Resident muscle stem cell myogenic characteristics in postnatal muscle growth impairments in children with cerebral palsy

Abstract

Children with cerebral palsy (CP), a perinatal brain alteration, have impaired postnatal muscle growth, with some muscles developing contractures. Functionally, children are either able to walk or primarily use wheelchairs. Satellite cells are muscle stem cells (MuSCs) required for postnatal development and source of myonuclei. Only MuSC abundance has been previously reported in contractured muscles, with myogenic characteristics assessed only in vitro. We investigated whether MuSC myogenic, myonuclear, and myofiber characteristics in situ differ between contractured and noncontractured muscles, across functional levels, and compared with typically developing (TD) children with musculoskeletal injury. Open muscle biopsies were obtained from 36 children (30 CP, 6 TD) during surgery; contracture correction for adductors or gastrocnemius, or from vastus lateralis [bony surgery in CP, anterior cruciate ligament (ACL) repair in TD]. Muscle cross sections were immunohistochemically labeled for MuSC abundance, activation, proliferation, nuclei, myofiber borders, type-1 fibers, and collagen content in serial sections. Although MuSC abundance was greater in contractured muscles, primarily in type-1 fibers, their myogenic characteristics (activation, proliferation) were lower compared with noncontractured muscles. Overall, MuSC abundance, activation, and proliferation appear to be associated with collagen content. Myonuclear number was similar between all muscles, but only in contractured muscles were there associations between myonuclear number, MuSC abundance, and fiber cross-sectional area. Puzzlingly, MuSC characteristics were similar between ambulatory and nonambulatory children. Noncontractured muscles in children with CP had a lower MuSC abundance compared with TD-ACL injured children, but similar myogenic characteristics. Contractured muscles may have an intrinsic deficiency in developmental progression for postnatal MuSC pool establishment, needed for lifelong efficient growth and repair.

INTRODUCTION

Cerebral palsy (CP) is the most common developmental movement disorder that affects two to three children per 1,000 live births (1, 2). Although it is primarily a nonprogressive perinatal brain alteration predominantly leading to spasticity, secondary impairments typically manifest progressively in the growing musculoskeletal system in postnatal development (2–4). CP is heterogeneous, in terms of walking ability based on the Gross Motor Function Classification System (GMFCS): some children are primarily ambulatory (I, II, and III), whereas others are minimally ambulatory (IV and V) (5), which remains fairly stable throughout childhood (6).

Skeletal muscles in children with CP have smaller muscles compared with typically developing (TD) children, secondary to impaired muscle growth from early on in life, which does not affect all muscles equally (7–13). Lower limb muscles, such as the gastrocnemius and adductors tend to be more affected in terms of limited growth (11, 12). Children with CP tend to have specific walking patterns such as toe walking (equinus gait) or crouch gait that increases the functional demand on certain muscles more than others (14). Muscle contractures develop in some muscles that limit range of motion of joints, independent of walking ability, requiring preventative rehabilitation, and in many cases eventually muscle lengthening surgeries. Muscle contractures also have alterations in extracellular matrix (ECM) components, which could contribute to their development (15–18).

Muscles are composed of contractile proteins-sarcomeres arranged along the length and girth of muscles, to facilitate force production and allow joint excursion (19). During postnatal development, growth occurs by the addition of sarcomeres along the length (20) and along the girth of muscles (21), i.e., myofibers become longer and bigger. Muscle contractures appear to have smaller myofiber cross-sectional area as compared with TD children (15, 22, 23). But this finding is inconsistent and heterogeneity with reduced growth with age has also been reported (24, 25). Contractured muscles in children with CP have overstretched sarcomeres (15, 26), and reduced number of sarcomeres along the length of the fiber (27), suggesting an overall inability to add sarcomeres during muscle growth, potentially leading to the development of contractures.

Satellite cells (muscle stem cells, MuSCs), the adult muscle stem cell population resides in their niche, between the sarcolemma of myofibers and basal lamina of the ECM. These cells are responsible for facilitating postnatal muscle growth, repair, and regeneration throughout life (28). Upon activation, MuSC exit proceed down a myogenic lineage of proliferation, differentiation, and fusion stages to facilitate myogenesis. MuSC also reestablish quiescence to maintain a mostly nonexhaustible stem cell pool throughout life (28). MuSC abundance can increase in response to exercise (29) or decrease with disuse immobilization (30). During early postnatal development, skeletal muscle myofibers accrue thousands of postmitotic myonuclei from MuSC (31–33). Recent studies show that in the postnatal period during the growth of myofibers, beyond providing myonuclei, MuSC in their ECM niche change dynamically to establish a MuSC pool, alterations of which lead to impaired growth (34–38). Overall, there is a reduction in MuSC abundance during development but there is an interaction between quiescent, activated, and proliferating cells to ensure that an appropriate MuSC pool is established by the end of the developmental period (37).

MuSC abundance, measured ex vivo by flow cytometry (39) or in situ by immunohistochemistry (25, 40), is significantly lower in some contractured muscles (hamstrings, biceps) in children with CP compared with TD children. In vitro cell culture studies, using hamstring muscles report MuSC-derived progenitor myoblasts from children with CP have greater proliferation rates, but an impaired capacity for differentiation and fusion leading to smaller myotubes, secondary to altered epigenetics (41, 42). On the other hand, in vitro results in progenitor myoblasts derived from gastrocnemius microbiopsies, expanded ex vivo, in younger children report myoblast fusion is enhanced but there is myonuclear dysregulation and impaired myotube formation (43). Furthermore, transcriptional profiles in MuSC-derived progenitor myoblasts as well as progenitor myotubes from mostly spinal muscles show significant alterations compared with children with TD (44). Overall, these cellular function and transcription studies indicate a cellular dysfunction in MuSC in muscle contractures during postnatal development.

The role of in vivo or in situ resident MuSC abundance, activation, and proliferation in contracture development in muscles of children with CP is currently lacking since in vitro studies may not entirely reflect resident muscle stem cell capacities in their niche (45). Our primary hypothesis is that muscles develop contractures due to reduced myogenic characteristics of resident muscle stem cells, either directly or secondary to alterations in ECM. We also hypothesize that resident muscle stem cells in contractured muscles are in a more primitive/altered state of postnatal muscle stem cell pool establishment. Since MuSC characteristics can be altered due to activity and some children with CP are primarily ambulatory and others primarily nonambulatory, we evaluated the effect of ambulation capacity on MuSC characteristics. In addition, we evaluated if CP muscle MuSC characteristics are similar to those from chronic injuries that occur later in postnatal development, such as following a torn anterior cruciate ligament (ACL) injury in otherwise TD children, which leads to significant muscle atrophy and alterations in MuSC (46–48). Utilizing biopsies from a large sample size of 36 children, we address this knowledge gap comprehensively by evaluating MuSC myogenic, myonuclear, and myofiber characteristics in situ with immunohistochemical labeling for MuSC abundance, activation, proliferation, myonuclei, myofiber sarcolemma, fiber type-1, and collagen content across two different contractured muscles in children with CP (adductors, gastrocnemius), and noncontractured muscle (vastus lateralis) in CP and TD. Our work here shows that although contractured muscles’ MuSC abundance may be higher they have lower myogenic potential. Our data also suggest MuSC have contributed to myonuclei accretion early on in postnatal development, but in contractured muscles MuSC pool may not be entirely established or is altered.

MATERIALS AND METHODS

Subjects and Muscle Biopsy Collection

Children with CP or TD who were undergoing lower limb surgery that allowed for open muscle biopsies to be obtained were recruited for this study. Children and/or their parents gave informed consent that was approved by the institutional review board (IRB) of Ann and Robert H. Lurie Children’s Hospital. IRB approval for the ethics of this study was in accordance with the guidelines of the 1964 Declaration of Helsinki. Open muscle biopsies (vastus lateralis, adductor, and gastrocnemius) were obtained by the surgeons in the operating room from children with CP during muscle lengthening or bony corrective surgery. Vastus lateralis muscle biopsies were obtained from children with otherwise typical development, undergoing anterior cruciate ligament reconstruction surgery using the quadriceps tendon. Thirty-six children participated in this study (CP = 30, TD = 6, CP: mean age 11.50 ± 0.63, sex: 18 males, 12 females, GMFCS: 17 I–III, 13 IV–V, TD: mean age 13.50 ± 0.76, sex: 3 males, 3 females) (Table 1). Muscle biopsies were briefly dried with a wipe, placed in a cryotube, quickly (25–30 s) snap-frozen in isopentane, appropriately chilled by liquid nitrogen, and stored at −80°C until used for immunohistochemistry (IHC).

Table 1.

Subject and quantification characteristics

	Cerebral Palsy Contracture Gastrocnemius	Cerebral Palsy Contracture Adductor	Cerebral Palsy Noncontracture Vastus Lateralis	Typically Developing ACL Injury Vastus Lateralis
Number of Subjects	9	11	10	6
Age	11.8 ± 0.9	12.5 ± 1.2	10.2 ± 0.8	13.5 ± 0.8
Sex	5 M/4 F	4 M/7 F	9 M/1 F	3 M/3 F
GMFCS, n	I–III (7), IV–V (2)	I–III (5), IV–V (6)	I–III (5), IV–V (5)	N/A
Myofiber number (per section)	1,842 ± 338	1,208 ± 223	2,395 ± 674	1,157 ± 360
Myofiber area	1.0 ± 0.0	1.0 ± 0.0	1.0 ± 0.0	1.0 ± 0.0
Myonuclear number	1.0 ± 0.0	1.0 ± 0.0	1.0 ± 0.0	1.0 ± 0.0
MuSC abundance	1.9 ± 0.1	1.9 ± 0.1	1.8 ± 0.2	2.0 ± 0.0
MuSC activation	1.0 ± 0.0	1.0 ± 0.0	1.0 ± 0.0	1.0 ± 0.0
MuSC proliferation	1.0 ± 0.0	1.7 ± 0.1	1.7 ± 0.2	1.8 ± 0.2
ECM collagen	1.0 ± 0.0	1.0 ± 0.0	1.0 ± 0.0	1.0 ± 0.0

Subject characteristics showing the muscle from which biopsies were obtained, number of subjects, average age, sex, and gross motor functional classification system (GMFCS) levels, with number per group in all contractured, noncontractured, and TD biopsies used for this study. GMFCS I indicates children who are highest functioning and V the lowest. Note: TD children do not have any GMFCS level assignment. Quantification for each protocol indicates average number of sections per subject. All data are represented as means ± SE. ACL, anterior cruciate ligament; ECM, extracellular matrix; MuSC, muscle stem cell; TD, typically developing.

Immunohistochemistry

Tissue was retrieved from the −80°C freezer, transferred to a cryostat on dry ice, and equilibrated in the cryostat at −22°C for an hour. A small piece of tissue, was cut with a chilled blade, placed in a mold with OCT (optimal cutting temperature compound), and mounted on a cryosectioning chuck onto the cryostat stage for sectioning. Muscle cross sections (10 µm thick) were cut and multiple sections (three to four) were obtained per glass slide. Slides were air-dried for an hour and placed in the −80°C freezer until utilized for IHC staining.

For IHC antibody labeling, slides were retrieved from the −80°C freezer and allowed to air dry for 1 h. Satellite cell labeling was based on a previously established protocol from the Center for Muscle Biology, University of Kentucky (49). Briefly, slides were fixed in acetone for 3 min at −20°C, and subsequently washed with PBS, endogenous peroxidases were blocked with 3% hydrogen peroxide, washed with PBS, and then blocked in 2.5% normal horse serum (NHS) (Vector Laboratories, S-2012) for 60 min. Slides were incubated with primary antibodies (details in the next paragraph and in Table 2) overnight at −4°C while rocking. Slides were then washed with PBS and incubated with biotin (as appropriate), and secondary antibodies. Slides were washed with PBS, stained with DAPI (1:10,000 of stock, Thermo Fisher Scientific, Molecular Probes, D1306) for 10 min and cover-slipped with VectaShield (Vector Laboratories, H-1000), or directly cover-slipped with VectaShield with DAPI (Vector Laboratories, H-1200).

Table 2.

Immunohistochemistry antibody specifics

	Primary Antibody				Secondary Antibody
Target	Species	Isotype	Dilution	Make	Secondary	Species	Isotype	Dilution	Make	Channel
Pax7	Mouse	IgG1	1:100	DSHB	Biotin-SP Conjugate	Goat anti-mouse	IgG1	1:500	Invitrogen
SA-HRP			1:500	Invitrogen	Superboost Tyramide reagent Alexa Fluor			1:500	Thermo Fisher	594
Dystrophin	Rabbit	IgG	1:100	Abcam	Alexa Fluor	Goat anti-rabbit	IgG (H + L)	1:250	Invitrogen	488
Ki67	Rabbit	IgG	1:200	Abcam
BA-D5	Mouse	IgG2b	1:75	DSHB	Alexa Fluor	Goat anti-mouse	IgG2b	1:250	Invitrogen	647
MyoD	Mouse	IgG2b	1:50	SCBT

Antibodies used in this study were primary and secondary fluorescent markers as indicated. Note: For MuSC abundance, Pax7 labeling was combined with a biotin-conjugated antibody, that was combined with SA-HRP and TSA superboost. DAPI was used to costain for nuclei. Ki67, cell cycle proliferation marker; BA-D5, type-1 fiber; DSHB, Developmental Studies Hybridoma Bank; MuSC, muscle stem cell; MyoD, myogenic determination factor; Pax7, satellite cell transcription factor; SA-HRP, streptavidin-horseradish peroxidase; SCBT, Santa Cruz BioTechnologies.

The primary antibodies used were as follows: anti-Pax7 (mouse, IgG1, 1:100, DSHB, concentrate), anti-Dystrophin (rabbit, IgG, 1:100, Abcam, ab15277), anti-BA-D5 (MHC1) (mouse, IgG2b, 1:75, DSHB), anti-Ki67 (rabbit, IgG, 1:200, Abcam, ab15580), and anti-myogenic determination factor (MyoD) [mouse, IgG2b, 1:50, Santa Cruz BioTechnologies (SCBT), sc-377460]. The anti-Pax7 and anti-BA-D5 developed by A. Kawakami and S. Schiaffino, respectively, were obtained from the Developmental Studies Hybridoma Bank (DSHB), created by the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) of the National Institutes of Health (NIH) and maintained at The University of Iowa, Department of Biology, Iowa City, IA. Secondary antibodies used were as follows: goat anti-mouse IgG1 biotin-SP-conjugated (1:1,000, Jackson Immuno Research, 115-065-205), Streptavidin-horseradish peroxidase (SA-HRP) (1:500, Invitrogen, S-911), SuperBoost Tyramide reagent Alexa Fluor 594 (1:500, Thermo Fisher, B40957), Alexa Fluor 488 goat anti-rabbit IgG (H + L) (1:250, Invitrogen, A-11034), and Alexa Fluor 647 goat anti-mouse IgG2b (1:250, Invitrogen, A-21242). More details on the primary and secondary antibodies are summarized in Table 2. To ensure appropriate Pax7 labeling with minimal nonspecific binding, one section in each slide was not incubated with the Pax7 antibody and acted as our negative control.

Histochemistry

To quantify extracellular matrix (ECM) content, we used a Sirius Red/Fast Green staining assay kit (Chondrex, Inc., No. 9046) for collagen/noncollagenous proteins and staining performed per the manufacturer’s instructions, with minor modifications. Briefly, slides were thawed from −80°C at room temperature, fixed in 4% glutaraldehyde for 60 s, washed with PBS, and then incubated in 0.2–0.3 mL per section of the dye solution containing Sirius Red/Fast Green for 2 h. Following incubation, slides were washed with water and then sequentially incubated for 1 min in 70%, 80%, 95%, 100% ethanol, and in xylene twice. Slides were cover-slipped in a xylene-based nonaqueous mounting media (Epredia Cytoseal XYL, No. 8312-4).

Image Acquisition and Analysis

An inverted microscope (Leica DMi8 fluorescence microscope, Mannheim, Germany) with a ×10 and/or ×20 objective was used to obtain tilescan images (image of entire cross sections) for each section, depending on the IHC protocol. We used an automated analysis software MyoVision (50) on ×10 tilescan images of Dystrophin/DAPI/BA-D5 to analyze cross-sectional area (CSA), myonuclei/fiber, and fiber type. CSA, myonuclei/fiber, and fiber types were calculated based on every viable fiber captured within the tilescan image in one section per subject (Table 1). Overall CSA, CSA by fiber type-1 and type-2 obtained from MyoVision were averaged across all fibers for one section per subject and averaged across subjects per muscle group. Frequency distribution of overall, type-1 and type-2 myofiber areas were binned in 250 µm² intervals and expressed as percentage of total number of fibers in each myofiber group in one section per subject. The frequency bins were averaged across subjects in each muscle group. Coefficient of variation of myofiber areas within one section per subject was calculated for each fiber type as a ratio of standard deviation to average values of the fibers in that section and averaged across subjects for each fiber type for each muscle group. Average myonuclear number per subject was calculated by averaging the myonuclei/fiber of all fibers that were detected to have nuclei by MyoVision in one section and averaged across subjects in each muscle group. Myonuclear domain was calculated by dividing each fiber area by its associated myonuclear number in the MyoVision-derived data, averaged across all the fibers in one section per subject, and averaged across all the subjects in each group. For Sirius Red/Fast Green histological protocol, slides were imaged in bright field as tilescans using a ×10 objective. Images were analyzed on ImageJ using the color deconvolution tool to yield %green area/total image area, and %red area/total image area. All red content was considered to be collagen protein whereas all green content was noncollagen protein. Collagen content was quantified as %red area to %green area, i.e., percent of collagen normalized to total muscle area.

Using ImageJ’s merge channel and channel tools functions, all Pax7+/DAPI+ cells along but outside the dystrophin fiber borders were quantified manually to assess MuSC abundance. Total MuSC abundance was recorded and reported as SCs/100 fibers. MyoD+/Pax7+/DAPI+ cells and Ki67+/Pax7+/DAPI+ cells were quantified similarly in ImageJ to assess activated MuSC and proliferating MuSC, respectively. Proliferating and activated SCs were expressed as percent of total SCs. For both abundance and proliferating MuSC, combined type-1 fiber labeling was used to quantify type-1 specific characteristics. The non-type-1 fibers were considered type-2 fibers and used as such. All quantification was done on one-to-three complete sections per subject per labeling protocol (abundance, activation, and proliferation), and averaged across them per subject. Details for each protocol are shown in Table 1. These values were compiled across all variables for statistical analysis, per independent group. To validate our manual quantification of MuSC with automated analysis, we reanalyzed our ×20 images using MuscleJ (51), for as many sections as technically possible. MuscleJ analysis software is an automated analysis program that considers quality of sections before performing many routine analyses, including MuSC quantification.

Clinical Measures

Clinical measures for severity of contracture and walking function were extracted for children with CP from their medical records from their preoperative clinical evaluation. We used a maximal range of motion (ROM), recorded by a trained physical therapist of either ankle dorsiflexion or hip abduction, for gastrocnemius and adductors surgeries, respectively. ROM was normalized to total physiological range for dorsiflexion or abduction and expressed as a percentage to calculate severity of contracture. For ambulatory children, who had an instrumented gait analysis before surgery gait velocity and normalized stride length during walking were also extracted from their gait assessment.

Statistical Analysis

Our independent variables for analysis were primarily muscle group (contractured or noncontractured muscle), myofiber type (type-1 or type-2), functional capacity (ambulatory or non-ambulatory), and diagnostic group (CP or ACL-injured TD). Our dependent variables were MuSC abundance, activation, proliferation, average myofiber cross-sectional area (CSA), myonuclear number, and collagen content. Grouped data were compared using t tests (paired or unpaired), one-way or two-way analysis of variance (ANOVA) tests, as appropriate. Only for ANOVAs with main effects post hoc comparisons of either Tukey’s test or Sidak’s multiple comparisons were performed, as appropriate. Kolmogorov–Smirnov tests were used to compare the frequency distribution of myofiber areas across muscles and fiber types. Spearman’s rank correlations were performed across MuSC abundance, myonuclear number, CSA, and age for each muscle group and fiber type and represented as correlation matrix heat maps. Linear regression analyses were performed to assess amount of common variance between collagen content, and MuSC abundance, activation, proliferation, as well as for manual quantification and MuscleJ quantification for MuSC abundance. Any outliers detected by a box plot were not included in the analysis, mentioned specifically in the results and only nonzero values were used in the regression analyses. All statistical analyses were performed using Prism 9.0 (GraphPad, San Diego, CA), with Brown–Forsythe tests built into them for assessing homoscedasticity for ANOVAs. All data are presented as means ± standard error of means (SEM) in the results. Exact statistical tests, sampling including number of fibers and subjects for each result are detailed in the results section, in Table 1, and in the figure legends.

RESULTS

Myofiber Area and Characteristics Are Altered in Contractured Muscles

Myofiber areas were assessed by immunohistochemical labeling for myofiber borders (Dystrophin), type-1 fiber (BA-D5) (Fig. 1A), and quantified using MyoVision (50). On an average 1,688 ± 232 fibers/section, per subject (Table 1) were quantified across contractured muscles (gastrocnemius, adductors), noncontractured muscles (vastus lateralis) in children with CP, and in typically developing children with an anterior cruciate ligament tear, leading to a temporary reduction in mobility (vastus lateralis). Average overall myofiber areas in contractured muscles were 747 ± 123 µm² and 1,081 ± 171 µm² in gastrocnemius and adductors, respectively (Fig. 1B). In contrast, the noncontractured (vastus lateralis) myofiber area was 1,465 ± 414 µm² and the TD ACL injured vastus lateralis myofiber area was 2,220 ± 660 µm² (Fig. 1B). Both the CP contractured muscles were significantly smaller than the TD ACL injured muscles (P < 0.05, Fig. 1B), but not the noncontractured muscle (P > 0.05). Similarly, only the CP gastrocnemius type-1 fibers were smaller than the TD ACL injured muscle, 718 ± 128 µm² vs. 2,133 ± 669 µm² (P < 0.05, Fig. 1B), whereas both gastrocnemius and adductor muscle type-2 fibers were smaller than TD ACL injured muscle, 801 ± 129 µm², 1,094 ± 166 µm² versus 2,246 ± 649 µm² (P < 0.05, Fig. 1B).

Figure 1.

Myofiber area characteristics. A: representative image showing dystrophin (myofiber border), myosin heavy chain 1 (type 1 fiber), and composite. B: average myofiber area overall, by type 1 fibers and type 2 fibers across contractured muscles [gastrocnemius (Gas), adductors (Add)], noncontractured muscles [vastus lateralis (VL)] in children with cerebral palsy (CP) and in typically developing children following an anterior cruciate ligament (ACL) injury (vastus lateralis). C: binned myofiber area frequency distribution overall, by type 1 fibers and type 2 fibers [n = 36, CP Gas (n = 9), CP Add (n = 11), CP VL (n = 10), typically developing (TD) ACL VL (n = 6)]. Myofiber areas and fiber types were quantified in an average of 1,688 ± 232 fibers/subject using MyoVision. All data are shown as means ± SE. Circles indicate male subjects, squares are female. Grouped data were compared by unpaired t tests (*P < 0.05), and distributions were compared with Kolmogorov–Smirnov tests (^CP Gas vs. CP VL, #CP Gas vs. TD ACL VL, @CP Add vs. CP VL, P < 0.05).

Frequency distribution of all myofiber areas per subject, binned in 250-µm² intervals, was calculated and grouped across subjects for each muscle group. Frequency distribution of myofiber sizes appears to be different overall, between noncontractured muscle, TD ACL injured muscle, and contractured muscles, with gastrocnemius having the greatest percent of smaller fibers in 250–500 µm² bins (38 ± 7%, P < 0.05, Fig. 1C). Type-1 fiber frequency distributions were different for both contractured muscles compared with noncontractured muscle and ACL injured muscle (P < 0.05, Fig. 1C), whereas only the gastrocnemius was different overall, and for type-2 (P < 0.05, Fig. 1C), based on Kolmogorov–Smirnov tests. In both the contractured muscles, coefficient of variation was greater in type-2 fibers than in type-1 fibers (68% and 65% vs. 51% and 53%, gastrocnemius and adductors, respectively, P < 0.01). Fiber-type proportions were variable across the muscles. Type-1 fiber proportions were 48 ± 3%, 44 ± 4%, 60 ± 8%, and 36 ± 6% for CP vastus lateralis, adductor, gastrocnemius, and TD ACL injured vastus lateralis, respectively. CP gastrocnemius was significantly different from TD ACL injured vastus lateralis (unpaired t test, P < 0.05). Contractured muscles appear to have altered fiber-type proportions, smaller myofiber areas, and a greater frequency of smaller fibers.

MuSC Abundance Is Greater In Situ in Contractured Muscles than in Noncontractured Muscles, Primarily in Type-1 Fibers

Satellite cells are the resident muscle stem cells (MuSCs) that are primarily required for postnatal muscle growth. We assessed if MuSC abundance was different between contractured and noncontractured muscles. MuSC in their niche was identified by using immunohistochemistry for satellite cell transcription factor (Pax7), myofiber borders (Dystrophin), and nuclei (DAPI), as well as type 1-fibers (BA-D5) (Fig. 2A). On an average, 1,705 ± 279 fibers per section, per subject were manually quantified in one-to-two sections, per subject using ImageJ for MuSC (Table 1). Contractured muscles had a greater abundance of MuSC compared with noncontractured muscles (CP Gas: 15.5 ± 1.5, CP Add: 16.6 ± 1.1 vs. CP VL: 10.2 ± 0.6, MuSC per one hundred fibers, one-way ANOVA, P < 0.001), with significant post hoc multiple comparisons using Tukey for both contractured muscles (Fig. 2B).

Figure 2.

Satellite cell abundance in noncontractured and contractured muscles in children with cerebral palsy. A: representative image of satellite cells (resident muscle stem cells, MuSCs) immunohistochemistry showing dystrophin (myofiber borders), 4′,6-diamidino-2-phenylindole (DAPI) (nuclei), Pax7 (satellite cell transcription factor), composite showing satellite cells in their niche (Pax7+/DAPI+ present along dystrophin), composite with MyHC1 (myofiber type 1), and Pax7 negative control slide, which was not labeled with Pax7 primary antibody (note similar fibers but no Pax7 expression). Scale bar (bottom right) is 50 μm. B: satellite cell abundance in noncontractured muscle [vastus lateralis (VL)] and contractured muscles [gastrocnemius (Gas), adductors (Add)] [n = 30, cerebral palsy (CP) VL (n = 10), CP Add (n = 11), CP Gas (n = 9)]. C: satellite cell abundance by type 1 & 2 fibers in the same muscles. On an average 1,705 ± 279 fibers/section were manually quantified in 1–2 sections per subject using ImageJ. All data are shown as means ± SE; circles indicate male subjects, squares are female. Grouped data were compared using an one-way ANOVA across muscles and a two-way ANOVAs for main effects of muscle, fiber types, interaction effects with post hoc Tukey’s or Sidak’s multiple-comparison tests (**P < 0.01, ***P < 0.001, ****P < 0.0001).

We next assessed if this increase in abundance was in both fiber types using a two-way ANOVA (muscle × fiber type). We found a main effect for muscle group, fiber types, and an interaction effect (P < 0.01, Fig. 2C). Increase in abundance in contractured muscles was only seen in type-1 muscle fibers (CP Gas: 22.3 ± 3.7, CP Add: 19.8 ± 1.4 vs. CP VL: 10.7 ± 1.2, MuSC per one hundred fibers, P < 0.001), but not type-2 fibers, based on post hoc comparisons using Tukey’s test. In addition, within contractured muscles there was a significantly lower abundance of MuSC in type-2 muscle fibers compared with type-1 muscle fibers in gastrocnemius muscle (9.0 ± 1.9 vs. 22.3 ± 3.7, MuSC per one hundred fibers, P < 0.0001), whereas in adductors this tended to significance (14.1 ± 1.0 vs. 19.8 ± 1.4, P = 0.0623), using post hoc Sidak’s test. MuSC abundance appears to be higher in contractured muscles, specifically in type-1 muscle fibers.

MuSC Myogenic Characteristics (Activation, Proliferation) Is Lower In Situ in Contractured Muscles than Noncontractured Muscles

We then evaluated if increased MuSC abundance in contractured muscles was due to greater percentage of myogenic (activated and/or proliferating) cells. Activated MuSC were characterized by labeling for satellite cell transcription factor (Pax7), nuclei (DAPI), and a myogenic regulatory factor (myogenic determination factor, MyoD) in contractured and noncontractured muscles. Total MuSC were quantified manually in ImageJ as Pax7+/DAPI+ cells, whereas activated MuSC were quantified as Pax7+/MyoD+/DAPI+ (Fig. 3A). There was a significant effect of muscle group on activated MuSC, based on one-way ANOVA (P < 0.05). In contrast to the MuSC abundance, the contractured adductors but not gastrocnemius had a significantly lower percent of activated MuSC than the noncontractured muscle (CP Gas: 1.05 ± 0.33, CP Add: 0.33 ± 0.15, CP VL: 2.83 ± 1.12, % Total MuSC, P < 0.05, Fig. 3B), based on post hoc Tukey’s test.

Figure 3.

Activated satellite cells in noncontractured and contractured muscles in children with cerebral palsy. A: representative image showing Pax7 (satellite cell transcription factor), MyoD (myogenic regulatory factor), 4′,6-diamidino-2-phenylindole (DAPI) (nuclei), composite (note the white arrows indicate a couple of examples of Pax7+/MyoD+ cells). B: activated muscle stem cells (satellite cells) (Pax7+/MyoD+) in noncontractured muscle [vastus lateralis (VL)] and contractured muscles [gastrocnemius (Gas), adductors (Add)] [n = 29, cerebral palsy (CP) VL (n = 10), CP Add (n = 10), CP Gas (n = 9)]. Quantification was performed using ImageJ. All data are shown as means ± SE; circles indicate male subjects, squares are female. Grouped data were compared using an one-way ANOVA across muscles with post hoc Tukey’s multiple-comparison tests (*P < 0.05). MuSC, muscle stem cells.

We then assessed proliferating MuSC in these muscles using labeling for satellite cell transcription factor (Pax 7), a cell cycle proliferation marker (Ki67), nuclei (DAPI), and type-1 fiber (BA-D5) (Fig. 4A). Similar to activated MuSC, one-way ANOVA showed an overall main effect of muscle group on proliferating MuSC (P < 0.0001). Again, in contrast with MuSC abundance and similar to activated MuSC, both contractured muscles had significantly lower proliferating MuSC than noncontractured muscles (CP Gas: 0.69 ± 0.31, CP Add: 0.57 ± 0.23, CP VL: 3.33 ± 0.59, % Total MuSC, P < 0.05, Fig. 4B), based on post hoc Tukey’s test. We then assessed if this decrease in proliferating MuSC in contractured muscles was in both fiber types, using a two-way ANOVA (muscle × fiber type). We found a main effect across muscle groups but not across fiber types on proliferating MuSC (P < 0.0001). In the contractured muscles both fiber types had a lower percentage of proliferating MuSC compared with noncontractured muscle [Type-1 (Gas, Add, VL): 0.73 ± 0.36, 0.73 ± 0.34, 3.85 ± 0.63, Type-2 (Gas, Add, VL): 0.15 ± 0.11, 0.32 ± 0.21, 3.00 ± 0.86, % Total MuSC, P < 0.001, Fig. 4C], based on post hoc Tukey’s test, but no significant differences were seen across fiber types within each muscle group. MuSC myogenic characteristics appear to be lower in contractured muscles in both fiber types.

Figure 4.

Proliferating satellite cells in noncontractured and contractured muscles in children with cerebral palsy. A: representative image showing Pax7 (satellite cell transcription factor), Ki67 (cell proliferation marker), 4′,6-diamidino-2-phenylindole (DAPI) (nuclei), composite (note the white arrows indicate a couple of examples of Pax7+/Ki67+ cells). B: proliferating satellite cells (Pax7+/Ki67+) in noncontractured muscle [vastus lateralis (VL)] and contractured muscles [gastrocnemius (Gas), adductors (Add)] [n = 30, cerebral palsy (CP) VL (n = 10), CP Add (n = 11), CP Gas (n = 9)]. C: satellite cell proliferation by fiber type 1 & 2 in the same muscles. Quantification was performed using ImageJ. All data are shown as means ± SE. Circles indicate male subjects, squares are female. Grouped data were compared using an one-way ANOVA across muscles and a two-way ANOVA for main effects of muscle, fiber types, interaction effects with post hoc Tukey or Sidak multiple-comparison tests (***P < 0.001, ****P < 0.0001).

Extracellular Matrix Content Alterations in CP Muscles and Associations with MuSC Characteristics

MuSCs are present in their niche surrounded by extracellular matrix (ECM) on one side and the myofiber sarcolemma on the other. Since ECM changes can influence MuSC function such as symmetric/asymmetric cell division and proliferation, we assessed if there was greater fibrosis. We measured collagen content using Sirius Red/Fast Green staining, where red labels the collagen proteins, while green the noncollagenous muscle proteins (Fig. 5A). The images were deconvoluted into red, green, blue, and quantified as percentage of total image area in ImageJ. Red to green percent area was calculated to assess collagen content in all muscles. There was a main significant effect of muscle group on collagen content on one-way ANOVA (P < 0.01). However, there was a lower percentage of collagen to noncollagen area in both contractured muscles [CP Gas: 4.7 ± 0.8, CP Add: 4.0 ± 0.7, CP VL: 9.0 ± 1.6, collagen to noncollagen area (%), P < 0.05, Fig. 5B], based on post hoc Tukey’s test, indicating they did not have increased fibrosis. Within each muscle group, we did not find any linear associations between collagen content and MuSC abundance, activation, and proliferation (P > 0.05). However, when we evaluated CP muscles overall, we found small but significant linear associations. Linear regression analyses show collagen content was negatively associated with MuSC abundance (r² = 0.16, P < 0.05), while it was positively associated with both MuSC activation (r² = 0.25, P < 0.05) and MuSC proliferation (r² = 0.42, P < 0.05). One data point for collagen content was detected as an outlier and not used for these analyses. Although contractured muscles do not show an increase in collagen content compared with noncontractured muscles, collagen content in CP muscles overall shows significant associations with MuSC characteristics.

Figure 5.

Extracellular matrix collagen in noncontractured and contractured muscles in children with cerebral palsy (CP) with overall associations with muscle stem cell (MuSC) characteristics. A: representative image of fast green, sirius red staining where the myofibers are green and extracellular matrix collagen is pink red. B: collagen to noncollagenous area in noncontractured [vastus lateralis (VL)] and contractured [gastrocnemius (Gas), adductors (Add)] muscles [n = 29, CP VL (n = 10), CP Add (n = 10), CP Gas (n = 9)]. Across all muscles significant associations between collagen content and MuSC abundance (C), MuSC activation (D), and MuSC proliferation (E). Composite image was deconvoluted into red, green, blue, and quantified using ImageJ. All data are shown as means ± SE. Circles indicate male subjects, squares are female. Grouped data were compared using an one-way ANOVA across muscles, with post hoc Tukey’s multiple-comparisons, and linear regressions (*P < 0.05, **P < 0.01).

Myonuclear Characteristics Are Similar in Contractured and Noncontractured Muscles but Vary in Fiber Types

Since MuSCs are the source of new myonuclei that are added during postnatal development, we assessed if myonuclear characteristics were altered between contractured and noncontractured muscles. Myonuclei were quantified in an average of 1,688 ± 232 fibers/section, per subject (Table 1) using MyoVision (50), which identified myonuclei per fiber on the basis of DAPI and location along dystrophin boundary, grouped by fiber type. Unpaired t tests showed that gastrocnemius myonuclear number was the lowest that tended to significance compared with both adductors (P = 0.0610) and vastus lateralis (P = 0.0628) (CP Gas: 1.26 ± 0.03, CP Add: 1.45 ± 0.08, CP VL: 1.50 ± 0.11, myonuclei/fiber, Fig. 6A). We then evaluated if myonuclear number was similar in both fiber types using a two-way ANOVA (muscle × fiber type). We found a main effect of fiber types but not of muscle group on myonuclear number (P < 0.0001). Both contractured muscle type-2 fibers, in contrast to MuSC abundance, had a higher number of myonuclei/fiber than type-1 fibers, (CP Gas: 1.16 ± 0.02 vs. 1.41 ± 0.07, CP Add: 1.28 ± 0.05 vs. 1.51 ± 0.09, type-1 vs. type-2, myonuclei/fiber, P < 0.001, Fig. 6B), but not noncontractured muscles (CP VL: 1.41 ± 0.08 vs. 1.49 ± 0.10), based on post hoc Sidak’s test. We also evaluated if myonuclear domain, i.e., average myofiber area per myonuclei was different between contractured and noncontractured muscles. Overall, although myonuclear domain was smallest in the gastrocnemius and largest in the vastus lateralis, this was not significantly different (CP Gas: 771 ± 104 µm², CP Add: 997 ± 140 µm², CP VL: 1377 ± 344 µm², P > 0.05).

Figure 6.

Myonuclear characteristics in noncontractured and contractured muscles in children with cerebral palsy. A: average myonuclei/fiber in noncontractured muscle [vastus lateralis (VL)] and contractured muscles [gastrocnemius (Gas), adductors (Add)] [n = 30, cerebral palsy (CP) VL (n = 10), CP Add (n = 11), CP Gas (n = 9)]. B: myonuclei/fiber in fiber type 1 and 2 in the same muscles. Quantification was performed using MyoVision based on dystrophin and DAPI immunohistochemistry. All data are shown as means ± SE. Circles indicate male subjects, squares are female. Grouped data were compared using unpaired t tests, a two-way ANOVA for main effects of muscle, fiber types, interaction effects with post hoc Sidak’s multiple-comparison tests (***P < 0.001).

Associations between MuSC Abundance, Myonuclear Number, Myofiber Cross-Sectional Area, and Age in Contractured and Noncontractured Muscles

To further understand the relationship between MuSC and developmental aspects of muscle growth, we looked at an exploratory correlation matrix between MuSC abundance, myonuclear number, average myofiber cross-sectional area (CSA), and age within contractured and noncontractured muscles, overall, and by fiber type. Positive associations were observed between MuSC abundance and myonuclear number in gastrocnemius muscle overall (r_s = 0.73, P < 0.05, Fig. 7A), in adductor muscle in type-1 fibers (r_s = 0.63, P < 0.05, Fig. 7B), but not in vastus lateralis muscles (Fig. 7C). Similarly, positive associations were seen between myonuclear number and CSA in gastrocnemius in type-2 fibers (r_s = 0.80, P < 0.05, Fig. 7A), in adductors in both in type-1 fibers (r_s = 0.65, P < 0.05, Fig. 7B), and in type-2 fibers (r_s = 0.68, P < 0.05, Fig. 7B), but not in vastus lateralis (Fig. 7C). A negative association was seen between MuSC abundance and CSA only in vastus lateralis type-2 fibers, (r_s = −0.77, P < 0.05, Fig. 7C), suggesting a reduction in MuSC abundance as a muscle grows. Although interestingly, only gastrocnemius type-1 fibers had a positive association between age and fiber CSA (r_s = 0.79, P < 0.05). Although MuSC has contributed to myonuclei early on in postnatal development, associations between abundance, myonuclear number, and CSA only in contractured muscles suggest their MuSC pool may not be entirely established or is altered.

Figure 7.

Heatmap for associations between muscle stem cells (MuSCs) abundance, myonuclear number, myofiber cross-sectional area, and age in children with cerebral palsy. Correlation matrix represented as a upper triangular matrix, across variables for muscles overall, for type-1 myofibers, and for type -2 myofibers in gastrocnemius (n = 9) (A), adductors (n = 11) (B), and vastus lateralis (n = 10) (C). All correlations are Spearman rank correlations (r_s, −1.0 to +1.0). Heatmap color indicates strength of association, with blue corresponding to +1.0, red to −1.0 and white is 0.0. Note that the main diagonal represents each variable’s correlation with itself. (*next to the correlations indicate P < 0.05). CSA, cross-sectional area.

MuSC Abundance and Myogenic Characteristics Are Similar between Ambulatory and Nonambulatory Children

Children with cerebral palsy are either able to walk independently, with the use of assistive devices (GMFCS I, II, and III) or are primarily nonambulatory, using wheelchairs for mobility (GMFCS IV, V). We wanted to evaluate if ambulatory capacity had an effect on MuSC characteristics in noncontractured and all contractured muscles, as a group. We evaluated MuSC abundance across muscles and ambulation using a two-way ANOVA (muscle × ambulation capacity). We found a significant main effect only of muscle group, but not ambulation capacity on MuSC abundance (P < 0.001, Fig. 8A). In both ambulatory and nonambulatory children, contractured muscles had greater MuSC abundance than noncontractured muscles (Ambulatory: 15.7 ± 1.1 vs. 9.0 ± 0.7, nonambulatory: 16.7 ± 1.5 vs. 11.5 ± 0.7, MuSC per one hundred fibers, contracture vs. non-contracture, P < 0.05, Fig. 8A), based on post hoc Sidak’s test.

Figure 8.

Satellite cell characteristics across gross motor functional classification system (GMFCS) levels in noncontractured and contractured muscles in children with cerebral palsy. Satellite cell abundance (A), activation (B), and proliferation in noncontractured muscles [vastus lateralis (VL)], contractured muscles [gastrocnemius (Gas), adductors (Add)] in ambulatory (GMFCS I, II, and III), and nonambulatory (GMFCS IV and V) children with cerebral palsy (n = 29 or 30, ambulatory noncontractured (n = 5), ambulatory contractured (n = 11 or 12), nonambulatory noncontractured (n = 5), nonambulatory contractured (n = 8) (C). All data are shown as means ± SE. Grouped data were compared using two-way ANOVAs for main effects of muscle, ambulation status, and interaction effects with post hoc Sidak’s multiple-comparison tests (*P < 0.05, **P < 0.01, ***P < 0.001). MuSC, muscle stem cells.

We next used a two-way ANOVA (muscle × ambulation capacity) to assess activated MuSC across muscles and ambulation. We found significant main effects of muscle, ambulation, and an interaction effect (P < 0.01, Fig. 8B). Surprisingly, only nonambulatory children had greater activated MuSC than ambulatory children in noncontractured muscles (5.02 ± 1.75 vs. 0.64 ± 0.42, % Total MuSC, P < 0.01, Fig. 8B), based on post hoc Sidak’s test. Post hoc comparisons also showed that only in nonambulatory children, noncontractured muscles had greater activated MuSC than contractured muscles (5.02 ± 1.75 vs. 0.69 ± 0.20, % Total MuSC, P < 0.001, Fig. 8B).

We then assessed proliferating MuSC in ambulatory and nonambulatory children using a two-way ANOVA (muscle × ambulation capacity). Similar to MuSC abundance, we only found a main effect of muscle but not ambulation capacity on proliferating MuSC (P < 0.0001, Fig. 8C). In both ambulatory and nonambulatory children, contractured muscles had greater proliferating MuSC than noncontractured muscles (Ambulatory: 15.7 ± 1.1 vs. 9.0 ± 0.7, Nonambulatory: 16.7 ± 1.5 vs. 11.5 ± 0.7, % Total MuSC, contracture vs. noncontracture, P < 0.01, Fig. 8C), based on post hoc Sidak’s test. Overall, ambulation capacity by itself does not appear to have a major impact on MuSC characteristics in contractured or noncontractured muscles.

Associations of MuSC Abundance with Clinical Measures of Contracture and Overall Walking Function

To assess amount of contracture, we expressed the maximal measured joint range of motion (ROM) angle for ankle dorsiflexion and hip abduction as a percent of total ROM. The lower the number greater the amount of contracture. Although there was a negative association between MuSC abundance in contractured muscles and degree of contracture (n = 14, r = −0.38), this was nonsignificant (P = 0.19). Since MuSC abundance can be increased with activity and exercise, although there were no main effects of ambulation, within the ambulatory children we assessed if there was any association between MuSC abundance and walking function. Overall, there was no association between MuSC abundance across all muscles and gait velocity, or stride length (n = 16, P = 0.18). However, within the gastrocnemius muscles, there was a negative association between abundance and walking function (n = 7, r = −0.82, P < 0.05 and r = −0.87, P < 0.05, gait velocity and stride length, respectively), i.e., children with lower walking function had a higher abundance of MuSC. We did not find many clinical correlates of MuSC abundance.

Myofiber and Myonuclear Characteristics in Ambulatory and Nonambulatory Children

We next evaluated if myofiber area and myonuclei, which are derived from MuSC were different between ambulatory and nonambulatory children. Average myofiber area was assessed in ambulatory and nonambulatory children between contractured, noncontractured muscles using a two-way ANOVA. We found a significant main effect of muscle, but not ambulation capacity on average myofiber area (Noncontractured: 1,803 ± 659 µm², 1,630 ± 618 µm², Contractured: 977 ± 140 µm², 863 ± 199 µm², ambulatory, nonambulatory, P < 0.05, Fig. 9A).

Figure 9.

Myofiber area and myonuclear characteristics across gross motor functional classification system (GMFCS) levels in noncontractured and contractured muscles in children with cerebral palsy. Average myofiber area (A), myonuclei/fiber showing an interaction effect between muscles and ambulation status (B), myonuclei/fiber in noncontractured muscles [vastus lateralis (VL)], contractured muscles [gastrocnemius (Gas), adductors (Add)] in ambulatory (GMFCS I–III), and nonambulatory (GMFCS IV and V) children with cerebral palsy [n = 30, ambulatory noncontractured (n = 5), ambulatory contractured (n = 12), nonambulatory noncontractured (n = 5), nonambulatory contractured (n = 8)] (C). All data are shown as means ± SE. Grouped data were compared using two-way ANOVAs for main effects of muscle, ambulation capacity, and interaction effects with post hoc Sidak’s multiple-comparison tests (#main effect of muscle, *P < 0.05).

We then assessed myonuclear number in contractured and noncontractured muscles in ambulatory and nonambulatory children using a two-way ANOVA (muscle × ambulation capacity). We found a significant interaction between the effects of ambulation capacity and muscle (P < 0.05, Fig. 9B) but no main effects of either on myonuclear number. Noncontractured muscles in nonambulatory children had greater myonuclear number than contractured muscles (1.60 ± 0.15 vs. 1.32 ± 0.05 myonuclei/fiber, Fig. 9C), which tended to be significant on post hoc Sidak’s test (P = 0.0527). Myonuclear number in noncontractured muscles in nonambulatory children were greater than in ambulatory children (1.60 ± 0.15 vs. 1.30 ± 0.03 myonuclei/fiber, Fig. 9C), which tended to be significant on post hoc Sidak’s test (P = 0.0686). Overall, ambulation capacity by itself does not appear to significantly impact myofiber or myonuclear characteristics in contractured or noncontractured muscles.

MuSC Characteristics in Children with CP Are Different from Musculoskeletal Injured Typically Developing Children

We wanted to assess if MuSC characteristics in children with CP were similar to other musculoskeletal injuries such as an anterior cruciate ligament (ACL) tear that leads to short-term altered mobility, in children with typical development (TD). We assessed this in the vastus lateralis in both groups. Children with CP had a lower abundance of MuSC compared with TD (CP VL: 10.2 ± 0.6 vs. TD: 19.6 ± 3.5, MuSC/ 100 fibers, unpaired t test, P < 0.01) (Fig. 10A). We next used a two-way ANOVA (group × fiber type) to assess if this decrease in abundance was in both fiber types. We found a main effect for muscle group but not between fiber types (P < 0.01). This decrease in abundance was seen in both type-1 muscle fibers (10.7 ± 1.2 vs. 21.9 ± 3.5, MuSC/100 fibers, P < 0.01), and in type-2 muscle fibers (9.4 ± 0.8 vs. 19.0 ± 3.8, MuSC/100 fibers, P < 0.01, Fig. 10B), based on post hoc comparisons using Sidak’s test. Activated MuSC were higher in CP but not significant (2.74 ± 1.07 vs. 1.52 ± 0.71, % Total MuSC, unpaired t test, P > 0.05, Fig. 10C), whereas proliferating MuSC were also higher in CP, which tended to significance (3.33 ± 0.59 vs. 1.30 ± 0.82, % Total MuSC, unpaired t test, P = 0.0598, Fig. 10D).

Figure 10.

Satellite cell characteristics in musculoskeletal-injured typically developing (TD) children and noncontractured muscles in children with cerebral palsy (CP). Satellite cell abundance (A), abundance by fiber type (B), activation (C), and proliferation in noncontractured muscles [vastus lateralis (VL)] in children with CP and musculoskeletal injury (torn anterior cruciate ligament, ACL) in typically developing children [n = 16, TD ACL VL (n = 6), CP VL (n = 10)] (D). All data are shown as means ± SE. Circles indicate male subjects, squares are female. Grouped data were compared using unpaired t tests and/or a two-way ANOVA for main effects of muscle, fiber types, interaction effects with post hoc Sidak’s multiple-comparison test (**P < 0.01). MuSC, muscle stem cells.

Manual Quantification of MuSC Abundance Was Largely Similar to Automated Quantification

Since the manual quantification was not performed in a blinded fashion, we validated our manual quantification of MuSC abundance in ImageJ by comparing it with automated quantification using MuscleJ (51), to ensure we avoided any systemic bias. We compared MuSC abundance using Pax7+/DAPI+ cells across the majority of sections (n = 68). No significant differences were seen between manual and MuscleJ quantification (220 ± 21 vs. 226 ± 22, Pax7+/DAPI+ cells per section, manual vs. MuscleJ, paired t test, P = 0.3842, Fig. 11A). Not surprisingly, our Pax7 negative control group (n = 23), that lacked a primary antibody for Pax7 had significant differences compared with both manual and MuscleJ (4 ± 1 Pax7+/DAPI+ cells per section, mixed-effects model, P < 0.0001, Fig. 11A). Linear regression also showed a strong association between manual and MuscleJ quantification (r² = 0.88, P < 0.0001, Fig. 11B). We also compared a subset of slides with no visible sectioning artifacts (n = 20) since MuscleJ is fully automated requiring high-quality sections and does not allow any user selection/elimination of regions of sections. We found in these sections, although improving quality of sections does increase the association (r² = 0.99), there is a slight but significant difference between them (246 ± 54 vs. 231 ± 52, Pax7+/DAPI+ cells per section, manual vs. MuscleJ, paired t test, P < 0.05).

Figure 11.

Validation of manual satellite cell quantification. A: comparison between manual and automated (MuscleJ) quantification of satellite cells (Pax7+/DAPI+) per section along with quantification in Pax7 negative control sections. B: significant association between manual and automated quantification across all sections (n = 68). All data are shown as means ± SE. Data were compared using t tests and linear regression (****P < 0.0001).

We also evaluated average MuSC abundance per section across sequential slides, roughly 40–50 µm apart using MuscleJ (n = 57 sections across 33 subjects), since MuSC nuclear length is around the thickness of a section (52) and abundance could potentially vary across the length of the muscle. Linear regression showed a strong association between slides per subject (r² = 0.89, P < 0.0001). However, there were also some regional differences in MuSC abundance across slides (242 ± 36 vs. 277 ± 41, Pax7+/DAPI+ cells average per section, slide 1 vs. slide 2, paired t test, P < 0.05). Similar results were observed for manual quantification using slides, which were roughly 80–100 µm apart (n = 32, r² = 0.95, P < 0.0001; 239 ± 33 vs. 263 ± 35, Pax7+/DAPI+ cells average per section, slide 1 vs. slide 2, paired t test, P < 0.01). With appropriate labeling protocols, automated quantification of MuSC abundance appears similar to manual quantification.

DISCUSSION

Children with cerebral palsy (CP) have a significantly impaired muscle development, secondary to a nonprogressive brain injury around birth, that affects some muscles predominantly leading to contracture development. Children are either primarily ambulatory; independently or with the use of assistive devices such as walkers, or are nonambulatory, primarily using wheelchairs for mobility. During postnatal muscle development satellite cells, i.e., muscle stem cells (MuSCs) are responsible for new myonuclei and muscle growth. In this study, we used immunohistochemistry to assess if resident MuSC myogenic, myonuclear, and myofiber characteristics in situ are altered between muscles that develop contractures and those that do not. In addition, we sought to assess if these characteristics in contractured and noncontractured muscles are different between ambulatory and nonambulatory children. And finally, we assessed if muscle characteristics in children with CP are similar to those with typical development (TD) following a common musculoskeletal injury, a torn anterior cruciate ligament (ACL).

Unsurprisingly, we found that average myofiber area in contractured muscles were smaller, with greater percentage of smaller fibers. Surprisingly, we saw that resident MuSC abundance in situ was greater in contractured muscles compared with noncontractured muscles, although this was primarily in fiber type 1. In contrast, myogenic characteristics in situ measured as activated MuSC and proliferating MuSC was lower in contractured muscles, supporting our hypothesis that altered myogenic potential might be responsible for development of contractures. Extracellular matrix collagen content measured histochemically was lower in contractured muscles, possibly important for establishment of an appropriate MuSC niche or myogenic progression. Accordingly, we report that when we consider CP muscle as a whole, collagen content and MuSC abundance are negatively associated, whereas MuSC activation and proliferation are positively associated. Myonuclear accretion, which comes from MuSC during postnatal development, was similar between contractured and noncontractured muscles. However, only in contractured muscles type-1 fibers had lower myonuclear number than type-2 fibers. Our correlational analysis also showed significant associations between myonuclear number and myofiber cross-sectional area, as well as between MuSC abundance and myonuclear number only in contractured muscles. Puzzlingly, we did not observe any major differences in MuSC, myonuclear, and myofiber characteristics between ambulatory and nonambulatory children, only between contractured and noncontractured muscles. Compared with ACL-injured TD children, CP noncontractured muscle has a lower abundance of MuSC but no differences in activated or proliferating cells. Finally, automated quantification of MuSC abundance with MuscleJ matches well with manual quantification and across multiple slides. To the best of our knowledge, this is the largest and most comprehensive in situ immunohistochemical analysis of MuSC characteristics in children with CP.

Impaired muscle growth is a key characteristic of children with CP. Prior macroscopic measurements of muscle volume have shown that impaired growth occurs from early on in life but does not affect all muscles equally (8, 10–12). Microscopic measurements have shown variable findings with some showing overall reduction in myofiber cross-sectional area (15), whereas others have primarily reported heterogeneity (24, 25, 53). Our data here clearly show contractured muscles have a greater percentage of smaller fibers and are overall smaller, in both fiber types. We also show that coefficient of variation in type-2 fibers is greater than that in type-1 fibers in contractured muscles, similar to what has previously been reported (24, 53). Interestingly, noncontractured muscle does not appear to be dramatically different from the typically developing (TD) children. Although this suggests that noncontractured muscle growth is unimpaired, the TD children in this study had a torn knee ligament for which they were undergoing surgery. In adults following this injury, there are significant alterations in the vastus lateralis, including atrophy and MuSC characteristics, which impairs recovery from these injuries in the long term (46–48). Although little prior information exists in ACL-injured children, it is likely that the atrophy in TD might be masking some of the changes.

Previous studies, including our own, have shown that MuSC of contractured muscle (hamstring, bicep) in children with CP exhibit a reduced MuSC abundance compared with TD (25, 39, 40). Our noncontractured group compared with TD agrees with this prior work in that MuSC abundance is reduced. However, in contrast to previous studies, we observed both contractured muscle groups, gastrocnemius and adductors had a greater MuSC abundance compared with noncontractured muscle, specifically in type-1 fibers. There are several possible explanations for this seeming inconsistency. First, the muscles are different suggesting some heterogeneity even within contractured muscles. The exact reason for reduction of MuSC abundance reported in prior studies is unknown. They possibly reflect an exhaustion of the MuSC pool or a failure to establish an appropriate MuSC pool. The amount of contracture might also be different between them, simple clinical measure of decreased range of motion do not capture the intrinsic biological similarities or differences between contractured muscles. Second, the developmental ages were similar in all the studies, roughly preadolescence since typically surgical correction for contractures is performed around then. However, the exact postnatal developmental timeline for establishment of a mostly quiescent MuSC pool is unknown in human studies. Based on murine studies it is known that MuSC abundance is highest during early stages, which reduces as myonuclei are established and as the MuSC pool is established (37). Consequently, our data of increased abundance could potentially reflect an earlier stage of development, suggesting a maturation defect in these muscles, which might lead to a reduction later on.

Muscle contractures and MuSC exhaustion are prominent features in a common childhood myopathy, Duchenne muscular dystrophy (DMD) (54, 55). In DMD, due to constant bouts of injury and repair, MuSC increases in abundance initially but as the disease progresses the MuSC pool undergoes extreme exhaustion (56–58). Similarly, muscle contractures are seen in children with brachial plexus birth palsy (59). Although it is unknown what happens to MuSC abundance in these muscles in children, a neonatal birth injury mouse model has investigated cellular aspects in these muscle contractures (60). Muscle contractures in these mice raise abundance of MuSC immediately after the plexus injury (60). Similar results are also seen in peripheral nerve injury, where there may be an increase in MuSC abundance before reduced abundance (61, 62). In the context of known MuSC abundance-trajectory during postnatal muscle development, our results of increased MuSC abundance may be either reflecting an altered myogenic-timeline to establish the MuSC pool in contractured muscle or a response to the initial brain injury. However, since we are looking at children 10–12 yr of age, the latter is likely not a reflection of an initial response from brain-injury and probably reflects an inherent state of the MuSC.

We used two common markers of MuSC activation—MyoD (63) and proliferation—Ki67 (64) that showed that in contrast to the increased MuSC abundance, these were decreased in MuSC of contractured muscles in situ, suggesting a lower myogenic potential. This matches with the in vitro studies in children with CP that show these cells are unable to form appropriately sized or organized myotubes (41–43). In contrast to our results, MuSC in DMD muscles have increased activation and proliferation (26, 65), whereas in brachial plexus injuries there is no difference (60). In adult healthy muscle the percentage of proliferating MuSC at baseline is fairly small (0.5%), which increases three- to fivefold with resistance exercise programs (64). Our results in noncontractured muscles show an approximately sevenfold increase compared with the baseline, suggesting robust muscle growth. However, the contractured muscles have similar low percentage of proliferating MuSC as adults at baseline. This again suggests this may be reflective of the state of postnatal establishment of MuSC pool or MuSC niche or showing alterations in responsiveness of the MuSC in response to growth stimulus.

MuSC are capable of undergoing asymmetric, symmetric cell division within their niche between the basal lamina of the ECM and the sarcolemma of the myofibers, which produce either committed myogenic progenitors or quiescent muscle stem cells (28). Previous studies have shown that ECM content in contractured muscle in children with CP is altered (15–18, 40), but it is unclear how that translates to their mechanical stiffness. Here, we use a histochemical method to measure collagen content in ECM, as a ratio of myogenic area, which shows this is reduced in contractured muscles. Importantly, we show in CP muscle as a whole that there are small but significant associations between collagen content and MuSC characteristics, which suggest a more intricate connection between MuSC in their ECM niche. Both activation and proliferation have a positive association with collagen content but abundance is negatively associated with collagen content. Alteration of ECM stiffness or associated signaling can directly modulate SCs’ ability to self-renew, proliferate, and differentiate (66–70). Consequently, our observation of lower MuSC myogenic characteristics might be related to alterations of ECM leading to altered state of MuSC quiescence, activation, and proliferation. Alternatively, changes in MuSC have been reported to alter the ECM negatively, whereas ablation of MuSC leads to increased collagen (71–73). Prior studies have also reported a decrease in ECM gene expression during the postnatal development period, whereas the MuSC pool is being actively established (34), suggesting a close interaction between them during this time.

Contractured muscles in children with CP have overstretched sarcomeres (15), suggesting they are under high amounts of tension. MuSC deformations by stretching act as activating stimuli to MuSC (74), and MuSC in CP muscle contractures undergo similar stretch-induced deformations as TD muscles, with increased elongation at increased sarcomere lengths (75). Overtime as muscles are unable to add more sarcomeres and develop overstreched sarcomeres, these MuSCs may have adapted to their elongated shape deformation and potentially reduced mechanically sensitive MuSC-activating mechanisms. This could be another potential explanation for the high SC abundance but, low activation/proliferation in contractured muscle.

Muscle fibers have numerous myonuclei that fuse into myofibers during development. During late embryonic development, MuSCs are established along the myofibers in their niche (76), and these are the source for myonuclei during postnatal development (31, 32) since myonuclei are postmitotic and unable to divide. This myonuclear accretion happens primarily during early postnatal development (33–35). Our data here show that myonuclear number and myonuclear domain in myofibers in muscles of children with CP are not statistically different across muscles, suggesting myonuclear accretion happened early on during postnatal development, regardless of the perinatal brain injury. However, both myonuclear number and domain are lowest in the contractured gastrocnemius muscle. During postnatal development, there is an increase in myonuclear domain as a muscle grows (33, 34). A smaller myonuclear domain might be reflective of an earlier phase of development or altered capacity. There is a more prolonged, dynamic role of MuSCs during postnatal development with two specific periods of activation, differentiation, and eventually an establishment of a quiescent MuSC pool by the end of the postnatal period (37). Positive associations between myonuclear number and myofiber cross-sectional area reduce in later postnatal development (34). Although our study is not longitudinal, some correlational analysis between MuSC abundance, myonuclear number, and myofiber cross-sectional area might be relevant to understanding postnatal development in contractured muscles. We observed an association between MuSC abundance and myonuclear number, as well as between myonuclear number and myofiber cross-sectional area only in contractured muscles. This may be reflective of an earlier phase of development and suggests that either the MuSC pool establishment in contractured muscles is ongoing or altered.

Children with cerebral palsy are either able to walk independently, with the use of assistive devices (GMFCS I, II, and III) or are primarily nonambulatory, using wheelchairs for mobility (GMFCS IV, and V). We did not find any significant differences in MuSC, myonuclear, and myofiber characteristics between ambulatory and nonambulatory children. Walking is typically established in children only around 12–18 mo of age, so some of the features of early postnatal development might be similar across children regardless of the developmental milestones. The primary variable that was different in both groups of children was whether the muscle was contractured or noncontractured, suggesting that the presence of muscle contractures might be driving some of the characteristics in MuSC, myonuclei, and myofibers rather than ambulation capacity. In addition, we did find that overall MuSC abundance was greater primarily in type-1 fibers in contractured muscles. This combined with reduced activation and proliferation might be reflective of a reduced ability of myogenic progression or could possibly be related to recruitment. Children with CP have a brain injury that typically leads to neuromuscular deficits, with an impaired ability to recruit stronger, type-2 fibers (77, 78). From early age, they might utilize mostly type-1 fibers in contractured muscles, which might be reflected in alterations in MuSC abundance. Our negative association between abundance and walking function also shows that children who are walking with higher velocities have a lower MuSC abundance in the gastrocnemius muscle.

In recent years, there have been numerous computational tools developed for automated or semiautomated quantification of muscle and cell characteristics, some of which have batch processing capabilities (50, 51, 79–83). Here we used both MyoVision and MuscleJ to comprehensively quantify, characterize and validate a large number of sections from 36 biopsies. This provides an illustration of the ability to combine manual and automated quantification, which might be necessary for reproducibility across different studies.

Our study, as with any utilizing biopsies from children, has a number of technical and practical limitations that we would like to address here to provide clarity and discuss potential confounds. First, inherently by comparing muscles with contractures and those without in children with CP, we are comparing different muscles. This variation in muscle sizes and development of contractures is the area of clinical interest that we are attempting to address here. We do not think our results in differences in MuSC characteristics necessarily reflect that since the muscles with contractures are also smaller. In addition, the large magnitude of difference we report between contractured and noncontractured muscles is unlikely due to primarily different muscle architecture. The architectural properties of vastus lateralis, adductors, and gastrocnemius muscles are fairly similar in adults without CP (84). Most of the literature is primarily on MuSC characteristics in vastus lateralis, mostly in adults but there is scarce to no data in the literature on MuSC characteristics in healthy adductor or gastrocnemius muscles in children or adults, due to challenges with obtaining muscle biopsies, particularly in children. Second, all the biopsies were obtained from children who were undergoing corrective surgery for their musculoskeletal impairments. Consequently, it might not reflect muscle properties in children who do not have muscle contractures or younger children. Future studies need to assess changes in MuSC characteristics in the same muscle, in the presence and absence of contractures, potentially with needle biopsies. Our children with TD had severe anterior crucial ligament (ACL) tears, necessitating reconstructive surgery. Our goal was not to compare with TD children, but rather with other chronic childhood injuries that limit mobility in the short term to evaluate how that alters muscle in the absence of a brain injury. This study depended entirely on immunohistochemistry and histochemical methods to quantify muscle characteristics in frozen muscle biopsies. As in all studies in biopsies, improper freezing techniques could lead to damage within the muscle fibers that could be a potential confound. Most of our protocols were focused on the myofiber border and muscle stem cells along it, which would not be affected by this. However, it could potentially affect the results for the collagen content since it is normalized to the muscle area. We used the same technique for freezing across all samples and do not think this skewed collagen content for any specific muscle group. Finally, although our study was cross-sectional, we interpret some results of myonuclei, myofiber cross-sectional area in the context of the role of MuSC in postnatal development. We think this is justified given the sparse amount of data in this area but the high clinical importance of understanding the possible roles of MuSC in development of muscle contractures.

Overall, muscle contractures in children with cerebral palsy appear to have a higher frequency of small fibers and smaller fiber size. Muscle stem cell abundance in these muscles is higher than noncontractured muscles but myogenic potential is lower. Myonuclear accretion during early postnatal development seems to have occurred across all muscles. Muscle stem cell pool establishment might still be ongoing or altered. Muscle stem cell characteristics are not different between ambulatory and nonambulatory children with cerebral palsy. Future longitudinal studies are needed during the course of postnatal development and before establishment of contractures to unravel the interaction of muscle stem cells, myonuclear number, and myofiber cross-sectional area for appropriate muscle growth.

DATA AVAILABILITY

Data will be made available upon reasonable request.

GRANTS

This study was supported, in part, by the NIH Grant HD094602 (to S. Dayanidhi), and Pedal-with-Pete/AACPDM Grant (to S. Dayanidhi and V. T. Swaroop).

DISCLOSURES

M. Encarnacion was a visiting medical student from Drexel University during the time she did work on this study. Currently, she is a Psychiatry resident at the University of California, San Francisco. None of the other authors has any conflicts of interest, financial or otherwise, to disclose.

AUTHOR CONTRIBUTIONS

R.E.K., V.T.S., and S.D. conceived and designed research; R.E.K., J.E.L., M.E., T.K., N.M.P., V.T.S., and S.D. performed experiments; R.E.K., T.K., M.E., and S.D. analyzed data; R.E.K., T.K., T.K., and S.D. interpreted results of experiments; R.E.K. and S.D. prepared figures; R.E.K. and S.D. drafted manuscript; R.E.K., T.K., J.E.L., M.E., T.K., N.M.P., V.T.S., and S.D. edited and revised manuscript; R.E.K., T.K., J.E.L., M.E., T.K., N.M.P., V.T.S., and S.D. approved final version of manuscript.