Behavioral improvement in dystrophic mdx23 mouse following repeated antisense oligonucleotides injections

Abstract

Duchenne muscular dystrophy (DMD) is a severe neuromuscular disorder caused by mutations in the DMD gene that disrupt the production of functional dystrophin proteins. Intellectual disability and neurobehavioral complications including autism spectrum disorder, attention-deficit disorders, and anxiety cumulatively occur in 33%–43% of the patients due to deficiency of multiple dystrophin isoforms produced in brain. Previous work also identified behavioral abnormalities in the mdx23 mouse model of DMD. In this work we mapped the expression of the different dystrophin isoforms in different areas of the mouse brain. Next, we determined the behavioral phenotypes that best differentiate mdx23 (lacking the Dp427 isoform) and wild-type mice. Finally, we investigated the response to intracisternal magna (ICM) injection of exon-skipping phosphorodiamidate morpholino oligomer (PMO) antisense oligonucleotides, which induces skipping of exon 23 and restores the reading frame on these phenotypes. PMO administration led to low, detectable, restoration of dystrophin protein and DMD exon skipping in different brain regions. Treated mdx23 male mice exhibited a small but significant rescue of their enhanced fear response. We conclude that ICM delivery of PMO leads to low levels of dystrophin restoration, but these levels are sufficient to elicit a modest behavioral phenotype in mdx23 mice.

Graphical abstract

Intracisternal magna (ICM) injection of exon-skipping phosphorodiamidate morpholino oligomer (PMO) antisense oligonucleotide, which induces skipping of exon 23 of the DMD gene and re-establishes the reading frame, restored low levels of dystrophin eliciting a modest improvement in the behavioral phenotype in the mdx23 mouse model of Duchenne muscular dystrophy (DMD).

Introduction

Duchenne muscular dystrophy (DMD) is a severe neuromuscular degenerative disorder which results from out-of-frame mutations such as deletions, duplications, or nonsense mutations in the DMD gene, located in chromosome Xp21.2.¹ DMD is one of the largest genes in the human genome with a size of approximately 2.5 Mb, and it is composed of 79 exons and 7 tissue-specific promoters.² The DMD gene generates various dystrophin isoforms either by alternative splicing, by using different internal promoters, or both. Dystrophin isoforms are named as Dp (dystrophin protein) followed by a number that represents the protein’s molecular weight in kDa. The three largest isoforms have a molecular weight of 427 kDa and are transcribed via three distinct tissue-specific promoters (M, C, and P) and a common set of exons (2–79).³ These Dp427 isoforms are virtually identical, although they contain unique amino termini encoded by their respective first exons, and are named based on the tissue of highest expression: muscle (Dp427m), cortex (Dp427c), and Purkinje cells (with Dp427p1 and Dp427p2 variants).⁴ Shorter dystrophin protein isoforms are expressed in neural and non-neural tissues such as Dp260 in retina, Dp140 in developing kidney and developing adult brain, Dp116 in peripheral nerves and cardiac muscle,⁵ while the shortest Dp71 and Dp40 isoforms are ubiquitously expressed. The expression of these isoforms is driven by promoters embedded in intronic sequences within the DMD gene with the exception of Dp40, which derives from alternative splicing leading to a different 3′ UTR of Dp71 and is the main dystrophin isoform expressed in the brain.³ DMD encodes the dystrophin protein Dp427m, a crucial member of the dystrophin associated protein complex (DAPC) with a critical role in maintaining the structural integrity of skeletal muscle. Expression profiles of the Dp427m isoform have been well characterized, in skeletal muscle in both human and animal models,⁶ whereas the number of studies on the localization of other isoforms in the CNS is currently limited.⁴

The clinical onset of DMD is between 3 and 5 years of age, with patients not only experiencing severe skeletal muscle deterioration but also facing progressive cardiac and respiratory insufficiency.^7,8 Moreover, individuals with pathogenic variants affecting only Dp427 isoforms (i.e., carrying mutations located upstream of intron 44) frequently present neurobehavioral comorbidities including autism spectrum disorder (ASD), attention-deficit/hyperactivity disorder, and anxiety disorders.^9,10

The lack of multiple brain dystrophins in patients with mutations located further downstream is also responsible for both intellectual disability and enhanced risk of neurobehavioral complications in DMD patients. These patients are indeed twice as likely to demonstrate emotional problems including anxiety or depression compared with patients with other chronic diseases.^10,11,12 Several studies have also demonstrated a correlation between multi-isoform loss and the cognitive profile of DMD patients, who globally exhibit intelligence quotient (IQ) scores one standard deviation lower than those of the general population.^13,14 While the IQ of patients only lacking the Dp427 isoform is within normal limits, the loss of Dp140 and Dp71 has the greatest impact on cognition and IQ, with patients carrying mutations affecting Dp71 having the most severe intellectual disability.^15,16 Considering the location of the internal promoters along the DMD gene, as the mutation affects genomic regions further distally along the gene, more Dp isoforms become affected resulting in the variable deficiency of brain.¹⁷

The effects of DMD mutations on behavior have also been studied in animal models of DMD. Previous research revealed that mdx23 mutant mice also show cognitive function-related behavioral abnormalities.¹⁸ The mdx23 mouse carries a nonsense mutation in exon 23, leading to a premature translation termination codon causing reduced levels of full-length dystrophin mRNA and a complete loss of Dp427 isoforms.¹⁹ These mice present long-term spatial and recognition memory deficits as well as altered fear memory and increased anxiety.^18,20,21 Further investigation of the patterns of dystrophin expression in the CNS can provide us with insights into the neurological aspects of DMD and related disorders.²² This could lead to the identification of potential targets for the development of pharmacological or genetic therapies for neurobehavioral comorbidities associated with the lack of dystrophin in the CNS. The US Food and Drug Administration (FDA) has approved several nucleic acid therapies for DMD, utilizing phosphorodiamidate morpholino oligomers (PMOs), including casimersen, eteplirsen, viltolarsen, and golodirsen. The nucleic acid therapies are strategized to skip particular exons using antisense oligonucleotides (ASOs) in order to restore the reading frame of the DMD gene, resulting in a shortened but functional transcript.²³ These treatments primarily target the skeletal muscle symptoms of DMD, but they do not address the neurobehavioral symptoms associated with brain dystrophin deficiency as they do not cross the blood-brain barrier (BBB). An important step toward developing therapies that could also address the CNS-related unmet needs of patients with DMD is to better understand the extent to which behavioral improvement can be achieved postnatally. Such approaches have the potential to alleviate the neurobehavioral comorbidities DMD patients develop.

Multiple pre-clinical studies have been conducted focusing on restoring dystrophin in the brain, employing various ASO modifications such as tricyclo-DNA^24,25 and PMO,²⁶ as well as AAV gene therapy, to deliver ASOs.²⁷ These studies demonstrated that partial restoration of dystrophin can lead to an improvement in the emotional behavior of murine models.²⁸ However, due to limitations in the efficacy, delivery of the treatment and its invasiveness, further research is required to determine the optimal levels of restored protein within specific regions of the brain necessary for improving the brain phenotype.

We hypothesize that acute and chronic stress have a more pronounced effect on DMD mouse models compared with wild-type (WT) mice. This differential stress response may result in a distinct baseline for DMD mice in neurobehavioral tests, attributed due to their heightened reaction to human handling stress. Acute and chronic stress are known to affect the nervous system differently in both humans and mice, which could influence the assessment of treatments targeting brain dystrophin. For instance, acute stress has been shown to enhance working memory in WT mice by increasing glutamatergic transmission in the cortex,²⁹ and to boost neurogenesis and memory formation by amplifying hippocampal temporal responses.^30,31

The aim of this study was to first elucidate both the neurobehavioral phenotype and the expression patterns of various dystrophin isoforms in the mdx23 mice, with the ultimate goal to: (1) better understand the neuropathological aspects of CNS comorbidities in DMD and (2) confirm findings of previous behavioral investigations of the mdx23 model assays, which had been suggested as reliable readouts in the pre-clinical evaluation of brain dystrophin restoration therapies, and explore the yield of novel tests. We then aimed to assess whether brain-targeted administration via the intracisterna magna (ICM) of PMO ASO therapy, which has been shown to restore muscle dystrophin expression in DMD patients,³² in 6-week-old mdx23 mice can achieve a sufficient biodistribution and exon skipping efficiency to restore dystrophin expression in different brain regions. This study therefore comprehensively examines the efficacy of PMO treatment in mdx23 mice by analyzing isoform expression at the molecular and protein levels in different brain regions, and by employing neurobehavioral tests to examine the rescue of key emotional behavior phenotypes, such as anxiety and increased fear response, characteristic of this mouse model.

Results

Acute stress increases immobility in mdx23 mice, but full recovery occurs within 24 h

In group 1 (Figure S1), we first investigated the influence of acute exposure to an aversive restraining event over 24 h. Open-field (OF) locomotor activity parameters, including the immobility of mdx23 mice were measured to determine the interval needed for mdx23 mice to show a complete recovery of motor function after exposure to a single restraint, as well as a repeated restraint post-recovery from the first. Mice were restrained by grasping both neck and dorsal skin folds between thumb and index, second and third fingers while securing the tail with the fourth finger; once restrained, mice were oriented with the abdomen facing upward and maintained in this position for 15 s. Gentle handling consisted in open cup holding of the mouse while maintaining a light grip at the base of the tail for 15 s.

Immediately after the first restraint mdx23 mice showed significantly higher immobility durations (Figure S2B), lower mobility durations (Figure S2I), and reduced ambulatory distances (Figure S2P) compared with WT littermates, while both mdx23 and WT animals followed similar locomotor behavioral patterns 6 and 24 h post-restraint after both session 1 (Figures S2C, S2D, S2J, S2K, S2Q, and S2R) and session 2 (Figures S2F, S2G, S2M, S2N, S2T, and S2U). This indicates that dystrophic mice fully recover their motor activity levels within 6 h after exposure to a single restraint. The difference observed between WT and mdx23 animals in all locomotor behavioral parameters (Figures S2E, S2L, and S2S) was reduced after a second restraint, performed 2 days later (Figure S1), while still remaining significant between the genotypes, suggesting that no habituation to this stimulus occurs in mdx23 mice after two restraint sessions. Based on these results, two new cohorts of mdx23 and WT mice (groups 2 and 3) were subjected to a set of behavioral assays as illustrated in Figure 1 to investigate the effects of repeated administration of a stressor (15 s restraint) on emotional and cognitive functions. Animals of both genotypes were assigned to receive a daily 15-s session of either manual restraint (stressed mice, experimental stress treatment) or gentle handling (naive mice, control stress treatment) starting from 3 days before the initiation of behavioral testing. The stress treatments were performed with a minimum interval of 12 h from a behavioral assay to allow complete recovery of activity levels in mdx23 mice and were accompanied by 5 min of OF testing pre- and post-treatment to examine locomotor and anxiety/fear-related behaviors throughout the in vivo experimental phase.

Figure 1.

Timeline of behavioral studies for experimental groups 2, 3, and 4

(A) Group 2 consisted of the following subgroups: WT-naive, n = 12; WT-stressed, n = 12; mdx23-naive, n = 12; mdx23-stressed, n = 14. Half of the animals for each subgroups underwent the social interaction test on day 3, whereas the others performed the social interaction test on day 4. (B) Group 3 consisted of the following subgroups: WT-naive, n = 12; WT-stressed, n = 12; mdx23-naive, n = 12; mdx23-stressed, n = 14. Stress treatment consisted in either 15 s of manual restraint (stressed) or 15 s of gentle handling (naive). (C) Timeline of behavioral assays performed on group 4.

During the 7 days of OF testing as described above, a general rise in baseline immobility time (pre-stress treatment) was observed (Figure 2A). This increase was mildly more pronounced in mdx23 mice but was not affected by gentle handling versus restraint. A general decrease in baseline time spent within the center area of the OF (Figure 2C) and total distance traveled (Figure 2E) was also detected across the same testing period. This decrease appeared more pronounced in WT mice, and unaffected by stress treatment.

Figure 2.

Daily open-field activity test results before and after the stress treatment in naive versus stressed, mdx23 versus WT mice, groups 2 and 3

The OF activity was monitored for 5 min before (baseline) and after 15 s of scruffing (stressed groups) or gentle handling (naive groups). (A–F) Line graphs and comparative analysis (mixed-effects model, REML fitting method, and Tukey as post-hoc analysis) of immobility times, times spent in the OF center area and distance traveled across treatment groups. n = 23–26 per treatment group. All data represented as mean (SD) (line graphs), post-hoc analysis can be found in Table S1; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

Remarkably, gentle manipulation initially induced greater fear and anxiety in mdx23 mice than in WT mice (groups 2 and 3). In the stressed mdx23 group, higher immobility time (Figure 2B, p < 0.0001), lower time spent in the center area (Figure 2D, p < 0.0001), and total distance traveled (Figure 2F, p < 0.0001) were significantly different compared with stressed and naive WT mice while no significant difference was observed between stressed and naive WT mice (post-hoc analysis can be found in Table S1). This suggests a heightened trait anxiety in mdx23 mice compared with WT mice. It is also noteworthy that times spent in the center area of the OF remained consistently low across the 7 testing days in restrained mdx23 mice, generally lower than those of unrestrained/restrained WT and unrestrained mdx23 mice. Nevertheless, the effect of stress treatment on this variable (time × stress treatment) was non-significant (Figure 2D).

Daily restraint stress elevates anxiety-like behavior in mdx23 mice

The light/dark box (LDB) and elevated zero maze (EZM) tests were performed to assess anxiety-like behaviors and exploration in both mdx23 and WT mice (naive and stressed). Anxious animals tend to spend more time in enclosed, dark areas of the LDB and EZM, exhibiting reduced exploration of the open, brightly lit sections of these apparatuses.

In the LDB assay, both naive and stressed mdx23 mice did not show any significant differences compared with stressed or naive WT littermates in either latency to enter the light box or in time spent in the light box parameters (Figures 3A and 3B). However, mdx23 animals demonstrated significantly lower distances traveled in the OF than WT littermates, regardless of stress treatment (Figure 3C). Overall, findings demonstrated no influence of the stress treatment on anxiety-related responses, and a specific effect on locomotor activity due to the loss of Dp427. Similarly, no significant differences between genotypes in latency to enter, duration of time spent, and number of entries into the open sections of the maze (Figures 3D, 3E, and 3G) were observed in the EZM assay. The effect of stress treatment could not be detected on any of the variables measured, but overall mdx23 mice showed a significant reduction in distances moved in the maze compared with WT littermates (Figures 3F and 3G, p < 0.05).

Figure 3.

Behavioral test results in naive versus stressed, mdx23 versus WT mice

(A–C) Light/dark box (LDB) test (mdx23 group 2 versus WT mice). (D–G) Elevated zero maze (EZM) test results (mdx23 group 3 versus WT mice). (H–J) Passive avoidance (PA) test results (mdx23 group 3 versus WT mice). (J) Percentage variation in latency to enter the dark chamber from acquisition to retention, calculated as the retention/acquisition latency ratio (shown in %). All data are represented as mean (SD), with n = 11–14 per group. Data were analyzed by two-way ANOVA with degrees of freedom = 1, followed by Holm-Šídák post-hoc tests; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. White and gray bars, respectively, represent naive and stressed mice.

To explore the effect of repeated stress on fear-related cognition phenotypes previously identified in mdx23 mice, naive and stressed mutant or WT animals were further tested in the passive avoidance (PA) assay,³³ which investigates associative learning and memory functions in response to a single aversive stimulus (foot shock). Performance in this test suggested a mild yet significant effect of genotype (Figure 3J, p < 0.05), but not stress treatment, in the retention of the association between the dark chamber of the PA apparatus and the aversive stimulus received within this chamber 24 h after the acquisition session. While no significant differences were observed in both naive and stressed mdx23 mice compared with WT littermates when the mean latency to enter the dark chamber during acquisition or retention were analyzed (Figures 3H and 3I), additional behavioral assays, namely the three-chamber social interaction test (Figure S3A) and the T-maze delayed alternation test (described in the supplemental information), were performed to further explore the effect of stress in mdx23 versus WT mice. No significant differences were observed in any of the parameters measured in either assay across experimental groups test between WT and mdx23 mice (Figures S3B–S3F and S4A–S4D).

Gene expression of full-length dystrophin isoforms in different brain areas of WT mice

Transcriptional expression of different Dp427 isoforms was investigated for the first time using qPCR in six distinct brain structures (cortex, cerebellum, hindbrain, hippocampus, olfactory bulb, and midbrain/striatum) of WT animals (11-week-old mice, n = 9). Dp427cm, which was previously known as muscle specific was observed in all six regions tested and Dp427c expression was consistently significantly higher than all other isoforms in all brain regions, except in the cerebellum where it was expressed in similar amounts compared with Dp427p1 (Figures 4A–4F). Notably, the cerebellum was the only region in which Dp427p1 and Dp247p2 transcripts were detected (Figure 4B). It must be noted that the olfactory bulbs presented high interindividual variability of expression within samples (Figure 4E).

Figure 4.

Dystrophin expression across the WT mice brains at RNA and protein levels

(A–F) Gene expression of Dp427 sub-isoforms in cortex, cerebellum, hindbrain, olfactory bulb, hippocampus, and midbrain/striatum (n = 6). (G) Ratio of the long isoform of dystrophin (Dp427) to the total protein produced in each brain and spine structure examined, as assessed via WES for different regions of the brain and the spinal cord of C57BL/10 mice (n = 4–6). To normalize, total protein was quantified per individual sample using the total protein detection module. Data were analyzed via repeated measures one-way ANOVA and are presented as mean (SD); ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

Analysis of Dp427 protein levels in various CNS regions of WT mice reveals no regional differences

Simple western blot (WES), a gel-free, blot-free, and hands-free protein characterization instrument, was used to quantify the amount of Dp427 protein present in different CNS regions (cortex, cerebellum, hindbrain, hippocampus, olfactory bulb, midbrain/striatum, and spinal cord) of WT mice. Quantification was performed relative to total protein levels measured individually in each sample for all brain regions. Quantitative Dp427 protein data obtained in this assay exhibited high interindividual variability in most of the CNS regions examined, including the cerebellum, hindbrain, and hippocampus, but no significant differences was observed in the relative amounts of Dp427 across different CNS regions of WT mice (Figure 4G).

PMO-treated mice displayed significantly lower anxiety in EZM 7 weeks post-PMO ASO injections

Findings from the first section of the study (Figures 2 and 3) indicated the EZM and the OF test as the behavioral assays most suitable to examine anxiety-related phenotypes correlating with the central role of Dp427 in mdx23 mice.

ICM-delivered PMO did not significantly reduce the heightened restraint-induced immobility behavior in the OF study at 2, 5, and 7 weeks post-treatment (Figures 5A–5D); moreover, PBS-treated and PMO-treated mdx23 cohorts presented similar rearing times (Figure 5E) and rearing counts (Figure 5F), whereas PBS-treated WT mice showed significantly higher values for both variables compared with the two mdx23 groups (Figures 5E and 5F, p < 0.0001) (post-hoc analysis can be found in Table S1). No significant differences were detected in total ambulatory distance traveled at 2, 5, and 7 weeks post-injection between PBS-treated and PMO-treated mdx23 mice (Figures 5G–5I); representative tracking plots obtained from OF tests 7 weeks post-injection can be found in Figure S5. Additionally, 5 weeks post-treatment, PMO treatment did not yield significant improvements in the hyper-anxious phenotype (i.e., avoidance of open areas) displayed by mdx23 mice during the EZM test both when measured manually (Figure 5J) and via automated scoring through ANY-maze (Figure 5L). However, at 7 weeks post-treatment, PMO-treated mdx23 mice exhibited a modest but significant increase in the time spent in the open sections of the EZM, when measured manually by an experienced scorer blind to genotype/treatment (Figure 5K; p < 0.01), while only a trend toward increase was observed when time in the open sections was automatedly scored (Figure 5M).

Figure 5.

Behavioral and anxiety-like assessments in PMO-treated mice over time

(A–C) Percentage of restraint-induced immobility. (D–G) Locomotor activities, including rearing (counts and time) and jumping counts. (H–J) Total distance traveled in the OF 2, 5, and 7 weeks post-PMO treatment. (K–R) Free exploratory activity in EZM was monitored 5 and 7 weeks post-treatment to assess anxiety-related behavior. Behavior in the EZM was measured using both manual scoring (I, J, M, N, Q, and R) by an experienced, experimenter in a blinded manner and automated scoring by the ANY-maze software (K, L, O, and P). n = 32–36 for 2 and 5 weeks post-injection and n = 18–21 for 7 weeks post-injection. Statistical analyses within a single time point: one-way ANOVA with Tukey’s post-hoc multiple comparisons, or Kruskal-Wallis test with Dunn’s post-hoc multiple comparisons. Statistical analyses across time points: mixed-effects model (REML) with Holm-Šídák’s multiple comparisons. All data are presented as mean (SD); ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

Additionally, other anxiety/exploration-related behavioral indexes measured in the EZM, i.e., the latency to enter open areas (Figures 5N and 5O), the number of stretch-attend postures (Figures 5P and 5Q), and the number of head dips (Figures 5R and 5S), indicated levels of anxiety and exploratory motivation intermediate between control WT and control mdx23 mouse groups for PMO-treated mdx23 animals. At 7 weeks post-PMO injection, all the above parameters presented significant differences in the comparison of PBS-treated WT versus PBS-treated mdx23 groups. However, the latency to first enter open EZM sections showed a trend toward decrease to PBS-treated WT levels in PMO-treated mdx23 mice 7 weeks post-injections, as no significant difference was detected for this parameter either between PBS-treated and PMO-treated mdx23 mutants, or between PBS-treated WT and PMO-treated mdx23 animals (Figure 5O). Similarly, a trend toward increase to PBS-treated WT levels was observed in the number of stretch-attend postures exhibited by PMO-treated mdx23 mice in the EZM (Figure 5Q), which did not significantly differ from those displayed by either PBS-treated mdx23 or PBS-treated WT animals.

PMO-treated mice have restored dystrophin protein and skipped DMD exon 23 at 5 and 7 weeks post-treatment

The exon skipping results showed that the highest amount of exon 23 skipped DMD occurred in cerebellum at both 5 and 7 weeks post-treatment; however, there was variation within the amount of skipped dystrophin in different samples (Figures 6A and 6B), with cerebellum showing significantly more skipped dystrophin compared with hindbrain, hippocampus, and midbrain/striatum 5 weeks post-injections (Figure 6A, p < 0.05). Dystrophin protein restoration was measured using WES, with each sample reading normalized to its respective total protein content. Then, dystrophin was quantified by being compared with a standard curve generated by adding different ratios of mdx23 (0% dystrophin) and WT (100% dystrophin). The olfactory bulb showed significantly higher dystrophin restoration compared with other areas including midbrain, cervical spinal cord, hippocampus, and hindbrain at both 5 and 7 weeks post-treatment (Figures 6C and 6D, p < 0.001 and p < 0.0001). The other regions including cerebellum, cortex, hindbrain, hippocampus, hindbrain, and midbrain presented similar levels of dystrophin restoration at both time points. Although both exon skipping (Figure 6A) and protein restoration (Figure 6C) peaked at 5 weeks post-injection, behavioral improvement in the EZM was observed 7 weeks following treatment. The time lag between molecular changes and behavioral effects might possibly be due to the time needed for the restored dystrophin to become functionally active in neural circuits.

Figure 6.

Dystrophin restoration 5 and 7 weeks post-PMO injection

(A and B) Exon skipping 5 (n = 8, 12-week-old mice) and 7 (n = 4, 14-week-old mice) weeks after the last treatment. (C and D) Protein restoration levels at 5 (n = 8, 12-week-old mice) and 7 (n = 4, 14-week-old mice) weeks post-PMO treatment. (E–K) Detection of restored dystrophin protein by immunohistochemistry in mouse brains 5 and 7 weeks post-PMO treatment. Purkinje cells were identified by their distinct microanatomical organization in a monolayer between the molecular layer and the granule cell layer, and their distinctive morphology, with large cell bodies containing large, vesicular nuclei, and a characteristic dendritic arbor within the overlying molecular layer (Dr. Rahul Phadke, personal communication). Representative images showing NCL-DYS1 (1:5), AF594 (1:1,000, red) and DAPI (blue) staining the mice 5 weeks post-treatment on sagittal fresh frozen brain sections in (E) WT, (F) WT, no primary, (G) mdx23, and (H–J) 3 mdx23 mice, 5 weeks post-PMO treatment. (K) 1 WT, (L) 1 mdx23, 7 weeks post-PBS treatment, and (M) 1 mdx23 PMO-treated, white arrows showing restoration of dystrophin protein in Purkinje cells within the cerebellum following treatment with ICM PMO. Brains captured at 40× at same exposure settings. Scale bars, 100 μM. (N) β-Dystrobrevin and (O) β-dystroglycan protein levels, normalized to the total protein of each sample, in the cerebellum of 7 weeks post-treatment mdx23 and WT mice. Statistical analysis: repeated measures one-way ANOVA with degree of freedom of 8 between treatments, all data represented as mean (SD); ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

Immunohistochemistry using NCL-DYS1 antibody against dystrophin was performed to visualize and localize the dystrophin protein restoration. The WT positive control showed sharp labeling of dystrophin in Purkinje cells (Figures 6E and 6K) in contrast to WT no primary antibody (Figure 6F) and mdx23 PBS-negative controls (Figures 6G and 6L). All three mdx23 mice at 5 weeks post-ICM PMO treatment (Figures 6H–6J) showed low but uniform levels of dystrophin protein restoration in the cell soma and dendrites of Purkinje cells within the cerebellum. In contrast, only one animal out of three had dystrophin protein restoration in the Purkinje cells within the cerebellum 7 weeks post-treatment with ICM PMO (Figure 6M).

The cerebella of mice with the highest dystrophin restoration (0.85%–6.1%) 7 weeks post-treatment underwent WES analysis to quantify the levels of β-dystrobrevin (Figure 6N) and β-dystroglycan (Figure 6O), which are DAPC components in the brain.³⁴ The cerebellum was chosen for analysis as a positive correlation was confirmed between protein restoration in this region and time spent in the open areas of the EZM at the same time point (Figure 7B). Similar amounts of β-dystrobrevin and β-dystroglycan were detected in cerebella from PMO-treated mdx23 and WT mice.

Figure 7.

Correlation of Dp427 protein restoration with the time spent in the open area during the EZM test 7 weeks post-PMO injections

(A) Cortex, (B) cerebellum, (C) hindbrain, (D) olfactory bulb, (E) hippocampus, and (F) midbrain/striatum. The asterisks represent individual mice’s DP427 protein restoration (Dp427/total protein assay) quantified using WES with the time spent in open area (s) during the EZM test when observed manually by an expert scientist. Statistical analysis: mixed-effect analysis with Dunnett’s multiple comparisons test was performed as well as simple linear regression of correlation of protein restoration to time in open area.

The correlation between regional Dp427 protein restoration 7 weeks post-injection and the time spent in open EZM sections at this stage was also assessed in the CNS structures (Figure 7). No statistically significant correlations were observed in most of the brain regions analyzed, i.e., cortex (Figure 7A), hindbrain (Figure 7C), olfactory bulb (Figure 7D), hippocampus (Figure 7E), and midbrain/striatum (Figure 7F), with r² values ranging from 0.02 to 0.32 and all non-significant. However, a significant positive correlation observed between restored cerebellar dystrophin and time spent in the open EZM areas (Figure 7B, r² = 0.36, p < 0.01), suggesting that higher Dp427 expression in this region may be associated with decreased anxiety in this assay.

Discussion

Research providing insights into CNS aspects of DMD is crucially needed to identify candidate brain targets and robust behavioral assessments for the development of drugs or genetic therapies that can rescue dystrophin expression in the brain, as well as establishing more comprehensive treatment approaches for DMD. In this study, we therefore first sought to identify neurobehavioral tests that might be used as outcome measures in pre-clinical studies aimed at restoring full-length dystrophin in the brain.

The most obvious phenotype of the mdx23 mouse³⁵ is a dramatically enhanced stress reactivity, characterized by abnormal defensive behavior (unconditioned fear response) in response to mild stressors such as scruff restraint. This was attributed to amygdala dysfunction and GABAergic disinhibition.^21,34,36

We implemented a wide range of behavioral approaches previously employed to characterize the mdx23 mouse model to further reveal interactions between stress reactivity and emotion/cognition in DMD. We thus examined the effects of acute and sub-chronic stress on mouse behavior³⁷ by comparing groups of animals exposed to either gentle handling or a mild stressor represented by manual restraint for the first time. Most of the previous behavior studies on the effect of restraint on mdx23 mice have exclusively compared stressed (i.e., restrained) WT and mdx23 mice,^21,38 or only in non-stressed (naive to restraint) animals.

To investigate the abnormally enhanced fear response to mild stress and enhanced anxiety profile characterizing the mdx23 mice, we chose to employ the acute restraint-induced immobility response in OF, the EZM, and the LDB assays. These tests have been attributed to altered amygdala functionality,^39,40 which has been suggested to be compromised in mdx23.²⁸ Deficits shown by mdx23 naive mice in the recall of long-term memories (24 h post-acquisition delay), which involves a circuit including hippocampus, prefrontal cortex, amygdala, and other subcortical forebrain structures,⁴¹ were assessed with tests involving both spatial (T-maze delayed alternation) and non-spatial memories (three-chamber social interaction).⁴² The three-chamber social interaction assay was also employed to examine reduced social behaviors previously reported in mdx23 mice.⁴³

Most of the behavioral tests were performed at age 8–10 weeks, as neurobehavioral complications do not worsen with age in DMD mouse models or patients, making the 8- to 10-week period ideal for determining the efficacy of potential therapies, while minimizing the effects of muscle and cardiac pathology.^44,45 There was substantial interindividual variability in some of the behavioral stress responses, such as the latency to enter in EZM, LDB, and PA tests, is likely due to various gene modifiers and/or environmental and epigenetic factors.^46,47 This variability may also be influenced by the molecular regulation of mechanisms linked to stress sensitivity or the individual levels of confidence and boldness in mice,⁴⁸ and has implications for the group size that will be necessary to assess the effects of potential therapies on these parameters, which may require 20–30 mice/group.

Enhanced fear responses and hyper anxiety in mice lacking Dp427

The acute stress represented by manual restraint stimulates the defensive posture that animals perform naturally in the presence of extreme threat.⁴⁹ It is believed that amygdala and its connection to the periaqueductal gray in midbrain play an important role in this defensive behavior.⁴⁹ Our results are in line with previous studies demonstrating that mutations in the DMD gene are associated with increased tonic immobility as a response to an acute stressor.⁵⁰ Importantly, these findings suggested that the 6-h interval for performing additional behavioral tests after restraint performance of mutant mice would not be affected by a reduction in locomotor activity in mdx23 compared with WT littermates.

Repeated restraint stress does not affect cognitive and memory function in mice

Chronic and sub-chronic stress is associated with increased anxiety,⁵¹ social withdrawal,^52,53 and cognition and learning deficits, as well as depression symptoms.⁵⁴ It is established that chronic stress activates the amygdala, prefrontal cortex, and hippocampus.^55,56

Our findings showed an enhanced fear response and higher anxiety in mdx23 mice, while no significant differences were detected due to genotype in tests exploring spatial and social cognition. However, moderate data variability was observed in some of the parameters measured during the chosen behavioral tests, such as the number of entries to the open section in the EZM test in both genotypes; these assays may therefore require higher sample sizes to reach statistical power in multi-factorial data analyses.²¹

It must be noted that, in both naive and stressed states, significantly reduced exploratory activity levels were measured in mdx23 animals compared with WT mice in both the LDB and EZM tests; this effect of genotype has been previously observed in OF studies undertaken by other researchers.⁵⁷ Our results are also in line with a study performed by Lindsay et al., who showed that mdx23 mice undergoing chronic stress in the form of repeated daily restraint exhibit reduced ambulatory distances in the OF.⁴⁸

Association of DMD and ASD has been reported,^58,59 and one of the main characteristics of ASD is the display of social communication and interaction deficits, such as reduced interest in peers and/or difficulty in maintaining relationships.⁵⁴ We found no effect of genotype on social responses or memory in WT and mdx23 mice subjected to a three-chamber social interaction assay, in line with previous findings.⁴³ Importantly, our experiments introduced an additional factor, repeated restraint-induced stress, in the analysis of social behavior in mdx23 mice, but this was not found to affect either sociability or social cognition performance of mutant versus WT mice.

Spatial learning and memory functions, which typically involve the hippocampus, were assessed using the T-maze delayed alternation and did not reveal any significant differences in the performance of WT and mdx23 mice, regardless of stress treatment, supporting findings of intact hippocampal-dependent spatial cognition in the absence of Dp427 obtained via Morris water maze.¹⁷

DMD isoform expression and full-length dystrophin protein quantification

It is known that the long isoforms Dp427c, Dp427p1, and Dp427p2⁴ are expressed at the mRNA level in humans. Here, we further characterized and localized the full-length isoforms in different WT mouse brain regions.

Dp427c is the highest expressed Dp427 isoform in all the brain regions examined in the adult WT mouse brain, except in the cerebellum where Dp427p1 and DP427p2 presented a similar amount of gene expression. Dp427m, the main muscle isoform,^18,60 is also expressed within different regions of the mouse brain which has not been reported previously. Among the different full-length (Dp427) isoforms Dp427c is the most highly expressed in the brain, particularly in cortical and hindbrain structures. The amygdala is one of the principal brain regions involved in modulating anxiety and fear responses^61,62,63; however, we could not analyze the level of dystrophin in this region due to its small size, which hampered accurate isolation by microdissection. Other brain structures involved in anxiety and fear-related processing are the hippocampus,^64,65 the striatum,⁶⁶ and the prefrontal and frontal cortex.^64,67,68 There was a similar pattern of dystrophin protein production in the various areas of the WT brain, although the hindbrain and the cervical spinal cord exhibited a trend toward lower protein production compared with other CNS regions. Additionally, it is important to note that the antibody used here recognizes all full-length dystrophins including Dp42c, m, p1, and p2. Although the specific role of Dp427 in the brain is still not completely clear, the presence of Dp427c in all the regions analyzed and the effect of the absence of dystrophin on the behavior of mdx23 mice bring us closer to fully elucidating its role within the brain.⁶⁹

PMO-treated mice present significantly less anxiety

mdx23 mice treated by ICM injection of three doses of the PMO inducing exon 23 skipping showed a small but significant rescue of an anxiety-related parameter in the EZM compared with untreated mice, when measured manually by an experimenter blind to genotype and treatment. We chose manual scoring for the EZM because certain behaviors are difficult to capture accurately with automated systems. Automated software typically registers an “entry” when a set percentage of the animal’s body crosses into a zone. However, this can overestimate true exploratory behavior, especially when animals only partially enter the open arm, such as when stretching in with their front paws while keeping their hind paws safely in the closed area. In contrast, manual observation allows us to apply the stricter, more widely used criterion of all four paws being in the open area to record an entry.⁷⁰ While manual scoring is more time-consuming and introduces some level of subjectivity, it is important to capture this specific aspect of behavior as accurately as possible. That said, we acknowledge this as a limitation and recognize the need for the field to develop improved automated tools that can better distinguish between full entries and exploratory postures.

Our results indicate that at least some of the neurobehavioral complications due to the lack of Dp427 could potentially be reversed after treatment via PMO injections postnatally.

The tested PMO concentration was well tolerated by the mice, and no toxicity was observed, resulting in 20% of exon 23 skipping in the cerebellum 5 weeks post-treatment and an average of 4% 7 weeks post-injection, after three ICM injections (3× ICM at 900 μg each, total dose 321 nmol). The restoration of the Dp427 protein was up to 10% in the olfactory bulb 5 weeks post-treatment, which was then reduced to an average of 5% in brains 7 weeks post-treatment. This indicates that the effect of PMO reduces after 7 weeks but dystrophin protein is still detectable. Although exon skipping efficiency appeared to be highest in the cerebellum, the level of Dp427 protein restoration here quantified was relatively low. Conversely, the olfactory bulb showed lower skipping efficiency but more pronounced protein recovery. This discrepancy may reflect differences in translational efficiency or protein stability across brain regions. Dp427 protein was also noticeable in the cerebellum, particularly the Purkinje cells by IHC. The restoration of Dp427 protein in the cerebellum of treated mice is of interest, as a positive correlation with the time spent in open EZM sections was observed; suggesting that the cerebellum may have a role in modulating anxiety and fear responses in DMD, while a fundamental role of the olfactory bulb seems not plausible.^71,72

A comparison of the behavioral performance of mdx23 mice, lacking Dp427 alone, to that of mdx52 or mdx4cv mice, lacking both Dp427 and Dp140, and to mdx3cv and Dmd-null mice, lacking all dystrophin isoforms, showed that all present enhanced anxiety, as do children with DMD (regardless of their genotype), supporting a role for the Dp427 isoforms in emotional disturbances.¹⁷ However, mdx52 mice display higher anxiety and deeper fear learning and memory impairments than mdx23 mice, suggesting that Dp140 loss worsens emotional disturbances.⁷³ Mild changes in short-term memory, discrimination learning, and movement patterns have been observed in mdx23 mice and mdx4cv mice, suggesting that Dp427 is involved in these functions.^69,74

Previous studies in the mdx23 mouse model demonstrated that partial restoration of Dp427 by intracerebroventricular (i.c.v.)-delivered PMO using micro-osmotic pumps directly to the amygdala,²⁸ or tricyclo-DNA administered intravenously⁷⁵ (as this ASO can cross the BBB), led to an improvement in the emotional behavior of the mice, improving their unconditioned fear response. In another study, intrahippocampal administration of adeno-associated virus vectors encoding a specific U7snRNA that induced skipping of exon 23 led to recovery of GABA_A receptor clustering and normalization of hippocampal synaptic plasticity, following partial rescue of hippocampal dystrophin.⁷⁶ Subsequent work showed that i.c.v. administration of tricyclo-DNA-ASO not only rescued expression of brain Dp427 isoforms but also significantly restored long-term memory retention of mdx23 mice, evidence that postnatal re-expression of brain dystrophin could alleviate some cognitive deficits associated with DMD.⁷⁷ Using a different mouse model, mdx52, that lacks Dp427 and Dp140,²⁵ Saoudi and colleagues showed that i.c.v. administration of tricycloDNA-ASO targeting DMD exon 51 resulted in partial Dp427 restoration in the brains of treated mice and significantly reduced anxiety and unconditioned fear.²⁴ Additionally, this partial rescue of Dp427 fully normalized the acquisition of fear conditioning, while fear memory tested 24 h later was only partially improved.³⁴

Our study provides further evidence that Dp427 deficiency in mice leads to neurobehavioral changes including an enhanced unconditioned fear response, a heightened anxiety profile, and a deficit in PA learning/memory. We conclude that assays such as the unconditional fear response and EZM tests represent robust tools to preclinically examine the effectiveness of candidate Dp427 restoration therapies in the brain. Using these methods, we showed that ICM injection of three doses of PMO into mdx23 mice induced exon skipping, led to minimal protein restoration in some brain areas and, despite very low levels of dystrophin protein rescue, resulted in a small but significant rescue of enhanced anxiety phenotypes. This is the first study demonstrating how PMO delivered via ICM injection can reach structures previously not targetable via other administration routes, and our findings also suggest that different parts of the circuitry are involved in the observed phenotypic improvement.

The observation that even low levels of dystrophin induced by exon skipping can improve neurobehavioral outcomes is encouraging and supports the development of more efficient ASOs with enhanced efficacy and biodistribution. The four exon-skipping ASO drugs approved by the FDA for DMD (eteplirsen, golodirsen, casimersen, and viltolarsen) are all PMOs, which are unable to cross the BBB and, as they are administered systemically, are not expected to cause any behavioral effects. Recent data appear to suggest that a novel PMO conjugated to Fab fragments binding the transferrin receptor could cross the BBB, but currently no information is available on whether this candidate therapy can rescue behavioral phenotypes, as studies are still in the early dose-ranging phases.⁷⁸

Materials and methods

Animals

mdx23 mice (C57BL/10ScSn-Dmd^mdx/J; Jackson Laboratory, stock no. 001801) were purchased from Charles River and shipped to the Trinity Biomedical Sciences Institute. Heterozygous females were crossed to C57BL/10ScSn male mice to generate mdx23 and littermate control (WT) males. Genotyping was performed with ear biopsy samples, as previously described.⁷⁹

Male WT and mdx23 mice were used for both behavioral tests and molecular analyses. Mice were housed in individually ventilated cages (Tecniplast) in a specific pathogen-free facility on a 12:12 h light/dark cycle with access to food and water ad libitum. All animal care and experimental procedures complied with the national and European legislation and were approved by the Irish Health Products Regulatory Authority (ref.: AE19136/P131) and by the Trinity College Dublin Animal Research Ethics Committee. mdx23 mice possess a spontaneous C-to-T transition in exon 23 of the DMD gene on the X chromosome, leading to an early termination codon at position 2983 (ENSMUST00000114000 chrX:g.83803333 C>T; p.Q995∗).²⁶

Behavioral assays

All the assays were performed in a blinded manner; mice were only identified by a unique ID and data matching to genotype or treatment group was performed after experiments had ended and behavioral measurements had been generated. Experimental room conditions matched maintenance conditions, and illumination was kept at around 200 lux.

Restraint/gentle handling and OF activity test

The OF activity test measures both motor and exploratory behavior as well as anxiety.⁸⁰ The time spent in the center versus wall-adjacent areas in the first 5 min of OF post-restraint, which is before animals habituate to the chamber, is considered as a behavioral index of anxiety (anxious animals will present more pronounced thigmotaxis), while other behavioral parameters measured in center versus wall-adjacent areas in the first 5 min post-restraint are used to assess motor and exploratory activity. Mice were restrained by grasping both neck and dorsal skin folds between thumb and index, second and third fingers, while securing the tail with the fourth finger, and re-orienting the animal so that the ventral part of its body faced the experimenter. Gentle handling of mice consisted in open cup holding while maintaining a light grip at the base of the tail. At the expiration of 15 s of restraint/handling, animals were released approximately in the center of the test arena. After 15 s, the mouse was released for 30 min (in the case of group 1) or 5 min (in the case of group 2 and 3) into a seamless OF activity chamber (27 × 27 × 20 cm; Med Associates) equipped with three 16-beam infrared (IR) beam arrays located on both the x and y axes for positional tracking and z axis for rearing detection. The animal’s movements in the chamber were automatically recorded via the Activity Monitor 7 software (Med Associates), with a threshold freezing duration set at >1 s (representative tracking plots can be found in Figure S5). Unconditioned fear responses induced by the short acute stressor, i.e., the 15 s restraint, were characterized by periods of tonic immobility, the duration of which was measured during the recording period. Complete lack of movement, except for respiration-related motion, was considered as immobility. The percentage of time spent immobile was calculated for group comparisons.⁸¹ Other behavioral indexes measured to examine locomotion and exploration levels across groups were distance moved and vertical activity (jumping, rearing) in the whole arena or in center versus wall-adjacent areas. OF assays were performed in the last 2 h before the dark phase, allowing overnight recovery of activity following restraint/gentle handling.

LDB test

The LDB test is an assay of anxiety-like behaviors in mice based on rodents’ innate avoidance of brightly illuminated spaces and their natural tendency to explore new environments. The test investigates how animals respond to mild stressors in an approach-avoidance conflict situation by presenting animals with well-lit yet novel surroundings to be examined. Hence, the test is performed without any previous training.^21,82 The LDB test was performed during the first half of the light phase using seamless OF activity chambers, as previously described.⁸³ The chambers were fitted with dark box inserts made of black IR-transparent acrylic, creating a dark enclosed compartment (13.5 × 27 × 20 cm; illumination: <5 lx) connected by a small door to an open compartment (13.5 × 27 × 20 cm) brightly lit by an overhead light source (illumination: 600 lx). Each mouse was placed at the center of the dark compartment from the hinged lid at the top of the insert, and was then allowed to freely explore the entire apparatus for 5 min. Step through latency, number of entries, and total time spent in the lit compartment, as well as other locomotor parameters were automatically scored using the Activity Monitor 7 software (Med Associates) with IR photobeam-based tracking.

EZM test

The EZM represents another test of anxiety-like behaviors in rodents in which anxiogenic stimuli are introduced by the presence of brightly lit, exposed areas in between wall-protected sections of the maze. A preference for these enclosed areas is interpreted as indicative of higher anxiety levels.⁸⁴ The apparatus consisted in an elevated ring-shaped runway with two opposite open (unwalled) and two opposite closed (walled) quadrants of equal dimensions (60 × 5 × 16 cm, 62 cm elevation from floor; Ugo Basile) illuminated by overhead lights at 500–600 lx (illumination in closed areas: 30–40 lx). A ceiling-mounted camera recorded the apparatus throughout testing for automated tracking and behavioral scoring using ANY-maze. At the start of each trial, all the mice were placed in the center of a closed quadrant and were then allowed to freely explore the maze for 5 min. Time spent and entries made in the open arms were used as indexes of anxiety-like behavior. Mice were considered to have entered one of the open sections of the maze if all four paws had crossed the threshold between closed and open areas; for automated scoring via ANY-maze, the entry criterion was defined as >90% of an animal’s body area being detected into the open section.

PA test

The PA test is a behavioral assay, which examines the test subject’s ability to learn and retain information about aversive experiences, in this case a mild foot shock.⁸⁵ PA of the dark chamber post-shock administration was tested 24 h after the learning trial, by placing the mouse in the start box and measuring the latency to enter the dark box (up to a maximum of 300 s) (retention trial). The testing apparatus consisted of a cage (47 × 18 × 25 cm; Ugo Basile) equipped with a grid floor and a transparent plastic cage lid and partitioned in two chambers of equal size (19 l x 9 w x 17 h cm). One chamber was enclosed by black walls and unlit (dark box), whereas the other was enclosed by white walls and illuminated by an LED white light (∼400 lumen) installed on the cage lid (start box). The two chambers were divided by an automated sliding door. The procedure entailed habituation, acquisition, and retention phases. The test mouse was first allowed to explore the dark box (door closed) for 120 s, then placed in the start box (door open) until spontaneous transitioning to the less-aversive dark box occurred. The next day, animals were re-exposed following this protocol three times to complete the habituation phase. Immediately after entering the dark box on the last of such trials, the animal received a single, inescapable scrambled foot shock (4 s duration, 0.3 mA intensity) (acquisition trial).

ICM administrations

M23D PMO (sequence: 5′-GGCCAAACCTCGGCTTACCTGAAAT-3ʹ),⁸⁶ a generous gift from Sarepta Therapeutics, was administrated using 6-week-old mdx23 and WT mice anesthetized via intraperitoneal injection of a ketamine (100 mg/kg)/medetomidine (1 mg/kg) mixture. ICM magna administrations were performed as previously described.³⁴ A 30-gauge stainless steel needle with a point 4 style bevel was attached to a Hamilton 25-μL syringe and curved 2 mm from the tip to a 45° angle, so that it was J-shaped. The needle was inserted through the skin at the base of the skull with the mouse head angled downward to expose the cisterna magna. The PMO solution (10 μL; 900 μg in PBS) was administered to 29 mice, 3 times, corresponding to a total injected dose of 2.7 mg, with 3 days of recovery between each infusion. The same amount of PBS was injected in 29 mdx23 mice and 29 WT mice.

Tissue collection

For mapping dystrophin section of the study, nine male WT mice and nine mdx23 mice with an average age of 11 weeks and, for the treatment efficacy section of the study, 15 PMO-treated mice, 14 PBS-treated mdx23 mice, and 15 PBS-treated WT mice were euthanized 5 weeks post-ICM injection, while 14 PMO-treated, 13 PBS-treated mdx23, and 14 PBS-treated WT were euthanized 7 weeks post-ICM injection by cervical dislocation and whole brains and spinal cord harvested. After sagittal hemisection, both brain hemispheres were microdissected for cortex, cerebellum, hindbrain, hippocampus, olfactory bulb, and midbrain/striatum regions. The right hemisphere was used for RNA quantification, while the left hemisphere was used for protein analysis. Brain tissue samples from the right hemisphere were submerged in RNAprotect (QIAGEN, catalog no. 76106) and incubated at 4°C overnight. The RNAprotect solution was then removed, and the tissues were stored at −70°C until analysis. For protein quantification, the microdissected brain regions from the left hemisphere were weighed, placed in cryovials, and snap-frozen in liquid nitrogen, then stored at −70°C.

Gene expression analyses

RNA extraction

The RNA was extracted using RNeasy Mini kit (QIAGEN, catalog no. 74104). Brain tissues were homogenized using TissueLyser II (QIAGEN, catalog no. 85300) after adding buffer according to the RNeasy kit after adding two tungsten carbide beads (QIAGEN, catalog no. 69997) into each tube at 30 cycles/s for 4 min. Samples were then centrifuged at 21,000 × g for 5 min at room temperature. The yield and purity of the samples were then measured using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific) and samples were stored at −70^oC until further use. The RNA samples were reverse transcribed using the high-capacity RNA to cDNA kit (Thermo Fisher Scientific, catalog no. 4387406) according to the manufacturer’s instructions. In order to minimize the gDNA contamination in the cDNA samples, as the probe for Dp140 and Dp40 recognizes gDNA, an additional step of DNase treatment (Invitrogen, catalog no. 18068015) was performed.

Quantitative PCR

Amplification was carried out on a StepOne real-time PCR detection system using TaqMan fast advanced master mix (Thermo Fisher Scientific, catalog no. 4444557), Ube2e1 gene (Integrated DNA Technology, assay ID: Mm.PT.58.12995636) as the housekeeping gene, and custom-designed TaqMan probes (Table 1; all the probes were ordered from Integrated DNA Technology except Dp427p1, which was ordered from Thermo Fisher Scientific as an MGB probe), according to manufacturer’s instructions on a StepOne real-time PCR detection system. Gene expression of different isoforms across the brain regions was performed using isoform-specific TaqMan probes, and to allow comparison between different probes, the primer amplification efficiencies were corrected using LinReg PCR software.⁸⁷ Gene expression analysis for the percentage exon skipping was performed as previously described via the absolute copy number method using gBlocks gene fragments (Integrated DNA Technology) and custom-designed TaqMan probes targeting unskipped and skipped gene fragments (Table 1).⁸⁸

Table 1.

TaqMan probes designed to map dystrophin across the mouse brain

Isoform	Forward primer	Reverse primer	Probe
Dp427m	GGACTGTTATGAAAGAGAAGATGTT	TGGCAGTTTTTGCCCTGTAAG	ACCTGCAGGATGGAAAACGCCTCC
Dp427c	GGCATGGAAGATGAAAGAGAAGA	GGCAGTTTTTGCCCTGTAAGG	ACCTGCAGGATGGAAAACGCCTCC
Dp427p1	CTTTCATCAGAAGAAACCTCAGACA	CAGCCAAATGCTTTCCTATGAAG	AAATTCTGCGGAGGCTG
Dp427p2	CAGGCTTCCCTAAAGATGAAAGAG	GGCAGTTTTTGCCCTGTAAGG	ACCTGCAGGATGGAAAACGCCTCC
Dp140	GCTGACTGTTCTGAGCTAAAATCG	GCCATCCTGGAGTTCCTTAATAAG	ACCAGAAGGGGGTTTTG
Dp71/40	CACTGCCTGTGAAACCCTTACA	TGGGTCTCGTGGCCTTTG	CCATGAGGGAACACC
Dp40	AACGTGAGTAGTGGCAGAAGCA	TTTTGGCTGGGAGGAGTTCA	AGCAAACTTGCATTTGATA
Exon 22-24 junction	GAGCTGTTTTTTCAGGATTTCAG	CGGGAAATTACAGAATCACATAAA	ATCCCCCAGGGCAGGCCATTC
Exon 23-24 junction	CAGGCCATTCCTCTTTCAGG	GAAACTTTCCTCCCAGTTGGT	TCAACTTCAGCCATCCATTTCTGTAAGGT

Protein expression analyses

Protein extraction

The protein samples were prepared on ice for extraction after adding 50 μL of a solution made of protease and phosphatase inhibitor mini tablets (Pierce, catalog no. A32959) to 950 μL of T-PER tissue protein extraction reagent (Thermo Scientific, catalog no. 78510) for each mg of tissue. Tissues were homogenized using an 850 Homogenizer (Fisherbrand, catalog no. 15505819) for 20 s at 10,000 rpm and placed in an ultrasonic ice bath (Branson 2510-DTH Ultrasonic Cleaner) for 5 min. The homogenates were centrifuged for 15 min at 14,000 × g at 4°C; the resulting supernatant was centrifuged again at 14,000 × g at 4°C for 5 min and kept at −70°C. The protein concentration per sample was quantified using the RC DC Protein Assay (Bio-Rad Laboratories, catalog no. 5000120).

Protein quantification

A simple WES (Bio-techne, catalog no. 004–600), a gel-free, blot-free, and hands-free protein characterization instrument, was used to quantify Dp427 levels in protein lysates from C57BL/10 and mdx23 male mouse brain samples, as previously described.⁸⁹ Given the relative low amount of Dp427 in brain, a cocktail of two monoclonal antibodies to dystrophin was used to enhance the signal for Dp427: Mandys1 (1:2000) purchased from the Wolfson Center for Inherited Neuromuscular Disease (CIND) and NCL-DYS1 (1:200) (Novocastra, lot. no. 6096908). To evaluate the restoration of dystrophin-dystroglycan complex, NCL-bDG antibodies (1:200; Leica Biosystems) against β-dystroglycan and DTNB (Proteintech, catalog no. 12045-1-AP) against dystrobrevin were used. In total 2 μg of protein for each sample was prepared for WES analysis following protocols for the 66–440 kDa separation module (Bio-techne, catalog no. SM-W005) for dystrophin quantification, and 12–230 kDa separation module (Bio-techne, catalog no. SM-W001) for β-dystroglycan and dystrobrevin quantification. The protein of interest’s signal was normalized to the total protein detection signal of each individual sample, which was run on the same plate, using the total protein detection module (Bio-techne, catalog no. DM-TP01) as previously described.⁹⁰ To quantify the protein restored after PMO injection, a standard curve was produced by mixing protein extracted from WT (100% dystrophin protein) and mdx23 mice (0% dystrophin protein). The area under the peak for each brain region at the correct size was then compared with the standard curve, in order to have an absolute quantification of protein restored in each brain region.

Localization of restored dystrophin

Fresh frozen brains were sectioned on a Leica CM 1860 UV cryostat (Leica Biosystems, Germany) with the chamber temperature set to −19°C and preferred thickness of 20 μm and were placed on Superfrost Plus Adhesion microscope slides (Fisher Scientific, catalog no. 10149870) and stored at −70°C prior to staining. The slides were thawed at room temperature for 30 min, sections were briefly fixed in pre-chilled 100% methanol for 2 min at −20°C and were washed in PBS. Next, the sections were circled with a hydrophobic pen before being incubated with blocking buffer (2% goat serum, 0.3% Triton X-100, 1% bovine serum albumin diluted in PBS) for 40 min at room temperature then with NCL-Dys1 (Novocastra, lot. no. 6096908) primary antibody diluted in blocking buffer at concentration of 1:5 overnight at 4°C. The slides were washed PBS before being incubated with the secondary antibody conjugated to Alexa fluorophore 594 (Invitrogen, catalog no. A-21135) 1:1,000 diluted in blocking buffer for 1 h at room temperature. All the images were acquired on a Zeiss observer 7 color fluorescent microscope using a 40× dry lens.

Statistical analyses

All statistical analyses were performed using GraphPad Prism 9 (GraphPad Software). A Student’s t test or, in case of data deviation from normality (assessed with the Shapiro-Wilk, Anderson-Darling, Kolmogorov-Smirnov, and D’Agostino-Pearson tests), Mann-Whitney U test was employed to compare individual behavioral outcome variables between two groups. For comparisons across multiple groups, normally distributed datasets were analyzed by one-way ANOVA followed by a Tukey’s multiple comparisons post-hoc test, whereas a Kruskal-Wallis one-way ANOVA on ranks followed by Dunn’s pairwise tests were employed for non-normally distributed data. In case of assays comprising multiple time bins or trials, behavioral indexes were compared by means of a two-way ANOVA (factors: genotype and time/trial) with repeated measures on one factor (time/trial), followed by a Holm-Šídák multiple comparisons test. If data did not meet requirements for use with a parametric model, a restricted maximum likelihood (REML) linear mixed-effects model was applied instead.

Data analysis for both group 2 and group 3 involved distinct approaches. For OF activity measurements, a two-way repeated measures ANOVA or mixed-effects model using the REML fitting method was applied, with factors being genotype, stress treatment, and time. Subsequent Tukey’s multiple comparisons post-hoc tests were performed. Anxiety and cognition test measures were examined across groups via two- or three-way ANOVA, incorporating factors such as genotype, stress treatment, and interaction stimulus where applicable, followed by Tukey’s multiple comparisons post-hoc tests. T-maze alternation rates, expressed as percentage proportions of mice alternating, were compared using a chi-squared analysis.

Data and code availability

Data are available upon request to the corresponding authors.

Acknowledgments

The authors are thankful to Sarepta Therapeutics for providing financial support (grant 557462) for developing this research project. The authors also wish to thank Dr. Rahul Phadke, neuropathologist at University College London Hospital, for validating the location of dystrophin in different brain regions.

Author contributions

A.A., M.D.M., C.F., T.E.G., E.S., S.T., S.S., M.S., J.A.P., and E.S. carried out the experiments. A.A. and M.D.M. designed and performed data analyses. A.A. and M.D.M. wrote the manuscript and prepared the figures. A.A., M.D.M., K.T., J.E.M., F. Montanaro, and F. Muntoni edited the manuscript. V.P.K., J.E.M., F. Montanaro, and F. Muntoni contributed to the conception of the project. F. Muntoni obtained funding for the study.

Declaration of interests

F. Muntoni is an investigator in Sarepta, Genethon, and Roche clinical trials. He is the PI of the Sarepta grant supporting this work (funding to UCL), and has participated in advisory boards and/or symposia for Sarepta, Roche, Dyne therapeutics, Solid, Wave, and Entrada.