Lipid nanoparticle delivery of antisense gapmers attenuates pathology in a mouse model of facioscapulohumeral muscular dystrophy

Abstract

Facioscapulohumeral muscular dystrophy (FSHD) is a progressive muscle-wasting disease caused by aberrant activation of the DUX4 gene in skeletal muscle. Antisense oligonucleotides (ASOs) targeting DUX4 have shown therapeutic potential, but challenges related to systemic delivery and efficacy limit their clinical utility. In this study, we explored lipid nanoparticle (LNP)-mediated delivery of optimized locked nucleic acid (LNA) and 2′-O-methoxyethyl (MOE) gapmers to enhance their therapeutic potential. In FSHD patient-derived muscle cells, gapmers reduced DUX4 expression and improved myogenic differentiation indices related to nontreated controls. In vivo studies demonstrated that LNP encapsulation increased gapmer bioavailability in skeletal muscle tissues compared to bare gapmers while maintaining a favorable safety profile. A single systemic injection in iDUX4pA mice moderately reduced DUX4 levels without significant functional improvement. However, repeated weekly injections resulted in over 50% knockdown of DUX4, repression of its downstream targets, improved grip strength and treadmill performance, and amelioration of muscle pathology. This study indicates that LNP-mediated gapmer delivery could be a candidate strategy to improve skeletal muscle biodistribution in iDUX4pA mice. This work provides preliminary proof-of-concept for further preclinical optimization and additional safety and efficacy studies.

Graphical Abstract

Introduction

Facioscapulohumeral muscular dystrophy (FSHD) is a rare, slowly progressive, and disabling skeletal muscle disease ¹. Typically beginning in the teenage years, FSHD is characterized by muscle degeneration with a distinctive descending pattern of weakness, first affecting the face, shoulders, upper arms, and trunk muscles, followed by the lower extremities ². Recent epidemiological assessments estimate FSHD prevalence at ~4–12 per 100,000 (≈0.3–0.8 million affected worldwide), with geographic variability and likely underdiagnosis; it remains the third most common muscular dystrophy^3–6. The pattern of inheritance is usually autosomal dominant; however, about one-fourth of cases occur de novo ^7,8. FSHD’s clinical impact is highly heterogeneous, even among affected individuals within the same family ^9–12. The disease causes lifelong disability with significant physical limitations, with approximately 20% of patients forced to be wheelchair-bound ¹³. FSHD does not occur disproportionately based on sex, though females often experience a greater disease burden ¹⁴. Some patients, especially those with early-onset or infantile FSHD, may experience extra-muscular symptoms, e.g., hearing loss, retinal vasculopathy, and cognitive impairment ^15–17. Rarely, cardiovascular involvement is also reported ^18,19. Currently, no cure or specific treatment for FSHD is available in the clinic ^8,20. Management strategies focus on palliative care, including physical therapies and surgical correction of affected body parts (e.g., scapulopexy) to manage symptoms and improve quality of life; however, these approaches do not address the underlying pathology of FSHD.

Arguably, FSHD is considered the most complex genetic disorders known and its pathobiology was a subject of extensive research ^21,22. A consensus now holds that FSHD stems from the epigenetic de-repression of the double homeobox 4 (DUX4) gene in skeletal muscle ^23,24. DUX4, located in the subtelomeric region of chromosome 4 (4q35), is typically expressed during early embryogenesis but becomes highly toxic in post-embryonic somatic cells ^23,25. On chromosome 4q35, there is a microsatellite repeat region, called the D4Z4 repeat region, that is typically 11–150 repeats long ^26–29. DUX4’s first two exons are found in each of these ~3.3 kb long D4Z4 units, while the third and final exon lies just beyond the last repeat. The distance between these exons, combined with tight DNA packaging and heavy methylation, keeps DUX4 heavily repressed ^29–31. However, in individuals with FSHD, this repression is lost, resulting in DUX4 activation in as few as 1:200–2000 muscle cell nuclei, contributing to the disease’s pathology ^32–34. FSHD is classified into two primary forms: FSHD1 and FSHD2. In FSHD1 (~95%), the D4Z4 array contracts to ≤10 repeats ^35–37. Importantly, this contraction leads to disease only if the 4qA haplotype, which perhaps through some transposon activity contains a polyadenylation site (PAS), is present, as this stabilizes the DUX4 mRNA ^38,39. In FSHD2 (~5%), D4Z4 length is ≥10 ^35,40; however, mutations in epigenetic regulators (e.g., SMCHD1, DNMT3B) reduce repression ^25,41,42. Mutations in these genes may lead to a loss of repression, allowing DUX4 to be expressed even with a normal-length D4Z4 array. In rare cases, FSHD1 and FSHD2 may co-occur in the same individual ⁴³. Once activated, DUX4 disrupts a network of genes important for muscle function, leading to inflammation, fibrosis, fat infiltration, and eventual muscle wasting, the hallmarks of FSHD pathology ^33,44–50. DUX4 activation also impairs myogenic differentiation, membrane resealing, and increases oxidative stress sensitivity ^48,51.

Given the central role of DUX4 in FSHD pathogenesis, suppressing its expression is a major therapeutic focus ⁸. Several antisense oligonucleotide (ASO) strategies have been explored to block DUX4 translation, degrade its mRNA, or alter splicing to lower toxic protein levels ^52–60. Early efforts using 2′-O-methyl (2’OMe) ASOs targeting the DUX4 PAS or exon 3 splice sites successfully reduced DUX4-positive nuclei and muscle atrophy in FSHD myotubes ^52,55. One of these sequences was tested as an octa-guanidinium dendrimer conjugated phosphorodiamidate morpholino oligomer (vivo-PMO) in a mouse model with recombinant adeno-associated virus-mediated DUX4 expression, which, upon intramuscular administration, showed reduced DUX4 expression ⁵⁵. Two parallel studies indicated that phosphorodiamidate morpholino oligomers (PMOs) targeting the PAS in exon 3 effectively downregulated DUX4 in myotubes and mouse models ^53,54. Systemic vivo-PMO reduced DUX4 and mitigating pathology in ACTA1-MCM; FLExDUX4 mice ⁵⁹. We previously designed locked nucleic acid (LNA) and 2′-O-methoxyethyl (2′MOE) gapmer ASOs, achieving robust silencing of DUX4 and its targets in patient-derived muscle cells, and local knockdown in FLExDUX4 mouse muscle ^56,57.We have also shown that subcutaneous injection with 2’MOE gapmers effectively reduces DUX4 expression, improves muscle function, and decreases inflammation ⁶¹. Aside from these studies, constrained ethyl (cEt) gapmer targeting exon 1 has shown promise in improving muscle pathology, although functional improvements in response to this strategy seemed limited ⁵⁸. Additionally, 2′-N-methanesulfonyl-2′-amino-locked nucleic acid (ALNA[Ms]) gapmers lowered target-gene expression and prevented force decline after subcutaneous dosing in ACTA1-MCM/FLExDUX4 mice⁶⁰.

These studies highlight the therapeutic potential of ASOs for FSHD but enhancing their efficacy and achieving reliable systemic delivery for clinical use remain significant challenges ^58,62–64. Several strategies are being explored to improve efficacy and consistent delivery, including chemical modification, cell-penetrating peptide conjugation, nanoparticle-mediated delivery, and the use of endogenous or smart-material-based delivery vehicles ⁶¹. Among these, lipid nanoparticles (LNPs) have emerged as a leading platform due to their successful use in clinically approved nucleic acid-based therapeutics ⁶⁵. Clinically validated for siRNA and mRNA (e.g., patisiran; COVID-19 vaccines)^62,66,67, canonical LNPs comprise an ionizable lipid, cholesterol, distearoylphosphatidylcholine (DSPC), and a polyethylene glycol (PEG)-lipid. In these formulations, the ionizable lipid is important for payload complexation and endosomal escape, while cholesterol and helper lipids (e.g., DSPC) modulate membrane fluidity and particle stability, and the PEG-lipid regulates circulation half-life and prevents aggregation ^68–70. Post-systemic administration, LNPs adsorb apolipoproteins (e.g., ApoE) and are internalized via receptor-mediated endocytosis, while endosomal acidification protonates the ionizable lipid, promoting endosomal escape and cargo release ^68–70. The default tropism is hepatic; however, extrahepatic delivery can be tuned by ionizable-lipid structure, particle size/PEG density, and surface ligands ⁶⁹ Skeletal muscle remains challenging due to endothelial barriers and relatively low endocytic flux; recent reviews outline emerging strategies for extrahepatic LNP targeting and remaining gaps ⁶⁸. Importantly, application of LNPs to gapmer-type ASOs has largely been unexplored in preclinical.

Here, we investigate LNP-encapsulated gapmers for DUX4 suppression in FSHD. We hypothesized that LNP-mediated gapmer delivery could enhance bioavailability in skeletal muscle, improve therapeutic efficacy, and mitigate safety concerns associated with bare gapmers. We designed LNA and MOE gapmers targeting DUX4 exon 3 and created a mini-library of LNP formulations to assess their efficacy. Using a battery of techniques, we demonstrated that LNP-encapsulation significantly improves cellular phenotypes in FSHD patient-derived cells and gapmer uptake in skeletal muscle tissues, leading to enhanced DUX4 knockdown, and functional benefits in a mouse model of FSHD. Importantly, we demonstrate that multiple intravenous injections resulted in significant improvements in muscle mass, strength, and endurance, addressing limitations of previous ASO therapies. This study provides the first evidence that LNP-gapmers offer a viable strategy for systemic DUX4 silencing in FSHD, laying the groundwork for preclinical development and potential clinical application of this strategy.

Results

Gapmer treatment reduces DUX4 expression and improves cellular phenotype in FSHD patient-derived muscle cells

We designed three LNA, e.g., LNAx, LNAy, and LNAz, and three MOE, e.g., MOEx, MOEy, and MOEz, gapmers targeting exon 3 of the DUX4 gene (Figure 1A). These gapmers were tested on FSHD patient-derived myotubes at different doses. To establish a baseline, we first profiled DUX4 expression at different time points in growth condition and differentiation conditions in healthy (WS234) and patient-derived (WS229) immortalized myoblasts. RT-qPCR analysis confirmed that DUX4 mRNA was undetectable in patient-derived myoblasts under growth conditions (Figure S1). Under differentiation conditions, DUX4 expression in FSHD patient-derived myotubes increased progressively, peaking at day 13 post-differentiation. The expression levels at post-differentiation days 14, 15, 18, and 21 were not statistically different from day 13.

Figure 1: DUX4 knockdown efficacy, cellular phenotype improvement, and in vitro safety assessment post-treatment with LNA and MOE gapmers.

(A) Relative positioning of LNA (deep blue) and MOE (light green) gapmers across exon 3 of the DUX4 gene. (B) Schematic representation of the experimental workflow for gapmer screening in patient-derived cells. Day −3: revival of myoblast cells; Day 0: seeding of myoblasts; Day 1: induction of differentiation; Day 14: treatment with gapmers (post-differentiation day 13); Day 15: RNA harvesting and analysis (post-differentiation day 14); Day 19: fixation and immunostaining (post-differentiation day 18). Cell-level toxicity in terms of apoptosis, cytotoxicity, and vitality relative to healthy cells was assessed 4 hours, 12 hours, 24 hours post-treatment. (C–E) Relative DUX4 expression levels in FSHD patient-derived myotubes post-treatment with LNA and MOE gapmers at 100 nM (C), 10 nM (D), and 1 nM (E) as determined by RT-qPCR. Each black dot represents a biological replicate (Total N = 5–9). (F) Representative immunostaining images of healthy control and FSHD patient-derived myotubes 4-days post-treatment with LNAx, LNAz, MOEx, and mock gapmers at 100 nM and 10 nM dose. Cells were immunostained with dystrophin (green, Alexa Fluor 488) and DAPI (blue). Scale: 100 μm (G) Quantification of myogenic fusion index (MFI) from the immunostaining images shown in F (n = 15–19). (H–I) Assessment of cell toxicity parameters, including apoptosis, cytotoxicity, and cell vitality, following treatment with 10 nM (H) and 100 nM (I) of gapmers (n = 5). Statistics, one-way ANOVA with Tukey’s multiple comparisons test; *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. Data are represented as mean ± standard error of mean. NT, non-treated, RNAiMAX, Lipofectamine RNAiMAX (a commercially available transfection reagent).

Since DUX4 expression plateaued at post-differentiation day 13, we chose this time point for screening our gapmers (Figure 1B). RT-qPCR analysis demonstrated that all gapmers effectively knocked down DUX4 transcripts. At a 100 nM dose, we observed a near-complete knockdown of DUX4 transcripts (Figure 1C). A dose-dependent knockdown effect was observed, with significant reductions (86.5–92.1%, compared to non-treated) at a 10 nM dose, while no significant reduction was observed at a 1 nM dose (Figures 1D, 1E). Moreover, the gapmers significantly reduced the expression of key DUX4 downstream genes, including ZSCAN4, TRIM43, MBD3L2, PRAMEF2, LEUTX, and KHDC1L in a dose-dependent manner (Figure S2) ^46,50,71. LNAx and LNAz among the LNA gapmers, and MOEx among the MOE gapmers, performed most efficiently in knocking down DUX4 expression and reducing the expression of DUX4’s downstream genes (Figures 1C–1E, S2). Also, these gapmers performed better compared to previously published designs, showing slightly enhanced efficacy in knocking down DUX4 transcripts (Figures 1C–1E, S3). In terms of cellular phenotypes, treatment with LNAx, LNAz, and MOEx significantly enhanced the myogenic fusion index in FSHD patient-derived cells, as indicated by the increased number of nuclei per syncytial myotubes relative to that of the non-treated cells, suggesting a positive impact on muscle cell differentiation and regeneration (Figures 1F, 1G). This was further supported by the finding that the expression of myogenic markers, e.g., myogenin, myomaker, and myomixer, were increased post-treatment (Figure S2). In addition, gapmer treatment at 10 nM dose did not negatively impact cell vitality or induce apoptosis in FSHD myotubes, while treatment at 100 nM dose showed signs of reduced vitality, as well as increased cytotoxicity and apoptosis levels (Figure 1H, 1I, S4).

These findings demonstrate both LNA and MOE gapmers are highly effective in significantly reducing DUX4 transcript levels, along with downstream gene expression, in vitro. Also, they promote myogenic differentiation in FSHD myotubes, consistent with previous studies ^56,57. However, higher doses of gapmers of both chemistries raise concerns about potential cytotoxic effects.

Lipid nanoparticle delivery enhanced the efficacy of the gapmers in vitro

Since 10 nM of gapmers produced a significant reduction of DUX4 expression without inducing cell-level toxicity, we aimed to further improve the efficacy of this dose using ionizable LNP delivery. Therapeutic benefit in FSHD is expected to correlate with the degree of knockdown of DUX4, therefore enhancing knockdown capability is critical. We hypothesized that LNP-mediated delivery of gapmers could enhance their efficacy while maintaining a non-toxic profile.

To identify the most effective LNP formulation for gapmer delivery into muscle cells, we conducted a comprehensive screening of ionizable lipids, creating a mini-library of eight LNP formulations (LNPa–LNPh). We initially tested these formulations in vitro to determine the optimal nitrogen-to-phosphate (N:P) ratio for gapmer delivery. The gapmer encapsulation efficiency by these LNP formulations increased with increasing N:P ratios (Figures 2A, S5). RT-qPCR analysis demonstrated that LNP-encapsulated gapmers improved DUX4 knockdown compared to untreated controls and commercial transfection agents (Figures 2B, S6). While increasing the N:P ratio enhanced knockdown efficacy, the effect plateaued at an N:P ratio of 4 (Figure 2B). Encapsulation with LNPa and LNPe seemed to make LNAx, LNAz, and MOEx outperform other formulations, while LNPb and LNPh ranked next to them (Figures 2B, S6). Also, treatment with LNPs lacking any gapmer payload did not alter DUX4 expression levels (Figure S6). However, N:P ratios of 4 or higher were associated with increased markers of cytotoxicity, apoptosis, and reduced cell vitality (Figure S7). In addition, LNP-gapmers at an N:P ratio of 3 improved cellular phenotypes, e.g., myogenic fusion index, to levels comparable to a 100 nM dose of unencapsulated gapmers (Figures 1F, 1G, 2C, 2D).

Figure 2: DUX4 knockdown efficacy and cellular phenotype assessment post-treatment with LNP-gapmers.

(A) Mean gapmer encapsulation efficiency into LNPs (LNPa–LNPh) at N:P ratio of 1–6 (n = 5). (B) Heatmap showing relative DUX4 expression levels in FSHD patient-derived myotubes following treatment with 10 nM LNAx, LNAz, MOEx, or mock (LNA) gapmers encapsulated in LNPa–LNPh at N:P ratios of 1–6, as determined by qRT-PCR. Each black dot represents a biological replicate (Total N = 4–6). (C) Representative immunostaining images of healthy control and FSHD patient-derived myotubes 4-days post-treatment with 10 nM LNAx, LNAz, MOEx, or mock gapmers encapsulated in LNPa, LNPb, LNPe, and LNPh at an N:P ratio of 3. Cells were immunostained with dystrophin (green, Alexa Fluor 488) and DAPI (blue). Scale: 100 μm. (D) Quantification of the myogenic fusion index (MFI) from the immunostaining images in panel C. The horizontal purple dotted line represents the average relative MFI achieved with 100 nM gapmers delivered via RNAiMAX (data from figures 1F and 1G; n = 15–19). Statistics, one-way ANOVA with Tukey’s multiple comparisons test; *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. Data are represented as mean ± standard error of mean. NT, non-treated, RNAiMAX, Lipofectamine RNAiMAX (a commercially available transfection reagent).

These findings demonstrate that LNP-mediated delivery improves the efficacy of gapmers for DUX4 knockdown in vitro while maintaining a favorable safety profile. An N:P ratio of 3 was chosen for further in vivo studies to maximize therapeutic potential.

LNP delivery improves gapmer bioavailability in skeletal muscle tissues and demonstrates safety in vivo

To assess the impact of LNP-mediated delivery on gapmer biodistribution upon systemic delivery in vivo, we assessed the bioavailability of bare and LNP-encapsulated gapmers in different tissues. Using a non-competitive hybridization-based ELISA designed to specifically detect the gapmers, we quantified their concentration in tissues samples harvested from wild-type B6 mice 2-days and 15-days post-retro-orbital injection with 11 mg/kg of each gapmer, either in bare or LNP-encapsulated form at an N:P ratio of 3.

Two days post-injection, we observed that LNP encapsulation significantly increased the bioavailability of gapmers across nearly all examined tissues, with the exception of the brain and heart tissues (Figure 3A). The gapmers, whether in LNP encapsulated or bare form, showed preferential accumulation in the liver and kidney, consistent with the well-established hepatic and renal tropism of systemically administered gapmers ⁷². However, encapsulation with LNPa resulted in a slightly attenuated enrichment in these clearance organs compared to LNAe. Specifically, LNPa increased gapmer bioavailability in the liver by 317.41 ± 28.23% and in the kidney by 226.6 ± 19.53%, while LNAe increased liver and kidney uptake by 330.67 ± 28.58% and 241.48 ± 20.88%, respectively. Importantly, both LNP formulations markedly enhanced skeletal muscle uptake, e.g., quadriceps (Qua), tibialis anterior (TA), diaphragm, and triceps brachii (TB), although the effect was less pronounced in the diaphragm. LNPa enhanced gapmer uptake in these tissues by an average of 77.52 ± 23.91%, while LNAe resulted in a similar average increase of 80.71 ± 23.60% in these muscle groups. Interestingly, MOE gapmers demonstrated slightly superior skeletal muscle uptake efficiency relative to their LNA counterparts, possibly due to differences in backbone flexibility or protein binding.

Figure 3: In vivo gapmer biodistribution and serum biochemistry profile 2-days post-injection with bare and LNP-encapsulated forms.

(A) Gapmer concentrations measured in different tissue samples, e.g., brain, diaphragm, heart, kidney, liver, lung, quadriceps, spleen, tibialis anterior (TA), and triceps brachii (TB) using ELISA based on the avidin-biotin affinity system (n = 5–6). (B) Serum biochemistry profile 2-days post-treatment. Hepatotoxicity markers include alkaline phosphatase (ALP), alanine transaminase (ALT), aspartate aminotransferase (AST), total bilirubin (T.Bil), total protein (T.Prot), albumin, globulin, and cholesterol. Renal markers include blood urea nitrogen (BUN), and creatinine. Creatine kinase (CpK) was included as a muscle injury marker. (C) Kidney injury molecule 1 (KIM1) levels in urine samples. Across the board, gapmers (11 mg/kg) were administered retro-orbitally in wild-type B6 mice either in bare or LNP-encapsulated forms (N:P ratio of 3). Tissue and blood samples were collected 2-days post-injection. Statistics, one-way ANOVA with Tukey’s multiple comparisons test; *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. Data are represented as mean ± standard error of mean. Each black dot represents a biological replicate (Total N = 5– 10).

Fifteen-days post-injection, we observed a marked decline in gapmer concentrations across all tissues compared to the earlier time point (Figure 3A), consistent with systemic clearance and tissue turnover over time (Figure S8). Despite this reduction, LNP encapsulation continued to provide a substantial bioavailability advantage, particularly in skeletal muscle tissues (Figure S8). Preferential accumulation of the gapmers in the liver and kidney remained evident; however, the relative enhancement by LNPs appeared less pronounced than at 2-days post injection, suggesting partial saturation of hepatic and renal uptake pathways or ongoing clearance of LNP-associated cargo. LNPa and LNAe both preserved improved tissue exposure, with LNPa showing slightly less pronounced liver and kidney accumulation relative to LNAe. Importantly, both LNP formulations maintained elevated levels of gapmers in skeletal muscle tissues, including the quadriceps, TA, and TB; however, the retention level in the diaphragm was no longer different between bare and LNP-encapsulated gapmers. While the overall magnitude of enhancement was reduced at this time-point, the increased retention levels in muscle tissues after 15 days underscores the durability of LNP-mediated delivery. LNP encapsulation improves the distribution and retention of gapmers in skeletal muscles, offering a clear advantage over bare formulations. This suggests that LNP composition can influence biodistribution and tissue residence time, enabling customizable approaches for targeted antisense delivery.

Together, these findings confirm that gapmers exhibit inherent hepato-renal tropism following systemic administration. LNP encapsulation does not eliminate this intrinsic distribution pattern; however, it substantially enhances gapmer bioavailability in peripheral skeletal muscles. This improvement is likely attributable to prolonged systemic circulation, reduced renal clearance, and diminished sequestration by the mononuclear phagocyte system (MPS) ^62,73,74.

To assess the safety profile of LNP-encapsulated gapmers, we analyzed hepatic and renal toxicity using serum biochemistry. Sera collected 2-days post-injection were assessed for alkaline phosphatase (ALP), alanine transaminase (ALT), aspartate aminotransferase (AST), blood urea nitrogen (BUN), creatine kinase (CpK), creatinine, total bilirubin (T.Bil), total protein (T.Prot), albumin, and globulin levels (Figure 3B). Most hepatic and renal toxicity markers were similar across treatment groups, except for elevated ALT in those treated with bare gapmers. Total protein, along with the albumin and globulin fractions, were elevated in all groups, suggesting a mild immune response.

Qualitative histological analysis of liver and kidney tissues harvested 2-days post-injection showed no signs of toxicity or immune damage (Figure S9). Additionally, we examined the expression of a panel of immune response markers, e.g., Il6, Tnfα, Il10, Cxcl10, and Hmox1, in liver and kidney tissues (Figure S10). In the liver, mice treated with LNP-encapsulated gapmers exhibited slight elevations in Tnfα and Hmox1, suggesting transient immune and oxidative stress responses. Il6 levels remained unchanged; however, a moderate increase in the anti-inflammatory marker Il10 was observed, likely reflecting a compensatory response to regulate inflammation and oxidative stress. Cxcl10 levels, a chemokine response marker, remained unchanged in LNP-treated groups but showed a mild increase in the bare gapmer-treated animals. In the kidney, immune activation was milder than in the liver, showing slight increases in Il6, Tnfα, and Il10, while Cxcl10 and Hmox1 levels were unchanged. We assessed kidney injury molecule 1 (KIM1) levels in urine samples, finding no significant changes (Figures 3C, S7). Importantly, 15 days post-injection, all serum biochemistry markers returned to levels similar to the control group (Figure S7). These findings suggest that LNP-encapsulated gapmers are generally well-tolerated, causing only mild and transient immune and oxidative stress responses with no significant evidence of overt toxicity in the measured panels over the study timeframe.

These findings suggest that LNP delivery can increase skeletal muscle exposure of gapmers, with tolerability supported by the set of readouts and timepoints investigated in this study. The results highlight the potential of LNP-encapsulated gapmers for therapeutic development.

A single systemic injection with LNP-gapmers did not improve functional performance in iDux4pA mice significantly

To evaluate the in vivo efficacy of LNP-gapmer therapy, we used doxycycline-inducible iDUX4pA; HSA-rtTA mice, referred to as iDUX4pA mice ⁷³. This transgenic model mimics key features of FSHD, including progressive myopathy, fibrosis, impaired muscle regeneration, and occasional hearing loss, with tunable control over disease onset and severity. To reduce variance due to mixed genetic background, we backcrossed the transgenes onto the FVB background ⁷³. We continue to refer to this model as iDUX4pA throughout. We fed 6-week-old female iDUX4pA mice doxycycline chow for 107 days, then administered a single intravenous dose of 11 mg/kg gapmers, either in bare or LNP-encapsulated form, before sacrificing them 7 days post-injection (Figure 4A).

Figure 4: Expression of DUX4 and its mouse-specific downstream genes in skeletal muscle tissues after a single intravenous injection of gapmer-LNP in iDUX4pA mice.

(A) Experimental workflow for evaluating the impact of a single injection of gapmers in iDUX4pA mice. Mice, aged 90 (± 3) days, were continuously fed doxycycline chow for 107 days to induce DUX4 expression and received a single retro-orbital injection with 11 mg/kg gapmers in bare or LNP-encapsulated forms. (B) DUX4 expression levels in quadriceps, tibialis anterior (TA), and triceps brachii (TB) muscles 7 days post-injection. (C–E) Expression of mouse-specific DUX4 downstream genes, e.g., Wfdc3 (C), Agtr2 (D), Serpinb6c (E) in quadriceps, TA, and TB muscles. Statistics, one-way ANOVA with Tukey’s multiple comparisons test; *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. Data are represented as mean ± standard error of mean. Each black dot represents a biological replicate (Total N = 3–5).

Gapmer treatments, whether in bare or LNP-encapsulated form, halted weight loss in iDUX4pA mice, with treated mice showing a slight post-treatment weight recovery (Figure S10). We evaluated the grip strength and treadmill running performance of mice on the initial day of treatment and at the conclusion of the study. Mice that received gapmer treatments, whether in bare or LNP encapsulated forms, showed improved performance compared to those that did not receive treatment (Figure S10). However, the observed differences did not reach statistical significance. RT-qPCR analysis indicated only modest reductions in DUX4 transcript levels in quadriceps (13.95–21.20%), tibialis anterior (13.41–19.84%), and triceps brachii (13.68–19.08%), with no significant differences between bare and LNP-gapmer treatments (Figure 4B). In addition to DUX4, we also looked into the expression of mouse-specific DUX4-related downstream genes, e.g., Wfdc3, Agtr2, and Serpinb6c (Figures 4C–E) ^58,73. While the treatments did not achieve more than 20% repression of DUX4, they nonetheless succeeded in reducing the expression of these target genes to some extent. A qualitative evaluation of Hematoxylin and Eosin (H&E) stained sections of TA muscles indicated no notable differences in pathological features, e.g., signs of fibrosis, inflammation, necrosis, centrally localized nuclei, and the presence of hypertrophic and hypotrophic fibers, when comparing the gapmer-treated group to the non-treated group (Figure S10).

Overall, a single injection of gapmers, whether bare or LNP-encapsulated, moderately reduced DUX4 expression and its downstream target genes. However, this treatment did not yield significant functional improvements or histopathological changes in iDUX4pA mice.

Multiple injections with LNP-gapmers were associated with improved functional performance in iDUX4pA animals

To evaluate the impact of repeated dosing, iDUX4pA mice were treated with multiple gapmer injections. Following 107 days of continuous doxycycline induction, mice received weekly intravenous injections of 11 mg/kg gapmers, either in bare or LNP-encapsulated forms, for six consecutive weeks. Mice were sacrificed 7 days after the final injection for sample collection and analysis (Figure 5A).

Figure 5: Body weight and functional performance following repeated intravenous injections of gapmer-LNP in iDUX4pA mice.

(A) Experimental workflow used to evaluate the effects of gapmer-LNP treatment following six weekly intravenous injections in iDUX4pA mice. 90 (± 3) days old iDux4pA mice were fed doxycycline chow for 107 days and then received six weekly injections of 11 mg/kg gapmers, e.g., LNAx, LNAz, and MOEx, in bare and LNP-encapsulated forms. The LNPs used were LNPa and LNPe, both at an N:P ratio of 3. The mice were sacrificed 7 days post-final injection. We did health monitoring and weekly functional performance assessments, including grip strength and treadmill running, until sacrifice. (B) Body weight normalized to day 0. (C–D) Forelimb (C) and total (D) grip strengths (GS) normalized to body weight. (E) Treadmill running time (TRT) at different time points throughout the study. (F–H) Weights of specific muscles, e.g., quadriceps (F), tibialis anterior (G), and triceps brachii (H), normalized to the corresponding body weights (BW). The purple downward arrows indicate the days when the mice received intravenous injections. Data are represented as mean ± standard error of mean. Each black dot represents a biological replicate (Total N = 3–5). FT, functional testing.

Gapmer treatments, especially with LNP encapsulation, significantly mitigated weight loss in iDUX4pA mice (Figures 5B, S11). By the terminal point, body weights were significantly higher in mice treated with LNP-encapsulated gapmers compared to non-treated controls (Figure S11). Specific muscle weights, including quadriceps, TA, and TB were significantly higher in LNP-gapmer-treated groups (Figures 5F–5H). Specifically, LNAx-LNPa and LNAx-LNPe showed the most pronounced increase in all three muscles. Also, quadriceps weights were significantly improved with LNAz and MOEx in LNP-encapsulated forms. Significant weight improvements were observed in the TA muscles of the mice recieving LNAz-LNPe and in the TB muscles of the mice receiving LNAz-LNPa and MOEx-LNPe. In addition, we observed improvements in grip strength and treadmill running performance following treatment, with LNP-gapmers showing better efficiency compared to bare gapmers (Figures 5C–5E, S11). Compared to treatment initiation day, fore limb grip strength showed significant improvement from the treatment initiation day across all treated groups, except for those receiving bare LNAz (Figure S11). LNP-gapmers, particularly those encapsulated in LNPe, showed the most substantial improvements in grip strength. Treadmill performance, measured as time to fatigue and total distance run, improved in all treated groups compared to baseline (Figures S11).

RT-qPCR analysis showed that LNP-gapmer treatment reduced DUX4 transcript expression by more than 55% in quadriceps, TA, and TB muscles compared to non-treated mice (Figure 6A). Among the formulations, LNPe-encapsulated gapmers exhibited better efficiency in reducing DUX4 levels. This was particularly evident in TA muscles, where LNP-gapmers significantly outperformed bare gapmers. Additionally, LNP-gapmer treatment led to a significant reduction in the expression of DUX4 target genes, e.g., Wfdc3, Agtr2, and Serpinb6c (Figures 6B–6D).

Figure 6: Expression of DUX4 and its downstream genes in skeletal muscle tissues following repeated intravenous injections of gapmer-LNP in iDUX4pA mice.

(A) DUX4 expression levels in quadriceps, tibialis anterior (TA), and triceps brachii (TB) muscles 7 days after the last injection. (B–D) Expression of mouse-specific DUX4 downstream genes, e.g., Wfdc3 (B), Agtr2 (C), Serpinb6c (D) in quadriceps, TA, and TB muscles. Statistics, one-way ANOVA with Tukey’s multiple comparisons test; *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. Data are represented as mean ± standard error of mean. Each black dot represents a biological replicate (Total N = 3–5).

The serum biochemistry profile and urinary KIM-1 analysis indicate that the LNP-gapmer treatment was generally well-tolerated in these mice (Figure S12). Importantly, mice that received the treatment demonstrated significantly reduced CpK levels, suggesting a decrease in muscle damage. Also, ALT levels showed a reduction across all treated groups. Histological analysis of TA muscles revealed marked improvements in pathological features in LNP-gapmer-treated groups. These included a significant reduction in centrally localized nuclei, a hallmark of muscle regeneration defects (Figure S13). Furthermore, the percentage of hypotrophic and hypertrophic muscle fibers in the TA and TB muscles demonstrated a significant reduction. While the percentages of both hypertrophic and hypotrophic fibers decreased, the mean Feret’s diameter of the muscle fibers remained unchanged, perhaps indicating a more uniform distribution in fiber size. Additionally, since macrophage infiltration is associated with inflammation and tissue damage in dystrophic muscle, we examined CD68⁺ macrophages in TA muscle sections via immunostaining (Figure S14). This showed reduced macrophage CD68⁺ cells in LNP-encapsulation gapmer-treated groups, compared to bare gapmers and non-treated controls. Also, we quantified Nox2, Nrf2, MuRF1, and Col3a1 expression to assess oxidative stress, inflammation, and fibrosis in the TA muscles. We noted, LNP-gapmer treatment decreased Nox2, MuRF1, and Col3a1 expression, indicating reduced ROS production, inflammation, and fibrosis, while Nrf2 expression increased, suggesting enhanced antioxidant response. In addition, we observed distinct variations in the toxicity profiles between bare and LNP-encapsulated formulations. In the kidney, bare gapmer treatment resulted in the upregulation of tubular injury markers (Havcr1, Lcn2) and inflammatory cytokines (Tnfα, Ccl2), consistent with the known renal accumulation of naked ASOs. In contrast, expression levels in LNP-treated groups remained comparable to controls, indicating that LNP encapsulation reduced the renal parenchymal exposure to gapmer-mediated stress (Figure S14). In the liver, LNP treatment resulted in a mild upregulation of macrophage markers (Adgre1) and chemokines (Ccl2) relative to untreated controls; however, this response was lower than that observed with bare gapmers. This likely reflects the engagement of the reticuloendothelial system for lipid clearance.

Overall, multiple injections of LNP-gapmers significantly reduced DUX4 expression, leading to enhanced muscle weight and improved functional performance. The treatment also mitigated histopathological abnormalities and inflammation, promoting muscle recovery. These results highlight the therapeutic potential of LNP-gapmers.

Discussion

Over the last couple of decades, ASO-based therapeutics have experienced significant advancements, with approximately two dozen ASOs receiving regulatory approval ⁷⁵. Among these, four morpholino-based exon-skipping ASOs have been greenlit by the U.S. Food and Drug Administration (FDA) for Duchenne muscular dystrophy, demonstrating the potential of ASO-mediated gene modulation for neuromuscular diseases ^76–79. Additionally, gapmer ASOs, e.g., Mipomersen for familial hypercholesterolemia and Inotersen for hereditary transthyretin amyloidosis, have gained traction, although Mipomersen was later withdrawn ^80,81. Despite these successes, ASO delivery remains a major challenge, particularly for systemic diseases like FSHD, where muscle tissue-specific targeting is essential ^20,62. Developing an effective therapeutic strategy for a disease like FSHD requires robust DUX4 suppression and a delivery platform that ensures tissue-specific uptake without compromising the overall safety of the patient ^8,20. Addressing this fundamental challenge, our study demonstrates that LNP encapsulation significantly enhances the bioavailability and efficacy of antisense gapmers targeting DUX4 while maintaining a manageable safety profile.

In this study, we demonstrate that antisense gapmers effectively suppress DUX4 expression both in vitro and in vivo, with significant improvements when delivered via LNP encapsulation. In patient-derived myotubes, we show that both LNA and MOE gapmers effectively silence DUX4 expression, resulting in reduced target gene expression and enhanced myogenic differentiation (Figure 1). Our study highlights potential negative effects of high dose gapmer treatment, as 100 nM concentrations reduced cell viability and induced apoptosis. In contrast, a 10 nM dose achieved significant knockdown with a favorable safety profile, indicating an optimal therapeutic window for FSHD. LNP encapsulation further improved the efficacy of 10 nM gapmers, lowering DUX4 expression without compromising cellular health and enhancing the myogenic fusion index (Figure 2). In vivo, we show that LNP-mediated delivery significantly enhances gapmer bioavailability in skeletal muscles (Figure 3). We noted a transient immune activation in the liver and kidneys of wild type mice, but no long-term toxicity was found, confirming the overall safety of LNP-gapmers. A critical aspect of our study was the evaluation of functional outcomes in the iDUX4pA mouse model, which recapitulates key pathological features of FSHD. A single systemic injection of gapmers, whether bare or LNP-encapsulated, resulted in only a modest reduction in DUX4 expression (Figure 4). However, repeated weekly injections over six weeks significantly improved therapeutic outcomes, resulting in increases in muscle mass, strength, and endurance. LNP-gapmer-treated mice outperformed controls, emphasizing the need for repeated dosing to achieve meaningful functional recovery (Figures 5–6). These results offer a proof-of-concept that LNP-gapmers can help enhance the efficacy and bioavailability of antisense gapmers for DUX4 inhibition in FSHD.

Various ASO chemistries, including 2’OMe ASOs, PMOs, gapmers composed of cEt, ALNA[Ms], LNA, and MOE gapmers, have demonstrated DUX4 knockdown and improvements in muscle pathology in vitro and in vivo ^52–60. Early studies with 2’OMe ASOs targeting the DUX4 PAS or exon 3 splice sites demonstrated partial knockdown and modest reductions in DUX4-positive nuclei in FSHD myotubes ^52,55. However, these strategies rely on passive transcript suppression rather than active degradation, limiting their overall effectiveness. More recently, PMO-based ASOs targeting the PAS in exon 3 exhibited efficacy in vitro and achieved localized suppression in mouse models following intramuscular administration ^53,55–59. Although Lu-Nguyen et al. (2021) showed that a vivo-PMO strategy led to systemic DUX4 knockdown in ACTA1-MCM; FLExDUX4 mice, the observed functional benefits were modest, likely due to incomplete suppression of DUX4-driven pathology ⁵⁹. Perceivably, antisense gapmers would supersede these ASO-based strategies in terms of effectiveness due to their ability to induce direct degradation of target mRNA transcripts via RNase H. In a simple head-to-head comparison, we compared LNA and MOE gapmers against a previously reported PAS-targeting PMO sequence in ACTA1-MCM;FLExDUX4 mice. Following intravenous administration, our gapmers showed comparatively superior performance in suppressing DUX4 upregulation compared to the PMO (Figure S15). Importantly, while PMO treatment resulted in variable knockdown efficacy across different muscle groups, potentially due to the limited systemic uptake of neutral backbones, the LNA and MOE gapmers consistently maintained robust silencing.

In this context, we previously designed LNA and MOE gapmer ASOs targeting exon 3, which effectively silenced DUX4 and its target genes in muscle cells from FSHD patients ^56,57. Both types of gapmers showed substantial knockdown of DUX4 in vitro and improved cellular phenotypes, with notable reductions in DUX4 expression in FLExDUX4 mice following local injections. Also, cEt gapmers targeting exon 1 improved muscle pathology upon systemic delivery in ACTA1-MCM; FLExDUX4 mice, but functional gains were limited, as evidenced from hanging grid and four-limb grip strength tests ⁵⁸. This highlights the challenges associated with systemic ASO therapy for FSHD. While ALNA[Ms] gapmers were recently reported to prevent muscle force decline in ACTA1-MCM; FLExDUX4 mice, the study lacked detailed pharmacokinetic and safety evaluations, making it difficult to assess the feasibility of long-term therapeutic application ⁶⁰.

Compared to previous approaches, the current study positions LNP-mediated delivery as an advancement in FSHD therapeutics. LNPs are well explored for their utility in delivering mRNA, siRNA, and DNA. However, to the best of our knowledge, this study reports the first systematic evaluations of LNP-enabled systemic delivery of gapmers in this context. Injection with bare gapmers resulted in a biodistribution pattern consistent with previously reported studies in murine models, characterized by predominant accumulation in the liver and kidney, and relatively limited uptake in skeletal muscle and central nervous system tissues ^72,82. We demonstrated that, systemic administration of LNP-encapsulated gapmers significantly enhanced bioavailability in skeletal muscle tissues, overcoming the hurdle of poor muscle uptake that has historically limited ASO-based strategies (Figure 3). Using wild-type animals, we observed that the gapmers showed a natural liver and kidney tropism. Encapsulation of the gapmers into the LNPs we used did not make them extra-hepato or extra-renal tropic. However, LNP encapsulation did also improve their skeletal muscle accumulation significantly. This increased bioavailability correlated with improved pharmacodynamic effects, as evidenced by enhanced DUX4 suppression and downstream gene regulation in iDUX4pA mice (Figures 4, 6, S10). The impact of LNP-gapmer therapy extended beyond molecular endpoints, leading to significant functional improvements in the iDUX4pA mouse model (Figures 4, 6, S10). While a single injection resulted in only modest reductions in DUX4 expression, repeated weekly injections over six weeks significantly improved muscle mass, grip strength, and treadmill endurance, highlighting the necessity of sustained suppression to achieve meaningful therapeutic outcomes (Figures 4, 6, S10).

From a methodical perspective, we primarily utilized RT-qPCR based quantification of DUX4 expression to rank the gapmers, following previous studies developing DUX4 targeting therapeutics. While we did try to assess the effect of the gapmers at a protein level, western blot analysis was not highly consistent in our hands. We acknowledge that endogenous DUX4 expression in FSHD myogenic cultures is extremely low and episodic, and highly stochastic. As a result, RT-qPCR-based quantification of DUX4 can be variable across differentiations and sampling, and small absolute differences may be sensitive to technical and biological noise ^83,84. Although we tried mitigating this by using multiple biological replicates per condition, we did not exhaustively assess donor-to-donor variability or differentiation difference across a larger panel of independent patient lines. Henceforth, the in vitro pharmacodynamic findings we report should be interpreted as supportive evidence that motivates broader validation across additional donors and independent differentiation batches, ideally incorporating orthogonal quantification approaches for low-abundance targets. We also recognise that apparent improvements in myogenic differentiation readouts can, in principle, be influenced by changes in cell number/viability and selection effects, particularly when treatments produce measurable cytotoxicity or apoptosis signals. In our study, we could not fully deconvolve whether observed differentiation-associated changes reflect direct biological improvement versus preferential survival of subpopulations under some dosing/formulation conditions. Accordingly, these differentiation-associated findings could be strengthened by adding more orthogonal differentiation metrics, and explicit normalisation to viable nuclei/cell number where appropriate.

In the in vivo studies, the 11 mg/kg dose that we used corresponds to a human equivalent dose of ~0.9 mg/kg when adjusted for body surface area allometry (K_m factor 12.3) ^85,86. While these appears to fall within and potentially below the range of currently approved ASO therapy doses, we acknowledge that simple allometric scaling does not account for interspecies physiological differences in terms of renal clearance, immunoresponse, metabolic stability, or target tissue receptor density ⁸⁷. Consequently, future translational efforts must focus on defining the minimum effective dose in large animal models and induced pluripotent stem cell derived cellular and organoid models to empirically validate the therapeutic index prior to clinical application ^88–90.

Another particularly novel aspect of our study is the comprehensive evaluation of safety in the context of LNP-gapmer therapy. Previous studies on ASO-mediated DUX4 suppression have largely focused on efficacy, with limited assessment of systemic toxicity. Our study fills this gap by integrating in vitro cytotoxicity assays, serum biochemistry profiling, histopathology, and immune marker analysis to assess potential adverse effects. Indeed, we report for the first time the in vitro cytotoxicity, apoptosis, and viability of cells upon gapmer treatment. While a transient immune activation was observed in the liver and kidneys following intravenous injections, these responses were mild a within the measured post-dosing intervals, and we did not observe overt hepatotoxicity or nephrotoxicity in the assessed targeted transcriptional panels.. The in vitro cytotoxicity assessments revealed that LNP encapsulation did not introduce additional toxicity beyond what was observed with bare gapmers, reinforcing its suitability for systemic delivery. A more comprehensive immunotoxicology assessment (e.g., broader innate immune profiling and longer follow-up) was outside the scope of the current study and will be important for future translational developments. This aspect is particularly relevant given that systemic ASO administration has historically been associated with dose-limiting toxicities, including hepatic inflammation and renal dysfunction. The iDUX4pA mouse model used in this study provides another critical strength ^64,65,73. Unlike the FLExDUX4 and ACTA1-MCM; FLExDUX4 models, which exhibit variable and mosaic DUX4 expression, the iDUX4pA model offers a more consistent and tunable disease phenotype under doxycycline induction ^91,92. Although this strain is less productive, perhaps owing to the heterozygous state of the iDUX4pA transgene, meaning only half of progeny carry it, their strong and reproducible pathology makes them a valuable tool for assessing the efficacy of DUX4-targeting therapies. The significant functional improvements observed in LNP-gapmer-treated mice further reinforce the potential for clinical translation. Despite these promising findings, several challenges remain. The LNP formulations used in this study were not specifically designed for muscle targeting (Figure 3). We took advantage of industry-standard formulations comprising both linear (e.g., LNPb) and branched alkyl tail architectures (e.g., LNPa, LNPh) to test the hypothesis that LNP encapsulation could improve the pharmacokinetic profile and biodistribution of gapmers. As we have observed, these LNPs showed substantial tropism in the liver and kidney tissues. Although this did not result in overt toxicity in our hands, the development of muscle-targeting LNP formulations could further enhance therapeutic efficacy while reducing systemic and/or off-target exposure. The data in this study indicate that LNP encapsulation enhanced the bioavailability of the gapmers in skeletal muscles, possibly due to a combination of distinct cellular uptake mechanisms and altered tissue pharmacokinetics; however, the exact mechanism warrants further investigation ^74,93. Unlike naked gapmers, or ASOs in general, which largely rely on inefficient fluid-phase pinocytosis or gymnosis, LNPs adsorb serum apolipoproteins (e.g., ApoE) which facilitate receptor-mediated endocytosis ^62,93,94. While this mechanism predominantly drives hepatic clearance, consistent with the liver tropism seen in our study, it also enables active uptake in peripheral tissues expressing lipoprotein receptors. Besides, given the inflammatory pathology of the iDUX4pA mice, we speculate that an enhanced permeability and retention (EPR)-like effect may play a critical role ^64,95–98. In dystrophic muscle, vascular permeability is often elevated; this likely facilitates the extravasation and interstitial accumulation of LNPs. In this context, the sequestered LNPs could function as a local micro-depot, gradually releasing, or leaching, the payload into the extracellular space for sustained uptake by adjacent muscle fibers. Once internalized, ionizable lipids promotes endosomal escape through membrane fusion upon acidification, thereby significantly enhancing the cytosolic bioavailability of the payload ^99,100. In the current study, we observed that LNPa, containing a branched lipid (SM-102), showed more favorable performance compared to formulations containing linear lipids ¹⁰⁰. This might be suggestive of specific lipid geometries offering advantageous features for muscle uptake or endosomal escape in myogenic cells. Deciphering the structure-activity relationship (SAR) of these lipids, with a focus on optimizing branching parameters, represents a critical scope for future investigations.

One important note to make here is that, while LNP-gapmers significantly improved DUX4 knockdown in vivo, their efficacy did not reach the near-complete suppression levels observed in vitro (Figures 1, 4, 6, S10). This discrepancy highlights the need for further optimization of gapmer chemistry and dosing regimens to maximize target engagement in skeletal muscles. Perhaps the development of muscle-targeting LNPs will also address this issue by improving the DUX4 knockdown efficacy in vivo.

Future efforts should focus on refining LNP compositions to increase muscle-specific tropism while minimizing effects in undesirable tissues. While our current LNP-mediated stability enhances systemic bioavailability, emerging artificial intelligence and machine learning strategies offer a powerful new avenue to accelerate the discovery of next-generation lipids with superior extrahepatic targeting capabilities ^101,102. These data-driven optimization strategies can revolutionize LNP design and development of formulations targeting the muscles and other organs. Alongside, incorporating small molecule oligonucleotide activity enhancers optimized for muscle tissue targeting could further potentiate DUX4 knockdown and enhance therapeutic outcomes ^103–107 Long-term studies will be necessary to evaluate the durability of treatment effects and potential adaptive immune responses associated with chronic ASO therapy. Moreover, exploring combination strategies that integrate ASOs with other therapeutic modalities, e.g., small molecules therapeutics, could provide synergistic benefits in mitigating FSHD progression.

In addition to the potential benefits in the context of muscle diseases, the LNP-based platforms for the delivery of complex nucleic acid cargoes, e.g, gapmers, may allow for physicochemical tuning (e.g., selective organ targeting or SORT) to redirect tropism toward other tissues, for example, the pulmonary system ¹⁰⁸. This suggests that similar LNP-gapmer, or LNP-ASO based strategies in general, could be explored in other pathologies requiring tissue-specific bioavailability, e.g., lung cancer, tuberculosis, etc ¹⁰⁹. The continued evolution of LNP chemistry toward precise tissue-specific tropism holds the promise of expanding the therapeutic window for both neuromuscular and extra-muscular indications.

Overall, we provide the first demonstration that LNP-mediated delivery significantly enhances the efficacy, bioavailability, and safety of antisense gapmers targeting DUX4. Compared to previous ASO-based strategies, LNP-gapmers achieve superior systemic knockdown, translating into measurable functional improvements in an established FSHD mouse model. These findings provide motivation for further optimization of the LNP-gapmer therapeutic strategy for clinical translation, underscoring the potential of antisense-based approaches to redefine the treatment landscape for FSHD.

Materials and methods

Ethics Statement

All experiments performed in this study received approval from the Research Ethics Office (REO) at the University of Alberta. The use of immortalized human cells was reviewed and endorsed by the Human Research Ethics Boards, REO (Pr00079871), while studies involving mice obtained authorization from the Animal Care and Use Committees (ACUC), REO (AUP00000365).

Gapmer Design, Synthesis, and Transfection

We designed three LNA and three MOE gapmers targeting DUX4 exon 3, outside the PAS sequence (Figure 1A, Table S1). Target sites were selected based on GC content and mRNA secondary structure predictions. We used RNAfold web server (rna.tbi.univie.ac.at/cgi-bin/RNAWebSuite/RNAfold.cgi) to predict secondary structures of DUX4 mRNA ^110,111. The DUX4 exon 3 sequence, including 50 bases of upstream intron 2 and downstream exon 3 sequences, was used as the input. Pre-mRNA folding was performed under default parameters, and the predicted structure with the lowest ΔG value was selected for designing the gapmers. We chose to focus on DUX4 exon 3 only, as it is specifically linked to the pathogenic DUX4 transcript, while excluding the PAS due to its potential similarity with other genomic sequences. The LNA gapmers were 15–16 nucleotides long, with the first and last three nucleotides modified with LNA chemistry to enhance binding affinity and nuclease resistance. The MOE gapmers were 20 nucleotides long, consisting of a central 10-nucleotide DNA segment, flanked by 5-nucleotide MOE-modified regions on both ends to increase stability and specificity. Also, we designed mock LNA gapmers as negative controls. All gapmers were fully phosphorothioated to improve nuclease resistance and were commercially synthesized by Exiqon and Integrated DNA Technologies (IDT).

Phosphorodiamidate morpholino oligomers (PMO)

A previously reported PMO sequence (FM10/PMO: 5′-GGG CAT TTT AAT ATA TCT CTG AAC T-3′), designed to target the DUX4 PAS, was synthesized commercially from Gene Tools, LLC (Philomath, OR) ⁵³. Lyophilized PMO was reconstituted in sterile PBS according to the manufacturer’s instructions.

Cell Culture

Immortalized human muscle cells used in the study were kindly provided by the Wellstone Program at the University of Massachusetts Medical School (MA, USA) through Dr. Jennifer Chen. Two cell lines were used: cell line WS229 and WS234. The WS229 (aka 15Abic CT#24, 0.1M) cell line is an immortalized myoblast line derived from the biceps brachii biopsy of a 66-year-old male patient genetically diagnosed with an early-adult onset of FSHD with 8 repeats of D4Z4 units ¹¹². The WS234 (aka 15Vbic CT#09, 0.1M) cell line is an immortalized myoblast line isolated from the biceps brachii of a 69-year-old female sibling of the individual who donated the WS229 cells, who was confirmed to be unaffected with FSHD with >10 repeats of D4Z4. Both cell lines were immortalized through stable CDK4/hTERT cassette integration.

Both cell lines were grown in a growth medium containing 15% fetal bovine serum (FBS) (Sigma, St. Louis, MO, USA), 0.055 μg/mL dexamethasone, 2.5 ng/mL recombinant human hepatocyte growth factor (Peprotech/Thermo Fisher Scientific, Cranbury, NJ, USA), and 10 ng/mL recombinant human fibroblast growth factor (Peprotech/Thermo Fisher Scientific, Cranbury, NJ, USA) in basal medium. The basal medium was prepared by adding 20% Medium 199 (Life Technologies, Carlsbad, CA, USA), 0.0282 μg/mL zinc sulphate (heptahydrate; Sigma-Aldrich, St. Luise, MO, USA), 1.4 μg/mL vitamin B12 (Sigma-Aldrich, St. Luise, MO, USA), and 2.5% penicillin-streptomycin (0.5% Penicillin-Streptomycin (Gibco, Grand Island, NY, USA) in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 4.5 g/L D-glucose, L-Glutamine, and 25 mM HEPES (4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid) (Gibco, Grand Island, NY, USA). To induce differentiation, the growth medium was replaced with differentiation medium containing 15% KnockOut Serum Replacement (KOSR; Life Technologies, Grand Island, NY, USA), 10 μg/mL insulin (Sigma-Aldrich, St. Luise, MO, USA), and 100 μg/mL human apo-transferrin (R&D Systems, Minneapolis, MN, USA) in BM. All cells were cultured at 37°C and 5% CO₂ ¹¹³. All cells were maintained at 37°C with 5% CO₂ until use.

Transfection for in vitro screens

For gapmer transfection, cells were seeded onto each well of a gelatin-coated (Sigma Aldrich, St. Luise, MO, USA) 6- or 24-well plate (BioLite/Thermo Scientific, Rochester, NY, USA) at a density of 50,000 cells/cm², grown, and then differentiated as described in the previous section (Figure 1B). For gapmer-only transfections, gapmers were diluted in PBS to achieve final concentrations of 100 nM, 10 nM, and 1 nM in the total transfection volume. The diluted gapmers were then directly added to the differentiation medium before application to the cells. For RNAiMAX-mediated transfection, DUX4-targeting gapmers or a mock gapmer were prepared at final concentrations of 100 nM, 10 nM, and 1 nM in 2% Lipofectamine RNAiMAX (Life Technologies, Carlsbad, CA) diluted in Opti-MEM^™ Reduced Serum Medium supplemented with GlutaMAX^™ (Gibco, Grand Island, NY). The transfection mixture was then further diluted 1:5 in differentiation medium before being applied to the cells. For LNP-gapmer transfections, gapmers were diluted in sodium acetate buffer (pH 4.0) to achieve final concentrations of 10 nM and 1 nM, while LNPs were prepared at N:P ratios ranging from 1 to 6 in the same buffer. The LNP-gapmer mixture was incubated at room temperature for 15 minutes to allow encapsulation. Following incubation, the LNP-gapmer transfection mixture was further diluted 1:5 in differentiation medium and recombinant human ApoE3 (PeproTech/Thermo Scientific, Rocky Hill, NJ, USA) was added in the transfection mixture at a standard concentration of 0.1 μg/mL before being applied to the cells.

In all experiments, WS229 cells treated with the transfection mixture but without gapmers or LNPs served as the non-treated control, while WS234 cells, grown and differentiated under identical conditions, were used as the healthy control.

RNA extraction, cDNA synthesis, and RT-qPCR

Total RNA was extracted from cultured cells using the RNeasy Mini Kit (QIAGEN, Germany). To ensure complete homogenization of the cell lysates, samples were processed twice through a QIAshredder column (QIAGEN, Germany), followed by on-column DNase treatment (RNase-free DNase set, QIAGEN, Hilden, Germany) to remove residual genomic DNA, as described previously ¹¹⁴. For RNA extraction from frozen tissue samples, a TRIzol-based method was employed on 20–30 sections (20 μm thick) following the manufacturer’s protocol, with a slight modification: tissue sections were incubated at −80°C overnight before vortexing, allowing for optimal homogenization and RNA yield.

From the extracted total RNA, 1400 ng of total RNA was used for cDNA synthesis using the SuperScript IV One-Step RT-PCR System (Invitrogen, Vilnius, Lithuania), as directed by the manufacturer, with 0.5 μg of oligo(dT)_12-18 (Invitrogen, Carlsbad, CA) as the primer in a final volume of 20 μL. A reaction containing nuclease-free water instead of RNA served as a negative control. For both cell and mouse derived samples, the synthesized cDNA was then used as a template for qRT-PCR in a QuantStudio 3 Real-Time PCR System (Applied Biosystems, Carlsbad, CA, USA). With the exception of genes ZSCAN4, TRIM43, and MBD3L2, the expression of all other genes in the cell samples were investigated using the SsoAdvanced^™ Universal SYBR^® Green Supermix (Bio-rad, Hercules, CA, USA), with forward and reverse primers added to achieve final concentrations of 0.4 μM each. Primer sequences for SYBR^®-based qRT-PCR are listed in Table S2. For ZSCAN4, TRIM43, and MBD3L2, pre-designed probe-based TaqMan^™ Gene Expression assays (ZSCAN4: Hs00537549_m1, TRIM43: Hs00299174_m1, and MBD3L2: Hs00544743_m1; Thermo Fisher, Waltham, MA) were used. Reactions were prepared using respective TaqMan^™ assays, synthesized cDNA, and TaqMan^™ Fast Advanced Master Mix (Thermo Scientific, Vilnius, Lithuania), following the manufacturer’s recommended protocol. For both SYBR and TaqMan reactions, the default “Fast” cycling program of the qPCR machine was used, (i) 95°C, 20 s, (ii) 40 cycles of 95°C, 1 s then 60°C, 20 s; for SYBR^® reactions, there was an additional step for melt curve construction. Expression levels were normalized to those of GAPDH or Gapdh and determined using a standard ΔΔCt method.

Immunocytochemisty

WS229 and WS234 cells were seeded at a density of 50,000 cells/cm² onto 24-well plates (BioLite/Thermo Scientific, Rochester, NY, USA) precoated with gelatin (Sigma-Aldrich, St. Louis, MO, USA) and differentiated into myotubes for 13 days before treatment as described above. Five days post-treatment, cells were fixed with 4% freshly prepared paraformaldehyde (Sigma-Aldrich, St. Louis, MO, USA) for 4–5 minutes, permeabilized with 0.25% Triton X-100 (Thermo Scientific, Rockford, IL) for 5 minutes, and blocked with 1% normal goat serum (Sigma-Aldrich, Oakville, ON, Canada) diluted in PBS containing 0.1% Triton X-100 for 20 minutes. For immunofluorescence staining, myotubes were probed with an anti-dystrophin primary antibody (ab85302, Abcam, Cambridge, UK) followed by an Alexa Fluor 488-conjugated anti-mouse IgG secondary antibody (Invitrogen, Eugene, OR, USA). Finally, nuclei were counterstained and the samples were mounted using ProLong Gold Antifade Mountant containing DAPI (4′,6-diamidino-2-phenylindole; Life Technologies, Eugene, OR, USA).

Myogenic fusion index evaluation

The myogenic fusion index (MFI) was determined by dividing the number of nuclei in myotubes (defined as cells containing at least two nuclei within a shared cytoplasm) by the total number of nuclei per field of view and then multiplying by 100 to obtain a percentage value. We used ImageJ software (National Institutes of Health) for counting nuclei. For each replicate, the MFI was averaged from three random fields of view. Randomly selected myotubes were quantified, and if fewer than 15 myotubes were present in a field of view, all available myotubes were included in the analysis. 15–18 wells per treatment group were analyzed for fusion index quantification. To normalize MFI across conditions, we calculated relative MFI as: relative MFI = mean MFI calculated of each view/MFI of healthy control.

Immunoblotting

We extracted proteins from in vitro cell cultures to obtain either whole-cell lysates or subcellular (cytoplasmic and nuclear/chromatin) fractions. For whole-cell extraction, cell pellets were directly suspended in lysis buffer containing 10 mM HEPES (pH 7.6), 0.5 mM EDTA, 100 mM NaCl, 1% Triton X-100, and 10% glycerol, supplemented with cOmplete^™ Mini EDTA-free protease inhibitor cocktail (Roche). For nuclear and chromatin fractionation, cell pellets were first suspended in hypotonic Buffer A (10 mM HEPES pH 7.6, 10 mM KCl, 1.5 mM MgCl₂, and 0.3% NP-40) supplemented with protease inhibitors. Following a 10-minute incubation on ice, samples were centrifuged at 3000 × g for 5 minutes at 4°C, and the supernatant was collected as the cytoplasmic fraction. To extract the chromatin-bound nuclear fraction, the remaining pellet was washed once with Buffer A and subsequently resuspended in Chromatin Buffer (20 mM Tris-HCl pH 7.5, 150 mM NaCl, 2 mM MgCl₂, 0.5% NP-40, and 10% glycerol) supplemented with protease inhibitors and 0.5 U/μL Benzonase nuclease (Sigma-Aldrich). The suspension was incubated on ice for 15–20 minutes with periodic mixing. All lysates were sonicated using a Sonic Dismembrator 60 (Thermo Fisher Scientific, Waltham, MA) equipped with a microtip probe at a 30% amplitude setting (short 0.5–1 s bursts for a total of ~10 seconds on ice). For the chromatin fractions, EDTA was added to a final concentration of 10 mM to halt nuclease activity, followed by a clarifying centrifugation at 16,000 × g for 10 minutes at 4°C. Total protein concentrations were quantified using the Pierce^™ BCA Protein Assay Kit (Thermo Fisher Scientific) following the manufacturer’s guidelines.

Protein samples (5 or 10 or 25 μg) were loaded onto pre-cast NuPAGE 4 to 12% Bis-Tris Mini Protein Gels (Invitrogen) for SDS-PAGE and then then transferred onto a polyvinylidene difluoride membrane (Immobilon-P, Millipore). Membranes were blocked in a solution of Tris-buffered saline containing 0.05% Tween 20 (TBST) and 5% skim milk (Fat Free Instant Skim Milk Powder, Carnation, Markham, ON) while shaking. Post-blocking, the membranes were cut and incubated with one of the following primary antibodies overnight at 4°C in blocking solution: (i) DUX4: anti-DUX4 antibody (E5-5, ab124699, Abcam) or human DUX4/DUX4c antibody (MAB95351100, R&D Systems), (ii) ZSCAN4: ZSCAN4 polyclonal antibody (PA5-32106, Thermo Fisher Scientific), (iii) TRIM43: TRIM43 polyclonal antibody (PA5-32040, Thermo Fisher Scientific), (iv) Histone H3: Histone H3 (D1H2) XP rabbit mAb (4499, Cell Signaling Technology), and (v) β-tubulin: anti-ß tubulin antibody (ab6046, Abcam). Post-overnight incubation in a cold room in darkness, membranes were washed in TBST and incubated for 1 hour at room temperature with the appropriate species-matched horseradish peroxidase (HRP)-conjugated secondary antibodies: anti-rabbit IgG (7074, Cell Signaling Technology), anti-mouse IgG (7076, Cell Signaling Technology; or 31430, Thermo Fisher Scientific), goat anti-mouse IgG1 (ab98693, Abcam), or goat anti-mouse IgG2b (M32407, Thermo Fisher Scientific). Protein bands were detected using ECL Select Detection Reagent (GE Healthcare) or Radiance Plus (Azure Biosystems) and visualized on a ChemiDoc Imaging System (Bio-Rad). Where necessary, we stripped the membranes using Restore^™ PLUS Western Blot Stripping Buffer (Thermo Fisher Scientific) according to the manufacturer’s instructions and re-probed.

Cytocompatibility assessment using ApoTox-Glo Triplex Assay

Cell vitality, cytotoxicity, and apoptosis were assessed using the ApoTox-Glo Triplex Assay Kit (Promega, Madison, WI, USA) according to the manufacturer’s protocol. Briefly, WS229 and WS234 cells were seeded at a density of 50,000 cells/cm² onto 96-well flat-bottom microplates (BioLite/Thermo Scientific, Rochester, NY, USA) pre-coated with gelatin (Sigma-Aldrich, St. Louis, MO, USA) and allowed to adhere and differentiate as described in the previous section. To assess cell viability and cytotoxicity, 20 μL of viability/cytotoxicity reagent was added to each well. This reagent contained glycyphenylalanyl-aminofluorocoumarin (GF-AFC), a fluorogenic substrate that generates a fluorescent signal in viable, metabolically active cells, and bis-alanylalanyl-phenylalnyl-rhodamine 110 (AAF-R110), which is cleaved by proteases released from damaged or dead cells, producing a cytotoxicity-specific fluorescent signal. After adding the reagent, plates were placed on an orbital shaker set at 300 rpm for 30 seconds, then incubated at 37°C for 40 minutes to allow the enzymatic reactions to proceed. Fluorescence was subsequently measured using a SpectraMax M3 Multi-Mode Microplate Reader (Molecular Devices, San Jose, CA, USA) with excitation/emission wavelengths of 400/505 nm for viability and 485/520 nm for cytotoxicity. Viable cells exhibited a reduction in AFC fluorescence, while cytotoxicity was indicated by an increase in protease-mediated fluorescence. For apoptosis detection, 100 μL of Caspase-Glo 3/7 reagent was added to all wells to measure caspase-3/7 activity, a key marker of apoptosis. Plates were once again placed on an orbital shaker at 300 rpm for 30 seconds, followed by incubation at room temperature for 40 minutes. Luminescence signals were then measured using a SpectraMax M3 Multi-Mode Microplate Reader to quantify apoptosis levels.

Flow cytometric assessments of cell viability and apoptosis

In vitro cell viability and apoptosis levels were assessed using the PE Annexin V Apoptosis Detection Kit I (BD Pharmingen^™, San Diego, CA) following the manufacturer’s instructions with some modifications. Briefly, 24-hrs post-treatment with 10 nM of the gapmers, we harvested, washed, and resuspended the cells in 1× binding buffer, and then stained with Phycoerythrin (PE) Annexin V and 7-Aminoactinomycin D (7-AAD) for 10–15 minutes at room temperature in the dark. Data were acquired on a BD LSR Fortessa SORP flow cytometer (BD Biosciences, Franklin Lakes, NJ) at the Flow Cytometry Core Research Facility at the University of Alberta and analyzed using a custom Python script. Single cells were isolated by gating on Forward Scatter Area (FSC-A) vs. Height (FSC-H) to exclude doublets and debris. The quadrant thresholds for PE Annexin V and 7-AAD were defined using negative/control samples, set at the 98^th percentile of the corresponding control intensity, and then applied uniformly across samples. Cells were classified as viable (Annexin V⁻/7-AAD⁻), early apoptotic (Annexin V⁺/7-AAD⁻), late apoptotic (Annexin V⁺/7-AAD⁺), or necrotic (Annexin V⁻/7-AAD⁺). For visualization, scatter plots were generated using a consistent number of randomly down-sampled singlet events per sample in each replicate, while percentages were calculated from all gated singlet events.

LNP Synthesis and Gapmer Encapsulation

Lipid nanoparticles were prepared following previously described methods with slight modifications ⁶⁶. Each LNP formulation consisted of 50% ionizable lipid, 38.5% cholesterol (MilliporeSigma, Burlington, MA, USA), 10% 1,2-distearoyl-sn-glycero-3-phosphocholine (DSPC; Avanti Research, Birmingham, AL, USA), and 1.5% 1,2-dimyristoyl-rac-glycero-3-methoxypolyethylene glycol-2000 (PEG2k-DMG; Avanti Research, Birmingham, AL, USA) at a total lipid concentration of 10 mM. The ionizable lipids used in this study were: LNPa – SM-102 (9-heptadecanyl 8-{(2-hydroxyethyl)[6-oxo-6-(undecyloxy)hexyl]amino}octanoate), LNPb – DLin-MC3-DMA (4-(dimethylamino)-butanoic acid, (10Z,13Z)-1-(9Z,12Z)-9,12-octadecadien-1-yl-10,13-nonadecadien-1-yl ester), LNPc – DODAP (1,2-dioleoyl-3-dimethylammoniumpropane), LNPd – DODMA (1,2-dioleyloxy-3-dimethylaminopropane), LNPe – nor-MC3 ((6Z,9Z,26Z,29Z)-pentatriaconta-6,9,26,29-tetraen-18-yl 4-(dimethylamino)butanoate), LNPf – DLin-KC2-DMA (2-[2,2-bis[(9Z,12Z)-octadeca-9,12-dienyl]-1,3-dioxolan-4-yl]-N,N-dimethylethanamine), LNPg – MF019 (6,8,26,28-tetrathiatritriacontan-17-yl 4-(dimethylamino)butanoate), LNPh – ALC-0315 ([(4-hydroxybutyl)azanediyl]di(hexane-6,1-diyl) bis(2-hexyldecanoate)). The ionizable cationic lipids, cholesterol, DSPC, and PEG2k-DMG were dissolved in ethanol at the specified molar ratios before LNP formulation.

For in vitro experiments, preformed lipid vesicles were loaded with the gapmers following a previously described method. In short, empty lipid vesicles were prepared by using a T-junction microfluidic mixer to combine the organic phase containing the lipids dissolved in ethanol with an aqueous phase of 25 mM sodium acetate buffer (pH 4.0) at a 1:3 flow rate ratio and a total flow rate of 20 mL/min. The empty vesicles were subsequently dialyzed against 25 mM sodium acetate buffer (pH 4.0) to remove the ethanol, then filtered through a 0.2 μm polyethersulfone membrane syringe filter (Cytiva Life Sciences, Marlborough, MA, USA), and concentrated using ultracentrifugal filters (10 kDa molecular weight cutoff, MilliporeSigma, Burlington, MA, USA) to a total lipid concentration of 20 mM. The empty vesicles were stored at 4°C before use. To form the final gapmer-LNPs, empty lipid vesicles were mixed with the gapmers, which were also dissolved in 25 mM sodium acetate buffer (pH 4.0), at a N:P ratio ranging from 1 to 6 for different samples. The resultant mix was incubated at room temperature for at least 15 minutes to enable the complexation of the negatively charged gapmers with the cationic lipid. PBS and cell culture medium (pH 7.4) were then added at a volume that was at least five times the volume of the sodium acetate buffer to increase the pH of the suspension, inducing the fusion of the lipid vesicles and the encapsulation of the gapmers into the final LNPs.

For in vivo applications, gapmers were dissolved in 25 mM sodium acetate buffer (pH 4.0), with their concentration adjusted to achieve an N:P ratio of 3:1 in the final LNP formulation. The organic phase and aqueous phase were combined using a T junction microfluidic mixer as described above ¹¹⁵. Following microfluidic mixing, the resulting LNP suspension was dialyzed against PBS (pH 7.4) at room temperature overnight. After dialysis, the gapmer-loaded LNPs were filtered and concentrated as described above. The final LNP formulations were stored at 4°C before use. The formulations were evaluated for gapmer concentration and encapsulation efficiency, and diluted in PBS to the desired concentration to achieve the target dose.

Encapsulation efficiency assessment of gapmers in LNPs

Right before use, the efficiency of gapmer encapsulation into the LNPs was assessed using the Quant-iT OliGreen assay (Life Technologiies, Eugene, OR, USA) in a 96-well plate format. A standard curve was prepared by serially diluting unencapsulated nucleic acid standards (10 μg/mL) in PBS across the first column, with each dilution plated in triplicate. LNP samples were diluted to approximately 2 μg/mL nucleic acid concentration and plated in triplicates under two conditions: with and without Triton X-100. TE buffer (with or without Triton X-100) was prepared, and 50 μL was added to respective wells. Quant-iT reagent was incorporated into the buffers immediately before use, and fluorescence was measured using a SpectraMax M3 Multi-Mode Microplate Reader (excitation/emission: 485/528 nm). Encapsulation efficiency was calculated as: encapsulation efficiency (%) = {(total nucleic acid–unencapsulated nucleic acid)/(total nucleic acid)} ×100, where total nucleic acid was determined from LNPs treated with Triton X-100, and unencapsulated nucleic acid from untreated LNPs. All measurements were performed in triplicate to ensure reproducibility.

Mouse husbandry, genotyping, injections, and sampling

All wild-type B6 mice (The Jackson Laboratory, Bar Harbor, ME, USA), FVB/NJ mice (The Jackson Laboratory), FVB; iDUX4pA (onX); HSA-rtTA (iDUX4pA), and ACTA1-MCM;FLExDUX4/+ transgenic mice used in this study were housed at the Health Sciences Laboratory Animal Services (HSLAS), University of Alberta, AB, Canada. Mice were maintained in individually ventilated cages under a 12-hour light/dark cycle with ad libitum access to standard rodent chow and water.

In vivo testing using iDUX4pA mice:

To induce DUX4 expression, six-week-old (± 2 days) female iDUX4pA mice carrying both the iDUX4pA and HSA-rtTA transgenes were fed doxycycline chow (625 mg/kg doxycycline, Envigo, Indianapolis, IN, USA). Genotyping was performed using PCR on DNA extracted from ear notch biopsies to confirm the presence of both iDUX4pA and HSA-rtTA transgenes. The iDUX4pA transgene was detected using Loxin primers (Forward: 5’- ATA CTT TCT CGG CAG GAG CA -3’; Reverse: 5’- CTA GAT CTC GAA GGA TCT GGA G -3’). The HSA-rtTA transgene was identified using HSA-rtTA Transgene primers (Forward: 5’- CGC TGT GGG GCA TTT TAC TTT AG -3’; Reverse: 5’- CAT CTC CAG ATC GAA ATC GTC -3’). To verify PCR efficiency, an internal positive control (IPC) amplicon (200 bp) was generated using HSA-rtTA IPC primers (Forward: 5’ - CAA ATG TTG CTT GTC TGG TG – 3’; Reverse: 5’ - GTC AGT CGA GTG CAC AGT TT – 3’).

All in vivo treatments were administered intravenously via the retro-orbital route using insulin syringes, with a maximum injection volume of 95 μL. Wild-type B6 mice and iDUX4pA mice identified through genotyping were randomly assigned to treatment groups. The researcher performing the injections was blinded to the treatment assignments. Healthy and nontreated control mice received an equivalent volume of PBS. At the designated endpoint, mice were humanely euthanized for sample collection.

In vivo testing using ACTA1-MCM;FLExDUX4/+ mice:

ACTA1-MCM;FLExDUX4/+ mice were generated by crossing hemizygous B6(Cg)-Gt(ROSA)26Sor^tm1.1(DUX4*)Plj/J (The Jackson Laboratory) and hemizygous B6.Cg-Tg(ACTA1-cre/Esr1)2Kesr/J (The Jackson Laboratory). The mice were genotyped using a previously described protocol ⁹². To induce strong DUX4 expression, 7-week-old (± 3 days) mice were assigned to different groups randomly, as described above, and were treated with 5 mg/kg tamoxifen via intraperitoneal injection. Mice received 10 mg/kg of gapmers or a previously published PMO treatment intravenously twice ⁵³. The treatment was administered prophylactically: the first dose was given 3 days prior to tamoxifen induction, and the second dose was administered on the day of induction. Nontreated control mice received an equivalent volume of PBS. Mice were humanely euthanized for sample collection 7 days after the final dose.

From all animals used in this study, blood samples were collected and kept at 4°C for 1 hour before being centrifuged at 4°C to isolate serum. Urine samples were also centrifuged similarly to remove debris and particulates. Muscle and organ tissue samples were mounted using tragacanth gum, rapidly frozen in liquid nitrogen–cooled isopentane, and left on dry ice for at least 15 minutes before being stored, as previously described ¹¹⁶. All serum, urine and tissue samples were subsequently stored at −80°C until experimental use.

ELISA-based quantification of gapmer uptake

To quantify gapmer biodistribution in vivo, total protein was extracted from 20–30 sectioned tissue slices (each 20 μm thick) using radioimmunoprecipitation assay (RIPA) buffer (Thermo Scientific) supplemented with cOmplete^™ Mini EDTA-free protease inhibitor cocktail (Roche), as described previously ⁸⁹. Gapmer concentrations were quantified from these extracts using a non-competitive ELISA-based method adapted from Yu et al. (2002) with slight modifications ¹¹⁷. Briefly, diluted tissue extracts and gapmer standards were incubated at 37°C for 1 hour in hybridization buffer (60 mM Na₂HPO₄, 0.9 M NaCl, pH 7.4, 0.24% Tween 20 in Milli-Q water) containing 0.025–0.05 μM of the template probe. The template probe was a biotinylated DNA oligonucleotide, complementary to the gapmer sequence, with an additional nine-base overhang (GAATAGCGA) at the 5’ end. The template probes were synthesized commercially by Integrated DNA Technologies. The hybridized mixture was then transferred into avidin-coated 96-well plates (NeutrAvidin^™, black, Thermo Scientific) in duplicates or triplicates and incubated at 37°C for 30 minutes to facilitate binding.

Following incubation, the plate was washed three times with wash buffer (50 mM Tris-HCl, 150 mM NaCl, pH 7.6, 0.1% Tween 20 in Milli-Q water) and twice with double-deionized water to remove unbound components. After washing, 0.067 μM of the ligation probe, along with 400 U/mL T4 ligase (New England Biolabs, Ipswich, MA, USA) and 0.05 mM ATP (New England Biolabs, Ipswich, MA, USA) in 1× One-Phor-All Plus buffer (10 mM Tris base, 10 mM Mg(CH₃COO)2, 50 mM KCH₃COO, pH 7.5 in Milli-Q water), were added to each well and incubated at room temperature for 2 hours. The ligation probe was a DNA oligonucleotide, synthesized commercially by Integrated DNA Technologies, that was complementary to the 5’ overhang of the template probe (5’-TCGCTATTC-3’) and modified with a 5’ phosphate group and a 3’ digoxigenin tag to enable subsequent detection. After incubation, the plate was washed again, and wells were incubated with 1:5000 alkaline phosphatase-conjugated anti-digoxigenin antibody (Sigma Aldrich, St. Luise, MO, USA) diluted in SuperBlock blocking buffer (Thermo Scientific, Rockford, IL, USA) containing 0.0025% Tween 20 for 30 minutes at 37°C. Following another wash step, wells were incubated with AttoPhos^® Substrate (Promega, Madison, WI) for 30 minutes at 37°C to generate the fluorescence signal. Fluorescence was measured using a SpectraMax M3 Multi-Mode Microplate Reader at excitation/emission wavelengths of 450 nm/580 nm emission (auto cut-off). A standard curve was generated using fluorescence values of known gapmer standards and was used to calculate the gapmer concentrations in the samples.

Serum biochemistry profiling and urinary KIM-1 assessment

Serum samples were analyzed commercially by IDEXX BioAnalytics (California, USA) to evaluate a comprehensive panel of toxicity markers, assessing hepatic, renal, and muscular function. The analysis included alkaline phosphatase (ALP), alanine transaminase (ALT), and aspartate aminotransferase (AST) for liver function, blood urea nitrogen (BUN) and creatinine for renal function, and creatine kinase (CpK) for muscle integrity and damage. Additionally, total bilirubin (T.Bil), total protein (T.Prot), albumin, and globulin were measured to assess overall systemic health and protein balance.

For the urinary kidney injury molecule-1 (KIM-1) assessment, urine samples were collected from the mice at the experimental endpoint. KIM-1 quantification, a key indicator of renal tubular injury, was performed using a single-wash, 90-minute sandwich ELISA assay according to the manufacturer’s protocol (Abcam, ab213477).

Histology

Frozen muscle samples were sectioned at 5–7 μm thickness and mounted on poly-L-lysine-coated slides. Sections were thawed at room temperature for 30 minutes, then stained with Mayer’s hematoxylin (Electron Microscopy Sciences) for 15 minutes to visualize nuclei. After hematoxylin staining, sections were rinsed under running tap water for 15 minutes, followed by counterstaining with eosin Y (Electron Microscopy Sciences) for 10 minutes to highlight cytoplasmic structures. The stained sections were then dehydrated through a graded ethanol series (50%, 70%, 90%, and 99%), cleared with a xylene substitute (Thermo Fisher), and mounted with Permount^™ (Fisher Chemical) for long-term preservation. Quantitative analysis, including centrally nucleated fiber counts and minimal Feret’s diameter measurements, was performed by blinded personnel to ensure unbiased evaluation.

For immunostaining, sections were prepared similarly and then incubated in a blocking solution containing PBS with 0.1% Triton X-100 (PBSTx) and 10% goat serum to prevent non-specific antibody binding. Next, sections were incubated overnight at 4°C with rat anti-mouse CD68 primary antibody (1:200, MCA1957T, Bio-Rad) diluted in the same blocking solution. On the following day, sections were washed once with PBSTx and twice with PBS, with each wash lasting 5 minutes. They were then incubated for 1 hour at room temperature in PBS containing DyLight^™ 488-conjugated goat anti-rat IgG secondary antibody (1:200, SA510018) to visualize CD68 expression. After secondary antibody incubation, sections underwent three additional PBS washes (5 minutes each). Finally, slides were mounted using Vectashield HardSet Antifade Mounting Medium with DAPI (Vector Laboratories, Burlingame, CA, USA) to preserve fluorescence and counterstain nuclei. Immunostained sections were imaged the next day using a Zeiss LSM 710 confocal microscope (Zeiss, Oberkochen, Germany). Quantification was conducted by blinded personnel to ensure objective analysis.

Imaging

Immunochemistry images were captures 24-hrs post-staining using a Zeiss Observer 5 fluorescence microscope equipped with an AxioCam 202 (Zeiss, Oberkochen, Germany) or Zeiss LSM 710 confocal microscope. Quantification was conducted by blinded personnel to ensure objective analysis. Visualization of histology slides were conducted using an Optika B-290TB microscope (Optika, Ponteranica, Italy) or ZEISS Axiolab 5 microscope.

Functional assessment

Forelimb and total grip strength were assessed using the Chatillon DFE II grip strength meter (Columbus Instruments, Columbus, OH, USA) to evaluate neuromuscular function and limb strength. For forelimb grip strength, each mouse was placed on the wire mesh platform of the grip strength meter, ensuring that only the front paws were allowed to grip the metal grid. The mouse was then gently and steadily pulled backward by its tail until it fully released its grip, at which point the peak force exerted was recorded. For total grip strength, the procedure was identical, except that the mouse was positioned so that all four paws gripped the mesh before being pulled away in a controlled manner. Each mouse underwent five trials, with 10–15-minute rest intervals between trials to prevent fatigue. Grip strength values were normalized to body weight for comparative analysis.

Treadmill running endurance was assessed using the Exer 3/6 animal treadmill (Columbus Instruments, Columbus, OH, USA) to evaluate exercise tolerance and muscle endurance. Mice were placed on the treadmill and subjected to the following incremental running protocol: 5 m/minute for the first 5 minutes as an acclimation phase, then speed increased by 1 m/minute every minute until the mouse reached exhaustion. Exhaustion was defined as the point at which the mouse failed to resume running within 10 seconds despite gentle nudges to encourage movement. All functional assessments were performed by blinded personnel to ensure unbiased data collection.

Statistical Analysis

All statistical tests were performed using Prism 10 (GraphPad Software, La Jolla, CA, USA). Unpaired two-tailed one-way ANOVA with post-hoc Tukey’s or Dunnett’s multiple comparisons tests were conducted as appropriate. Normality of residuals were tested using Shapiro-Wilks test, where applicable. Analyses were performed at α-levels of 0.05, 0.01, 0.001, and 0.0001 (p = 0.05, p = 0.01, p = 0.001, and p = 0.0001, respectively).

Anwar and colleagues report that lipid nanoparticle (LNP) delivery significantly enhances the efficacy and safety of antisense gapmers targeting DUX4 in facioscapulohumeral muscular dystrophy. These proof-of-concept findings point to the potential of LNP platforms to help address delivery challenges in gapmers and other RNA-based therapies for complex neuromuscular disorders.

Data availability statement

The datasets generated and analyzed during the current study are available from the corresponding author upon reasonable request. Any additional information required to interpret, verify, or extend the findings of this study can be obtained by contacting the corresponding author.