy iPSCs was unsuccessful. Therefore, it remains a limitation of this study that the efficacy of the two human shSNORD115 candidates has not yet been validated in a disease-relevant or patient-derived neuronal model.

5. Conclusions

This study provides proof-of-concept evidence that inhibition of Ube3a-ATS using siSnord115 or LV-shSnord115 can induce paternal Ube3a expression in mouse primary neurons and that similar inhibition through LV-shSNORD115 reactivates UBE3A expression in healthy human iPSC-derived neurons. While these findings demonstrate promising in vitro efficacy, the observed reduction in Snord115 and Snord116 highlights the need for careful assessment of off-target and safety profiles, particularly given the potential for unintended modulation of imprinted genes. Further studies are warranted to evaluate durability, allele-specificity, and in vivo relevance, including the use of AAV-mediated shRNA delivery targeting SNORD115 to assess feasibility in animal models. While AAV vectors offer efficient and neuron-preferential gene delivery, future studies should evaluate their applicability in vivo through a comprehensive safety and feasibility assessment to ensure long-term safety and stable expression before therapeutic translation.

6. Patents

A provisional patent has been filed for the anti-UBE3A-ATS siRNAs/shRNAs and their use (with RKB, HRK, and SJG as co-inventors).