Research Article – PLOS Currents Muscular Dystrophy Wed, 17 Oct 2018 20:45:27 +0000 en-US hourly 1 Neurodevelopmental Needs in Young Boys with Duchenne Muscular Dystrophy (DMD): Observations from the Cooperative International Neuromuscular Research Group (CINRG) DMD Natural History Study (DNHS). Wed, 17 Oct 2018 13:45:54 +0000 Introduction: Duchenne muscular dystrophy (DMD) is the most common X-linked neuromuscular condition manifested by progressive skeletal muscle weakness, cardiopulmonary involvement and cognitive deficits. Neurodevelopmental symptoms and signs are under-appreciated in this population despite the recognition that cognition has a major impact on quality-of-life. We describe the neurodevelopmental needs in a large cohort of young boys with DMD from the DMD Natural History Study (DNHS). We explore the association between neurodevelopmental needs and DMD mutation location, and with glucocorticoid use.  

Methods: We prospectively evaluated 204 participants between ages 4 to less than 9 years of age with DMD as part of a large, longitudinal, international DNHS. We obtained parent- or primary care-giver report of neurodevelopmental needs as part of their study visit. We assessed the relationship between parent/care-giver neurodevelopmental needs and DMD mutation location, and glucocorticoid use.

Results: The neurodevelopmental needs that were most commonly reported included speech delay (33%), mild developmental delay (24%), significant behavioral problems (16.5%), language impairment (14.5%), learning disability (14.5%), attention-deficit hyperactivity disorder (5%) and autism spectrum disorder (3%). Neurodevelopmental needs were more commonly reported by care-givers in those with DMD mutations downstream of exon 51. There was no relationship between care-giver reported neurodevelopmental needs and glucocorticoid use.

Conclusion: Neurodevelopmental needs are highly prevalent in young boys with DMD. Care-givers report higher neurodevelopmental needs when subjects have DMD mutations downstream of exon 51. Early interventions aimed at cognitive health are critical to improve the quality-of-life of individuals with DMD.

Trial Registration: NCT00468832



Neurodevelopmental needs are clinically important but often overlooked co-morbidities in Duchenne muscular dystrophy (DMD), the most common X-linked neuromuscular disease that affects 1 in 5000 live births1. Autism spectrum2,3, attention-deficit hyperactivity3,4,5, and obsessive compulsive disorders3 are 4 times more common in DMD than in typically developing children. Developmental delay—particularly expressive language delay—is a core neurological symptom in DMD, and was described as early as Duchenne himself 150 years ago6,7. Over the past several years, there has been increasing recognition of the higher-order cognitive skills that are affected in DMD. These include working memory and executive function deficits8, academic under-performance9, and increased utilization of school resources10. As the life-expectancy in DMD has increased tremendously with survival into the third decade of life, it is increasingly important to identify potential strategies for early and improved screening of neurodevelopmental needs in DMD, and to evaluate pragmatic interventions to improve cognitive health11. Most importantly, neurodevelopmental disorders and cognitive disabilities have a significant negative impact on the quality-of-life on a daily basis, and affects overall health maintenance.

Dystrophin plays a critical role in brain function12. There are several brain-specific dystrophin isoforms, including the full-length dystrophin (dp427), and shorter dystrophin isoforms (dp260, dp140, and dp71) generated by tissue-specific promoters. The following are the unique first exons for the shorter dystrophin isoforms: exon 30 for dp260, exon 45 for dp140, and exon 63 for dp7113,14. In humans, dystrophin is highly expressed by neurons in the cerebellum, the cerebral cortex, and hippocampus. In neurons, dystrophin localizes within the pre-synaptic density (PSD95), a region that is important for dendritic spine formation and synaptic function15.

Few investigators suggest a differential vulnerability in the severity of cognitive involvement when DMD mutations are between exons 45-50, suggesting a critical role for the dystrophin dp140 isoform in cognitive function16,17. Not only do individuals with mutations between DMD exons 45-50 have lower intellectual capacity and impaired information processing compared to individuals with DMD mutations in exons 1-44, they also have smaller total brain volume and gray matter volume18.

We systematically evaluated the prevalence of neurodevelopmental needs, stratified based on DMD mutation location, in a large prospectively followed cohort of 204 boys between ages 4 to less than 9 years with DMD. These 204 boys were recruited as part of the DMD Natural History Study (DHNS) conducted by the Cooperative International Neuromuscular Research Group (CINRG)19. We describe the neurodevelopmental needs in this young cohort at study enrollment as reported by the parent or primary care-giver.

We present data that supports that young boys with DMD have a high prevalence of neurodevelopmental needs as reported by parent or care-giver. Further, boys with DMD mutations between exons 45-50 reported higher cognitive problems. There was no relationship between neurodevelopmental needs and glucocorticoid use. We conclude that there is an unmet, critical medical need in DMD to develop pragmatic solutions for early detection and intervention of neurodevelopmental needs during a window of neurodevelopmental plasticity.


The study was conducted in accordance with the Declaration of Helsinki (2000) and the principles of Good Clinical Practice according to the International Conference on Harmonization. All study participants provided written consent prior to study enrollment. Each site has its IRB which approved the study. At Children’s National Health System, the institutional IRB approved the study (#0159)

Study participants

The study was conducted in accordance with the World Medical Association Declaration of Helsinki. All study participants provided written consent prior to study enrollment. The study was approved by the ethical review committee at each of the participating sites. Study subjects were recruited from academic institutions within the United States and at other international sites between the years 2006 and 2014. Subjects in this data analysis were part of the DNHS, a 10-year multi-center, international, prospective longitudinal study of DMD ( NCT00468832). A detailed study description is available as described in McDonald et al.19. We chose to evaluate the 204 participants who were between age 4 to less than 9 years in order to identify neurodevelopmental needs in early childhood. Briefly, study participants were evaluated every three months during the first year of study enrollment, every 6 months during the second year of enrollment, and annually thereafter. The assessments performed at each study visit included physical examination, reviews-of-body systems, medication history, muscle strength quantification, functional measurements, and quality-of-life questionnaires completed by the parent or primary care-giver.

Dystrophin isoform (dp140) assignment

Information regarding DMD mutation was available in 152 of the 204 participants. The anonymized study participants’ genetic information including exon boundaries of the DMD mutation was reviewed by a genetic counselor. We categorized mutations upstream of DMD exon 44 as dp140+, and those with mutations downstream of DMD exon 51 were categorized as dp140 negative. All subjects whose mutation fell between DMD exon boundaries 45 to 50 were assigned into the dp140 intermediate group. The categorization of study participants is summarized in Figure 1. There were 53 subjects in the dystrophin dp140+ category, 52 subjects in the dystrophin dp140- category, and 48 subjects in the dystrophin dp140 intermediate category.

Fig 1

Fig. 1: Categorization of participants based on DMD exon boundary.

Evaluation of neurodevelopmental needs

We evaluated several responses that relate to neurodevelopmental needs from the quality-of-life questions completed by the parent or primary care-giver. These domains focused on twelve neurodevelopmental items, including reports of speech delay, cognitive and learning impairments, and were collected as part of the review of participant medical and developmental history. Reports of study participants attending individual or group therapy, being under-the-care of a psychiatrist or psychologist, and four assessments of school-related activities, (full- or part-time special education, utilization of class-room services, and being on an individualized educational plan (IEP)) were also collected. An IEP refers to the special education, services and additional accommodations that are made available to a child with disability that allow them to attain educational goals and academic milestones. Lastly, pediatric quality-of-life data were captured from all study participants. Here we report the physical, social, emotional, and school sub-scores, along with the total score, from the Pediatric Quality of Life Inventory, version 4 (PedsQoL)20. The Pediatric Quality of Life (PedsQL) Inventory is a proxy-report measure designed to measure core health dimensions in children from 5 to 17 years old. The measure consists of 23 items in four scales: physical functioning, emotional functioning, social functioning, and school functioning.

Statistical Analysis

Descriptions of the study cohort are provided either as frequencies (N and %) or as summary statistics (N, mean, standard deviation (SD), median, minimum, and maximum) as appropriate for each domain type. Each domain was first compared independently among the three dystrophin dp140 mutation categories. These statistical comparisons were performed using one-way analysis of variance or Kruskal-Wallis tests for continuous outcomes, and Fisher’s exact tests for categorical outcomes. Those neurodevelopmental or cognitive domains showing evidence of a significant relationship with predicted dystrophin dp140 isoform expression were further analyzed using logistic regression. In each logistic regression model, neurodevelopmental or cognitive domain (dichotomous yes/no response) was the dependent variable, and dystrophin dp140 category was the predictive variable. Dystrophin dp140(-) category was considered the reference group, and an additional test comparing the dystrophin dp140(+) and dystrophin dp140 intermediate group was also performed.

In order to assure that neither age nor glucocorticoid use were confounding factors in our data analysis, we compared outcomes between glucocorticoid use (yes/no at baseline visit), and between 1-year age intervals.

For the statistical comparison of responses obtained for the twelve neurodevelopment-related outcomes, we excluded those that answered “Don’t know” or “Not applicable” due to the ambiguity of the response. The rationale to exclude these responses was that we were not able to meaningfully interpret the response, especially for the school-related resources. The “not applicable” response in this setting could be due to the child not attending school, or lack of neurodevelopmental needs, and therefore not needing school services.

All statistical analyses were performed on the data collected at the first study enrollment visit. The significance level for all statistical tests was set at 0.05. Data was analyzed using Stata V15 software (College Station, TX).


Study characteristics and demographic information

A total of 204 participants from North America, Europe, Israel, Australia, and India were included in this young cohort of the DNHS. To determine equity in access to health care, we note that most of the study participants were from developed countries (n=176), whereas 28 participants were from India, a developing country. The mean age of the study participants was 6.4 years (range, 4.0 – 8.9 years). The majority of the study participants identified their race as Caucasian (72.6%), with 17.7% identifying as Asian, 3.5% as Other, 1.0% as Pacific Islander, and 0.5% as Black (0.5%). A minority of the participants (4.9%) reported their race as unknown. Ethnicity was reported as non-Hispanic by 93% of participants and as Hispanic by 7.8% participants. These demographic characteristics, along with anthropometric measures, are summarized in Table 1.

Table 1

Summarizes the demographic and anthropometric details of the study participants.

Characteristic Total cohort DP140 Negative DP140 Intermediate DP140 Positive P-value comparing DP140 groups
N (%) N (%) N (%) N (%)
Mean ± SD Mean ± SD Mean ± SD Mean ± SD
Median (min, max) Median (min, max) Median (min, max) Median (min, max)
Age (years) 204 52 48 53 0.99
6.4 ± 1.4 6.5 ± 1.4 6.5 ± 1.3 6.5 ± 1.4
6.4 (4.0, 8.9) 6.6 (4.0, 8.9) 6.4 (4.3, 8.9) 6.2 (4.1, 8.9)
Weight (kg) 204 52 48 53 0.67
22.0 ± 5.7 22.3 ± 5.9 21.4 ± 6.2 22.4 ± 5.7
20.7 (13.4, 48.2) 21.0 (13.4, 37.1) 20.0 (15.0, 48.2) 21.8 (13.5, 41.0)
Calculated height (cm) 203* 51 48 53 0.85
116.6 ± 9.2 116.6 ± 9.2 117.0 ± 8.9 116.0 ± 9.6
115.7 (95.4, 139.9) 117,8 (96.9, 135.4) 116.3 (99.9, 139.9) 114.9 (95.4, 134.3)
Race 0.25
Caucasian 148 (72.6%) 39 (75.0%) 31 (64.5%) 39 (73.6%)
Black 1 (0.5%) 0 (0.0%) 1 (2.1%) 0 (0.0%)
Pacific Isl. 2 (1.0%) 0 (0.0%) 0 (0.0%) 2 (3.8%)
Asian 36 (17.7%) 11 (21.2%) 13 (27.1%) 6 (11.3%)
Other 7 (3.4%) 1 (1.9%) 1 (2.1%) 4 (7.6%)
Unknown 10 (4.9%) 1 (1.9%) 2 (4.2%) 2 (3.8%)
Ethnicity 0.22
Non-Hispanic 188 (93.2%) 47 (90.4%) 45 (94.8%) 52 (98.1%)
Hispanic 16 (7.8%) 5 (9.6%) 3 (6.3%) 1 (1.9%)

*Height was not reported in one participant during the first year of study.

Study subjects and their treatment with oral glucocorticoids

Sixty-one percent of participants (n=124) were on oral glucocorticoids at the time of their baseline study visit (Table 2). Of these 124 participants, 79 participants (39%) were either on prednisone or prednisolone, and 44 participants (22%) were on deflazacort. One participant reported glucocorticoid use but did not indicate the name of the medication being prescribed. The duration of cumulative use of glucocorticoid ranged from no previous use to 4.5 years.

Table 2

Summarizes the glucocorticoid use in study participants.

Total cohort DP140 Negative DP140 Intermediate DP140 Positive P-value comparing DP140 groups
N (%) N (%) N (%) N (%)
Mean ± SD Mean ± SD Mean ± SD Mean ± SD
Median (min, max) Median (min, max) Median (min, max) Median (min, max)
Is the study participant taking steroids at baseline 0.39
Yes 124 (60.8%) 29 (55.8%) 33 (68.8%) 34 (64.2%)
No 80 (39.2%) 23 (44.2%) 15 (31.3%) 19 (35.9%)
Total lifetime steroid use at baseline (days) 204 52 48 53 0.34
288 ± 382 291 ± 390 241 ± 340 355 ± 423
95 (0, 1665) 53 (0, 1182) 95 (0, 1665) 208 (0, 1579)
Type of steroid taken* 0.92
None 80 (39.4%) 23 (44.3%) 15 (31.9%) 19 (35.9%)
Prednisone 48 (23.7%) 12 (23.1%) 13 (27.7%) 15 (28.3%)
Deflazacort 44 (21.7%) 11 (21.2%) 11 (23.4%) 11 (20.8%)
Prednisolone 31 (15.3%) 6 (11.5%) 8 (17.0%) 8 (15.1%)

Care-giver reported neurodevelopmental needs

The response completion of neurodevelopmental needs identified by the care-giver was variable depending on the question, ranging from 168 to 174 respondents (82 – 85%) of the 204 participants at the first study visit. The following neurodevelopmental needs were most commonly reported: speech delay (33%), mild developmental delay (24%), significant behavioral problems (16.5%), language impairment (14.5%), and learning disability (14%). Other neurodevelopmental needs reported by the parent or primary care-giver included attention-deficit hyperactivity disorder (5%) and autism spectrum disorder (3%). Summaries of responses from the entire cohort are shown in Supplementary Table 1.

With regards to therapy services, most respondents had not received any (Supplemental Table 2). With regards to utilization of school-based resources at the first study visit, of the 169 respondents, an equal number of participants were on an IEP (n= 65) as not on an IEP (n=65) (Supplemental Table 3). Thirty-six participants (22%) reported that an IEP was not applicable in their child’s case, though thirteen of these participants (36%) were under the age of 5 years. The commencement of school is age 5 years in developed countries. These thirty-six participants were excluded from the subsequent analyses since no further information was available from them. Sixty-three percent (n=106) reported receiving no services in the school compared to 27% (n=46) receiving classroom services (Supplemental Table 3). Of those not receiving classroom services, 10% (n=17) responded that they were not applicable and 1% (n=2) did not know whether their child was receiving any services. Fourteen of the “no responses” and 10 of the “not applicable” responses were in participants under the age of 5 years.

When participants were asked whether they had been evaluated by a mental health provider (psychiatrist or psychologist), most of them participants had not been (Supplemental Tables 4, 5).

When study participants were stratified based on 1-year age intervals at study enrollment, parent or primary care-giver of participants identified between boys between 7 to 8 years of age as having more behavioral, neurodevelopmental and cognitive needs (significant behavioral problems, speech delay, language impairment, learning disability) (Table 3). Similarly, IEP and utilization of services in the classroom were more often reported in participants 7 years to 8 years of age (Table 4). Of the 169 respondents, 141 (83%) report that they had never been evaluated by a psychologist.

Table 3

Summarizes neurodevelopmental challenges based on age groups.

Have you ever been diagnosed with- Age group No Yes Don’t know P-value*
Significant behavioral problems 4 to <5 25 2 1 0.59
5 to <6 36 6 2
6 to <7 27 6 0
7 to <8 29 8 1
8 to <9 25 6 0
Depression 4 to <5 27 0 1 0.74
5 to <6 44 0 0
6 to <7 33 0 0
7 to <8 36 1 1
8 to <9 31 0 0
Autism 4 to <5 27 1 0 0.21
5 to <6 42 0 2
6 to <7 32 1 0
7 to <8 35 3 0
8 to <9 30 0 0
Speech delay 4 to <5 16 12 0 0.15
5 to <6 28 13 3
6 to <7 24 9 0
7 to <8 21 17 0
8 to <9 25 6 0
Language impairment 4 to <5 23 4 1 0.20
5 to <6 37 4 3
6 to <7 26 4 3
7 to <8 28 10 0
8 to <9 28 2 1
Learning disability 4 to <5 26 2 0 0.018
5 to <6 38 3 3
6 to <7 27 3 3
7 to <8 25 12 1
8 to <9 28 3 0
Sensory integration disorder 4 to <5 27 1 0 0.51
5 to <6 40 1 3
6 to <7 32 0 1
7 to <8 33 3 2
8 to <9 30 1 0
Cognitive impairment 4 to <5 26 2 0 0.31
5 to <6 36 4 1
6 to <7 31 0 2
7 to <8 32 4 2
8 to <9 29 1 1
ADHD 4 to <5 27 0 1 0.28
5 to <6 42 1 1
6 to <7 31 1 1
7 to <8 32 4 2
8 to <9 27 2 2
Mental retardation 4 to <5 27 1 0 0.59
5 to <6 42 2 0
6 to <7 33 0 0
7 to <8 37 0 1
8 to <9 30 1 0
Mild developmental delay 4 to <5 17 10 1 0.24
5 to <6 32 10 1
6 to <7 29 4 0
7 to <8 27 10 0
8 to <9 23 6 2
Pervasive developmental disability 4 to <5 27 0 1 0.14
5 to <6 42 0 1
6 to <7 32 0 0
7 to <8 34 2 2
8 to <9 28 0 2

*P-value calculated using total who answered yes or no (excludes those indicating they don’t know)

Table 4.

Summarizes the use of services based on age groups.

Outcome Outcome response – No Outcome response – Yes Age group OR P-value 95% CI
Classroom services 14 3 4 to <5 1.00
26 11 5 to <6 1.97 0.35 0.47 – 8.27
26 7 6 to <7 1.26 0.77 0.28 – 5.63
15 20 7 to <8 6.22 0.011 1.51 – 25.62
25 5 8 to <9 0.93 0.93 0.19 – 4.50
Individual therapy 27 1 4 to <5 1.00
42 0 5 to <6
31 2 6 to <7 1.74 0.66 0.15 – 20.29
33 4 7 to <8 3.27 0.30 0.35 – 31.04
30 1 8 to <9 0.90 0.94 0.05 – 15.10
Group therapy 27 1 4 to <5 1.00
41 1 5 to <6 0.66 0.77 0.04 – 10.98
32 1 6 to <7 0.84 0.91 0.05 – 14.13
35 3 7 to <8 2.31 0.48 0.23 – 23.51
30 1 8 to <9 0.90 0.94 0.05 – 15.10
Seeing a psychologist 25 3 4 to <5 1.00
40 2 5 to <6 0.42 0.36 0.06 – 2.67
25 7 6 to <7 2.33 0.27 0.54 – 10.06
29 8 7 to <8 2.30 0.25 0.55 – 9.61
22 8 8 to <9 3.03 0.14 0.71 – 12.85

Glucocorticoid use and reported neurodevelopmental needs

Of the 204 participants, 124 participants (60.8%) were on oral glucocorticoid. When cognitive domains were compared between glucocorticoid users and glucocorticoid non-users, there was no relationship between glucocorticoid use and any of the cognitive domains (Table 5). The utilization of classroom services or mental health professionals was also not significantly different between these glucocorticoid users and glucocorticoid non-users (data not shown). In addition, no significant difference in the total quality-of-life score or sub-scores was detected between these two groups (data not shown).

Table 5.

Summarizes the relationship between neurodevelopmental and behavioral challenges and glucocorticoid use.

Have you ever been diagnosed with- Cumulative lifetime steroid use No Yes Don’t know P-value*
Significant behavioral problems <6 months 80 11 4 0.15
≥ 6 months 62 17 0
Depression <6 months 94 0 1 0.45
≥ 6 months 77 1 1
Autism <6 months 89 3 2 0.99
≥ 6 months 77 2 0
Speech delay <6 months 66 27 2 0.20
≥ 6 months 48 30 1
Language impairment <6 months 79 11 5 0.39
≥ 6 months 63 13 3
Learning disability <6 months 78 10 7 0.38
≥ 6 months 66 13 0
Sensory integration disorder <6 months 89 2 4 0.41
≥ 6 months 73 4 2
Cognitive impairment <6 months 83 7 3 0.76
≥ 6 months 71 4 3
ADHD <6 months 88 3 4 0.47
≥ 6 months 71 5 3
Mental retardation <6 months 92 3 0 0.63
≥ 6 months 77 1 1
Mild developmental delay <6 months 68 22 4 0.86
≥ 6 months 60 18 0
Pervasive developmental disability <6 months 89 0 4 0.21
≥ 6 months 74 2 3

*P-value calculated using total who answered yes or no (excludes those indicating they don’t know)

Neurodevelopmental needs based on DMD mutation location

We compared the reported cognitive domains affected in the three subcategories of DMD study participants (dystrophin dp140(+), dystrophin dp140(-), and dystrophin dp140 intermediate). Several statistical associations were noted between the three subcategories of study participants, with differences between subcategories in domains of cognitive and language impairments, learning disability, and utilization of classroom services.

For those cognitive domains that showed a statistically significant relationship with DMD mutations affecting the dystrophin dp140 isoform (exon 45-50), further analysis using logistic regression showed that study participants in the dystrophin dp140 intermediate category were more alike those in the dystrophin dp140(+) category (Table 6). Most of the statistical differences detected were between the dystrophin dp140 intermediate and dystrophin dp140(-) categories.

Table 6.

Summarizes the cognitive outcomes based on the three DMD genotype subgroups.

Cognitive outcome Mutation Response – No Response – Yes OR p-value 95% CI P-value comparing Dp140Intermediate to Dp140Negative
Behavior problems Negative 34 10 1.00 0.21
Intermediate 37 3 0.28 0.07 0.07–1.09
Positive 40 8 0.68 0.47 0.24–1.91
Language impairment Negative 31 11 1.00 0.59
Intermediate 38 3 0.22 0.031 0.06–0.87
Positive 42 5 0.34 0.06 0.11–1.06
Learning disability Negative 33 10 1.00 0.26
Intermediate 38 1 0.09 0.023 0.01–0.71
Positive 42 4 0.31 0.07 0.09–1.09
Classroom services Negative 17 19 1.00 0.49
Intermediate 32 6 0.17 0.001 0.06–0.50
Positive 36 10 0.25 0.004 0.10–0.65
Cognitive impairment Negative 36 7 1.00
Intermediate 39 0
Positive 44 2 0.24 0.08 0.05–1.20

Dystrophin dp140 intermediate participants were significantly less likely to report affected cognitive domains than dystrophin dp140(-) participants. In the domains of cognitive impairment and behavioral problems, there was no statistical significance between the three groups. Participants in the dystrophin dp140 intermediate group were significantly less likely to have a diagnosis of significant language impairment (OR=0.22, 95% CI 0.06 – 0.87) and learning disability (OR=0.09; 95% CI 0.01 – 0.71). Further, a statistically significant relationship between the utilization of classroom services and DMD mutation subcategory, with both the dystrophin dp140(+) and dystrophin dp140 intermediate categories being less likely to utilize classroom services than those who were dystrophin dp140(-) (OR=0.17, 95%CI 0.06-0.5; OR=0.25, 95%CI 0.10-0.65, respectively). There were no statistically significant differences between the dystrophin dp140(+) and dystrophin dp140 intermediate categories in any of the cognitive domains. There were also no statistically significant differences in the utilization of mental health professionals or quality-of-life assessments between the three subcategories (data not shown).

Quality-of-life assessment

Quality-of-life was assessed using the PedQoL questionnaire and a total of 141 (69%) care-givers completed the questionnaire. One questionnaire was completed per participant. Mean (±SD) sub-scores for physical, emotional, social, and school domains were all below 50 (41.9 ± 21.2, 25.4 ± 18.3, 34.2 ± 10.5, and 29.2 ± 20.0, respectively). The mean total score for all 141 participants was 33.9 ± 16.0 and ranged from 4.4 to 70.7 (Supplemental Table 6). None of the participant in the study scored the maximum score of 100.


In this report, we present data as assessed by the DNHS to support the prevalence of neurodevelopmental needs in young boys with DMD. To our knowledge, this is the first prospectively collected data of neurodevelopmental needs from a large well-characterized, internationally representative cohort of young boys with DMD. We also stratified boys with DMD mutations into three sub-categories based on DMD mutations likely to affect the dystrophin dp140 isoform to characterize in greater detail the neurodevelopmental needs in this cohort. By contrast, both Felisari et al.16 and Bardoni et al.17 evaluated cognitive impairment based on assessment of intellectual capacity through intelligence quotient, while our data is based on parent- or primary care-giver report.

Speech delay was the most commonly reported neurodevelopmental need in our cohort. Speech delay was reported by 33% of our respondents, and is consistent with reports of higher prevalence of speech delay in DMD7,10. Speech delay is a predictor of later education and classroom services utilization. Soim et al. describe that receiving speech therapy is statistically associated with an increased prevalence of grade repetition10. Thus, speech delay in DMD has implications in the developmental trajectory of affected boys, and suggests that these children may need additional resources to achieve optimal educational outcomes. Such additional resources include earlier psycho-educational assessment in boys with DMD identified with speech delay. Recognition of early learning differences would potentially facilitate establishing appropriate school support and prevent grade repetition. Not only would it improve the quality of school experience for these children, it would also help improve long-term educational outcomes. Our collective experience as professionals engaged in the care of DMD families, psycho-educational testing often times, does not occur until age 7 or 8 years, thus delaying the care-givers and educators awareness of the unique neurodevelopmental needs in young boys with DMD. This lack of awareness may be the reason that parent-reported concerns were particularly more frequent in boys between the ages of 7 to less than 8 years in our cohort.

The reported prevalence of attention-deficit and autism that is described in our cohort is different from those reported from earlier studies. For example, Ricotti et al. evaluating boys with DMD of ages 5 to 17 years noted that the prevalence of autism spectrum was 21%21. They used both a neurodevelopmental questionnaire and a structured diagnostic interview. Pane et al. found that nearly 32% of boys with DMD fulfilled criteria for attention deficit-hyperactivity using DSM-IV criteria. The mean age of their cohort was 12.6 years4. Thus, the differences reported in our cohort is likely methodological in nature. Furthermore, the mean age of our cohort was 6.4 years, and full manifestations of attentional challenges and hyperactivity may not be fully manifest.

To our knowledge, there is limited data on the long-term educational and employment impact of cognitive needs in DMD. We note that parent-reported concerns were particularly frequent in boys between ages 7 to less than 8 years (Tables 4 and 5). Most of the study participants in our data are from developed countries, and age at school entry is approximately 5 years. We reason that as they transition from pre-school to primary school, neurodevelopmental needs may become increasingly apparent. This may be in addition possibly to the lack of awareness of the unique neurodevelopmental needs in this population by care-givers and educators.

School and emotional sub-scores scored among the lowest average sub-scores with PedsQoL, demonstrating that these two sub-domains may be most affected by disease burden in this cohort (Supplemental Table 6). While the exact reasons for these findings are unclear, our data suggests that neurodevelopmental and behavioral needs may influence school-related experiences in young children. Further detailed analyses in this area are warranted to understand the impact of neurodevelopmental needs in educational outcomes.

The use of glucocorticoids has been shown to be associated with behavioral and mood changes22. In this young cohort of boys with DMD, we found that approximately 60% of them were on oral glucocorticoids at the time of entry to study. Consistent with other reports23,24, we did not find a statistically significant association between reported neurodevelopmental concerns and glucocorticoid use. Population-based longitudinal data does support that evaluated longitudinally, glucocorticoid use is associated with behavioral difficulties in older boys with DMD5. The cross-sectional nature of our data analysis and the younger age of our cohort may explain this difference with existent data.

Clinical and genetic heterogeneity in DMD has been well-described25,26,27 and it has previously been noted that cognitive outcomes may predict motor, cardiac and respiratory outcomes in DMD28. Boys with DMD who had poor cognition, also had worse cardiac outcome, with two-thirds of these boys having cardiac involvement before age 12. This data suggests that overall health maintenance in DMD may be dictated in part by cognitive health, or a genetic modifier, or both.

Our study has some limitations. First, the DNHS was not designed to assess in detail the neurodevelopmental needs of this cohort. Therefore, there was no collection of objective psychometric data during the study. Existent data does suggest that neurodevelopmental needs of DMD are present even during infancy29. Chieffo et al. report that young boys with DMD showed clear developmental concerns when followed prospectively30.

The second limitation of our study is that some of the recorded responses were ambiguous, and some parent- or care-giver reported concerns have overlapping features. For example, there was lack of distinction between “cognitive and language impairment” on the questionnaire. Neurodevelopmental needs may exist on a continuum and may overlap; for example, there may be co-existent speech delay and language impairment. Particularly in DMD, two or more neurodevelopmental disorders can co-exist, highlighting the overlap in developmental pathways affected. Despite these limitations, our data highlights the clear neurodevelopmental burden in this medically vulnerable population. Our data suggests that future studies in DMD should consider the use of standardized reports such as patient-reported outcome measures or the muscular dystrophy quality-of-life questionnaire. Further, specific recommendations in future studies in DMD should consider standardized assessment of cognitive ability.

Taylor et al. reported that the cumulative loss of multiple brain-specific dystrophin-isoforms may increase the burden of neurocognitive deficits in DMD31. The existence of multiple brain-specific dystrophin isoform suggests a redundancy in some of its functions in the developing brain. We chose to particularly focus the analysis in DMD region exon 45-50 based on the current clinical and research priority to understand better the heterogeneity within this sub-group based on the available exon 51 skipping therapy, and recent published report on the structural and functional compromise to the developing brain due to DMD mutations that affect dystrophin dp140 isoform18.

Our data highlights a critical, yet often unmet medical need in this population. It is a widely held notion that cognitive impairments are irreversible in neurodevelopmental disorders. This notion is being challenged by the emergence of new scientific data that modulation of receptors can result in improvement of cognitive function32. Our data provides a focal point on how future studies in DMD should aim to address neurodevelopmental needs more comprehensively through collection of objective psychometric data longitudinally. This approach will provide important scientific data to enable development of pragmatic interventions to improve educational and workplace integration for individuals with DMD, many of whom now have the capacity to attain independency as adults in society.

Supporting Information

S1 Table

Description of cognitive outcomes at the baseline visit.

Have you ever been diagnosed with… No* Yes Don’t Know Total answered
Significant behavioral problems 142 (83.5%) 28 (16.5%) 4 174
Depression 171 (99.4%) 1 (0.6%) 2 174
Autism 166 (97.1%) 5 (2.9%) 2 173
Speech delay 114 (66.8%) 57 (33.3%) 3 174
Language impairment 142 (85.5%) 24 (14.5%) 8 174
Learning disability 144 (86.2%) 23 (13.8%) 7 174
Sensory integration disorder 162 (96.4%) 6 (3.6%) 6 174
Cognitive impairment 154 (93.3%) 11 (6.7%) 6 171
ADHD 159 (95.2%) 8 (4.8%) 7 174
Mental retardation 169 (97.7%) 4 (2.3%) 1 174
Mild developmental delay 128 (76.2%) 40 (23.8%) 4 172
Pervasive developmental disability 163 (98.8%) 2 (1.2%) 6 171

*Percentages calculated using total who answered yes or no (excludes those indicating they don’t know).

S2 Table

Description of therapy outcomes at the baseline visit.

Have you ever had… No In the past Occasionally Daily or regularly Total answered
Individual therapy 163 (95.3%) 3 (1.8%) 2 (1.2%) 3 (1.8%) 171
Group therapy 165 (95.9%) 4 (2.3%) 1 (0.6%) 2 (1.2%) 172

S3 Table

Description of school related outcomes at the baseline visit.

Have you ever had… No* Yes Not applicable Don’t know Total answered
IEP or 504 plan 65 (39.2%) 65 (39.2%) 36 (21.7%) 3 169
Services in the classroom 106 (62.7%) 46 (27.2%) 17 (10.1%) 2 171
Part-time special education 119 (71.3%) 27 (16.2%) 21 (12.6%) 4 171
Full-time special education 135 (81.8%) 8 (4.9%) 22 (13.3%) 5 170

*Percentages calculated using total who answered yes, no, or not applicable (excludes those indicating they don’t know)

S4 Table

Description of psychiatry use at the baseline visit.

Have you ever seen a… No Yes Total answered
Psychiatrist 158 (94.1%) 10 (6.0%) 168

S5 Table

Description of psychology use at the baseline visit.

Have you ever seen a… Never Weekly Monthly Quarterly As needed Total answered
Psychologist 141 (83.4%) 2 (1.2%) 2 (1.2%) 4 (2.4%) 20 (11.8%) 169

S6 Table

Description of PedsQL assessments at the baseline visit.

PedsQOL score N Mean ± SD Median (min, max)
Physical sub-score 141 41.9 ± 21.2 40.6 (0, 100)
Emotional sub-score 141 25.4 ± 18.3 25 (0, 75)
Social sub-score 141 34.3 ± 10.5 35 (0, 100)
School sub-score 133 29.2 ± 20.0 25 (0, 80)
Total score 141 33.9 ± 16.0 32.6 (4.4, 70.7)

*Percentages calculated using total who answered yes or no (excludes those indicating they don’t know)


DNHS was funded through grants from the U.S. Department of Education/NIDRR (H133B031118 and H133B090001), U.S. Department of Defense (W81XWH-09-1-0592), the National Institutes of Health (UL1RR031988, U54HD053177, UL1RR024992, U54RR026139, G12RR003051, 1R01AR061875, and RO1AR062380), and Parent Project Muscular Dystrophy.

Data analysis for this manuscript was supported by Award Number UL1TR001876 from the NIH National Center for Advancing Translational Sciences to C.S. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the National Center for Advancing Translational Sciences or the National Institutes of Health.

Mathula Thangarajh was supported the American Brain Foundation/American Academy of Neurology Clinical Research Fellowship (2015 – 2017), and also gratefully acknowledges the support by National Institutes of Health grant (R25NS088248-02; Principal Investigator Dr. Meurer).

The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. The authors take full responsibility for the contents of the manuscript. The views presented in this report do not represent the views of the Department of Defense or the U.S. Government.

Competing Interests

Mathula Thangarajh has provided consultant services to PTC Therapeutics. Christopher Spurney has no conflict of interest to disclose. Heather Gordish-Dressman has served as a consultant for AGADA BioSciences and Solid GT, and is a co-founder and part owner of TRiNDS. Paula R Clemens serves on data safety monitoring boards for Pfizer and NIH and has served as a consultant to NS Pharma and Sanofi/Genzyme. She currently receives grant support from NS Pharma, Sanofi/Genzyme, Amicus, MDA, Department of Defense, and NIH. She is board member for TRiNDS. Eric P Hoffman has served on advisory committees for AVI BioPharma, as a consultant with AGADA BioSciences, the Gerson Lehrman Group, Medacorp, and Lazard Capital, is cofounder, board member, and shareholder of ReveraGen, is a co-founder and part owner of TRiNDS, and has a patent pending related to the treatment of Duchenne muscular dystrophy. Craig McDonald has served as a consultant for clinical trials for PTC Therapeutics, BioMarin, Sarepta, Eli Lilly, Pfizer, Halo Therapeutics, Santhera Pharmaceuticals, Cardero Therapeutics, Catabasis, Marathon, and Mitokyne, outside the submitted work; serves on external advisory boards related to Duchenne muscular dystrophy for PTC Therapeutics, Eli Lilly, Sarepta, Mitokyne, and Marathon; and reports grants from the US Department of Education/National Institute on Disability and Rehabilitation Research (NIDRR), National Institute on Disability, Independent Living, and Rehabilitation Research (NIDILRR), US National Institutes of Health (NIH)/National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS), US Department of Defense, and US Parent Project Muscular Dystrophy, during the conduct of the study. Erik Hendricson has served as a consultant for Santhera Pharmaceuticals, Cardero Therapeutics, Bristol-Myers Squibb, Genzyme, and PTC Therapeutics, and has served as an external advisory board member for Parent Project Muscular Dystrophy and as an executive committee member of the Cooperative International Neuromuscular Research Group. RTA has served as a consultant for Santhera Pharmaceuticals. There are no patents, products in development or marketed products to declare. The competing interests do not alter the authors’ adherence to all the PLOS policies on sharing data and materials.

Data Availability

Data presented in this manuscript was collected as part of the DMD Natural History Study (DNHS). The CINRG executive committee has restrictions on who accesses the data, as there are many datasets within this study. Data requests can be made to Lauren Morgenroth ‎[].

Corresponding Author

Mathula Thangarajh, MD, PhD

Department of Neurology

Children’s National Health Systems

Washington, D.C. 20010

Phone: 202-476-2771

Fax: 202-476-2864


A Review of Mathematical Models for Muscular Dystrophy: A Systems Biology Approach Fri, 16 Feb 2018 14:35:04 +0000 Muscular dystrophy (MD) describes generalized progressive muscular weakness due to the wasting of muscle fibers. The progression of the disease is affected by known immunological and mechanical factors, and possibly other unknown mechanisms. These dynamics have begun to be elucidated in the last two decades. This article reviews mathematical models of MD and models that could be used to study molecular and cellular components implicated in MD progression. A biological background for these processes is also presented. Molecular effectors that contribute to MD include mitochondrial bioenergetics and genetic factors; both drive cellular metabolism, communication and signaling. These molecular events leave cells vulnerable to mechanical stress which can activate an immunological cascade that weakens cells and surrounding tissues. This review article lays the foundation for a systems biology approach to study MD progression. 



Muscular dystrophy (MD) describes generalized progressive muscular weakness due to the wasting of muscle fibers. While this is an umbrella term used to describe a wide range of muscle wasting diseases, the two types that have been most extensively mathematically modeled are Duchenne’s (DMD) and Becker’s (BMD) 1,2, with DMD being the most common childhood form of MD in males. Children affected by this X-linked recessive disorder have an average life expectancy in the mid-twenties, typically becoming fully wheelchair dependent by their teens3. Most forms of MD leave striated muscle cells with reduced contractile abilities, leading to a wide range of phenotypic expression in patients from fatigue to drooping eyelids.

Previous research attributes pathogenesis of DMD and BMD to either absent or partial forms of the dystrophin protein4. From a cellular perspective, the basic contractile unit of the muscle is the sarcomere. The dystrophin protein is located between the sarcolemma, the outer membrane of the sarcomere, and outer layer of myofilaments, providing a scaffold for muscular contraction. Weakness typically begins in extremity muscles, propagating in a proximal–distal direction, until ultimately affecting the diaphragmatic muscles responsible for breathing3,5. Hypertrophy of cardiac muscle cells is an additional complication associated with both DMD and BMD. The loss of function associated with both typically leads to premature death6.

Epidemiological impacts of MD remain difficult to pinpoint definitively. Each subset of MD contains its own range and pattern of pathogenesis and progression. We can further classify subsets of MD in (ordered in decreasing rates of prevalence): dystrophinopathies, laminopathies, dystroglycanopathies, sarcoglycanopathies, and alternative congenital forms7,8. In the United States, dystrophinopathies have a prevalence of about 15 cases per 100,000 people9. The following list describes the rates of prevalence of MD per 100,000 cases in Northern England8:

  • Myotonic muscular dystrophy is the most common adult onset MD. The prevalence rate is about 10.6.
  • With respect to dystrophinopathies that manifest as dystrophin absence or deficiency, DMD and BMD have been most extensively studied. Incidence of dystrophinopathies is about 8.5.
  • Facioscapulohumeral muscular dystrophy (FSHD) has a prevalence rate of about 4.
  • Laminopathies, dystroglycanopathies,and sarcoglycanopathies are typically classified as Limb-Girdle MD (LGMD). The prevalence rate is about 2.3.
  • Collagen VI deficiencies that take form in Ullrich Congenital MD (UCMD, also known as Ullrich Scleroatonic MD) have a prevalence rate around 0.13.
  • Most alternative congenital forms have rates below one.

The goal of this paper is to offer a survey of mathematical models relating to MD, as well as identifying potential opportunities for quantitative research that have yet to be explored within the context of MD. In the context of this paper, mathematical models refer to quantitative approaches that start from mechanistic principles and result in simplified representations of the phenomena that is studied. The modeling formalisms include algebraic expressions, ordinary and partial differential equations, discrete dynamical systems, agent-based models, etc. Mathematical methods differ from statistical methods in that the latter are applied posteriori to broad families of data representations, usually to test hypothesis of significance whereas mathematical models can be used a priori. Clearly, we exclude from this paper statistical analysis of experimental data. However, we include in the scope of the present work strictly computational methods such as network reconstruction.

Each section describes either progression or pathogenesis of several forms of MD. Fig 1 illustrates how the sections relate in some forms of MD. At their roots, most forms of MD have genetic causes; the section Genetic Models presents models of gene-gene interactions intended to discover pathways either specific to a given form of MD or shared between multiple forms of MD. Vulnerabilities caused by genetic mutations result in an increased probability of muscle damage due to normal stress-strain of muscle usage; the section Muscle Models considers simulations of stress-strain during muscle contraction. Muscle contraction requires ATP originated in the mitochondria of which dysfunctions have been implicated in MDs; the section Mitochondrial Models discusses models of cell death initiated by mitochondria as a result of muscle cell damage. Both muscle damage and cell death initiate an acute immune response; this response attempts to regenerate the damaged or dead muscle tissue. Apoptotic signals from immune cells during muscle repair further antagonizes the mitochondrial response. In a patient with some forms of MD, the mitochondrial and immunological responses cause extended periods of inflammation; chronic immune responses have been implicated in the deterioration of muscle fibers in MD patients. The section Immunological Models introduces models of inflammation — both acute and chronic — and tissue repair.


Fig. 1: Conceptual linkage between the sections of this article.

Most types of MD are genetic disorders. Vulnerabilities due to genetic mutation result in increased stress-strain on muscles during contraction. Mitochondrial and immunological responses continue for extended periods; this causes a chronic immunological respone that perpetuates MD progression.

Genetic Models

Most types of MD are believed to have genetic origin4. The genes implicated in MD have functions in numerous biological processes, thus obscuring a univocal characterization of MD pathogenesis. Although each form of MD is caused by different genes, MDs often share similar phenotypes. Finding genetic pathways shared by different MDs could allow for interventions at the downstream node. Furthermore, studying these pathways allow for a greater understanding of how these genes normally interact.

The genetic causes of observed phenomena can be inferred computationally through the reconstruction of gene regulatory networks (GRNs). The characterization of a GRN usually starts with experimental measures of many genes at at least two conditions or points in time. Gene regulatory networks (GRNs) can provide insight regarding subtleties in disease development; graph theoretic models like Boolean Networks10, Bayesian graphs11, and Petri Net12 have emerged as important tools for understanding and reconstructing GRNs. To develop models of GRNs, researchers may employ database like OMIM13 or computer programs like Snoopy14 that use experimental genetic expression data to reconstruct such networks. GRNs can then be studied to find nodes that have more parent genes, are up-regulated or down-regulated by parent genes, affect more downstream genes, or express more often by different but related diseases like MDs. For time dependent data or complex processes, though, directed graphs may be impossible to construct. Instead continuous models like dynamic Bayesian networks15 and temporal Bayesian classifiers16 can be constructed. Although continuous models are unable to answer qualitative questions about the GRN, continuous models can estimate quantitative values. Lack of data hampers continuous models ability to explain large complex GRNs.

Published Genetic Models About MD

Relatively few GRNs have been constructed for MD; among the few successful studies are Grunwald et al. (2008)12 and Tucker et al. (2006)16.

Grunwald et al. (2008)12 created a Petri net of the downstream gene affected by dystrophin (DMD and BMD). The model incorporated sixty-four places and eighty-eight transitions. The authors used Muaritius maps to analysis this model to elucidate the important nodes through knockout experiments. The knockout of the RAP2B-calcineurin pathway caused a 78% disturbance of the whole model. This — the authors concluded — made that pathway important for future study in DMD/BMD therapeutics.

Tucker et al. (2006)16 constructed a continuous model of a GRN involving DMD, LGMD2C, and LGMD2E by using temporal Bayesian classifiers. The model was trained using mouse models (mdx, bsg, gsg) of each form of MD. the model predicts that genes Dlk1, Dusp13, and Casq2 play a part in all three MDs and should be studied further as a possible influence of MD pathogenesis. The role of Dlk1 and Casq2 as shared pathways in MD has been independently verified by statistical studies of gene expressions by Vinciotti et al. (2006)17.

Modeling Other Implicated Genetic Pathways

The task of elucidating gene regulatory networks is a vast area of research (see e.g. DREAM challenge 200818); given the correct data most of these model schemes could be used to study MD. A modern review on this subject can be found at Chai et al. (2014)19. Fig. 2 illustrates the extracellular and intracellular domains of a muscle cell, relating proteins embedded within and associated with the nucleus, the dystrophin-glycoprotein complex (DGC) and the extracellular matrix (ECM). The nuclear outer membrane (OM) is an integral part of the rough ER (Fig 2). Destabilized lamina release proteins that diffuse along the OM to the ER. To aid in future modeling, Fig. 3 includes the known single nucleotide polymorphisms (SNPs) that have been strongly implicated in MD and the genes affected. Publicly available databases can be used to investigate each SNP’s related genes, proteins, and pathways (e.g. SNPedia20, OMIM13, and 1000 Genome Project21).

Fig 2

Fig. 2: Location of Genes Implicated in MD

A diagram of cell structure proteins implicated in MD pathogenesis. The corresponding SNPs for each type of MD shown here are given in Fig. 3. The proteins affected by many types of MD have yet to be elucidated.

2016-12-31 (3)

Fig. 3: SNPs Implicated in MD

SNPs Implicated in different types of MD. SNP IDs correspond to

Muscular Models

Although vulnerabilities caused by mutated genes, poorly regulated immune responses, and damaged mitochondria perpetuate MD, muscle wasting is the main phenotypical effect of MD5. Damage to muscle fibers caused by contractions exposes muscle tissue to the effects of chronic immune activation and mitochondrial dysfunction. Fibrosis due to chronic immune activation reduces vulnerable muscle cell’s contractual force and flexibility. Understanding these interactions by using mathematical models could offer a macroscopic view of MD progression.

One muscle cell contains thousands of sarcomeres interlaced and primed for contraction. A single sarcomere is made up of parallel thin and thick filaments called actin and myosin; these entwined filaments pull on each other during contraction22,23 . MD leaves muscles vulnerable to unhealthy contraction which exacerbates damage in a cycle that proves catastrophic to muscle cells perpetuating MD progression24,25.

Fig 3

Fig. 4: Sarcomere Regional Classification.

Muscle contraction occurs when calcium ions allow crossbridge sites on myosin to bind with actin and slide actin into the H zone. This action called a powerstoke is followed by ATP binding to the crossbridge allowing the crossbridge to release the actin; Rebinding and powerstroke may than occur again until calcium ions depart the muscle22,23.

Published Muscular Models About MD

Fibrotic tissue accumulation caused by chronic inflammation, contractures, and vulnerabilities in cell structure due to genetic mutation, complicate the process of modeling muscle tissue in MD. Of those attempted, Virgilio et al. (2015)26 and Zöllner (2015)27 created models describing the macro-scale complexities of fibrotic tissue and contractures. Lammerding et al. (2004)28 considers cell deformation in the cytoskeleton due to lamina-deficiency.

Model of fibrosis: Virgilio et al. (2015)26 created a simulation to study the effects of various pathologies including MD. The simulation used an agent based model to create fascicle geometry with the addition of fibrosis and fatty tissue infiltration followed by the use of the micromechanical model described by Sharafi and Blemker (2010)29. This model, using only in silico methods, agreed with the results that fibrosis aggravates the symptoms of DMD.

Model of contractures: MD patients can experience higher-degree of ankle plantarflexion due to contractures; “toe-walking” can make plantar flexor muscles such as the gastrocnemius, soleus and peroneus therapeutic targets. Although the model concerns with long term use of high heels, Zöllner’s model (2015)27 could be used to describe toe-walking in MD. Due to removal of sarcomeres, fascicles shorten with long term toe walking causing less than optimal motion range and increased muscle stress. This increased stress coupled with DCG vulnerabilities accelerates fibrosis and adipose replacement.

Model of cytoskeleton: Nuclear lamina-deficiencies are linked with cytoskeletal defects in muscle cells30,31. Among the mechanisms for these deficiencies are: (i) Destabilized DGC due to cytoskeletal compromise within the plasma membrane leads to the aggregation of mechanical stress in contracting muscle cells28. (ii) Histology in lamina-deficient diseases is commonly concomitant with maligned and internalized nuclei in muscle fibers30,31. (iii) Chronically strained A/C deficient lamins result in cells experiencing higher rates of apoptosis28,32. Intriguingly, regenerating muscle fibers also share high levels of internalized nuclei- the distinction between pathology and regeneration is unclear presently3132.

Lammerding et al. (2004)28 modeled nuclear misalignment and cytoskeletal stress in lamin A/C-deficient mouse embryo fibroblasts. A/C lamins form compartments for splicing factors as well as RNA polymerase II transcription. Inhibition of A/C laminis suppresses RNA polymerase II-dependent transcription in mammalian cells, while its as a scaffold in nuclear compartments remain unknown. A sinusoidal force with amplitude 0.6 nanonewtons (nN) and with frequency 1 Hz, offset 0.6 nN, was applied through a magnetic trap. Cylindrical coordinates (r,θ) were used to measure bead displacement with the magnetic bead at the origin and θ=0 for the force direction. The equation represents the induced strain field described by the analytic cell mechanics model proposed by Bausch et al. (1998)33 ur is the radial component of the induced bead displacement as a function of the applied force, F, cell stiffness μ* the characteristic cut of radius κ-1, the distance from the magnetic bead center r, and the polar angle θ. Thus,

Fig. 5: Lammerding et al. Model

is the radial component of the induced bead displacement as a function of the applied force, F, cell stiffness the characteristic cut of radius , the distance from the magnetic bead center r, and the polar angle . Assuming the cytoskeleton to be incompressible, we can set .

K0 and K1 are modified Bessel functions of respective order 0 and 1 with:

Fig. 6: Bessel Function of Order 1

Letting for incompressibility and metallic bead contact radius of , parameters and were computed by fitting Fig. 6 to the bead displacement.

In this model, the cytoskeleton was exposed to the same biaxial strain as the membrane; for each cell type, nuclear deformation increased approximately linearly with applied membrane strain. Both lamin A/C deficient cells result in decreased nuclear stiffness and altered nuclear mechanics. Under resting conditions, they found that the integrity of the nuclear envelope was maintained in A/C deficient cells. However, under pressure, the control cells maintained nuclei integrity far more than lamin-deficient cells, though both could be ruptured. Increased vulnerability of A/C deficient cells to mechanical stress comparatively was also concluded, with an increase in both necrotic and apoptic cell fraction28. Necrosis mediated through nuclear rupture is not wholly attributed to vulnerability to mechanical stimulation in this model; only about 3-5% of A/C deficient nuclei ruptured.

Modeling Other Implicated Muscular Pathways

Future muscle models into MD could follow several pathways including crossbridge modeling and whole tissue stress-strain simulation. Models of crossbridges combine the physiology of muscle contraction with ATP production in mitochondria. Stress-strain simulation allows for quantification of damage to tissue that activates the immune response.

Cross-bridge models: The crossbridge cycle allows for the ratcheting of muscles during contraction. To understand crossbridges, we must describe the regions of the sarcomere. The A band is made up of thick myosin filaments; throughout contraction its length remains constant centrally localized to the H zone. Actin filaments are laced with the protein tropomyosin. At rest, tropomyosin covers the hinges that catch on actin. Alternatively, the I band is composed of thin actin filaments that alter its length between myosin filament pairs. Structures called Z discs fetter actin filaments at their opposite ends. In S1 regions, myosin edges are hinged and highly flexible, catching on actin and releasing in an indefinite binding cycle (Fig. 4). Upon their release, myosin filaments perform a power stroke catalyzed by the hydrolysis of ATP. ATP is the means for crossbrige formation; it is the release of phosphate during ATP hydrolysis that contracts the S1 region22,23. Calcium provides the means for binding whereas ATP drives contraction. Upon the release of calcium, the protein troponin pushes the tropomyosin to expose binding sites for actin. Upon exposure and a threshold level of ATP expression, the contraction cycle begins34.

Smith et al. (2013)35 described this procedure using a directed graph. The model uses twenty-five components in four domains to describe (mostly protein) interaction during a muscle contraction. The model is a micro scale representation of muscle contraction since it only incorporates one crossbridge. This model could be used to show how lack of ATP due to mitochondrial defects in MD could effect muscle contraction.

Huxley (1957)36 proposed an early mathematical description of this power stroke process. Distance from binding site to crossbridge is taken as the independent variable while function is the probability of a crossbridge being bound at position at time . This yields a conservation law of:

Fig. 7: Crossbridge Conservation Law

where is the velocity of actin to myosin, is the binding rate of crossbridges at position , and is the unbinding rate of crossbridges at position .

The rate of energy release, , by ATP is:

Fig. 8: Rate of Energy Release by ATP

where is the total number of crossbridges at , and is the energy released by a single crossbridge.

This model captures the key physiological properties of crossbridges but fails to illuminate the biochemical interactions; Pate (1989)37 proposed a model based on the Huxley model rectifying the biochemical interaction issue.

The weakening of cardiac muscles due to fibrotic/adipose tissue replacement and metabolic dysfunction is known as cardiomyopathy. This weakening can cause fatigue, swelling in extremities, arrhythmias, chest pain, and possibly heart failure. Cardiomyopathy is the end result of most moderate to severe forms of MD38,39. Understanding how MD results in cardiomyopathy could save lives.

Tewari et al. (2016)40 expanded on Huxley’s concepts with a model of cardiac muscle that connects metabolic and mechanistic crossbridge actions. Using five state variables, the model allows for experimenting with lack of ATP and Ca2+ in cardiac muscle contraction. Additionally, passive muscle action is modeled using spring dynamics. The model was able to fit several experimental data sets of cardiac contraction under different ATP and inorganic phosphate concentrations. The model does not account for three dimensional dynamics nor fibrotic/adipose tissue influences. An expansion of this model to include calcium has also been created41.

Stress-strain models: Levels of healthy isometric force production can be maintained under the strain of intense activity. Rest commenced with the buffering of calcium levels in the cytosol reduces mechanical stress to a point; inappropriate high or low levels of activity leave muscles vulnerable to damage. Furthermore, both types of activity are associated with increased levels of cytosolic calcium additionally linked with fiber damage and apoptosis42.

A.V. Hill (1938) 43 created a Force-Velocity relationship model to understand isotonic and isometric muscle contractions; his work provides a basis for multibody, dynamic musculoskeletal modeling and simulation.

Fig. 9: Hill Model

relates rate of muscle contraction (shortening length), v, to the load p where and are constants given by experimental data and is the isometric force.

Although Jewell and Wilkie (1958)44 demonstrated that this model lost accuracy when exposed to sudden changes to muscle length, Hill’s model remains pivotal and is integrated in many other models.

Van der Linden et al. (1998)45,46 attempted whole tissue interaction simulations examining aponeurosis/muscle under stress, limited by computational power, they were restricted to 2D and simplified 3D models. Johansson et al. (2000)47 improved upon van Leeuwen-Kier’s (1997)48 model of squid tentacles by separately modeling active muscle attributes with Hill’s force-velocity model and describing passive elements as a hyperelastic material. Unfortunately since they used parameter values based on van Leeuwen-Kier’s research and ANSYS, Johansson et al. failed to significantly improve upon the predictions made by van Leeuwen-Kier. Yucesoy et al. (2002)49 introduced a similar model expanding on Van der Linden’s work. Like Johansson et al., Yucesoy et al. separated muscle into the extracellular matrix (passive) and myofiber (active); from that they created a linked fiber-matrix mesh that fuses a passive element and an active element. This “two domain” approach allowed a glimpse into the interaction of muscle fibers and the extracellular matrix. In an effort to understand the effects of geometries on muscle tissue, Sharafi and Blemker (2010)29 devised a model where actual rabbit muscle biopsies were estimated with linear functions. They captured both the geometries of fibers (microscopic level) as well as the fascicles (macroscopic). Since they were modeling muscle stress, Sharafi-Blemker only modeled the passive elements of muscle which were described as a hyperelastic, nearly incompressible material. In opposition to assumptions made by earlier modelists, Sharafi-Blemker discovered that fascicle display anisotropic characteristics.

Mitochondrial Models

Mitochondria are organelles inside a typical cell. Mitochondria produce ATP, which is the currency for sarcomere contraction. In a single cell, there can be hundreds of mitochondria; these organelles are responsible for far more than ATP synthesis. Mitochondria also regulate cell life-death cycle with implications to the immune system. Mitochondria synthesize protein encoded in mitochondrial DNA (mtDNA) in addition to dividing independently as needed inside the cell. mtDNA is vulnerable to mutations that perpetuate the mitochondrial cascade, with rates up to ten times higher than nuclear DNA mutation50,51.

Poorly regulated mitochondria have been implicated in several forms of MD. The distinct role of dystrophin in DMD and BMD’s pathogenesis and progression remains unknown; despite devastating consequences in its absence, the dystrophin protein comprises a mere 0.002% of healthy skeletal muscle proteins52 . Mitochondrial Myopathies are a group of disorders closely related to — and in some articles/websites included as a form of — MD. An accumulation of defective mtDNA could be the cause of these myopathies53.

Published Mitochondrial Models About MD

Mitochondria’s shape, number, and energy processing power are constantly in flux; mitochondrial dynamics alternate between fusion or fission. This fluctuation allows for the dispersal of mtDNA mutation. Tam et al. (2013)54 and Taylor et a. (2003)55 created stochastic models emulating these dynamics.

Tam et al. (2013)54 produced a stochastic model examining rates of fusion-fission that would optimize mitochondrial function and minimize clonal expansion in neighboring mitochondria. This model classifies fusion-fission events in mitochondria using the following criterion:

  • Fusion events feature nucleoid exchange between mitochondria; one is emptied out to the other and marked for degradation.
  • Fission sites appear close to the fusion event, regionally containing original nucleoid distributions from precursor mitochondria.
  • Healthy fission features low levels of exchange of nucleoids, with only mitochondrial matrix contents being mixed.
  • Larger and longer mitochondria have higher propensities for fission, and smaller ones are more likely to fuse.

The model predicts that protective nuclear retrograde signaling could rescue the mitochondrial cascade through the promotion of mitochondrial nucleoid replication propensity up to sixteen times the basal rate, increasing stochasticity by neutralizing clonal mutant aggregation. Benefits of nuclear retrograde signaling are limited by rates of fusion-fission. Within this simulation, rate of mitochondrial fusion-fission plays a significant role in clonal expansion. Slow exchanges of mtDNA result in homoplasmy, where interventionist retrograde signaling could compound the issue by increasing rate of nucleoid replication. Higher rates of fusion-fission result in a heteroplasmic steady state; increasing levels of mitochondria in cells mix nucleoids faster. These patterns persist regardless of mitochondrial presence in the cell or their replication parameters. A cytoskeletal, cellular model that considers mitochondrial movement independent of fusion-fission, as well as mitochondrial morphology in a differentiated cell context, is needed to further conclude potential therapeutic benefits of retrograde signaling54.

Table 1: Propensity Equations for Tam et al. (2013)

This table lists the equations used in Tam et al. (2013)54 to update the number of mtDNA in the mitochondria where rmax+1 – is the maximum copy number for amplification of mtDNA, Mmito– is the number of mitochondria in a cell at a given time kD – the rate of autophagy, and VF,max – is the maximum propensity of fission.

Propensity Propensity equations
Upregulated propensity for nucleoid replication due to higher ratio of mutant mtDNA
Propensity for mitochondrial autophagy
Propensity for mitochondrial fission

Taylor et al. (2003)55 also created a probabilistic model. Using this model, the authors were able to estimate mtDNA mutation rate as approximately 5·10-5 mutations per genome per day. This rate guarenteed that clonal expansion in stem cells after 80 years matched known literature. The model provides a simpler example simulating similar results to Tam et al.

Modeling Other Implicated Mitochondrial Pathways

Future research into the effects of mitochondrial dysfunction could follow several paths including models of apoptosis and ATP productions. As discussed above, a newer model studying the effect of accumulations of deleterious mtDNA could be considered for both Mitochondrial Myopathy and DMD. As chronic inflammation has been implicated in several forms of MD, models that simulate apoptosis and necrosis could help explain how acute inflammation turns chronic. Understanding ATP production would also be useful since all muscle contractions require ATP.

Apoptosis models: Byproducts of mitochondrial metabolism include small amounts of electrons that leak from inner membrane complexes and attach themselves to oxygen, forming free radicals called reactive oxidative species (ROS). In a healthy immune response, free radicals such as superoxide and nitric oxide are produced by macrophages for destruction of foreign species. Small amounts of the free radical superoxide produced by the mitochondrion are neutralized by antioxidant enzymes such as superoxide dismutase. Mutated mtDNA leak more ROS in a degenerative mitochondrial cascade essentially poisoning vulnerable cells through ROS release50,51.

Table 2: Equations for Mitochondrial Models.

These equations model the behavior of proteins during apoptosis. Fusseneger et al. (2000)56 assumes isotropic conditions, and Huber et al. (2010)57 allows for converting isotropic conditions into anisotropic conditions. All and are kinetic constants.

Fusseneger et al. Equations56 Variables
– Concentration of FAS-FAS ligand complex.
– Concentration of FAS receptor.
– Concentration of FADD protein.
– Concentration of FAS-FASL-FADD complex.
– Concentration of cytosolic cytochrome c.
– Concentration of Apaf-1 protein.
– Concentration of Apaf-1-cytochrome c complex.
– Concentration of procaspase-8.
– Concentration of procaspase-9.
– Concentration of active caspase-8.
– Concentration of active caspase-9.
– Concentration of executioner procaspase.
– Concentration of active executioner caspase.
– Concentration of Bcl-2.
– Concentration of Bcl-X_L.
– Concentration of FLIPs.
– Concentration of ARC.
Huber et al. Equations57 Variables
concentration of a given protein (n=1,2,..,23)
– Chemical reactions given by usual Mass/Kinetic action.

Intrinsic apoptosis results from a stressed cellular response. Severe ROS damage can result in cellular necrosis whereas the release of cytochrome c to the cytosol from the inner membrane of the mitochondria triggers intrinsic apoptosis. Depolarization in MOMP (mitochondrial outer membrane permeabilization) in stressed cells triggers the MOMP cascade. MOMP stress markers in survival-apoptotic dynamics feature members of the Bcl-2 protein family and BH-3 only proteins56,58. When triggered by MOMP, mitochondria release cytochrome c and Smac (second mitochondrial-derived activator of caspases) into the cell’s cytosol. By binding with XIAPs (x-linked inhibitors of apoptosis), Smac allows for cytochrome c, Apaf-1, and ATP to combine into apoptosome and cleave procapase-9 forming the initiator caspase-9; cleavage of procaspase-3 by caspase-9 then forms the executioner caspase-3 which in turn activates executioner caspase-6,7 and creates a positive feedback loop by cleaving more caspase-9. These executioners finalize the death of the cell. Apoptosis will not occur if threshold levels of effector capases are not reached56,57,58.

Extrinsic apoptosis occurs when an extracellular self-destruct order is given. Stressed cells signal macrophages to engage with apoptotic cells through phagocytosis as a protective mechanism. Extracellular death ligands act as messagers of these orders by binding with FAS (CD95) death receptors. Subsequently, FAS receptors cluster allowing the binding with FADD (FAS-associated death domain). Recruited by FADD, multiple procapase-8 compile and mutually cleave forming capase-8. This new protein cleaves pro-apoptotic, BH3-only protein Bid forming tBid (Truncated Bid). These interactions lead up to MOMP activation and subsequent cascade; the MOMP activation pathway bridges extrinsic and intrinsic apoptosis. Multidomain proteins are activated by apoptotic tBid activation, which can be inhibited due to protective Bcl-2 proteins56,58.

Fussenegger et al. (2000)56 proposed a model simulating apoptosis to study caspase activation and inhibition (See Table 2). The model confirms experimental observations that Bcl-2 above a critical level effectively inhibits procaspase-9 activation but fails to adequately inhibit procaspase-8 activation, and suppression of FADD’s binding to FAS/FASL complex blocks caspase-8 activation but has little effect on caspase-9 activation. The model assumes isotropic reactions with a well mixed single domain and omits proteins including Bid/tBid, reactions like caspase-8 cleaving of Bid, and bundles executioner caspases 3,6,7 into a single variable. Furthermore, intrinsic and extrinsic apoptosis were not distinguished. Since Fussenegger et al., several models have been proffered to redress omissions.

Albeck et al. (2008)59,60 introduced a model concentrating on the extrinsic apoptosis death switch as well as MOMP interactions. The model represents both cytosol and mitochondria as two separate domains interacting after MOMP with parameters trained by live-cell imaging of HeLa cells. Similar to Fussenegger et al., Albeck et al. bundles many proteins with similar properties — such as caspase-8 and -10 are represented as a single variable C8 — to simplify the model and all reactions are isotropic. However, unlike the previous model, Albeck et al. incorporates a time delay mechanism to compensate for the delay of death ligand reception to MOMP as oppose to the quick death of the cell post-MOMP. Intriguingly, western blot fails to show enough XIAP pre-MOMP to properly inhibit caspase-3. Albeck et al. concluded that another protein/reaction must exist to account for this discrepancy. The model did confirm that MOMP occurs after proapoptotic Bcl-2 proteins reach a certain level depended on the physiological state of the cell. An alternate stable state — partial cell death — is predicted by the model.

To eliminate the isotropic assumption, Huber et al. (2010)57 combined previous models with one-dimensional diffusion PDEs. The typical mass reactions and kinetics are extended by a PDE (Table 2) where vn(x,t) is the chemical reactions. Although their goal was to investigate anisotropic reactions in MOMP, the reaction-diffusion equations remain applicable to wider studies.

Metabolism model: Smith et al. (2013)35 created a directed graph model for ATP production. The model describes the interactions of fifty-nine genes, signals, and transcription factors in 4 domains. Indication of activation or inhibition is included in each interaction although no time scale is given. This network lays the groundwork for future modeling using ODEs and PDEs.

Immunological Models

Immunology describes the coordination of innate and adaptive immune systems to redress threats of the pathogenic, viral, and parasitic kind in addition to the filtration and disposal of excessively damaged cells. Unfortunately, while lifesaving and essential for protection, immune system responses can include self-induced attacks in areas of the body that become so damaged they leave immune cells unable to recognize their healthy peers1,61.

The innate immune system has been implicated in perpetuating multiple forms of MD62,63. Vulnerabilities from missing proteins that stabilize the DCG promote a switch from acute inflammatory response to chronic response; the results of extended inflammation are adipose tissue replacement and muscle fibrosis1,64. This has been confirmed with microscopic analysis of damaged muscle cells lacking dystrophin. Cellular analysis further indicates abnormal, elevated, levels of macrophages, helper and cytotoxic T cells in damaged tissues61.

Published Immunological Models About MD

Cyclic immunological activation purported by mechanical stress is associated with coexisting restorative and degenerative processes in muscle cells; immunological players act both as starters and finishers of apoptosis. CD8+ (cytotoxic T-cells) initiate apoptosis in compromised cells, and the cellular remains of apoptosis are eaten by macrophages61,65. Two models, Dell’Acqua and Castiglione (2009)1 and Jarrah et al. (2014)2, have been proposed using a predator-prey system to study chronic immune activation by employing a log-normal distribution:

Fig. 10: Lognormal Equation

This equation is used to simulate damage to muscle tissue in the following models where is the time to reach peak damage, is the standard deviation of damage, and the magnitude of peak damage at time .

to analyze initial (acute) muscle damage. Both models utilize concentrations of the previously listed immune response helpers in addition to concentrations of healthy, damaged, and regenerating muscle cells.

Dell’Acqua and Castiglione (2009)1 is generated by five ODEs in addition to a conservation law (Table 3) describing the immune response of DMD in the mice model. They used COPASI’s66 optimization methods on experimental mdx mice data from Spencer et al. (2001)61 and Wehling et al. (2001)62 to find the best fit parameters. The set of equations used is

Table 3: Dell’Acqua-Castiglione’s Equations

where , , and are kinetic parameters. is a lognormal distribution describing damage to normal tissue. The final equation provides a conservation of tissue.

Equations Variables
M = concentration of macrophages
H = concentration of CD4+
C = concentration of CD8+
N = percentage of normal muscle fibers
D = percentage of damaged muscle fibers
R = percentage of regenerating muscle fibers

Jarrah et al. (2014)2 (Table 4) refines Dell’Acqua-Castiglione’s model by adding an additional ODE which allows the conservation law to be implicit. The parameters of this model were derived from recent experimental data of mdx mice. Both models assume that the missing dystrophin in the muscle causes damaged muscle cells to initiate the immune response which contributes to their own damage until eventual apoptosis. The set of equations used is

Table 4: Jarrah et al. Equations

where , , and are kinetic parameters. is a lognormal distribution describing damage to normal tissue. This system is similar to the above except for the lack of an explicit conservation of tissue equation. However, the system still has an implicit conservation of tissue with .

Equations Variables
M = concentration of macrophages
H = concentration of CD4+
C = concentration of CD8+
N = percentage of normal muscle fibers
D = percentage of damaged muscle fibers
R = percentage of regenerating muscle fibers

and the definition of variables is the same as Table 3. Initial conditions have levels of cytotoxic T cell levels at 0; this changes when the impulse damage represented by equation Fig. 10 sets the system into motion. When h=0, the impulse damage is negated and the system remains in a stable state; allowing h>0, the model ensures that T helper cells draw cytotoxic T cells to the damaged region. Damage caused by the immune system reaches a peak in weeks four through eight until the presence of the players wanes. By week twelve, the decreased presence of macrophages, CD4+ and CD8+ T cells (week fourteen) results in diminished levels of degeneration and restoration.

Both models display regions of bistability. Depending on the initial damaged caused and M0, the system collapses to healthy muscle stability or approaches a stability with heterogeneous mixtures of healthy and damaged muscle. This suggests that immune response to muscle damage could be a major contributor to DMD’s pathophysiology1,2 which has been shown experimentally61,62.

These models indicate that the strength of the immune response and maintenance of the positive feedback system relies upon moving past these threshold points to enter another stability state. Driving the system into these recovery regions could prove to be a potential therapeutic target for redressing the role of the inflammatory response.

Modeling Other Implicated Immunological Pathways

The above models incorporate only cellular interactions in an chronic immune response; any multi-scale model would need to include molecular signals that are used in an acute and chronic immune responses to muscle damage. Veltman et al. (2017)67 and López-Caamal et al. (2012)68 introduced models describing molecular signals.

Inflammasome Model: Inflammasome has been implicated in initiating inflammation after an injury or infection occurs by stimulating caspase-1 and activating IL-1β. Both LGMD2B and DMD have shown upregulation of inflammasome69.

A recently published model, Veltman et al. (2017)67, describes inflammasome interactions by using an ODE model to study molecular signaling input to IL-1β activation output. The model also includes inhibitors of inflammasome — IL-10 (signaled by IFNβ) and IFNγ. The model illustrates that negative feedback by IFNγ has less effect on IL-1β than the inhibition by IL-10/IFNβ. This result matches experimental research. The model does not account for anisotropic dispersion nor cellular interactions.

AKT/mTOR Model: Another important avenue for research would be propagation of immune cell signals. López-Caamal et al. (2012)68 created a model describing AKT/mTOR pathway activation by IGF-1. IGF-1 is an important protein released by M2 macrophages during muscle repair that promotes muscle growth, and AKT/mTOR has been implicated as the main signal pathway to promoting muscle growth. The model simulates a positive feedback and diffusion of AKT/mTOR signals once activated. The model also is used to speculate on consequences of signals from other sources like AMPK or pharmaceuticals.

Areas of Future Quantitative Research

Characterizing pathology and pathogenesis of MD requires further study; future models could accelerate therapeutic discovery by testing potential pathways in silico as well as detecting new therapeutic pathways. There are areas for computational research that have not yet been explored mathematically. These pathways may be worthwhile exploring to better understand MD disease development and providing treatments that may delay onset or progression.

The creation of rAAV/AAV1 (Recombinant Adeno-associated virus, Adeno-associated virus 1) delivery system and CRISPR/Cas system indicates the possibility of a genetic cure for MD70,71. Heller et al. (2015)72 overexpressed the human α7 integrin Gene, ITGA7, using the AAV1 delivery system in mdx mice. ITGA7 is a skeletal muscle laminin receptor (Fig. 1) whose overexpression does not cause an immune response in mdx mice. Protective benefits of the DGC were restored with ITGA7 overexpression; lifespans were also prolonged. Xu et al. (2015)73 used CRIPSR/Cas9 to remove mutated exon 23 with the dystrophin genomic region to restore dystrophin expression in the DGC of mdx mice. Clinical trials are currently taking place for LGMD2D74. Mendell et al. (2012)75 wrote a review outlining future work in gene therapy. Both systems require a relatively small number of injection sites. Population dispersion models and GRNs could help in the development and effective administration of both rAAV/AAV1 and CRISPR/Cas systems. Population dispersion and diffusion models could be used to predict the outcome from a series of injections to the spread of the corrected genes throughout the body. Potentially, these models could indicate the most effective injection sites. Beyond the correction of defective genes, both systems could target genes that promote muscle growth and regeneration76. GRNs may help in developing new therapies by finding pathways that both system could target and thereby accelerate the body’s muscle regeneration. These types of treatments could be applicable to both MD patients and in sports medicine.

Myostatin — a TGF-β protein — has long been recognized as a possible therapeutic option for MD. Early murine testing for several MD phenotypes produced mixed results. For DMD (mdx mice) and LGMD2F (scgd{-/-} mouse), Parsons et al. (2006)77 concluded positive results for mice treated early in development before widespread necrosis occurred. Treating sgcg{-/-} mice (modeling LGMD2C) resulted in positive muscle physiology including increase fiber size, muscle mass, and grip force in addition to reduce frequency of apoptosis; however, muscle histology remained unfazed signifying lack of pathology change78. A possible solution proposed by Rodino-Klapac (2009)76 used AAV1 to genetically edit Follistatin (the major myostatin inhibitor). A few experiments show that the new treatment has few side effects and shows similar improvements as other myostatin inhibitors in LGMD2A (Calpainopathy). Future research is needed to show if myostatin inhibition is a means to maintain pathophysiology in MD patients. GRNs and physiological muscle models could be used to understand effective usage of myostatin. GRNs could discover new inhibitors and enzyme activators of myostatin; this may allow targeted genetic editing to regulate myostatin similar to Rodino-Klapac. GRNs might also explain why these methods will work with some types of MD but fail with other types. Physiological models may demonstrate the increase in fiber size, muscle mass, and grip force due to myostatin therapies.

A plethora of models could be used to describe cellular, molecular, and pharmaceutical interactions of the immune system. Precise molecular, mathematical models bridging arginine metabolism with oscillations in macrophage phenotypic expression could be used to model nitric oxide mediated cytotoxicity as well as fibrosis during satellite cell proliferation. Mechanical models for macrophage infiltration and molecular models for macrophage phenotype oscillation could also be useful to better characterize chronic immune system activation due to structural defects. Integration into musculoskeletal simulation may be useful to model the immunological role in MD pathogenesis and progression. Agent based models could be used to imitate immune cells.

MD disease progression also results in alterations to pathophysiology such as gait and muscle atrophy. Noninvasive studies with patients, especially children, could be critical to create a staged model for gait devolution and morphology. Quantifying degrees of eccentric contraction using musculoskeletal simulation could possibly explain selective degeneration in DMD79.


With new developments in computational power and data availability, a growing amount of research is using a systems biology approach to understand pathogenesis and progression of disease. Effective and integrated in vitro and in silico models could inform biological phenomena, even without the need of a living subject. For instance, over the last few decades, collagen hydrogel with muscle derived cells (CHMDCs) have promised to revolutionize in vitro experiments and tissue engineering. For CHMDCs to reach the envisioned use, verification by use of mathematical simulations are needed. Recently while examining shape and design, Hodgson (2015)80 used a combination finite elements and agent based analysis to illustrate the lines of principle strain and cell migration in CHMDCs confirming earlier in vitro work by Smith et al. (2012)81 . As MD is a rare disorder, the use of mathematical models could help elucidate the underlining mechanisms of the disease that might not be easily detectable given the limited subject pool.

Although genetic studies have implicated genes as the cause of many types of MD, relatively little is know about about common pathways between these genes that may affect pathogenesis and create similar phenotypes; this necessitates the use of mathematical models describing GRNs (Gene Section). Common genes like Dik1, Dusp13, and Casq21623and there downstream pathways provide future prospects for therapeutic intervention.

Mitochondrial and immunological mechanisms further MD progression. Mathematical models of apoptosis and mitochondrial fission/fusion help to understand intracellular processes that directly affect cellular vitality and death. Tam et al.54 mitochondrial fission/fusion model related mitochondrial health with nuclear mechanisms to rescue mutated mtDNA; unhealthy mitochondria can trigger intrinsic apoptosis with the release of cytochrome c. Fusseneger et al.56 and Albeck et al.59,60 furthered the study of apoptotic mechanics by creating models to simulate the processes; both models agree with already published results. Extrinsic apoptosis can be signaled by immune cells. Immunological mathematical models display a larger view of cellular interactions that bridge the gap of molecular actions and tissue level muscle damage. Damage caused by stress and weakened cellular structure is repaired and debris removed by extrinsic apoptosis and phagocytosis.

Mathematical models of muscles under contractile stress are essential to understanding the long term development of most types of MD. Partial differential equations and agent based models of anisotropic strain from contraction display where cellular rupture and immune response will likely occur. Physical therapeutic and pharmaceutical interventions can be targeted to such areas of high stress to stymie MD progression. Expanding cellular models of the immune response and combining with molecular signals could create a more comprehensive view of muscle tissue regeneration and damage caused by chronic inflammation. Future research could incorporate multiple levels of models into a unified simulation to give a whole view of the progression of MD.

Future research that sheds light on MD disease dynamics, which is likely to occur through mathematical modeling, will provide the means to engage and perhaps ultimately bypass biological systems coordinating together to exacerbate degeneration.


The authors received no specific funding for this work.

Competing Interests

None of the authors have a financial conflict of interest. M.H. was differential diagnosed (and genetically confirmed recently) with Limb-Girdle Muscular Dystrophy in 2004. A.C. was diagnosed with Progressive Mitochondrial Myopathy in 2010 through a muscle biopsy and differential diagnosis. She worked for the Foundation for Mitochondrial Medicine since 2013 as an intern and now is a volunteer as needed.

Data Availability

This review article does not include any data. All information underlying this study is within the paper.

Corresponding Author

Juan B. Gutierrez,

Author Contribution

J.G. conceived of the paper and advised A.C. and M.T. on the organization of the manuscript. A.C. conceived of the mitochondrial subsection and created the sarcomere regional classification graphics. M.T. conceived of the muscle subsection and heavily revised the manuscript after reviewer suggestions. Both A.C. and M.T. contributed to the genetic, mitochondrial, immune, muscle, areas of future quantitative research and discussion sections. All authors reviewed the manuscript.

]]> 0
Epigenetic Regulators Modulate Muscle Damage in Duchenne Muscular Dystrophy Model Thu, 21 Dec 2017 13:43:11 +0000 Histone acetyl transferases (HATs) and histone deacetylases (HDAC) control transcription during myogenesis. HDACs promote chromatin condensation, inhibiting gene transcription in muscle progenitor cells until myoblast differentiation is triggered and HDACs are released. HATs, namely CBP/p300, activate myogenic regulatory and elongation factors promoting myogenesis. HDAC inhibitors are known to improve regeneration in dystrophic muscles through follistatin upregulation. However, the potential of directly modulating HATs remains unexplored. We tested this possibility in a well-known zebrafish model of Duchenne muscular dystrophy. Interestingly, CBP/p300 transcripts were found downregulated in the absence of Dystrophin. While investigating CBP rescuing potential we observed that dystrophin-null embryos overexpressing CBP actually never show significant muscle damage, even before a first regeneration cycle could occur. We found that the pan-HDAC inhibitor trichostatin A (TSA) also prevents early muscle damage, however the single HAT CBP is as efficient even in low doses. The HAT domain of CBP is required for its full rescuing ability. Importantly, both CBP and TSA prevent early muscle damage without restoring endogenous CBP/p300 neither increasing follistatin transcripts. This suggests a new mechanism of action of epigenetic regulators protecting dystrophin-null muscle fibres from detaching, independent from the known improvement of regeneration upon damage of HDACs inhibitors. This study builds supporting evidence that epigenetic modulators may play a role in determining the severity of muscle dystrophy, controlling the ability to resist muscle damage. Determining the mode of action leading to muscle protection can potentially lead to new treatment options for muscular dystrophies in the future.



There is no cure to date for muscular dystrophies caused by absent or malfunctioning Dystrophin, be it lethal Duchenne or milder Becker type. Current therapies aim at alleviating the symptoms and delaying the progression of a disease that can be life threatening. One promising pharmacological treatment is to inhibit histone deacetylases (HDACs) 1. HDACs can regulate gene transcription in muscle progenitor cells, by controlling the activity of myogenic regulatory factors and MEF2 family proteins 2,3. HDACs promote chromatin condensation, inhibiting gene transcription until myoblast differentiation is triggered and HDACs are released. Similarly, blocking HDACs leads to hyper acetylated chromatin, inhibiting condensation and therefore facilitating transcription. HDAC inhibitors ameliorate the dystrophic phenotype by promoting myogenesis and improving regeneration in dystrophic muscles 1,4,5. Several studies show that follistatin upregulation by HDAC inhibitors is responsible for boosting regeneration in dystrophic muscles 6,7 . While studies have been focusing on blocking HDACs to promote hyper acetylated chromatin and transcription, the potential of directly modulating histone acetyl transferases (HATs) on a muscle dystrophy context remains unexplored. CBP (CREB Binding Protein) is a nuclear transcriptional co-activator with HAT activity, belonging to the p300/CBP family 8. It is ubiquitously expressed and acetylates both histone and non-histone proteins to regulate transcription. CBP was shown to functionally activate MyoD by acetylation and to directly interact with MEF2C 5. CBP expression in zebrafish muscle was reported recently 9. We used the zebrafish model, extensively characterised and widely used for studying muscular dystrophy 10, to explore the effects of overexpressing CBP in a Dystrophin-null background. We have found that overexpressing this single HAT rescues the dystrophic phenotype as efficiently as blocking HDACs. Moreover, we observed that both treatments inhibit the early appearance of dystrophic fibres, prior to any effect on regeneration could take place. We report a follistatin-independent mechanism protecting against fibre damage.


To test whether Dystrophin absence affects HAT’s expression we performed quantitative real-time PCR (qPCR) for p300 and/or CBP (crebbp) transcripts (Figure 1A). Two-way ANOVA shows highly significant differences [F(1,24)=31.5, p<0.0001] between HAT genes expression in controls and embryos injected with well described Dystrophin morpholino cocktail (also see Materials and methods, S1 Fig A). Multiple comparisons indicate strong downregulation of crebbpb and p300b and milder but significant decrease of p300a expression (p values: crebbpb = 0.0010, p300a = 0.0247, p300b = 0.0045). Therefore, the absence of Dystrophin significantly affects HAT expression.

Taking these results, we set to determine whether the dystrophic phenotype could be rescued by overexpressing RFP-tagged murine CBP in dystrophin-null embryos (dmdta222a/ta222a). Nuclear and dose dependent expression of the transgene was confirmed (S1 Fig B,C). To test the hypothesis that CBP overexpression improves regeneration, embryos were analysed at 2 dpf, as fibre tips start detaching from somitic borders 10, and 6 dpf, when regeneration upon damage is underway 11,12. While dystrophin-null embryos show characteristic dystrophic muscles from early stages, overexpressing CBP clearly reduced damage not only at later stages but, unexpectedly, also at 2 dpf (Figure 1B). Fast swimming was also restored, a characteristic failure of dystrophic embryos. Absence of damage as early as 2 dpf cannot be explained by stimulating regeneration, which takes several days 11,12, suggesting that CBP overexpression may act at an earlier phase preventing muscle degeneration in the first place. We focused on investigating this observation further and testing whether only CBP overexpression or also HDAC inhibition could have a protective effect. Different concentrations of CBP were tested and rescue at 2 dpf compared with TSA treatment, previously used in zebrafish and considered the most efficient pan-HDAC inhibitor 1,13 (Figure 1C-E; S1 Fig D). On average, 83% of dystrophin-null (dmdta222a/ta222a) embryos showed clear dystrophic muscles at 2 dpf (Figure 1C, dmd-/-). The prevalence of a dystrophic signature was drastically reduced to 28% upon treatment with TSA and 27% when a low dose of CBP (+CBPlow) was used. A 10-fold increase of the CBP dose (+CBPhigh) improves the rescue significance (p<0.0001). Therefore, CBP overexpression is at least as efficient as HDACs inhibition in protecting dystrophin-null muscles from damage. Similar results were obtained in the Dystrophin morphant background for both HDACs inhibitor and CBP (Figure 1D,E). Immunostaining for RFP-CBP shows that rescue is obtained in morphants even when detectable levels are seen in few muscle nuclei (S1 Fig D). Overexpression of a CBP form lacking active HAT domain leads to mild rescue indicating that the acetylase activity of CBP is essential for full rescue (+ΔHAT; Figure 1D,E).

Since CBP/p300 expression is generally downregulated in Dystrophin-depleted embryos (Figure 1A), we tested whether CBP overexpression or TSA act though increasing endogenous crebbp/p300 levels. Dystrophin-depleted embryos were co-injected with CBP (107 molecules) or treated with TSA and qPCR was performed at 2 dpf (Figure 1F; S1 Fig E). One-way ANOVA revealed statistically significant differences between treatment groups for crebbpa expression [F (2, 9) = 8.915, p = 0.0073**], Fisher’s LSD post-hoc test finding the decrease upon TSA treatment statistically significant (p = 0.0164*). Independent one-way ANOVAs revealed absence of statistically significant differences in the expression of crebbpb [F (2, 9) = 3.340, p = 0.0823], p300a [F (2, 9) = 1.099, p = 0.3741] and p300b [F (2, 9) = 0.6550, p = 0.5425] transcripts. These results show that neither CBP nor TSA significantly increase any of the transcripts, crebbpa being even decreased upon TSA treatment (Figure 1F). Therefore, rescue is not due to restoring endogenous CBP/p300 transcripts levels.

While the mechanism of action of HDAC inhibition leading to improved regeneration is not fully understood, follistatin upregulation is a key event 6,7. We questioned whether the follistatin pathway could also be involved in the early protection against muscle damage observed in TSA treatment and CBP overexpression. Quantitative analysis of the two zebrafish follistatin transcripts, fsta and fstb, expression was performed at 2 dpf (Figure 1G). Dystrophin depletion does not significantly affect fsta or fstb levels. There were no statistically significant differences in the expression of fsta between groups as determined by one-way ANOVA [F (3, 12) = 1.288, p = 0.3232]. The expression of fstb is statistically significantly different between groups [F (3, 12) = 7.680, p = 0.0040**], Fisher’s LSD post-hoc test finds the decrease upon TSA treatment statistically significant (p = 0.0013**). Summarising, TSA treatment significantly downregulates the expression of fstb while not affecting fsta and both transcripts remain unaltered by CBP overexpression. Therefore, the protection against muscle damage granted by CBP overexpression or TSA treatment does not rely on stimulating follistatin transcription but likely on a follistatin-independent pathway.


Fig. 1: Figure 1. Epigenetic regulators in zebrafish dystrophic embryos.

A) Quantitative PCR analysis of p300/CBP transcript expression at 2 days post fertilisation (dpf) in control and morpholino-depleted embryos (MOdmd). (B) Confocal sections of Tg(actc1b:mCherry)pc4 embryos immunostained for Dystrophin (green). Well organised pattern of muscle fibres (mCherry, red) and nuclei (DAPI, blue) in siblings, while heavily disrupted in dmdta222a/ta222a embryos (arrows), showing collapsed and wavy fibres which ends misalign with somatic borders (dashed lines). Overexpressing CBP (dmdta222a/ta222a + CBP; injection of 107 RNA molecules of plasmid per embryo at one cell stage) has a clear positive effect on dystrophic embryos both at 2 and 6 dpf. (C, D) Frequency of dystrophic phenotype (see Materials and Methods) on 2 dpf Dystrophin-null embryos, either dmdta222a/ta222amutants (C) or depleted with MOdmd (D). Overexpressing CBP (low = 107, high = 108 molecules of plasmid RNA injected per embryo) rescues the dystrophic phenotype of both dmdta222a/ta222a(C) and morphant (D; 107 molecules) embryos as efficiently as blocking HDACs (TSA), even at lower concentrations (C,D). A mutated CBP form lacking the HAT domain (ΔHAT; 107 molecules) significantly decreases the rescue efficiency (C). Individual values for independent experiments are plotted, bars represent the average ± SD and boxes above plots indicate total number of embryos analysed. (E) Representative confocal sections of 2 dpf embryos immunostained for Dystrophin (green) to confirm Dystrophin presence in siblings and depletion in MOdmd embryos, counterstained with DAPI (blue). Both TSA (200 nM) and CBP (107 molecules) are able to rescue the muscle phenotype of MOdmd embryos, while ΔHAT injected embryos most often show muscle damage (107 molecules). nt = neural tube. (F,G) Quantitative PCR analysis of crebbp, p300 and follistatin endogenous transcripts expression at 2 dpf. See text for details on ANOVA (A,F,G) or T tests (B,C), performed with cut-offs: (*) < 0.05, (**) < 0.005, (***) < 0.001, (****) < 0.0001. ns = non significant.


This study shows for the first time that either CBP overexpression or pan-HDAC inhibition are able to prevent, or at least delay, early stages of muscle damage in zebrafish dystrophic muscles. This observation adds up to the recognised positive effect of HDAC inhibitors on muscle regeneration upon damage 1. Other authors observed that treating mouse dystrophic muscles (mdx) with deacetylase inhibitors was able to confer resistance to contraction-coupled degeneration 14. However, this was suggested to be mediated by follistatin through satellite cell number increase. We show that damage protection occurs independently from regeneration. Follistatin upregulation is not involved in protecting early zebrafish muscle from damage, supporting the idea that the two rescuing mechanisms triggered by HDAC inhibition are independent. That the effect of TSA on follistatin expression differs according to the cellular context is not surprising. What we report here is a muscle-cell autonomous effect of TSA preventing degeneration, and in this context we observed a downregulation of follistatin. Other authors focused on a later effect on regeneration of already damaged muscles, mediated by satellite cells stimulation 1,5,6. In this context, follistatin was found to be upregulated. However, it was shown that TSA promotes follistatin upregulation in the fibroadipogenic progenitors rather than in the muscle satellite cells, and the stimulating effect is paracrine 7,15. Interestingly, overexpression of a single HAT, CBP, achieves a similar level of protection as pan-HDAC inhibition. The strong downregulation of fstb by TSA but not CBP should be investigated in future work to determine whether these agents may act through different mechanisms of action. Our results prompt for investigating similar mechanisms in other animal models and identifying new therapeutic targets, possibly with more restricted and controlled effects than HDAC inhibitors. This study strengthen the idea that epigenetics plays a role in the progression of symptoms in some patients with Dystrophin mutations. This could explain the variation in severity in Becker muscular dystrophy patients with identical mutation and the variable progression of DMD. Doubt persists on the mechanism leading to protection. To further understand the mechanisms of action triggered by either HDAC inhibition or CBP a high throughput analysis needs to be employed in future studies.



Fish used were wild-type Danio rerio, dmdta222a/+ and dmdta222a/+ crossed to Tg(actc1b:mCherry)pc4 to facilitate identifying dystrophic muscles. All animals were handled in a facility certified by the French Ministry of Agriculture (approval ID A-31-555-01) and in accordance with the guidelines from the European directive on the protection of animals used for scientific purposes (2010/63/UE), French Decret 2013–118. Staging and rearing was performed as described 16.

Plasmids, morpholinos and HDACs inhibition

pCS2+mRFP-CBP and pCS2+mRFP-ΔHAT plasmids were created by digesting by BamH1 full length CBP of mouse origin, or CBP deleted for amino acids 1430 to 1570 from pCMV-HA-CBP and pCMV-HA-CBPΔHAT (provided by A. Harel-Bellan) and inserting them into pCS2+mRFP-N1 vector (Addgene) digested by BamH1. To produce capped RNA to inject, the vectors were linearized with NotI and the mMESSAGE mMACHINE SP6 kit (Ambion) was used. Injections were performed at one-cell stage with the indicated amounts of RFP-CBP and RFP-CBPΔHAT RNA. RNA copy number was calculated using the NEBioCalculator v1.7.1 (BioLabs). Mouse CBP shares 85% protein identity to both zebrafish CBP-A (the product of crebbpb) and zebrafish CBP-B (the product of crebbpa), according to the standard NCBI BLAST tools. Zebrafish protein CREBBPA (also called CREBBPb or CBP-B) corresponds to protein ID ENSDARP00000081684, while CREBBPB protein ID (also called CREBBPa or CBP-A) corresponds to ENSDARP00000086306.

Morpholinos were from Gene-tools and were used as a cocktail as described previously (dmd-MO1 at 4 ng per embryo and dmd-MO6 at 7.5ng/embryo 13.

Embryos were exposed to Trichostatin A (TSA, Sigma) for 24 hours, from 24 to 48 hpf. TSA was added to fish water for a final concentration of 200 nM as described previously 13,17.

Frequency of dystrophic embryos and rescue analysis

Phenolthiourea (0.003%) was added at 24 hpf to inhibit pigmentation and facilitate muscle analysis. To determine the dystrophic phenotype frequency each embryo was dechorionated individually and observed free moving, then anaesthetized with tricaine (0.2 mg/ml) to facilitate close observation of muscles. The embryos were split into two groups, “normal” or “dystrophic” phenotype. Dystrophic muscles were characterised by abnormal birefringence, supported by observing disruption of the actin reporter pattern when Tg(actc1b:mCherry)pc4 fish were used, and impaired movement 10,18. After sorting by phenotype, immunohistochemistry (details below) was performed to identify embryos negative for Dystrophin expression (i.e., dmdta222a/ta222aor dmd-morphants) within both the “normal” and “dystrophic” phenotypical groups. The plotted “frequency of dystrophic embryos” refers to the relative amount of Dystrophin-negative embryos that effectively show a dystrophic phenotype. Rescued embryos are those that despite being Dystrophin-negative do not show a dystrophic phenotype upon a specific treatment. Note that the rescue analysis is blind in the sense that the phenotype analysis is done prior to any possibility of knowing the genotype of the embryo, or the state of Dystrophin expression.


Standard protocols were used. Embryos were fixed in cold methanol for dystrophin staining, or otherwise in paraformaldehyde 4%. Primary antibodies were mouse anti-Dystrophin (MANDRA1 (7A10), Developmental Studies Hybridoma Bank) and rat anti-RFP (Chromotek). Secondary antibodies were goat anti-mouse Alexa-488 and goat anti-rat Alexa-546 (Molecular Probes). DAPI was used to counterstain nuclei.

Microscopy, software and statistics

An inverted Zeiss 710 LSM with a 20x/1.0 W Plan-Apochromat and a 40x/1.3 Plan-Apochromat objective were used for Z-stack acquisition. Acquisition and maximum intensity projections were made with ZEN 2009/2010 (Zeiss). Images were uniformly contrasted with Adobe Photoshop CS6. Illustrations were made in Adobe Illustrator CS6. GraphPad Prism 6 was used for graph plotting and statistical analyses.

Real time quantitative PCR

Four samples were analysed for each condition, each sample consisting of 15 embryos at 2 days post fertilisation (dpf) for a total of 60 embryos per condition. Total RNAs were extracted with the EZNA total RNA kit I (Omega Bio-tek), and reverse-transcribed with the qScript cDNA synthesis kit (Quanta biosciences) according to the supplier’s instructions. A C1000™ Thermal Cycler with CFX96™ Real time System (BioRad) was used to perform the qPCR and analyses were performed on BioRad CFX Manager 2.0, according to the manufacturer’s instructions. SsoFast™ EvaGreen® Supermix (Bio-Rad) was used according to the manufacturer’s instructions. The following primer sequences were used: EF1α Fwd 5’- GAT GCA CCA CGA GTC TCT GA -3’; EF1α Rev 5’- TGA TGA CCT GAG CGT TGA AG -3’; m-crebbp Fwd 5′- TGC CAA GTT GCC CAT TGT G -3′; m-crebbp Rev 5′- TTG TTG GTT TCG CTT GTC ACT -3′ ; z-crebbpa Fwd 5’- CGA AAA GTG GAA GGG GAC AT -3’; z-crebbpa Rev 5’- TTC TCT TCC AGC TCT TTC TGG -3’; z-crebbpb Fwd 5’- CAG GTT CCT CAA GGG ATG G -3’; z-crebbpb Rev 5′- CCA TCA TGG CTT GAG CTT G -3’; p300a Fwd 5′- CAC CTT CCT CAA CAC CAC AGT -3’; p300a Rev 5′- GCA TAG CAT TCT GGT CTG CTC -3′; p300b Fwd 5′- ATA TGG CCG TCA GGG TTT ATC -3′; p300b Rev 5′- CTC GTG TCT CCA GAA AGT TGC -3′; fsta Fwd 5’- GAT GCA AAA TGA ACA GGA GGA -3’; fsta Rev 5’- GAC TTC AAA AGG GCA CAT TCA -3’; fstb Fwd 5’- GCA TGG ACT GAG GAG GAT GTA -3’; fstb Rev 5’- CAC CTC TTT CCA GAA CCA CAA -3’. All experiments included a standard curve for each primer pair and control for genomic DNA contamination. Data were normalized to EF1α and expressed relative to the respective control that is set as 1 in boxplot with whiskers representing the median, 0.25 percentiles, maximum and minimum values.


This work was supported by grants from the Fondation de l’Association pour la Recherche contre le Cancer (Fondation ARC ;; SFI20121205590); Association Française contre les Myopathies (AFM ;; 20102); joint call Alliance pour la Vie et pour la Santé (AVIESAN ; et Plan Cancer ( (P036407) and Institut National du Cancer (INCa ;; P030972) to L.V and a fellowship from the Fondation pour la Recherche Médicale (FRM ; ; SPF20130526627) to F.B.

Competing Interests

The authors have declared that no competing interests exist.

Data Availability

The values behind statistics and graphs presented here are deposited at FigShare, DOI: 10.6084/m9.figshare.5570038.

Corresponding Authors

Fernanda Bajanca,

Laurence Vandel,

Supplementary Information


S1 Fig. (A) Wild-type and MOdmd injected zebrafish embryos at 2 dpf, showing drastic reduction of Dystrophin on embryos injected with morpholino cocktail compared with wild type expression. No secondary developmental defects were observed. (B) Control quantitative PCR with specific primers for crebbp of mouse origin performed on 2 dpf control embryos (non-injected) and embryos injected at one-cell stage with 107 (low), 5×107 (medium) or 108 (high) molecules of mRFP-CBP. Note that: i) endogenous zebrafish crebbp transcripts are not detected in control non-injected embryos; ii) the number of transcripts detected at 2 dpf increases linearly with the increasing number of transcripts injected at one cell stage. (C) Representative confocal sections of 2 dpf wild type embryos injected at one-cell stage with 10-fold doses of RFP-tagged CBP: 107 (low) and 108 (high) RNA molecules per embryo. Embryos were immunostained for exogenous CBP (αRFP; red), Dystrophin (green), and nuclei counterstained with DAPI (blue). Note the distribution of exogenous CBP, accumulated in the nuclei of a small subset of muscle cells when a lower dose is injected (B, red) compared with more widespread expression when a higher dose is injected (C, red). (D) Representative confocal tiled section of a 2 dpf embryo, co-injected at one-cell stage with MOdmd plus 107 RNA molecules (low dose) of RFP-tagged CBP expression plasmid. Panels show immunostaining for Dystrophin (green), exogenous CBP (αRFP; red) and nuclei counterstained with DAPI (blue). Note that despite Dystrophin expression is undetectable the muscles are not damaged, even if only a small subset of total nuclei express exogenous CBP. (E) Control for z-crebbpa and z-crebbpb qPCR primers specificity: neither detects the exogenous CBP of mouse origin. PCR was performed on template pCS2+mRFP-CBP cDNA at concentrations: 100 ng/µl (lines 2-4), 10 ng/µl (lines 5-7) and 1 ng/µl (lines 8-10), and primers: m-crebbp (lines 2, 5, 8), z-crebbpa (lines 3, 6, 9) and z-crebbpb (lines 4, 7, 10). A 100-bp ladder (Biolabs) was used on lines 1 and 11.

]]> 0
Influenza A Virus Infection Damages Zebrafish Skeletal Muscle and Exacerbates Disease in Zebrafish Modeling Duchenne Muscular Dystrophy Wed, 25 Oct 2017 12:25:58 +0000 INTRODUCTION: Both genetic and infectious diseases can result in skeletal muscle degeneration, inflammation, pain, and/or weakness. Duchenne muscular dystrophy (DMD) is the most common congenital muscle disease. DMD causes progressive muscle wasting due to mutations in Dystrophin. Influenza A and B viruses are frequently associated with muscle complications, especially in children. Infections activate an immune response and immunosuppressant drugs reduce DMD symptoms. These data suggest that the immune system may contribute to muscle pathology. However, roles of the immune response in DMD and Influenza muscle complications are not well understood. Zebrafish with dmd mutations are a well-characterized model in which to study the molecular and cellular mechanisms of DMD pathology. We recently showed that zebrafish can be infected by human Influenza A virus (IAV). Thus, the zebrafish is a powerful system with which to ask questions about the etiology and mechanisms of muscle damage due to genetic and/or infectious diseases.

METHODS: We infected zebrafish with IAV and assayed muscle tissue structure, sarcolemma integrity, cell-extracellular matrix (ECM) attachment, and molecular and cellular markers of inflammation in response to IAV infection alone or in the context of DMD.

RESULTS: We find that IAV-infected zebrafish display mild muscle degeneration with sarcolemma damage and compromised ECM adhesion. An innate immune response is elicited in muscle in IAV-infected zebrafish: NFkB signaling is activated, pro-inflammatory cytokine expression is upregulated, and neutrophils localize to sites of muscle damage. IAV-infected dmd mutants display more severe muscle damage than would be expected from an additive effect of dmd mutation and IAV infection, suggesting that muscle damage caused by Dystrophin-deficiency and IAV infection is synergistic.

DISCUSSION: These data demonstrate the importance of preventing IAV infections in individuals with genetic muscle diseases. Elucidating the mechanisms of immune-mediated muscle damage will not only apply to DMD and IAV, but also to other conditions where the immune system, inflammation, and muscle tissue are known to be affected, such as autoimmune diseases, cancer, and aging.



Skeletal muscle is critical for homeostasis because skeletal muscle is required for breathing, posture, locomotion, metabolism, thermoregulation, and the immune response. Muscle tissue is remarkably plastic and can increase or decrease in mass in response to genetic and environmental factors. Muscle degeneration is a serious health issue that can reduce lifespan and quality of life. Muscle wasting can be caused by aging, injury, disuse, medications, genetic mutations, and infectious or inflammatory diseases. Understanding how muscle growth, regeneration, and degeneration are regulated in response to genetic and environmental insults alone and in combination is an important undertaking in order to be able to promote muscle health in cases of sickness and disease.

Skeletal muscle damage occurs in response to some genetic and infectious or inflammatory diseases. The most common, genetic muscle wasting disease is Duchenne Muscular Dystrophy (DMD), which is caused by mutations in the DMD gene. The most common viral infections that cause muscle pain and weakness are Influenza A and B viruses1 . Many strides have been made towards elucidating the mechanisms of muscle degeneration due to DMD mutations. Dystrophin is required in muscle fibers for sarcolemma integrity2 and in muscle stem cells for proper polarity and asymmetric cell division3. However, much less is known about the etiology of Influenza-induced muscle damage and nothing is known about the consequences of Influenza infection in the context of patients with genetic muscle wasting diseases.

Biopsies from patients with genetic muscle diseases or muscle complications of Influenza infection show biomarker and histological similarities, suggesting that these conditions may share common mechanisms of muscle damage. Creatine kinase was upregulated and correlated with poor outcome in patients with IAV muscle complications4. Creatine kinase upregulation is also used in the diagnosis of DMD. The first histological report of muscle biopsies from IAV (H1N1)-infected people found muscle necrosis, fibers with variable diameters, atrophic round fibers, atrophic angulated fibers, type 1 and 2 fiber atrophy, and type 1 fiber predominance5. The findings from these biopsies are similar to reports of DMD histopathology, which include fiber size variability, fiber necrosis, regeneration, inflammation, and connective tissue deposition6. It is not known whether Influenza infection exacerbates muscle damage in the context of genetic muscle diseases.

Here, we characterize muscle damage and assay innate immune/inflammatory markers in muscle in response to IAV infection in vivo. We also determine the consequences of IAV infection on muscle degeneration in an animal model of DMD. To our knowledge, IAV has never been experimentally induced in any model of DMD. We used zebrafish embryos for these experiments as it has been shown that zebrafish embryos can be infected with human isolates of IAV7 and sapje/dmd mutant zebrafish are a well-characterized model that recapitulates aspects of human disease and has greatly contributed to our knowledge of DMD mechanisms and potential treatments8,9,10,11,12,13,14,15,16,17. Therefore, zebrafish embryos are well-suited for the study of the combinatorial effects of genetic and infectious diseases on muscle degeneration. We find that systemic IAV infection potentiates fiber damage accompanied by markers of heightened inflammation in zebrafish muscle tissue. We also find that IAV infection greatly exacerbates the extent of fiber damage in zebrafish modeling DMD. Taken together, these results show an important gene-environment interaction between the pro-inflammatory innate immune response and the DMD gene in skeletal muscle.


Ethics statement

Zebrafish (Danio rerio) used in this study were handled in accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The protocols used in this study were approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Maine (Protocol Number: A2013-06-03).

Zebrafish care, staging, and husbandry

Zebrafish were maintained in the Zebrafish Facility at the University of Maine, Orono. The facility was maintained according to IACUC standards. IACUC approved guidelines for zebrafish care followed the standard procedures ( of a 14 hour light, 10 hour dark cycle at 28°C. Embryos were obtained from natural spawnings of these adult fish and grown at 28°C or 33°C. Fertilized eggs were collected and raised in egg water (60 μg/ml Instant Ocean sea salts; Aquarium Systems, Mentor, OH). Developmental staging was performed according to18. Transgenic lines used were the Tg(fli1:GFP) line19, Tg(NFkB:EGFP)nc1 line20, and the Tg(BACmpo:gfp)i114 line21. The mutant line used was the sapjeta222a line8. Approximately 900 embryos were used for these studies.

IAV and Evans Blue Dye (EBD) injection

Influenza A/PR/8/34 (H1N1) virus was purchased from Charles River Laboratories, aliquoted upon arrival, and stored at -80°C. Embryos were manually dechorionated at 2 days post fertilization (dpf) with fine forceps (DuMont no. 5). Prior to injections, 2 dpf fish were anesthetized in tricaine solution and lined up on a 3% agarose gel in a Petri dish before being injected into the Duct of Cuvier (DC) with 1.5 nl (~1×104 EID50) of wild-type A/PR/8/34 IAV or 4 nl [~6×102 plaque forming units (PFU) per embryo] of NS1-GFP A/PR/8/34 IAV in PBS with a final concentration of 0.25% phenol red. For experiments involving EBD, phenol red volume was replaced with EBD2,22 and zebrafish were injected as before. Sterile PBS including phenol red or EBD was injected into the DC of 2 dpf control zebrafish. Injection volumes were calibrated using a micrometer slide. Following injection, zebrafish were maintained at 33°C until fixation or euthanasia and egg water was changed daily. Euthanasia was carried out by immersion in 300mg/L MS222 buffered with Sodium Bicarbonate for 10 minutes. Microinjection was controlled by an MPPI-2 pressure microinjector (Applied Scientific Instruments) and pulled microcapillary pipettes (Sutter Instruments, Novato, CA) were used to inject the virus or PBS.

Phalloidin staining and immunohistochemistry

Alexa Fluor 488 phalloidin (Molecular Probes) staining involved fixing embryos in 4% Paraformaldehyde (PFA) for 4 hours (h) at room temperature (RT), washing 5 times for 5 min each (5 × 5) in PBS-0.1% Tween20, permeabilizing for 1.5h in PBS-2% Triton, washing 5×5 and then incubating in phalloidin (1:20) for 1–4h at RT or overnight at 4°C. All antibodies (Abs) were diluted in block (5% w/v Bovine Serum Albumin (BSA) in Phosphate Buffered Saline (PBS) with 0.1% Tween20). Ab staining followed phalloidin staining or started with blocking for 1h at RT, incubating in 1° Ab overnight at 4°C, washing for 2–8h in block at RT, incubating in 2° Ab overnight at 4°C, then washing for at least 1h in PBS-0.1% Tween. 1° Abs: anti-β-Dystroglycan 1:50 (Novocastra); anti-Dystrophin 1:50 (Sigma D8043); anti-Paxillin 1:50 (BD Biosciences). 2° Abs: GAM/GAR 488, 546, 633 1:200 (Invitrogen).


Images were obtained on a Zeiss Imager-Z compound microscope with ApoTome attachment running Axiovision software or an Olympus Fluoview IX-81 inverted microscope with FV1000 confocal system. Linear adjustments were made to images in Adobe Photoshop and figures were collated in Adobe Illustrator. Fixed and stained zebrafish were deyolked and mounted in PBS for imaging. Live imaging for EBD experiments involved anesthetizing zebrafish in tricaine embedding them in low melt agarose and imaging in 24 well plates with glass bottoms.

Quantitative polymerase chain reaction (qPCR)

Total RNA was extracted from whole embryos at 24 hours post infection (hpi) by homogenizing 10 fish, treating with TRIzol reagent (Invitrogen, Carlsbad, CA) and subsequently storing at −80°C. RNA was extracted according to the manufacturer’s protocol. Reverse transcription reactions to synthesize cDNA were performed according to manufacturer’s instructions using Bio-Rad iScript reagents (Bio-Rad Laboratories, Hercules, CA). Bio-Rad SsoFast EvaGreen reagents were used for qPCR reactions according to manufacturer’s instructions. qPCR was performed on a Bio-Rad I-cycler IQDetection system using cycling parameters described previously23. Gene expression was normalized to the corresponding beta-actin value using the delta delta ct method to determine relative transcript abundance.


Zebrafish embryos were injected and maintained as described above. For mortality experiments, egg water was changed daily and mortality (defined as lack of a discernible heart beat) was monitored and recorded from 0-5 days post infection (dpi). Deceased zebrafish were removed each day. The sapje phenotype of disrupted birefringence was determined in live zebrafish using two polarized light filters.


IAV infects muscle cells and causes muscle damage

Symptoms of IAV muscle complications include pain, tenderness, weakness, and problems with ambulation. These symptoms usually resolve in a week regardless of treatment; however, muscle complications can be severe. IAV has been shown to directly infect cultured human muscle cells24 Despite this observation, IAV is only rarely recovered from muscle biopsies25 and muscle biopsies show inconsistent infiltration of immune cells. Therefore, it is unclear to what degree muscle fiber infection, inflammation, or both contribute to the muscle complications caused by IAV infection. To study the effects of IAV infection on skeletal muscle tissue in vivo, we utilized the zebrafish model. We previously showed that IAV infects and replicates in zebrafish embryos when administered via intravenous injection7. We used this model of IAV infection to ask whether IAV can enter and infect muscle cells and the effect of IAV infection on muscle tissue.

To investigate the effect of IAV infection on skeletal muscle structure, 2 dpf zebrafish were infected with human IAV (H1N1, A/PR/8/34), fixed, and then stained with phalloidin to visualize F-actin at 24 and 48 hpi (hours post infection). Phalloidin staining in PBS-injected zebrafish at 24 and 48 hpi revealed the normal segmented, highly ordered, parallel arrays of muscle fibers (Fig. 1A-B). Zebrafish infected with IAV displayed foci of muscle degeneration at 24 hpi and 48 hpi (hours post infection) (Fig. 1C-D, white arrowheads). Muscle damage worsened over time: damage was more frequently observed in infected zebrafish at 48 hpi than 24 hpi (Fig. 1C-E). Sites of muscle damage were more prevalent in the anterior muscle segments than the posterior muscle segments of infected zebrafish at 24 hpi (Fig. 1F). These data show that injection of IAV into the bloodstream of zebrafish embryos has an impact on skeletal muscle tissue and results in areas of muscle degeneration; however, the data do not determine if the muscle damage observed is an indirect consequence of a systemic Influenza infection or if muscle cells are infected by IAV.

To determine if IAV infects zebrafish muscle cells, we injected 2 dpf zebrafish with a fluorescent reporter strain of IAV (NS1:GFP) where GFP expression signifies that translation of a recombinant viral gene occurred in an infected host cell7,26. Injection of this strain of IAV into zebrafish embryos resulted in punctate GFP fluorescence throughout embryos by 24 hpi whereas GFP fluorescence was not observed in PBS-injected control zebrafish embryos (Fig. 1G-H). To determine if any of the cell types expressing NS1:GFP were muscle cells, we imaged infected zebrafish under higher magnification. We occasionally observed zebrafish muscle cells expressing NS1:GFP (Fig. 1I-I1). The NS1:GFP fluorescent reporter strain of IAV is approximately ten-fold less infectious than wild-type IAV26. Therefore, infection of zebrafish with NS1:GFP did not cause foci of muscle damage and we were unfortunately unable to determine whether fiber degeneration was limited to IAV-infected muscle fibers or not. Our results show that IAV can enter and infect zebrafish muscle cells in vivo and suggest that muscle degeneration, pain, and weakness may be, at least in part, due to direct infection of muscle cells by IAV.

IAV muscle Fig 1

Fig. 1: Human IAV infects zebrafish muscle cells and causes muscle fiber damage

All embryo images are side mounts, dorsal top, anterior left. Panels A-D are phalloidin stained to visualize F-actin. White arrowheads denote retracted fibers. (A) PBS-injected control at 24 hpi (3 dpf). (B) PBS-injected control at 48 hpi (4 dpf). (C) IAV-infected embryo at 24 hpi. (D) IAV-infected embryo at 48 hpi. (E) Quantification of the proportion of muscle segments per embryo with damaged fibers in IAV-infected embryos over developmental time. Fiber damage is more frequently observed at 48 hpi than at 24 hpi. (F) Spatial location of damaged fibers along the anterior-posterior axis of IAV-infected embryos at 24 hpi. The frequency of damaged fibers peaks in segments 5-9, which is in the anterior of the fish near the Duct of Cuvier (the site of injection). (G) PBS-injected control at 24 hpi (3 dpf). (H) NS1-GFP-injected zebrafish at 24 hpi. Note the punctate green fluorescence in infected cells throughout the body. (I) NS1-GFP-injected zebrafish at 24 hpi. Higher magnification view of a GFP-positive, infected muscle fiber. (I1) Merged panel of NS1-GFP and brightfield images.

IAV infection results in fiber detachment and decreased sarcolemma integrity

We showed that IAV infection causes muscle degeneration in zebrafish. The phenotype in IAV-infected zebrafish is similar to what is seen in zebrafish models of congenital muscle diseases. Work in these zebrafish disease models showed that detached muscle fibers can occur via at least two different etiologies: (1) loss of muscle membrane (sarcolemma) integrity and subsequent fiber death, or (2) detachment of muscle fibers from their surrounding ECM prior to loss of sarcolemma integrity and cell death. Injection of the fluorescent, cell impermeable Evans blue dye (EBD) can be used to discriminate between these two possibilities. If EBD penetrates long and/or retracted muscle fibers, sarcolemma damage has occurred. Loss of sarcolemma integrity occurs in zebrafish models of DMD and dysferlinopathy2,27; however, retracted fibers do not take up EBD in a zebrafish model of Merosin-deficient congenital muscular dystrophy (MDC1A)28. Interestingly, in the two zebrafish models of primary dystroglycanopathy, each caused by a different mutation in the dag1 gene, retracted fibers were reported to remain impermeable to EBD29 or to take up EBD only after retraction30. We sought to determine the etiology of muscle fiber retraction in IAV-infected zebrafish using the EBD permeability assay.

For this experiment, 2 dpf zebrafish were infected with IAV as before except that the phenol red in the injection solution was replaced with EBD. Live zebrafish were imaged at 24 hpi. In PBS-injected Tg(fli1:GFP) zebrafish, where GFP expression driven by an endothelial-specific promoter allows for visualization of the vasculature in vivo19, EBD remained within blood vessels (Fig. 2A, white arrowheads from top to bottom point to EBD in intersomitic vessels, the dorsal aorta, and the caudal vein). Uptake of EBD into muscle cells was not observed in PBS-injected zebrafish (Fig. 2A-C). In IAV-infected Tg(fli1:GFP) zebrafish, EBD leaked out of the vasculature and was taken up by long muscle fibers (Fig. 2D, F, black arrowheads). In addition to permeating some long muscle fibers, EBD was taken up by retracted fibers in IAV-infected zebrafish (Fig. 2E-F, white arrowheads denote EBD in retracted fibers). These data show that IAV infection increases vascular permeability as well as compromises sarcolemma integrity in attached and retracted fibers in zebrafish embryos.

Fig 2

Fig. 2: IAV infection compromises the sarcolemma

All embryo images are side mounts, dorsal top, anterior left. (A) Tg(fli1:GFP) zebrafish embryo with labeled endothelial cells (green) DC-injected with PBS plus EBD (red). EBD remains in the vasculature. White arrowheads point to EBD in an intersomitic vessel (top left), the dorsal aorta (middle), and the caudal vein (bottom right). (B-C) Wild-type zebrafish injected with PBS plus EBD. (B) Cropped EBD panel. (B1) Cropped EBD and brightfield panels merged. (C) EBD panel. Note that muscle fibers are impermeable to EBD in PBS-injected zebrafish. (D) Tg(fli1:GFP) zebrafish embryo DC-injected with IAV plus EBD (red). EBD leaked out of the vasculature and penetrated muscle fibers (black arrowheads). (E-F) Wild-type zebrafish injected with IAV plus EBD. (E) Cropped EBD panel. (E1) Cropped EBD and brightfield panels merged. (F) EBD panel. Note the uptake of EBD by long (black arrowheads) and retracted fibers (white arrowheads) indicative of sarcolemma damage.

To investigate the possibility that the detached fibers in IAV-infected zebrafish could also be due to disrupted adhesion to the ECM, we performed antibody staining to visualize intracellular proteins that localize to the myotendinous junction (MTJ) and are known to be involved in stable muscle fiber-ECM attachments. We assessed the localization of beta-Dystroglycan, Dystrophin, and Paxillin proteins in the muscle fibers of IAV-infected zebrafish. In mock-infected zebrafish, no retracted fibers were observed (Fig. 3A) and beta-Dystroglycan localized to MTJs (white arrow in Fig. 3A1-2) and to neuromuscular junctions (white arrowhead in Fig. 3A1-2). In IAV-infected zebrafish, detached fibers were clearly visible with phalloidin staining (white arrows in Fig. 3B, C, D). Many retracted fibers in IAV-infected zebrafish were observed to have beta-Dystroglycan, Dystrophin, or Paxillin still localized to their detached end (white arrowheads in Fig. 3B1-2, C1-2, D1-2). These data show that some fibers can retract in IAV-infected zebrafish without disruption to the localization of their intracellular MTJ anchoring complexes. These data are consistent with previous experiments that show fiber detachment and then death can be due to an extracellular disruption in adhesion. Altogether, our results suggest that IAV infection damages and causes muscle fiber death in zebrafish via at least two (not necessarily mutually exclusive) mechanisms: (1) loss of sarcolemma integrity and (2) failure of muscle fiber-ECM adhesion external to the sarcolemma.

Fig 3

Fig. 3: IAV infection disrupts muscle fiber-ECM adhesion

All embryo images are side mounts, dorsal top, anterior left of zebrafish at 24 hpi. Lettered panels show phalloidin staining for F-actin in red. Panels numbered 1 show immunohistochemistry for beta-Dystroglycan, Dystrophin, or Paxillin proteins in green. Panels numbered 2 are merged images of phalloidin and antibody staining. (A-A2) Phalloidin and beta-Dystroglycan staining in a PBS-injected zebrafish. White arrow in A1 denotes MTJ localized beta-Dystroglycan and white arrowhead in A1 points to neuromuscular junction localized beta-Dystroglycan. (B-B2) Phalloidin and beta-Dystroglycan staining in an IAV-injected zebrafish. White arrow in B points to a retracted fiber and white arrowheads in B1-2 highlight beta-Dystroglycan staining at the unattached end of a retracted fiber. (C-C2) Phalloidin and Dystrophin staining in an IAV-injected zebrafish. White arrow in C points to a retracted fiber and white arrowheads in C1-2 highlight Dystrophin staining at the unattached end of a retracted fiber. (D-D2) Phalloidin and Paxillin staining in an IAV-injected zebrafish. White arrow in D points to a retracted fiber and white arrowheads in D1-2 highlight Paxillin staining at the unattached end of a retracted fiber. These results showing that some retracted fibers retain the localization of ECM adhesion proteins suggest that muscle fibers-ECM adhesion can be disrupted external to the sarcolemma.

An inflammatory innate immune response is elicited in muscle tissue upon IAV infection

Although we have demonstrated that human IAV infects zebrafish muscle cells and causes muscle fiber damage and degeneration, it is still unclear to what degree the innate immune/inflammatory response contributes to muscle complications of IAV infection. This is because of inconsistent detection of immune cells in muscle biopsies from IAV-infected humans. The distinction between IAV-induced myopathy (muscle degeneration due to a defect within muscle) and IAV-induced myositis (muscle degeneration due to the pro-inflammatory innate immune response) is an important one to make because these conditions would likely respond to different treatments. The immune response is well conserved between humans and zebrafish (in terms of molecules, signaling pathways, and cell types/functions); however, only the innate immune response is functional in zebrafish at the time points of our experiments31. We have previously shown that systemic IAV infection elicits an innate immune response in zebrafish embryos/larvae7,32, but inflammation in muscle tissue was not specifically examined.

We used the accessibility of the zebrafish model to ask whether IAV elicits an innate immune response in zebrafish muscle cells by infecting in multiple transgenic lines of zebrafish. First, we used the NFkB reporter line20 to investigate activation of NFkB-dependent, pro-inflammatory innate immune signaling cascades at the molecular level. Transgenic embryos (2 dpf) were injected with PBS or IAV as before and then imaged for GFP at 24 hpi. In PBS-injected embryos, GFP expression was observed in certain cell types, such as neuromasts of the lateral line, but not in muscle cells (Fig. 4A). This result shows that NFkB signaling is normally not active in zebrafish muscle cells at 3 dpf. In IAV-injected zebrafish, in addition to the basal level of active NFkB signaling seen in controls, we observed GFP-positive muscle fibers (white arrowhead in Fig. 4B). Muscle fibers expressing GFP occurred throughout the anterior-posterior axis of infected zebrafish; however, GFP-positive muscle fibers were more prevalent in the anterior muscle segments (Fig. 4B). This distribution of GFP-positive muscle fibers correlates with the increased frequency of muscle degeneration in anterior muscle segments of IAV-infected zebrafish (Fig. 1F). Our data show that NFkB transcription factor-dependent, pro-inflammatory innate immune signaling cascades are turned on in muscle cells in IAV-infected zebrafish.

We additionally interrogated the innate immune response at the molecular level by performing qPCR to assess pro-inflammatory cytokine gene expression. We previously showed that IAV infection elicits an antiviral innate immune response in zebrafish (as assayed by differential expression of Interferon and Myxovirus Resistance Gene A family members)7. Here, we found that IAV-infected zebrafish had approximately 11-fold more interleukin 1, beta mRNA and 3.5-fold more interleukin 8 mRNA than mock-infected zebrafish at 24 hpi (Fig. 4C). We did not detect a change in tumor necrosis factor a mRNA expression in IAV-infected zebrafish compared to PBS-injected zebrafish at 24 hpi (Fig. 4C). Together, our data show that IAV infection activates pro-inflammatory, innate immune cell signaling cascades in zebrafish muscle and culminates in changes to downstream target gene expression.

Next, we infected 2 dpf zebrafish expressing GFP under the control of the neutrophil-specific myeloperoxidase (mpo or mpx) promoter21 to look at the innate immune response to IAV infection in muscle at the cellular level. In zebrafish at this developmental stage, neutrophils mainly localize to the caudal hematopoietic tissue but can be recruited to sites of injury and/or infection21,33. We also conducted phalloidin staining to allow for visualization of muscle tissue structure. In PBS-injected zebrafish, muscle tissue appeared normal and few GFP-positive neutrophils were observed in muscle (Fig. 4D-D2). Phalloidin staining of IAV-infected zebrafish at 24 hpi revealed sites of damaged muscle fibers, especially in anterior muscle segments (Fig. 4E). Interestingly, neutrophils were found not only to infiltrate muscle tissue in IAV-infected zebrafish (Fig. 4E1), but were observed specifically localized to the unanchored ends of retracted fibers (Fig. 4E2, white arrowheads). The role of neutrophils at the ends of retracted fibers is unknown; however, our data clearly show infiltration of neutrophils in muscle tissue and place neutrophils at the right time and place to play a role in IAV-induced muscle degeneration. Taken together, our data show that inflammation is activated at the molecular and cellular levels in zebrafish muscle in response to IAV infection. Our data also suggest the hypothesis that muscle complications due to IAV infection are may be myositis.

Fig 4

Fig. 4: Molecular and cellular markers of inflammation are present in the muscle tissue of IAV-infected zebrafish

All embryo images are side mounts, dorsal top, anterior left of zebrafish at 24 hpi. (A) PBS-injected Tg(NFkB:GFP) zebrafish embryo which expresses GFP in cells where NFkB transcription factor-dependent gene expression is occurring. NFkB signaling is active in the lateral line system, but not in muscle cells. Inset panel is a merge of fluorescence and brightfield imaging. (B) IAV-injected Tg(NFkB:GFP) zebrafish embryo. NFkB-dependent gene transcription is turned on in muscle cells in response to IAV infection. Inset panel is a merge of fluorescence and brightfield images. (C) Quantification of pro-inflammatory cytokine mRNA expression in IAV-infected zebrafish compared to PBS-injected zebrafish at 24 hpi. Expression of interleukin 1, beta (11.1 +/- 5.2-fold increase; 3 biological replicates; 3 independent experiments) and interleukin 8 (3.7 +/- 1.4-fold increase; 3 biological replicates; 3 independent experiments) increases in response to IAV while tumor necrosis factor a expression remains unchanged (1.1 +/- 0.5-fold increase; 2 biological replicates; 2 independent experiments). (D-E2) Lettered panels show phalloidin staining (red), panels numbered 1 show GFP-positive neutrophils, and panels numbered 2 are merged. (D-D2) PBS-injected Tg(mpx:GFP) zebrafish. Note that not many neutrophils are present in muscle tissue. (E-E2) IAV-infected Tg(mpx:GFP) zebrafish. Note the retracted muscle fibers and the infiltration of muscle tissue by neutrophils. White arrowheads in E2 point to neutrophils localized to the unanchored ends of retracted muscle fibers.

IAV-infected sapje/dmd mutant zebrafish embryos have an increased incidence of muscle damage

While muscle pain and weakness are unpleasant complications caused by infections, they normally resolve within a week in otherwise healthy individuals. However, severe skeletal or cardiac muscle complications due to infection can occur in immunocompromised patients and can be life-threatening for people with genetic muscle diseases34,35,36. Skeletal muscle damage is known to be exacerbated in patients with Fukuyama congenital muscular dystrophy upon infection with coxsackie or enteroviruses34 or Parvovirus B1935. Parvovirus B19 infection in siblings with MDC1A led to severe myocarditis in both patients and was fatal in one case36. While individuals with muscular dystrophies are strongly advised to get vaccinated against Influenza viruses due to the potential for respiratory complications, it is unknown, to the best of our knowledge, how IAV infection impacts skeletal muscle tissue in the context of a genetic muscle disease.

To address this question, we infected 2 dpf zebrafish modeling DMD (sapje/dmd mutants) with IAV and analyzed muscle tissue structure with phalloidin staining at 24 hpi. Wild-type siblings injected with PBS showed no signs of muscle degeneration in anterior (Fig. 5A) or posterior (Fig. 5B) muscle segments. Sapje/dmd mutant embryos injected with PBS displayed foci of muscle degeneration in anterior and posterior muscle segments (Fig. 5C-D, white arrowheads). IAV-infected, wild-type siblings had sites of muscle degeneration mainly in anterior muscle segments (Fig. 5E-F, white arrowhead). Sapje/dmd mutant embryos infected with IAV displayed severe muscle degeneration in virtually every muscle segment along the anterior-posterior axis (Fig. 5G-H, white arrowheads). Quantification of the muscle segments with damaged fibers showed that IAV infection and dmd mutation have a synergistic effect (i.e. more than an additive effect) in terms of causing muscle degeneration in zebrafish (Fig. 5I). We also tracked survival for 5 days post infection (dpi) (i.e. 7 dpf). PBS-injected sapje/dmd siblings or mutants all survived until 5 dpi (Fig. 5J, blue lines). Mortalities were initially observed in IAV-infected, wild-type siblings at 2 dpi, but most mortalities occurred between 4 and 5 dpi (Fig. 5J, green line). There was no difference in total mortality between IAV-infected, wild-type siblings and IAV-infected sapje/dmd mutants at 5 dpi; however, mortalities in IAV-infected sapje/dmd mutants occurred in greater numbers at earlier time points compared to infected wild-type siblings (Fig. 5J, purple line compared to green line). Our data suggest that underlying dmd mutations predispose individuals to muscle damage upon IAV infection and that these individuals are more likely to succumb earlier in the time course of the infection. Altogether, our investigation of skeletal muscle tissue in our zebrafish model of IAV infection shows that muscle is an important in vivo target of IAV, that IAV-induced fiber damage is associated with inflammation in muscle tissue, and that muscle complications caused by IAV are greatly exacerbated in a genetic muscle disease model.

Fig 5

Fig. 5: Muscle damage caused by IAV infection is exacerbated in the zebrafish model of DMD

All embryo images are side mounts, dorsal top, anterior left of zebrafish at 24 hpi stained with phalloidin to visualize F-actin. White arrowheads point to foci of muscle damage. (A) Anterior muscle segments of a PBS-injected wild-type sibling embryo. (B) Posterior muscle segments of a PBS-injected wild-type sibling embryo. Note the lack of muscle damage present. (C) Anterior muscle segments of a PBS-injected sapje/dmd mutant embryo. (D) Posterior muscle segments of a PBS-injected sapje/dmd mutant embryo. Note that certain muscle segments in the anterior and the posterior have foci of muscle damage. (E) Anterior muscle segments of an IAV-injected wild-type sibling embryo. (F) Posterior muscle segments of an IAV-injected wild-type sibling embryo. Note the damaged fibers in the anterior muscle segments. (G) Anterior muscle segments of an IAV-injected sapje/dmd mutant embryo. (H) Posterior muscle segments of an IAV-injected sapje/dmd mutant embryo. Note the presence of damaged fibers in every imaged muscle segment of this embryo. (I) Quantification of the average number of muscle segments with damaged fibers per embryo at 24 hpi. Note that the prevalence of fiber damage in IAV-infected sapje/dmd mutants is more than would be predicted from adding the prevalence of damaged fibers of IAV-infected zebrafish and sapje/dmd mutants together. (J) Plot tracking survival for 5 days post injection. All PBS-injected wild-type siblings and sapje/dmd mutants lived for 5 dpi (blue lines). Most mortalities were observed in IAV-infected wild-type siblings between 4 and 5 dpi (green line). IAV-infected sapje/dmd mutants succumbed to the infection earlier than their wild-type siblings with more mortalities occurring on the first and second days post infection. Survival curves from individual experiments representative of three replicates.


Here, we investigated the effects of an infectious disease on skeletal muscle tissue alone and in combination with a genetic muscle disease. We found that human IAV can infect zebrafish muscle fibers and cause fiber damage via loss of sarcolemma integrity and/or loss of ECM adhesion external to the sarcolemma. Additionally, we showed that molecular and cellular markers of inflammation are present in muscle tissue in response to IAV infection. Finally, we showed that an infectious disease in combination with a genetic muscle disease greatly worsens the severity of muscle tissue degeneration. Taken together, our results show that gene-environment interactions are important regulators of muscle tissue structure, function, and health.

We used a model in which transparent embryos/larvae can be infected with a virus that when translated by host cells generates a fluorescent product. This allows for the visualization and tracking of infected cells in vivo and in real time. This model could be utilized to extend our findings in ways such as testing the effect of aqueous chemicals on the tropism of IAV for skeletal muscle fibers as well as determining the subcellular localization of GFP puncta in NS1-GFP-infected muscle cells based upon co-localization with fluorescent markers for subcellular organelles. Data from experiments such as these could be used to inform and develop better therapeutic options for preventing and treating skeletal muscle infection by IAV. Inflammation of the heart tissue is another important non-pulmonary complication of IAV infection. Zebrafish infected with IAV display a swollen pericardium, suggesting evidence of viral myocarditis. This model is uniquely suited for the study of viral myocarditis because zebrafish embryos/larvae don’t require heart function due to their small size and aqueous environment where oxygen is freely diffusible for the first 7 dpf37. Therefore, zebrafish research could provide insights into the mechanisms underlying viral myocarditis and potential treatments for this condition, which is fatal and thus impossible to study in vivo in mammalian models of IAV infection.

We provide evidence that one etiology of muscle fiber death in IAV-infected zebrafish is loss of sarcolemma integrity, similar to dystrophinopathies. This suggests that cytoskeletal disruption may contribute to muscle degeneration upon IAV infection. Influenza virus has been found to associate with and induce changes to the plasma membrane-associated cytoskeleton in infected chick embryo cells38. IAV proteins NP and M1 were found to bind to host cell cytoskeletal elements39,40; and, using superresolution or FRET microscopy, the IAV HA protein was recently observed to be localized to actin-rich membrane regions in infected cells41,42. Increased actin polymerization at the plasma membrane of infected cells promotes proper assembly and budding of IAV43,44. The binding of Influenza viral proteins to microfilaments and the reorganization of the cytoskeleton in infected cells, which seems to serve the virus, might be particularly detrimental to skeletal muscle fibers given the critical role of stabilized cytoskeletal-ECM linkages for muscle structure and function. The IAV-induced cytoskeletal disruptions and increased permeability of muscle fibers suggested by our EBD data may be mediated by p38MAPK, Rho/ROCK, and PKC pathways, which have been shown to be involved in changes to the cytoskeleton and permeability in IAV-infected pulmonary microvascular endothelial cells45.

A second etiology of fiber death that was found to occur due to IAV infection in zebrafish skeletal muscle tissue is loss of adhesion to the ECM external to the sarcolemma. This could be due to remodeling of the muscle tissue ECM by inflammation. ECM remodeling is achieved via a family of proteinases called matrix metalloproteinases (MMPs). In the lungs of IAV-infected mice, MT1-MMP was found to remodel collagen and blocking MT1-MMP protected infected mouse lungs from tissue damage46. Given that laminin-211 is a principle muscle ECM component in zebrafish at the developmental stage of our experiments28,47, it would be interesting to determine the effect of IAV-induced inflammation on the expression and activity of laminin-211-degrading MMPs as well as laminin-211 architecture and binding interactions. Altogether, both IAV- and host-initiated remodeling of the cytoskeleton and the muscle ECM could be mechanisms underlying the muscle fiber damage that we observed to occur in IAV-infected zebrafish.

Finally, our examination of IAV-induced muscle degeneration in vivo suggests that muscle complications of IAV are likely caused by a combination of direct infection and inflammation. We detected cellular and molecular markers of inflammation in IAV-infected zebrafish muscle tissue. The inability to consistently detect immune cell infiltration of muscle in IAV-infected humans may be due to varying biopsy locations within and between muscle groups as well as non-standardized timing of biopsies with regards to symptom onset/resolution and infection time course. Our results suggest that anti-Influenza vaccines and other precautionary measures to prevent this infectious disease and the activation of the inflammatory immune response in people (especially those with genetic muscle diseases) is a very important endeavor not just to protect against pulmonary complications of IAV infection, but skeletal muscle damage as well. If IAV is contracted, prompt treatment to reduce muscle tissue inflammation could protect against increased muscle fiber permeability, detachment from the ECM, and fiber death.

Corresponding Author

Clarissa Henry

Data Availability

Our data is up on figshare with the accession number 10.6084/m9.figshare.5499958 <>. The link is


This work was supported by the National Institutes of Health Grant RO1GM087308 ( to C.H.K., the March of Dimes Award number #1-FY14-284 ( to C.A.H., and the National Institute of Child Health and Human Development Award numbers 5RO3HD077545 and R15HD088217 ( to C.A.H. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing Interests statement

The authors have declared that no competing interests exist.

Can Quantitative Muscle Strength and Functional Motor Ability Differentiate the Influence of Age and Corticosteroids in Ambulatory Boys with Duchenne Muscular Dystrophy? Fri, 08 Jul 2016 11:20:49 +0000 Background: In the absence of a curative treatment for Duchenne Muscular Dystrophy (DMD), corticosteroid therapy (prednisone, deflazacort) has been adopted as the standard of care, as it slows the progression of muscle weakness and enables longer retention of functional mobility. The ongoing development of novel pharmacological agents that target the genetic defect underlying DMD offer hope for a significant alteration in disease progression; however, substantiation of therapeutic efficacy has proved challenging. Identifying functional outcomes sensitive to the early, subtle changes in muscle function has confounded clinical trials. Additionally, the alterations in disease progression secondary to corticosteroid therapy are not well described making it difficult to ascertain the benefits of novel agents, often taken concurrently with corticosteroids.

Objective: The purpose of this study was to examine outcome responsiveness to corticosteroid therapy and age at the onset of a natural history study of ambulatory boys with DMD.

Methods: Eighty-five ambulatory boys with DMD (mean age 93 mo, range 49 to 180 mo) were recruited into this study. Fifty participants were on corticosteroid therapy, while 33 were corticosteroid naïve at the baseline assessment. Within each treatment group boys were divided in two age groups, 4 to 7 years and 8 and greater years of age. The Biodex System 3 Pro isokinetic dynamometer was used to assess muscle strength. Motor skills were assessed using the upper two dimensions (standing/walking, running & jumping) of the Gross Motor Function Measure (GMFM 88) and Timed Motor Tests (TMTs) (10-meter run, sit to stand, supine to stand, climb 4-stairs). Two way analysis of variance and Pearson correlations were used for analysis.

Results: A main effect for age was seen in select lower extremity muscle groups (hip flexors, knee extensors and ankle dorsiflexors), standing dimension skills, and all TMTs with significantly greater weakness and loss of motor skill ability seen in the older age group regardless of treatment group. Interaction effects were seen for the walking, running, and jumping dimension of the GMFM with the naïve boys scoring higher in the younger group and boys on corticosteroid therapy scoring higher in the older group. The TMT of climb 4-stairs demonstrated a significant treatment effect with the boys on corticosteroid therapy climbing stairs faster than those who were naïve, regardless of age. Examination of individual items within the upper level GMFM dimensions revealed select motor skills are more informative of disease progression than others; indicating their potential to be sensitive indicators of alterations in disease progression and intervention efficacy. Analysis of the relationship between muscle group strength and motor skill performance revealed differences in use patterns in the corticosteroid versus naïve boys.

Conclusion: Significant muscle weakness is apparent in young boys with DMD regardless of corticosteroid treatment; however, older boys on corticosteroid therapy tend to have greater retention of muscle strength and motor skill ability than those who are naive. Quantification of muscle strength via isokinetic dynamometry is feasible and sensitive to the variable rates of disease progression in lower extremity muscle groups, but possibly most informative are the subtle changes in the performance characteristics of select motor skills. Further analysis of longitudinal data from this study will explore the influence of corticosteroid therapy on muscle strength and further clarify its impact on motor performance.



Duchenne muscular dystrophy (DMD), the most common neuromuscular disease of childhood, is characterized by a progressive myopathy that results in muscle wasting, weakness, and a loss of ambulation. In the absence of a curative treatment, corticosteroid therapy (prednisone, deflazacort) has been adopted as the standard of care, despite the adverse side effects seen in some individuals. Corticosteroids preserve cardiac and pulmonary function and enable the retention of motor skills and walking for an additional few years. 1 Age and disease progression are often considered when determining the optimal time to initiate corticosteroid therapy, resulting in variability in the age of onset and dosing regiments. 2,3

Recent advancements in the development of novel pharmacological agents that target the genetic defect underlying DMD, offer hope for a significant alteration in disease progression and ultimately lifespan. The primary focus of intervention in this population is the preservation of muscle function, and with therapeutic efficacy dependent upon muscle integrity at the onset, younger boys have been identified as the best candidates for clinical trials. However, the inherent heterogeneity of disease progression, the added confounder of normal maturation, and a limited repertoire of clinical outcome measures have challenged substantiation of intervention efficacy in this population. With the dramatic increase in the number of treatment agents in development for DMD and the need to differentiate the therapeutic efficacy of these therapies, the psychometric properties of clinical outcome measures have received greater scrutiny, and new disease specific outcome measures have been developed.4,5 Increasingly, evidence suggests that outcome measures targeting specific areas of interest or a carefully selected battery of outcomes are most effectively in identifying and differentiating therapeutic efficacy. 4

The purpose of this study was to characterize how corticosteroid therapy alters the natural history of boys with DMD using a battery of outcome measures selected to define patterns of muscle weakness, their relationship to function, and the features predictive of loss of ambulation in a large cohort of boys with DMD using a prospective multi-site study protocol. Since a number of the newer pharmacological agents are taken in addition to corticosteroids, it is important that we have a sound understanding of how disease progression in DMD has been altered by corticosteroids, within a multi-dimensional framework, in order to be able to determine the efficacy of newer agents. In this paper, a cross-sectional analysis of baseline assessments of quantitative muscle strength, functional motor skill ability, and timed motor tests were examined to determine measurement responsiveness to differences among age and treatment groups as well as the relationships among the outcome measures.

Materials and Methods

Eighty-five ambulatory boys with DMD (mean age 93 mo, range 49 to 180 mo) were recruited into this multi-site study, thirty-two from Shriners Hospital for Children, Portland, OR, thirty-three from Shriners Hospital for Children, Sacramento, and twenty from University of California, Los Angeles, California from 2006-2014. Two boys were removed from analysis due to later reclassification as probable cases of Becker muscular dystrophy. Of the 83 participants, 50 participants were on corticosteroid therapy (mean age 104 mo, range 53 to 180 mo), while 33 were corticosteroid naïve (mean age 77 mo, range 49 to 153 mo) (Table 1) at the baseline assessment. Participants were divided into two age groups, 4-7 and ≥ 8 years of age, within each treatment group (corticosteroid and naïve). Age groups were selected based on historical trend in the literature for the outcomes of interest to enable comparison of study results. Within the naïve group a significant number of boys were of a younger age than in the corticosteroid group (Table 1). All participants on corticosteroid therapy for ≥1 month were allocated to the corticosteroid group due to documentation of the early therapeutic effects of corticosteroid therapy on muscle tissue using MRI/MRS in boys with DMD.6 Boys on corticosteroid therapy had been on therapy a mean of 29.3 mo, range 1-150 months. Ten percent (n=5; ages 53, 74, 84, 90, 97 months) had been on corticosteroid therapy for less than 4 months. Whether boys were on corticosteroid therapy was dependent upon the cooperative decision of the local site physician caring for the patient and family. Study participation did not influence this decision-making process. All study participants/parents signed informed assents and consents approved by local Institutional Review Boards (Oregon Health Sciences University, University of California, Davis, and University of California, Los Angeles). Study inclusion criteria included: 1) a diagnosis of DMD as determined by clinical evaluation, family history, genetic testing, 2) male, 3) four years of age or older, 4) ability to walk independently at self-selected speed for 10 minutes, 5) ability to cognitively understand directions for testing procedures. Assessments were completed in one three-hour visit when feasible. If this schedule was too arduous for the child, assessments were performed over two days. All assessment protocols and procedures were standardized across centers. Efforts were made to hold assessing clinicians constant at each center.


Table 1: Descriptive Statistics


Volitional Muscle Strength

Volitional muscle strength was assessed using the Biodex System 3 Pro isokinetic dynamometer. Select muscle groups were assessed unilaterally (based on hand dominance) as prior investigation has reported comparable strength bilaterally in boys with DMD.7,8 Isometric hip flexor strength was assessed in supine at 45° hip flexion, while extensor strength was assessed at 85° hip flexion. The Biodex pediatric knee attachment was used for assessing strength about the knee. Isometric knee flexor strength was assessed in sitting at 30° knee flexion, while extensor strength was assessed at 90° knee flexion. The isometric protocol about the hip and knee consisted of three five-second contractions performed consecutively by each muscle group with 10 second rests between contractions. Isokinetic concentric knee flexor and extensor strength was assessed at 60 °/sec, with the subject completing three consecutive arcs of knee extension and flexion through their full volitional range. Isometric ankle dorsiflexion and plantarflexion were assessed in semi-reclined sitting with the ankle in 0-5° of plantarflexion. The isometric protocol about the ankle consisted of three five-second contractions, alternating plantarflexion and dorsiflexion, with 10 second rest between each contraction. The order of muscle strength testing was held constant with strength about the ankle assessed first, followed by knee, and lastly the hip. The peak torque output for each muscle group and contraction type was used in the analysis. All torque values (Nm) were normalized by body weight (kg).

Functional Motor Skills

The skills included in the Standing (13 items) and Walk/Run/Jump (24 items) dimensions of the Gross Motor Function Measure (GMFM-88) were used in this study to assess motor skill ability in boys with DMD. While the GMFM has gained acceptance for its use in children with cerebral palsy and traumatic brain injury (TBI), it was originally developed in response to the paucity of measures in the field of developmental disabilities that demonstrated responsiveness to change in gross motor function over time. The gross motor skills included in the GMFM are based on normal development to capture the natural influence of maturation on motor skill development and the resultant alterations in skill acquisition due to motor deficits.9 All skills are typically mastered by five years of age. The GMFM has been shown to be valid, reliable and responsive in children with Down syndrome and spinal muscular atrophy, thus it is reasonable to consider the GMFM as an indicator of changes in muscle function in DMD.9,10,11,12 Although delayed, boys with DMD can demonstrate maturational motor skill acquisition in their younger years, prior to motor skill plateau and the subsequent onset of regression due to progressive muscle weakness; therefore, motor skill assessment can be informative. The GMFM is presently listed on the National Institute of Neurological Disorders and Stroke (NINDS), Common Data Elements (CDE) site ( as a recommended instrument for use in DMD. In this study dimension point score sums were used in the analysis.

Timed Motor Tests

Four timed motor tests (TMTs) were assessed in this study including; assuming standing from supine on the floor, assuming standing from sitting on a bench (hips and knees at 90 degrees), climbing four standard stairs with assist of a railing, and ten-meter run on a level surface. Subjects were instructed to complete all measures as quickly as possible and were timed in seconds.


Two-way analysis of variance (ANOVA), with treatment group (corticosteroid/naïve) and age (4-7 years, ≥ 8 years), was used to determine whether significant differences in the outcome variables of interest were revealed for the main effects of treatment and age, and whether interaction effects were present. Pearson product moment correlations were used to examine the relationships between outcome variables. Significance was set at p=.05.


Volitional Muscle Strength

No significant main effect for treatment group (corticosteroid therapy versus naïve) was seen in muscle strength; however, a main effect for age (4-7 years versus ≥ 8 years) was seen in select muscle groups. Isometric hip flexor strength (p=.037)(Figure 1), isometric and isokinetic concentric knee extensor strength (p=.017, p=.001)(Figures 2-3) and isometric ankle dorsiflexor strength (p=.024)(Figure 4) all demonstrated a significant decrement with age regardless of treatment group. No significant interaction effects were seen.

Hip strength

Fig. 1: Hip isometric muscle strength (group means)

Knee strength isometric

Fig. 2: Knee isometric strength (group means)

Knee strength con

Fig. 3: Knee isokinetic concentric muscle strength (group means)

Ankle strength

Fig. 4: Ankle isometric muscle strength (group means)

Functional Motor Skills

No significant main effect for treatment group (corticosteroid therapy versus naïve) was found in Standing dimension skills; however, a main effect for age (p=.011) was seen with lower standing scores in the older age group. A significant interaction effect was seen for Walking/Running/Jumping dimension skills (p=.046), with scores reversing across age. The higher scores in the naïve group when 4-7 years of age indicates a less advanced disease process and probably delay in initiation of corticosteroid therapy; conversely, the lower scores in the naïve group when ≥ 8 years may indicate greater severity of disease effect without the influence of corticosteroid therapy (Figure 5). In an attempt to identify the motor skills that are potentially more sensitive to incremental changes in muscle strength each individual Standing and Walking/Running/Jumping skill was analyzed.


Fig. 5: GMFM Standing and Walking, Running, & Jumping dimensions (group means)

Standing Skills

Of the standing skills, six were not discriminatory in this study due to the inclusion criteria of this study (able to walk 10 minutes). The seven Standing skills that involved position transition demonstrated a significant main effect for age, with lower scores seen in the older age groups (Table 2). A significant interaction effect was seen for the skill of ‘lowering to the floor with control’ (p=.004) as the naïve group scored higher than the corticosteroid group when 4-7 years of age; however, the corticosteroid group demonstrated greater ability when ≥ 8 years.


Table 2: GMFM Standing dimension skill items (group means, standard deviation, p-values)

Walking, Running & Jumping Skills

Similar to the finding that simple stance skills were not discriminatory, simple walking and running skills did not differentiate among age or treatment groups. Complex ambulatory skills (walking while carrying a large object with both arms, walking within an 8-inch pathway, climbing stairs with use of a railing) and jumping skills (jumping ‘up’ or ‘forward’ with two feet simultaneously) did demonstrate a main effect for age with skill decrement across age (Table 3). An interaction effect was seen for obstacle navigation while walking (stepping over a stick (p=.01) and jumping down (jumps off 6 inch step with both feet simultaneously (p=.036), as the naïve group scored higher than the corticosteroid group when 4-7 years of age; however, the corticosteroid group demonstrated greater ability when ≥ 8 years. ‘Hopping on one foot’ and ‘walks up 4 steps/walks down 4 steps alternating feet’ (without use of a railing), the most difficult skills determined by Rasch analysis of GMFM, did not demonstrate a main effect for age or treatment group. This finding suggests these skills may be informative early indicators of muscle strength in boys with DMD as only 20-30% of the steroid naïve group and 30-40% of the corticosteroid group were able to perform these skills.

WRJ skills

Table 3: GMFM Walking, Running, & Jumping dimension skill items (group means, standard deviation, p-values)

Relationship between Volitional Strength and Functional Motor Skills

Analysis of the relationships between lower extremity muscle strength and functional motor skills revealed significant fair (.25-.50) to moderate (.50-.75)13 positive correlations indicating that greater muscle strength was related to a greater ability to perform functional motor skills (Tables 4, 5). Significant relationships between muscle group strength and functional skills were more prevalent in the corticosteroid group than the naïve group. Standing skills correlated with hip extensor/flexor and knee extensor (isometric and isokinetic concentric) strength for corticosteroid group, while only hip and knee extensor (isokinetic concentric) strength were significant for the naïve group (Table 4, 5). During Walking, Running & Jumping skills both the extensor/flexor strength about the hip and knee correlated significantly for the corticosteroid group, while only knee extensor strength (isometric and isokinetic concentric) correlated for the naïve group.

Correlation CS

Table 4: Corticosteroid group: Correlations for muscle strength, functional motor skills and TMTs (Pearson r and p-values)


Table 5: Corticosteroid Naïve: Correlations for muscle strength, functional motor skills and TMTs (Pearson r and p values)

Timed Motor Tests

The only TMT to demonstrate a main effect for treatment was ‘climb 4 stairs’ (p=.042) as the corticosteroid group demonstrated significantly greater ability at > 8 years of age than the naïve group. A main effect for age was found for all TMTs; ‘run 10 meters’ p=.043, ‘supine to stand’ p=.026, ‘sit to stand’ p=.001, and ‘climb 4 stairs’ p=.002, with the older age group taking significantly longer to perform each TMT (Figure 6).


Fig. 6: Timed Motor Tests (group means)

Relationship between Volitional Strength and Timed Motor Tests

As in the functional motor skills, significant relationships between muscle group strength and TMT were more prevalent in the corticosteroid group than the naïve group. All correlations between muscle strength and TMT were negative indicating that as muscle strength decreased the time to perform the TMT increased. In the corticosteroid group ‘run 10 meters’ demonstrated a significant negative correlation with force production by hip and knee extensors/flexors and ankle plantar flexors; ‘stand from supine’ and ‘climb 4 stairs’ demonstrated a significant negative correlation with strength of hip flexors, knee extensors, and ankle plantar flexors; and ‘stand from sit’ correlated only with knee extensors (Table 4). The naïve group demonstrated significant correlations between force produced by knee extensors (isokinetic concentric) and ankle dorsiflexors and ‘sit to stand’, while only knee extensors (isokinetic concentric) correlated with ‘climb 4 stairs’ (Table 5).


Volitional Muscle Strength

While quantitative methods of measuring strength have become more prevalent, they remain limited, and thus the relationship between the muscle group strength and functional mobility remains difficult to ascertain. Reliable methods of monitoring alterations in muscle strength are vital to understanding the progression of DMD, albeit, volitional strength measurement is dependent upon the child’s ability to reliably produce maximal efforts. Hand held dynamometers (HHD) are common in the clinical environment due to their low cost, portability, and universality; however, isokinetic dynamometers (ie Biodex System) provide a more psychometrically sound assessment of volitional strength as well as real-time visual feedback during force production, which assists in motivating and reinforcing maximal effort in children. Presently, studies that have used isokinetic dynamometers for strength assessment in DMD boys are limited, thus the sensitivity and specificity of the method of quantitative measurement has not been established and standardized in this population; however, emerging evidence is encouraging.14

In a natural history study McDonald et al (1995) assessed quantitative muscle strength about the knee using isokinetic dynamometry in boys with DMD 6-12 years of age (treatment regime unspecified) and age-matched controls. By six years of age boys with DMD demonstrated an isometric knee extensor/flexor strength of 50%/32% of controls, and an isokinetic concentric knee extensor/flexor strength of 22%/28%, respectively. A relatively linear decrement in strength from 6-12 years of age, with the rate of decline in greater in extensor than flexor muscle groups, was seen in DMD boys.7,15 A recent analysis of strength using isokinetic dynamometry in boys with DMD of similar age (5.6-12 years of age) (71% on corticosteroid therapy) reported isometric knee extensor/flexor strength to be 18%/40% of controls, and isokinetic concentric knee extensor/flexor strength to be 21%/39%. Longitudinal analysis revealed boys with DMD had an increase in isometric and isokinetic concentric knee extensor/flexor strength when <7.5 years of age; conversely, decrements were seen in these muscles in the boys over 7.5 years of age, with greater decline after 9 years of age.14 While the variation in strength reported in boys with DMD in the studies above could be attributed to the heterogeneous nature of DMD or differences in treatment regimes, it should be noted that variability in muscle strength is typical in normal development. With normally developing children 4-12 years of age strength variability is high within each age group and increases with age.14,15 The notable finding in boys with DMD is that muscle weakness is significant early in the disease process, and despite the reported gains in strength in boys with DMD <7.5 years, their strength profile diverges from that of normal children 4-12 years of age who demonstrate a gradual and almost linear increase in muscle strength, regardless of gender, until the onset of puberty.

This study supports the above noted studies in that significant weakness was seen in the younger boys (4-7 years) relative to age-matched norms, and the decline in muscle strength across age was significant. Comparison of isometric and isokinetic concentric knee extensor/flexor strength of DMD boys, 4-7 years of age, in this study with a normative age-matched data16,17 revealed that the corticosteroid group had 45%/48% of isometric and 39%/36% of isokinetic concentric knee extensor/flexor strength of controls, respectively; while the corticosteroid naïve group had 43%/41% of isometric and 38%/30% of isokinetic concentric knee extensor/flexor strength. At ≥ 8 years of age the corticosteroid group demonstrated 17%/18% of isometric and 19%/21% of isokinetic concentric knee extensor/flexor strength of controls, respectively, while the naïve group had 14%/13% of isometric and 15%/15% of isokinetic concentric knee extensor/flexor of controls. Although significant differences in the strength of the corticosteroid and naïve groups were not found in this study, the boys on corticosteroid therapy did tend to be slightly stronger. Recent examination of the impact of corticosteroid therapy on muscle tissue using magnetic resonance imaging (MRI) and magnetic resonance spectroscopy (MRS) in combination with isokinetic dynamometry6 in boys with DMD (5-6.9 years of age) revealed a suppression of muscle inflammation and fatty tissue infiltration in the muscles studied. While tissue changes were associated with significantly greater isometric knee extensor strength in the corticosteroid group than an age-matched naïve group, tissue changes in the soleus were not associated with significant strength differences, indicating the need for further investigation of the relationship between tissue changes and volitional strength.

While isokinetic dynamometry has been shown to be a sensitive and responsive method of measuring muscle strength in boys with DMD14 standardization of testing procedures (i.e. data normalization, testing positions and velocities) is needed to allow for valid comparison of results across studies. Future studies combining the use of isokinetic dynamometry, MRI, and select functional outcomes have the potential to greatly enhance our understanding of the primary and compensatory changes in muscle function characteristic of DMD and to expedite the identification of therapeutic agents that alter muscle integrity and function.

Functional Motor Skills

Boys with DMD are known to lose functional motor skill capability with progressive muscle deterioration and ultimately loss of ambulation; however, the pattern and rate of skill loss is variable. Methodological review of the DMD literature has suggested that motor scales that assess a wide range of abilities may be more informative in characterizing different phases of the disease process, whereas disease-specific motor tests might be more appropriate for short term efficacy studies.18 Due to the functional ability of the boys at the onset of this study and the interest in the exploring ambulatory skill performance, only the Standing and Walking, Running & Jumping skill dimensions of the GMFM-88 were used. The preliminary findings of this study demonstrate that these dimensions were sensitive to changes in gross motor function across age in boys with DMD; additionally, the Walking, Running, & Jumping dimension was able to discriminate differences in skill ability between the corticosteroid and corticosteroid naïve groups.

Standing Skills

In ambulatory boys with DMD skills that involved transitions in and out of standing were the items discriminatory across age in this ambulatory cohort, with greater skill loss seen in the older boys with more advanced disease. ‘Assuming stand through half-kneel’ and ‘lowering to the floor from standing’, which are not usually seen in disease-specific motor scales for DMD, appear to be informative indicators of disease progression and possibly therapeutic efficacy. Lowering to the floor from standing requires generation of a controlled eccentric contraction of the extensor muscles of the hip, knee and ankle, the greatest force a muscle can produce, to control the lowering of the center of mass. This skill was sensitive to age and treatment in this study, and thus may be a sensitive early indicator of disease progression and intervention efficacy.

Walking, Running & Jumping Skills

Simple walking activities were not discriminatory in this cohort, however, walking items that included additional tasks or constraints were better indicators of disease status and the possible mitigating influence of corticosteroid therapy. Prior investigation has reported that the progression of muscle damage in DMD is workload dependent, and thus can be seen first in movements of the whole body against gravity, with jumping and running (which require both feet to be off the ground simultaneously) being the first skills to fail.19 While running skills were less responsive to early disease progression in this cohort, jumping skills were already significantly impaired in the most of the boys, with hopping skills never attained in the majority of the boys in this study. Hopping skills normally emerge at three years of age, with most able-bodied children able to hop at least once. By five years of age mastery is achieved with hopping ten times on one foot easily attained. One legged hopping is considered to be the most advanced jumping skill and has been shown to be strongly related to age and thigh muscle strength20 , which is known to be affected early in boys with DMD, thus is a early indicator of gains or losses in strength. Descending and ascending 4 stairs (no rail), per Rasch analysis of the GMFM, is second in difficultly, preceded by one-legged hopping. Most of the boys with DMD in this study were unable to stair walk without rail use.

Since the onset of this study a variety of disease-specific functional scales that assess different ranges of motor abilities have been developed and validated for DMD, including the Motor Function Measure (MFM)21 and the North Star Ambulatory Assessment (NSAA),5,22 thus the repertoire of appropriate outcome measures for DMD is expanding. The NSAA was developed specifically to identify functional change and therapeutic efficacy in ambulatory boys with DMD and is used in a number of clinical trials.5,14,22,23 Eleven of the 17 NSAA skills are equivalent to skills assessed within the Standing and Walking, Running & Jumping Dimensions of the GMFM. Differences are seen between the scales in how the motor skill is graded with the GMFM stratifying responses over a 4-point scale, while the NSAA uses a 3-point scale. Further investigation of the various dimensions of the GMFM is needed to determine its sensitivity to change in this population; however, its skill range, specificity within dimension, and scoring methodology may be more discriminatory than other assessments and thus more likely to detect incremental change in motor skill performance in boys with DMD.9

Timed Motor Tests

Timed motor tests (TMTs) such as the 10 meter walk, rising from floor, rising from chair, and walking up 4 steps have become the standard approach to determining peak motor performance.8,24,25 The use of TMTs in boys with DMD has become an accepted method for motor function staging and predicting clinically meaningful endpoints in motor function, such as loss of ambulation, which is informative to patients and their families.26,27,28 The benefits of TMTs are proposed to be ease of use and responsiveness to change29,30,31 , and the ratio data generated that allows for parametric analysis; however, the reliability of TMTs has been questioned, as the potential for random error in some measures (stand from sit) is significant.24 In this study TMTs were sensitive to disease progression with age, but only the time to ‘climb 4 stairs’ indicated a difference in ability between the corticosteroid and naïve groups. Further study is indicated to determine if TMTs can reliably detect the early, subtle changes in muscle function induced by newly introduced therapeutic agents.

Relationship between muscle strength and motor skills

A relationship between muscle strength and functional motor skill ability is known in DMD, but not well understood. Muscle weakness in DMD has been reported to have a proximal to distal progression, with involvement of large proximal hip and shoulder girdle muscle noted first followed by distal extremities and the trunk.32 Various muscle groups have been reported to be predictive of ambulatory loss;19,33 however, recent studies of muscle tissue changes using MRI revealed considerable variation in disease progression across thigh muscles, and within select muscles within muscle groups.19,32 Thus, it is probable that the contribution of various muscle groups to the eventual loss of ambulatory skills may be more variable than appreciated within this population. A recent study of boys with Becker muscular dystrophy suggested that functional motor skills are muscle group dependent, and the patterns of change are reflective of a given disease progression, thus clinical study designs should select outcomes relevant to disease progression when evaluating intervention efficacy.3 In this study the relationships between muscle group strength and motor function varied based on the motor ability being measured and treatment group. While the steroid group demonstrated relationships between the strength of various muscle groups and functional motor skills (GMFM, TMTs), these relationships were dramatically reduced in the naïve group. Whether this finding is indicative of the various compensatory strategies employed as lower extremity weakness progresses, or other contributing factors warrant further investigation.


The baseline data analysis of this natural history study indicates that the outcomes measures utilized in this study were sensitive to the age related differences in strength and motor function that are characteristic of disease progression boys with DMD; however treatment effects were less likely to be identified. These findings reflect the difficulty inherent in obtaining the statistical power needed to substantiate intervention efficacy in the small, heterogeneous samples sizes that are characteristic of DMD clinical studies. Isokinetic dynamometry revealed variability in the muscles affected, which has been corroborated with magnetic resonance imaging (MRI) of thigh muscle tissue,32 with significant muscle weakness present in lower extremity muscle groups at an early age. Most informative was the sensitivity of the GMFM dimensions to age and disease related changes in muscle strength and resultant motor skill performance in this cohort, specifically the select motor skills that offer the potential to be sensitive indicators of early disease progression, and thus probable indicators of intervention efficacy. The differences seen in the relationship between muscle group strength and motor skill performance for boys on corticosteroid therapy versus those who are naïve indicates further study is needed to determine the compensatory strategies (i.e. muscular, postural, etc.) employed to preserve functional ability with decrements in muscle strength. Analysis of the longitudinal data of this study will further delineate the psychometric attributes of these outcome measures and advance evidence of how corticosteroid therapy alters the natural history of disease progression in DMD.

Competing Interests

We have read the journals policy and we have the following conflicts:

Susan Sienko, Research Associate: Shriners Hospitals for Children. “Biomechanical Analysis of Gait in Individuals with Duchenne Muscular Dystrophy” with Sussman, (PI), 1/06 – 12/15.

Cathleen E Buckon, Research Associate: Shriners Hospitals for Children. “Biomechanical Analysis of Gait in Individuals with Duchenne Muscular Dystrophy” with Sussman, (PI), 1/06 – 12/15.

Eileen Fowler, Local Principal Investigator (UCLA): Shriners Hospitals for Children. “Biomechanical Analysis of Gait in Individuals with Duchenne Muscular Dystrophy” with Sussman, (PI), 1/10 – 12/15.

Anita Bagley, Local Principal Investigator (NCal): Shriners Hospitals for Children. “Biomechanical Analysis of Gait in Individuals with Duchenne Muscular Dystrophy” with Sussman, (PI), 1/06 – 12/15.

Loretta Staudt, Research Associate (UCLA): Shriners Hospitals for Children. “Biomechanical Analysis of Gait in Individuals with Duchenne Muscular Dystrophy” with Sussman, (PI), 1/10 – 12/15.

Mitell Sison-Williamson, Research Associate (NCal): Shriners Hospitals for Children. “Biomechanical Analysis of Gait in Individuals with Duchenne Muscular Dystrophy” with Sussman, (PI), 1/06 – 12/15.

Kathy Zebracki: None.

Craig M McDonald, Local CoI: (NCal): Shriners Hospitals for Children. “Biomechanical Analysis of Gait in Individuals with Duchenne Muscular Dystrophy” with Sussman, (PI), 1/06 – 12/15.

Michael D Sussman, Principal Investigator: Shriners Hospitals for Children. “Biomechanical Analysis of Gait in Individuals with Duchenne Muscular Dystrophy”, 1/06 – 12/15.

]]> 0