Pax7-Positive Cells/Satellite Cells in Human Extraocular Muscles

Abstract

Purpose: We quantified and investigated the distribution of Pax7-positive cells/satellite cells (SCs) in the human extraocular muscles (EOMs).

Methods: An immunofluorescence multiple-marker method simultaneously combining two SC markers (Pax7, NCAM), detection of the basement membrane (laminin) and cell nuclei (4′,6-diamidino-2-phenylindole [DAPI]), was used on the anterior, middle, and posterior portions of EOMs from five healthy donors. Pax7-positive cell and SC content, myonuclear content, myofiber cross-sectional area, and myonuclear domain were analyzed in single cross-sections. Between 3915 and 13,536 myofibers per muscle cross-section and myofibers from the entire EOM cross-section were analyzed for quantification of Pax7-positive cells per myofiber (Pax7/F).

Results: The number of Pax7/F in the human EOMs varies along the length of the muscle with twice as high Pax7/F in the anterior part of the EOMs, but within the range of what has been previously reported for normal adult limb muscles. Furthermore, there are Pax7-positive cells in positions other than the classical SC position and the myonuclear domain size of adult EOMs is noticeably smaller than that previously reported for other adult skeletal muscles.

Conclusions: Previous data on differences in Pax7-positive cell/SC abundance between EOMs and limb muscles must be reconsidered and the characteristics of different Pax7-positive cell populations further investigated. Higher numbers of Pax7-positive cells in the anterior portion of the EOMs may have a bearing for strabismus surgery involving sectioning of the muscle fibers.

The extraocular muscles (EOMs) differ from other skeletal muscles in the body in many respects¹ and, in particular, with regard to their response in neuromuscular diseases. The EOMs are selectively spared in muscle dystrophies^2,3 and far less affected than limb muscles in amyotrophic lateral sclerosis (ALS).^4–7 A higher number of satellite cells (SCs), together with a heightened state of SC activation in the EOMs has been proposed to be a possible instrumental factor for their selective sparing in neuromuscular disease.^8,9 

Satellite cells are resident muscle stem cells, per definition located between the plasma membrane and the basement membrane of myofibers.¹⁰ During myogenesis, muscle progenitor cells proliferate and provide myoblasts that differentiate and fuse into multinucleated myofibers. From the end of the fetal period, muscle progenitor cells are found wedged between the basement membrane and the plasma membrane of the myofibers, forming the future SC pool.^11–13 Satellite cells express the paired box transcription factor Pax7¹⁴ and are essential for skeletal muscle postnatal growth and repair.^15–17 Myofibers are long, cylindrical cells containing multiple nuclei that normally are located close to the plasma membrane. Because the myonuclei are postmitotic, additional nuclei gained during muscle growth or cells needed for repair must come from another cell source, the SCs.¹⁵ In their native position between the basal lamina and plasma membrane of myofibers, SCs normally are quiescent in adult muscles, but can respond to environmental stimuli and reenter the cell cycle to generate myoblasts that will differentiate and fuse with existing myofibers or self-renew to maintain or increase the SC pool.^18–20 The cytoplasm-to-myonuclei ratio, also known as myonuclear domain,²¹ is the theoretical volume of myofiber cytoplasm supported by each myonucleus. The size of the myonuclear domain increases throughout postnatal development and is flexible in adult muscle fibers.^22–25 

A highly reliable method to identify and quantify SCs has become available.²⁶ This immunofluorescence multiple-marker method allows the simultaneous visualization of two SC markers, the transcription factor Pax7 (expressed in the nucleus)¹⁴ and CD56/neural cell adhesion molecule (NCAM, present in the cell membrane),²⁷ the nucleus and basement membrane, making the identification of SCs more accurate than previous strategies relying on one SC marker or the use of the different markers on consecutive muscle sections.²⁶ In the process of developing the multiple-marker method, we compared the results obtained in a large number of different human muscles. Interestingly, our preliminary data for EOMs showed large differences compared to the SC numbers in human EOMs published so far.⁹ Furthermore, although the number of SCs per fiber has become the predominant parameter to quantify SCs using light microscopy, the number of SCs per myonucleus and the size of the myonuclear domain also are important parameters to understand the biology of muscle homeostasis and muscle mass adaptation. These parameters are of particular interest in the context of myofiber size of the EOMs being very small²⁸ and similar to that of limb myofibers at early age.^29,30 We have shown previously that the EOMs are far less affected than the limb muscles of terminal ALS patients⁴ and one possible factor involved might be the reported difference in SC abundance between EOM and limb muscles. However, given the background above, it became apparent that a thorough and extended evaluation of the SCs in the human EOMs, including not only the number of SC per fiber, was necessary before we could proceed with studies on EOMs of patients with ALS or other pathology. 

In the present study, we applied the multiple-marker method and compared the number of Pax7-positve cells per myofiber (Pax7/F) and Pax7-positive cells per sublaminar nucleus (Pax7/N), as well as the size of the myonuclear domain in cross-sections along the length of normal human EOMs. 

We found lower numbers of Pax7/F in the EOMs than reported previously,⁹ and distinct differences in the frequency of Pax7-positive cells along the length of the muscle. We also found that a considerable proportion of Pax7-positive cells were located outside of the classical SC niche, and found that the myonuclear domain size of adult EOMs differs dramatically from that of adult limb muscles.³⁰ 

Methods

Muscle Samples

Eight EOMs (four superior rectus, two medial rectus, and two lateral rectus), collected at autopsy from one female and four male human donors (aged 42, 47, 71, 75, and 79 years) within 25 hours after death, were used for the study. None of the donors was known to suffer from any neuromuscular disease. The muscles were collected and studied with the approval of the Regional Ethical Review Board in Umeå, Sweden, following the recommendations of the Declaration of Helsinki. 

Each EOM was longitudinally orientated under the microscope, to keep the muscle fibers as well-oriented as possible, and mounted with OCT cryomount (HistoLab Products AB, Gothenburg, Sweden) on cardboard covered with aluminum foil. The muscles then were quickly frozen in propane cooled with liquid nitrogen and stored at −80°C until further use. Later, the muscles (18–28 mm long) were brought to −24°C and 5- to 7-mm-long samples comprising the whole muscle cross-section were cut out from the middle, the most anterior, and the most posterior part of the muscles and remounted with OCT cryomount for cross-sectioning. Five-μm-thick whole muscle cross-sections were cut at −23°C using a Leica CM3050 cryostat (Leica Biosystems, Nussloch, Germany), collected on gelatin-coated slides and stored in the freezer until further processing for immunohistochemistry the same day. 

Immunohistochemistry

The multiple-marker method²⁶ was used for identification and quantification of muscle progenitor cells, myonuclei, and myofibers. In brief, the sections were brought to room temperature and postfixed in 2% paraformaldehyde (PFA) for 8 minutes and rinsed in 0.01 M PBS containing 0.05% Tween 20 (PBST) for 3 × 5 minutes. The sections were immersed in 5% nonimmune donkey serum (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA, USA) for 15 minutes before each antibody incubation. After each antibody incubation, the sections were rinsed for 3 × 5 minutes in PBST, but PBS was used for the final rinse. First, the sections were incubated (1:10, +4°C overnight) with a CD56/NCAM monoclonal antibody (No. 347740; BD Biosciences, Europe, Plymouth, UK). For visualization of this antibody, donkey anti-mouse FITC (No. 715-095-151; Jackson ImmunoResearch Laboratories, Inc.) was applied at a concentration of 1:50, at 37°C for 30 minutes. Subsequently, a Pax7 monoclonal antibody (1–1.5 μg/mL; Developmental Studies Hybridoma Bank, Iowa City, IA, USA) and a polyclonal antibody against laminin (PC128, 1:15,000; The Binding Site, Birmingham, UK) were applied (37°C for 60 minutes). Thereafter, for visualization of these antibodies, the sections were incubated with donkey anti-mouse Rhodamine Red X (No. 713-295-151, 1:500) and donkey anti-sheep Alexa Fluor 647 (No. 713-605-147, 1:300; both from Jackson ImmunoResearch Laboratories, Inc.) for 30 minutes at 37°C. Nuclei were identified with 4′,6-diamidino-2-phenylindole (DAPI) provided in the mounting medium (Vectashield; Vector Laboratories, Inc., Burlingame, CA, USA). 

For further visualization of Pax7-positive cells and their surrounding environment, 30-μm thick longitudinal sections were prepared for confocal microscopy. Sections were cut as above and collected on Superfrost Plus (Gerhard Menzel GmbH, Braunschweig, Germany) slides. The sections were immunolabeled using a modified multiple-marker protocol: the fixation in 2% PFA was carried out for 30 minutes; 0.05% Triton X-100 was applied for 30 minutes; 10% nonimmune donkey serum was used for 30 minutes: all washing steps were done for 3 × 10 minutes; CD56/NCAM was visualized by donkey antimouse Alexa 488 (1:300; Jackson ImmunoResearch Laboratories, Inc.); the time for incubation with Pax7 and laminin primary antibodies was extended to 2 hours; the time for incubation with all secondary antibodies was extended to 1 hour, and Prolong Gold with DAPI (Molecular Probes, Thermo Fisher Scientific, Inc., Halethorp, MD, USA) was used as mounting medium. 

During evaluation of the staining protocol, we noted difficulties in using NCAM as a single marker for SCs in human EOMs and in addition we found Pax7-positive cells in nontypical positions. Therefore, in the quantification processes outlined below, we based our data on Pax7-positive cells. Please see the beginning of the Results section for a more detailed explanation of why this approach was chosen. 

Image Acquisition and Selection for Analysis

All quantitative analyses were done exclusively on single sections because the use of serial sections can bias quantification of structures thicker than the 5-μm section (such as satellite cells). 

Sections of 5 μm were evaluated using a Nikon eclipse E800 microscope (Nikon, Inc., Melville, NY, USA). Digital images were captured with a ×20 objective using a SPOT RT KE slider camera (Diagnostic Instruments, Inc., Sterling Heights, MI, USA). Images were either captured at constant intervals evenly distributed over the whole muscle cross-section or, in most cases, the whole cross-section was captured by consecutive images. 

The numbers of Pax7-positive cells, myonuclei, and myofibers, as well as the myofiber cross-sectional area were assessed, using a counting frame for unbiased two-dimensional quantification of myofibers with varying cross-sectional area³¹ and Adobe Photoshop CS6 software (Adobe Systems, Inc., San Jose, CA, USA). 

For more detailed imaging of Pax7-positive cells in different positions, thick muscle sections were studied under a Nikon A1R confocal (LSM) imaging system (Nikon Instruments Europe BV, Amsterdam, The Netherlands) with a Nikon Eclipse Ti-E inverted microscope controlled by Nikon NIS Elements interface. Images were captured with an ×60 objective. Optical section thickness was 0.45 μm. The stacks acquired were exported to ImageJ (version 2.0.0, Build 2b76d0d5e2, Open source software licensed under Creative Commons, http://imagej.nih.gov/ij/; provided in the public domain by the National Institutes of Health, Bethesda, MD, USA). Individual Pax7-positive cells were identified within the stack. Cells close to the upper and lower limits of the section were excluded and 3D projections of the stack with all four channels—Yellow (NCAM), Red (Pax7), Blue (DAPI), and Gray (laminin), were generated and exported to AVI format. 

Image Analysis

For the assessment of Pax7/F in the anterior, middle, and posterior portions, all acquired images (12–48 per section) were used. 

For each donor and EOM portion, the number of Pax7/F was calculated by summing up all Pax7-positive cells, that is, Pax7-positive cells associated with the myofibers and interstitial Pax7-positive cells included by the counting frames and dividing them by the total number of myofibers included in those counting frames. 

A randomized subset of ≥9 images from each muscle cross-section was used for the assessment of Pax7/N, the myonucleus-to-myofiber ratio (N/F), the myofiber cross-sectional area, and myonuclear domain. This was done for the anterior portion of the EOM. In the middle and posterior portions, images were stratified before randomization for equal representation of global and orbital layer. Note that the orbital layer is absent in the most anterior part of the EOMs.³² The orbital and global layers were defined based on general morphology, myofiber size, and variation in myofiber diameter. The orbital layer was recognized by its typical location, smaller and more homogeneous myofiber size, surrounding the global layer in a C-shaped form. Sporadic pictures taken where the boundary between the two layers was uncertain were discarded. 

For assessment of Pax7/N, the total number of Pax7-positive cells in a section/layer was divided by the total number of myonuclei plus Pax7-positive nuclei in that section/layer and expressed as a percentage. The number of N/F was calculated by dividing the number of myonuclei in a section/layer by the total number of myofibers in that section/layer. Care was taken to correctly identify all myofibers and their myonuclei, leaving out capillaries and any other cells. 

The cross-sectional area of the myofibers was assessed using the Leica QWin image processing and analysis software (Leica Microsystems, Schweiz AG, Heerbrug, Switzerland). On each captured image, the inner circumference of all myofibers included by the counting frame was traced manually guided by the laminin labeling. Frayed, obliquely sectioned, or otherwise distorted myofibers were discarded. Thereafter, using a custom-made macro, the software automatically calculated the area of each individual myofiber. Finally, the cross-sectional area of each myofiber in each muscle section was exported to SPSS (IBM, Armonk, NY, USA) for further analysis. At least 1200 myofibers were measured for each layer/portion of each subject. In total, the areas of 42,642 myofibers were measured. 

To calculate the mean myonuclear domain size, the sum of cross-sectional areas of the myofibers in each section/layer was divided by the total number of myonuclei present in that section/layer. 

In this study, we investigated the myonuclear domain using the two-dimensional method, as previously done by other groups.^{22,26,30,33,34} Under this protocol, myonuclear domain is calculated as surface area per myonucleus instead of a volume. There are other methods to calculate myonuclear domain, for example, by extrapolating two-dimensional data into a volume, based on the thickness of the sections and the average myonuclei length,^35,36 or by direct measurement of longitudinal isolated myofibers.^37,38 Each method has its own advantages and disadvantages,³⁹ but to facilitate comparison with studies that are of relevance for the current investigation, the two-dimensional approach was chosen for this study. 

An image montage comprising all molecular markers used was assembled to visualize the distribution of Pax7-positive cells throughout the muscle cross-sectional area for two anterior, two middle, and two posterior EOM cross-sections. 

Pax7-positive cells also were found in non-SC positions. Therefore, the localization of Pax7-positive cells was carefully analyzed in a subset of ≥9 randomly selected images from the anterior, middle, and posterior portions of each donor. 

Depending on their position in relation to the myofiber basal lamina and surrounding structures, Pax7-positive cells (>1820) were classified into four distinguishable positions. For sporadic Pax7-positive cells (n = 10) where a certain classification was not possible, these cells were omitted from the calculation of percentages outlined below. 

All ratios involving total number of Pax7-positive cells described above were recalculated using only the number of Pax7-positive cells in the SC position, to determine SC/F, SC/N. 

Statistical Methods

All estimates were investigated with graphs, Q-Q plots, and Shapiro-Wilk tests to probe for nonnormal distributions. For comparison of Pax7/F, Pax7/N, N/F, cross-sectional area, and myonuclear domain between the global and orbital layers as well as between the anterior, middle, and posterior portions, a repeated measures ANOVA was used with a Sidak post hoc correction and a Mauchly's test of Sphericity, which was confirmed to be nonsignificant. Differences were considered significant for P < 0.05. Where appropriate, separate P values for specific pairwise comparisons are presented. Estimates are presented as a simple arithmetic mean ± SD. All statistical analyzes were done using the SPSS Statistics version 21.0 (IBM). 

Results

Similar to human limb muscles, the vast majority of SCs in all portions and layers of the EOM were positive for Pax7 and NCAM (Pax7^pos/NCAM^pos). We have shown previously that a small subpopulation of SCs in limb muscles are Pax7^neg/NCAM^pos26. In EOMs, detectable Pax7^neg/NCAM^pos SCs were very rare. However, a reliable assessment of different Pax7/NCAM subgroups of SCs was not possible in the EOMs due to the presence of a varying number of myofibers with weak NCAM labeling in their periphery and cytoplasm (white asterisks in Fig. 1), meaning that in those myofibers it would be very difficult to differentiate with certainty between a myonucleus and a Pax7^neg/NCAM^pos SC, or whether a Pax7-positive SC was inherently positive for NCAM or simply surrounded by NCAM staining from the myofiber cytoplasm. In addtition to SCs, Pax7^pos/NCAM^pos cells also were found in the interstitium and located in association with other structures in the EOMs. The majority of these interstitial Pax7-positive cells, regardless of where in the EOM they were found, also were positive for NCAM. However, quantification of Pax7^neg/NCAM^pos interstitial cells was not possible, since NCAM as a single marker for SCs is absolutely dependent on the cell being located in the classical satellite cell position inside the myofiber basal lamina. Therefore, we based our quantitative estimates on Pax7-positive cells only. Furthermore, basing the quantification on Pax7-positive cells also permitted comparisons with previous studies on human EOMs that had used the same primary antibody (further elaborated in Discussion, please see below). Despite the complications outlined above it is our experience that the multiple marker method facilitates the identification of Pax7-positive cells, and adds valuable morphologic information. Thus, we decided to keep NCAM in the protocol for this study. 

Figure 1

 

Cross-section from the middle part of a lateral rectus muscle labeled for NCAM, Pax7, laminin, and DAPI. Muscle progenitor cells in three different positions are shown in left, middle, and right columns. Row (a–c) shows images of NCAM labeling (green). Row (d–f) shows merged images obtained with the green (FITC), red (Rhodamine Red X), and white (Alexa 647) fluorescence filters. NCAM labeling is seen in yellow (green + red), Pax7 labeling in red, and laminin in white. Row (g–i) shows merged images of all markers and, therefore, NCAM appears in yellow (green + red), Pax7 in pink/purple (red + blue), laminin in white, and DAPI in blue. Row (j–l) shows images of laminin labeling alone (white). In the left column, a Pax7- and NCAM-labeled cell inside the basal lamina of the myofiber, that is, in the SC position, is shown (vertical large arrows). In the middle column, a Pax7- and NCAM-labeled cell associated with the myofiber, but clearly separated by a basal lamina is shown (horizontal large arrows). In the right column, a Pax7- and NCAM-labeled cell surrounded by a basal lamina is seen in the interstitium (oblique large arrows), completely separated from muscle fibers. Weak intracellular labeling with the antibody against NCAM is shown in some myofibers (asterisks), in the left and the right columns. Capillaries (e.g., small thin arrows in row [d–f]) are identified by their intense labeling of the basal lamina with laminin. In contrast, interstitial, nonmyogenic cells lack labeling for laminin and are seen as blue nuclei between muscle fibers (e.g., see large arrowheads in [g, i]). Autofluorescence due to lipofuscin granules is visible in all fluorescence filters (e.g., see small arrowheads in all columns) and autofluorescence due to connective tissue/elastin between myofibers is especially visible in the green (FITC) fluorescence filter ([a, b], [d, e], and [g, h]). Scale bar: 10 μm.

In the anterior portion of the EOMs, the mean number of Pax7/F was more than double (0.076 ± 0.022) that of the middle (0.028 ± 0.005) and posterior (0.028 ± 0.007; P < 0.017 and <0.014, respectively) parts of the EOMs (Table 1). This higher abundance of Pax7-positive cells also was reflected in the proportion of Pax7/N, which was significantly higher in the anterior (15.7 ± 3.7%) than in the middle and posterior (7.9 ± 0.8% and 8.3 ± 0.9; P < 0.039 and <0.050, respectively) portions. 

Table 1

 

Comparison of Estimates From Different Portions of the EOMs

 

The number of myonuclei per myofiber (N/F) was similar for the anterior and middle portions (0.393 ± 0.045 and 0.365 ± 0.031), whereas it was significantly lower in the posterior portion (0.293 ± 0.020) than in the middle and anterior portions (P < 0.031 and 0.016, respectively). 

During quantification of Pax7/F in subject I, the anterior portion of the left lateral rectus muscle was found to contain similar numbers of Pax7-positive cells (0.031 Pax7/F) as those of middle and posterior parts of the other specimens. Therefore, two additional sections closer to the tendon and >2 mm apart were analyzed for the numbers of Pax7/F in that sample. Both sections showed similar results (0.041 and 0.032 Pax7/F) to those observed in the first cross-section (0.031 Pax7/F). To rule out the possibility that this sample (the left lateral rectus of subject I) did not include the most anterior part of the muscle and to determine whether there were possible subject- or muscle specific differences, additional sections from the anterior parts of the medial rectus muscle and the right lateral rectus muscle from the same subject also were analyzed. In the anterior parts of those muscles, the numbers of Pax7/F was 0.073 and 0.058, respectively. Therefore we suspect that the left lateral rectus sample, which was the shortest muscle analyzed (18 mm), had been erroneously resected too far posterior of the tendon insertion into the globe, and did not comprise the most anterior part of the muscle. 

The proportion of Pax7-positive cells in different niche conditions was similar for the anterior, middle, and posterior portions of the EOMs. Pax7-positive cells were identified in four different positions depending on their position in relation to the myofiber basal lamina and surrounding structures (Fig. 2.) with indisputable exemplars observed in sections from all five donors. The vast majority of Pax7-positive cells (80.73%) were closely associated with a myofiber either in the position inside or outside the myofiber basal lamina, while other Pax7-positive cells were identified as individual cells in the interstitium between myofibers (16.42%) or close to the basement membrane of other structures (2.85%), such as small blood vessels and nerves (Fig. 2). Thick (30 μm) longitudinal sections were analyzed with confocal microscopy (Fig. 3) to ensure that interstitial Pax7-positive cells were indeed independent from adjacent myofibers and not simply connected further away than what is discernible from a 5-μm section. Indeed, Pax7-positive cells completely lacking attachment to any myofiber were found in the interstitium (Fig. 3, see also Supplementary Video S1 for 3D rendering of typical interstitial Pax7-positive cell). 

Figure 2

 

Proportions of Pax7-positive cells in different niche conditions, based on 1827 categorized Pax7-positive cells. The vast majority of Pax7-positive cells, irrespective of their position, were NCAM-positive (yellow circle). Pooled data for the anterior, middle, and posterior portions of the EOMs are presented: 80.7% of the Pax7-positive cells (pink) were found in positions close to a myofiber (a–d). These myofiber-associated cells were either in the classical SC position (62.3%, [a, b]), inside the myofibre basal lamina (white) or separated from the immediate proximity of the myofiber plasma membrane as suggested by distinct staining for laminin in the space between the two cells (18.4%, [c, d]). Other Pax7-positive cells were clearly separated from myofibers, lying in the interstitium either surrounded by a distinct basement membrane or not ([e], 16.4%) or observed close to other structures (referred to as other Pax7^pos cells in Table 2), such as blood vessels and nerves ([f], 2.8%).

Table 2

 

Localization of Pax7-Positive Cells

 

Figure 3

 

Confocal longitudinal image rendering of Pax7-positive cells in different typical positions. Subpanels (1) display the combined staining for NCAM (yellow), Pax7 (red) laminin (white), and the nucleus (blue). Subpanels (2) display only NCAM (yellow), Pax7 (red), and laminin (white), and subpanels (3) display laminin (white) and the nucleus (blue). (a) A classical Pax7-positive satellite cell (white short arrow) in sublaminar position, corresponding to the left-hand column in Figure 1 and category (a) in Figure 2. (b) A myofiber-associated Pax7-positive cell (white arrowhead), corresponding to the middle column in Figure 1 and category (d) in Figure 2, close to the end of a myofiber and partially surrounded by a laminin envelope of its own. (c) An interstitial Pax7-positive cell (white long arrow), corresponding to the right-hand column in Figure 1 and category (e) in Figure 2, situated in the interstitium, surrounded by elastin fibrils (seen in green in subpanel [c2] and in turquoise in subpanel [c1]) and completely detached from any adjacent myofiber. An animated 3D rendering of the interstitial Pax7-positive cell is presented in Supplementary Video S1. Each step up/down in the panels corresponds to a 1.5-μm step along the z-axis of the confocal stack. m, myofiber; c, capillary. Scale bar: 10 μm.

Individual data on the proportion of Pax7-positive cells associated and not associated with myofibers in the anterior, middle, and posterior portions of the EOMs is presented in Table 2. 

The numbers of Pax7-positive SCs per fiber (Pax7^+sc/F) and the proportion of Pax7-positive SCs per sublaminar nucleus (Pax7^+sc/N, Table 3) in the anterior, middle, and posterior portions of EOMs showed significant differences mirroring those mentioned above for Pax7/F or Pax7/N. 

Table 3

 

Comparison of Orbital and Global Layers and Classical SC Versus All Pax7 Cells

 

No apparent differences in the pattern of distribution of Pax7-positive cells could be observed in the whole cross-section montages of the anterior, middle, and posterior parts of the EOMs (Fig. 4). In the middle and posterior portions, neither the number of Pax7/F (P < 0.482) nor the proportion of Pax7/N (P < 0.950) differed significantly between the orbital and global layers (Table 3; Figs. 4, 5). 

Figure 4

 

Visualization of the distribution of Pax7-positive cells in the anterior (a, b), middle (c, d), and posterior (e, f) portions in human EOMs. The position of each Pax7-positive cell has been marked with a red dot, as the Pax7-positive cells are otherwise not visible at this magnification. The left-hand panels (a, c, e) are schematic pictures showing the muscle contour in black lines, whereas rifts and previously fat-filled compartments are delineated with thinner lines. Larger blood vessels and nerves are marked in blue. The boundary between the global (GL) and orbital (OL) layer is delineated with a dashed line in the middle (c) and posterior (e) portions, where both layers are present. Because these are freshly frozen specimens, the boundary between layers has a more “ragged” appearance than in EOMs that are obtained following perfusion and fixation. The right-hand panels (b, d, f) show a montage of the original fluorescence images. Please note that, like in the left-hand panels, the Pax7-positive cells have been marked with large red dots, as these cells are otherwise not visible at this magnification. Scale bar: 1 mm.

Figure 5

 

Low-power magnification (×20) of Pax7-positive cells (white arrows) in the orbital (a) and global (b) layers of a superior rectus EOM. The combined staining of NCAM (yellow), Pax7 (red), and laminin (white) are shown. Some myofibers exhibited weak staining with NCAM (asterisks). Scale bar: 50 μm.

The myonucleus-to-myofiber ratio (N/F) did not differ between the orbital and global layers in neither the middle nor posterior portions of the EOMs (P < 0.16, please see Table 3). Interestingly, the lower N/F of the posterior portion did not correlate with any difference in myofiber cross-sectional area, which was similar in the global layer of the anterior 242 ± 47 μm²), middle (262 ± 64 μm²), and posterior (270 ± 49 μm²) portions, and similar for the orbital layer of the middle (190 ± 46 μm²) and posterior (202 ± 34 μm²) portions. Two-dimensional myonuclear domain was similar for the global layer of all three portions (anterior, 641 ± 126 μm²/N; middle, 706 ± 180 μm²/N; posterior, 843 ± 143 μm²/N), with the posterior portion being somewhat, but not significantly, higher (P < 0.068) than the other two portions. The orbital layer of the middle and posterior portions had significantly (P < 0.003) smaller myofiber cross-sectional area than their corresponding global layer (mean difference, 71 μm² smaller than global layer of corresponding portion). The two-dimensional myonuclear domain also was significantly smaller in the orbital layer of the middle (543 ± 160 μm²/N) and posterior (720 ± 156 μm²/N) portions compared to corresponding global layer (P < 0.005 mean difference, 143 μm²/N smaller than global layer of corresponding portion), but there was no statistically significant difference in the two-dimensional myonuclear domain between the orbital layer of the middle and posterior portions (P < 0.057). 

Discussion

The major findings of the present study were: the number of Pax7/F in the middle and posterior part of human EOMs was lower than previously reported for normal adult EOM⁹ and limb muscles,^26,30,40,41 the number of Pax7/F varied along the length of the muscle with higher numbers of Pax7/F in the anterior part of the EOMs, there were Pax7-positive cells in other positions than the classical SC position in human EOMs, and the myonuclear domain size of adult EOMs was noticeably smaller than that previously reported for other adult skeletal muscles.^22,30 

Most immunohistochemical studies on human limb muscle SCs have used either antibodies against CD56/NCAM or Pax7 as markers for SC quantification.^30,42–44 Using our multiple-marker method, we have shown previously that Pax7 and NCAM, in combination with staining for the basement membrane and nuclei, are needed for a reliable identification of all SCs in human limb muscles.²⁶ We found that 94% of SCs in human limb muscle express Pax7 and NCAM, but there also are small numbers of Pax7^pos/NCAM^neg and Pax7^neg/NCAM^pos cells.²⁶ Similar results were observed by Verdjik et al.,⁴⁵ who reported that >96% of CD56/NCAM-positive SCs also were Pax7-positive. 

We realized early on that a reliable distinction between Pax7^pos/NCAM^pos, Pax7^pos/NCAM^neg, and Pax7^neg/NCAM^pos SCs would be possible only in the variable proportion of EOM myofibers that are devoid of NCAM staining, making any quantification of these subsets invalid. Furthermore, Pax7-positive cells also were found outside of the classical SC position in the human EOMs and, therefore, we had to decide whether only SCs or all cells identified by Pax7, regardless of the cell position, should be included in our quantification. The inclusion of interstitial Pax7-positive cells also necessarily excluded any interstitial Pax7^neg/NCAM^pos cells, because the use of NCAM as a single SC marker is absolutely dependent on the classical SC position, as NCAM can be found in other cells and structures outside the myofibers. The current body of knowledge based on in vitro and in vivo studies clearly supports the use of Pax7 as a muscle specific marker for skeletal muscle progenitor cells.^19,46–49 Although EOMs and other skeletal muscles arise from different founder stem cells with distinct myogenic programs, Pax7 is expressed in quiescent and activated SCs and is essential for their muscle progenitor function, survival and expansion in all postnatal skeletal muscles.^19,50–52 We chose, therefore, to include all Pax7-positive cells in our primary quantification and express our data as the number of Pax7/F. The inclusion of all Pax7-positive cells and the exclusion of any Pax7^neg/NCAM^pos SC in our quantification also facilitate the comparison with other studies on human EOM using Pax7 as the only SC marker.⁹ 

There is no formalized consensus on the minimum number of myofibers or SCs required for an accurate measurement. In fact, the minimum number of myofibers will vary depending on the inherent variance of the individual muscle sample studied, which in turn is influenced by the type of muscle studied, the variability of myofiber size, variability of fiber types or the age of the subjects analyzed. Typically, studies of SC content in adult limb muscles base their estimates on 200 to 400 myofibers per biopsy.^{33,41,45,53,54} To take the unknown variance of Pax7 cell distribution in EOMs into account, we opted to base our estimation of Pax7/F on a very large number of myofibers, 3915 to 13,536 per whole muscle cross-section. In the previous study on the number of Pax7-positive SCs in human EOMs, four microscopic fields in four serial sections were used, but the numbers of myofibers analyzed was not presented.⁹ 

In summary, although we cannot exclude the possibility of differences among EOMs, the use of the multiple-marker method and the sample size of the present study provide a solid and reliable data set to calculate estimates of Pax7-positive cells in human EOMs. 

Our results on the number of Pax7/F in human EOMs differ from those previously reported.^9,55 In a study of four human donors with similar age range to our donors and adult cynomolgus monkey EOMs, the numbers of Pax7/F in the middle portion was reported to be approximately 7% (i.e., 0.07) for humans and between 7% and 8% (i.e., 0.07–0.08) for monkey EOMs.⁹ These values are much higher than those found in the middle or posterior portions of the EOMs in our study, which ranged from 0.024 to 0.036 and from 0.022 to 0.040 Pax7/F, respectively. However, in the anterior part of the EOMs from our donors, the number of Pax7/F (0.043–0.103) was more in line with that previously reported for the middle portion of human EOMs.⁹ Given the clear difference in the numbers of Pax7/F found across the length of the EOMs, we speculated that the higher numbers of Pax7/F reported previously for the middle portion of human EOMs⁹ were possibly obtained from the more anterior parts of the muscles. Remarkably, in that study,⁹ Pax7-positive cells were virtually absent in the limb muscles of humans and monkeys, conflicting the commonly held notion that SCs are Pax7-positive and readily detectable in limb muscles.^14,26,56 In contrast, in a recent study using a multiple staining approach similar to that used here, the numbers of Pax7-positive cells per myofiber found in the EOMs and limb muscles of mice were similar.⁵⁷ 

In a study of human inferior oblique EOMs using c-met as a marker for SC quantification, values as high as 34% (i.e., 0.34 SC/F) were reported.⁵⁸ Unfortunately, c-met is a very unreliable marker for SC in human muscle, as we have shown previously.⁵⁹ In addition to SCs, c-met also labels non-SC nuclei in myofibers, other structures, such as myofiber membranes and small vessels, and it does not label all Pax7-positive nuclei.⁵⁹ 

Compared to adult limb muscles, 0.02 Pax7^SC/F or 0.03 Pax7/F in the middle portions of human EOMs can be considered a relatively low “SC” density. The numbers of SC/F identified using antibodies against Pax7 and/or NCAM in previous studies on different limb muscles of healthy human adults report mean values of 0.04 to 0.12 SC/F^26,30,40,41 and the wide span may partly reflect differences in antibody sensitivity and specificity as well as differences related to age, physical activity, and type of muscle. An alternative expression of SC density is the proportion of SCs per sublaminar nucleus (SC/N). Our EOM middle portion estimate of 7.9% Pax7/N is slightly higher, whereas the estimate of Pax7^sc/N (5.0%) is more comparable to the proportion of SC/N previously reported for some adult limb muscles (range, 1.4–5.18%).^26,30,40,41 However, keeping in mind that EOM myofibers are small, we argue that low numbers of Pax7/F while high proportions of Pax7/N in our study simply reflect the low numbers of myonuclei per myofiber in the EOMs. 

The present study revealed variable abundance of Pax7-positive cells along the length of the EOMs. A systematic quantification of SCs close to the myotendinous junction, to the best of our knowledge, has not been performed in human muscles, but a previous morphologic electron microscopy study on deep muscles of the spine remarked that there is a high prevalence of SCs near the myotendinous junctions.⁶⁰ Similarly, a developmental study on chicken reported that SCs are more abundant close to the tendon.⁶¹ The higher abundance of Pax7-positive cells in the anterior part may be clinically relevant for the success of strabismus surgery procedures involving sectioning of the muscle tissue itself in contrast with procedures involving only section of the tendon or plication of the EOMs. Further studies are needed to determine the role of these cells in the regulation of final muscle length in surgical procedures that are likely to trigger a response of the SCs to heal muscle fibers. The difference in Pax7-positive cells along the length of the EOMs must be taken into consideration in future studies and when different data on Pax7-positive cell or SC abundance is compared. 

In our study, we found no difference in Pax7/F and Pax7/N between the orbital and global layers. Earlier studies on limb muscles have shown a relationship between myofiber area and SC numbers.²⁶ Also, in hypertrophic training studies, increases in myofiber area are associated with a parallel increase in SC/F.^30,40,62,63 However, myofiber areas in EOMs are practically one order of magnitude smaller than in limb muscles. Therefore, the size difference between the orbital and global layers, though statistically significant, might be too small to become statistically evident in our Pax7/F ratios. If there is a true difference of Pax7/F between the orbital and global layers, it can be expected to be less than 0.018 (based on our 95% confidence intervals). Similarly, for our Pax7/N ratios, any difference can be expected to be less than 0.048. 

The finding of an isolated, slightly lower N/F in the posterior portion of both layers in the EOMs was an unexpected outcome of our study. Our data suggested a 20% lower N/F in the posterior portion compared to the middle portion, yet unaccompanied by any change in myofiber area. 

The myonuclear domain, an estimate of the cytoplasmic region governed by a single myonucleus, has been shown to vary in response to physical activity and natural growth.²⁴ In adults, the mean myonuclear domain of limb muscle fibers tends to vary between 1600 and 2000 μm² per myonucleus.^26,30,33 In adulthood, individual myofibers can increase their cross-sectional area with and without accretion of new myonuclei, but to accommodate substantial hypertrophy, additional myonuclei, supplied by the SCs, are required.⁶⁴ During infancy, mammals and birds have small myonuclear domains that become progressively larger during natural growth.²² In our study, the mean myonuclear domain of EOMs varied between 543 and 843 μm²/N depending on the portion and layer studied. To our knowledge, this is the smallest myonuclear domain size of adult human muscles reported to date. In a study of myonuclear domain progression from infancy to adulthood²² using a similar method, the youngest infant muscles all clustered around a myonuclear domain of 500 μm². Furthermore, cross-sectional area, N/F, and SC abundance for children at early age show striking similarities with their correlates in our data on EOMs.^22,30,65 

Interestingly, the biomechanical loading of EOMs is remarkably stable across the lifespan, as the eye and orbit undergo much less relative growth⁶⁶ from birth to adulthood than other parts of the musculoskeletal system. Accordingly, postnatal myofibers in EOMs reach their adult cross-sectional diameter at a younger age than other muscles.⁶⁷ In summary, we suggest that our results for the human EOMs reflect a muscle with a physiological relationship between myofiber size, number of myonuclei, and Pax7-positive cells that resembles limb muscle of early age. 

To our knowledge, the occurrence of a mean of 18.1%, and as high as 39.8% (subject V), of Pax7-positive cells located in the interstitial space, independently from a myofiber, has not been reported previously in healthy human skeletal muscle. The positive staining for Pax7 and NCAM strengthens their identification as muscle progenitor cells. Small numbers of Pax7-positive cells with a distinct basal lamina, but in close apposition to the myofiber have been observed previously in normal trapezius muscle biopsies²⁶ and in paravertebral muscle biopsies from Duchenne muscle dystrophy patients.⁶⁸ A recent study by Formicola et al.⁵⁷ supports our finding as they also report a population of interstitial Pax7-positive cells in mice EOM. Interestingly, dystrophic limb muscles and EOMs have similarities in that they have more extracellular space than normal limb muscles. It cannot be excluded that the extracellular space influences the localization of interstitial Pax7 cells. 

Further studies are needed to determine the biological role of the interstitial Pax7-positive cells in EOM. One interesting possibility is that these interstitial Pax7-positive cells have never occupied the traditional SC position during EOM formation and, therefore, may represent a unique population of conserved muscle progenitor cells. In a recent study,⁴⁹ isolated Pax7-positive cells from young and old mice EOMs were found to possess superior growth and renewal capacities compared to Pax7-positive cells isolated from limb muscles, but the reason why these cells perform better in vitro is currently unknown. In the same study (Stuelsatz et al.⁴⁹), it was shown that the number of Pax7-positive cells per milligram of muscle tissue was significantly higher in the EOMs than in limb muscle tissue. Other studies, also using cell sorting based on different markers, also have reported higher numbers of potential stem cells per milligram of tissue in the EOMs than in the limb muscles of reference.^69,70 Our results of low number of Pax7/F in the EOMs do not contradict these studies, as the total number of myofibers per milligram in the EOM becomes many times larger than that of limb muscle, if we take into consideration the diminutive size of the myofibers in the EOMs in relation to those in limb muscles. Consequently, even though the number of Pax7/F is 2-fold lower in the EOMs, the number of Pax7-positive cells per mass unit becomes considerably larger. 

The importance of the stem cell niche, the biomolecular environment between the myofiber plasma membrane and basal lamina, has been discussed frequently.^71–73 Studies on mutant mice indicate that Pax7 expression, homing of limb muscle progenitor cells to the niche and the production of a common basement membrane with the myofiber rely on Notch signaling.⁷⁴ However, lack of Notch signaling does not explain the interstitial location of myogenic progenitor cells in human EOM, since they were identified by their Pax7 expression and were surrounded commonly by a basement membrane. It currently is not known whether the intimate SC myofiber contact within the niche is crucial for the ability of myogenic precursor cells to provide new myonuclei during myofiber growth or hypertrophy. 

In summary, the present study indicates that previous data on a higher abundance of Pax7 cells per myofiber in the human EOMs than that reported previously for limb muscles must be reconsidered. Our finding of a higher Pax7-positive cell prevalence in the most anterior portion of the EOMs prompts further studies on the role of these cells in the regulation of final muscle length after strabismus surgery, particularly in procedures that involve section and repair of the muscle fibers. The report of Pax7-positive cells in nontraditional SC positions prompts further investigation of their inherent characteristics and whether they represent a more stem cell–like population. 

Acknowledgments

The authors thank Anna-Karin Olofsson for excellent technical assistance, Franziska Marschinke and Catarina Conde for help with data collection, and Irene Martinez Carrasco from Biochemical Imaging Center Umea for excellent help with confocal microscopy. 

Supported by grants from the Swedish Research Council (K2012-63x-20399-06-3; Stockholm, Sweden), County Council of Västerbotten (Cutting Edge Medical Research and Central ALF, Umeå, Sweden), The Swedish Association of Persons with Neurological Disabilities (NHR–Neuroförbundet, Stockholm, Sweden), Ögonfonden (Umeå, Sweden), Kronprinsessan Margaretas Arbetsnämnd för Synskadade (KMA, Valdemarsvik, Sweden), and The Medical Faculty, Umeå University (Umeå, Sweden). 

Disclosure: M. Lindström, None; A.E. Tjust, None; F. Pedrosa Domellöf, None