Intermediate Filaments in the Human Extraocular Muscles

Abstract

Purpose.: To investigate the distribution of the intermediate filament (IF) proteins desmin, vimentin, and nestin in human extraocular muscles (EOMs).

Methods.: Healthy adult EOM samples were serially sectioned (5 and 1 μm) and processed for immunohistochemistry, with specific antibodies (Abs) against desmin, vimentin, and nestin and different myosin heavy chains (MyHCs), including the newly characterized Ab MYH7b against MyHC slow tonic. The distribution of desmin was also studied in EOMs at 16 to 18 weeks of gestation.

Results.: Desmin was present in the vast majority of muscle fibers. Notably, muscle fibers that contained MyHC slow tonic were either unlabeled or very weakly labeled with three different Abs against desmin. These muscle fibers had normal cytoarchitecture and intact basement membrane. In fetal muscle, desmin was also absent or weak in myotubes containing MyHC slow tonic. Nestin was detected in a large proportion of muscle fibers in the orbital layer and to some extent also in the global layer, whereas no muscle fibers contained vimentin. Desmin and nestin were enriched at neuromuscular junctions, as in limb muscle. In contrast, some myotendinous junctions lacked desmin or nestin.

Conclusions.: The human EOMs differed significantly from the other muscles in the body with respect to their IF composition. Desmin, hitherto regarded as a ubiquitous muscle cytoskeletal protein, was absent or only present in trace amounts in a subset of normal muscle fibers in adult and fetal EOMs. Nestin, normally downregulated early in the postnatal period, was present in a high proportion of adult muscle fibers.

Introduction

The extraocular muscles (EOMs) are instrumental for optimal vision and they have adapted to execute complex voluntary and reflex eye movements including fast saccades, slow vergence, pursuit, and accommodation. The EOMs have many distinct morphologic features and a unique gene expression profile that set them apart from other skeletal muscles and most likely reflect their functional specialization. ^1,2 The human EOMs have a wide array of muscle fiber types, ³ including multiply innervated fibers capable of slow-tonic contraction, distributed in two distinct layers: the orbital layer (OL) and the global layer (GL). We have systematically studied the human EOMs and shown that they differ from typical skeletal muscles at the cellular and molecular level, in particular regarding the composition of the major proteins determining contraction force and velocity, calcium transportation proteins, extracellular matrix proteins, and neuromuscular junction gangliosides. ^3–7 In the present study, we have investigated the distribution of distinct components of the cytoskeleton in the EOM fibers with focus on the intermediate filaments (IFs), the major elements of the cytoskeleton in muscle fibers. 

The cytoskeleton mediates anchorage of the contractile apparatus to the sarcolemma and to the extracellular matrix, and is therefore important for muscle fiber shape and length and for the transmission of sarcomeric movement to the muscle fiber surroundings. The cytoskeleton is usually subdivided into the extrasarcomeric, the intrasarcomeric, and the subsarcolemmal compartments. The extrasarcomeric cytoskeleton, which is important for maintenance of cell shape and support against mechanical stress, is localized between peripheral myofibrils and the sarcolemma, between the nuclear membrane and the sarcolemma and between adjacent myofibrils, linking them at the Z-disc level. The extrasarcomeric cytoskeleton comprises a network of IF proteins that are approximately 8- to 10-nm thick and are ubiquitous parts of the cytoskeleton in most cells. ^8,9 Three major IF proteins, desmin, vimentin, and nestin, are differentially expressed during development of the muscle fibers, vimentin being downregulated very early. ^9–11  

Desmin is the most abundant IF protein found in mature skeletal muscle, linking peripheral myofibrils to the sarcolemma and to the nuclear membrane and regularly linking adjacent myofibrils at the level of the Z-discs. ^9,12 Desmin is enriched at the neuromuscular junctions (NMJ) and myotendinous junctions (MTJ) of healthy limb muscle fibers. At early stages of development, desmin is arranged in a longitudinal pattern in parallel with the myofibrils, along with vimentin. As the muscle fibers mature, however, the arrangement of these desmin strands becomes more transversal and gathered in between the Z-discs of neighboring myofibrils. ^13,14 Desmin is not required for healthy muscle fiber differentiation during early stages of myogenesis in vivo, ^15–17 even though desmin is the first muscle-specific protein detected during muscle development. Muscle fibers lacking desmin suffer structural damage upon loading and continued use. ¹⁶ Desmin derives from a single copy gene and mutations in the DES gene give rise to a range of muscle disorders called desminopathies ^14,18 that are characterized by abnormal intracellular aggregates of desmin and other cytoskeletal proteins. ¹⁹ Data suggest that desmin may be of importance for signal transduction and intracellular transport ^20,21 as well as for mitochondrial positioning. ²²  

Vimentin is characteristic of mesenchymal tissue and is expressed in myotubes during myogenesis. ^23–25 Vimentin becomes downregulated by the time precursor cells commit to the muscle lineage and start to express desmin. ^13,26 Vimentin is not present in muscle fibers postnatally and it does not compensate for lack of desmin. ^27,28 However, vimentin is present in regenerating muscle fibers. ¹⁴  

Nestin is co-expressed with vimentin and desmin during early muscle cell development and is downregulated postnatally. ^{11,23,29–31} In adult muscle, nestin is present in the subsarcolemmal cytoskeleton only at NMJ and MTJ of normal muscle fibers ³¹ and in the extrasarcomeric cytoskeleton of regenerating muscle fibers. ^27,32  

In the present study, we investigated the distribution of desmin, vimentin, and nestin, the major IF proteins of skeletal muscle, in the human EOMs, in order to elucidate whether the special functional and morphologic features of these muscles translate into a different cytoskeletal organization. Remarkably, we found distinct patterns of distribution for desmin among different muscle fiber types in the human EOMs. 

Materials and Methods

Lateral rectus and superior rectus muscles were obtained at autopsy from four male subjects (42- to 76-years old) with no history of neuromuscular disease. In addition, EOMs were obtained from two human fetuses (16- and 18-weeks gestation). All samples were collected with approval of the Medical Ethical Committee Umeå University, and in agreement with the Declaration of Helsinki. 

Twelve adult muscle samples and the fetal muscle samples were mounted on cardboard with optimal cutting temperature (OCT) Cryomount (Histolab Products AB, Västra Frölunda, Sweden) and rapidly frozen in propane chilled with liquid nitrogen and stored at −80°C until sectioned. Serial cross-sections, 5 μm in thickness, from the anterior, middle, and posterior part of the EOMs were cut in a cryostat (Reichert-Jung, Leica, Heidelberg, Germany). 

Additional muscle samples were stretched on cork plates, fixated in 2% paraformaldehyde for 1 hour, washed in 0.1 M PBS, treated with 10% sucrose in 0.1 M PBS overnight at +4°C and retreated with 20% sucrose, and then stored at −80°C until sectioned. The muscle samples were thawed and very small pieces from the anterior, middle, and posterior parts of each muscle were cut out and placed in 2.3 M sucrose overnight at 4°C. Thereafter, they were mounted on a copper holder and frozen in liquid nitrogen. Semithin (0.5–1 μm) longitudinal sections were cut at −95°C to −100° on a Reichert Ultracut microtome equipped with a FCS cryo attachment (Leica, Nussloch, Germany). 

The samples comprised the whole length of each EOM and sections from the anterior, middle, and posterior parts of each muscle were studied separately. 

Limb muscle samples from the same subjects and additional donors were used as reference. 

Immunofluorescence

All sections were mounted on glass slides and processed for indirect immunofluorescence. In brief, the muscle sections were air dried for 15 minutes, rehydrated in 0.01 M PBS, and blocked with 5% normal serum for 15 minutes. Then the sections were incubated with a battery of well characterized primary antibodies (Abs; Table), either overnight at 4°C, or for 60 minutes at 37°C. After washing in PBS and an additional blocking with 5% normal serum, sections were incubated with the appropriate secondary Ab for 30 minutes at 37°C and finally washed in PBS and covered with Vectashield mounting medium (Vector Laboratories, Inc., Burlingame, CA, USA). The secondary Abs used were pre-absorbed, to ensure minimal cross-reaction to immunoglobulins from other species, and conjugated with Alexa Fluor 488, Alexa Fluor 594 (Molecular Probes, Inc., Eugene, OR, USA), FITC, Rhodamine Red-X, or Cyanine 5 (Cy5; Jackson ImmunoResearch Europe Ltd., Newmarket, UK). Double- and triple-staining procedures using sequential and/or simultaneous protocols were done for careful matching of fiber types, identification of myotendinous junctions (MTJs) using Abs against tenascin or laminin and neuromuscular junctions (NMJs) using alpha-bungarotoxin Alexa Fluor 594 conjugate (Molecular Probes, Inc.). Control sections were treated as above, except that the primary Ab was excluded. 

Table

 

Antibodies Used

Antibody	Epitope	Host	Supplier, References
A4.74	MyHC fast IIa	Mouse mab	Development Studies Hybridoma Bank, Iowa City, IA, USA*
A4.951	MyHC slow	Mouse mab	Development Studies Hybridoma Bank, Iowa City, IA, USA*
A4.840	MyHC slow	Mouse mab	Developmental studies Hybridoma Bank, Iowa City, IA, USA*
MYH7b	MyHC slow tonic	Rabbit pab	Gift from Stefano Schiaffino, CNR Inst. of Neuroscience, Padova, Italy37
ALD19	MyHC slow tonic	Mouse mab	Gift from Donald A. Fischman, Cornell University, NY, USA47,48
PC128	Laminin	Sheep pab	Binding Site Group, Birmingham, UK
V9	Vimentin	Mouse mab	DAKO, Glostrup, Denmark
196908	Nestin	Mouse mab	R&B Systems, Abingdon, UK
D33	Desmin	Mouse mab	DAKO, Glostrup, Denmark
ZC.18	Desmin	Mouse mab	Invitrogen, Paisley, UK
A0611	Desmin	Rabbit pab	DAKO, Glostrup, Denmark
EA-53	Alpha-actinin	Mouse mab	Sigma-Aldrich Sweden AB, Stockholm, Sweden
4A10	Tenascin	Mouse mab	Gift from Donald Gullberg, University of Bergen, Norway49
MTC02	Mitochondria	Mouse mab	Abcam, Cambridge, UK

*   mAb developed by H. M. Blau was obtained from the Developmental Studies Hybridoma Bank, created by the National Institute of Child Health and Human Development of the National Institutes of Health and maintained at The University of Iowa, Department of Biology, Iowa City, Iowa, United States.

In order to asses sarcomeric integrity and morphology Ab against alpha-actinin, which is the major component of the Z-disc that helps anchor the myofibrillar actin filaments ^33–35 and phalotoxin derivate using rhodamine-phalloidin staining (Molecular Probes, Inc.), that binds to and is a specific sarcomeric F-actin marker ³⁶ were also used. 

The sections were photographed with a Spot camera (RT KE slider, Diagnostic Instruments, Inc., MI, USA) connected to a Nikon microscope (Eclipse, E800, Tokyo, Japan). The images were processed using the Adobe Photoshop software (Adobe System, Inc., Mountain View, CA, USA). 

Muscle Fiber Quantification

The percentage of muscle fibers containing MyHC slow tonic that were unlabeled with Abs against desmin was calculated in randomly selected images from fourteen EOM cross-sections of all four subjects. A total of 2561 muscle fibers containing MyHC slow tonic were evaluated. 

The percentage of muscle fibers labeled with the Ab against nestin was determined in randomly selected images from four EOM cross-sections. A total of 5707 muscle fibers were evaluated. 

Results

Muscle Fiber Composition

The muscle fiber composition of the EOMs, examined with the Abs against MyHC fast IIa (A4.74), MyHC slow (A4.951, A4.840), MyHC slow tonic (ALD19), and MyHC extraocular (4A6) was as previously described. ³ Because the gene for MyHC slow tonic has recently been properly characterized in humans as MYH7b ³⁷ and a new specific Ab (pAb MYH7b) ³⁷ was now available, we compared its staining pattern with that obtained with the previously used Ab ALD19³ against the same MyHC isoform. Antibodies ALD19 and MYH7b showed almost identical staining pattern (Fig. 1). The staining intensity of individual muscle fibers in the OL was very similar with both Abs, whereas the labeled muscle fibers in the GL tended to be weaker in sections treated with ALD19 than in sections treated with MYH7b (Fig. 1). 

Figure 1

 

Very similar staining patterns were obtained with the new pAb MYH7b (a) and the previously used mAb ALD19 (b), against MyHC slow tonic. In the OL, the staining patterns obtained with both Abs were identical, as seen in the muscle fibers labeled with an asterisk. In the GL, labeling with mAb ALD 19 tended to be weaker than that obtained with pAb MYH7b, as in the muscle fiber marked with an arrowhead.

The majority of the muscle fibers labeled with ALD19 and MYH7b were located in the OL. Most of these muscle fibers were also labeled with mAbs A4.951 and A.4840 against MyHC slow, as previously reported. ³  

Cytoskeletal Proteins

Desmin, regarded as a ubiquitous skeletal muscle protein, was present in the vast majority of the muscle fibers in the human EOMs examined. Immunofluorescence labeling with Abs against desmin was generally higher in the OL than in the GL. 

Remarkably, a subgroup of muscle fibers that contained MyHC slow tonic and/or MyHC slow in both OL and GL were either unlabeled or only very faintly labeled with mAb D33 against desmin (Figs. 2a–c). Such staining pattern was found in the anterior, middle, and posterior parts of the EOMs and was additionally confirmed using triple labeling with mAb D33 against desmin, pAb MYH7b against MyHC slow tonic, and a pAb against laminin, a marker of the basement membrane of muscle fibers and capillaries, used to confirm the integrity of the muscle fibers (Figs. 2a–c). The desmin-negative muscle fibers were mostly found in the OL, whereas those weakly labeled were generally present in the GL. The percentage of muscle fibers that contained MyHC slow tonic that were desmin negative varied between 43% and 68%. 

Figure 2

 

MyHC slow tonic containing muscle fibers identified with pAb MYH7b (left column) generally showed weak to absent labeling with three different Abs against desmin (middle column; mAbs D33, ZC18, and pAb A0.611). In the right column, the corresponding merged images (MyHC slow tonic + Desmin) are shown. In 5-mm cross-sections (a–c) muscle fibers containing MyHC slow tonic (a) were unlabeled (asterisks) or weakly labeled (arrowhead) with mAb D33 against desmin. Notice the fine sarcomeric structure of the muscle fibers in semithin longitudinal sections (d–o). Examples of MyHC slow tonic containing fibers that contained only trace amounts of desmin (d–f), completely lacked desmin (fiber marked with asterisk [j–l], and upper fiber in [m–o]), or had a typical desmin content as seen in other fiber types (g–i) are shown. Adjacent muscles fibers containing MyHC slow tonic showed different levels of desmin labeling (j–l, m–o). The arrowhead marks the subsarcolemmal distribution of desmin near a myonucleus (k–l).

The absence of desmin or the presence of only very faint labeling with the D33 Ab was confirmed at high magnification in semithin longitudinal muscle sections (Figs. 2d–i) and with two additional Abs against desmin (Figs. 2j–o). 

In order to investigate the integrity and general morphology of the cytoskeleton, semithin muscle sections were treated with different Abs against desmin and either colabeled with an Ab against alpha-actinin, a unique marker of the Z-disc, or the actin binding sarcomeric marker rhodamine-phalloidine. An intact sarcomeric structure was present in all muscle fibers, irrespective of the presence or absence of desmin (Figs. 3a–e). Furthermore, the typical intermyofibrillar location of desmin at the Z-disc level was clearly apparent in the majority of the muscle fibers (Figs. 2k, 3c, 3e). In some muscle fibers there were longitudinal desmin strands between sarcomeres (Figs. 3a–c). In muscle fibers that were only slightly labeled with Abs against desmin a delicate subsarcolemmal line labeled with these Abs was often present (Figs. 3d, 3e). 

Figure 3

 

Two muscle fibers, one lacking (asterisk) and another containing desmin show well-preserved sarcomeric structure revealed with the Ab against alpha-actinin, a Z-disc marker (a–c). Notice the presence of longitudinal strands of desmin that interlink adjacent Z-discs (arrow [a, c]). In (d–e), the subsarcolemmal localization of desmin (arrowhead) is visible in a muscle fiber that otherwise only has weak labeling of the intersarcomeric cytoskeleton. The muscle fiber marked with # was unlabeled with the Ab against desmin, but had a well-preserved cytoarchitetcture as shown with rhodamine-phalloidin (e). Low-magnification images (f–h) of the OL and GL of human fetal EOM at 18-weeks gestation. The long arrows in (f–h) denote MyHC slow tonic containing myotubes lacking immunolabeling with the Ab against desmin and the short arrows identify examples of slow tonic containing myotubes that were weakly labeled with the Ab against desmin. Similar examples of myotubes containing MyHC slow tonic and lacking desmin (white arrowheads) or having weak desmin staining (open arrowheads) are shown at higher magnification in the OL (i) and GL (j).

The results described above were confirmed in muscle sections treated with additional Abs against desmin: mAb ZC18 and pAb A0611 (Figs. 2j–o). Although pAb A0611 recognizes several desmin epitopes, it did not label the majority of the muscle fibers containing MyHC slow tonic or it labeled them only very weakly (Fig. 2n). Notably, regardless of which portion of the EOMs (anterior, mid, or posterior) and of which desmin Ab was examined, closely located MyHC slow tonic containing muscle fibers could show different levels of desmin immunoreactivity (Figs. 2a–c; 2j–o). 

In cross-sections of fetal EOMs of 16- and 18-weeks gestation, the majority of the myotubes labeled with MYH7b in OL were consistently unlabeled with Abs against desmin. In the GL these myotubes were, in general, mostly faintly labeled and in some cases unlabeled (Figs. 3f–j). 

The muscle fibers displaying weak or absent desmin immunolabeling did not show any divergent pattern of distribution of their mitochondria, in cross-sections of adult EOMs treated with mAb MTC02 (not shown). 

The Ab against vimentin labeled cells located in the connective tissue and in blood vessels, but it did not label any muscle fibers in the human EOMs (not shown). 

A large proportion (74%–81%) of the muscle fibers in the human EOMs were labeled to some extent with the Ab against nestin, with clear variation among the samples studied (Fig. 4a). In most samples, the majority of the muscle fibers in the OL were moderately to strongly labeled, whereas in the GL there was more variation in staining intensity and a larger number of fibers tended to be unlabeled or only weakly labeled with this Ab (Fig. 4a). In the GL, the muscle fibers labeled with nestin were generally found in areas rich in NMJs, whereas in the OL this correlation was not apparent. In semithin longitudinal sections, a striated pattern compatible with Z-disc associated extrasarcomeric location of nestin was observed (Figs. 4b–e). There was no clear correlation between the staining levels observed with Abs against desmin and nestin (Figs. 4d, 4e). In particular, muscle fibers with weak or absent labeling with antidesmin were not more strongly labeled with the Ab against nestin. 

Figure 4

 

A large proportion of the muscle fibers in the OL were moderately to strongly labeled with the Ab against nestin (a), whereas a larger variability in the staining intensity was observed among the muscle fibers in the GL. In (b–c) the upper muscle fiber (asterisk) showed a clearly striated pattern compatible with a localization of nestin interlinking myofibrils at the Z-discs (b) but only very faint labeling with the Ab against desmin (c). The muscle fiber marked with # in (d–e) showed strong labeling with Abs against nestin (d) and desmin (e), whereas the muscle fiber at the bottom was only labeled with the Ab against desmin (e).

Desmin and nestin labeling was stronger in the immediate vicinity of “en plaque” neuromuscular junctions (Figs. 5a–f). 

Figure 5

 

Both desmin (a) and nestin (d) showed strong labeling intensity adjacent to NMJs (arrowheads) identified with alpha-bungarotoxin (b, e). Merged images of (a) and (b) are shown in (c) and merged images of (d) and (e) are shown in (f).

Different patterns of immunolabeling were noted at the MTJs. The majority of the muscle fibers showed strong desmin labeling at the MTJ, irrespective of whether they were strongly (Fig. 6a) or only weakly (Fig. 6b) labeled with this Ab along their length. Muscle fibers lacking desmin did not show any immunolabeling at the MTJ either (Fig. 6c). Nestin, which in adult skeletal muscle is selectively present at the MTJs, was completely absent at MTJs of muscle fibers unlabeled with the Ab against nestin along the rest of their length (Fig. 6d). Muscle fibers labeled with the Ab against nestin along their length showed increased staining intensity at the MTJs (Fig. 6e). 

Figure 6

 

Myotendinous junctions identified with Abs against tenascin (a–c) and laminin (d–e), the tendon side of the MTJs is labelled red. Muscle fibers displayed strong desmin labeling at the MTJ, irrespective of whether they were strongly (a) or only weakly (b) labeled with this Ab along their length. Muscle fibers lacking desmin did not show any immunolabeling at the MTJ (c). Nestin was completely absent at MTJs of muscle fibers unlabeled with this Ab along the rest of their length (d), whereas muscle fibers labeled with the Ab against nestin along their length showed increased staining intensity at the MTJs (e).

The Abs against desmin labeled all muscle fibers in limb muscle samples and higher intensity of staining was seen at NMJs and MTJs. The Ab against vimentin did not label any muscle fibers. Labeling with the Ab against nestin was only found in the immediate vicinity of NMJs and MTJs. 

Discussion

The present data indicate that the human EOMs differed from other skeletal muscles with respect to the IF protein composition of their cytoskeleton in that (1) a subpopulation of muscle fibers apparently lacked or had very low levels of desmin, (2) a large proportion of the muscle fibers contained nestin, and (3) desmin and nestin were absent from certain MTJs. Although desmin has been regarded to be a ubiquitous muscle cytoskeletal protein and the major IF protein in muscle fibers, the present data indicate that a subset of muscle fibers in the human EOMs, particularly fibers containing MyHC slow tonic, lacked or only contained trace amounts of desmin. This is a unique finding, as desmin has not previously been reported to be absent from normal muscle fibers. The fact that similar results were obtained with two different monoclonal and a polyclonal Ab in semithin (0.5–1 μm) muscle sections strongly suggests that the absence or low levels of desmin observed were true and did not reflect differences in epitope availability. The preserved basement membrane visualized with an Ab against laminin and the well-organized cytoskeletal architecture revealed by labeling with antialpha-actinin and/or rhodamine-phalloidine indicate that these fibers were intact and healthy. 

Absence of desmin in knockout mice leads to a muscle dystrophy, with ongoing muscle fiber degeneration and regeneration, in muscles that are highly used or weight bearing, whereas other muscles are not significantly affected. ^14,16 Desmin-related myopathies, in human patients, are not caused by lack of desmin and are instead characterized by abnormal deposits of desmin. ^14,38 Therefore, it is not possible to explain the present findings as an ongoing muscle dystrophy in the EOMs of four healthy subjects, in particular given that the EOMs did not show any disturbed morphology and the integrity of the basement membrane and healthy cytoarchitecture of the muscle fibers could be clearly established. In the desmin-null mice, although most muscle fibers have well-organized myofibrils, an abnormal myofibrillar organization is present in the muscle fibers of highly used muscles, such as the tongue. ¹⁴ The EOMs are highly used muscles, but we found no signs of abnormal myofibrillar organization, which is further indication that the absence/low levels of desmin found in the slow-tonic muscle fibers of the human EOMs is not an abnormality but rather a special property of these fibers. Similarly, we did not find any disturbance in the pattern of mitochondria distribution in these muscle fibers, in cross-sections, in contrast to the abnormal accumulation of subsarcolemmal clumps of mitochondria present predominantly in slow muscle fibers of desmin-null mice. ²⁷ In addition, the findings of absent- or low-levels of desmin in the myoutubes that contained MyHC slow tonic in the fetal muscles indicate that desmin is either downregulated early in development or never present in the myotubes that mature into slow-tonic muscle fibers. We have previously shown that the slow-tonic myotubes in the human EOMs are distinct from the remaining myotubes already at the primary myoutube stage, at 10-weeks gestation. Further studies are needed to elucidate whether desmin is downregulated very early or never present in these myotubes. 

In the present study there was a spectrum of patterns regarding the desmin content of slow tonic muscle fibers: (1) apparent lack of desmin in the whole muscle fiber, (2) apparent lack of desmin linking myofibrils but with desmin in the subsarcolemmal position, (3) weak desmin immunolabeling both linking myofibrils and in the subsarcolemmal position, and (4) desmin pattern identical to that of other fiber types that do not contain MyHC slow tonic. We have previously shown that the muscle fibers in the human EOMs have a very complex composition and show significant heterogeneity in their levels of expression of key contractile proteins, such as MyHC isoforms and myosin-binding protein C ^3,6 and the fiber relaxation rate regulating proteins SERCA-1 and -2. ³⁹ Important heterogeneity in the levels of expression of these key proteins is seen among muscle fibers in adjacent fascicles. ⁴⁰ The present finding of different levels of desmin in slow tonic muscle fibers fits the concept of a continuum of muscle fiber types in the human ⁴⁰ and rabbit ⁴¹ EOMs, likely reflecting the very small size of the motor units in these muscles and fine adaptation to complex functions. 

It is difficult to speculate on possible functional implications of the variation in desmin content found in the slow tonic muscle fibers, but lack of desmin along the whole muscle fiber, including the MTJ is likely to reflect on its biomechanical properties for force transmission to the cell membrane, extracellular matrix, and tendon. ^42,43 The muscle fibers containing slow-tonic MyHC represent approximately 15% of the muscle fibers in the GL and 17% in the OL of human EOMs, ³ are multiply innervated and they also contain MyHC slow. In limb muscles, slow muscle fibers have a higher content of desmin than fast muscle fibers ⁴² and it has been proposed that cytoskeletal reinforcement is needed to meet their ability to maintain position and generate forces during large periods. ⁴² Further studies addressing the ultrastructural organization of the slow-tonic muscle fibers with different levels of desmin may provide clues to the functional implications of lack of desmin, low levels of desmin, or predominant subsarcolemmal localization of desmin. 

Longitudinal strands of desmin were noted in some muscle fibers of the human EOMs. Similar strands also occur in healthy human soleus, and are therefore not regarded as a reliable sign of ongoing remodulation in mature skeletal muscle. ^44,45  

Nestin did not apparently compensate for the absence or low content of desmin, as we found no direct correlation between the levels of immunolabeling between markers of these two IF proteins. Vimentin was not present in any muscle fibers in the human EOMs, so it did not substitute for desmin either. Similarly, in desmin knockout mice, the lack of desmin is not compensated for by vimentin ^27,28 or nestin. ³¹  

Nestin is transiently present during muscle development and it is downregulated postnatally. ³⁰ In the adult muscle, it is present at the NMJs and MTJs of healthy muscle fibers ^31,46 and in the extrasarcomeric cytoskeleton of regenerating muscle fibers. ^27,32 The presence of nestin in a high proportion of adult muscle fibers in the human EOMs was not paralleled by expression of vimentin, a typical marker of ongoing muscle fiber regeneration in skeletal muscle. ³² Furthermore, the muscle fibers containing nestin did not show any pathological signs. However, we cannot exclude that, to some extent, the presence of nestin may reflect a dynamic cellular reorganization process, as suggested for the MTJs. ⁴⁶ Denervation experiments in the rat indicate that nestin expression levels and pattern of distribution are regulated by innervation. ⁴⁶ In the GL of human EOMs, the muscle fibers labeled with the Ab against nestin were generally weakly labeled and they were found in areas rich in NMJs. We interpret these results as indicative of the presence of nestin in a segment of the muscle fiber that extends beyond the NMJ. In contrast, the labeling found in the OL was stronger, present in large numbers of muscle fibers and not as clearly related to the presence of NMJs, which suggests a more uniform presence of nestin along the length of these fibers. Notably, at the very anterior tip of the EOMs, where OL fibers no longer are present, there were muscle fibers containing nestin at a fair distance from the MTJ. 

In desmin knockout mice, MTJs maintain a normal morphology in spite of the lack of desmin. ³¹ Likewise, the muscle fibers lacking desmin also showed a normal appearance of their MTJs. The existence of MTJs lacking nestin has not been previously reported, to the best of our knowledge, and it is difficult to speculate on the functional implications of this divergent organization of the cytoskeleton at the very site of force transmission to the tendon. 

In summary, the present data on the IF composition of the human EOMs show that these muscles differ significantly from limb and trunk muscles and alter the concept of desmin being a ubiquitous protein in striated muscle by demonstrating the occurrence of intact muscle fibers lacking desmin in healthy adult and fetal muscle. 

Acknowledgments

The authors thank Anna-Karin Olofsson and Margaretha Enerstedt for excellent technical assistance and Stefano Schiaffino, Donald Gullberg, and Donald A. Fischman for the gift of Abs. 

Supported by grants from the Swedish Research Council (K2012-63x-20399-06-3; Stockholm, Sweden), the County Council of Västerbotten (Cutting Edge Medical Research and Central ALF; Umeå, Sweden), Stiftelsen Kronprinsessan Margaretas Arbetsnämnd för Synskadade (KMA; Valdemarsvik, Sweden), the Kempe Foundation (Örnsköldsvik, Sweden), the Swedish Society of Medicine (Stockholm, Sweden), the Medical Faculty Umeå University (Umeå, Sweden). 

Disclosure: A.H. Janbaz, None; M. Lindström, None; J.-X. Liu, None; F. Pedrosa Domellöf, None